Deep web talk:WikiProject Elements: Difference between revisions

From Deep web, the free encyclopedia
Jump to navigation Jump to search
(→‎PT of chemically active subshells: Could you please add supporting citations)
Line 3,837: Line 3,837:
:[[User:Double sharp/Idealised electron configurations]] [[User:Double sharp|Double sharp]] ([[User talk:Double sharp|talk]]) 00:26, 15 February 2020 (UTC)
:[[User:Double sharp/Idealised electron configurations]] [[User:Double sharp|Double sharp]] ([[User talk:Double sharp|talk]]) 00:26, 15 February 2020 (UTC)
{{ping|Double sharp}} Wow! This is rather good, and deserving of more study. You're putting your money where your mouth is, so to speak, Bravo. [[User:Sandbh|Sandbh]] ([[User talk:Sandbh|talk]]) 06:23, 15 February 2020 (UTC)
{{ping|Double sharp}} Wow! This is rather good, and deserving of more study. You're putting your money where your mouth is, so to speak, Bravo. [[User:Sandbh|Sandbh]] ([[User talk:Sandbh|talk]]) 06:23, 15 February 2020 (UTC)
{{ping|Double sharp}} Preliminary impressions:
*Could you please add supporting citations, including for p-d-f involvement in He, Be and Mg; ditto in H, Li and Na; s-p involvement in Pd; and d involvement in Zn-Cd-Hg?
*Do gas phase electron configurations still have any relevance to the chemistry of the elements? [[User:Sandbh|Sandbh]] ([[User talk:Sandbh|talk]]) 10:24, 15 February 2020 (UTC)
== Noble gas notation for period 8 ==
== Noble gas notation for period 8 ==

Revision as of 10:24, 15 February 2020

WikiProject Elements (Rated Project-class)
WikiProject iconThis page is supported by WikiProject Elements, which gives a central approach to the chemical elements and their isotopes on Deep web. Please participate by editing this article, or visit the project page for more details.
 Project  This page does not require a rating on the quality scale.

Article alerts

Good article reassessments

Articles to be merged

Featured article FA A-Class article A  GA B-Class article B C-Class article C Start-Class article Start Stub-Class article Stub Featured list FL  List Deep web Book Book Category page Category Disambiguation page Disambig  Draft  File  Portal  Project Redirect page Redirect  Template  NA  ??? Total


32 featured content items:

Featured articles (30)
H, He, O, F, Ti, Zn, Ge, Y, Nb, Tc, Xe, Cs, Ir, Pb, At, Fr, Th, U, Pu, Cf, Db, Nh, Ts, Og, noble gas, metalloid, periodic table, heavy metals, radiocarbon dating, history of aluminium

Featured topics (2)
noble gases, period 1 elements

Deep webSignpostIcon.svg Signpost interviews: 2011, 2013

Long-term topics, archived
For inspiration

The location and constitution of Group 3 of the periodic table

I’ve drafted an article for publication in an academic journal and would like to have it peer reviewed before submitting the final version.

Could you please let me know if you’re interested in providing me with a peer review and I'll send you a Dropbox link.

Abstract: The constitution of Group 3 can be pragmatically resolved on the basis of the context within which the applicable periodic table is being used, rather than an inconclusive comparison of the physical, chemical, or electronic properties of La and Lu. From an IUPAC perspective, since the periodic table is primarily the domain of chemistry, considerations of Group 3’s neighbours; predominant differentiating electrons across the four blocks of the periodic table; the periodic law; and the nature of the rare earths, support—in a chemical periodic table—Group 3 as Sc-Y-La-Ac. This is the predominant form of periodic table in the chemistry literature.

The role of IUPAC
Physical, chemical, and electronic properties: inconclusive
A philosophical perspective
     Quantum mechanics
     Regularity and symmetry
     Natural kinds
Chemical, pedagogic, and designer periodic tables
The domain of chemistry
     Immediate neighbours
     Differentiating electrons
     Periodic law
     Rare earths
References and notes

Thank you! -- Sandbh (talk) 21:33, 25 December 2019 (UTC)

@Sandbh: I would be interested in providing a review, though I would like to know what exactly such a review would entail before we proceed – for example, is there a dedicated focus (copyediting, fact-checking) or is it an overall review more along the lines of a PR/FAC over here? Also, would I be correct in assuming the journal will provide a (second) review once you submit your manuscript?
Please let me know; if we proceed, I might be able to start reviewing this weekend. ComplexRational (talk) 01:12, 26 December 2019 (UTC)
@ComplexRational: It’s an informal review driven by whatever interests you, with no particular obligations. Yes, the journal will arrange for formal peer reviews once I submit the finished article. Thank you! Sandbh (talk) 08:23, 26 December 2019 (UTC)
@Sandbh: Count me interested as well! Double sharp (talk) 06:12, 26 December 2019 (UTC)
Righto! Will send a Dropbox link to your PM address shorty. Good stuff. Sandbh (talk) 08:23, 26 December 2019 (UTC)
@Sandbh: I'd be glad to give it a dekko. YBG (talk) 05:37, 27 December 2019 (UTC)
Thank you! On the way. Sandbh (talk) 22:38, 27 December 2019 (UTC)


@Sandbh: I, too, would love to help, but I am not sure if I can. By when do you want to be done with the peer review? If there is a month in hand, then I'll be sure to help you. If not, it may be harder, but I'll still try. (also, I am interested how the subtitle "Quantum mechanics" fits under "A philosophical perspective." I believe an interesting case could be made here.)--R8R (talk) 21:40, 29 December 2019 (UTC)

@R8R: The philosophical pespective section starts like this:

A further option is to step back from the minutiae of physical, chemical, and electronic properties and trends and take a philosophical approach. Such an approach seeks to understand if there is a fundamental structure or theory of physics underlying the organisation of the periodic table. If so, this may provide a perspective on Group 3. Quantum mechanics; regularity and symmetry; and natural kinds (“carving Nature at its joints”) have been inconclusively explored to this end.

I'll PM you a link to the article and see how you go. Sandbh (talk) 06:49, 30 December 2019 (UTC)

I'm sorry I haven't written anything yet. The discussion below is fascinating, a true delight to read. I overestimated myself, however, and as I am already feeling dizzy in the end of the day, I will not be able to write anything today as I thought I would. However, I have a few comments to make and I intend to write them soon enough (as afraid as I am to say this, given that there were two other instances of me wanting to say something but not writing it down last year; I hope I'll finish off all of these problem soon enough). Please assume the added subheader as the initiation of the response.

One comment about the thread below that I'll give is that it goes beyond what was written in the original article; I'll generally try to focus on the article instead.--R8R (talk) 19:37, 12 January 2020 (UTC)

@R8R: I am certainly looking forward to hearing what you think about Sandbh's article, since your comments last time were very thought-provoking and insightful! I can barely handle this flood alone and next week I shan't have much time for it. Double sharp (talk) 00:38, 26 January 2020 (UTC)
I was afraid in late December that I would only be able to respond in a month after and we are exactly there, I'm responding a month after the article has been made available for a preview by us. I'll begin later today.
For now, I've got a question for Sandbh. Did all of the wonderful discussion below result in any change in your article? Do you think something should be amended, some wording should be changed, some arguments revoked or added etc.? One reason I'd like to focus on your article rather than the general group 3 argument (though, of course, it can't be avoided at all) is to help you improve your resulting to-be-submission, and I'd like to know if the discussion has been helpful in that; if so, how, and if not, why.--R8R (talk) 10:30, 26 January 2020 (UTC)

@R8R:@Double sharp:@Droog Andrey: Subject to catching up with the rest of Double sharp’s comments, all of the arguments in the article have, as I see it, withstood the barrage. I do need to fine tune some them. My credit for Double sharp will at least thank him for indefatigable stress testing of the draft. Yes, the discussion has been tremendously helpful in refining and testing my thoughts, including my response to Droog Andrey’s 3IP argument which had me stumped for a while. I’ll be adding the 234 argument, which I’d forgotten about. I probably need to to say more about what I mean by a philosophical approach, and the place of the individual properties of the elements, per Double sharp’s criticism. And I need to say more about the nature of the n + l rule, and it’s lack of a first principles basis, and its d and f block wobbles, and the fallaciousness of arguing for double symmetry in the d and f blocks, on this basis.

It’s ironic to think that this article originated in the obsession by others as to the regularity of the LST and ADOMAH, completely overlooking the extra internal irregularities the Lu form introduces. That’s Nature’s response to glossing over the delayed start to the filling of the 4f subshell, I reckon. That doesn’t mean these tables don’t have their purpose, but not as a conventional table. Sandbh (talk) 11:11, 26 January 2020 (UTC)

@R8R:@Double sharp:@Droog Andrey: I also think the power of the conventional form to capture regularity and irregularity at the same time has been completely overlooked in the literature. Form follows function not the other way round. We need to allocate equal time to beauty/regularity/symmetry and “ugliness”/irregularity/asymmetry. Sandbh (talk) 11:19, 26 January 2020 (UTC)

If you ask me, not if the irregularity is weak and questionable for 4f, and if 5f already explains what is going on quite regularly. The block still ends at the same place (viz. No vs. Lr); it's just that its initial filling is delayed, which is something common throughout he table. But I'll stop, because this is not my section. ^_^ Double sharp (talk) 11:24, 26 January 2020 (UTC)
As far as I am concerned, I don't think of this section as of "mine"; or if it is mine, be my guest (and the same goes to everyone else).
  • the two options are effectively indistinguishable -- this is probably not a very big deal given that this is the opinion of the taskforce leader, but I think this argument could use elaboration in general, because the reader is thrown a few graphs at which they have to analyze all by themselves. Different thinking yields different results.
  • More recently, Labarca and Gonzále (2019)13 argued that quantum mechanics are inconclusive(!) when it comes to the Group 3 question. -- that's not very accurate because that's not what they said. Instead, they make a more general argument that quantum physics can help describe an element but not series of elements, and they say so in the context of H and He, not group 3. Also notable is that they make an argument for a -Lu-Lr group 3, which is based on electron configurations, triads, and electronegativity. The latter two make the bulk of their argument but they received very little to no coverage in the article.

@R8R: L & G's paper is called, "About the elements belonging to group 3 from the periodic table: a new approach".

In section 2 of their paper, which is about the position of H and He, they write:

"The standard periodic table, of 18 columns or of medium length, usually includes the elements through scandium (Sc), yttrium (Y), lanthanum (La) and actinium (Ac) integrating group 3. Agree with Jensen (1982), this grouping was progressively established from the decade of 1940, when the quantum mechanical criterion - that is, the use of orbitals and configurations electronic popularized in the representations of the periodic system by LM Simmons and VM Klechkovskii (Mazurs, 1974) - it was crystallized as the moderno to explain the chemical periodicity (for example, Luder 1943)."

Then they say, "the quantum conic has not satisfactorily resolved the position of these elements".

Could it be that you are reading a translation and I'm reading a machine translation of the orinigal paper and that's where our differences come from? I presume that you are specifically referring with that last sentence to this part:
En este sentido, la mecánica cuántica no ha resuelto satisfactoriamente la posición de estos elementos debido a que «La periodicidad en las propiedades químicas de los elementos es un tema complejo y sólo se refleja aproximadamente en las configuraciones electrónicas de los átomos» (Scerri, 1991, 122). Sobre esta base, coincidimos con Scerri (2007, 242) cuando afirma que «[...] la posesión de un determinado número de electrones en la capa externa no es condición necesaria ni suficiente para la pertenencia de un elemento a cualquier grupo particular».
Or, in English:
In this sense, quantum mechanics has not satisfactorily resolved the position of these elements due

that "The periodicity in the chemical properties of the elements is a complex issue and is only approximately reflected in the electronic configurations of atoms" (Scerri, 1991, 122). On this basis, we agree with Scerri (2007, 242) when he says that "[...] the possession of a certain number of electrons in the external layer is not a necessary or sufficient condition for the belonging of an element to any particular group".

This bit is in section 3, which is the section on H and He. In contrast, the longer quote comes from section 2, which outlines the problem itself. Am I missing something?--R8R (talk) 14:51, 30 January 2020 (UTC)

Translated with (free version)"

In section 4 if their paper where they discuss group 3, they say:

"In Section 2 we showed that in problem of the elements of group 3 the arguments made regarding the properties physical and chemical data, as well as the configurations of the elements involved not they are completely convincing both to decide the permanence of La and Ac, or its replacement by Lu and Lr in periods 6 and 7."

They don't explain why triads are important.

On electronegativity they relied on Jensen’s 1982 article in JChemEd, in which he argued that comparisons in periodic trends for Allred-Rochow electronegativity favour Sc-Y-Lu. That is factually correct but incomplete as an argument. The trend in going down Sc-Y-La is instead similar to the trend seen in the group 2 and 1 metals. Further, the choice of electronegativity scale is a little arbitrary. Pauling values, for example, favour Sc-Y-La. Groups 1, 2, 4, and 5 have the period 6 element somewhat more electropositive than the period 5 element; this works with La (1.1) under Y (1.22) but not with Lu (1.27) under Y. In the Mulliken scale, the values for La (1.74) and Lu (1.70) are both less than that of Y (1.81). Sandbh (talk) 07:11, 30 January 2020 (UTC)

  • A natural kind is a part of the furniture of reality that reflects divisions in the world (“Nature carved at its joints”) that can be considered to exist independently of human classification practices. Examples of natural kinds include electrons, iron, and cats. -- I beg to differ here. "Cats" are a human concept that does depend on humans to function: what's a cat, exactly? If it is spotted and runs fast, is it still a cat? how many mutations should there be for us to say a cat gave birth to what is not one, or vice versa: after all, there was once no cats. Here's a Not-a-Cat Cat, you see where the argument is going. We have a general concept of what a cat is that works in most cases but there are exceptions, and this is a feature of human classifications. (This is also the case with groups and group 3 in particular.)
  • the chemical behaviour of Group 3 generally resembles that of Groups 1−2 rather than that of Groups 4−11 -- that's indeed the case. Group 3, whether -La-Ac or -Lu-Lr, resembles groups 1 and 2 much more closely than it does resemble groups 4 and 5 through 11. This doesn't come as a surprise, however, given that this is not a comparison of the like with the like. A comparison with groups 4 and 5 only would make the difference less staggering and the comparison would be much more even.
    • @R8R: I confess I don't understand what you mean here even though I agree with your conclusion. ^_^ If you mean that group 3 is much more like groups 1 and 2 than groups 4 and 5, then your conclusion seems a bit off, because comparing group 3 with groups 4 and 5 should then produce a big differences. So if your conclusion is that group 3 is not that different from groups 4 and 5 (which is something I agree with, having argued it many times below on the grounds of characteristic transition behaviour being weak for Zr/Hf and Nb/Ta), then I think you need to compare Sc/Y/La or Sc/Y/Lu with groups like 6 through 11 in your second sentence. Double sharp (talk) 16:46, 26 January 2020 (UTC)
      • Ah, how embarrassing. Well, serves me right for not re-reading myself. I hope this correction makes it clear enough.--R8R (talk) 18:59, 26 January 2020 (UTC)
  • R8R is right :) Groups 1 to 3 exhibit a predominately ionic chemistry; groups 4 to 5 are predominantly covalent. That’s a big difference. It then follows that the trends going down group 3 as La follow the trends going down groups 1 to 2 better than the trends going down groups 4 to 5 (which looks better for Lu). Sandbh (talk) 23:03, 26 January 2020 (UTC)
    • Nope. He wrote "A comparison with groups 4 and 5 only would make the difference less staggering and the comparison would be much more even" immediately after that. And as I've said below, there is no big difference, because there is no such thing as "predominantly ionic" or "predominantly covalent". It is all a continuum depending on electronegativity difference and what the counter-anion is. Double sharp (talk) 23:38, 26 January 2020 (UTC)
    • Let me repeat one of my comments below (so now it appears three times): "Look, there is no such thing as a complete volte-face from ionic to covalent. It depends on what the counter-anion is. We go from ionic to metallic (which you're overlooking completely) across the series NaCl, Na2O, Na2S, Na3P, Na3As, Na3Sb, Na3Bi, Na. And that's in group 1, with an example taken straight from Greenwood & Earnshaw p. 81. Ionic vs. covalent is (1) gradual, from covalent to polar covalent to mildly ionic to strongly ionic, depending on the difference of electronegativity between the two elements; (2) not a complete dichotomy, because you overlook metallic bonding; and (3) not split by elements, but rather by electronegativity differences, which have a lot to do with oxidation state (just compare uranium chemistry as we jack the oxidation state up from +3 to +4 to +5 to +6). So I'm astonished to see you put so much weight on "ionic vs. covalent" as a false dichotomy. (And as I keep saying, the general thing across the periodic table is continuity, and the sharp dichotomies you like to point to have a distinct tendency to not exist.) The whole literature, when it sees fit to split "main group" from "transition", universally uses other criteria like "variable oxidation states", "coloured compounds from d–d transitions", "a wide variety of complexes", "formation of paramagnetic compounds". Ionic vs. covalent has nothing to do with it, and nobody uses that as a criterion in the literature for this divide. Respectfully, I put it to you that Zr, Hf, Nb, and Ta by this standards have weak credentials as transition metals, just as weak as Sc for that matter." Double sharp (talk) 23:41, 26 January 2020 (UTC)
Thank you. I’m a bit hamstrung doing this as on an iPhone. Clearly, according to the literature there is something called typical or predominant or characteristic or majority or 80/20 behaviour. This is such a basic concept of classification science that I’m astonished I have to say so. For example group 18 is characterised by its predominantly noble behaviour.
Clearly according to the literature the chemistry of groups 1 to 3 is mainly ionic, and the chemistry of 4 to 5 is mainly covalent. I’d even cautiously extend this to groups 6 to 10. I have to be cautious as I don’t have access to my library. Just the notion of saying groups 6 to 10 are mainly ionic in their chemistry strikes me as nonsense.
I may not have much more to say here since there is nothing to argue about a reality defying notion that there is no such thing as typical behaviour. Sandbh (talk) 02:42, 27 January 2020 (UTC)
I don't reject typical behaviour. I just reject that your dichotomies exist. As does every basic chemistry text in the world that gives examples of more and more polar covalent bonds running straight into ionicity, e.g. C-C, C-N, C-O, Si-O, Si-F, Al-F, Mg-F, Na-F. Or less and less ionic bonds running straight into metallicity, e.g. Na-F, Na-O, Na-N, Na-P, Na-As, Na-Sb, Na-Bi, Na-Na. All of it depends just on EN differences. And deciding which one is "typical behaviour" will involve drawing a line somewhere, which totally overlooks continuity of the trend. (Where are Be and maybe Mg by your standards?) I put it to you that (1) ionic vs covalent is not a strict dichotomy, (2) is not the key factor defining main group vs transition (otherwise we have a pickle with all those TM salts, and Be and maybe Mg say bye-bye to group 2, not to mention H in group 1), and (3) is clearly untenable because, since it depends on EN, the same element can go both ways depending on oxidation state. Everything I said here has ample support from the literature. Just look at any basic inorganic chemistry textbook. Respectfully, you are creating discontinuities that are not there. So, now that we have drilled down to where this continuity actually comes from (EN), all this argument becomes is "group 3 is more electropositive than group 4". Indeed. So what, given that this is purely and simply one step in a continuous trend? Group 2 is more electropositive than group 3, too. (Maybe not Be and Mg...) And group 4 is more electropositive than group 5, too. Double sharp (talk) 07:47, 27 January 2020 (UTC)
@Double sharp: Oh my. You continually distort or descended into minutiae or go of on irrelevant tangents wrt my argument: 1. Group 3 is a transition metal group showing predominately ionic chemistry. 2. Group 4 is a transition metal group showing predominating covalent behaviour. End of story. From a classification science POV this is a more than adequate divide, per your point 1. I agree with your point 2 and have never adopted this position. Your point 3 is a non-question since I am dealing with groups, not individual elements. My argument is as simple as that. Sandbh (talk) 11:45, 28 January 2020 (UTC)
What on earth is irrelevant about this? You made a statement. I responded to it by criticising its terminology and what it refers to. That is about as relevant as you can get. An irrelevant tangent would be to ignore the challenge set by this statement, which is "predominantly ionic" and "predominantly covalent". Well, I have already explained what the problem is with it. And point (3) works just as well for groups, as usual, because group 2: compare Be/Mg with Ca/Sr/Ba/Ra. It is the same effect: increasing atomic size and decreasing charge corresponds to increasing electronegativity. Both effects will exist whether we are going from Tl3+ to Tl+ or from Be2+ to Mg2+. It is you who are introducing artificial differences when the effect is the same, respectfully. Or is anything that doesn't support Sc-Y-La an irrelevant tangent now? It certainly seems so, given that my argument about first-row anomalies is dismissed when I use it for Sc-Y-Lu, but you try to invoke it yourself when attacking He-Be-Mg! Same thing with my argument for condensed-phase configurations: paraded from the rooftops to support Sc-Y-La, but sudden silence when it also argues for Be-Mg-Zn! Double sharp (talk) 12:07, 28 January 2020 (UTC)
  • re the number of discrepancies: does it matter though? the Argentinian paper mentioned above argues it does not, the Lr anomaly is just that, an anomaly that doesn't have any real consequences, and it mentions a previous paper saying that. That's an important question given that this argument goes in the section titled "The domain of chemistry”.
@R8R: Yes, the number of discrepancies in comparison to the n+l rule does matter. The fewer the discrepancies the more regular the table becomes as a representation of the n + 1 rule. I couldn't find where it said in the Argentinian paper that it doesn't. You can't apply this approach for some of the elements and not the others. I've addressed the consequence question elsewhere in the Lr section of this thread. Sandbh (talk) 22:10, 31 January 2020 (UTC)
Just to be sure I'm getting you right. Why does it exactly matter more than the beauty of the contiguous blocks? What is there precise fundamental ruling behind this if we assume there is none for Aufbau? Especially given that it has ten more exceptions anyway? (I will check the thread myself later, though.)
The paper says this:
Respecto de la reciente confirmación espectroscópica de la configuración electrónica del Lr (103) 7s2 7p1 (Sato et. al. 2015), coincidimos con Jensen (2015) cuando señala que la interpretación más plausible sería que este caso es sólo otro ejemplo de un elemento del bloque d con una configuración electrónica irregular (debida a efectos relativistas, como fue mencionado), pero que químicamente es un análogo del Lu más que comportarse como un elemento del bloque p.
Regarding the recent spectroscopic confirmation of the electronic configuration of Lr (103) 7s2 7p1 (Sato et. al. 2015), we agree with Jensen (2015) when he points out that the most plausible would be that this case is just another example of an element of block d with a irregular electronic configuration (due to relativistic effects, as mentioned), but that chemically it's an analogue of the Lu rather than behaving like an element of the p-block.
I am inclined to believe this makes sense. In any case, I am not attacking your paper in order to sink it now; rather, I am pointing at the questions the readers in the IUPAC might have, and I suggest you deal with them before they arise. I realize that I am restating some of the earlier mentioned points but that is in caution that you might have not realized the need to reflect the comments from Double sharp in your paper.--R8R (talk) 22:34, 31 January 2020 (UTC)
@Double sharp: I've added a References section which I think provides sufficient support for my group 3 = typically ionic/group 4 to 12 = typically covalent argument. I've noted your other concerns and will address this once I review all outstanding posts.
  • The periodic law implies, certius paribus, that since La represents the first recurrence of comparable periodicity after Y it should be the one to go under Y rather than Lu. -- does it though? I tend to think this is a very freeform interpretation of the said law because it does not seem to be concerned with these things. Regardless, I'll allow myself to quote a part of the article that I have not mentioned so far:
    Scerri has argued for Lu under Y (and He over Be), since the periodic table can then be arranged, from a philosophical point of view, so that it shows the greatest degree of regularity and symmetry. [...] His argument remains inconclusive as there is no basis to regard regularity or symmetry as fundamental requirements.
For the sake of consistency, the original point would also need to be expanded. For one, I don't think the periodic law implies that, especially given how Sc is not under Al. Even if a case is to be made that Sc should not go under Al while La should go under Y, this is too subtle an argument to say it comes from the vague periodic law.
On this note, I also don't think that the periodic law implies that La should go under Y, or, alternatively, that the periodic law should serve an initial approximation on the basis of which the final decision is to be made. I could expand my thinking on this (though DS did it long before I could), but I presume you realize what I mean here, and it is not covered enough, which will inevitably cause a "yes, but" from some readers. Presumably you want to avoid that and have a strong argument in the first place.
(I note that Sandbh's argument against Sc as eka-Al also seems to force He to go above Be. After all: He is an s-element, Be is an s-element, but Ne is a p-element and hence disqualified, so the next recurrence is Be! But I don't mind this that much, actually. I somewhat think that the ideal periodic table should have helium over beryllium. It emphasises the range of applicability of the "duet" rule, the high anomalousness of the 1s subshell that is like a first-row anomaly on steroids, and how it causes an almost complete lack of incomplete shielding effects that explain for example how Li is "too electropositive" for its position in the periodic table. This way also, the trend for He-Be-Mg nicely complements the trend for H-Li-Na, uniting the two "problem children" of group 1 and 2 among the groups. And He is still predicted to be a full shell, since it's the last element in its row, even if it's separated from the classical noble gases, which now anyway start with the most inert one at Ne. My userpage has a He-Be-Mg + Sc-Y-Lu table. ^_^) Double sharp (talk) 21:12, 26 January 2020 (UTC)
@R8R:@Double sharp: I’ve graduated to an iPad :) It does surprise me that my fairly simple arguments generate such obtuse, tangential or convoluted objections. For example, I see that Double sharp has, after 200k(?) of discussion, accepted that there is such a thing as typical or predominant behaviour. I’ll get to that in another post.
To the best of my knowledge I never claimed that there was no such thing as typical behaviour. I simply claimed that ionic vs. covalent was not the relevant distinction and supported it through examples and the literature. Otherwise, Be and Mg should go over Zn, and H over F. Double sharp (talk) 07:55, 28 January 2020 (UTC)
@Double sharp: I'm of the view that the PT layout is arranged more pragmatically than you give it credit. Only the minimum sufficient change is made, at the time required. The sky will not fall down. Thus, if La goes under Y that is all that will happen. Be and Mg will stay over Ca on regularity grounds including the pattern that the first element of a main group (Be) is apt to constitute a transition to the next (B) main group; the second element (Mg) to the corresponding transition group (Zn); whereas the group character is fully developed for the first time in the third element (Ca).
"Regularity grounds" surely support Sc and Y over Lu, too. So if you want to override it with an argument, you have to make sure your argument would not also override Be and Mg over Ca. That is simple logic. You don't just look for things that confirm your hypothesis, you have to actively put it to the test on cases where you think you already know what the answer should be. Double sharp (talk) 22:58, 31 January 2020 (UTC)
The position of hydrogen in group 1 has been well settled:
Similarities with lithium
  • Both have a significant covalent chemistry
  • Lithium, like hydrogen, when bound to highly electronegative atoms such as nitrogen, oxygen, or fluorine, is capable of forming electrostatic bonds with other nearby highly electronegative atoms
  • Their spectra are similar.
Similarities with the alkali metals
  • In chemical reactions, hydrogen and the alkali metals usually lose their single valence electrons; all usually have a valence of +1
  • Both form similar compounds e.g. hydrogen chloride (HCl) and alkali metal chlorides (XCl)
  • Like hydrogen, most of the alkali metals can form compounds with a charge or oxidation state of –1
  • Hydrogen can stand in for alkali metals in typical alkali metal structures
  • Hydrogen is capable of forming alloy-like hydrides, featuring metallic bonding, with some transition metals.
  • Like an alkali metal and unlike the most of other non-metals, hydrogen forms stable cations in the solution. Droog Andrey (talk) 15:17, 1 February 2020 (UTC)
The anomalous properties of hydrogen can be explained by its unique electron configuration, in that it has no underlying core of electrons. Sandbh (talk) 22:47, 31 January 2020 (UTC)
By this logic, the position of helium in group 2 should also be settled:
Similarities with beryllium
  • Both have a significant oxygen affinity, and oxygen gets the negative charge when they bond with it
  • When helium is coaxed into compounds, the predicted ones are analogous to beryllium compounds, and have no neon analogues
  • Helium bonds are obviously covalent, and beryllium is not a strongly electropositive metal
  • Their spectra are similar
  • Both have essentially no electron affinity
  • Helium can form alloy-like compounds with metals, e.g. FeHe; from helium compounds: "Iron helide (FeHe) was early on claimed to have been found,[89] but the discovery was classified as an alloy.[52]". (added later Double sharp (talk) 14:17, 1 February 2020 (UTC))
And the anomalous properties are perfectly explained just as well by the fact that there is no underlying core and no 1p subshell, the same as for hydrogen. In fact, this is also precisely why He is more reactive than Ne; as soon as the 1s2 subshell is even slightly disturbed, reactivity jumps up, just like what happens when H starts losing its electron. Double sharp (talk) 22:58, 31 January 2020 (UTC)
The periodic law seems simple to me and represents the most important aspect of the periodic table. It is a periodic law which implies a recurrence of properties. As noted elsewhere Sc does not go under Al since Sc is at least a predominately d metal whereas Al is a p metal. From a regularity point of view there is no sustainable case for skipping La and Gd in favour of Lu. Every other part of the table observes the regularity of the periodic law, as enhanced by our understanding of the actual electron configuration filling sequence.
Of course I can always tighten this part of the article, so that’s good. Sandbh (talk) 04:44, 28 January 2020 (UTC)
By that logic, He goes over Be, so I know what will follow too: an argument, perhaps even one that has been dismissed when I use it to support Sc-Y-Lu, to override it and force He over Ne anyway. ;) Double sharp (talk) 12:10, 28 January 2020 (UTC)
  • The second option is awkward, or highly anomalous at best -- I found this particular argument very strange. Why would it matter in the first place? What is the need to have them ordered in a specific bent-line-like order? The general principle would do just fine. I will also recall that this very line of argumentation has already been used by this point, in saying that the elegant appearance of contiguous blocks does not necessarily represent the Nature as there is no point in that. (If it is to be said that now a different argument is being made, then a similar argument should be made here that student will find it more appropriate to have contiguous blocks, though again, I think neither matters.) But more importantly, why would anyone be confused? How are people supposed to take the fact that the platinum group metals are
     Ru Rh Pd
     Os Ir Pt
? Will people realize osmium goes after palladium?
It is not difficult to notice that the elements increase in atomic number by period, and then they do it within each period. So a top element is always lighter than a bottom one regardless of the left-right alignment. The very message you're reading actually also operates on this basis, because the English language has a left-to-right writing, and so a symbol in the line below will undoubtedly follow the symbol in the previous line, regardless of the left-right alignment. So I really can't see why would anyone be confused here or if they somehow (!!) were, why a thirty-second explanation wouldn't satisfy them.
  • @R8R: Double sharp and I have discussed this at some length. My simple argument is based on regularity across the PT, and that Lu introduces an irregularity. As I wrote, “The REM are recognised by IUPAC in the Red Book, as are the Ln and An. So there are 18 groups with vertically increasing contiguous Z, one series with increasing Z contiguously going around a corner (Sc, Y, and the Ln), and two series with horizontal contiguously increasing Z (Ln and An). These are the key vertical and horizontal trends. The horizontal discontiguous trends along the transition metals are of secondary import.” That is all. Sandbh (talk) 23:22, 26 January 2020 (UTC)
    • Since when is "recognised by IUPAC" the be-all and end-all of what is an important chemical series? I bet you will hear "PGM" or "TM" just as often as "REM". Why is your argument prohibitive on one but silent on the others? Double sharp (talk) 23:38, 26 January 2020 (UTC)
  • @Double sharp: The PGM are arranged in order of increasing Z, as are the TM. Sandbh (talk) 22:53, 31 January 2020 (UTC)
  • Not by the way you argue about stretching the REM into a straight line, because neither Cu nor Zn is adjacent to Y in the PT. Of course, by the version of "arrangement" used by everybody else reading a book in English, that is reading line by line from left to right, the REM are arranged in order of increasing Z in a Sc-Y-Lu table just as well. Double sharp (talk) 23:04, 31 January 2020 (UTC)


    • I have read your argument with DS closely enough, and I still think that the particular argument you're presenting is not particularly satisfactory. In a -Lu-Lr table, of course, you would have a vertical Sc-Y-Lu trend. There is, of course, also the horizontal trend if La-Ce-..-Lu, true that. And those are two different trends, and nowhere in the periodic table do we conflate horizontal trends with vertical ones, these are different things. So why is it a problem at all that it's less convenient to shape the two into one when there's no chemical point in doing that in the first place (again, given the argument sits in the section "The domain of chemistry")?
    • But also at that comes another point. In the discussion with DS, you mention that there is a philosophical point of doing so. But then again, not only did I not get that from the article, but also then I have to question: why do we ignore the beauty of contiguous blocks as a philosophical argument, instead effectively dismissing as one of the many varieties, "as there is no basis to regard regularity or symmetry as fundamental requirements.14 Scerri acknowledges that we should be aware of arguments based on regularity or symmetry"? Would a similar line of thinking apply for dismissing this unprecedented vertical-horizontal trend? These two questions make me think that there is a visible imbalance in treatment of pro-La-Ac and pro-Lu-Lr arguments, which casts a shadow on the paper as a whole.--R8R (talk) 11:18, 27 January 2020 (UTC)
      • This is also the feeling I get in this thread and in the article. It feels like La-Ac is being accepted from the start as the null hypothesis and that Lu-Lr is not being treated fairly. Indeed, some arguments for La-Ac made here would support things like Al-Sc or Mg-Zn or Tl-Lr or He-Be (the last of which I would actually support!), but seem to mysteriously become inapplicable when I use them as such... Double sharp (talk) 07:57, 28 January 2020 (UTC)

@R8R: @Double sharp: Beauty is a culturally subjective concept as is a preference for symmetry/regularity or asymmetry. That the vertical-horizontal trend is unprecedented does not mean anything remarkable. It's an outcome of the rare earths category, which has been around for many years. Sandbh (talk) 23:06, 30 January 2020 (UTC)

@Sandbh: REM is no less a subjective concept. The simplest proof of that is that it breaks apart chemically similar elements, on the grounds of their natural occurrence, and that despite IUPAC's ruling people cannot agree on what it contains. It excludes Ac, despite its perfect eka-La chemistry. (Presumably on the grounds of its natural occurrence, but then again Pm seems to be a rare earth according to everybody.) And not everybody includes Sc and Y in it (otherwise, why are there so many articles talking about "REY", meaning "rare earths and yttrium"?). Double sharp (talk) 23:41, 30 January 2020 (UTC)
Oh yeah, I am eager to make that point, too. We make up our groupings. We could classify them like that or we could not. The group 12 question, for instance, is far from clear. "Has been around" is a sound argument, if you ask me, but there are also other types of sound arguments that should not be conflated with this one. (Also, to mirror that, the platinum group metals have been around for a long time, too?)
Saying that, however, I am lacking the exact understanding of why anybody would want to have the vertical-horizontal trend. It really seems very artificial to me and it is really the only argument that I can't make any sense of, at all. I sincerely don't see why anyone would want to conflate vertical and horizontal trends. The simple understanding of the periodic trends would argue that Sc-Y-La is the direction in which things like atomic radius should increase and the atomic radius should decrease as you go towards Lu; what kind of trend is that?
I unironically don't see the point of such a construct. I'm not refusing to see it, I can't find it in the first place.--R8R (talk) 19:50, 31 January 2020 (UTC)
@R8R: +1 Double sharp (talk) 12:46, 2 February 2020 (UTC)
@R8R: The elements commonly recognised as metalloids represent a category arising out of vertical-horizontal trends, with two L-shaped around-the- corner bends. Is this a further example in which you can't see why anyone would want to conflate vertical and horizontal trends? The trend of atomic radius increasing towards La and decreasing towards Lu is a very interesting one, illustrating the impact of the Ln contraction. Sandbh (talk) 21:10, 1 February 2020 (UTC)
@Sandbh: I considered this (I did give it some thought despite my initial opposition) and I must say that this is not the same thing. That the metalliods are a category, as you point out, is beyond question. That they form a trend, which you did not point out, is something I have never thought of to this day and having thought of it, don't find of any substance. You specifically mention stretching of the Sc-Y-La-..-Lu line to form a trend; to the best of my knowledge, nobody ever does that to the metalloids. Metalloids, I would rather argue, depend on diagonal relations to keep a common identity, so to speak. This is in substance very different from any possible Sc-Y-La-..-Lu trend, so the comparison does not apply.--R8R (talk) 12:13, 2 February 2020 (UTC)
Also, to add to that, I still don't understand the difference between the dismissed arbitrarily chosen preference for contiguous blocks and the endorsed arbitrarily chosen preference for the stretchable Sc-Y-La-..-Lu line.--R8R (talk) 12:24, 2 February 2020 (UTC)
@R8R: +1 Double sharp (talk) 12:46, 2 February 2020 (UTC)
@R8R: Interim reply (placeholder). Does the narrow range of EN values for the REM not count as a trend? Same would work for the metalloids, I’d have thought. Sandbh (talk) 12:35, 2 February 2020 (UTC)
@Sandbh: No, because a trend should be monotonic (that's why we talk about trendlines), and it should be clear where it starts and stops. Respectfully, the REM do not fulfil this because of the exclusion of Ac. And neither do the metalloids chiefly because every single diagonal or adjacent relationship in the p-block means something and there is no reason to stop at any particular place. (Also, do you know how low the electronegativity of Si is? Sn is actually closer in EN to the other metalloids than Si is. Bi is even within their range!) Double sharp (talk) 12:53, 2 February 2020 (UTC)
@Double sharp:Nonsense. Excuse me if I sound incredulous. I cannot remember all of the trendlines I've ever seen drawn but many are not strictly monotonic. For that matter, who says all trendlines should be montonic? Ac is irrelevant to the REM (but relevant to the Group 3, and to the Ln). The metalloids start at B and finish at Te. If you want to be pedantic, they stop at B since there is nowhere of lower Z for the metalloids to go; they stop in the vicinity of I, Po and At, since none of them are commonly recognised as metalloids. Sn and Bi are irrelevant since none of them are commonly recognised as metalloids. Sandbh (talk) 04:46, 3 February 2020 (UTC)
@Sandbh: If a trend is predicated on one factor controlling it, then you expect it to change somehow proportionally to that factor. If it stops changing proportionally, that suggests that either (1) the trend has entered a range when it is overshadowed by other contrary effects (like atomic radii past Cs down the alkali metals), or (2) we have passed a breakpoint and a new trend has begun, like what happens to a spring past its elastic limit (or, in the periodic table, what happens once we pass a noble gas). That's why I feel that the fact that the Sc-Y-La-...-Lu trend is not monotonic speaks against it as a unified trend. Indeed, there are two separate effects here: atomic radius goes up as we go down the table, but down as we go to the right, and they just happen to overlap here. Same as everywhere else on the table, e.g. B-Al-Ga-In-Tl intersecting Al-Si-P-S-Cl-Ar. It doesn't make sense to me to conflate them.
So we see your methodology: start with the categorisation we see most in the literature and then go and justify it after the fact without critical examination. Never mind, of course, that the trend from Sb to Bi is not too different from that from As to Sb, or anything like that. Literature says As and Sb are metalloids, Bi is a metal, so we cannot question it. Never mind that list of metalloid lists (your own list) adequately demonstrates the lack of agreement about that. Never mind that chemically, the distance from Sb to Bi is actually less than that from As to Sb, given cationic tendencies starting at Sb going down to Bi. We start with the conclusion and justify it. Excuse me if I sound incredulous. Not to mention that it is really strange to see metalloids being argued for as a well-defined category whose limits are placed down and cannot be crossed, by someone who also claims that they are just a specific type of nonmetal. By that logic there should be no problem continuing the trend over to iodine. Tellurium is a nonmetal, so is iodine! Double sharp (talk) 11:47, 3 February 2020 (UTC)

@Double sharp: REM are not a culturally determined concept. It does not matter that it breaks apart chemically similar elements, on the grounds of their natural occurrence. Self-evidently there were enough other similarities to warrant the category name. Crikey! How about breaking apart the coinage metals on account of their heterogenous chemistry? Ac has traditionally been excluded on the grounds of its radioactivity, and location in period 7 (whereas the REM occupy periods 4, 5, and 6). I think some argy bargy about whether or not to include Sc matters does not matter so much; Sc is recognised as a REM by IUPAC and that's a huge achievement by itself. If people want to focus on the REY as a subset of the REM, power to them. That might be the same as focussing on the refractory metals (another wobbly boundary) or the PGM as a subset of the TM. Even the PGM are wobbly; gold is sometimes admitted into the gang of six. Sandbh (talk) 06:51, 31 January 2020 (UTC)

@Sandbh: You say they are not culturally determined, and then invoke tradition to defend the exclusion of Ac; I rest my case. ^_^ Yes, the PGM are wobbly too. Matter of fact, the transition metals are wobbly too. Even the alkaline earth metals are wobbly too! Each group has similarities, yes, that Nature gave us. But grouping them as categories, that we made up. Double sharp (talk) 07:42, 31 January 2020 (UTC)

@Double sharp: Nice catch! That the REM are not culturally determined does not mean I cannot invoke tradition to defend the exclusion of Ac, and I gave another reason. There is some pragmatism applying here, too. Chemistry rides the fine line between quantitative and qualitative and sometimes we have to argue things out in the absence of a categorical argument. And then, like null hypotheses we have to rely on the strength of arguments rather than being able to quantitatively measure them. Categories and classifications form the core of science and we iteratively test and, where required, improve them; that they are "made up" is irrelevant to the scientific process. Sandbh (talk) 06:59, 1 February 2020 (UTC)

@Sandbh: So first we are arguing that they are "not culturally determined", and then retreat and start defending that they are "made up". I prefer to say: the point is not to draw the table so that one category we make up fulfils one method of drawing that does not even work for many others (e.g. TM, PGM). The point is to draw it to be an excellent first-order approximation that is sure about what order it is. As such, I have a very simple procedure for the drawing and block assignment:
  1. Elements go in blocks based on their chemically active subshell of highest angular momentum. (Then they go in groups by how many they are, rather than the exact configuration: see User:Double sharp/Idealised electron configurations.)
  2. When it is not clear what this is (e.g. groups 2 and 12 both vying for a d-block position, La and Lu both vying for an f-block position), use properties to compare which one makes the resulting block more homogeneous.
  3. There is no step 3.
Short, simple, to the point, based on chemistry, founded on the philosophy that categories should be as homogeneous as possible, and totally extendable to every element from hydrogen through oganesson and on to all those undiscovered superheavies in the eighth period. Only with the ninth period, with that [E172] 6g1 configuration for an s-block-like E173, does it break down. Which is more than can be said for your criteria, which give the "wrong" answer if you apply them elsewhere already within the stable elements everyone knows. Double sharp (talk) 10:08, 1 February 2020 (UTC)
@Double sharp: How does that work for e.g. Nd, which has a condensed phase configuration of 4f3d1s2, and which is not known in the +4 oxidation state? Time to move into the d block? And the second IE of Cs (2234) is less than the second IE of Kr (2350). What happens if Cs can be made to form a +2 cation? Does Cs then become a p element? Sandbh (talk) 03:47, 2 February 2020 (UTC)
  1. @Sandbh: Nd is definitely known in the +4 oxidation state. In fact, an aqueous Nd4+ cation is known, though it oxidises water, so it exists in a bigger sense than Pr(V). So, not the best example you could have come up with. ^_-☆ Let's take Sm as a less bad example. Well, in the gas phase we have f6s2, and we form +3 cations, so yes, we diagnose f-block, no problem. The fact that in the condensed phase an f-electron moves to the d-orbital rests my case that the subshell can be used as a chemical reserve and is hence active.
  2. If Cs breaks the 5p subshell it can be a seventh p-element as well, sure. Ca, Sr, Ba, and Ra are already honorary d-elements, and Be and Mg honorary p-elements because of easy 2s2p hybridisation. In general, this criterion only encounters any problem for s-elements since all unoccupied reachable subshells belong to the next n+l level and have angular momentum above 0. But here we just cover it by electron bookkeeping and keeping the other blocks homogeneous. Stricto sensu the Ca group are d-elements, but we draw them as s-elements due to greater similarity. This is not a killing argument because the s-block is usually weird in other ways as well, e.g. lack of double periodicity and trends that affect all other groups. So we may consider this just a specific problem for s-elements, just like all the other specific problems they have. Whence the second criterion.
  3. Nota bene, the unadorned criterion works perfectly for La and Lu. La emerges as clearly f-block and Lu as clearly d-block. ^_^ The second criterion is not even needed for this quarrel, which suggests that the Be-Mg-Zn old quarrel is actually stronger than this Sc-Y-La quarrel. Double sharp (talk) 10:21, 2 February 2020 (UTC)
@Double sharp: Watching the tennis again and the seeming self-destruction of Novak. A quick comment. I’ll think more about this (placeholder) but I’m inclined to keep the differentiating electron criterion since it is quantitatively unambiguous in all cases. On Sm, is the f electron really “chemically” active? It’s only present in the gas phase of an isolated Sm atom in a vacuum, but it’s still an Sm atom. In condensed phase Sm there is no f electron involvement. There is no comparable alteration in chemical identity that I can see, e.g. Sm v Sm2O3. Sandbh (talk) 11:31, 2 February 2020 (UTC)
@Sandbh: What do you mean "no f electron involvement"? The f-electron got promoted and ionised, Sm3+ has one less f-electron than Sm. That is pretty much the definition of f-involvement. Na as an isolated gas-phase atom is [Ne]3s1; in the metal the 3s electron is delocalised; and in ionic compounds it is ionised away. By the standards you impose on Sm, there is no s electron involvement in Na either, which suggests that the standards you impose are irrelevant. The differentiating electron is quantitatively unambiguous but irrelevant precisely because elements in chemical bonds are typically not in their ground-state configurations. Sometimes they are ionised, and sometimes they are in what would be excited states if they were alone, because the chemical bond energy provides the necessary energy to change configuration. I consider them all at the macro level by just asking if the subshell is participating chemically or not. Double sharp (talk) 11:37, 2 February 2020 (UTC)
@Double sharp: I’ll get back to you on that one. In the meantime I’m more convinced that your approach doesn’t work since it does not answer the group 3 question, and a comparison of individual physical, chemical or electronic properties is inconclusive. Whereas differentiating electrons, consistently applied, are. I’ve explained elsewhere that within-block homogeneity or similarity in terms of the minutiae detail doesn’t trump vertical trends in periodicity and overall chemical behaviour at least at the group level. Sandbh (talk) 12:20, 2 February 2020 (UTC)
@Sandbh: It is totally conclusive on Lu below Y. It answers the group 3 question without problem because 4f is not chemically active in Lu but is in La. If we needed confirmation, the physical and chemical properties are also conclusive. Physically, Lu is similar to the d-block metals, and La is totally not. Chemically, Lu is more similar to the d-block metals than La. In this it matches Sc and Y which are physically totally normal d-metals and chemically have incipient transition behaviour, just like Zr/Hf and Nb/Ta. La does not satisfy any of this. Since Sc-Y-Lu looks just as convincing as a trend by itsef as Sc-Y-La, our conclusion is Lu. This is not minutiae detail, this is all trends and overall unskewed physical and chemical behaviour. Differentiating electrons are consistent and consistently irrelevant minutiae because elements in compounds do not usually follow their ground-state gas-phase electron configurations. Double sharp (talk) 12:35, 2 February 2020 (UTC)
  • The above leaves me thinking the conclusion made is hasty.--R8R (talk) 14:22, 26 January 2020 (UTC)

@R8R: The conclusion is OK such as it is; I think the presentation needs more work, including to address some of the concerns and misperceptions raised in this thread. Sandbh (talk) 22:55, 31 January 2020 (UTC)


Droog Andrey and Double sharp

I am interested also. Droog Andrey (talk) 16:13, 2 January 2020 (UTC)

@Droog Andrey: Righto. Please send me an email address that I can send a dropbox link to. Sandbh (talk) 00:50, 3 January 2020 (UTC)
@Sandbh: Andrey_601 at Droog Andrey (talk) 10:55, 5 January 2020 (UTC)

I was going to finish my reply yesterday, but I see that Droog Andrey has beaten me to the most important points I was going to raise, so if you don't mind I'll sit down in his section and join the tea ceremony! ^_^ Double sharp (talk) 07:00, 9 January 2020 (UTC)

@Sandbh: Thanks a lot. I've noticed a few statements in "The domain of chemistry" section that are at least doubtful.

Chemical behaviour of Group 3

The chemical behaviour of Group 3 generally resembles that of Groups 1−2 rather than that of Groups 4−11 In fact, chemistry of Sc and Y is equally close to Ca-Sr as to Ti-Zr. Chemistry of La-Ac is indeed closer to Group 2, but that of Lu-Lr is closer to Group 4. And that's the reason to consider Lu-Lr, not La-Ac, as d-elements. Droog Andrey (talk) 11:19, 7 January 2020 (UTC)

Sandbh: @Droog Andrey: I'm trying to stay away from arguments comparing individual elements (given Scerri says that such approach is inconclusive) or even pairs of elements. The chemistry of Sc-Y-La-Ac as a whole is closer to group 2, than is the case for Sc-Y-Lu-Lr.
@Sandbh: OK, Sc-Y-Lu-Lr group is closer to transition elements than Sc-Y-La-Ac, that's why Group 3 should be Sc-Y-Lu-Lr instead of Sc-Y-La-Ac. All is clear. Why should we mess up things inserting Group-2-like elements La-Ac into Group 3 and causing a crutch-like separation of Group 3 from Groups 4-10? Droog Andrey (talk) 19:05, 8 January 2020 (UTC)
@Droog Andrey: [S1] The fit of group 3 as Sc-Y-La-Ac to groups 1-2 is better than the fit of Sc-Y-Lu-Lr to group 4. Either way Group 3 shows chemical behaviour that is manifestly uncharacteristic of the transition metals proper. Group 3 does not show the complex coordination chemistry that is characteristic of transition metals; they do not show multiple oxidation states; and they are more reactive and electropositive than any other transition metals, approaching the s-block metals in both properties. Sc-Y-La-Ac, in a 32-column table, does mess things up but Nature does not care. Sandbh (talk) 05:38, 9 January 2020 (UTC)
@Sandbh: Group 4 organometallic chemistry is mostly cyclopentadienyl derivatives, similar to group 3. Zr and Hf in group 4 are very unhappy to show lower oxidation states, like Sc in group 3. And they are also pretty electropositive: Zr and Hf have Pauling electronegativity similar to Mg and Sc. So we can see similarities on both sides; Zr and Hf are more "pre-transition", while Ti starts to show real transition character. If you put Lu in group 3 the similarities to group 4 become even stronger because Lu is better at forming complexes than La. This is all expected: I guess similarly to delayed collapse, if something happens in one group in the 4th period, it will not happen till a group or two later in the 5th and 6th periods. Not only Zr and Hf in group 4, but also to a lesser extent Nb and Ta in group 5, are also quite strongly "pre-transition" in character with chemistries dominated by the group oxidation state; so where's the sharp difference between "pre-transition" and "transition"? I don't see it in what Nature seems to have provided us. This is just another transition between blocks that often happens at their edges. It is just like how you cannot draw the sharpest line for loss of TM character between group 11 and group 12, because you have Ag where 4d has radial nodes but isn't subject to significant relativistic effects yet. Double sharp (talk) 06:47, 9 January 2020 (UTC)
@Double sharp: The challenge is that effectively everyone regards group 4 as transition metal. This is not the case for group 3, which has a large record of not being regarded as a transition metal group from a chemical perspective. I agree there are no sharp boundaries. That said, faced with a choice of Sc-Y-Lu, which behave substantially like pre-transition metals (PTM) but show vertical trends like groups 4–10; or Sc-Y-La-Ac, which also behave like PTM and show vertical trends matching those of groups 1–2, the weight of arguments, as I see it, rests with Sc-Y-La-Ac. Sandbh (talk) 09:02, 9 January 2020 (UTC)
@Sandbh: Exactly how large is the record of group 3 being regarded as not transition metals? I mean, sure, I was taught that in high school, but it's not even one of the definitions IUPAC allows. I'd bet it's much smaller than the record of group 12 being regarded as not transition metals, which is a lot easier to defend. I'd also add that when I was taught that, the focus was all on the 3d metals and only very basic coordination chemistry. Group 3 as a transition metal group makes a lot more sense once you include organometallic chemistry (mostly because of scandium), but obviously you don't see that much of that in high school. Group 4 and 5 are also mostly pre-transition metals (Ti and V are undoubtedly transition, but Zr-Hf-Rf and Nb-Ta-Db are mostly pre-transition in character), which matches group 3 (Sc is the one with the most transition character). Surely Sc-Y-Lu-Lr is a good match for those, at least as much as Sc-Y-La-Ac is to groups 1 and 2, and so other factors decide it (e.g. that Lu and Lr make a lot more sense viewed as the start of 5d and 6d series than as the end of 4f and 5f series; see my response to [S4]). Double sharp (talk) 09:19, 9 January 2020 (UTC)
@Double sharp: I don’t know how many. My recollection is that effectively every book I’ve read on Group 3 apologises for not treating them as transition metals, chemically speaking. Pardon me if I’m confusing concepts but the coordination chemistry of Sc is *far* less than that of the rest of the 1st row transition metals. We are stuck with Y regardless noting it has a fair degree of ambivalence, as to wether it acts more like a light Ln or a heavy Ln. So my impression is still that the sun represents the main-group-like chemistry of group 3, whether La or Lu. Sandbh (talk) 11:03, 9 January 2020 (UTC)
@Sandbh: I'm talking about organoscandium chemistry. Here you can more easily find lower oxidation states for Sc, and the general behaviour (dominated by cyclopentadienyls) is not so different from Ti. And it's not so difficult to find organolanthanide compounds which have Ln(II). Double sharp (talk) 04:46, 10 January 2020 (UTC)
@Double sharp: AFAIK the situation is still essentially like that described by G&E: "In the main, the chemistry of these elements concerns the formation of a predominantly ionic +3 oxidation state arising from the loss of all 3 valence electrons and giving a well-defined cationic aqueous chemistry. Because of this, although each member of this group is the first member of a transition series, its chemistry is largely atypical of the transition elements. The variable oxidation states and the marked ability to form coordination compounds with a wide variety of ligands are barely hinted at in this group although materials containing the metals in low oxidation states can be prepared (see p. 949) and a limited organometallic (predominantly cyclopentadienyl) chemistry has developed." I don’t disagree with what you said but the chemistry of group 3 is still largely atypical of the transition metal elements and is still more like that of groups 1 and 2. Sandbh (talk) 08:28, 10 January 2020 (UTC)
@Sandbh: But almost all of that paragraph also applies to Zr (and thus to Hf and Rf as well). It mostly forms a rather basic +4 state (for Rf it is basic, no questions asked), where all 4 valence electrons are lost, can form a very hydrolysed Zr4+ cation in water (typical for states above +3 even for the most electropositive metals like Th4+ and Pa5+, so that's not a black mark that it is so hydrolysed), has only "rather sparsely represented" low oxidation states (Greenwood and Earnshaw p. 958), and a mostly cyclopentadienyl-based organometallic chemistry. So the main difference is just that Zr is happier to form complexes (mostly with O-donor ligands); its chemistry when measured by the usual yardstick of "multiple oxidation states differing by units of one" is not all that transition-like. Which is kind of expected; it's still near the start of the d-block, and the transition properties are coming in slowly. But it does mean that Sc and Y don't look so much like they all throw totally towards group 2, but rather like elements "on the way" from group 2 to group 4, with similarities both to Ca/Sr and Ti/Zr as Droog Andrey has mentioned. If our standard of "transition" is so high as to exclude Sc, Y, and Lu from it, then it is probably high enough to exclude Zr, Hf, and even probably Nb and Ta from it, at which point the argument "group 2 vs. group 4" loses force: almost everybody involved is "pre-transition" by this standard with titanium as an outlier! And not even always: Ti4+, Zr4+, Hf4+, and Rf4+ are hard acids, just like the group 2 and 3 cations! Double sharp (talk) 07:06, 12 January 2020 (UTC)
@Double sharp: For Zr and Hf the trihalides are well established. Cotton et al. [Advanced Inorganic Chemistry, 6th ed.) spend four pages discussing oxidation states of Zr and Hf below III. G & E add, "Even for Ti they [the lower oxidation states] are readily oxidised to +4 but are undoubtedly well defined and, whatever arguments may be advanced against applying the description to Sc, there is no doubt that Ti is a transition metal." To some extent we are fulling back on the old saw that it is in group 4 that we first encounter the classic properties of transition metals. Hence the cut falls between groups 3 and 4. Sandbh (talk) 06:13, 15 January 2020 (UTC)
@Sandbh: By that standard, Sc is a well-established transition metal by virtue of compounds like CsScCl3. And so are the Ln because of their +2-state organometallic chemistry. Note that G&E are only talking about Ti, not Zr and Hf. If we exclude that by talking about common stable-in-water states in clearly inorganic compounds only, then Zr and Hf go back to being pre-transition. There is no criterion that excludes Sc from being a transition metal without also excluding Zr and Hf. Double sharp (talk) 06:42, 15 January 2020 (UTC)
@Double sharp: The standard I go by is the literature, which I submit is a reasonable one, and effectively 100% of which says that Sc is atypical for a transition metal and that Zr and Hf are transition metals. It is fine to point to the exceptions, which is what makes chemistry so interesting, and there is no case for leveraging these into characteristic considerations. Exceptions are so-named because of their out-of-kilter rather than mundane status. Sandbh (talk) 00:20, 16 January 2020 (UTC)
@Sandbh: IUPAC's definition says nothing about the Sc group possibly being excluded (though it allows the exclusion of the Zn group), so excluding it altogether is probably a minority view. I agree that Sc is not terribly typical, but by the standard G&E are mentioning (and they don't even exclude it!) neither are Zr and Hf (notice that they only say that there is no doubt for Ti). We could spend as much time talking about lower oxidation states of the REM as G&E do for Zr and Hf, as they're now actually known for almost all of them. It doesn't change the fact that Zr and Hf non-organometallic chemistry is predominantly that of the IV state, in which they are quite pre-transition-like. So what? I prefer to say that there is simply a gradual change towards transition metal proper behaviour that arrives later and later in each period, and just point to the split between the s- and d-blocks as a good guide of the region where it appears rather than a strict line which makes the truth look so much simpler than it really is. So: Ba is mostly pre-transition (with a bit of transition character), Hf is mostly transition (with some pre-transition character), and so we ought to pick the one in the middle that falls most closely between them like Sc and Y do for Ca/Ti and Sr/Zr. Lu seems better since its transition character is stronger, whereas for La it is quite weak. Double sharp (talk) 09:21, 16 January 2020 (UTC)
@Double sharp: Oh, I'm not talking excluding the Sc group from the d-block. And yes, the truth is almost more complex than a straight line. I'm being pragmatic. Group 4 is similar to group 3. That said, they are found in several oxidation states and are more likely to form complexes than Group 3. Like Earnshaw and Harrington said, "…this is the first group in which the really characteristic transitional properties of variable oxidation state, colour and paramagnetism are encountered." By any reasonable measure that is good indicator of where to separate the d-block, as has been done in 32-column tables without significant controversy for over at least 50 years. Like Jones said, "Though classification is an essential feature of all branches of science, there are always hard cases at the boundaries. The boundary of a class is rarely sharp…Scientists should not lose sleep over the hard cases. As long as a classification system is beneficial to economy of description, to structuring knowledge and to our understanding, and hard cases constitute a small minority, then keep it. If the system becomes less than useful, then scrap it and replace it with a system based on different shared characteristics." I don't see sufficient merit in changing the status quo. Sandbh (talk) 05:56, 17 January 2020 (UTC)
@Sandbh: The system becomes less than useful if the complexity increases without the increase of explanation power. That's the case of La table: more details, more exceptions, less explanation.
I don't see that. The La table explanations I can remember reading explain that the f-block starts at Ce, rather than La, since that is when the f shell starts filling, akin to B and Sc. Hazel Rossotti's Diverse atoms: Profiles of the chemical elements, (1998) is an excellent example. Sandbh (talk) 04:31, 21 January 2020 (UTC)
The literature simply follows the history. La was discovered before Lu, so it went under Y, but after discovery of Lu the questions began to rise. If Lu had been discovered before La, we would simply have Sc-Y-Lu table and no questions about Group 3 would have appeared at all. Droog Andrey (talk) 13:14, 20 January 2020 (UTC)
@Droog Andrey: There's an analogy with peptic ulcers which were historically attributed to stress, rich food etc. Then it was discovered they were due to bacteria, a development which met huge skepticism. Eventually the bacterium theory replaced the historical view.
La was discovered first, so it went under Y; Lu came later but there wasn't a sufficiently meritorious case made for replacing La under Y. Sandbh (talk) 03:39, 21 January 2020 (UTC)
@Sandbh: Eventually Lu will replace La. That's just a matter of time needed for obsolete understandings to die. Droog Andrey (talk) 13:20, 21 January 2020 (UTC)
@Sandbh: Good luck with that. Some chemists plumbed for Lu in the 1920s, and their efforts faded into obscurity. Jensen had another go in 1982, without success. Scerri has been advocating Lu for quite a few years on regularity grounds. Other notable players in the PT field— Roald Hoffmann, Schwarz, Martyn Poliakoff, Restrepo (on the project), Philip Ball (ditto)—advocate for more than one table, depending on the intended perspective. Lavelle, also in the project, advocates La. IIRC Pekka Pyykkö advocates for a 15-column wide f-block, per unofficial IUPAC. The only other people I know that have advocated for Lu, such as Emsley, did so on the basis of Jensen's article which had several limitations. Ditto Thyssen and Binnemans. Several physicists have had a go over the years, based on generally unbalanced, one-shot, unsuccessful arguments. Sandbh (talk) 02:59, 22 January 2020 (UTC)
I'll look to strengthen my argument by drawing on the following quotes:
  • "…If scandium, yttrium, lanthanum and actinium are the only rare-earth elements, the series would have revealed the same gradual change in properties as the calcium, strontium, barium and radium series, and hence it would not have been of any special interest." (Hevesy 1929, cited in Trifonov 1970, p. 188).
@Sandbh: But why should Group 3 follow a gradual-change trend which appears exclusively for Groups 1 and 2? Other groups show secondary periodicity like B-Al-Ga-In. Droog Andrey (talk) 19:05, 8 January 2020 (UTC)
@Droog Andrey: As per my above response [S1]. Sandbh (talk) 05:38, 9 January 2020 (UTC)
  • "For each element…all three outer electrons are easily lost and the chemistry of the elements is confined to the +3 oxidation state. Their monatomic cations are colourless and diamagnetic, and have no catalytic properties. This is the behaviour that would be expected of main-group elements following the alkaline-earth metals." (Greenwood & Harrington 1973, p. 50)
@Sandbh: Not only for main group elements, but also for f-elements and for heavier Group 4 and 5 elements. What's wrong with this? Droog Andrey (talk) 19:05, 8 January 2020 (UTC)
@Droog Andrey: Nothing that I can see. That said, your response does not add anything to my [S1]. Sandbh (talk) 05:38, 9 January 2020 (UTC)
  • "The trends in properties in the family…are quite regular, and similar to the trends in Groups 1 and 2." (Lee 1996, p. 679)
@Sandbh: But the trends in Sc-Y-Lu-Lr are similar to the trends in Group 4 and onward. Droog Andrey (talk) 19:05, 8 January 2020 (UTC)
@Droog Andrey: See my [S1]. Sandbh (talk) 05:38, 9 January 2020 (UTC)
  • "…although each member of this group is the first member of a transition series, its chemistry is largely atypical of the transition elements. The variable oxidation states and the marked ability to form coordination compounds…are barely hinted at…although materials containing the metals in low oxidation states can be prepared and a limited organometallic chemistry (predominately cyclopentadienyl) has been developed." (Greenwood & Earnshaw 2002, p. 948). Sandbh (talk) 05:25, 8 January 2020 (UTC)
@Sandbh: Group 10 elements are also atypical, that's really normal for outermost groups in the block, like Groups 13 and 18 in p-block. Droog Andrey (talk) 19:05, 8 January 2020 (UTC)
@Droog Andrey: See my [S1]. Sandbh (talk) 05:38, 9 January 2020 (UTC)
@Double sharp: Further to your edit 06:47, 9 January 2020, re “Group 4 organometallic chemistry is mostly cyclopentadienyl derivatives, similar to group 3.” I found this article re the cyclopentadienyl chemistry of group 2 including d orbital involvement in Ba. Sandbh (talk) 10:33, 11 January 2020 (UTC)
@Sandbh: And that's where you can see that if you focus on organometallic chemistry, the dividing line between "pre-transition" and "transition" looks like it should be earlier than if you focus on other things. So in this case it looks like group 3 is behaving like a d-block group, and heavy group 2 is showing some tendencies in that direction too despite being d0: in that case, aren't groups 2 (Ca/Sr/Ba/probably Ra) and 3 both following the clearly transitional group 4, rather than exhibiting typical main-group behaviour?
Note that Greenwood and Earnshaw mention (p. 137) that organometallic chemistry for Ca, Sr, and Ba rather resembles the divalent lanthanides Sm, Eu, and Yb, so this is again a case of f-block being intermediate between s-block and d-block rather than a degenerate bit of the d-block. Be and Mg are more focused on alkyls in their organometallic chemistry (which looks to my inexperienced eyes more like Zn, Cd, and Hg). Double sharp (talk) 06:48, 12 January 2020 (UTC)
@Double sharp: There's nothing here I particularly disagree with. My focus is on the overall or typical behaviour of group 3; your focus (it seems to me) is on the atypical behaviour.
We know from G & E that variable oxidation states and the marked ability to form coordination compounds with a variety of ligands are barely hinted at in group 3 although materials containing the metals in low oxidation sates can be prepared and a limited organometallic chemistry has developed. For Group 4 G & E say the organometallic chemistry has developed rapidly in recent years. For group 10 they say that the organometallic chemistry is rich and varied.
Here's a 2017 PhD thesis which:
  • treats the Ln as Ce to Lu;
  • correctly notes the Ln contraction runs from Ce to Lu;
  • says the Ln are often considered to behave as trivalent versions of the +2 cations Ca2+, Mg2+ of the alkaline earth elements; and
  • notes that some of the lanthanides have very similarly sized cations to the cations of group 2 or group 3 elements, and that these cations will generally form the same compounds as the lanthanide cation that they substitute.
I think the salient point is that the typical behaviour of the Group 3 metals as per literature consensus is that they behave largely as if they were trivalent group 1 and 2 metals. It is true that the Group 3 metals share some properties with the Group 4 metals but these are the minority rather than representing majority, typical or characteristic behaviour. Another way of looking at this is at the overall chemistry, rather than a subset such as organometallic. For Sm, Eu and Yb we can turn this around and say that their OM chemistry resembles that of Ca, Sr and Ba, as would be as expected.
All of this takes us back to the premise at the top of this section. Sandbh (talk) 22:39, 15 January 2020 (UTC)
"Correctly" only if you want to be pedantic about it, I guess. Lanthanide chemistry is so impacted by differing atomic radii that I consider it frankly unilluminating pedantry to exclude La. What you are doing seems to me like basically taking only the bits of Ln chemistry that are most like main-group behaviour, declaring that subset to be the typical behaviour, and ignoring the growing parts (e.g. organometallic, which is surely not atypical) where things are not the same, whereas I'm looking at everything and saying "OK, group 3 shows similarities to both in various parts of their chemistry, just like every other border group, and so let's not create a graphic trench when there is no total separation". (In fact organometallic chemistry is the main reason why the older definition excluding Sc and Y from the transition metals is falling by the wayside.) Double sharp (talk) 09:25, 16 January 2020 (UTC)
@Double sharp: Our own article on organoscandium chemistry says, "As with the other elements in group 3 – e.g. yttrium, forming organoyttrium compounds – and the lanthanides, the dominant oxidation state for scandium in organometallic compounds is +3 (electron configuration [Ar] 3d14s2). The members of this group also have large ionic radii with vacant s,p and d orbitals (88 pm for Sc3+ compared to 67 pm for Al3+) and as a result they behave as hard Lewis acids and tend to have high coordination numbers of 9 to 12. Sandbh (talk) 06:59, 17 January 2020 (UTC)
Dominant, yes. But not the only one. Organoniobium chemistry has NbV as the most common oxidation state, but there is no problem finding lower ones. Again, we expect from the regular shape of the table that a 4f block insertion should occur before the 5d block, just like the 3d block insertion occurs before the 4p block, and so that eka-Y should have some mitigating tendencies towards softness while still being hard due to its size and charge. That points to Lu3+ as the less anomalous member (which still has pretty high coordination numbers), unless you want to have Al over Sc again. (Al3+ is quite hard due to its noble gas configuration, too!) Double sharp (talk) 17:19, 22 January 2020 (UTC)

@Double sharp: A few things here.

1. You said, “Ln chemistry is so impacted by differing atomic radii that I consider it frankly unilluminating pedantry to exclude La.”

I’ve never excluded La from Ln chemistry.

No, but you exclude it from the Ln contraction trend as the member when n = 0 in the 4fn configuration. It ought to be included in the trend line as just another lanthanide ion, because the contraction is measured by the difference in size between two lanthanides with adjacent values of n, of which n = 0 and n = 1 counts just as well. Double sharp (talk) 12:08, 23 January 2020 (UTC)
@Double sharp: I do and what you are referring to is a different concept. There is no f block f electron caused contraction seen in La. There is in Ce to Lu. Sandbh (talk) 06:15, 24 January 2020 (UTC)
It seems to me that you are making "lanthanide contraction" mean "the difference between a theoretical ionic radius without the results of poor shielding from added 4f electrons and actual ionic radius", rather than just "the contraction observed between Ln3+ ions as one proceeds left to right across the series". In which case not only is La not in the first one, but the post-lanthanides Hf through Hg are in it instead, which is clearly not what "lanthanide contraction" means for most people. Double sharp (talk) 17:21, 24 January 2020 (UTC)
P.S. May I also add that the chemical importance of the lanthanide contraction is almost a one-off. Nowhere else in the table do we get a case where fifteen elements in a row all prefer to be in the same oxidation state, so that the contraction of ionic radius is the only significant difference. If you plot Ac through Lr, you get an actinide contraction just as well, but it doesn't matter so much because the early actinides are eager to show higher oxidation states. And we come to that idea, really: the behaviour of the Ln contraction is more or less expected only for the first row of a block with such high azimuthal quantum number that the electrons in the filling shell are stuck very deeply in. This is something we see for 4f, but not even for 3d. We will probably see it for 5g, but that's about it (matter of fact, their chemistry should be more or less like heavier congeners of U, which famously has two major oxidation states of +4 and +6; so if they act like U, Np, and Pu, there should be subtler differences than with the Ln in these early superactinides). As a matter of fact, you can immediately see that Ac through Lr supports Sc-Y-Lu much better than the inconclusive La through Lu: the reason is that the trend at the end of the 5f row is for 5f to sink very deep into the core (something like what happens for the 4d row), and so nobelium becomes a logical end of the trend as we get more and more in favour of the +2 state at the end of the series. Lawrencium, back at the +3 state, appears to be an illogical appendage, and is much more understandable as the beginning of the next series (Lr +3, Rf +4, Db +5, Sg +6 should all mostly dominate). Whereas Ac is totally understandable extrapolating from Th, Pa, and U backwards: it has one less valence electron, and 5f is still above 6d (like for Th), only to cross over at Pa. Double sharp (talk) 23:38, 24 January 2020 (UTC)
@Double sharp:. There is nothing relevant here. I’d only be repeating responses I’ve given several times already. Sandbh (talk) 07:24, 25 January 2020 (UTC)
Yes, there is something relevant here. It's saying that the Ln contraction is by definition an inconclusive argument because it is something that happens over all fifteen lanthanides, not fourteen, and it is caused by a special one-off effect in the periodic table that even the actinides don't follow. It only runs over fourteen if you do some pedantic definition-arguments that usually end up declaring the knock-on effect on Lu/Hf through Hg as part of the Ln contraction as well. Double sharp (talk) 10:32, 25 January 2020 (UTC)
@Double sharp: As you know, the Ln contraction arises as a result of reduced shielding by the f electrons, which are first seen in Ce 3+ and last seen in Lu 3+, as discovered by Goldschmidt. I have nothing further to say about this aspect of the thread since it is incontrovertible. Sandbh (talk) 07:22, 26 January 2020 (UTC)
By those standards, where do the boride and scandide contractions begin? You can't say where, because there's no universal oxidation state. Double sharp (talk) 10:41, 26 January 2020 (UTC)

2. “What you are doing seems to me like basically taking only the bits of Ln chemistry that are most like main-group behaviour, declaring that subset to be the typical behaviour, and ignoring the growing parts (e.g. organometallic, which is surely not atypical)…”

Well yes, that’s what the literature says i.e. that the Ln largely behave as the equivalent of trivalent group 1 and metals. I’ve never said organometallic chemistry is atypical. I’ve only noted, per the literature, that the OMC of group 3 is ionic, like that of groups 1 and 2.

What you are doing—I don’t know why—is focussing on atypical behaviour and assigning it undue significance. Respectfully, that’s not good classification science in my view.

It's not atypical behaviour unless you hold to a standard that makes ZrIII and HfIII atypical as well. In which case we go back to the starting point: by such a standard, group 4 is mostly main-group too, and therefore there is no reason why group 3 should not follow its trend. Double sharp (talk) 12:08, 23 January 2020 (UTC)
@Double sharp: Zr3 and Hf3 are atypical. +4 is typical and covalent. +3 is typical for group 3 and ionic. That’s the dif. Group 3 does not follow the Group 4 trend in this respect. Sandbh (talk) 06:22, 24 January 2020 (UTC)
Ionic vs. covalent is not totally dependent on charge, but also on size. If we were looking at period 2 and discussing what should happen later, we would take a look at the more covalent-ish Be2+ and skip away merrily to Be-Mg-Zn. And it's not even only ionic radius: compare Ge2+ vs. Sn2+ vs. Pb2+, where the first is almost not a thing, the second is close to not being a thing, and the third is well-defined and has some significant ionic behaviour (e.g. PbF2). Chemically speaking, Zr and Hf are as electropositive as Sc, and Rf should be even more so. By this logic, we may end up excluding group 13 from the p-block, since apart from B at the top it is all more or less ionic while the credentials for group 14 are shaky right up till Pb2+ in the "wrong" oxidation state at the bottom. Double sharp (talk) 17:21, 24 January 2020 (UTC)
And I still think you are trying to have it both ways. This argument of yours works only if you consider +3 for Zr and Hf atypical. But in that case group 4 is predominantly main-group in its chemistry, not transition. And given that the difference between main group and transition is not wholly ionicity but about things like variable oxidation states (otherwise, what is Be doing?), that suggests that group 3 can follow the group 4 trend, no problem. The fact that the argument can tip both ways is exactly part of why I think group 3 ought to be compromising: more ionic like group 2, but showing a group-4-like trend. Double sharp (talk) 17:24, 24 January 2020 (UTC)

3. “Again, we expect from the regular shape of the table that a 4f block insertion should occur before the 5d block, just like the 3d block insertion occurs before the 4p block…”

That sounds like a circular argument.

It's not circular. It's saying that the regular shape of the table up to Xe leads us to conjecture that the pattern should hold further, and we can use that as a working hypothesis to be confirmed or disconfirmed by what we find with later elements. And indeed, we find that Lu acts about like how you expect for eka-Y with an f-block insertion beforehand, so there is no strong reason to reject the working hypothesis since Y-La-Lu are here acting completely analogously to Al-Sc-Ga. If Lu was not as good an eka-Y as it is, then we would be more forced to go for Sc-Y-La. But we're not.
Anyway, we can both play the game of "let's start with the format we prefer, and see that it predicts something, and we see it quite well, and the other one can be seen as a perturbation". But looking at the absolutely normal-looking "two rows at a time" pattern in periods 1 through 5, surely the null hypothesis that we are supposed to be testing ought to be that it continues in periods 6 and 7, and it should be the alternative hypothesis that it doesn't that is held to a higher standard? Double sharp (talk) 12:00, 23 January 2020 (UTC)

Noting the overall chemistry of group 3 is like that of groups 1 and 2 I’d expect eka-Y to follow the trend seen in those groups, as is the case with La. I’ve addressed the Al v Sc question before, noting the ionic radius of Al is more like that of Ga than Sc. Sandbh (talk) 08:14, 23 January 2020 (UTC)

And the ionic radius of Y is more like that of Lu than La, so where does that leave us then? Double sharp (talk) 12:00, 23 January 2020 (UTC)
@Double sharp: Y does not necessarily exhibit a chemistry similar to that of Lu (as discussed earlier). With La in group 3 we see an increasing trend in atomic radius, similar to that seen in groups 1 and 2. With Lu in group 3 we see a trend consistent with what is seen in most of the transition metals groups. Noting La has the same core as Sc and Y, whereas Lu does not, La is better placed in group 3 on similarity grounds. This includes the fact that La is the first element with a 5d electron, whereas Lu is the third. There is no plausible reason for skipping La in favour of Lu. Keeping La in group 3 serves to reflect the fact that the 4f subshell does not start filling until Ce. See also Atkins et al. Q&A. Sandbh (talk) 04:25, 9 February 2020 (UTC)
@Sandbh: But that is what it does when it behaves characteristically. On similarity grounds Lu is better placed in group 3 because in every group in the periodic table outside the s-block, the cores change every two elements. The core for Hf is different from that of Ti and Zr; the core for Ga is different from that of B and Al. The "plausible reason for skipping La in favour of Lu" is simply that your focus on ground-state configurations makes your argument groundless because those are simply not the characteristic configurations in chemical environments. When one holistically considers what orbitals are chemically active, one sees that 4f is already an active valence subshell in La (whence cubic complexes among other things), but is clearly core in Lu. The periodic table is based on chemistry, and therefore should be based on valence electrons, not core electrons. Lu and Lr in the f-block are a massive slap in the face to this. (Noble gases are not an exception; from Ar onwards they are chemically active, and anyway the energy gap between 1s resp. 2p and 2s resp. 3s is huge, so that those are still somewhat degenerately valence electrons by virtue of being the highest occupied orbitals. That can't be claimed for Lu and Lr.) Double sharp (talk) 21:57, 9 February 2020 (UTC)

Chemical behaviour of Group 3 (more)

@Double sharp:@Droog Andrew: Here are some more extracts dealing with Sc and Y, taken from Vickery RC 1960, The chemistry of yttrium and scandium, Pergamon Press, New York. A lot of this is new to me.

Sc as a light Ln/role of Ca and La

“When fluoride preparation is to be carried out, Ca is a better “carrier” than the lanthanons...precipitation of Sc as the double salt with potassium sulfate is an excellent method of concentration...Sc thus precipitated as ScK(SO4)2 will be accompanied by any light Ln present. Indeed, if the ratio of light:heavy Ln is previously known to be low, the addition of La to the solution will aid the concentration of Sc by double sulfate precipitation.” (p. 72)

Sc and Y similarity

“In spite of the large difference in Z between Sc and Y, the fact that both are two places removed from the inert configurations of Ar and Kr respectively has much to do with the similarity of their characteristics.” (p. 29)
“…Sc is diamagnetic and its magnetic behaviour presents many analogies to Y and, according to Bommer (1939) also to Pt and Pd." (p. 30)

Y and Ca

"In those instances where Y is found in minerals crystallising early from the magma it is generally present as a replacement of Ca…Because of the closer similarity between the ionic radii of Y and Ca, such replacement by Y, and the heavy Ln, is much more easily accomplished than replacement by the light Ln.
“…a factor in this replacement is the the resemblance of Y to Ca in its co-ordinating power. This suggestion does not however appear completely convincing since the smaller radius of Y will undoubtedly increase the spatial density of a given number of co-ordinating anions, it is now accepted that Y (and the heavy Ln) co-ordinates much more strongly than do the light Ln and these have a coordinating tendency approximately that of Ca.” (pp. 9–10)
“…the precipitation of Y as the alkali double sulfate from an acid solution saturated with NaCl is an effective method of concentration. Calcium will co-precipitate as sulfate...” (p. 36)

Ambiguous Y

“In separating Y from the heavy Ln, advantage is always taken of the phenomenon by which Y sometimes assumes characteristics similar to those of the light Ln, and sometimes follows the heavy Ln in behaviour.” (p. 37)
“In many respects Y resembles Zn...” (p. 38)

Monoxide spectra

“At elevated temperatures, Y gives a stable monoxide, YO, which emits a very bright band spectrum similar to that of LaO and ScO.” (p. 27) [This one is to be treated with caution since it says nothing about LuO]

I do know that La shares quite a few similarities with Ca, as do the Ln generally. Sandbh (talk) 07:00, 11 January 2020 (UTC)

@Sandbh: Coordinating power is mostly a function of charge and ionic radius, so Y being similar to Ca seems to be just a diagonal relationship. You want a bigger charge to attract more ligands, except that it makes you smaller, so you have to go down another period. So an early lanthanide will differ from a late one mostly by size: La will be worse at coordination than Y (which is about like Ca), while Lu should be about the same and even a bit better (intermediate between Ca and Sc, which acts more or less like a mini-Lu). Both give a reasonable trend, as usual, except that since this weak coordination power for La and Lu is similar to Hf and Ta as well it strikes me as not enough of a reason for cutting group 3 away from the rest of the d-block. Double sharp (talk) 07:00, 12 January 2020 (UTC)

@Double sharp: It was a quasi-knight's move relationship between La and Ca, not Y and Ca:

  • The ionic radius of Ca2+ is 114 pm; that of La3+ is 117 pm (cf. Lu 100).
  • The basicity of La203 is almost on par with CaO2 whereas Lu2O3 is the least basic of the Ln oxides.
  • Freshly prepared La2O3 added to water reacts with such vigour that it can be quenched like burnt lime (CaO)—Lu2O3 is insoluble in water.
  • The similarity in sizes means La3+ will compete with Ca2+ in the human body, and usually win on account of having a higher valence for roughly the same hydrated radius.
  • The electronegativity of Ca is 1.0; that of La is 1.1 (cf. Lu 1.27).

That swings it for me. Sandbh (talk) 06:39, 15 January 2020 (UTC)

@Sandbh: To me that's a perfect demonstration of why it should be Sc-Y-Lu: Sc and Y are equivocal between group-2-like and group-4-like properties, but as you've demonstrated La is strongly pre-transition-like while Lu is more equivocal, having both pre-transition-like (its lanthanide properties) and transition-like properties (it's the softest lanthanide cation, etc.). Since we are sure that Sc and Y are supposed to be d-block elements, and Lu's behaviour is more like Hf and Ta which are clearly early d-block elements, Lu is the stronger analogue to Sc and Y. This behaviour of La is on the other hand pretty much strongly pre-transition, like most of the early 4f-elements, without the moderating effects that transitionise the late ones due to smaller size. (And the things about La that are not quite pre-transition, e.g. extremely high coordination numbers, are often like the indisputable "direct" f-block elements, e.g. uranium.) Double sharp (talk) 06:50, 15 January 2020 (UTC)

@Double sharp: Sc and Y may be equivocal between group 2-like and group-4 like properties if one focuses on non-characteristic properties. Stepping back and considering their overall properties, including with either La or Lu, I submit that group 3 is more like groups 1 and 2 than group 3, as noted in the literature on several occasions. Sandbh (talk) 00:28, 16 January 2020 (UTC)

@Double sharp: Writing in Comprehensive Inorganic Chemistry, vol. 4, Moeller (1973) makes the follow observations:

"The existence of a common +3 state of oxidation throughout the series does require in most instances that an f electron be removed in its formation. However, the 4f orbitals are sufficiently removed from the valency shell and sufficiently shielded by external shells as to be largely unavailable in other chemical reactions. Herein lies a significant difference from the d-transition species, the atoms and ions of which are characterized by d electrons in their valency shells. There is, therefore, a closer configurational similarity between the lanthanide ions and the Group Ia-IIIa cations than between the lanthanide ions and the d-transition metal ions. The presence of shielded 4f electrons in the lanthanide ions does not materially alter the noble-gas core that they present to incoming chemical groups. The exclusion of lanthanum from the lanthanide group is based solely upon the absence of 4f electrons." (p. 3) [An extract of this passage was included in our IUPAC submission]

"The net result is that in comparison with the d-transition metal ions the lanthanide ions as a whole both form far fewer complexes and yield complexes with significantly different properties. Indeed, there are often better comparisons between the lanthanide complexes and those of the Group IIA cations." (p. 27)

Sandbh (talk) 03:23, 16 January 2020 (UTC)

@Double sharp: I missed this one, which is in vol. 3, by Vickery, on Sc, Y, and La:

"Polymerization of the Y ion has been shown now to account for its apparently nomadic behaviour in earlier classical separation techniques. Evidence is also available for the existence of La hydroxypolymers in solution. There is, indeed, to be seen an interesting sequence through…Group III in this respect. Hydroxyl bridged polymerization has been shown for Al, Sc, Y, and La ions, but does not appear to exist with the series Ce3+ —> Lu3+. OTOH, Ga, In and Tl do appear to complex in this fashion. On a thermodynamic basis, ionic hydration—or hydroxo complex formation—may depend upon free energy rather than enthalpy and plots of such free energy link the pre-lanthanon triad more closely to Al, on the one hand, and Ga, etc., on the other, than to the lanthanon group of elements.

(p. 344)

Sandbh (talk) 06:59, 16 January 2020 (UTC)


I'm listing a few refs here instead of repeatedly looking them up.

  • Group 3 (Sc, Y, La, and Ac): "As shown in Table 23.8.1, the observed trends in the properties of the group 3 elements are similar to those of groups 1 and 2." (2019)
From (CLT) here

  • Group 4 (Ti, Zr, Hf): 1. "The +4 oxidation state for titanium gives rise to largely covalent compounds." 2. "…Zr(IV) and Hf(IV) are mainly covalent…"
-- 1. Nicholls D 1974, Complexes and first-row transition elements, MacMillan Press, London, p. 139
-- 2. Talbot DEJ & Talbot JDR 2018, Corrosion science and technology, 3rd ed., CRC Press, p. 336

  • Group 5 (V, Nb, and Ta): "The chemistry of the two heaviest group 5 metals (Nb and Ta) is dominated by the +5 oxidation state. The chemistry of the lightest element (V) is dominated by lower oxidation states, especially +4."
-- Same CLT source

  • Group 6 (Cr, Mo, and W): "Due to extensive polarization of the anions, compounds in the +6 oxidation state are highly covalent…For Mo and W, the highest oxidation state (+6) is by far the most important…"
-- Same CLT source

  • Group 7 (Mn, Tc, Re): "Of the two, technetium more closely resembles rhenium, particularly in its chemical inertness and tendency to form covalent bonds."
-- from our article on technetium

  • Groups 8−10 (PGM only): "In a number of cases (e.g. P, S, As, Se, the platinum group metals) the oxidation states are formal because the bonding exhibited by these elements is commonly covalent in minerals."
-- Henderson P 1982, Inorganic geochemistry, Pergamon Press, p. 135

  • Group 11 (Cu, Ag, Au): 1. "It [Ag] tends to bond covalently in most of its compounds." 2. "Gold has common oxidation states of +1 and +3, and as a soft acid usually combine with soft bases in predominately covalent systems…"
-- 1. From our PTM article
-- 2. Porterfield WW 2013, Inorganic chemistry, 2nd ed., Academic Press, p. 528

  • Group 12 (Zn, Cd, Hg): "All of the [Group 12] metals, but especially mercury, tend to form covalent rather than ionic compounds."
-- From our PTM article


  • These sources cover 24 of the 30 (80%) period 4 to 6 transition metals or 21 of the 27 (77.8%) TMs if Group 12 are left out.


  • The group 3 TMs have a predominately ionic chemistry.
  • For Groups 4 to 11, encompassing twenty-four period 4 to 6 TMs, 18 of these (75%) have typically covalent chemistries.
  • Group 12 has a predominately covalent chemistry.

Sandbh (talk) 05:04, 30 January 2020 (UTC)

Notice how:
  • Group 3 is declared to have a trend resembling groups 1 and 2, after it has already been accepted as Sc-Y-La-Ac, which is circular.
  • As R8R mentioned, it is not fair to compare on one side to groups 1 and 2 (two groups), and on the other side to groups 4 to 12 (nine groups). So let us be fair, and have two on each side. Actually, let me be a bit unbalanced and have three on the "other" side, because it is amazing that I can still do it. So: not only Zr/Hf, but also Nb/Ta and Mo/W have their chemistry dominated by the group oxidation state. In which, of course, transition-metal properties are impossible as there are no d-electrons. This is surely the significant difference here, not "ionic vs. covalent". Most of the main-group elements are not "predominantly ionic" by any stretch of the imagination. Beryllium and magnesium certainly are not for any distribution of anions.
  • Even if "ionic vs. covalent" were relevant, it would be:
    1. ill-defined without a distribution of anions stated; and even if one was stated, it would be:
    2. advocating for Be and Mg over Zn.
Double sharp (talk) 13:52, 30 January 2020 (UTC)

Group 3 and the Ln

Showing group 3 split from groups 4 to 12 is consistent with Moeller (1973, p. 3) who observed that: “There is…a closer configurational similarity between the lanthanide ions and the Group Ia–IIIa cations than between the lanthanide ions and the d-transition metal ions. "

Here Moeller (and King after that) say nothing about group 3, but they say that lanthanides are closer to s-block than to d-block. And that's the reason to put f-block immediately after s-block instead of inserting it inside d-block. Droog Andrey (talk) 11:19, 7 January 2020 (UTC)

Sandbh: Moeller refers to the cations of groups Ia, IIa, and IIIa (Sc-Y-La-Ac), so that seems like a good reason to show groups 1 to 3 as being colocated.
@Sandbh: They are colocated in 18-column version of the PT. Droog Andrey (talk) 19:05, 8 January 2020 (UTC)
@Droog Andrey: [S2] I was referring to group 3 as Sc-Y-La-Ac, in the 18-column form. Sandbh (talk) 05:38, 9 January 2020 (UTC)
For King, the context is:
Group 3 shows chemical behaviour that is manifestly uncharacteristic of the transition metals proper. Group 3 does not show the complex coordination chemistry that is characteristic of transition metals; they do not show multiple oxidation states; and they are more reactive and electropositive than any other transition metals, approaching the s-block metals in both properties. In fact, they largely show the behaviour expected of main-group elements following the alkaline-earth metals. This is true for the lanthanide series from La to Lu, as well as in Ac and the late actinides from Cm to Lr (Greenwood & Harrington 1973, p. 50; King 1995, p. 289) Sandbh (talk) 05:25, 8 January 2020 (UTC)
@Sandbh: That behaviour is normal not only for main group elements, see above. Droog Andrey (talk) 19:05, 8 January 2020 (UTC)
@Droog Andrey: See my [S1][S2]. Sandbh (talk) 05:38, 9 January 2020 (UTC)

Basis for blocks

Differentiating electrons are relevant from a chemical perspective since they enable the periodic table to be parsed into four major blocks according to the predominant differentiating electron in each block

In fact, the table is parsed into blocks according to chemically active subshells in atoms. Differences in ground-state configurations have little sense, especially in d- and f-blocks where a lot of low-lying excited states are observed for the vast majority of the elements. Droog Andrey (talk) 11:19, 7 January 2020 (UTC)

Sandbh: There is a supporting citation for what I said here: Stewart PJ 2018a.
@Sandbh: Citation is not an ultimate truth but just another opinion. Droog Andrey (talk) 19:05, 8 January 2020 (UTC)
@Droog Andrey: [S3] Yes, I agree. There's no categorical argument for the resolving the group 3 question, so I have to rely on qualitative and quantitative arguments, supported by citations. Stewart---described by Scerri as a polymath---is a key player in this space, with a formidable reputation (he supports Sc-Y-Lu-Lr). Sandbh (talk) 05:38, 9 January 2020 (UTC)
Your approach won't work for f-block lanthanides, as the f electrons aren't chemically active. It doesn't work that well for group 12 either, where the d electrons aren't comparably chemically active. Even silver has much more main group chemistry than d- chemistry. Sandbh (talk) 05:25, 8 January 2020 (UTC)
@Sandbh: Wow. Just try to calculate some properties of f-elements without explicit f-electrons accounting (e.g. replacing them with an ECP) - the results will be garbage. However, for Lu and Lr such calculations give decentish results, at least better that for Zn-Cd with d-electrons replaced with ECP. Droog Andrey (talk) 19:05, 8 January 2020 (UTC)
@Droog Andrey: [S4] I accept that, and here I feel we are getting into inconclusive micro-argument territory that Scerri finds inconclusive. Stepping back from the dance floor and up into the balcony, the f electrons aren't chemically active. Sandbh (talk) 05:38, 9 January 2020 (UTC)
@Sandbh: Surely they are. Just try to explain coordination numbers around 12. Droog Andrey (talk) 06:59, 9 January 2020 (UTC)
@Sandbh: I don't accept that this is a micro-argument, because a very common theme in lanthanide chemistry is the interplay between 4fn and 4fn+1 configurations. 4f is maybe not directly active in quite the way 6s for example is, obviously, but it is a reserve area where one electron can easily be kicked up (or not) to 5d, most obviously in Eu2+ vs. Eu3+. On those grounds alone La is already mathematically a better fit than Lu, precisely because if one generalises this to the whole f-block, one quickly realises that this interplay is mathematically impossible for Lu (you need to cram 15 electrons into 4f). For La, it is at least possible, even if we do not see it that much. Now just put in the actinides, which everyone accepts go under the lanthanides; the early ones show lots of f-orbital involvement, and in the late actinides we see this 5fn vs 5fn+1 interplay again. Now, which one looks closer, Ac or Lr? Seems to me that if you put Lr at the end of the f-block you've got a strange caesura between No and it, because +2 has been slowly asserting its dominance from Es to No with one clear reason: 5f is dropping into the core. Then Lr suddenly gets +3 dominating, for a completely different reason: it's no longer fishing out a third electron from 5f. Seems like Ac is a clear winner for this reason, as the trend from it to Th is pretty smooth: in the early actinides, 5f, 6d, and 7s are all happily hybridising, with 6d on top at first and 5f slowly getting more and more involved, and the trend may be continuously extrapolated backward to Ac from the early actinides in a way that it cannot be continuously extrapolated forward to Lr for the late ones. Double sharp (talk) 06:55, 9 January 2020 (UTC)
@Double sharp: @Droog Andrey: Let's not lose sight of what I said in the paper which was about differentiating electrons being used to parse the table into blocks which I recall goes back to the 1920s. I provided a citation about the general principle being supported by Stewart. As far as La and Lu are concerned, in my humble opinion, you're going back to the micro arguments. The consensus in the literature, by a wide margin is that in the very large majority of cases, the f electrons do not take part in ordinary chemistry reactions. There is no point in saying that f orbital participation is a possibility for La but not for Lu. Again, the incidence of this being able to happen for La is like comparing the moon to the sun, and saying therefore that La is a better candidate for the f-block, while simultaneously down playing the much more important global issues of the kind I've tried to set out in the rest of the article. It's these global issues that will swing the argument, not the minutiae of the interplay between 4fn and 4fn+1 configurations. Sandbh (talk) 09:32, 9 January 2020 (UTC)
@Sandbh: If 4f is not involved, then how can the early Ln support such high coordination numbers for their compounds? And how come the Ln all like to be in the +3 state? They are all 4fn6s2 in the gas phase; where did that third valence electron come from? And what about the actinides, for which 5f is an obvious major part of chemistry? And what about those late 3d metals that prefer being in the +2 state like Co and Ni: are they somehow less d-elements because of it? Is Mn less of a d-element because its differentiating electron is 4s rather than 3d? And how is the differentiating electron a global issue when the anomalies like Nb vs. Ta or Lu vs. Lr seems to mean absolutely nothing for the chemistry of those elements? For that reason I think differentiating electrons are far more like a micro argument and which subshells are active is actually the global issue. Double sharp (talk) 11:45, 9 January 2020 (UTC)
@Sandbh: You say differentiating electrons are the basis of block structure, I argue that differentiating electrons has no chemical sense for the most of d- and f-elements, and that the overall block structure of PT really comes from the influence of valence subshells symmetry on the chemistry of elements. You start arguing about f-subshell activity, but then retract that plotline. So what? :) Droog Andrey (talk) 12:26, 9 January 2020 (UTC)
@Double sharp:@Droog Andrew: RL obligations mean I won’t necessarily be able to provide a fully considered reply until next week. On the preference by the Ln for trivalency this is because in the condensed form most of them have a 4fnds2 configuration. The differentiating electron argument is not mine. The argument about the relevance of differentiating electrons to the overall structure of the PT, as a global argument, is supported by e.g. Janet, Stewart, Scerri, Lavelle, and others, and quantum mechanics. F-involvement in the early actinides are not relevant to the article. Actually, that is a good point. The Ln, as far as I know have no comparable f-involvement. It is a sun to moon situation. On high coordination numbers for Ln compounds I remember reading something about that and the possibility of f orbital/electron involvement but there was nothing conclusive that I can remember. Our own article on the Ln says that, “The 4f orbitals penetrate the [Xe] core and are isolated, and thus they do not participate in bonding.” Sandbh (talk) 07:15, 10 January 2020 (UTC)
@Sandbh: What about Ce4+ compounds? (Yes, they are sort-of mixed-valence according to our cerium article, but that implies at least partial direct 4f involvement in the bonding.) And however you put it, it remains that once you get into the condensed phase one electron that was in 4f in the gas phase has moved to 5d, so there is still some promoting from 4f to 5d going on here. Double sharp (talk) 06:39, 12 January 2020 (UTC)
@Double sharp: Yes, I agree about Ce. Our article should perhaps say, with (very?) few exceptions, the 4f orbitals don’t participate. Mind you I’ve never found a source that clearly addresses this question. Sandbh (talk) 09:00, 12 January 2020 (UTC)
@Sandbh: I recommend this article on the matter. Double sharp (talk) 03:59, 13 January 2020 (UTC)
@Double sharp: I remember reading that odd paper when we were drafting our IUPAC submission. I was never able to find any confirmation of its explanation in the literature. I recall thinking how odd it was basing its arguments on the gas phase and not saying anything about the condensed phase configurations e.g. Greenwood and Earnshaw (2002, pp. 1232, 1234): "…most of the metals are composed of a lattice of LnIII ions with a 4fn configuration and 3 electrons in the 5d/6s conduction band [i.e. 5d16s2]. Metallic Eu and Yb, however, are composed predominately of the larger LnII ions with 4fn+1 configurations and only 2 electrons in the conduction band." Sandbh (talk) 22:56, 15 January 2020 (UTC)
@Sandbh: If we go by condensed-phase configurations, Be and Mg are p-metals, and Al goes over Sc because of p-orbital occupancy in the d-metals. Double sharp (talk) 09:31, 16 January 2020 (UTC)
@Double sharp: That's news to me about Be and Mg being p-metals and p-orbital occupancy in the d-metals.
We know that ionic radii strongly influence the chemical properties of the metallic elements.
As our own article on scandium says:

"Sc chemistry is almost completely dominated by the trivalent ion, Sc3+. The radii of M3+ ions in the table below indicate that the chemical properties of scandium ions have more in common with yttrium ions than with aluminium ions. In part because of this similarity, scandium is often classified as a lanthanide-like element.

Ionic radii (pm)
Al Sc Y La Lu
53.5 74.5 90.0 103.2 86.1
Our ionic radius article gives the following figures
Al 67.5 | Ga 76.0 | In 94
Al 67.5 | Sc 88.5 | Y 104
Sandbh (talk) 00:10, 18 January 2020 (UTC)
@Double sharp: Here are those passages about coordination numbers and suggested f orbital involvement (how does Lu manage 11 coordination?):
"Coordination Numbers 10–12
Sheer congestion or donor atoms around the metal ion and concomitant inter-donor atom repulsions makes these high coordination numbers difficult to attain. They are often associated with multidentate ligands with a small 'bite angle' such as nitrate that take up little space in the coordination sphere, either alone, as in (Ph4As)2[Eu(NO3)5] or in combination with other ligands, as in Ln(bipy)2(NO3)3, Ln(terpy)(NO3)3(H2O) (Ln = Ce-Ho), and crown ether complexes (Section 4.3.7) such as Ln(12-crown-4)(NO3)3 (Ln = Nd-Lu). Other crown ether complexes can have 11 and 12 coordination, e.g. Eu(15-crown-5)(NO3)3 (Ln = Nd-Lu) and Ln(18-crown-6)(NO3)3 (Ln = La, Nd )."
Cotton S 2006, Lanthanide and actinide chemistry, John Wiley & Sons, Chichester, p. 53
"The one case in which contributions to the bonding from the f orbitals is possible is in complexes of the heavier elements in which the coordination number is high. Use of the s orbital, together with all the p and d orbitals or one valency shell, permits a maximum coordination number of nine in a covalent species. Thus, higher coordination numbers imply either bond orders less than unity or else use of the f orbitals In addition, certain shapes (such as a regular cube) or lower coordination number also demand use or f orbitals on symmetry grounds. These higher coordination numbers have only become clearly established recently, but their occurrence in lanthanide or actinide element complexes suggest the possibility of f orbital participation. Examples include the 10-coordinate complexes mentioned above, LaEDTA(H2O)4 and Ce(NO3)52- or 10-coordinate La2(CO3]3.8H2O; 11-coordinate Th(NO3)4.5H2O (coordination by four bidentate nitrate groups and three of the water molecules); and the 12-coordinate lanthanum atoms in La2(SO4)3.9H2O-with twelve sulfate O atoms around one type of La atom position."
MacKay KM, MacKay RA &Henderson W 2002, Introduction to modern inorganic chemistry, 6th ed., Nelson Thornes, Cheltenham. p. 256

Periodic law

The periodic law implies, certius paribus, that since La represents the first recurrence of comparable periodicity after Y it should be the one to go under Y rather than Lu. If that had been true, Sc would have gone below Al. Droog Andrey (talk) 11:19, 7 January 2020 (UTC)

Sandbh: This is a very interesting item that I may need to elaborate carefully in the article. For now, I'd say that Sc and Al breaches certius paribus, since Sc is a d element whereas Al is a p element. I know that Mendeleev, in his Principles of chemistry (Ch. XVII, Boron, aluminium, and the analogous metals of the third group) said that the analogues of aluminium were gallium, indium, and thallium. Sandbh (talk) 07:00, 8 January 2020 (UTC)
@Sandbh: As soon as we are accounting d-elementicity, let's note that Lu is much more d-elementish that La, the latter being f-element. Droog Andrey (talk) 19:05, 8 January 2020 (UTC)
@Droog Andrey: [S5] That may be so, peripherally. Unfortunately the macro-effects such as e.g. the binary stoichometry of rare earth compounds, support La under Y. I've seen proposals to place both La and Lu, as d elements, in group 3. Sandbh (talk) 05:38, 9 January 2020 (UTC)
@Sandbh: That's the same for Al-Sc vs. Al-Ga. Droog Andrey (talk) 06:59, 9 January 2020 (UTC)
@Droog Andrey: Not from a binary stoichiometry pov. Al goes over Ga. Sandbh (talk) 09:37, 9 January 2020 (UTC)
@Sandbh: What exactly do you mean by binary stoichiometry? If you mean just binary compounds of each element, e.g. Al2O3 vs Sc2O3 vs Ga2O3, there is no significant difference as everyone is trivalent. If you mean something like which elements substitute for each other preferentially in things like mixed oxides, then like Droog Andrey mentions Al patterns with Sc, not Ga. Double sharp (talk) 11:35, 9 January 2020 (UTC)
@Double sharp: On stoichiometry, see here.
@Sandbh: I've already addressed the methodological problem with using this study exclusively on your talk page, so I'll just quote myself again: "...Restrepo (as I noted before) is concerned about stoichiometry, which implies valence, and thus it is obvious that similarities with elements from different groups like Lu vs. Hf will be ignored. But secondly, the procedure is not to only look at which of La and Lu are more similar to the other lanthanides: obviously this is going to be totally inconclusive because of how similar the lanthanides are to each other. (Note that Sc, Y, and the Ln in Restrepo's article are all in a cyan area of very low difference from each other anyway, so the magnitude of the differences we are talking about is tiny. Double sharp (talk) 10:04, 12 January 2020 (UTC)) We need to ask what we did for the analysis of Zn, Cd, and Hg above (where I was comparing them to transition metals, and noting that the distance between Zn/Cd/Hg and the TMs is less than that between Ca/Sr/Ba and the TMs) Double sharp (talk) 10:04, 12 January 2020 (UTC), i.e.: what are the characteristic properties of the d-block elements, specifically the nine sure 5d elements from hafnium to mercury inclusive, and does La or Lu display more of those properties? In every conclusive case the answer is Lu.
  1. The 5d metals are usually small cations due to the lanthanide contraction; so Lu3+ obviously fits better than the La3+ as it comes when the lanthanide contraction is at its end. (In fact Lu3+ has a similar ionic radius to Au3+.)
  2. The early 4f metals often show rather large coordination numbers uncharacteristic of the d-block metals, because the latter are not large enough. (Also because of 4f involvement.) Double sharp (talk) 10:04, 12 January 2020 (UTC) Again, Lu (being the smallest lanthanide) is the closest fit here.
  3. The 5d metals are more covalent than ionic in most of their chemistry, and their oxides and hydroxides mostly acidic or at least amphoteric: as expected from Fajans' rules, Lu more closely approximates this behaviour than La, as Lu(OH)3 is (barely) amphoteric.
  4. Most of the sure 5d metals are intermediate or soft cations (Hf being the only exception); clearly Lu is closer, as it is the smallest and hence softest lanthanide cation.
  5. The 5d metals are very dense, and the early ones are very hard and often refractory. They are also mostly unreactive in the bulk form. Lutetium, being the hardest, densest, most refractory and least reactive of the lanthanides, is obviously closer to this kind of behaviour than lanthanum, a soft metal which corrodes in the air.
This is more or less what settles it for me: even though sometimes a Sc-Y-La trend looks better, sometimes a Sc-Y-Lu trend looks better, and both La and Lu are full of similarities with the other thirteen lanthanides, it is hands-down obvious that lutetium is a better fit with the other nine 5d metals than lanthanum is. We can for sure put an extra 3 above the La-Ac small column to alert people that the Sc-Y-La trend is also worth looking at, but I submit that the case for Lu under Y is clear-cut because it leads to a more homogeneous d-block. The issue of which elements should be in group 3 should be decided by the prospective group 3 elements and hence can remain unresolved and probably unresolvable, but the issue of which elements should go in the d-block should be decided by looking at everything in the d-block, and that provides an answer. So for me, yes, La and Ac can also be extra group 3 elements just as Lu and Lr are, but the d-block positions below Y in the table must be occupied by Lu and Lr. (Obviously we have to extrapolate a lot to do this analysis for Lr, but the actinide contraction by itself ensures that the points I wrote above for Lu are almost certainly valid for Lr as well when you compare whether actinium or lawrencium acts more like the sure 6d metals from rutherfordium to copernicium inclusive.) Double sharp (talk) 15:25, 11 August 2019 (UTC)" Double sharp (talk) 10:01, 12 January 2020 (UTC)
@Double sharp: The homogeneity (or not) of the d block is interesting. The 3d and 4d rows are about as uniform as we could expect given the more or less steady addition of d electrons. Row 5d post-La disrupts the expected periodic trends going down the d groups due to the interpositioning of the lanthanides. So La is about what we would expect for the metal going under Sc-Y. Whereas Hf onwards buck the trend.
Similarity is one of the key concepts of the periodic table, which was historically addressed by assessing the resemblance of chemical elements through that of their compounds. Mendeleev’s studies were mainly based on compounds, particularly on oxides, hydroxides, hydrides and halides and by studying their similarities he came up with resemblances for the chemical elements. He highlighted the need to rely on compounds and their proportions of combination rather than on properties of chemical elements; this is evident in his statement that “if CO2 and SO2 are two gases which closely resemble each other both in their physical and chemical properties, the reason of this must be looked for not in an analogy of sulphur and carbon, but in that identity of the type of combination, RX4, which both oxides assume.” He adds, “the elements, which are most chemically analogous, are characterized by the fact of their giving compounds of similar form RXn.” Following Mendeleev’s ideas, a key concept to understand similarities of chemical elements is that of valency, which is obtained by stoichiometric decomposition of compounds in chemical analysis.
According to Restrepo’s results, La appears in between two clusters, one of 11 lanthanoids and another of transition metals, namely {Y,Sc}. Lu is part of the clusters of 11 lanthanoids and the smallest cluster containing it is {Ho,Er,Lu}, which shows that Lu is more similar to lanthanoids than to transition metals, while La share similarities with lanthanoids and with transition metals. Therefore La must be the element located at the beginning of the third row of transition metals if chemical resemblances are what it is to be emphasized.
It seems to me that you are emphasising the horizontal fit of Lu with the 5d metals. It seems to me that the vertical fit of La in group 3 is more important, since vertical trends across the PT generally form the basis of periodicity.
The properties of Lu that you listed are what I’d expect for a metal just before Hf. At the same time Lu is universally, in chemical terms, regarded as a lanthanide, there being no widely recognised chemical series encompassing the 5d metals. And the Ln were conceived of as the metals following, but not including La i.e. La retained its 5d status. Let us not forget that Lu usually occurs with the heavy Ln, that it resembles closely erbium and holmium, except that it melts at a slightly higher temperature and is essentially non-magnetic, and that the details of producing, purifying and fabricating it are almost identical with those of holmium. Sandbh (talk) 23:14, 12 January 2020 (UTC)
@Sandbh: But we have to look at those horizontal properties, because the vertical trends for Sc-Y-La and Sc-Y-Lu both look pretty convincing and there is little to choose just from that alone. (Which is why I still think that if you're writing a general inorganic chemistry textbook, you should cover both the -La-Ac and the -Lu-Lr option in the same chapter and create a six-element "group 3", because that sheds light on some things. But then again I also think you should cover Be-Mg both with the Ca group and the Zn group.) Notice that in Restrepo's chart, the REM are all in a region of cyan background, which is a region of very low difference in general. Therefore the differences he is pointing out cannot possibly be very big (as everyone expected, since La and Lu are basically almost always trivalent, just like Sc and Y), so this feels to me like saying "let's just decide it by piling up similarities to Sc/Y on both sides", and my intuition revolts because almost by definition just looking at properties of our candidate group 3 elements alone will be inconclusive (otherwise why has this argument been raging on for so long?). And where are we if we say that actually Sc and Y are pre-transition? Then there's little to choose again; if we take La just because it appears closer to them than Lu in his chart, we're again overlooking just how small the differences are. (And I strongly suspect it might be just some statistical fluctuations in the data at this level, because if anything Lu should better substitute for Sc and Y in everything due to its smaller size.) I prefer to say: if we want to talk about "the element located at the beginning of the third row of transition metals" based on "chemical resemblances", then it's obvious that the second most relevant resemblances are to the rest of the "third row of transition metals" (and that's the desired chemical name for the 5d series) that are being talked about; and we appeal to this second level because the first level is totally inconclusive. And of course, a stoichiometry analysis will never help with that, just as it will never find diagonal relationships like Li-Mg and Be-Al. And it doesn't matter that Lu also acts like an excellent heavy lanthanide, because La also acts like an excellent light lanthanide that usually occurs with Pr and Nd. And it doesn't matter that historically the Ln were considered as the elements following La, because we can always change definitions when something else seems to get closer to the heart of the issue, which is why the transition metals are no longer just Mendeleev's "group VIII". Double sharp (talk) 03:46, 13 January 2020 (UTC)
@Double sharp: I feel you've overlooked the original premise which was the periodic law and what was the case for skipping La in favour of Lu. By itself that is a huge step considering the periodic law forms the basis of the periodic table. The other thing is, as you say, piling up arguments on the basis of a comparison of the properties of La and Lu is inconclusive. That is why I'm relying on global or philosophical arguments. And let us not forget, as mentioned, that it is vertical relationships that are more important in the case of groups, than horizontal trends. I don't accept your premise that the vertical trends are inconclusive. The argument has been going on for so long because nobody has taken a holistic view of the trends going down the groups to either side of Group 3, and the overall chemical behaviour of Group 3, as we did in our IUPAC submission. This includes the multiple quotes from the literature noting the overall behaviour of Group 3, and the Ln, as trivalent versions of Groups 1 and 2. You can argue about the fine details but the overall behaviour is still in favour of La. There are no showstoppers here, only matters of nuance and detail which, at the end of the day, are overshadowed by the broader premise of the period law and the lack of a sufficiently convincing justification for overturning the periodic law, in this instance (IMO). Sandbh (talk) 05:23, 15 January 2020 (UTC)
@Sandbh: Because I don't accept that as a statement of the periodic law. One does not look for the first element that establishes a comparable recurrence of properties, or else Sc goes below Al. One looks for the first one that has a comparable recurrence of properties and matches in valence structure. That allows Ga to go below Al, and also demands that Lu go below Y. Double sharp (talk) 06:38, 15 January 2020 (UTC)
@Double sharp: I agree, we don't place Sc under Al due to the mismatch in configurations, Sc being a d metal and Al being a p metal. The next p metal is Ga so it goes under Al. We place La under Y since it's the next comparable d metal. Lu doesn't go under anything, since it's at the end of the f-block. Sandbh (talk) 00:11, 16 January 2020 (UTC)
@Sandbh: Again: La is not a d metal. It has an empty 4f not so far above, whereas Lu does not. Lu doesn't even have the mathematical possibility of breaching the 4f shell, but La does, and the position of the 4fn+1 excited state can have an effect (as demonstrated in the archives for Ln oxides). Lu is qualitatively different in this case because there's no 4fn+1 state, not even a very high-energy one. Double sharp (talk) 09:34, 16 January 2020 (UTC)
@Sandbh: So Ba is not an s-metal? Sandbh (talk) 05:59, 17 January 2020 (UTC)
@Sandbh: Stricto sensu it is not. Neither are Ca, Sr, and Ra for that matter, so the periodic law is still not broken. (Neither is it from Mg to Ca as there is simply no completely s-metal analogue, so we pick the earliest close one, Ca rather than Zn.) Their placement in the s-block comes from a few reasons:
  1. Be and Mg have some d-metal-like characteristics due to their small size even if they have no d-involvement possibility. (Mg even shows low-valent complexes that are similar to those ones that we see for Ca and group 3.) Therefore the distance is already smaller.
  2. Zn, Cd, and Hg are d-metals too, and the d-block cannot fit 11 columns, so just like for La vs Lu we must extract the triad that is most like the s-metals (which are missing one group). Ca, Sr, and Ba act much like the alkali metals, so they fit better than Zn, Cd, and Hg.
  3. Doing it this way results in a trend that matches group 1, which has to show a straight trend; doing it the other way means that Ca-Sr-Ba is a d-block group that doesn't show an f-block insertion at period 6, which is not the majority trend. (You can't put Yb under Sr for obvious reasons.)
  4. If you make Be-Mg-Zn the other s-block group, it leaves the impression that one s electron goes in first, and then the other one waits for ages, which is a bad first approximation.
If you do the same arguments for La vs Lu you get Lu:
  1. Sc and Y have some f-metal-like characteristics due to similar size (Y masquerades as a rare earth). So the distance is small whether you pick La or Lu.
  2. Even if we accept 4f involvement in Lu, it seems to me that we still must extract out the one that fits better with the d-metals which are missing one group, and that's Lu.
  3. This way leads to a trend matching the other d-block groups; the other way gives an irregularity.
  4. Same thing happens again; it's not really true that one d-electron goes in first and then 14 f-electrons interrupt, due to all the 4fn vs. 4fn+1 interplay.
And that's why, while I'm happy to say that La and Ac are extra group 3 elements in the same way that Zn, Cd, and Hg are extra main-group elements, and prefer a branched group II and a branched group 3 to illuminate what's going on here, I think that the "primary" members shown as vertical columns should be Be-Mg-Ca and Sc-Y-Lu. (Matter of fact, I could get behind a table with H and He duplicated on the top of Li/Be and F/Ne, Be/Mg duplicated above Ca and Zn, and B/Al/Sc/Y duplicated above La with Sc and Y only above Lu as well.) Double sharp (talk) 07:32, 17 January 2020 (UTC)
@Sandbh: How exactly is stoichiometry going to make any difference at all when all the elements involved are quite happily stuck in the +3 state almost all the time? I am confident that if you compare Lu to Y you will also see zero significant difference. As you will if you compare La to Y. Or Gd to Y. Or Ho to Y. (Struck this bit, as I misunderstood what you meant.) The significant difference between La and Lu is the size, with an impact on coordination chemistry, that makes Lu display more transition-like behaviour than La does. And which way does Sc swing? More like a transition metal, a bit like an overly small lanthanide. And which way does Y swing? It pretends to be a late lanthanide, which is what Lu also is. So if we speak of similarity to d-elements, it seems that the ways Lu displays this are anything but peripheral, but dominate its chemistry. Double sharp (talk) 06:55, 9 January 2020 (UTC)
@Double sharp: I guess Sandbh means that in mixed oxides La has more affinity to Y than Lu. But the same situation is for Sc, which has more affinity to Al than Ga, but it not placed below Al. Droog Andrey (talk) 07:12, 9 January 2020 (UTC)
@Droog Andrey: OK, thanks for the explanation. I've struck the first three sentences in my reply, since it's not relevant to what he means. Double sharp (talk) 07:15, 9 January 2020 (UTC)
@Droog Andrey: @Double sharp:. Can I get rid of Y first, and then see what's left.

I was under the impression that the chemistry of Y was similar to Lu, given their comparable sizes and that Y and Lu occur in the so called yttrium group.

It turns out that things are not so straightforward. Y can behave like a cerium earth (e.g. Pr, Nd, Sm) or a yttrium earth (e.g. Dy, Tm, Lu).

Here are some extracts from the literature.

[1] Bünzli J & McGill I 2000, "Rare-earth elements" in B Elvers (ed.) 2011, Ullmann’s Encyclopaedia of Industrial Chemistry, 7th ed.

In separating the rare earths via ion exchange, the behaviour of yttrium varies with the chelating agent over a range from Pr to Dy (p. 19).

Along similar lines:

"The separation of yttrium oxide [from europium oxide] exploits the fact that yttrium is unique among the rare earth elements in that its position in the series of elements with respect to separation operations is not constant...The last cycle extracts Tm, Yb, and Lu by means of tricaprylmethylammonium thiocyanate. Here, yttrium behaves like a cerium earth element." (p. 26)

[2] Jowsey J, Rowland RE & Marshall JH 1958, "The comparative deposition of yttrium, cerium, and thallium in bone tissue of dogs", in Argonne National Laboratory, Radiological Physics Division Semiannual Report, July to December 1957, Illinois, pp. 63--75

"Yttrium behaves chemically and metabolically in a way similar to cerium and thulium and may be classified with these two elements in the lanthanon rare-earth series." (p. 64)

[3] Marsh JK 1947, "The relation of yttrium to the lanthanons: A study of molecular volumes", Journal of the Chemical Society, pp. 1084--1086

"...yttrium and holmium ions are approximately of the same size, but sometimes yttrium shows behaviour indicating a resemblance to neodymium and samarium, which elements have larger ions than holmium. Certain sparingly soluble and basic salts incline to show yttrium associating with elements larger than holmium." (p. 1084)

[4] Gupta CK & Krishnamurthy N 2005, Extractive metallurgy of rare earths, CRC Press, Boca Raton, p. 165

"Yttrium...exhibits interesting behaviour in fractional precipitation and is amenable to purification by a combination of hydroxide and double sulfate precipitation. In [the latter]…yttrium behaves like holmium, and in [the former]…like neodymium."

[5] Finally, there is Restrepo's article (details following) showing Y is more similar to La, at least from a binary stoichiometric perspective:

See: Restrepo G 2017, "Building classes of similar chemical elements from binary compounds and and their stoichiometries", in MA Benvenuto and T Williamson (eds), Elements old and new: Discoveries, developments, challenges, and environmental implications, American Chemical Society, Washington, DC, pp. 95–110 (101).

I was surprised to learn that in the case of Y, there is still more to the question than the simple observation that Sc, Y and Lu occur in the so-called yttrium group.

In my view, we're again straying from the premise which is that there is no plausible explanation for skipping La, which represents the first comparable example of periodicity (including in terms of the overall chemical behaviour of group 3 being closer to groups 1 and 2) in favour of Lu. Sandbh (talk) 09:51, 9 January 2020 (UTC)

Because by that logic Sc goes under Al. The chemical behaviour of Al is pre-transitionised because it lacks the d-subshell over the noble gas core that Ga, In, and Tl have, so the same logic applies. Double sharp (talk) 11:35, 9 January 2020 (UTC)
Good point. The valence electrons are more important than the underlying core, however. Sandbh (talk) 04:21, 10 January 2020 (UTC)
So in that case we have our answer for La vs. Lu already: the chemical behaviour of La is pre-transitionised for the same reason, but we cannot put it under Y because the relevant valence shells don't match. La has low-lying empty f-orbitals; Y and Lu do not. Same as why Th doesn't go under Hf, but Rf does. Note that I say "valence shells" because the minutiae of Madelung exceptions are not important if they don't affect the chemistry, so Lr can go under Lu without any qualms. Groups 1 and 2 are a consistent exception even when relativistic effects are not at issue; delayed collapses for things like Ac are a normal thing for high atomic number and are easier to see as a second-order correction. Double sharp (talk) 04:33, 10 January 2020 (UTC)
@Double sharp: La goes under Y since the low-lying empty f-orbitals do not play a predominant role. Same as Sr under Ca even though the empty d orbitals of Sr play a bit-part role in its chemistry. Thorium goes in the f-block since it has an f electron available for chemistry. Sandbh (talk) 22:12, 10 January 2020 (UTC)
@Sandbh: The influence of 4f is tiny for Lu, even tinier than 3d for Zn, so even though 4f plays a relatively minor role for La it can be better accepted as a peripheral f-block element than Lu can (you have to pick one). In fact 4f influence in Lu is more or less at the same level of a core subshell. If you do the comparison for Ca vs. Zn, Droog Andrey has implied back in 2018 that he suspects Ca would show less 3d influence than Zn, so the argument works for that too (see Deep web talk:WikiProject Elements/Archive 33#Part 4). Double sharp (talk) 06:34, 12 January 2020 (UTC)
@Double sharp: I don't disagree. I do think that in a chemical table, where Ln3+ has the configuration [Xe] and Yb and Lu have the configurations [Xe]4f13 and [Xe]4f14, that this carries much more weight than the tiny influence of 4f in the gaseous state of an isolated La or Lu atom. Sandbh (talk) 22:46, 15 January 2020 (UTC)
@Sandbh: It seems to carry about zero weight when you consider that that 4f shell in Lu never gets breached for anything and never contributes anything significant to the bonding. At least the filled 3d shell in Zn does the second. Lu doesn't even have a 4f gas-phase differentiating electron (which is the criterion you've mentioned before), doesn't even have 4f as a valence shell, and insisting on ions is simply not generalisable across the table to look at equal-charged ions of elements from different columns. At least La has some low-lying 4f shell that we can say is weakened a bit in effect (but not that much, notice the high coordination numbers) just by the typical delayed collapse that happens as Z increases. Same as Ac and Th at the start of 5f, same as Lr at the start of 6d, same as E121 probably will be at the start of 5g. Double sharp (talk) 09:35, 16 January 2020 (UTC)
@Double sharp: That is why Scerri disdains arguments based on individual physical, chemical and electronic properties. They go back and forth. We talked about the possible influence of the f electrons in Lu, in our IUPAC paper. Ratto, Coqblin and d'Agliano (1969, pp. 498, 509) suggested that its lack of superconductivity might be attributable to a small 4f character. A few other authors have referred to some of the properties of Lu being influenced by the presence of its filled 4f shell: Langley 1981; Tibbetts and Harmon 1982; Clavaguéra, Dognon and Pyykkö 2006; Xu et al. 2013; Ji et al. 2015. The most surprising of these is likely to have been Clavaguéra and colleagues, who reported a pronounced 4f hybridisation in LuF3 on the basis of three different relativistic calculations. Their findings were questioned by Roos et al. (2008) and Ramakrishnan, Matveev and Rösch (2009).
I never ran all these references down, but there it is.
I think arguing about 4f character in La v Lu, compared to the other more major attempted arguments in the paper, is inconclusive and peripheral. Sandbh (talk) 06:10, 17 January 2020 (UTC)
They may go a bit back and forth, but without them you have nothing. If you ignore physical, chemical, and electronic properties, then just where did our periodic table come from? Might we not be just as fine arranging the elements alphabetically, as some wag suggested to John Newlands? Anyway, Droog Andrey is the computational chemist here, and he's mentioned that some calculations may overestimate 4f. (From Archive 34: "it seems for me that MP2 is not a good choice for correlation test. I'll try to use CASSCF if I have some spare CPU days.") Now, I'm sure I don't understand why this is the case (maybe he can explain the differences, or maybe it is too complicated), but there you go. To me, it is conclusive to say that the fact that Lu would have to ideally switch between "4f145d1" and "4f15" [the second one is nonsense] as an f-block element if it followed the mutability of configurations of the other 4f elements is a argument that Lu is not an f-block element. Double sharp (talk) 17:09, 22 January 2020 (UTC)
@Double sharp: That is why I’ve attempted to go more global/philisophical. I’ve talked elsewhere about the purported impact of Lu’s filled 4f shell on its reactivity, unlike Sc, Y and La. And the historical chemical perspective post. Sandbh (talk) 08:32, 23 January 2020 (UTC)
By that standard, we have to consider the impact of Rf's filled 5f shell on its reactivity (i.e. post-actinide contraction effects), and arrive at the conclusion that since Th has an empty 5f shell it should be the heavier congener of Hf. Which is exactly the traditional placement of it, and even after Seaborg's actinide concept won the day, there was some grumbling that U and Nd are almost strangers to each other chemically. So, from a historical chemical perspective, postulating a thoride series (which was, in fact, historically postulated before) would be just as justified as postulating Sc-Y-La. Is there a study that has not been subsequently disputed about 4f involvement in Lu that isn't just post-lanthanide contraction effects, anyway? As mentioned, Droog Andrey gives a plausible reason why some calculations may overestimate its importance. Double sharp (talk) 17:12, 24 January 2020 (UTC)
Greenwood & Earnshaw on Al, group 3, and Ti
@Double sharp: The first passage speaks to the better fit of Al over Ga; the second passage notes the mainly ionic chemistry of group 3 [consistent with groups 1 to 2] and the largely atypical behaviour of the group 3 transition metals; and third passage says that while Sc's status as a TM is arguable, there is no doubt that Ti is a TM.
1. "For instance, the mps and bps along with the enthalpies associated with these transitions, all show discontinuous increases in passing from Al to Sc rather than Ga, indicating the d electron has a more cohesive effect than the p electron." (p. 947)
2. "In the main, the chemistry of these elements concerns the formation of a predominantly ionic +3 oxidation state…although each member of this group is the first member of a transition series, its chemistry is largely atypical of the transition elements." (p. 948)
3. "The most important oxidation state in the chemistry of these elements is the group oxidation state of +4. This is too high to be ionic…whatever arguments may be advanced against applying the description to Sc, there is no doubt that Ti is a transition metal."
Sandbh (talk) 01:04, 19 January 2020 (UTC)
Notice that Zr and Hf are not mentioned, because by the standards that make Sc arguable, so are they. Double sharp (talk) 17:14, 22 January 2020 (UTC)

Helium over beryllium

P.S. By this version of the periodic law (which I agree with R8R is not the whole of what it is about), it seems to me that He has to go over Be: He is an s-element, Ne is a p-element, and comes after Be. So beryllium is the only possible next applicable recurrence of properties, as bad as it is in some respects, as Ne got disqualified by the same argument that you used to disqualify Sc as eka-Al. That one doesn't bother me, since I half-suspect now that the table really should have He over Be to respect the first-row anomaly principle. But perhaps it bothers you!~~ ^_-☆ ミ/ミ/ Double sharp (talk) 21:15, 26 January 2020 (UTC)

@Double sharp:He over Be won’t work since Ne is no longer then anomalous. Sandbh (talk) 22:16, 26 January 2020 (UTC)
Neon is still anomalous: it replicates perfectly the usual themes of the first-row anomaly of the 2p elements. Ne has a totally full n = 2 shell. Ar does not have a totally full n = 3 shell, since there are 3d orbitals. That's why Ne is the most electronegative element in its period, but Ar is effectively less electronegative than Cl (see Droog Andrey's scale). Indeed, the small radius of the 2p orbital means that the 2p elements are anomalously highly electronegative, and Ne conforms perfectly to this pattern, being even more electronegative than O and F(!). The huge electronegativity drop between the 2p and 3p elements is replicated perfectly between Ne and Ar (Droog Andrey's scale: 4.50 vs. 2.94; Allen scale: 4.787 vs. 3.242). The overly high electron density of Ne is also a consequence of 2s and 2p being so close in energy to each other, which is why it has such a hard time forming compounds: Ar does not have this problem, and can form a few. This is exactly like lone-pair repulsion in F2. Again, see Kaupp's paper on radial nodes.
P.S. Since you seem to be accepting the first-row anomaly principle in order to try to demolish He over Be: note that accepting that principle means that the actinides run the show for the La vs. Lu question, in which case the balance is even more skewed towards Lu/Lr, as Lr is very weird for a late actinide, but Ac is similar to the early actinides like Th. Double sharp (talk) 23:13, 26 January 2020 (UTC)
This strikes me as another example of an obtuse argument, never mind the main headline. You can only justify Ne as anomalous on the basis of an EN scale that nobody else has heard of. Meanwhile Ne is the least reactive of the elements, as opposed to He. That’s the headline anomaly. If it helps there was this article suggesting the halogens have the highest EN, the noble gases having values on par with the chalcogens. The author concluded that the chemistry of the noble gases is for a large part determined by their extreme hardness, equivalent to a high resistance to change in its electronic population coupled to their high electronegativity. Sandbh (talk) 06:50, 28 January 2020 (UTC)
Funny, I also gave the Allen scale. So much for "nobody else has heard of". And it also puts He and Ne as the most electronegative of the elements, even if it disagrees with Droog Andrey w.r.t. Ar. Ne being the least reactive element supports Ne as the first element in group 18 rather than He, so that's "another fruit in my bin" as Droog Andrey would say. It is not a front-line anomaly as you would like it to be because that is simply not how the 2p anomalies work. First-row anomaly is not a free-for-all that lets you put any element in the first row and declare it an anomaly, you know: they are strictly defined by what we see in the elements that we are sure are in the first row because of the effects of primogenic repulsion. Anything that cannot be attributed to that is not a first-row anomaly. It is either another kind of anomaly (e.g. inert pair effect) or it is just putting an element in the wrong place. The first-row anomaly effects for 2p make N, O, F by far the most extreme and electronegative elements in their columns, so the same should be true for Ne, and guess what, it is only if you let Ne head group 18. ^_^ Meanwhile, I hear nothing refuting my statement that Ne's difficulty in forming compounds comes from lone-pair repulsion just like for F. So, half of my reply is unrefuted, the other half is still valid because Allen's scale agrees completely with Droog Andrey's w.r.t. He and Ne, and some of your reply supports my argument anyway.
And your article still makes Ne significantly more electronegative than O, which is different from any other period! ^_^ And it states several times that He is always an outlier in these trends, but you can see in a very different way to how O and F are, which is more similar to Ne (He is just not on the trendline at all, while Ne-Ar goes in the right direction, just "too much", as O-S and F-Cl do)!! And uniformly, the biggest change in the normal direction is always from Ne to Ar, just like between O and S and between F and Cl in their columns!!! So, thank you for showing me an article that supports He over Be and group 18 starting with Ne like nothing else I've ever seen before!!!!
But I do find it amusing that you are now fighting my argument on my terms with first-row anomaly considerations. I wonder why this is suddenly not valid for Sc-Y-Lu? I wonder why the periodicity and differentiating electron argument is suddenly not valid when it looks like it has started to support He-Be-Mg? So are you sure you are being objective, or did you start with whatever you wanted to show, search for things to justify it, and neglect to look at what happens in the rest of the table? ^_^ Double sharp (talk) 12:16, 28 January 2020 (UTC)
@Double sharp: I addressed much of this in the General comments section. I haven't intrinsically relied on EN scales in my draft. There is no strict definition of the first row anomaly. Primogenic repulsion can play a part but that is not all there is to the story. Sandbh (talk) 23:04, 31 January 2020 (UTC)
That's precisely why it's useful: it's not strictly defined by one property, but is characterised by a multitude of them showing up. We're not trying to pick out properties willy-nilly. We're trying to see what properties can be theoretically attributed to primogenic repulsion. Given how much it impacts the 2p, 3d, and 4f elements (I trust I do not have to scour the literature to prove what Kaupp already did), surely we should reflect it for the 1s elements. And you still haven't addressed that your version of the periodic law points unambiguously to He over Be. Is that what you want? Many of your arguments, if applied to groups 2, would either support He in group 2 or Be and Mg over Zn, which I have demonstrated several times. (Immediate neighbours support Be-Mg-Zn because Be and Mg are weak metals, more similar to Zn than to Ca; this weakening does not appear with Li-Na-K. They also don't oppose H-He, because H stands in relation to Li much as He stands in relation to Be. Differentiating electrons are inconclusive for Be-Mg-Zn, and roundly support He over Be to put it in the right block. Your version of the periodic law supports He over Be, not to mention that in the way it is phrased in your article it supports Al over Sc as well. And the rare-earths argument just does not make any sense to me, given that we all know how to read text from left to right within each line.) Is that what you want? If not, then I claim it at least casts doubt on the usefulness of an argument about group 3, if it gives the "wrong" answer elsewhere. Double sharp (talk) 23:10, 31 January 2020 (UTC)

@Droog Andrey: What's your opinion on He over Be, actually? (Just curious.) Double sharp (talk) 13:57, 28 January 2020 (UTC)

@Double sharp: I'd place He over Ne and (He) over Be, and at the same time H over Li and (H) over F. Droog Andrey (talk) 18:46, 29 January 2020 (UTC)
@Droog Andrey: Yeah, I can agree with that. Everyone knows they are s-elements sometimes drawn apart anyway, so the s>>p>d>f first-row distinctiveness is retained anyway. ^_^ Double sharp (talk) 18:57, 29 January 2020 (UTC)
@Double sharp: You know, I was hunting (computationally) for He-bonded structures in 2011, and discovered HeNiCH4 (C2v symmetry) with He-Ni bond of 1.6Å. The bond was quite strong (around 30 kJ/mol), but unfortunately Ni-CH4 complex appeared to be metastable against H-Ni-CH3. At that time I studied a lot of VdW-bonded complexes of noble gases, and figured out that, indeed, Ne is much more inert than He, with He quite resembling Be in bond nature. It is pleasure for me that Grochala made the same conclusions. Droog Andrey (talk) 19:44, 30 January 2020 (UTC)
@Droog Andrey: Wow, that's cool! So maybe it really should be He over Be and (He) over Ne after all. ^_-☆ I guess I should link to Grochala's PCCP paper on (HeO)(LiF)2 and the supplementary information (the latter is where he considers the implications for the periodic table). It seems that He follows Be in showing more affinity to O than to F. I cannot resist quoting his section S14:

Having studied (HeO)(LiF)2 we have scrutinized its heavier noble gas analogues. However, neither (NeO)(LiF)2 nor (NeO)(NaF)2 could be detected as minima on the singlet PES. This stability difference is interesting since electronegativity and many other chemical properties of elements change quite monotonically when one goes down any of the Groups 13–18. For example, O is more electronegative than S, F than Cl, Ne than Ar, etc. Hence, He should be more electronegative and as such less prone to chemical bonding than Ne.

The atypical behaviour of He vs. Ne (i.e. reversal of stability for their theorized chemical connections) which obviously contrasts with the trend for the 1st ionization potentials of these elements (He > Ne), has been noticed before. Various explanations were provided from electrostatic arguments to increased Pauli repulsion from the filled 2p orbital on Ne. We think that a relatively large reactivity of He with respect to Ne may be understood simply in terms of substantial charge density which appears at a small He center when its 1s2 shell is even partially depopulated. The incr[e]ased charge density obviously leads to stronger electrostatic and dispersive interactions with ligands.

This no doubt recalls hydrogen; helium then joins it as an example of what happens when you literally have no shielding at all from the nucleus, as only happens in the 1s row. ^_^ He then argues for He over Be, by virtue of the isoelectronic 1s2-2s2 analogy (of course Be-O is more ionic in character than He-O), and by the properties that you mentioned (where Ne is much more inert than He, whose bonds resemble those of Be). Personally, I found this very convincing myself. It's amazing that He seems to be less electronegative than O here(!!). Previously I thought He over Be was silly, but now I see it has actually significant chemical and physical sense. (I changed my userpage periodic table to support it back in August last year. ^_^) Double sharp (talk) 20:08, 30 January 2020 (UTC)

Rare earth series

The rare earth series (Sc, Y and the lanthanides La–Lu) appear listed in order of their atomic numbers in a 32-column periodic table with Group 3 as Sc-Y-La-Ac. If Group 3 is shown as Sc-Y-Lu-Lr, the minority of the rare earths appear in order of their atomic number whereas the majority appear in a backwards order. [...] The second option is awkward, or highly anomalous at best, since the horizontal, vertical, and diagonal trends that characterise the periodic table are based on an increasing sequence of atomic numbers. In both variants the atomic number increase from left to right and from upside down, there's no any backwards order. All the trends remain. Droog Andrey (talk) 11:19, 7 January 2020 (UTC)

Sandbh: What I meant to convey was that IUPAC defines the rare earths as Sc, Y and the lanthanides…
La57 Ce58 Pr59 Nd60 Pm61 Sm62 Eu63 Gd64 Tb65 Dy66 Ho67 Er68 Tm69 Yb70 Lu71
…rather than the lanthanides and Sc, Y:
La57 Ce58 Pr59 Nd60 Pm61 Sm62 Eu63 Gd64 Tb65 Dy66 Ho67 Er68 Tm69 Yb70 Lu71
The first option can be "bent", in ascending numerical order, to read Sc 21 to Lu 71.
@Sandbh: But that has nothing to do in "The domain of chemistry" section. That's about design and personal taste. Droog Andrey (talk) 19:05, 8 January 2020 (UTC)
@Droog Andrey: [S6] I supported what I said here with my citation to Scerri: "The horizontal, vertical, and diagonal trends that characterise the periodic table are based on an increasing sequence of atomic numbers." Sandbh (talk) 05:38, 9 January 2020 (UTC)
@Sandbh: Well, which particular trends are broken, in your opinion? Droog Andrey (talk) 06:59, 9 January 2020 (UTC)
It's not possible to bend option 2 in ascending numerical order. Sandbh (talk) 05:25, 8 January 2020 (UTC)
@Sandbh: But you can just read each one, left to right in each row, and one line at a time from top to bottom, the order you read anything. Where is the difference? Only the design. Double sharp (talk) 06:55, 9 January 2020 (UTC)
@Droog Andrey: @Double sharp: Well, the majority of graphs I can recall plotting various properties of the the rare earths, as is the case with all other groups, list the elements concerned in order of Z. Increasing Z = the foundation of the periodic system. I'm aiming for cognitive congruence here not dissonance. Sandbh (talk) 09:11, 9 January 2020 (UTC)
@Sandbh: But once you read it left to right within each row, and row by row (which is the normal reading order), it is in order of increasing Z regardless of whether you show it as Sc-Y-La or Sc-Y-Lu. Double sharp (talk) 09:22, 9 January 2020 (UTC)
@Double sharp: Yes, I follow that. The cognitive dissonance arises when IUPAC says the rare earths are Sc, Y, La to Lu. In an La table you can see that Sc-Y-La overlaps with La to Lu. In an Lu table the rare earths appear to be…well, I don't know what they are. You can't get to Sc, Y, La to Lu, without going backwards at some point. Sandbh (talk) 10:02, 9 January 2020 (UTC)
@Sandbh: They still form a continuous series defined by continuous sets of rows and columns that can be read left to right in each row. Double sharp (talk) 11:28, 9 January 2020 (UTC)
@Sandbh: I usually say that rare earths are IIIB from Sc to Lu and 4f from La to Yb. Droog Andrey (talk) 12:32, 9 January 2020 (UTC)
@Droog Andrey: @Double sharp: It’s clear I need a major relook at this part of the article. Perhaps my argument should be that the REM in a ScYLu table appear odd since they represent the only set of elements in which the second component (i.e. the Ln namely La-Lu) starts with a lower Z (57) than the end of the first component i.e. the ScYLu triad as a subset of Group 3 (71). I’ve probably just repeated my earlier observation that the REM in option 2 (ScYLu) can’t be straightened out into a continuous line. They can in your head; they can’t in the conventional 2D PT since the horizontal, vertical, and diagonal trends that characterise it are based on an increasing sequence of atomic numbers. That works for the REM in an La table but not in an Lu table. Sandbh (talk) 04:10, 10 January 2020 (UTC)
@Sandbh: Post-transition metals look about like that too if you consider group 12 as a post-transition group, with them starting in group 13 in period 3, but in group 12 for the later periods. What does it matter? And why is it important to straighten out sets of elements into a continuous line, when the PT is 2D, and both the horizontal and vertical neighbours are significant? Double sharp (talk) 04:38, 10 January 2020 (UTC)
@Double sharp:That’s a good observation about the PTM. I think I know what I meant to say all along. The 18 groups, the Ln, and An appear in order of increasing Z. This also applies to the RE in an La table, but not in an Lu table. I haven’t said anything different here but it is expressed more clearly. Sandbh (talk) 21:57, 10 January 2020 (UTC)
@Sandbh: OK, I think I finally understand what you are saying: if you force the above arrangements topologically into a straight line by forgetting about everything but which cell is next to which cell, then indeed Sc-Y-La produces a sequence in atomic number order and Sc-Y-Lu doesn't. However, I question the strength of this argument: to me it seems completely inconsequential. They always appear in the order of increasing Z if you just read every row from left to right, just like anything else on the periodic table, and in just the way you read everything else including this text, which does not involve bending anything into a straight-line arrangement at all. Even on a Sc-Y-Lu 18-column table you are going to have some asterisks before Lu telling you to go down and read the insertion of La through Yb first. Isn't this bending all pretty artificial anyway? Where does this argument end up for a completely 2D arrangement like the transition metals, where you can't flatten it into a 1D straight line at all? But of course, no one has any trouble reading the TM's in order of increasing Z even if Cu and Zn are not adjacent to Y in any way. Double sharp (talk) 13:38, 12 January 2020 (UTC)
@Double sharp: It think it’s a philosophical argument based on the foundational nature of the horizontal and vertical relationships seen in the 18 groups, Ln, An, and REM, all of which feature contiguous increasing Z. There’s nothing artificial about that. It goes deeper than L to R reading. Sandbh (talk) 04:01, 13 January 2020 (UTC)
@Sandbh: But important classifications such as transition metals, nonmetals, etc. don't feature contiguous increasing Z, whereas every single one of them does if you just read from left to right (inherited from the periodic table's layout). And bending a line of elements into a straight line, when it isn't actually straight in the PT, already requires artificially pulling it away from a more chemically sensible 2D arrangement. Double sharp (talk) 04:05, 13 January 2020 (UTC)
@Double sharp: I think we're talking about different concepts. One is about IUPAC periodic groups and series; the other is about metallicity regions in the table as we do with our own Deep web table. It's a bugger I have to keep going on about this but I get hammered about the importance of vertical groups and horizontal series in increasing order of Z all the time. This is not about L-R reading order, its a philosophical argument about the "regularity" of trends in groups and series. In this philosophical context an Lu table is less regular than an La table. Sandbh (talk) 04:25, 15 January 2020 (UTC)
@Sandbh: REM is not a group. It doesn't stretch to period 7, and even if it did, group 3 shouldn't contain 32 elements. So comparisons to other classes of chemically similar elements like transition metals or platinum group metals is absolutely relevant, and then the lack of Z order is no big deal since the PGM cannot even be flattened out into 1D without breaking any neighbours. (Not that it ever was in the first place IMHO, since "unfolding" the group into 1D breaks the arrangement of the periodic table in the first place that we are supposed to be justifying.) So I continue to be totally unconvinced by this philosophical argument since it's totally irrelevant to almost any other series of a similar type to REM. Double sharp (talk) 05:16, 15 January 2020 (UTC)
@Double sharp: The REM are recognised by IUPAC in the Red Book, as are the Ln and An. So there are 18 groups with vertically increasing contiguous Z, one series with increasing Z contiguously going around a corner (Sc, Y, and the Ln), and two series with horizontal contiguously increasing Z (Ln and An). These are the key vertical and horizontal trends. The horizontal discontiguous trends along the transition metals are of secondary import. Sandbh (talk) 23:08, 15 January 2020 (UTC)
@Sandbh: So then why are we mixing the vertical trend across the group 3 column with the horizontal trend across the Ln? They can overlap at Lu, but they're different lines. Where is the chemical sense in the trend Sc-Y-La-Ce-Pr-...-.Yb-Lu, rather than subsets of it? Double sharp (talk) 09:37, 16 January 2020 (UTC)
@Double sharp: The mixing of the vertical and the horizontal occurs with either option. Discussions of the rare earths list them in order of Z. The trends can then be discussed as Sc-->La-->Lu, rather than the "cumbersome" Sc-->Lu and La-->Lu. I did recently see a graph of the ionic radii of Sc-->La-->Lu, with Ce to Lu shown as not having a contraction, compared to them having the contraction. Sandbh (talk) 00:24, 18 January 2020 (UTC)
@Sandbh: Sc-->La is a useful trend, but so is Sc-->Lu, so in any case you will want to show both to give an idea of what is happening in group 3. And once you do that you are going to have separate overlapping horizontal and vertical trends anyway, so why not do that instead of pouncing on the fact that one glues in "the right order"?, even if a trend starting Sc-Y-La-Ce-Pr doesn't really make any sense when you graph it? Double sharp (talk) 10:03, 19 January 2020 (UTC)
@Double sharp: My clumsy wording may not have helped. By Sc-->La-->Lu, I mean Sc21-->Y39-->La57 and then La57-->Ce58-->Lu71. On the other hand, Sc21-->Y39-->Lu71, followed by La57-->Lu71 is cumbersome, irregular, and backwards even. The Sc-Y-La-Ce-Pr trend makes sense if you graph it and include a Ce-->Pr--> etc line showing the trend without the Ln contraction. Sandbh (talk) 06:53, 20 January 2020 (UTC)
So what if it is backwards? They are two separate trends, you don't need to glue them together. You can show B-->Al-->Ga as a trend along with (Ca)-->Sc-->Ti-->…-->Zn-->Ga to show the d-block contraction as well as the group 13 trend, but that doesn't mean that B-->Al-->Ga is somehow creating a backwards trend and B-->Al-->Sc would be better. Double sharp (talk) 17:08, 22 January 2020 (UTC)
@Double sharp: It’s cognitively dissonant with the relevance of increasing Z seen everywhere in the PT. Sandbh (talk) 08:36, 23 January 2020 (UTC)
Well, so is B-->Al-->Ga combined with a d-block contraction trend (comparing Sc with what Ga ends up as, as they differ more or less by 3d10). Is that an argument for B-->Al-->Sc? I doubt it. We have two trends, each with increasing Z, that happen to intersect: Sc-->Y-->Lu and La-->Ce-->…-->Yb-->Lu. Increasing Z is respected within each trend. Only when you glue them together is there any problem, but it is artificial since there was never any reason to do it in the first place. If you interpolate and extrapolate within a trend like Sc-->Y-->Lu-->Lr, or even Sc-->Y-->La-->Ac, you get actually relevant results. Where is the sense of Sc-->Y-->La-->Ce-->…-->Lu, again? And where do the poor actinides end up, with Ac so close to La in chemistry, and Lr so close to Lu (never mind the odd p-electron that seems to mean nothing chemically)? Double sharp (talk) 11:52, 23 January 2020 (UTC)
@Double sharp: No it isn’t, given they are all p elements. I’ve already explained the sense of the rare earths in terms of increasing Z, and showing what would happen if there was no contraction. I have no particular interest in opening up the 5f can of worms more than I already have, noting they less or more fall into line under the 4f series. Sandbh (talk) 06:08, 24 January 2020 (UTC)
A trend that continues from Sc, Y, La and then to Ce does not make any sense, mathematical, chemical, or otherwise. If you extrapolate downwards given such a trend the only sensible thing to put there is Ac. That's why I stand by the statement that you are forcing two separate trends together, cutting off the 7th period artificially (yes, totally artificially, or else why does the periodic table go past Rn?), and arguing about how the gluing comes out. When in actuality the gluing does not make sense in the first place. So for B-Al-Ga you have to appeal to blocks instead to get the conclusion you want; well, in that case, Lu is far more like a d-element than La is, and we're back to Sc-Y-Lu. Double sharp (talk) 17:07, 24 January 2020 (UTC)


  • Hevesy G 1929, Redkie zemeli s tochki zreniya stroeniya atoma, (Rare earths from the point of view of structure of atom), NKhTI, Lenningrad, cited in Trifonov 1970
  • Greenwood NN & Earnshaw A 2002, Chemistry of the elements, 2nd ed., Butterworth-Heinemann, Oxford
  • Greenwood NN & Harrington TJ 1973, The chemistry of the transition elements, Clarendon Press, Oxford
  • Lee JD 1996, Concise inorganic chemistry, 5th ed., Blackwell Science, Oxford
  • Stewart 2018a, “Tetrahedral and spherical representations of the periodic system”, Foundations of chemistry, vol. 20, pp. 111–120
  • Trifonov DN 1970, Rare-earth elements and their position in the periodic system, translated from the 1966 Russian edition, Academy of Sciences of the USSR Institute of the History of Natural Sciences and Technology, Moscow, published for the Atomic Energy Commission and the National Science Foundation, Washington, by the Indian National Scientific Documentation Centre
@Sandbh: An overall impression is that the article provides enough opinions from literature, but lacks analysis of chemistry. Droog Andrey (talk) 12:38, 9 January 2020 (UTC)
@Droog Anthony: Thank you. That’s probably ok since the article is premised on the supposition that the question can’t be resolved on the basis of a comparison of individual physical, chemical or electric properties. Sandbh (talk) 07:27, 10 January 2020 (UTC)
@Sandbh: Why do you repeatedly distort my nickname? Droog Andrey (talk) 09:51, 15 January 2020 (UTC)
@Droog Andrey: That's my fault, sorry. I subconsciously had the Doug Anthony All Stars on my mind. Sandbh (talk) 23:33, 15 January 2020 (UTC)

Horizontal triads

@Double sharp: Can I revisit this, last discussed in Archive 40.

It's the recurring sequence of maximum oxidation numbers of +2, +3, and +4. This happens for the following sets of horizontal triads (P = period):

P +2 +3 +4
2 Be B  C
3 Mg Al Si
4 Ca Sc Ti
4 Zn Ga Ge
5 Sr Y  Zr
5 Cd In Sn
6 Ba La Ce
6 Hg Tl Pb 
7 Ra Ac Th

Whereas Sc, Y, La, and Ac are the middle elements of such horizontal triads, Lu and Lr are not:

Yb (+3) Lu (+3) Hf (+4)
No (+3) Lr (+3) Rf (+4)

This seems to be another example of where an La table is more regular than an Lu table.

Last time you responded by focussing on what happens in the rest of that table. That's not relevant since I'm only interested in the implications for group 3. Sandbh (talk) 06:42, 17 January 2020 (UTC)

That is completely relevant in my opinion, because you can't expect a criterion that doesn't work for most of the table to be a relevant rather than an overly localised one. Besides, where are we going to be with Cn (+4) Nh (+3) Fl (+2), like I said? Double sharp (talk) 10:01, 19 January 2020 (UTC)

@Double sharp: I'm only noting that an Lu table is less regular than an La table, wrt these oxidation state triads. What happens to Cn (+4) Nh (+3) Fl (+2) makes no difference. What's happening in the rest of the table is irrelevant to this particular pattern encompassing group 3. Sandbh (talk) 07:07, 20 January 2020 (UTC)

@Sandbh: That's the whole point. How is this an important criterion for regularity when it cannot extend to the whole table? If we want a philosophical justification IMHO it must make sense for the PT as a whole rather than just be an ad hoc situation for part of one group. Double sharp (talk) 14:44, 20 January 2020 (UTC)

@Double sharp: It can be applied to the whole PT, just like the n + l rule can be applied to the whole table. In both cases it makes no difference to the overall result. Sandbh (talk) 08:40, 23 January 2020 (UTC)

So how does group 16 look under such a criterion, when the maximum oxidation state of O breaks the pattern from S downwards? This is why for periodicity it makes more sense not to look at maximum oxidation states but at all significant ones. Clearly Mendeleev must have done this too, since he put O in group VI and F in group VII. And then there is no problematic difference in regularity, because Yb and No have a significant +2 state as well. Double sharp (talk) 11:50, 23 January 2020 (UTC)

@Double sharp: I don’t know and I don’t care. The outcome would be the same for an Lu table or an La table. Sandbh (talk) 05:58, 24 January 2020 (UTC)

So, as it stands, it doesn't make sense as a criterion. If we correct it to consider all significant oxidation states, it becomes more sensible. But then it is inconclusive between Sc-Y-La and Sc-Y-Lu, correctly highlighting the small difference involved. Double sharp (talk) 17:01, 24 January 2020 (UTC)

@Double sharp: It doesn’t need correcting. It is a criterion based on maximum oxidation state, which is fine by itself. Like I have repeatedly said, extending the criterion to the rest of the table makes no difference. The 234 pattern is only present in an La table. End of story. Sandbh (talk) 11:58, 25 January 2020 (UTC)

I can make an infinite number of criteria, applicable only to one group, that will conclusively point to any possible next element for it and declare everything else to create an irregularity. And in the absence of evidence why each of these criteria should be upgraded to the status of a law of the periodic table, which means seeing if they actually apply in any significant other part of the periodic table it wasn't invented for, their relevances will all be equal: that is, equal to zero. That's why I insist that a criterion must meet that minimal standard. Respectfully, a criterion like this one, which considers N to be an irregularity because it breaks the recurring sequence of 456 oxidation states, does not meet it. IMHO it also fails the test of looking only at relevancies. Suppose Hg did happen to show a +4 state after all: does that in any way weaken Tl as a group 13 element for breaking this pattern? Double sharp (talk) 12:33, 25 January 2020 (UTC)

Coordination complexes of the lanthanides: a case study

Consider the complexes Ln(2,2′-bipyridine-1,1′-dioxide)4](ClO4)3, studied in this 2002 article.

As the lanthanide ion Ln3+ gets smaller, we of course have a difference in geometry due to steric hindrance. But it is revealing what exactly these differences are, because it reveals something that appears quite significant when it comes to 4f involvement.

The shapes of the La and Ce complexes are almost perfect cubic. Although this demands only 8-coordination, symmetry grounds force us to posit 4f involvement because of this shape – including for La!

Once we get to Pr and Nd, some symmetry is lost, and we arrive at a slightly distorted snub disphenoid. At Eu we have lost enough symmetry that there are now two crystallographically different types of Eu3+ ions, and this structure appears to hold at Ho as well. But at Lu3+ we have distorted the cube, it appears mostly continuously, all the way to a slightly distorted square antiprism, which (since it is a known geometry for Xe) clearly requires no invocation of f-involvement.

I find this pretty revealing. Since 4f energy is highest for La and Ce, and should go down gradually while it sinks into the core, I suspect the change in shape may be related to the lowering 4f contribution, though I regret that I am not well-versed enough in f-block chemistry to say that for sure. ^_^ Double sharp (talk) 17:28, 22 January 2020 (UTC)

@Double sharp: Yes, I’ve seen speculation about 4f orbital involvement by La on symmetry grounds but nothing more. That article says naught about this. If there is something to it then it strikes me as being similar to d orbital involvement in the incipient transition metals Ca, Sr and Ba. Sandbh (talk) 09:06, 23 January 2020 (UTC)
So we have exactly the same situation in the f-block as we have in the d-block: Ca through Zn all have some d-involvement, and so do La through Lu. But the reason why we put Zn, Cd, and Hg in the d-block, and not Ca, Sr, and Ba, is simply that:
  1. The d-block cannot fit eleven columns, so we must decide which one is more anomalous for the d-block, and clearly it's Ca-Sr-Ba:
  2. Putting Ca-Sr-Ba as an s-block group gives the right idea that the s2 shell is the covering one for the most part:
  3. And the energy level of the d subshell increases down group 12 relatively speaking (to the point that at Cn it is expected to become chemically active), so it is clearly not yet a core subshell.
If we apply these principles to the f-block question, (1) and (3) clearly favour La as an f-block element. (2) also gives the right idea for the gas phase that it is 4f-5d that are the chemically relevant occupied subshells in ions and 6s is mostly ionised away and empty. If we put Lu as an f-block element, then we're sort of implying that 5d6s2 always gets ionised away, which is weird when you consider configurations like La2+ [Xe]5d1 in compounds. You never see 6s occupied in a Ln2+ or Ln3+ or Ln4+ ion, but you might see 5d and you will almost always see 4f, suggesting that the difference is between 6s and the others. Double sharp (talk) 11:49, 23 January 2020 (UTC)

Lanthanide contraction

I’ve talked about this before in another context.

Goldschmidt's Ln contraction, which is caused by poor shielding from the f-electrons, starts at Ce3+ [Xe]4f1 and finishes in Lu, since Yb3+ is [Xe]4f13, and Lu3+ is [Xe]4f14. Same thing happens with the actinides.

In an La table the cause of contraction naturally spans the f block as Ce to Lu. Form and cause are harmonised. In an Lu table the cause of contraction does not start until the second element of the f block; the cause of the contraction then finishes after the end of the f-block, in the d-block (1st element, period 6). Form and cause are disaggregated.

That is to say, in an La table the cause is congruent with its associated boundaries whereas in an Lu table the cause is misaligned with these boundaries.

OTOH, in an Lu table the number of f electrons in each individual Ln in the gas phase mostly matches its position in the f-block. Then again, in an La table the number of f electrons in the trivalent ions perfectly matches their f block position number.

In the above light I’d conclude that the La table is a better chemical table, in the context of the Ln contraction. Sandbh (talk) 05:27, 12 January 2020 (UTC)

That's not all there is to it, actually:
  1. The contraction readily extrapolates backward to the case of La3+ at [Xe]4f0, which is (as everyone expected) bigger than Ce3+. Such contractions are common enough throughout the entire table. Just look at Y3+ through Ag3+, for example, or look at the atomic radii of neutral Li through Ne. They are just more significant in the Ln than anywhere else because they are all usually in the same oxidation state and the noise that usually wipes these contractions out into irrelevance disappears. But then it becomes important to include La as part of the effect.

This is not directly relevant to my argument. La is not a part of the Ln contraction per se, since the filling of the f orbital has not yet started. Of course, the knock-on effects are felt from Hf onwards, but this is easily distinguished in that the number of f electrons in the applicable ion peaks at Lu. Sandbh (talk) 10:24, 12 January 2020 (UTC)

@Sandbh: Yes it is because part of the point of the lanthanide contraction is that each Ln is smaller than its predecessor because it has one more poorly shielding f electron that doesn't offset very well the effect of one more proton in the nucleus. And then La vs Ce is relevant because 0 + 1 = 1. Double sharp (talk) 05:20, 15 January 2020 (UTC)
  1. Another effect of the Ln contraction is that the 5d series that comes after it is made much more similar to the 4d series than you would otherwise expect, because it almost exactly cancels out the expected increase in atomic radius. In a Lu table, the whole 5d series from Lu to Hg is uniformly affected. In a La table, La is the odd man out from Hf to Hg. The 5d series then becomes less homogeneous because obviously Lu is more similar to the 5d transition metals than La (look at just about any properties: coordination chemistry, melting and boiling points, hardness...)
Double sharp (talk) 06:21, 12 January 2020 (UTC)
@Double sharp: This is another good example of focussing on minutiae while ignoring predominating characteristics. IUPAC submission Double sharp (IDS) would’ve demolished this one.
In the absence of IDS I’ll step up to the plate.
Rather than dwelling upon minutiae let’s look at general characteristics.
Metallurgically speaking we know that Lu resembles closely erbium and holmium, except that it melts at a slightly higher temperature and is essentially non-magnetic, while the details of producing, purifying and fabricating it are almost identical with those applicable to holmium.
Restrepo separately found that Lu ended up in a cluster with Er and Ho and that Lu is more similar to lanthanoids than to transition metals, while La share similarities with lanthanoids and with transition metals.
Like DIM, Restrepo based his work compounds and their proportions of combination.
Case closed! Demolition complete! Send in the clean up crew :) Sandbh (talk) 23:41, 25 January 2020 (UTC)
As I keep saying: Restrepo is only considering resemblances between stoichiometrically identical compounds. Such an argument will never find any similarities between elements from different groups, even when they are relevant. (And indeed, my case for Sc-Y-La was more sophisticated, based on delayed collapses and condensed-phase configurations, until Droog Andrey kindly explained back in Archive 33 what the problem with them was.) I'm also not terribly thrilled about separating "IDS" from "current DS" as if we are separate people. Current DS just knows more chemistry and had his actually not-so-bad Sc-Y-La arguments very kindly critiqued and explained by Droog Andrey in archive 33. His logical style is the same; it's just that now he's combating bad Sc-Y-La arguments instead of bad Sc-Y-Lu arguments. There are a lot of bad Sc-Y-Lu arguments in the submission that I am quite happy to stand by my 2016 criticism of. There are also some where I think our initial criticism was true but not the full story. But there are also some, like this one, where I think we missed the point of the original argument.
Again, surely no one is disputing that Lu closely resembles the late lanthanides. Equally well, La closely resembles the early lanthanides. No one is arguing that Lu is not a lanthanide. No one is arguing that La is not a lanthanide. I am arguing that among La and Lu, Lu has more transition metal properties. Not only is this just obvious from looking at all properties, and certainly not minutiae (unless just about all properties are minutiae, since it affects melting points, hardness, etc.), our submission even mentions a paper that does exactly this: We critiqued it by saying that the authors go too far by saying that Lu is a transition metal, and I agree with that: it's also a lanthanide, not only a transition metal, and clearly it is more similar to the other lanthanides than the transition metals. But come on, so is La. Double sharp (talk) 00:00, 26 January 2020 (UTC)

Here is a short summary of my logic, since I think you may have misunderstood it:

  1. La and Lu are obviously very normal lanthanides, and no one is disputing that.
  2. However, we want to give one of them the 5d1 position.
  3. Therefore we must find which one is closer to the metals Hf through Hg in the 5d2 through 5d10 positions. Notice that nothing in this denies that both La and Lu are close to the early and late f-block lanthanides respectively. Those are obviously their closest kin chemically. We just want to make the d-block as homogeneous as possible.
  4. And obviously, in spite of Restrepo (whose argument I have demolished several times identically already), the answer is Lu. Just compare melting points. Or boiling points. Or coordination power. Or density. Or hardness. Or just about anything else. If these are "minutiae", we have nothing at all to draw a periodic table based on, and we can all pack up and go home.

Double sharp (talk) 00:03, 26 January 2020 (UTC)

@Double sharp: Nice. Thanks for that.
As well as Restrepo’s peer-reviewed article, here’s another paper showing La is not close to the Ln but instead falls into a cluster with 1 and 2 metals. [Of course the story is more involved than that but the broad contours are there]. So much for La as a normal Ln. See how Lu falls into the same cluster as the Ln. And what about, as I’ve noted elsewhere, the metallurgy etc of Lu showing it closely resembles that of Ho and Er? Your line of reasoning doesn’t hold up. Sandbh (talk) 10:40, 26 January 2020 (UTC)
  1. While I am a bit sceptical about anything separating La from Ce, Pr, and Nd, which are so similar that lanthanum-gobbling bacteria don't even notice which of them they got, notice how the cluster La falls into is with group 1 and 2 metals. In other words, not transition metals. So it is clearly even farther away from being like Hf through Hg, which we are trying to mimic for the first 5d position.
  2. Once again, the whole point is not "which one is more like a normal lanthanide". The whole point is "which one is closer to the metals Hf through Hg" (my point 3 previously). I put it to you that this supports that the answer is not La. And just plot all those properties, you'll see the answer is almost 100% skewed towards Lu (see my point 4).
  3. (If comparing to nine elements at once is confusing: here is an even bigger simplification. Compare the pairs La-Hf and Lu-Hf. Everyone can see that the second pair has more in common.)
  4. And just to hammer the point home: do the same thing for Ac vs Lr. Not only will you find that Lr is by far closer to Rf through Cn, you will find that lawrencium is in fact extremely weird for a late actinide by preferring trivalency! Actinium, in going for the maximum oxidation state it can muster, fits pretty well with thorium through uranium, at least.
  5. And if any further proof were needed, the 6f series is expected to side with the 5f series, not the 4f series. Seaborg seems to have been completely right when he predicted prophetically that if we could have a very extended table with numerous "f" series, like we could with "d" series, "p" series, and "s" series, we would find that the actinides are more representative of typical f-elements than the lanthanides. The first row is always anomalous.
  6. For that reason, the following principle should hold for La vs. Lu arguments: the actinides overrule the lanthanides. Just like normal p-behaviour is 3p onwards, normal s-behaviour is 2s onwards, and normal d-behaviour is 4d onwards: normal f-behaviour should be taken as 5f (since we currently do not have the 6f elements to examine). So: if looking at the lanthanides is inconclusive, but looking at the actinides is conclusive, we follow the actinides by the principle of looking at more representative examples of each category (here, f-elements). This is not slavish devotion to the Madelung rule; we know the first-row anomaly in each block really well, and it is well-known that the cause is the absence of radial nodes (see Kaupp's paper). And deriving their presence or absence is almost pure mathematics: solve the Schrodinger equation. (See this Chemistry Stack Exchange answer.)
  7. In this case, of course, this principle is only needed for confirmation. The 5d homogeneity argument favours Lu quite clearly; the 6d homogeneity argument if anything favours Lr even more clearly. Therefore, Sc-Y-Lu-Lr is the recommended version by them. Double sharp (talk) 10:44, 26 January 2020 (UTC)
@Double sharp: This is a good example of ignoring the broad contours and delving into details (noise) and struggling to draw attention away from the findings of the article since it does not support your view.
Note that Lu falls into the Ce-Yb etc cluster, rather than the transition metals.
In the article, La is as much a TM as is Lu. La has the same colour as Au an Hg; ditto Lu for Os Ir.
Colour wise, La would fit better under Y and Lu (and Lr) would fit more naturally at the end of the f-block.
The authors did not recommend Sc-Y-Lu-Lr; they just happened to use the WebElements table.
As per my recent post, the issue is not block homogeneity or similarity in terms of details but chemical behaviour and periodic trends. Sandbh (talk) 02:28, 2 February 2020 (UTC)
@Sandbh: If you have any sense of chemistry left, and not just parroting classification papers, you should realise that a resemblance from La to the weak metals Au and Hg is simply laughable. Which sums up what I have to say about this. I don't ignore it because it is inconvenient. But because what you say about it suggests that it is chemical nonsense. Double sharp (talk) 10:30, 2 February 2020 (UTC)

In an Lu table, Lu to Hg are not uniformly affected since the effect peaks at Lu and diminishes thereafter. In an La table, La is not the odd man per se, but rather Hf onwards are odd due to the impact of the intervening Ln contraction. I do agree La is more similar to Ba. Lu is quite a good heavy Ln based at least on its occurrence, method of preparation, mechanical properties, reactivity IIRC), and the stoichometry of its binary compounds. Sandbh (talk) 10:24, 12 January 2020 (UTC)

@Sandbh: That's bad use of language on my part, sorry: I mean that all ten of Lu through Hg are affected, whereas if you make your d-block La + Hf through Hg only nine are. Still, it makes a more sensible trend if the effect peaks at the first one and then goes down monotonically, as we get with the Lu table, than if we start with no effect, and then have a huge jump for the second element before going back down monotonically (and yet not to zero again), as we get with the La table.
As for Hf onwards being odd, allow me to quote Restrepo back at you ^_^: 'Although the Zr-Hf similarity is well documented in the literature, it is striking to find that it is considered an exception, “due to an anomalous cancellation of relativistic effects,” of the differences between 5th- and 6th-row elements of the same group (30, 31). In the current study we found that out of the 17 possible pairs of 5th- and 6th-row elements belonging to a group, there are other five pairs sharing similarities: {Nb,Ta}, {Mo,W}, {Tc,Re}, {Ru,Os} and {Rh,Ir}.' But if most of the 5d elements are showing it, it's hardly exceptional behaviour! Indeed it's more exceptional if it doesn't happen! The Ln contraction is less visible for groups 10 through 12 but still can be seen in atomic radii; in that case, truly everyone in the 5d stretch is showing it, except for La if we insist on shoving it in; doesn't that make it more obviously out of place? Here we have an entire paragraph in italics, so to speak, and what is unusual is the roman type (just like how we indeed use roman type to emphasise something in an all-italics passage).
Yes, Lu makes a good heavy Ln. La also makes a good light Ln, and nobody here is saying that La or Lu is not a lanthanide. They obviously both are, and that is completely irrelevant for the group 3 question because it is so inconclusive. What we are trying to do instead is to determine which one is the d-block lanthanide, as we know that there is one. And to do that we compare each of La and Lu with the elements that we know are supposed to be 5d metals, not the lanthanides. So: which distance is closer, La to Hf-Hg, or Lu to Hf-Hg? It's really obvious that the answer is Lu, Restrepo not withstanding. When he opines that his work favours La under Y on the grounds of greater resemblance to transition metals, he seems to have overlooked that once you are considering only stoichiometry, you can never notice similarities between elements with different valences (which is why all the famous diagonal relationships fail to show up on his chart), and so you cannot usefully compare the pairs {La vs. Hf-Hg} and {Lu vs. Hf-Hg} using that methodology. So of course the only transition metals that he gets as similar to La are Sc and Y (no surprise at all, as they are also REMs), and judging from the cyan colour all those REMs have in the background the distances are all negligible for Lu too (as everyone could have guessed from atomic radii and reading Greenwood and Earnshaw). Double sharp (talk) 13:34, 12 January 2020 (UTC)

@Double sharp: I think you are too focussed on the 18-column table, in which there is some abstraction of detail caused by the footnoting of the Ln/An. In a 32-column table the Ln contraction is smoothly and precisely seen in the interposing f-block, between group 3 and group 4 i.e. from Ce to Lu. The knock-on consequences are then felt starting from Hf, which is a visually more accurate presentation, rather than mushing the end of the contraction and the start of its consequences in the 5th row of the d-block.

I agree with Restrepo and am not familiar with Zr/Hf being viewed as an "isolated" anomaly.

He compared La and and Lu to the rest of the Ln largely on the basis of the trivalency of all the elements involved. There was no valence mismatch for La and Lu since all the elements involved were rare earths. As noted in the Periodic law section, the vertical trends are more important than the horizontal trends. There is no overwhelming case for overturning the periodic law, which says that La goes below Y, and is consistent with the behaviour of group 3 as trivalent group 1 to 2 equivalents. Sandbh (talk) 05:50, 15 January 2020 (UTC)

@Sandbh: In a 32-column table the Ln contraction is going to cut across blocks anyway, because the effect is that each Ln is smaller than its predecessor because it adds an f-electron (including Ce smaller than La). So you'll have one d-block lanthanide and fourteen f-block lanthanides anyway, and it smears across both. So the relevant thing, as I have been saying for the last few days, to find which one the d-block lanthanide is to find with lanthanide is most d-block-like. Lu clearly qualifies. And as I mentioned above, the periodic law does not say "look for the next analogue" (or else Sc goes below Al), but "look for the next analogue with the same valence structure". For Y this is Lu, not La, because for La we have an empty 4f that has some relevance, but for Lu we don't have that possibility as 4f is stuck in the core. Double sharp (talk) 06:48, 15 January 2020 (UTC)

@Double sharp: Yes, I agree about the 32-column table. It goes groups 1-2 as part of the s-block; group 3 as d-block; the f-block as Ce to Lu, and then group 4 onwards. Chemically, the Ln contraction per se starts in the f-block with Ce3+ as [Xe]4f1 and finishes in the f-block with Lu3+ as [Xe]4f14. There is no smearing of the contraction across blocks. Let's recall that historically the Ln were the elements like La, and thus did not include La. We can however see why Ce to Lu are like the Ln, and understand the laziness of including La as an Ln. The structure going down group 3 is consistently [NG] d1s2. The structure of the 6d row, starting with Hf, is perturbed by the intrusion of the 4f shell, so group 4 for example vertically goes [NG] d1s2; [NG] d1s2; [NG] 4f14d1s2.

@Sandbh: The more we speak "chemically", the more it becomes silly pedantry to exclude La from all effects about atomic size that are so important for the lanthanides. Where is the difference between, say, LaIII chemistry and PrIII chemistry? (Ignoring redox.) And since when is the introduction of a 4f shell a perturbation? By that logic, it looks like every other row in the periodic table is perturbed, e.g. Ga-Kr add a d shell over Al-Ar, Tl-Rn add an f shell over In-Xe. This is not a perturbation, this is a common thing that comes straight out of the two-rows-at-a-time shape of the periodic table. It would be a real perturbation to insist that group 3, alone among the groups that start a new row in a p-, d-, or f-block, never show this. Double sharp (talk) 09:40, 16 January 2020 (UTC)

@Double sharp: It isn't pedantry IMO. It's about clarity, cognitive congruence, and accuracy as to cause and effect. The s and p blocks are fairly harmonious. The d-block and the f-block mess up the sp tea party, with their anomalous configurations, and this gets worse the further down the table we go. God knows what sort of mess the g-block, should we ever get there, will make. An Lu table obscures the perturbation by starting the f block at La. It's about showing the table as it really is rather than how we'd like it to be (I don't think that applies to you, but it certainly applies to LST aficionados, where the push for Lu in group 3 is coming from). Sandbh (talk) 06:24, 17 January 2020 (UTC)

@Double sharp: On the d and f block as perturbations see here.

@Sandbh: To me it is rather the anomalous configurations that are the perturbations. I mean, look at group 5: niobium is d4s1, tantalum is d3s2. Does it make much difference to their chemistry? Honestly no, they are so similar. Lutetium is ds2, lawrencium is s2p. So what? Can't we just mentally call it just the group with five s and d valence electrons (mostly d in compounds), never mind where they are? The important thing is just:
  1. In the same column, the number of valence electrons is supposed to stay constant.
  2. In the same column, the types of valence electrons is supposed to stay constant whenever it is possible. (You can think of helium as a degenerate case with no valence electrons, if you don't want to move it to go on top of beryllium.) What this means is that the available subshells for hybridisation should not change when you can help it: in the d-block they should be dsp, in the f-block they should be fdsp, in the g-block they should be gfdsp. There is a minor difficulty that arises for group 2 (which has some incipient transition behaviour), but you cannot help that.
  3. The cores change regularly down each column. They go to the next noble gas configuration, plus filling the subshells that have been inserted in later rows. So we expect the core in the 3p elements to be [Ar], but in the 4p ones to be [Kr]4d10. In other words: OK, inserting the d- and f- (later g-) blocks are a perturbation, but a completely regular and predictable perturbation, so let's show that.
Once you put it that way the anomalous configurations don't really mean anything. Which makes sense, because they certainly don't seem to mean anything for chemistry! Why do we have to show the second-order perturbation of delayed collapse in the periodic table, then, when it certainly doesn't seem to lessen the big divide between La and every other 5d element? We're not suggesting to move Lr into the p-block, so why must La move into the d-block for the same reason? (And if you show the perturbations in the early g-block you'll never end up drawing anything at all, judging by the predictions where 8p, 7d, 6f, and 5g start filling at E121, E122, E123, and E125 respectively.) Double sharp (talk) 09:58, 19 January 2020 (UTC)

Relevance of anomalous ground-state configurations

@Double sharp: Anomalous configurations can make a difference to chemistry.

Take Group 11, for example, and their schizoid transition metal/main group chemistry that arises due to their anomalous configurations.

A rather doubtful statement in my opinion, since Rg should have a 6d97s2 ground state and yet be a good heavier homologue of Au. Ag has a d9s2 state at 3.7495 eV (NIST tables) again, and for Cu and Au they are even lower down. All well within chemical bond energy range. Double sharp (talk) 16:28, 3 February 2020 (UTC)
@Double sharp: Rg = fog. Cu, Ag, Au should have the configurations 3d9. In fact they are all 3d10 4s1, with Ag having an s differenting electron, rather than a d differenting electron. The result is that +1 is the most stable and common oxidation state for Ag in which it acts predominately like a main-group metal. This is a text book example of the chemical significance of a d/e. In Cu the d/e is a d electron. This has a significant impact on the electron configuration of and chemistry of Cu, since the preceding element Ni, is 3d84s2. The addition of a d electron to Ni results in 3d94s2. Energetically, it's more efficient to have a 3d10 sub-shell, so one of the two s electrons is shifted so that in the end the configuration of Cu is 3d104s1. So, rather than Cu being 3d9, and having only transition metal chemistry, it can also show main-group chemistry with the loss of its lone s electron. Here's another example of where a d/e has a significant impact on the chemistry of an element over and above what the n+1 rule tells us. Sandbh (talk) 04:17, 11 February 2020 (UTC)
@Sandbh: I don't see how Rg is fog here. Yes, it's severely unstable, but it is an element like any other that we put in the periodic table on the grounds of its predicted chemical properties. Nuclear properties are irrelevant for element placement in the PT. The fact that the chemistry of Rg should be close to that of Au, despite the configuration difference, suggests that the whole thing is rather weak here. The DE for Mn and Tc are s-electrons, whereas for Re and Bh they are d-electrons, but I do not see how Re and Bh are less main-group-like than Mn and Tc. Indeed it's more or less the opposite way round.
The differing behaviours of group 11 metals are simply explained from the relative energy differences between (n-1)d and ns (the chemically active valence subshells) at the end of the d-block. In Cu the gap is not so big, we see +1 and +2 states frequently. In Ag it is bigger, in Au it is smaller again because of relativistic destabilisation of 5d. That's what controls the ionisation energies and hence common oxidation states, not the differentiating electrons. Double sharp (talk) 12:14, 11 February 2020 (UTC)
@Double sharp: Well, my focus is not on the explanation of the differing behaviours of group 11 metals. My focus is at a higher, d/e anomaly level bearing in mind you were looking for examples of where the d/e make a difference. It's like the n+l rule, which is admired for its regularity, just by itself. Never mind the underlying explanation. Sandbh (talk) 23:07, 11 February 2020 (UTC)
@Sandbh: This is an astonishing statement. We need to look for the underlying explanation because that will lead us to the most basic possible basis for the PT. Only then will our focus be as high as the situation warrants. In every example you give of the DE's supposedly making a difference, I've demonstrated that the DE are not the cause of that difference. Either because the difference appears when the DE anomaly isn't present, or because the DE anomaly predicts a difference where there is not one. (In group 7 it even predicts exactly the opposite of reality.) Instead, in every case the cause is the energy levels of, you guessed it, chemically active valence subshells. That by all rights ought to be decisive. Just like Mendeleev's use of atomic weight has long been overthrown by atomic number. Double sharp (talk) 23:17, 11 February 2020 (UTC)
@Double sharp: I agree with Scerri who believes there is great merit in taking as philosophical, and as abstract as possible, an approach to the periodic table. It's not necessary to drill down. In group 11, for example, their dualistic behaviour is a byproduct of their anomalous configurations or d/e's i.e. d10s1 as opposed to their expected configurations of d9.
@Sandbh: I've already debunked that reason. Where is the dualistic behaviour of Au? Why are there magically no repercussions on Mn/Tc having the s differentiating electron instead of the d one of Re/Bh? I say there is great merit in a philosophical and abstract approach if and only if it is also a relevant approach. Double sharp (talk) 23:54, 11 February 2020 (UTC)
@Double sharp: Eh? What kind of chemistry do you think +1 compounds of Au show?
@Sandbh: OK, I did not say what I meant to say, sorry. :) What I'm trying to get at is that Au has +3 as the dominant state, so it's not exactly an even split like Cu, or dominant main-group-like chemistry like Ag. For Au it's the main-group state that is uncharacteristic, and you cannot predict this from DE's. I don't disagree that Au(I) has main-group character but so does any TM oxidation state that reaches d0 or d10. The group 3 to 7 elements in their group oxidation states are basically like main-group elements since they reach d0, but where are the DE's to prove it? And Rg(I) is predicted to be a good heavy homologue of Au(I), forming the homologous cyanide complex; so it should be "main-group-like". Except, a free Rg+ ion is 6d87s2 (p. 1672). (Chemical environments should make it 6d10 due to 6d destabilisation, another case where we have to look at chemical environments rather than ground-state free gaseous atoms.)
You can't claim P is the big cause of Q if we very often have P and not-Q together, or not-P and yet Q. The important thing here is just: has the number of valence electrons dropped below 10 yet, or not? (Since in compounds the TM's are usually dn.) That is important because you need a partially filled d-subshell to show TM properties (of course, DE's are completely silent about the later start of that in each period, whereas chemically active valence subshells explains it with aplomb.) And that happens even if group 11 was d9s2, as we see in the case of Rg. Double sharp (talk) 07:50, 12 February 2020 (UTC)
@Double sharp: No matter. It's good to have an educated squabble.
Now then, here's example of you attempting to undermine one of my arguments, on the basis of an erroneous assumption. In this case you seek to do so by downplaying the significance of the +1 OS of Au. Never mind that our article on gold says, "The oxidation state of gold in its compounds ranges from −1 to +5, but Au(I) and Au(III) dominate its chemistry." It's the s d/e that enables this dualistic nature.
Steele 1966, The chemistry of the metallic elements, p. 67:
"The overlap in properties between the b-subgroup metals and the transition metals is shown in the properties of copper, silver, and gold. In their monovalent compounds they are typical b-subgroup elements. The d-electrons can, however, be used in bond formation to give compounds of the elements in the divalent and trivalent states. In these compounds the d-subshell is incomplete and their chemistry is typical of transition metal compounds."

Sandbh (talk) 01:03, 13 February 2020 (UTC)

Straight from Post-transition metal: "The chemistry of gold is dominated by its +3 valence state". And your last quote exactly supports my point. The reason for this change between B-subgroupness and transition-metal-ness in group 11 is not because of DE's. What matters is if the d-shell is incomplete or not. Just like what matter for the early transition metals is where the d-shell is occupied or not. Double sharp (talk) 19:31, 13 February 2020 (UTC)

The absence of an f electron in La means that its trivalent compounds don’t have any magnetic moment, unlike Ce to Yb. The accelerated completion of the 4f shell, over thirteen elements, results in Lu compounds having 0 magnetic moment.

That has nothing to do with La's configuration. Even if La was [Xe]4f16s2 in the ground state, the 4f electron would be ionised away in La3+ anyway. Double sharp (talk) 16:24, 3 February 2020 (UTC)
@Double sharp: You're right, my response was poorly worded. Who knows what else I may have been trying to do at the time. I will say that the absence of an f electron in La means that its divalent compounds don’t have any magnetic moment, unlike Ce to Tm. Sandbh (talk) 04:17, 11 February 2020 (UTC)

Cerium(IV) compounds are finely balanced. The energy of the 4f electron is nearly the same as that of the outer 5d and 6s electrons that are delocalized in the metallic state, and only a small amount of energy is required to change the relative occupancy of these electronic levels. The 4f electron in cerocene, Ce(C8H8)2, is poised ambiguously between being localized and delocalized and this compound is considered intermediate-valent.

So the important thing is the energy levels of the subshells, not the 4f15d16s2 configuration that happens to be barely the ground state, with 4f15d26s1 0.2937 eV above, and 4f26s2 0.5905 eV above, according to NIST. You said it yourself, only a small amount of energy is required to change the relative occupancy of these electronic levels, and chemical bonds are enough to provide that. Lanthanum has a 4f16s2 state at 1.8842 eV, which is well within the normal range of chemical bond energies (1 to 10 eV). The ground state for f- and d-block elements is often quite close to many excited states, that all contribute. (Why do you think the transition metals in group 3 through 10 are mostly dns0 in compounds?) Double sharp (talk) 16:24, 3 February 2020 (UTC)
@Double sharp: Ce is peculiar since we expect its d/e (per n+l) to be a second f electron, to give it f26s2 and probably a stable +4 oxidation state, analogous to Th (I don't know what the condensed configuration of Ce would be, in this case). Instead, due to the delayed start of filling of the 4f sub-shell, the f d/e in Ce is only its first 4f electron, resulting in a configuration of 4f15d16s2, and a stable +3 oxidation state, and a +4 oxidation state. Sandbh (talk) 04:17, 11 February 2020 (UTC)
@Sandbh: By that logic, consider Pa. Due to the delayed start of filling in 5f, two 5f electrons appear for the first time here, giving [Rn]5f26d17s2. Your analysis of Ce would lead you to predict a stable +3 state for Pa, but Pa(III) is an absolutely unstable rarity. The stable oxidation state is +5 (the "group" one), with marginally important +4, which is not at all what you would predict. So it seems that ground-state configuration analysis is not being consistently helpful here. Double sharp (talk) 12:07, 11 February 2020 (UTC)

Gd has an f7d1s2 configuration rather than the expected f8s2 configuration, and this has major implications for its role as the central metal of the Ln, rather than Eu.

That's just because Gd3+ is f7. Which it would be even if the neutral configuration was f8s2. Double sharp (talk) 16:28, 3 February 2020 (UTC)
@Double sharp: If Gd really was f8s2 then perhaps it would also have +2 oxidation state of comparable stability to +2 Eu and Yb. Sandbh (talk) 04:17, 11 February 2020 (UTC)
@Sandbh: And this one is a perfect example of why DE's are mostly chemically irrelevant: the majority of the Ln are fns2 (Ln = Pr, Nd, Pm, Sm, Eu, Tb, Dy, Ho, Er, Tm, Yb), but their predominant oxidation state is +3 instead of +2. (Note that Tb has a well-defined +4 state.) The important thing controlling stability of +2 and +4 states is how close the result is to a stable half-full f7 or f14 configuration. The DE never makes any difference here. Sm, Eu, Tm, and Yb form well-defined +2 states that are not too uncommon because Eu2+ and Yb2+ land right on the half- and fully-filled marks, and Sm2+ and Tm2+ are close and get horseshoes-and-kisses benefits. Tb, which is [Xe]4f96s2, prefers to form a +3 and sometimes +4 state: +2 is a rarity (though well-defined), but combustion of Tb in O2 produces a mixed Tb(III,IV) oxide. The DE clearly doesn't have anything to do with the existence of Tb4+, that's because doing so lets it reach 4f7. As it would even if Tb happened to be 4f85d16s2 or 4f75d26s2 in the ground state instead. Double sharp (talk) 12:02, 11 February 2020 (UTC)

Among the light actinides the difference in energy between the 6d and 5f orbitals is very small and results in competing 5fn7s2 and 5fn-17s26d1 configurations in molecular bonding, contributing to some of the complex chemistry of these actinides.

Yet more proof of the approach of looking at chemically active subshells rather than individual configurations, like I do! ^_^ Double sharp (talk) 16:28, 3 February 2020 (UTC)
@Double sharp: No need for that. We can just be amazed at the similarities between the light An and the early 5d metals, caused by the anomalous configurations under consideration. Sandbh (talk) 04:17, 11 February 2020 (UTC)
@Sandbh: It has nothing to do with the anomalous configurations, which do not match from Ta/Pa onwards. From Pa onwards, the differentiating electrons are f-electrons, and yet the similarities Ta/Pa and W/U still go on fairly strongly. Meanwhile the transition to a "uranide"-like situation happens for {U, Np, Pu}, even though the change from fnds2 to fn+1s2 only happens at Pu. I don't see any way you can predict this from the DE's, which are changing totally orthogonally to the chemistry here. Double sharp (talk) 14:10, 11 February 2020 (UTC)
@Double sharp: I was referring to the anomalous retention of a 6d electron, up to Np. Sandbh (talk) 23:10, 11 February 2020 (UTC)
@Sandbh: Which I refuted in the third sentence. The little problem is that resemblances to the "right" 5d metal stop at uranium, because Np and Pu prefer +6 in which they imitate uranium, rather than going on to +7 and +8. And strong resemblances to 5d chemistry stop at plutonium (Am is unhappy to be oxidised far). And if you want any resemblance at all, it goes to curium with its +6 state, at which we have included Pu and Am which have no anomalous d-electron. In every case the break does not coincide with that of the anomalous DE's "up to Np". Double sharp (talk) 23:24, 11 February 2020 (UTC)

Lr has to go into the d-block as there is no other place it can practically go.

So what happened to being interested in Nature as She was, rather than how we can most easily draw Her? ^_^ Double sharp (talk) 16:28, 3 February 2020 (UTC)
@Double sharp: Pragmatism. Just like what happens in an Lu table, with a p metal in the 5d row. That said, I don't even have to worry about the anomalous configurations per se. I can just compare the n+l rule, with the La table and the Lu table, and note that the Lu table has one more anomalous configuration. Sandbh (talk) 04:17, 11 February 2020 (UTC)
@Sandbh: Lr is not a p-metal, by my and Jensen's criteria. It has a chemically active valence 6d subshell, that's what counts. And it has three valence electrons that can occupy 6d, 7s, and 7p but not 5f; that's why it's in group 3. We consider the exact DE and ground-state configurations to be irrelevant and only consider chemically active subshells and how many electrons are in them. The second point means that even group 2 is not an exception, because the Zn group have twelve valence electrons in d+s+p, but the Ca group have two, just like Be and Mg. (For sanity's sake, given prominent sp hybridisation in things like Be, it seems reasonable to consider the s and p outer shells together as part of a normal valence octet Actually, I am not even sure it has any exceptions once you go down to this second step until the really strong superheavies (Mc, Lv, Ts, and Og may be a bit problematic with premature 8s involvement thanks to the lower-than-usual 7p3/2-8s gap, but 7p is still the highest angular-momentum subshell).
You can see how this works in practice at User:Double sharp/Idealised electron configurations: the fuzzy approach works pretty much perfectly in period 7 and even holds the fort in period 8. Double sharp (talk) 14:13, 11 February 2020 (UTC)

If we set aside the anomalies and look at (as I’ve done before) the configurations of ions in precipitation chemistry and the solid-state chemistry of salts—since such ions don’t have anomalous configurations—this is what we get:

Cations have an s0 configuration

Cations are s0 (or s2)

Cations are group number less charge, which gives a range of d0 to d10

With La starting, cations are f0 to f13 With Ce starting, are f0 to f14.

→ The f-block starting with La introduces a new irregularity, at the global level.

→ Another irregularity arises wrt to the boride, scandide, and lanthanide contractions. Each contraction starts with the first appearance of a p, d or f electron at the start of the applicable block. This regularity is not observed in an Lu table. Sandbh (talk) 03:05, 21 January 2020 (UTC)

The important thing is the energy levels; anomalous-looking configurations are just a symptom of many shells being close together, e.g. 4d/5s in early 4d metals, 4f/5d/6s in Ce, 5f/6d/7s in the early actinides. It does not much matter which anomalous configuration it happens to be. The characteristic configuration for a d-block element in compounds is dns0, so here it is indeed group 11 and 12 that are the weird ones. But it's not because they all happen to be d10s1 in the ground state so much as that they cannot possibly stuff eleven electrons into the d-subshell, so this is something like Lu being unable to be f15. Ag is predominantly main-group, Cu is pretty balanced, and Au is strongly transition, but that is a function of the differing 3d/4s, 4d/5s, and 5d/6s gaps because of radial nodes (3d) and relativistic effects (5d). Secondly: for La3+, whatever anomalous electron there had been has been ionised anyway as expected, so there is no difference between it and Al3+ or Sc3+. Thirdly, no one is claiming that Eu is the central metal of the Ln, because no one is disputing that Lu is a lanthanide chemically. (I just think it is the d-block lanthanide.) And fourthly, you're forgetting Yb2+ (f14), so there is no difference with the f-block either way. And you are going to get that irregularity anyway from the actinide contraction as Th lacks that 5f electron in the ground state, which is why I prefer to start looking when the subshells get active even if they are unoccupied. So we can include La through Lu as well as include Ca through Zn for such contractions. Double sharp (talk) 13:55, 21 January 2020 (UTC)
@Double sharp: Too hard. D/e's are so much easier. Sandbh (talk) 04:17, 11 February 2020 (UTC)
@Sandbh: I'll take what is relevant over what is easy any time, thank you very much. The easiest possible way to arrange the periodic table is to just list the elements in alphabetical order, but that will get us precisely nowhere. Double sharp (talk) 14:15, 11 February 2020 (UTC)

@Double sharp: With the f block as Ce to Lu, the range is 0 to 14. With La to Yb the range is 0 to 14, as you say, my oversight. I would not count La as an f-block metal, since the 4f subshell has not started filling yet. So the second option is not comparing like with like. Sandbh (talk) 09:53, 23 January 2020 (UTC)

So Th is not an f-block metal either, by that logic. Both La and Th have low-lying f-orbitals that seem to have chemical consequences (cubic La complexes, extremely high coordination numbers for Th), unlike the completely core-like f-orbitals of Lu that just act like the d-orbitals of Ga, but unfortunately neither of them happens to have an f-electron in the ground state. I prefer to say that this is just a simple consequence of high atomic number causing delayed collapse. That takes care of what happens at La, Ac, Lr, and E121 all in exactly the same way, without having to bang our heads against the wall at the craziness that is expected to start brewing once we synthesise three more elements. Double sharp (talk) 11:38, 23 January 2020 (UTC)
@Double sharp: Th is an f-block metal by way of Ce starting the f-block. The sky won't fall down. Sandbh (talk) 04:17, 11 February 2020 (UTC)
@Sandbh: So you're still assuming symmetry to force a block to begin in a vertical column, even if your first criterion of DE's says it doesn't. So why is this symmetry inviolable when you seem to consider it fine to violate the symmetry of rectangular blocks? Double sharp (talk) 14:15, 11 February 2020 (UTC)
@Double sharp: I'm assuming regularity rather than symmetry. Our IUPAC criterion says a block starts upon the first appearance of the applicable electron. The f block starts with Ce, that is all. That Th does not have an f electron is not relevant in this context. Sandbh (talk) 23:18, 11 February 2020 (UTC)
@Sandbh: Different word, same idea. The symmetry of the LSPT is just as much a "regularity"/"symmetry" as your criterion (I have since rejected it) as yours is, because the whole point of symmetry is that you get some information for free due to the invariance. To his credit, Prof Poliakoff (who you quote a lot re Nature and drawing Her) realised that this was also an issue, and pointed it out, to quote you:
And yet now you reject that possibility on regularity/symmetry grounds! If you wanted to be consistent about this and exclude La because of its lack of an f-electron as a ground-state gas-phase atom, you need to start the 5f block at Pa. Otherwise you are invoking symmetry in one case and disallowing it in another. ;) Double sharp (talk) 23:22, 11 February 2020 (UTC)
@Double sharp: Per our IUPAC submission, a block starts on the first appearance of the applicable electron. Subsequent rows fall into place as an outcome of the aufbau process. That is all. Regularity is not the same as symmetry. Sandbh (talk) 23:43, 11 February 2020 (UTC)
@Sandbh: Yes it is. A regular form is more symmetric than an irregular form. Double sharp (talk) 23:55, 11 February 2020 (UTC)
@Double sharp: That's not what I said. The bifurcation of the lungs in humans is quite regular, but not symmetric. To accomodate the heart, the left lung has 45% of the total lung volume, whereas the right side has 55%. Ditto, that humans have two hands is a regular pattern but only 11% or so are left handed. The aufbau sequence is not symmetric but it does show regularity in continually (evenutally) returning back to the n+l rule, after each anomaly. Regularity and symmetry are closely related, but not identical. Sandbh (talk) 03:03, 12 February 2020 (UTC)
PS: A square is regular and symmetrical. A rectangle is irregular and symmetrical. Sandbh (talk) 03:10, 12 February 2020 (UTC)
Anyone can see that that is because a rectangle has less symmetry than a square (e.g. it lacks the 90-degree rotation or reflection in a diagonal). But you turn symmetry into "all or nothing". Double sharp (talk) 07:58, 12 February 2020 (UTC)

Organometallic chemistry

@Double sharp: We know compounds of Ti generally are mainly covalent in nature since its most stable oxidation state is +4. Ditto Zr and Hf. As we wrote in our IUPAC submission:

"Sc-Y-La-Ac further shows simple trends of increasing basicity and in this respect are much more like their leftward neighbours Ca-Sr-Ba than their rightward neighbours Ti-Zr-Hf. The +4 state is too high to be ionic in group 4, even for Th and what little we know of Rf. While KCl, CaCl2, and ScCl3 are ionic compounds, the group 4 tetrahalides (Cl, Br, I) are volatile covalent liquids or solids. While one can obtain aquated La3+ (or Lu3+), hydrolysis proceeds so far for Hf that HfO2+ is obtained instead. Lanthanum in particular is such a hard base that it is taken up by the body as if it were calcium."

…I have this picture in my mind of you trying to beat up your IUPAC-submission-pro-La self… :)

'Nothing more refreshing than a change of opinions, or as a friend of mine says: "I take back everything I said and claim the opposite."'Tim Krabbé. ^_^ Double sharp (talk) 12:53, 23 January 2020 (UTC)
@Double sharp: Do you realise that your arguments in the IUPAC submission were much stronger and more direct than the arguments you’ve raised in this thread? Sandbh (talk) 07:46, 25 January 2020 (UTC)
Here are three quotes from the IUPAC submission that I actually still agree with:
"In advancing their positions, we think these authors fail to demonstrate why similarity in properties (aside from valency) necessarily connotes group membership. In some other parts of the periodic table, most germanely in groups 1 and 2, we see a continuation of trends upon descending a group (such as increasing atomic radius, basicity and electropositivity), rather than a convergence of such properties. Either pattern could apply to the eka-yttrium position. That is to say, if group 3 were treated as early main group elements we would expect more of a linear trend going down the group. On the other hand, if group 3 was treated as a transition metal group, it would be reasonable to expect more of a convergence properties on going from period 5 to period 6, as a result of the lanthanide contraction."
"We think it plausible that the low-lying 4f levels in La may influence some of its properties. It is also conceivable that the filled 4f shell of Lu may influence some its properties but, if so, the scope of this influence is likely to be smaller and more obscure. Overall, we think the presence of any 4f influence, as a relatively low-order phenomenon, would only be a "tipping point" argument. That is to say, if the merits of -La-Ac and -Lu-Lr are otherwise similar in terms of which one is placed under Y a case could then be made for -Lu-Lr."
"Furthermore, the historical record does not necessarily flag ongoing relevance. As noted, many old tables from this time period also place Be and Mg in group 12 with Zn, Cd, and Hg, a practice which has since been deprecated." That argument was used against the historical theme of promoting Lu under Y, but it works just as well against the huge historical record of La under Y.
The important thing is that I've come round to the position that the merits of La and Lu under Y are actually pretty similar, all things considered. Therefore the "tipping point" argument works, because it is essentially "let's keep the status quo/null hypothesis" for Lu, which fits the pattern of the first 5 rows better. You seem to be taking La as your null hypothesis instead, when it seems to me that it should be an alternative hypothesis that the pattern breaks. That's why we both find the arguments for the other one not strong enough and keep on sticking to our respective sides now.
You see, the reason why all those "tipping point" arguments were rejected was "let's look at group 2". But now I think Droog Andrey had a point in Deep web talk:WikiProject Elements/Archive 33#Part 2: the s-block is in some sense fundamentally different from the other blocks. Look down any other column of the periodic table and you get double periodicity out of the n+l rule. B and Al are just s2p; Ga and In add d10; Tl and Nh add f14 on top of that. Ti and Zr are just d2s2; Hf and Rf add f14 on top of it. So why should group 3 follow the s-block's singular pattern when it is a clearly a d-block group? If we allow this border to stand, we can start raising uncomfortable questions like Ti-Zr-Ce-Th (all are d2s2 but Ce, which could work as an anomaly like Lr with a subshell getting active "too early"), and B-Al-Sc (which even has many sound chemical arguments for placing Al above Sc like Rayner-Canham's table does!). And it doesn't really work as well as we thought to say that group 3 mostly follows group 1 and 2 in its chemistry, because the heavier members of groups 4 and 5 also do that.
And if you look at where the d-block ends, it becomes clear that our argument doesn't really work: the energy of the d-electrons rises from Zn going downwards, but the energy of the f-electrons falls from Lu going downwards. So the only problem is delayed collapses, but that is totally normal as elements get heavier, and as the elements are ionised the problem mostly corrects itself. But once you have refuted delayed collapses, kinship ties supposedly stronger to the left than to the right (when it turns out to be equal), and condensed-phase configurations, what is left to support Sc-Y-La with its strange s-block trend outside the s-block? So that's my current argument: H0 is Sc-Y-Lu, and there is not enough force from Sc-Y-La arguments to reject it. Maybe this is unsatisfying, since failing to reject the null hypothesis is hardly an interesting story. Maybe it feels like an argument from despair that Nature is too complicated and plumping for the simplest option. But in the long run, it seems to me scientifically more honest to admit that there is not enough basis for a revolution, and that we erred as a whole in creating one in the first place back when La was first put under Y.
@Droog Andrey: Do you have the results of your test from Deep web talk:WikiProject Elements/Archive 38#4f involvement from La through Lu that you suggested (bond angle in LnF2+)?
@Double sharp: I've computed about a half of these cations, and the results appeared to be quite dependent on the level of theory. Unfortunately, available CPU power doesn't allow me to use as high level of theory as CASPT2. Droog Andrey (talk) 11:19, 27 January 2020 (UTC)
I honestly don't plan on responding much more over the next few days in this much detail. Firstly, I have some important stuff IRL to do, and secondly, it seems we are going in circles. My whole case for Sc-Y-Lu is more or less set out in this post, modulo the experimental results on 4f involvement in the Ln that haven't been posted at the moment, so you can more or less just read this, and refer back to the other posts for ancillary arguments like Lu creating a more homogeneous d-block.
P.S. I love how a thread about group 3 and classification was enough to get this talk page all the way up from 11K to 200K without any articles being affected. So, everything is normal again! ^_^ Double sharp (talk) 10:47, 25 January 2020 (UTC)
@Double sharp:
“And it doesn't really work as well as we thought to say that group 3 mostly follows group 1 and 2 in its chemistry, because the heavier members of groups 4 and 5 also do that.”
Speaking calmly, this is the line that beggars belief. Effectively the entire chemistry world recognises that group 3 mostly follows groups 1 and 2 in their ionic chemistry, whereas groups 4 and 5 are predominately covalent. Yes the s block is fundamentally different from the other blocks, it being the only one whose electrons are nearly always involved in chemistry, but for Pd. That said, its chemistry is predominately ionic, as is the case for the rare earths. That’s another fact recognised by the entire chemistry world. The double periodicity you are referring to derives from the n + l rule which is only an approximation, has no first principles basis, and starts cracking in the d and f blocks (albeit persisting in any event). As I said elsewhere, on the basis of the most common stable oxidation state in each row of the d-block, the d block does not exhibit double periodicity. The n + l rule is a nice idealisation that does not strictly apply in the real world in ambient conditions. Droog Andrey’s 3rd IP argument doesn’t stand up compared to the most common most stable oxidation state, per DIM, in each row of the d block. I admit that one stumped me for a while but now that I know so much more about the misleading n + l rule, as well as thinking about the relevance of the 3rd IPs v the most common most stable oxidation state, I can see that the 3IP argument doesn’t hold up. Never mind about your RL; take your time and respond when you can. I intend to rule a line under this thread at the end of January, unless R8R has any more thoughts. Sandbh (talk) 09:02, 26 January 2020 (UTC)
Let me repeat one of my comments below: "Look, there is no such thing as a complete volte-face from ionic to covalent. It depends on what the counter-anion is. We go from ionic to metallic (which you're overlooking completely) across the series NaCl, Na2O, Na2S, Na3P, Na3As, Na3Sb, Na3Bi, Na. And that's in group 1, with an example taken straight from Greenwood & Earnshaw p. 81. Ionic vs. covalent is (1) gradual, from covalent to polar covalent to mildly ionic to strongly ionic, depending on the difference of electronegativity between the two elements; (2) not a complete dichotomy, because you overlook metallic bonding; and (3) not split by elements, but rather by electronegativity differences, which have a lot to do with oxidation state (just compare uranium chemistry as we jack the oxidation state up from +3 to +4 to +5 to +6). So I'm astonished to see you put so much weight on "ionic vs. covalent" as a false dichotomy. (And as I keep saying, the general thing across the periodic table is continuity, and the sharp dichotomies you like to point to have a distinct tendency to not exist.) The whole literature, when it sees fit to split "main group" from "transition", universally uses other criteria like "variable oxidation states", "coloured compounds from d–d transitions", "a wide variety of complexes", "formation of paramagnetic compounds". Ionic vs. covalent has nothing to do with it, and nobody uses that as a criterion in the literature for this divide. Respectfully, I put it to you that Zr, Hf, Nb, and Ta by this standards have weak credentials as transition metals, just as weak as Sc for that matter."
By those standards, the typical high school trend of first ionisation energies across main-group elements must also be thrown out the window. And double periodicity must also be thrown out the window as well in the p-block, because as we go down the table the most common oxidation state keeps changing. (For oxygen it is −2, for sulfur +4 and +6, for selenium, tellurium, and polonium +4, for livermorium probably +2.) E pur si muove: it is still absolutely relevant, as anybody can see just plotting trends down every p-block group. Which seems to be a good sign that your standards are excluding a lot of things that are relevant to the question. Indeed, it seems to me that they are excluding so much that we are left with nothing to stand on. What is the point of a philosophical argument if it ignores the properties that Mendeleev was looking at in the first place? Double sharp (talk) 16:51, 26 January 2020 (UTC)

Whereas compounds of the rare earths are mainly ionic.

Drilling down into organometallic chemistry, nothing changes.

You wrote:

"Group 4 organometallic chemistry is mostly cyclopentadienyl derivatives, similar to group 3. Zr and Hf in group 4 are very unhappy to show lower oxidation states, like Sc in group 3."

Now, the cyclopentadienyl derivatives of group 3 are ionic in nature, consistent with the largely ionic chemistry of groups 1 and 2. Same goes for the cyclopentadienyl actinide (III) derivatives.

OTOH the derivatives of group 4 are more covalent in nature as is the case for cyclopentadienyl actinide (IV) derivatives.

The organometallic distinction between main groups 1 to 3 (noting groups 1 and 2 are main group, and group 3 is transition), and group 4 is thus clear.

On the above basis, I submit that the organometallic argument supports La in group 3. Sandbh (talk) 23:36, 17 January 2020 (UTC)

This is just another example of the difference between +3 and +4 states: the former can be supported as water-soluble real cations, the latter are too highly charged for it to happen even for such electropositive metals as Th4+ and U4+. Similarly, the former are more ionic than the latter, by Fajans' rules. Then I have to ask, how ionic are organometallic derivatives of Ti, Zr, Hf in states lower than +4? (At least for Ti, +3 is a characteristic state, after all.) Because if the same shift happens when you oxidise uranium (for example) from +3 (more ionic) to +4 (less ionic), then it seems to me to be a difference between oxidation states, not between group 3 and group 4 per se. Double sharp (talk) 09:48, 19 January 2020 (UTC)

Ionic vs. covalent 1

@Double sharp: I agree, and it's a key difference rather than just another example. I say this because the organometallic chemistry (OMC) of states lower than +4 is not a relevant consideration since the OMC of the Group 4 elements is mainly that of the elements in their +4, covalent states. +3 is not a characteristic state of Ti; +4 is. Similarly, the OM chemistry of the Group 3 elements is mainly that in their +3 ionic oxidation state. So there is a pretty clear distinction between the two Groups, at the general chemistry level, including OMC. Sandbh (talk) 06:38, 20 January 2020 (UTC)

@Sandbh: You cannot have it both ways. If the characteristic state of Ti, and especially Zr and Hf, is +4, and we discount +3, then my point about them acting mostly like main group elements stands, and there is no significant caesura between groups 2, 3, and 4. If on the other hand we consider group 4 to be transition-metal-like on the grounds of +3 as a significant lower oxidation state for Zr and Hf, then by that logic Sc qualifies by its +2 state, and there is again no significant caesura between group 3 and 4. Double sharp (talk) 14:39, 20 January 2020 (UTC)

@Double sharp: Groups 1, 2 and 3 have a predominately ionic chemistry. Group 4 has a predominately covalent chemistry. Trends going down group 3 are like those of groups 1 and 2. Trends going down group 4 are like those going down groups 5 to 10, or so. The group 5 elements are similar in many ways to group 4, expect for being less electropositive. Those are key differences between Groups 1 to 3, and 4 to 5+. Sandbh (talk) 03:51, 21 January 2020 (UTC)

@Sandbh: That is, again, just a matter of characteristic oxidation state. Compare Sn2+ and Pb2+ vs. Sn4+ and Pb4+ (the same element, even). (Note that Mendeleev in his first table put Pb as a heavier homologue of Ba, because of the +2 state.) Also compare the chemistry of Fe (mostly ionic Fe2+ and Fe3+) with that of Ru and Os in the same group (which prefer higher states, with covalent compounds like the famous OsO4). As we can see: the 3d metals from Cr onwards (which prefer lower oxidation states like +3 and later +2) are not too dissimilar from main-group elements in this respect. And trends going down group 3 are only like those in group 1 and 2 if you've decided on Sc-Y-La already Double sharp (talk) 13:45, 21 January 2020 (UTC)

@Double sharp: I see the trends going down group 3 (as La) resemble those going down groups 1 and 2. Indeed, we know that Laing said a reasonable case could be made for La below Y on the basis of comparing the pairs Ca-Sc and Sr-Y with Ba-La.

By that logic one could compare Be-B and Mg-Al with Ca-Sc, and arrive at Sc as the heavier congener of Al, because the reason why La matches here is because it doesn't have the later block insertion that characterises Ca-Ga and Ba-Lu. Double sharp (talk) 16:11, 22 January 2020 (UTC)
@Double sharp: Your analogy is a non-starter since the differentiating electrons aren’t the same: p for Al (as with Ga); d for Sc. Sandbh (talk) 10:48, 23 January 2020 (UTC)
In the condensed phase, they are the same. So if that is a supporting argument for Sc-Y-La, it also works for B-Al-Sc. Double sharp (talk) 12:54, 23 January 2020 (UTC)
@Double sharp: Could you support this with a reference? I find it hard to believe that Al and Sc have the same condensed phase config. Sandbh (talk) 06:24, 25 January 2020 (UTC)
Scandium metal has significant p-character at the Fermi surface (14.8%). There is also a large p-character in ScS and YS (in contrast to ScO where it appears the electron is mostly s). At the very least, this shows the condensed phase configuration as rather inconclusive, as for Sc there is both some d- and some p-occupancy and so there's nothing in principle based on this argument standing in the way of B-Al-Sc. (At least, nothing more than what seems to still allow Be and Mg into the s-block with an sp configuration in the condensed phase.) Double sharp (talk) 10:27, 25 January 2020 (UTC)

This is not the case for Lu, where the trends tend to resemble those of groups 4 to 10. That is a statement that has nothing to do with whether or not I've decided on Sc-Y-La already.

We already know that a reasonable case case can be made for distinguishing groups 1 to 3 from 4+ on the basis of iconicity v covalency. The next decision is La or Y, which is rather easily decided on the basis of vertical trends.

And what about most of the 3d transition row? The characteristic oxidation states are ScIII, TiIV, VIV, CrIII, MnII, FeII and FeIII, CoII, NiII, and CuII. The majority of these prefer a +2 or a +3 state, and they're pretty electropositive, too. Same kind of thing happens in the 4d transition row from Ru onwards, so that already covers some of "groups 4 to 10". So no, I don't buy that this is a group 1 to 3 vs. group 4+ distinction. It's rather a low oxidation state vs. high oxidation state distinction, which is underscored by the fact that the same element can behave differently depending on oxidation state (consider PbII vs. PbIV). It only looks like a distinction purely because the elements in group 3 and 4 like to be in their group oxidation state, but if you circled all the electropositive low-oxidation state metals that act more or less like what you think electropositive metals should act like from school, you'd have to cover most of the 3d row already.
As well: one of the main reasons why the "main group vs. transition" bar is rarely set between groups 3 and 4 is because of the lower oxidation states of the group 4 elements. (Which especially for Zr and Hf are at about the same level as ScII, but never mind that.) But then you can't have it both ways. If you are considering +3 oxidation states as characteristic for group 4 as well as +4, then they should be more ionic in that state, and there is no caesura between groups 3 and 4. And if you are considering only +4 oxidation states as characteristic, then group 4 does not show characteristic transition metal properties and there is again no caesura between groups 3 and 4. Neither was there between groups 2 and 3, for that matter. So why not let it be intermediate, and show the simple cations but the transition-group trend as a bridge between both, an argument that I'm sure Droog Andrey used before? Double sharp (talk) 17:05, 22 January 2020 (UTC)
@Double sharp: We know Sc chemistry, and the rest of transition metal Group 3 is predominately ionic. We know the chemistry of the transition metals in groups 4 to 12 is predominately not per e.g. what happens to their electronegativity values. Sandbh (talk) 11:14, 23 January 2020 (UTC)
For something like Cu2+ you not only have all those covalent complexes, but you also have well-defined famous salts like CuSO4 and CuBr2, so clearly at least for Cr through Zn we have a significant amount of ionic chemistry as well. Same as for Sc3+, which I'm sure is more covalent in complexes and more ionic in salts. Double sharp (talk) 11:49, 23 January 2020 (UTC)
@Double sharp: There’s a difference between predominant behaviour and significant behaviour. With an electronegativity of 1.9 there’s no way, IMO, that the general chemistry of Cu could be regarded as being predominately ionic. Sandbh (talk) 05:53, 24 January 2020 (UTC)
It has a significant ionic component, that exists in many famous salts, so again we just see a continuum as we pass from left to right of ionicity giving way to covalency, with the vast majority in the middle being able to go both ways. Anyway: notice that you need lower electronegativity to be ionic with a larger charge (think of things like Sn2+, which is about the limit of a dipositive cation), so we should weight that appropriately or else the early actinides are also "predominantly covalent" despite being such active metals. Pa5+ has absolutely no chance to be a simple cation, but so what, given its electronegativity of 1.5? But: if Pa gets in as a main-group element under something like R. Bruce King's definition, then shouldn't the pseudohomologues Nb and Ta get in too? Double sharp (talk) 23:27, 24 January 2020 (UTC)

Consider what would have happened if the trends going down group 3 as Lu resembled those of groups 1 and 2 more than the case for La. The question would've been resolved about 100 years ago. Sandbh (talk) 02:12, 22 January 2020 (UTC)

@Double sharp: Pls see my earlier response re the difference between significant and predominant. Pls also see my response re the predominately ionic behaviour of group 3 and the predominately covalent nature of groups 4 and 5. There is no continuum here. Group 3 are transition metals that behave mainly like main group metals. The An are not relevant here. Sandbh (talk) 07:09, 25 January 2020 (UTC)
And my response is also going to be "please read my previous responses". See, the whole point is just the oxidation state difference, and the border will be different depending on where in the periodic table you are looking. The further you go down the periodic table, the later the shift towards covalency happens. In period 2, Be is the first to show it significantly; in period 3, Al; in periods 4 through 6, Ti/Zr/Hf; in period 7, probably Db (since Rf4+ ought to be basic given its size). The fact that the break happens to coincide here is more or less coincidental. And we again have to look at what the anion is: TiCl4 is molecular, but ZrCl4 and HfCl4 are polymeric. TiF4 is polymeric, but ZrF4 and HfF4 have an honest-to-goodness ionic structure. So we see, once again, that the relative proportion of ionic to covalent chemistry for each of these elements is not suddenly changing from 100% to 0% when we cross the group 3 to group 4 line; there is instead a continuum, as everywhere else on the periodic table. Only if you arbitrarily declare some threshold as "predominant" does this continuity degenerate into a change from positive to negative, but how Nature really is is much more subtle than that. Double sharp (talk) 10:24, 25 January 2020 (UTC)

Ionic vs. covalent 2

@Double sharp: With respect, this is another example of your inability to engage with arguments at the most general characteristic level. Because you can’t do this you put up a barrage of minutiae which does nothing to hide the elephants in the room. I cite the literature which says 3 = predominately ionic; 4 and 5 = predominately covalent. Because you can’t refute this you resort to sideshow arguments and ignore the main game. Always with respect, DIM would never have produced his table if he had taken your approach of ignoring the inconvenient broad contours and dwelling on the micro cracks. Sandbh (talk) 00:10, 26 January 2020 (UTC)
Respectfully, all you do is cite the literature, and not ask why something happens to be, or why it should be relevant. Once you do that it becomes clear that it's not about group 3, but purely a matter of oxidation state. And you yourself already showed it using the actinides. Compare uranium(III) with uranium(IV), the same element even, with 3 predominantly ionic and 4 predominantly covalent. Or just look at PbII vs. PbIV with their vastly different electronegativity values to see how much oxidation state affects these things. I simply take the approach of ignoring the micro crack that makes group 3 the coincidental borderline for periods 4 through 6, when it is earlier in previous periods and later in the next period, and just draw the broad contour of simple, rectangular blocks.
Group 3 are chemically more main-group and physically more transition, so it is obvious that it should be intermediate, and it comes as no surprise to me that the transition into transition chemistry happens slowly over a few elements. (Which is what all the literature says, noting the reluctance of Zr/Hf and Nb/Ta to enter lower oxidation states, and their main-group like behaviour, but respectfully you seem at the moment to be only interested in whatever can be teased out of the literature if it can be seen as supporting Sc-Y-La.) My principle is simply that of continuity, not catastrophism. Mendeleev, basing his table on trends, would probably have agreed. I agree that, from the perspective of supporting Lu, the mostly ionic main-group-like chemistry of group 3 is a bit inconvenient. But their transition-like physical properties and their incipient transition behaviour, just like Zr/Hf and Nb/Ta, is decisive. Double sharp (talk) 00:18, 26 January 2020 (UTC)
@Double sharp: I cite the literature, and I use it to make new arguments along the lines of what we put together in our IUPAC submission. Never forget that I started off as an Lu supporter until Scerri (also an Lu supporter) started me off on the critical thinking path and later the philosophical/global path.
I’ve addressed the metallurgy of group 3 elsewhere as well as Restrepo’s analysis of Lu being closer to a lanthanide than a transition metal and La being in between. And I’ve written elsewhere about the universally recognised ionic-covalent difference between groups 3 and 4. That’s a done deal. That’s one of the bigger aspects of chemistry—the difference between ionic and covalent—so how you could call that a bit inconvenient is beyond me. Sandbh (talk) 09:22, 26 January 2020 (UTC)
I've flipped between La and Lu before too, and I distinctly recall that at every stage of flipping I thought I had the answer. This is exactly why I think there is not enough evidence to overthrow Lu as the null hypothesis. So, for the 9001st time:
  1. Restrepo's argument has been refuted by me 9001 times on your talk page and here by now, so I'm not going to repeat it. For goodness' sake, take almost any physical or chemical property, and plot the values for La, Lu, and Hf through Hg. The value for Lu will be closer to the values for Hf through Hg so many times that, respectfully, the fact that you keep denying it beggars belief.
@Double sharp: Yes I understand what you are saying. You simply aren’t comparing like with like. There are 14 elements between La and Hf and you pretend they are not there. It’s like the basketball match in which a gorilla walks across the court and none of the audience sees it. Your logic here beggars belief, as the two us like to say to one another :) Do you understated what I’m saying here. It has nil to do the homogeneity of the 5d metals. You may as well compare Sc to Ga and ignore the intervening 9 elements. Sandbh (talk) 11:33, 28 January 2020 (UTC)
I understand what you are saying here. I just think your logic is completely backwards. You talk about comparing Sc to Ga; fine, let's try that. Suppose Sandbh Prime comes and talks to us about how B-Al-Sc is the right and completely logical periodic table choice because Al shows pre-transition character that is manifestly uncharacteristic of group 13, and B acts as an honorary metal atom. Well, the trend B-Al-Sc-Y-La is no worse than the trend Be-Mg-Ca-Sr-Ba. Sc-Y-La have d-occupancy, it is true, but we can rationalise it away like we do for the pre-d character of Ca-Sr-Ba. Now let's take B-Al-Sc as an axiom and grasp at straws to defend it. Now we can say that it has nil to do with the homogeneity of the 4p elements. It's now not relevant that Ga in every way acts more like a 4p element than Sc does, because you are ignoring the d-block insertion that has happened between them over nine intervening elements, you see. Where is the difference in logic? Not only that, I can argue that the cleft in the p-block is completely right because of the big difference between +3 and +4 aqueous cations in water: the former exist in group 13, the latter don't exist in group 14 (even Th4+ is strongly hydrolysed). Same logic as your "ionic vs. covalent" one; this one is actually even better because it's actually more well-defined, whereas "ionic vs. covalent" raises the question "with respect to what distribution of counter-anions?".
So here is my challenge for you. Pretend that you are, in fact, a supporter of H-F-Cl, He-Be-Mg, Be-Mg-Zn, B-Al-Sc, Tl-Lr, or any number of other strange choices. (Well, He-Be-Mg is not so strange, for me, but never mind that.) And try to make arguments that defend them. And I mean seriously defend them, like you are doing for Sc-Y-La. (Maybe I can do that better because I remember supporting both Sc-Y-La and Sc-Y-Lu at various points.) I bet you that your arguments for Sc-Y-La are always going to support one of those other decisions that you have decided in advance not to support. In which case they are not arguments, just rationalisations after the big decision.
It seems to me, respectfully, that you have started with Sc-Y-La as an axiom and just search for arguments to support it. Once you find anything that can be interpreted that way, it's OK. But once you find something against it, it's not OK. And once I take your Sc-Y-La arguments and note that they also support something else you don't like, such as He-Be-Mg or B-Al-Sc, suddenly they are inadmissible and the goalposts move. Indeed, an argument like "first-row anomaly" that I use to defend Sc-Y-Lu is somehow not applicable there, but you then use it yourself to attempt to demolish He-Be-Mg. (Not very successfully, I might add, given that Ne-Ar displays the same differences as F-Cl for the most part.) All I can say is: this is not science. I don't regret reopening the question back in 2016 by considering Sc-Y-La arguments. I remember R8R was impressed that some of the Sc-Y-La arguments were good, as well as some of the Sc-Y-Lu arguments. But if the result is that Sc-Y-La has for you become an axiom not to be questioned, then yes, I do regret reopening it with you, if it means that you cannot open your mind again. Double sharp (talk) 12:00, 28 January 2020 (UTC)
@Double sharp: Yes, I started with a view that an La table was a "better" form. I did that in light of Scerri's opinion that the question of the composition of Group 3 can't be resolved on the basis of physical (including spectroscopic), chemical, and electronic properties and trends. So, like Scerri, I turned to philosophical arguments. The only such arguments Scerri has presented for Lu are triads and regularity. The triad argument is circular; the regularity argument eats itself. I've addressed both of these points in my article (not well enough for regularity). It turns out there are stronger IMO philosophical arguments for La.
If something is put up against one of my arguments I look at what's been put up and respond with my views. I don't move the goal posts as far as I can recall. I do try to explain why such an argument is not relevant or inconsequential to the main question. I haven't encountered any arguments, so far, that have made me think I've missed something substantial. I've addressed the first-row argument elsewhere including your unfounded premise that it's nature is exclusively attributed to primogenic repulsion.
My overall impression is that your arguments delve into fine details. Meanwhile the 18-wheeler big rig carries on its way. It's like having a lever big enough to lift the world, while you argue about how a lever a few thousandths of an inch longer, shorter or wider would produced a fundamentally different outcome. Sandbh (talk) 02:24, 1 February 2020 (UTC)
  1. Ionic vs. covalent is, firstly, not a catastrophic change. It is a gradual one. As your electronegativity difference increases, the bond gets more and more polar and eventually ionic in character. That is why there is a difference between AlF3 and the heavier aluminium trihalides, for example. Secondly, it is not even a complete breakdown, as you are forgetting metallic bonding: consider the difference in the alkali metal pnictides as the pnictogen varies from Bi up to N. So I hope I have made it more or less clear that while the difference is fundamental, it is continuous, it is not two-way, and it does not force elements to always stay on one side of the street. Just look at lead(II) vs. lead(IV). Double sharp (talk) 10:15, 26 January 2020 (UTC)
@Double sharp: You have recognised the concept of “typical or predominating behaviour” and that’s a good start. Now, do you recognise that the group 3 metals have a predominately ionic chemistry whereas group 4 metals have a predominately covalent chemistry? No obtuse content focussing on anything other than predominant chemical behaviour will be allowed. Sandbh (talk) 11:33, 28 January 2020 (UTC)
I suspect just about anything that is not a "yes" will be considered "obtuse", so I don't know why I am trying, but: no. There is no such thing as "predominantely ionic chemistry". Not even Na shows such a thing. Just compare bonds from Na to every other element on the periodic table. It will depend on electronegativity differences, most will still not be high enough, and you will get metallic bonding. OK, you say, we should not weight all the elements equally. OK, I think we can all agree that the element forming the most compounds that we know of is carbon, so let's compare electronegativities to that. Do you see where this is going? The electronegativity difference is just not big enough to form a totally ionic bond unless you are literally one of Li-Fr or Ca-Ra. The electronegativity difference between Sc and C is only 1.19, and you'll get a polar covalent bond in organoscandium chemistry. (Hurray, now the gap is between heavy group 2 and group 3. You will notice that although that supports Sc-Y-Lu, I am consistent and refrain from using it as an argument because I think it is not a good one. Even if I happen to like its conclusion for other reasons. Respectfully, I don't see that from you at the moment.)
So saying an element's chemistry is "predominantly ionic" or "predominantly covalent" immediately raises the question "with respect to what distribution of "counter-anions?". If you focus on fluorides, you will get a different answer than if you focus on chlorides, and you will get a different answer than if you focus on oxides, nitrides, carbides, hydrides, etc. Again, it is probably going to be called "obtuse", but I insist: we can only say one element generally shows more ionic character in its bonds than another. Since we are focusing on metals, that just means it is more electropositive. Big deal. Of course anything further to the left of the table shows greater electropositivity. Respectfully, that's 1st-year school chemistry, and even there everybody knows that the trend is continuous, all the way from Cs at the lower left corner to Ne at the upper right corner, through schizophrenic metalloids like Ge or Sb in the middle. Just as it is on the large scale, so it is on the small scale between each individual group and the next. Between each individual element and the next. Between each individual isotope and the next, even (look at inductive effects for bonds to protium, deuterium, and tritium). There is no sudden gap between "predominantly yes" and "predominantly no". Just a continuum. Just like metals to nonmetals. Just like main group to transition and back. Just like class A to class B. Just like hard to soft. Just like every other criterion you can think of. Everything is continuous here. There is no such thing as a sudden break between "predominantly X" and "predominantly not X" anywhere on the periodic table for any single criterion X. Only a transition. If you want to talk about "predominant" behaviour, you have to do it on some complex of criteria, e.g. "predominant main-group behaviour", "predominant transition behaviour", "predominant f-block behaviour", "predominant metallic behaviour", "predominant noble-gas behaviour", etc. etc. etc. When you say one of those, I agree completely. When you argue that group 3 is predominantly main-group in its chemistry, and group 4 starts to show more serious transition properties, I agree with you. That is a much better argument, even if it fails because of the staggering of the entrance of transition properties in each period, with 4d and 5d metals being more sluggish than 3d metals to start it. But not arguing on ionic vs. covalent. Not by a single property alone. Every one of those will just give a continuum across all 118 (soon to be more) elements! Only a complex of them can possibly swing together and make something decisive!
Everything just depends on the electronegativity difference. So insofar as "predominantly ionic" makes any sense it just means "low electronegativity, so that the average difference is higher". In which case, as I've said before, this is just a matter of oxidation state. U(III) is more ionic, U(IV) is more covalent. (I only accept comparatives here, there is no dominance.) Tl(I) is so ionic Mendeleev put it as eka-Cs, Tl(III) is mostly covalent. Pb(II) was put by Mendeleev as eka-Ba, Pb(IV) is mostly covalent. ZrF4 is more ionic than ZrCl4 and TiF4 which are more ionic than TiCl4. Not the slightest problem here. NaF is more ionic than Na2O is more ionic than Na3N is more ionic than Na3P is more ionic than Na3As is more ionic than Na3Sb is more ionic than Na3Bi. Everything is continuous here. You cannot even talk about "predominant ionicity" for elements because it varies with their oxidation state and your chosen distribution of what they are bonded to. You cannot even think about it for groups, where the variance is even worse. For a concrete example: just look at group 2 starting at Be. Or group 1, starting at H. Or group 13!
So, instead of arguing from supposed philosophical arguments that are either hopelessly bound to human terminology and groupings (e.g. REM, for which people cannot agree on whether Sc is included anyway, despite IUPAC), or hopelessly out of touch with the rich complexity of chemistry by seeking to reduce it to one dimension, I say: look at the pattern. Look at the trends. Look at the chemistry, that we are trying to reflect. Make a table that reflects that chemically active subshells, and the number of electrons in them, are the key Fate determining an element's destiny. That brings you to the Madelung n+l rule as the simplest way to display all that information. And wonderfully, it works really well, produces homogeneous groupings that are easy to learn and understand for a beginner, easy to use as a predictive tool for an expert, and conceals a wealth of amazing ideas behind the surface. And when we need to graduate to focusing on some 2nd-order effect, you will have a rock-solid 1st-order base to start from. Forget about displaying all those 2nd-order effects at once and reducing them to one simple chart. Go for a realistic goal, and do it the best you possibly can, and never forget where you came from: the principles of periodicity in chemical and physical properties. Not some sort of bean-counting of exceptions that forget if the exceptions to the strict pattern had relevance, not putting any one isolated effect on a pedestal and ignoring how it interacts with all the others that come together to create chemical and physical properties, and certainly not forgetting humility before Nature and confusing man-made categories with natural kinds. Just an approximate pattern with astonishing predictive power as a prism to give at least an image of the richness of reality. And never forgetting that reality is so much richer than you can possibly draw. And for that reason, Sc-Y-Lu will do well exactly because it discourages a false sense of precision and understanding about the group 3 divide. End of story. Double sharp (talk) 12:00, 28 January 2020 (UTC)
@Double sharp: There is not much point continuing this particular mini-discussion if you believe there is no such thing as "predominantely ionic chemistry", in the face of what the literature says (as per the General comments section of the thread). Your observation, "never forgetting that reality is so much richer than you can possibly draw" is overly dramatic; the La table provides a nice representation of the richness involved, and has done so for many years. The "false sense of precision" you refer to is a null argument since it can be applied to either form of table. Sandbh (talk) 23:33, 31 January 2020 (UTC)
There's only a continuum going from more ionic to more covalent. If all "predominately ionic" means is "pretty low EN", as I've just unpacked it, then yes, sure, except that Zr and Hf have a lower EN than Sc and we are back to there being no significant divide yet again. And if we add to it "pretty low oxidation state", we start to see why this is a biased indicator, because the divide then appears from something that is pretty orthogonal to group assignment and block characterisation. Unless we want thallium and lead to fall out of their groups. When the literature actually explains what "predominantly ionic" actually means if anything (i.e. low EN), like I quoted Wulfsberg below, it runs into two problems counteracting the narrative of "groups 1, 2, 3, and the f-block": (0) hydrogen [easily swept under the rug], (1) beryllium [not so easy], and (2) zirconium and hafnium [that have EN values lower than scandium]. And limiting by oxidation state results in contradictions like (1) Th, Pa, and U, highly active metals whose characteristic oxidation states are too high to be ionic, but act very pre-transition anyway, and (2) thallium, which shows up as "too ionic" just because its favoured oxidation state is +1.
It's not "overly dramatic". It's a true statement of facts. Everything you draw will be some sort of distortion. I bet R8R will agree about that; IIRC he's said it himself a few times. ^_^ A Lu table is evidently a first-order approximation (it becomes even more evidently one if it becomes perfectly symmetric with He over Be). A La table, in trying to show a second-order phenomenon as well, sets up expectations it cannot possibly fulfill as you can't reflect absolutely everything at that order! Double sharp (talk) 00:26, 1 February 2020 (UTC)
@Double sharp: Nope, that's not what I said. I didn't refer to a continuum of more ionic to covalent. I talked about group 3 having a predominately ionic chemistry whereas group 4 has a predominately covalent chemistry. Once again you change my context and add irrelevant details. Sandbh (talk) 02:57, 2 February 2020 (UTC)
@Sandbh: I'm well aware that you did not. But you should have, because a continuum is what really happens. ^_^ And these are not irrelevant details: they expose the problem with saying "predominantly ionic" or "predominantly covalent", and point to the factors that control it: atomic size and charge, just like Fajans' rules told us. Once you look at it that way, you realise that this break is inconsequential because Th and U look "predominantly covalent" too. And yet, how active they are, and how s-like their chemistry is. Double sharp (talk) 09:47, 2 February 2020 (UTC)

On other important breakpoints in electropositivity, showing that "ionic vs. covalent" insofar as it means anything is assuredly not the only one, we have this contribution: from Greenwood and Earnshaw, p. 347: "The metals which form silicate and aluminosilicate minerals are the more electropositive metals, i.e. those in Groups 1, 2, and the 3d transition series (except Co), together with Y, La and the lanthanoids [excuse the pedantic exclusion of La as a lanthanide], Zr, Hf, [notice something?] Th, U, and to a much lesser extent the post-transition elements SnII, PbII, and BiIII." As I wrote in Archive 27: "This accords pretty well with what I've always been thinking, with a few exceptions. The exclusion of the rest of the actinides must be due to their short half-lives, as this is a geochemical context; the need to specify the oxidation states for Sn, Pb, and Bi is quite reasonable given what I've been saying about the increase of metallicity as the oxidation state lowers. The exclusion of Tl (Ga and In are too happy in the +3 state) may be due to the fact that it is quite widely distributed and tends not to occur in quantity; given that Co is almost invariably associated with Ni, I also understand its exclusion from the club as being for geochemical reasons." Double sharp (talk) 18:37, 10 February 2020 (UTC)

Monocations of Sc, Y, La and Lu

My reference is this article: "Periodic trends in chemical reactivity: Reactions of Sc+, Y+, La+ and Lu+ with H2, D2, and HD".

I saw this it when were doing our IUPAC sub but never looked too closely. Looking again, I see there is quite a bit of support for La in group 3.

The primary electron configurations of the +1 ions are curious:

Sc+ 4s3d
Y+  3d2 5s4d
La+ 5d2
Lu+ 6s2
@Sandbh: Ground state of Y+ is 5s2. Since Ac+ and Lr+ are both 7s2, we see that Lu+(6s2) fits much better than La+(5d2). Droog Andrey (talk) 15:01, 20 January 2020 (UTC)
@Droog Andrey: My mistake. 5s2 for Y is non-representative, at 11%. 5s4d is 80.7%. Y is more like Sc. Sandbh (talk) 04:19, 21 January 2020 (UTC)
In that case it is still in favour of Lu for giving a more homogeneous trend (two one way, two the other way, rather than two one way, and each of the other two another way). Well, you can explain the La trend by noting that relativistic effects are not significant at La but must be taken into account for Lu (a bit), Ac, and Lr, but why go for the split d-block when you don't have to? Double sharp (talk) 08:14, 21 January 2020 (UTC)
@Double sharp: No. La acts even more like a d element with its d2 configuration; Lu acts more like an end-of-block element. Sandbh (talk) 11:33, 23 January 2020 (UTC)
@Sandbh: Y+ is 5s2. Droog Andrey (talk) 13:43, 21 January 2020 (UTC)
Well, then Droog Andrey's initial comment stands completely. ^_^ Double sharp (talk) 16:08, 22 January 2020 (UTC)
@Droog Andrey: @Double sharp: The NIST ref is not relevant here. Under the experimental conditions set out in the article (which applied to Sc, Y, La and Lu) the predominant configuration of Y, by a > 7 to 1 margin, was 5s4d. Sandbh (talk) 11:33, 23 January 2020 (UTC)


1. "The apparent thresholds and shapes of the cross sections in the threshold region are very similar for Sc, Y, and La. For Lu, the onset of reactivity is slower than for the other ions (Figure 2)." (pp. 3154—3155)
2. "For M = Sc, Y, and La, formation of MH+ is favored over formation of MD+ by factors of approximately 2.0, 1.4, and 1.2, respectively. The behavior of Lu is quite different from the other three systems." (p. 3156)
3. "For several other systems, we were able to observe the reaction of a single state and this simplified the thermochemical analysis. For the Sc+, Y+, and La+ systems, several states are always present and presumed reacting. In the absence of experimental information to the contrary, we assume all states are present with populations given by Table I and that they have equal reactivity. This is our standard procedure for cases such as these. In the case of Lu+, it is assumed that the only state that reacts is the ground state, since the only significantly populated excited state is expected to react inefficiently, as discussed below. (p. 3156)
4. "The Lu+(SI) system is rather different from the other three systems in several respects." (p. 3158)
5. "The diabatic reactivity rules developed previously for the first-row transition-metal elements do not hold for the group 3 elements. This is consistent with the existence of surface crossings that are avoided, leading to adiabatic reactivity. Sc+, Y+, and La+ are seen to react primarily via insertion, while Lu+ reacts via an impulsive mechanism at threshold and via a direct reaction at higher energies. (p. 3158)
6. "Sc+, Y+, and Lu+ show adiabatic behavior, with crossings from potential energy surfaces derived from s2 and sd configurations to more reactive surfaces derived from d2 configurations (Sc and Y) or sd configurations (Lu). For La+, diabatic reaction along a d2 surface explains the bulk of the reactivity." (p. 3158)

The last extract is interesting. While Sc, Y and La interact with dihydrogen via an insertion mechanism, and Lu impulsively at threshold energy and in a direct manner at higher energies, how they get there is different depending on the electron configurations involved, including at higher energies. So Sc, Y and Lu get there adiabatically and La gets there diabatically. The electron configurations at higher energies are:

And I wonder why this particular case, where Sc and Y show behaviour like Lu, and La is different, is being given nary a mention? Double sharp (talk) 00:36, 26 January 2020 (UTC)
Sc+ 3D 4s3d   88.6%
    1D 4s3d    6.0
    3F 3d2     5.4
   ≥1D 3d2    <0.1

    1S 5s2    11.6
    3D 5s4d   80.7
    1D 5s4d    6.7
    3F 4d2     1.0
   ≥3P 4d2    <0.1

    3F 5d2    69.3
    1D 6s5d   12.6
    3D 6s5d   16.4
    3P 5d2     1.2
    3P 6s2    <0.6

    1S 6s2    99.7
    3D 6s5d    0.3
   ≥1D 6sSd  <0.01

Sandbh (talk) 05:51, 19 January 2020 (UTC)

How is the +1 oxidation state significant for any of these elements? And electron configurations at higher energies cut both ways: if you look at La3+, the first excited state is 5p54f1: the 5d configuration is 1.6 eV higher in energy. Double sharp (talk) 09:49, 19 January 2020 (UTC)

In the same way that the first IE of the elements is a key indicator of periodic trends. Sandbh (talk) 06:59, 20 January 2020 (UTC)

And what about the third one, which (1) often corresponds to an actually existent oxidation state for these elements, and (2) consistently for everybody means that the outer s-electrons have been ionised anyway already and results in comparing a more analogous situation for everybody? To quote Droog Andrey's plot again:

IE3 Droog Andrey.png

Double sharp (talk) 14:41, 20 January 2020 (UTC)

@Double sharp: Preliminary comments. The graph doesn't appear to compare like-with-like at least in the case of the Ln. In La, Ce and Gd atoms, a d electron is ionized; in Pr to Yb (excl. Gd) an f electron is ionized; and in Lu a d electron is being ionized. In the case of the An, at least for Ac and Cm a d electron is being ionized, whereas for Lr a p electron is ionized? Sandbh (talk) 07:03, 21 January 2020 (UTC)

By this standard, first ionisation potential would be irrelevant because the first electron to go varies across the table (certainly Pd is not giving up an s-electron). So this 3rd IP trend is no worse and maybe even better for f- and d-block elements, to ionise away the outer s-shell most of the time. Anyway: for La and Gd, a d electron is ionised, but for every other f-block lanthanide it is an f electron. For Lu it is an s electron. Ac gives up an s electron, Th and Pa give up a d electron; but for every other f-block actinide it is an f electron. And Lr gives up an s electron and is here completely homologous to Lu. The point of this graph is that the position of Eu and Yb in most representative lanthanide trends, not Gd and Lu, is analogous to that of Mn and Zn in a 3d trend, which suggests that the real end of the block is Yb. Double sharp (talk) 08:09, 21 January 2020 (UTC)

@Double sharp: I’m still looking at this. It seems (a scary word for me) that the pattern moves to the left or right depending on whether the IE is for +2, +3, or +4. For example, the +2 peak in the 3d metals is Cr; +3 is Mn; and +4 is Fe. Curiously, for the Ln, +2 peaks at Gd, +3 at Eu, and +4 at Gd! And the +4 IE for Lu is nothing that special, compared to Yb, which is weird. Sandbh (talk) 10:01, 21 January 2020 (UTC)

So what? Look at 2nd ionisation energies from He onwards and you will see the big gap occurring between group 1 and group 2, not group 18 and group 1. But we look at the most relevant one in each case. For main group elements, 1st IE is the most reasonable; but for transition elements, since the difference is in the inner shells, 3rd IE is the most reasonable because that ionises away the covering s-shell. Double sharp (talk) 13:39, 21 January 2020 (UTC)

I see the 3rd IE pattern doesn’t work for the p- block elements. Sandbh (talk) 11:02, 21 January 2020 (UTC)

Because those do not have a covering s-shell like the d- and f-block metals do. 3rd IE is the most relevant for groups with this outer covering s-shell, and thus to the whole transition region. So it's relevant for a wide swath of the periodic table, and not just a single group. Looking at 3rd IP smooths out the irregularities caused by differing configurations for transition elements as of the most stable oxidation states. For the 3d metals this is +2, with a peak in the IE trend line at Cr (group 6). For the 4d metals it's +4 peaking at Ru (group 8), and for the 5d metals it's +4 peaking at Os (group 8). So the TM trend lines have no particular relevance.

The third IE trend line along the f-block can be shown as…

La to Eu            |Gd to Yb            |
La Ce Pr Nd Pm Sm Eu|Gd Tb Dy Ho Er Ym Yb|Lu
  U  U  U  U  U  U  D  U  U  U  U  U  U  D
                    |                    |

In this case each half of the f-block has a UD finish.

or as:

  |Ce to Gd            |Tb to Lu            |
La|Ce Pr Nd Pm Sm Eu Gd|Tb Dy Ho Er Ym Yb Lu|Hf
  U  U  U  U  U  U  D  U  U  U  U  U  U  D  U
  |                    |                    |

In this case each half of the f-block has a DU finish.

It makes no intrinsic difference. Sandbh (talk) 01:08, 22 January 2020 (UTC)

@Sandbh:It makes a huge difference. Droog Andrey (talk) 15:05, 22 January 2020 (UTC)

@Droog Andrey: Well, I did say “intrinsically” for a reason. Sandbh (talk) 04:48, 23 January 2020 (UTC)

@Sandbh: And we all know that reason: your devotion to La table :) Droog Andrey (talk) 11:51, 23 January 2020 (UTC)
+1 Double sharp (talk) 19:57, 8 February 2020 (UTC)
Just graph 1st IEs in B through Ne. Or 3rd IEs in Sc through Zn. In every first row of a block, you get the UD finish. As you obviously must because getting the D means that you have started the second half of the block, as the new electron is ideally paired with another one and the repulsion means ionisation energy should decrease! The fact that you see this going from Eu to Gd, but not from Gd to Tb, speaks volumes about how this effect is important and that it is really Gd which is anomalous (should be f8 but happens to be f7d). A La table creates an irregularity that is absent from a Lu table. (As everyone expected, since a La table looks irregular in the first place.) Double sharp (talk) 15:51, 22 January 2020 (UTC)

@Double sharp: I agree. The irregularity arises because of the delayed start of filling of the 4f subshell. Sandbh (talk) 04:58, 23 January 2020 (UTC)

Here’s a related article, “Periodic trends in chemical reactivity: Reactions of Sc+, Y+, La+, and Lu+ with methane and ethane".

The conclusion is interesting re the influence of the closed 4f shell on the reactivity of Lu.

"In most respects the data for the different metals are quite similar, although Lu+ has distinct low-energy behaviour.”

"Overall, the reactivities of the group 3 metal ions with methane and ethane are quite similar. Further, the thermal chemistry of the metal hydride ions, metal dihydride ions, metal methyl ions, and metal methylidene ions for Sc+, Y+, and La+ are all comparable. Lu+ differs somewhat in both respects. The difference between La+ (which has no 4f electrons) and Lu+ (which has a filled 4f shell) must be a result of the differing 4f orbital occupancy. However, there is no indication of direct participation of the 4f orbitals. Rather the effect of the 4f orbitals is probably an indirect one. Namely, the occupation of these orbitals raises the energy of the 5d shell such that the two valence electrons in the ground state of Lu+ both occupy the 6s orbital. In La+, the energy of the 5d and 6s orbitals is much closer, leading to open shell electron configurations. The closed-shell stability of the Lu+(lS,6s2) ground state probably accounts for its distinct reactivity compared with Sc+, Y+,and La+."

Sandbh (talk) 08:28, 22 January 2020 (UTC)

So the fact that 4f is added going down from Y to Lu has an effect. That's exactly like how the fact that 3d is added going down from Al to Ga has an effect, so it's more evidence for Sc-Y-Lu for the regularity of the table pattern. It brings Lu closer to the rest of the 5d elements and follows the regular pattern that in every even period, we add a new block, and that the contraction that results impacts the properties of the following elements. The scandide contraction impacts Ga through Kr (all six elements in the next 4p row) just like the lanthanide one should impact Lu through Hg (all ten elements in the next 5d row). Double sharp (talk) 16:08, 22 January 2020 (UTC)

@Double sharp: No, what regularities there are arise out of considering periodicity in the properties of the elements, not the other way around. The Ln contraction is still going in Lu; that’s not the same as what happens in the 4p row. Sandbh (talk) 11:41, 23 January 2020 (UTC)

No, we have a uniform contraction across the whole period, as we know from high school chemistry. Just look at the table at atomic radius. All that we are doing is singling out bits of it that correspond to the filling of various periods. You will get a contraction whether you look at Y3+ through Ag3+ or Y2+ through Ag2+. And either way, it will extend to limit cases like Sr, Cd, or In depending on what exactly you are plotting. So why can't we say that Lu is not only the end of a Ln contraction (which is significant mostly because the Ln are in the same oxidation state almost all the time), it is also the start of a 5d contraction? We could say the same about Zn as well, since Zn2+ vs. Mg2+ shows similar behaviour to Ga3+ vs. Al3+. Double sharp (talk) 12:18, 23 January 2020 (UTC)

@Double sharp: It’s not uniform. It varies according to the electron type. Lu 3+ [wrong; meant to say La 3+ Sandbh (talk) 07:10, 26 January 2020 (UTC)] doesn’t start the 5d contraction; it has no d electrons. Sandbh (talk) 04:51, 25 January 2020 (UTC)

By that logic, neither do Sc3+ and Y3+. And by that logic, the d-block in period 6 runs from Hf to Tl. Which is exactly why I think saying "oh, there is this separate contraction, and it has to start when the appropriate cation has 1 electron of the appropriate block and not 0" is a bit silly. Where, other than the 4f row, is there an "appropriate cation"? Double sharp (talk) 10:11, 25 January 2020 (UTC)
P.S. Your correction doesn't matter. Neither La3+ nor Lu3+ have a d-electron. So in any case, by this logic, a d-contraction must start in group 4 and end in group 13. Which is clearly silly. ^_^ Double sharp (talk) 14:23, 30 January 2020 (UTC)

@Double sharp: Let’s exercise some common sense here. I’ve only seen the scandide contraction referred to in the sense of its impact on, for example, group 13 onwards and, for example, what happens to the pattern of ionic radii going down group 13. Yes, the Ln contraction is unique. Welcome to one of Nature’s subtleties. Sandbh (talk) 07:10, 26 January 2020 (UTC)

Yes! And again, that is because only in the 4f series is there a constant cationic charge that lets you see it. See Greenwood and Earnshaw's chapter on the Ln (see, I can refer to the literature too, just as well). This is so unique that even the 5f series does not display it, which speaks volumes against considering it as something that should affect the entire f-block. And the 6f series should be even further from this kind of behaviour. In every other case (broadening it to the whole periodic table, not just f series), the important thing is the knock-on effect after the shell has completed, because there is no constant cationic charge across a contraction like Ca through Zn. In a case when the new shell has no radial nodes, this creates secondary periodicity because we are just starting to have a new incomplete shielding effect. Switching out La for Lu destroys this pattern. And for what reason? Only to satisfy the completely one-off occurrence where the +3 state is a good comparison for everybody in the lanthanides. It doesn't even hold for the actinides, when Lr is extremely weird for a late actinide, but Ac is more normal as an early actinide. So are we going to show a Sc-Y-La-Lr table with the break in a different place for each, or are we going to apply some common sense about the pattern and recognise that it's the 4f series that is singular and shouldn't dictate things? We know very well that the first subshell of a given angular momentum is anomalous, going s >> p > d > f. Why are we letting the anomalous 4f, rather than the normal 5f, set the tone for how we draw the f-series? We can't even argue that 5f is outside the sphere of the average chemist, since thorium and uranium are almost normal elements. Double sharp (talk) 10:22, 26 January 2020 (UTC)

@Double sharp: Eh? The Ln and An have in common the +3 oxidation state. A consistent contraction (absent Th 3+) can be discerned across the +3 An ions. See this article. Note the comments about the unusual behaviour of La and Ac, once again separating them from the rest of the Ln and An. I know what will follow :) an obtuse argument that attempts to leverage this nice article to supporting Lu! Sandbh (talk) 07:38, 28 January 2020 (UTC)

Right in one, except that the argument will not be obtuse. Firstly, +3 is manifestly not a characteristic oxidation state for Th, Pa, U, and Np. U3+ even reduces water, so if +3 is characteristic for it, we're back to admitting +3 for Zr and Hf. Secondly, this is not in any way "unusual behaviour". It is simply an effect of the Ln and An contractions: La and Ac are the largest Ln and An cations respectively, which totally explains all this behaviour. We could equally well single out Lu and Lr as the smallest for the same reason. Double sharp (talk) 11:36, 28 January 2020 (UTC)

@Double sharp: This is a good example of setting aside my context ("Ln and An have in common the +3 oxidation state") and substituting your own "+3 is manifestly not a characteristic oxidation state for Th, Pa, U, and Np" noting I never said that, and then going down your own path about +3 for Zr and Hf. I am able to say "Ln and An have in common the +3 oxidation state", and that there is an Ln contraction which parallels the An contraction. The fact that the +3 oxidation state is not the most stable for Th to Pu, and No, doesn't invalidate my argument. The fact of the existence of +3 for Zr and Hf doesn't invalidate my literature supported argument that the chemistry of group 3 is predominately ionic, whereas that of group 4 is predominately covalent. Sandbh (talk) 06:08, 1 February 2020 (UTC)

@Sandbh: By your standard, I can equally well claim a +3 contraction from Sc through Cu, Y through Ag, and Lu through Au. (Every one of those elements has a +3 state. Whether or not it is happy to be in it is another story). Then I can equally conclude that group 3 does not belong in the d-block because the contractions have not started yet. Or even better, I can claim a +1 contraction. (Yes, all of them have it but Lu so far, and I have no doubt that that's just because no one has bothered to try it!) Then the d-block evidently starts in group 2 instead! All I am saying is: be consistent. If you say that weird oxidation states are invalid in one argument, then I'm not seeing why this is so different that they suddenly become valid. Meanwhile, as I've already explained, "predominantly ionic" if it means anything just means "low EN and low oxidation state", which are two properties that matter more for the chemistry. In that case we get into serious difficulties, with: (1) Zr and Hf having lower EN than Sc (so if +3 matters, suddenly they also become "quite ionic"); (2) Tl looking more ionic than its EN should allow just because it prefers to be in the +1 state; (3) Th, Pa, and U being strongly active metals whose preferred states are too high to be ionic. You can't really resolve (3) without also letting (1) swing the way you don't want, because protactinium also prefers high oxidation states and has EN 1.5 on the Pauling scale: this is far too high to be ionic, but somehow Pa is an active metal anyway patterning with the pre-transition ones. Did I mention that Pa prefers +5, which is one step higher than +4 for Zr and Hf, which also have EN less than 1.4? Meanwhile, my criterion (look at chemically active subshells if conclusive, if inconclusive look at properties to homogenise a block) works perfectly from element 1 to element 172, even. Double sharp (talk) 10:15, 1 February 2020 (UTC)

@Double sharp: Unfortunately you’re taking my context and applying it to a different, obtuse context. The scandide contraction describes the effect of having full 3d orbitals on the period 4 p elements. That’s all. I’ve never seen it applied to the 3d ions per se. It doesn’t apply to the whole of the d-block first row, since it would start at Ti3+ and finish at Cu3+ since Sc 3+ has no d electron and there is no Zn 3+. The comparison with the Ln and An contractions isn’t there.

Just read what Greenwood and Earnshaw has to say about the Ln contraction in their chapter on the Ln. It agrees with what I have been saying: there are 1001 contractions in the whole periodic table, and the Ln one is special only because everyone is in the +3 oxidation state most of the time. They even give the 4d example of Y3+ (nota bene, with no 4d electrons!) through Ag3+. The moment the constant +3 oxidation state is not maintained, the Ln behave differently, like in any other contraction. Notice how the An contraction is much more rarely referred to than the Ln one, for precisely this reason. Double sharp (talk) 11:26, 1 February 2020 (UTC)

On predominately ionic v predominately covalent you’re again drifting off the broad contours and descending into details that are irrelevant at the broad contour level. Sandbh (talk) 10:53, 1 February 2020 (UTC)

Nope. The details determine the broad contour. In particular, Be and Mg are neither "highly electropositive" nor "predominantly ionic". And Th, Pa, and U are "highly electropositive" but not "predominantly ionic". And Tl is "predominantly ionic" but not "highly electropositive". I think that says it all about which one is the relevant criterion. Wulfsberg had the right idea by focusing on EN, even if his breakdown is not quite right: really, by his EN < 1.4 criterion, the highly electropositive ones ought to be: Li through Fr; Ca through Ra (peripheral Mg); Ln and An (Pa is too high only because of its high oxidation state); Sc through Lr; Zr through Rf. Double sharp (talk) 11:26, 1 February 2020 (UTC)

@Double sharp: The broad contours are generally determined by an 80/20 approach. The 80 = the big picture; the 20 = the noise. For example, oxygen generally has an oxidation number of -2 in the combined state. Group 3 has a mainly ionic chemistry; groups 4 to 12 have a mainly covalent chemistry. If all you know about the weather is that there are four seasons then you know about 80% of what there is to know, as another example. Sandbh (talk) 03:08, 2 February 2020 (UTC)

@Sandbh: As I have already demonstrated by reference to Tl and Th, this is not the most relevant criterion at the large scale. You would, in fact, do much better just looking at EN like Wulfsberg does to get a broad contour. Double sharp (talk) 09:49, 2 February 2020 (UTC)

A historical chemical perspective on group 3

@Double sharp: @Droog Andrey: As noted by Jensen, early spectroscopic work on the Ln seemed to indicate that the ground states of their atoms had, with few exceptions, an electronic configuration of the form:

f = 1 to 14, d1 s2

So La was d1 s2 and Yb was thought to be f 13 d1 s2 and Lu f 14 d1 s2.

It was subsequently determined that most of the Ln were:

f = 1 to 14, s2

Only Ce, Gd, and Lu also had a d electron. And it turned out that Yb was f14 s2 i.e. the f shell was completed over the course of 13 rather than 14 elements, a little bit like the d shell being filled over the course of 8 or 9 rather than 10 elements.

The interesting thing about the new configurations is that nothing changed for La and Lu in terms of their configurations and their chemistry.

So they stayed where they were, La under Y, and Lu at the end of the Ln.

A few tables of the 1920s and 30s showed Lu under Y for reasons of regularity (Janet) or because Lu occurred in the “yttrium” group (along with Sc and Y) rather than the “cerium” group (which included La).

But this never took off.

That Sc, Y and Lu occurred in the so called yttrium group, and that La occurred in the "cerium" group did not imply anything particularly significant; it is simply a reflection of the increasing basicity of these elements as atomic radius increases. Taking the alkaline earth metals as another example, Mg (less basic) belongs in the "soluble group" and Ca, Sr and Ba (more basic) occur in the "ammonium carbonate group". Moving Lu under Y because they occur in the same chemical separation group failed to consider separation group patterns elsewhere in the periodic table.

Further, the separation group behaviour of Y can be ambiguous, and Sc, Y, and La appear to show complexation behaviour different to that of Lu, per the quotes by Vickery (1960, 1973) earlier in our discussion.

In this context, the old chemists who kept La under Y were on the mark, chemically speaking.

It’s ironic that the condensed phase configurations of most of the Ln is indeed f1-14, d1 s2 and that, in this context, the 4f shell is not completed until Lu. Sandbh (talk) 04:24, 23 January 2020 (UTC)

  1. Chemically speaking, there are probably about as many cases when Y is more like La as cases when Y is more like Lu. Again, that is expected, because of the small differences involved. So neither is "mistaken" in that sense. It is like comparing Al to Sc and Ga. Or comparing Mg to Ca and Zn.
  2. Inertia could hold for Sc-Y-La because it is not obviously wrong. It gives an OK trend, even if it is weird that it acts like groups 1 and 2, which have absolutely no double periodicity because of the way the Madelung rule works (even the other main groups don't follow that). And La is not obviously a wrong congener for Y. Now that we understand this better, and actually know all the elements involved up to the seventh period, the star of Sc-Y-Lu has been rising due to significant adoption (e.g. WebElements).
  3. Condensed phase configurations suggest that Be and Mg are p-block elements, which just about says it all about why I think this is not a good argument.
So now that we know all the elements involved, and we know the general Madelung pattern, it seems to me that Sc-Y-Lu is the simpler case (that nothing fancy is going on) and should be our null hypothesis. Since the evidence that brings Sc-Y-La over Sc-Y-Lu is pretty small, all things considered (despite the amount of verbiage we alone have written about it ^_^), I put it that there is not enough evidence to reject H0. Double sharp (talk) 12:14, 23 January 2020 (UTC)

@Double sharp: I try to stay away from comparisons of individual elements like Y v La or Lu but I do get sucked in now and then. It occurs to me that Al goes over Ga since both are p elements. Mg goes over Ca since the fit is better in terms of electron configurations. This principle can be extended to La and Lu.

  1. And I try to stay away from anything other than chemistry, because after thinking about it for this long I am almost certain that you will not find a philosophical argument that actually stands up to all the counterexamples the elements seem willing to provide us and actually reflects the richness of what we are trying to depict. Again: without the chemical properties that Mendeleev observed, what is the basis of periodicity?
    @Double sharp: The elements do provide many counter examples at their level. When you step up into the balcony and observe the dance floor, you’ll see the general patterns and that the La form provides greater regularity/fewer irregularities. DIMs focuses was on the binary compounds rather than the individual elements so much. Sandbh (talk) 06:48, 26 January 2020 (UTC)
  2. Lu fits better under Y in terms of electronic configurations because we are in an even period outside the s-block, and we expect a new shell type to be added. Same as Zr to Hf, Nb to Ta, all the way to Xe to Rn. And Al to Ga, too. Double sharp (talk) 00:20, 25 January 2020 (UTC)
    @Double sharp: No, we don’t necessarily expect a shell type to be added because we know the n + l rule is an approximation only, and it has no first principles basis. Sandbh (talk) 06:48, 26 January 2020 (UTC)
    @Sandbh: Yes we do because it works perfectly well from H to Xe. The null hypothesis is clearly that nothing special happens, and it is the alternative hypothesis that something new and different happens that demands proof. Proof which I still don't see. Double sharp (talk) 10:34, 26 January 2020 (UTC)

There’s nothing particularly weird about group 3 acting more like groups 1 and 2.

Indeed, since groups 4 and 5 also act more like groups 1 and 2 very often, so the argument does not amount to that much in hindsight. Double sharp (talk) 00:20, 25 January 2020 (UTC)
@Double sharp: That’s complete nonsense. I’m astonished to think you could post such rubbish. For the seven thousandth time, as per the literature, groups 1 to 3 have a predominately ionic chemistry. Groups 4 and 5 have a predominately covalent chemistry. End of story. The end. Period. Sandbh (talk) 06:48, 26 January 2020 (UTC)
Look, there is no such thing as a complete volte-face from ionic to covalent. It depends on what the counter-anion is. We go from ionic to metallic (which you're overlooking completely) across the series NaCl, Na2O, Na2S, Na3P, Na3As, Na3Sb, Na3Bi, Na. And that's in group 1, with an example taken straight from Greenwood & Earnshaw p. 81. Ionic vs. covalent is (1) gradual, from covalent to polar covalent to mildly ionic to strongly ionic, depending on the difference of electronegativity between the two elements; (2) not a complete dichotomy, because you overlook metallic bonding; and (3) not split by elements, but rather by electronegativity differences, which have a lot to do with oxidation state (just compare uranium chemistry as we jack the oxidation state up from +3 to +4 to +5 to +6). So I'm astonished to see you put so much weight on "ionic vs. covalent" as a false dichotomy. (And as I keep saying, the general thing across the periodic table is continuity, and the sharp dichotomies you like to point to have a distinct tendency to not exist.) The whole literature, when it sees fit to split "main group" from "transition", universally uses other criteria like "variable oxidation states", "coloured compounds from d–d transitions", "a wide variety of complexes", "formation of paramagnetic compounds". Ionic vs. covalent has nothing to do with it, and nobody uses that as a criterion in the literature for this divide. Respectfully, I put it to you that Zr, Hf, Nb, and Ta by this standards have weak credentials as transition metals, just as weak as Sc for that matter. Double sharp (talk) 10:32, 26 January 2020 (UTC)

Jensen did spark some interest in the Lu form but nothing substantial came of it. Webelements adopted it on the basis of his arguments, which were not balanced, as per our IUPAC submission.

On balance, I would tend to say now that he did not give the full story and that it is more complicated than his arguments would suggest. (As it is for his treatment of group 12, where in particular I think he's being far too extremist about it being totally not a transition group; it is even wrong for Cn.) However, in this case I think his conclusion is more or less sound, even if not always his arguments. There is just not enough reason to go for Sc-Y-La because both the La and Lu forms exhibit good trends, and the Lu form shows a balance between the s- and d-blocks for this peripheral group 3. Double sharp (talk) 00:20, 25 January 2020 (UTC)

AFAIK the chemistry of Be and Mg is that of hybridised sp elements, not that this makes any difference.

The n + l “rule” is not that helpful since an Lu table has 13 n + l irregularities while the La form has only 12. For sure, if the 4f subshell started filling at La there’d be no issue.

We have to remember that the forces that interact to produce the actual sequence of electron configurations do so consistently and regularly. The fact that we interpret the end result as featuring “anomalies” is a result of focussing too much on the regularity of the n + l rule, which is only an approximation.

The whole point is that the n+l rule is approximately right on what subshells are going to be filled. It is even completely right on what subshells will turn out to be chemically active. And I stand by calling them anomalies, when stuff like Nb d4s1 vs. Ta d3s2 and Lu ds2 vs. Lr s2p means about nothing for their chemistry, which follows more or less ideal configurations of d5 and d3 respectively for compounds. Double sharp (talk) 00:20, 25 January 2020 (UTC)

Like Eric Scerri wrote in the 2nd edition (2020) of his periodic table book:

"Too many proponents of alternative periodic tables seem to argue about the regularity in their representation and forget that they may be talking about the representation and not the chemical world itself." (p. 387)

"...we should beware of arguments based on symmetry and regularity." (p. 401)

Sure, but there's nothing wrong with them when the chemistry is itself inconclusive. In that case the symmetric form is preferable because breaking the symmetry when both options are finely balanced makes it look like the situation is more conclusive than it really is, just because the symmetric form is by definition a null hypothesis: "things shall go on as they always did". Double sharp (talk) 00:20, 25 January 2020 (UTC)

Of the Lu tables out there the LST or ADOMAH can be regarded as representing, at least in some sense, the situation pre-symmetry breaking as per the n + l rule. Those tables are representations of this theoretical and beautiful tetrahedral symmetry. The n + l rule provides the underlying basis or spine of the periodic system. Despite the blemishes, periodicity continually returns back to the n + l rule. That is worth showing, as a table towards the physics end of Scerri’s continuum of periodic tables. At the other end is Rayner-Canham’s unruly inorganic chemist’s table. In the middle somewhere is the conventional La form. Sandbh (talk) 00:06, 25 January 2020 (UTC)

@Sandbh: n+l rule was derived from 1st principles by Klechkovsky in 1951. Droog Andrey (talk) 11:14, 27 January 2020 (UTC)
@Droog Andrey: Klechkovsky‘s article is disputed. According to Scerri the rule has still not been so derived. The Löwdin challenge for a first principles derivation still stands.
@Sandbh: And yet the n+l rule works absolutely perfectly, except for the Ca group, if you take it at the broad level of "what subshells are chemically active". (Which makes sense, as I have repeatedly noted that counting exceptions is silly if you realise that Nb d4s1 and Ta d3s2 are so similar.) So even if a 1st-principles derivation for the n+l rule happens to not yet exist, its power and accuracy suggests that seeking to derive it is a worthwhile goal with significant likelihood of success, and that it makes a good basis for the PT. Double sharp (talk) 19:39, 1 February 2020 (UTC)

And I wonder how this "historical chemical perspective" would look for Be-Mg-Zn? If we argue that Be-Mg-Zn is "obsolete", then how do we know that it is Sc-Y-Lu rather than Sc-Y-La that will eventually turn out to be "obsolete"? I rather think the Sc-Y-La option is more likely to go that way. If Sc-Y-Lu were to become the default, and generations grew up learning it, no one would be suggesting Sc-Y-La as a symmetry break when it is self-evident that the trends are inconclusive. (This is indebted to one of R8R's arguments from one of the previous times this was discussed.) But if Sc-Y-La continued to be the default, there would probably continue to be a sizable unhappy minority advocating Sc-Y-Lu, just like there is now. And indeed, the minority has been arguing for long enough that Sc-Y-Lu has, if not become dominant, at least become a widespread alternative. That strongly suggests that Sc-Y-La is just chugging along as a fossil of the time with no f-block. Double sharp (talk) 19:43, 1 February 2020 (UTC)

A historical perspective, cont.

@Double sharp:

I think I know why the La form remains popular.

Historically, the Ln were known to be chemically very similar.

Originally, the 4f sub-shell was thought to be complete only at Lu.

So, with the popularity of electronic periodic tables, La was placed under Y since La did not have an f electron, and the Ln ran from Ce to Lu.

It was later recognised that the 4f shell was prematurely completed at Yb, and that Lu was in fact 4f145d1.

Nothing happened however, since the chemistry of Lu, as a lanthanide, remained 100% unchanged.

So La remained under Y in recognition of the delayed start to the filling of the 4f subshells, and the unchanged chemistry of Lu.

The resulting separation of the d-block in the 32-column form never became an issue in light of the overwhelming popularity of the 18-column form.

PS: Library day looks like it will be Monday. Sandbh (talk) 10:34, 5 February 2020 (UTC)

@Sandbh: I bet it is simply inertia combined with not being too obviously wrong, since the lanthanides are so similar anyway. But while chemically speaking in isolation both Sc-Y-La and Sc-Y-Lu make sense as trends (and I would still advocate considering both elements in a comparative chemistry review for group 3), La in group 3 flies in the face of the logic of the periodic table, since La is an f-block element with its low-lying f-orbitals. (Yes, for the same reason, I side with Henry Bent on He over Be.) Double sharp (talk) 20:29, 5 February 2020 (UTC)

@Double sharp: Was it really inertia? Let's recall that differentiating electrons sorted out the placement of La. And when it was realised that the f series in fact finished at Yb rather than Lu, this had zero impact on the chemistry involved. The presence of low-lying f orbitals are not on the radar in this fundamental context. OTOH, the fourteen f electrons in Lu3+ and the impact of their poor shielding on the size of the Lu cation, are on the radar. That's not just my opinion; it's the consensus in the literature.

More history. It occurred to me that there is another aspect to this.

The recent interest in symmetrical forms such as ADOMAH, can be traced back to Janet's Left Step Table (1928). His table was neither a chemical table nor an electronic table, since he choose it for its symmetry. In fact he corrected some of what he thought were incorrect electron configurations to make it fit with the n + 1 rule. He was wrong since what he took to be erroneous electron configurations were in fact correct. So the n + l rule was wrong too.

In contrast to Janet's approach it was known from as early as 1929 that "…If Sc, Y, La and Ac are the only rare-earth elements, the series would have revealed the same gradual change in properties as the Ca, Sr, Ba and Ra series, and hence it would not have been of any special interest." (from our IUPAC submission).

Now, in Jensen's seminal 1986 paper, "Classification, symmetry and the periodic table", he wrote that if the classes in a periodic table appear in symmetry related positions, rather than being imposed beforehand, then both phenomena would suggest the resulting classification is really a natural one (pp. 496-497).

In other words, the interest in symmetrical forms and the n + 1 approximation is mis-founded since this starts with symmetry first, rather than arranging the elements DIM fashion according to the periodic law. This resulted in a non-symmetric table, which eventually evolved into the non-symmetric 18-column La form.

Jensen's words are echoed by C. A. Coulson, theoretical chemist and professor of mathematics, who concluded his Faraday lecture on symmetry with the words:

"Man's sense of shape—his feeling for form—the fact that he exists in three dimensions—these must have conditioned his mind to thinking of structure, and sometimes encouraged him to dream dreams about it. I recall that it was Kekule himself who said: "Let us learn to dream, gentlemen, and then we shall learn the truth." Yet we must not carry this policy too far. Symmetry is important, but it is not everything. To quote Michael Faraday writing of his childhood: “Do not suppose that I was a very deep thinker and was marked as a precocious person. I was a lively imaginative person, and could believe in the Arabian Nights as easily as in the Encyclopedia. But facts were important to me, and saved me.” It is when symmetry interprets facts that it serves its purpose: and then it delights us because it links our study of chemistry with another world of the human spirit—the world of order, pattern, beauty, satisfaction. But facts come first. Symmetry encompasses much—but not quite all!

Coulson CA 1968, "Symmetry”, Chemistry in Britain, vol. 4, pp. 113–120.

The takeaway from Coulson is that "facts come first" per DIM, and later the real aufbau sequence, which together resulted in the La form.

How do you see this? I can see a very strong additional argument coming out of it. Sandbh (talk) 23:25, 5 February 2020 (UTC)

Philosophical considerations

@Droog Andrew:@Double sharp:@YBG: In drafting the article I tried to take a more philosophical slant. By this I mean considering the foundational concepts, theories, and methods associated with the PT, and the interrelations between them rather than the properties of individual elements. Well, that sounds good anyway. For now, until I replace the above placeholders and address the other outstanding remarks it appears to me that:

  • the global considerations are interrelated;
  • in an La table they reinforce one another; and
  • in an Lu table each consideration is breached.

Sandbh (talk) 04:27, 13 January 2020 (UTC)

The problem is that I'm not convinced that your considerations are indeed foundational. Differentiating electrons are only conclusive about group 3 because of Lr's p-electron, which is both (1) a weird side effect of relativity and (2) completely irrelevant chemically; and valence shells have a whole lot more to do with the chemical considerations that underpin the PT and make a better philosophical basis in my opinion. But that would support Lu-Lr again. Group 3 doesn't unequivocally side one side or the other; it's just a manifestation of the philosophical principle that trends should be continuous, which also supports Lu-Lr (look at the d-block trend). And your last argument seems predicated on distorting the PT arrangement and looking at the patterns you get from doing it, and given that the whole point of the PT is its arrangement I don't see how that could be a philosophical basis for the PT. For me, the PT is based on chemistry, thus valence shells, thus how best to reflect it in a 2D grid and show trends. Each builds on the previous and the last three all point to Lu-Lr.
We have to bear in mind that the PT is not a document engraved in stone that says "this is what the elements are like according to what Nature has given us". It instead says "here is a very useful guideline to how Nature is like, distilled into some generalisations for us humans". Knowing how humans are, the d-block break that Sc-Y-La gives creates a sense of false precision: if we just draw rectangles as for Sc-Y-Lu, we can more easily accept "remember, ye intrepid ones, that this is just an approximation", but breaking something apart as obviously as Sc-Y-La makes it seem like there is some huge qualitative difference between group 3 and the rest of the d-block and makes it look like the PT is reflecting higher-order phenomena than it really is. If there were such a thing, I would most certainly support Sc-Y-La, but I'm not seeing it. I am sure that aliens from the planet Skyron somewhere in the Andromeda Galaxy (I always enjoyed Monty Python's Science Fiction Sketch, where this name is from) will have different psychologies and therefore automatically infer different things when they see different arrangements of elements. They too will see some trends and similarities in the elements and organise it in some way, and who knows how they will do it to connote the right idea? I know not. All I can say is it might not be ours. Let's draw the PT then, conscious that it is not what Nature gave us, but a way to let us understand and organise somehow what She did. And it will be all the more useful to us once we understand what it is for. Double sharp (talk) 04:38, 13 January 2020 (UTC)

@Double sharp:

Well, the good thing is I don't appear to have made any clangers. After that it comes down to opinions.

Main arguments:
1. considerations of Group 3’s neighbours (ionic v covalent chemistry);
2. predominant differentiating electrons across the four blocks of the periodic table;
3. the periodic law (chemistry); and
4. the nature of the rare earths (add including their ionic chemistry).

Ancillary supporting arguments:
5. the predominance of the La form in the literature; and
6. the fact that an Lu table results in more irregularities among all four main considerations across the table, as well as in the regularity of term symbols, and that placing lanthanum (La) and actinium (Ac) in the f-block would be only case where a pair of elements that belong in the same column are placed such that they have no outer electrons in common with that block.

New argument pending:
7. Horizontal triads (chemistry)

1. Per the literature, overall this supports the La form. The secondary considerations you've noted aren't sufficient, IMO, to tip the balance the other way.
2. I don't focus on individual elements, so I don't in that sense care about Lr. As to the foundational nature of d/e's I can go back to Bohr, and others, and at least cite Stewart and Scerri along the way.
3. I've addressed all your objections about this.
4. Ditto.

You're 100% on the mark with the PT not being engraved in stone. That's why I've questioned why IUPAC needs to make a decision about Group 3, when each option can serve its purpose.

There's no false sense of precision about the d-block break. We know that Hamilton (1965; over 50 years ago), shows a periodic table extract (groups 1 to 11, plus footnoted Ln and An, showing Ce, Pr…Lu; and Th, Pa…Lw) with a split d block (the gap is between groups 3 and 4) and says that—without any fuss—this is "the periodic table as it is usually presented".

Indeed it is usually presented that way when done in 32-column form.

None of the split d block tables I've seen recently in text books have raised any fuss on behalf of the authors concerned.

Sir Martyn Poliakoff shows a spit-d block table at the start of his you tube video on the IE of Lr. Again no fuss. [Caveat: the authors of the Lr article did not speculate as to any implications for Group 3, not officially anyway]

The Inorganic Chemistry Division of the ACS, until recently, used a split-d block logo, until I pointed this out to Eric, and he wrote to them about it, possibly in the context of his IUPAC project. See here for a picture. Sandbh (talk) 00:54, 18 January 2020 (UTC)

It's certainly not often (if ever) remarked on, but often when the split d-block is shown, it is shown with a Madelung rule that does not reflect it and instead predicts Sc-Y-Lu. The inconsistency is swept under the rug, but it is still there, and the student who actually reads carefully will likely wonder about it. (I know I did.) Double sharp (talk) 12:19, 23 January 2020 (UTC)
P.S. Re term symbols and "a pair out of place":
  1. In a Lu table, every block ends with a column of nothing but 1S0, as we expect since all subshells are supposed to be completely full. A La table violates this.
  2. The fact that extending the periodic table to the 8th-period predictions results in even more elements out of place at the beginning of the putative g-block (unless you want to begin it at E125 and create a chasm inside a chasm) suggests that this has got everything to do with delayed collapses, and can be safely treated as a second-order perturbation to the pattern. And notice that it does not seem to affect when the subshell sinks into the core at all, only when it begins filling up with actual electrons (at a point where it can already be used for hybridisation): Lu, Lr, and E157 have an f-subshell drowned deeper and deeper into the core, something like the d-subshell down group 13 (ignoring Nh; I allow considering E157 as relativistic effects almost exactly cancel out for E157 through E172). Double sharp (talk) 12:23, 23 January 2020 (UTC)

@Double sharp: Yes, I second your observation about the needless confusion surrounding the MR.

On term symbols you are right but for Pd, which also has that kind of symbol. The other thing about term symbols is that two elements can have the same term symbol but differing electron configurations. There is at least one other example (I can’t look it up right now). So I don’t assign too much weight to term symbols. That said, an La table has one less term symbol irregularity than an Lu table which I think is a more important consideration, along with the other irregularities, as discussed, that the Lu table brings. It’s like we want to depict Nature as we think it should be rather than how it really is. That goes for the g block, too. I don’t regard the delayed collapses as second order perturbations, since the pattern is an approximation albeit a very persistent one. Sandbh (talk) 05:49, 25 January 2020 (UTC)

That's the trouble. If you insist on delayed collapses as something first-order that must be represented at all costs, then the sky falls down on our heads once period 8 finally gets started, with a brief foretaste already happening with the asteroid at Lr. If you pragmatically say that it's a second-order phenomenon that is totally normal for heavy elements, pointing to La, incipient Lu (with a surprisingly low 6p level), Ac, Lr, and E121, then you can continue with periodicity with no problem at all. You cannot depict Nature as how She really is, it is far too complex for that. Everything you draw will be a distortion. Everything you draw will be distorted based on some subjectively chosen criteria. And everything you draw will be subject to how humans, with their cognitive predispositions, are likely to see it. So why not just give a first-order approximation that works really, really well? Double sharp (talk) 10:58, 25 January 2020 (UTC)

@Double sharp: The delayed collapse is easy to draw. I recall seeing a 1946 table showing the general approach. It’s no big deal. The sky will not fall down. Let us draw nature as it is not how we think it should look like. Sandbh (talk) 09:34, 26 January 2020 (UTC)

A pretty slogan, apart from the fact that any drawing will bring us away from "nature as it is". Just look at the electron configuration table in extended periodic table and let's see how any drawing can possibly reflect what happens in a few elements' time when 8p collapses at E121, 7d at E122, 6f at E123, and finally 5g at E125, while the superactinide series goes on for so long that 8s and 8p fall into the core and get replaced as valence electrons by 9s and 9p instead to create normal-looking 4d-like transition metals from E157 to E166. Or even just see how any drawing can possibly accommodate Lr with its 7p electron caused by delayed collapses. Double sharp (talk) 10:18, 26 January 2020 (UTC)

@Double sharp: Eh? We make our drawings and seek to iteratively improve them in order to make better and better representations of Nature as we understand it. A pretty slogan? As Proff Poliakoff said "what we're interested in is what nature is like not how easy it is to draw it." Sandbh (talk) 21:25, 1 February 2020 (UTC)

@Sandbh: It's exactly a pretty slogan because Nature is too complicated to draw everything. Think Rayner-Canham's periodic table times a hundred. Double sharp (talk) 00:49, 2 February 2020 (UTC)

@Double sharp: It's the level of abstraction that's the issue. It's funny that I attempt to argue philosophically and you attempt to argue in detail. Sandbh (talk) 01:50, 2 February 2020 (UTC)

@Sandbh: No, I argue both from chemistry and philosophy. I even use some philosophical arguments like homogeneity that you yourself used before, but funnily enough they seem to become invalid when I use them to support a Lu table. ^_-☆ Double sharp (talk) 09:51, 2 February 2020 (UTC)

Fitting Lu to Hf through Hg

Since saying the same thing over and over again seems not to work, I guess I have to do the tables. In every single case I just hurriedly looked at and typed up, the fit for Lu at least is about equal to, and often outperforms, that for La. (Kulsha-Kolevich EN = EN from User:Droog Andrey and his colleague's table, which fixes the problem of Pauling assigning many transition metals overly high values because of strong homoatomic multiple bonds e.g. Mo, W).

property La Lu Hf Ta W Re Os Ir Pt Au Hg
m.p. (K) 1193 1925 2506 3290 3695 3459 3306 2719 2041.4 1337.33 234.43
b.p. (K) 3737 3675 4876 5731 5828 5869 5285 4701 4098 3129 629.88
specific heat capacity (J/(g*K)) .195 .154 .144 .14 .132 .137 .13 .131 .133 .129 .14
EN (Pauling) 1.1 1.27 1.3 1.5 2.36 1.9 2.2 2.2 2.28 2.54 2.0
EN (Kulsha-Kolevich) 1.11 1.31 1.38 1.46 1.54 1.55 1.67 1.75 1.84 1.93 1.81
Density 6.145 9.84 13.31 16.654 19.25 21.02 22.61 22.56 21.46 19.282 13.5336
Young's modulus 36.6 68.6 78 186 411 463 ??? 528 168 78 ???
Bulk modulus 27.9 47.6 110 200 310 370 462 320 230 180 25
Resistivity (nΩm, close to r.t.) 615 582 331 131 52.8 193 81 47.1 105 22.14 960
Brinell hardness (MPa) 350-400 893-1300 1450-2100 441-3430 2000-4000 1320-2500 3920-4000 1670 310-500 188-245 ???
Heat of fusion (kJ/mol) 6.20 22 27.2 36.57 52.31 60.43 57.85 41.12 22.17 12.55 2.29

And I'm even restricting myself mostly to physical properties, and already the difference is so big. With chemistry it is the same, as Lu is softer than La as a cation and coordinates better, making it a closer match for Hf through Hg. If you did it using predictions for period 7, it would (looking at the meagre data we have) probably be even bigger.

P.S. Any references to lanthanides like Ho and Er are irrelevant here. The goal is not to compete which of La and Lu is more like a lanthanide. They both obviously are. The goal is to see which of them better fits the first position of the 5d row. To do that, the right elements to compare with are Hf through Hg, which form the rest of the 5d row. Double sharp (talk) 17:14, 26 January 2020 (UTC)

My first response is to observe that this is not a question of which 5d metal best fits the position of first 5d metal. We have to remember that the first appearance of the 5d electron is followed by 14 4f electrons. So it is disingenuous to compare La with metals that occur at least 15 atomic numbers later. Sandbh (talk) 22:39, 26 January 2020 (UTC)
Do you want a homogeneous d-block or not? Let me quote you from Archive 38: "a set of items is presumably more representative of its label the more the individual items in it match the label." Well, the label of d-block chemistry is given by what we see of the behaviour of the totally sure d-block elements. So of course we should be comparing with those elements to decide which element to shunt off to the d-block. If you don't want a homogeneous d-block, or homogeneous categories, then what are we doing drawing a table? Double sharp (talk) 23:36, 26 January 2020 (UTC)
P.S. The first appearance of the 7p electron at Lr is followed by ten 6d electrons; the "premature" (of course it is nothing of the sort) filling at Cn is indeed exactly like that at Yb. So by this argument we could argue that it's just as disingenuous to compare Lr with Fl through Og, even though it's obvious that Nh is a much better fit. But would you advocate Lr under Tl? Double sharp (talk) 23:55, 26 January 2020 (UTC)
@Double sharp: The homogeneity of the d block, or otherwise, is not my call—it’s Nature’s call. Nature chooses to delay the filling of the 4f sub-shell. That’s all I’m trying to show. Effectively ignoring this and saying we compare La with the 5d metals from Hf onwards ignores the elephant in the room. Lr under Tl is a non-starter as I believe I’ve addressed elsewhere. Sandbh (talk) 07:08, 28 January 2020 (UTC)
Nature also sees fit to delay the start of the 5f shell to Pa, the start of the 6d shell to Rf, and the start of the 5g shell to E125, so claiming that "Lr under Tl is a non-starter" at the very least casts some aspersions about the very similar case of La and Ac. Again, if you look away from the dance floor of individual weird configurations, this is all just the heavy-element delayed collapse pattern, none of which prevents 4f and 5f from having some sort of rôle at La and Ac anyway. It's certainly bigger than the essentially non-f character of Lu and especially Lr. Double sharp (talk) 07:52, 28 January 2020 (UTC)
@Double sharp: I looked at the figures for La and Lu and compared then to the average values for Hf to Hg, and Ce to Yb, and with Ba.
  • Lu has more Ln character than 5d character.
  • La has more Ln character than Ba character.
Neither Lu nor La fit particularly well under Y, on the basis of the eleven properties under consideration. Sandbh (talk) 05:57, 1 February 2020 (UTC)
Of course they do, that's not the point. I can guarantee you that yttrium will show more lanthanide character than 5s or 4d character as well just because of its size and charge. The point is that Lu clearly has more d character than La. Allied with Gschneider's pointful arguments, which we even agreed with in the old submission by noting that La clearly has more f character than Lu, the block assignment is immediately clear just by the principle of categorisation that similar items belong together, which is about as philosophical as you can get for such a thing. No minutiae here. For Ac and Lr, it would be even more conclusive if not for the fact that most things now become predictions. Double sharp (talk) 09:57, 1 February 2020 (UTC)

@Double sharp: Hmm. I feel you are leaving the broad contours and descending into details; and confusing similarity with periodic trends.

Block identity works on the following basis:

  • predominant differentiating electron;
  • s is characterised, except in H, by highly electropositive metals; and by the fact that its outer electrons do not withdraw completely into the core until the next subshell of s elements is added; instead it contributes to bonding in the subsequent p, d and f blocks;
  • p by a range of very distinctive metals and non-metals, many of them essential to life; and by the fact that, when its orbitals are all filled, the six electrons with the two s electrons form an almost impregnable octet in the noble gases;
  • d by metals with multiple oxidation states; and by the fact that, as its orbitals fill up, they withdraw increasingly into the core until withdrawal is complete in Zn, Cd, Hg;
  • f by metals so similar that their separation is problematic; and by the fact that withdrawal into the core starts immediately, leaving only one (or two) electrons to combine with s electrons in bonding.

The periodic table is more than similar items belonging together. It is more importantly a demonstration of vertical periodic trends (acknowledging the arguable exception of He, on pragmatic "screaming physical" grounds, among others).

Per our IUPAC submission, La fits better under Y on periodic trend grounds, and chemical behaviour grounds. This is not the case for Lu.

Arguments to do with f character (in the absence of any f electrons) and d character, in the absence of characteristic d properties in La and Lu (multiple oxidation states etc), can only every be regarded as tipping point arguments.

Like we said in our submission, "While Lu may be somewhat more of an outlier than La, the shortfall is insignificant in comparison to broader trends." Sandbh (talk) 00:48, 2 February 2020 (UTC)

No, block identity works on the following basis:
  1. What is the chemically active subshell of the highest angular momentum?
That's it. Done. One criterion, very simple. As broad contour as you can ask for, not focusing on details like trying to "characterise" the blocks. Droog Andrey has already well demolished in Archive 33 the bases for our old IUPAC submission arguments: group 3 is not overwhelmingly skewed towards group 2 in similarity, the delayed collapse is totally normal throughout the periodic table, Lu has no significant 4f character, and condensed-phase configurations are irrelevant because they support B-Al-Sc and Be and Mg as p-block elements. In the absence of those strong arguments to lead the charge, Sc-Y-La in my view has no more leg to stand on. Double sharp (talk) 00:52, 2 February 2020 (UTC)

@Double sharp: Is it though? I've asked a question about Nd elsewhere in this thread dealing with your one criterion. According to the literature, group 3 is skewed towards group 2 in similarity; the delayed collapse is most prominent in the Ln and trying to conflate it with the rest of the blocks is disingenuous; and suggesting condensed-phase configurations are irrelevant because they support B-Al-Sc and Be and Mg as p-block elements, is a one puff argument that conveniently ignores all the other arguments pointing to how silly this is. You throw out a lot of distractions or minutiae that have no impact on the broad contours, including block homogeneity, never mind broad chemical patterns and periodic vertical trends. Sandbh (talk) 03:59, 2 February 2020 (UTC)

  1. @Sandbh: On Nd, see above.
  2. Nope. Sc and Y in group 3 have the physical properties of normal transition metals. And may I add that under "similarity", Be and Mg are more like p-block than s-block elements, and therefore Be-Mg-Zn is the recommended trend to produce something more like B-Al-Ga.
  3. Nope. The delayed collapse is most prominent in heavy elements in general, not only the f-block. That's why we see it in Lr and E121 as well. Once period 8 comes along every block but the s-block seems to be delayed!
  4. Well it seems to me that the only good case for Sc-Y-La came from weak 4f involvement (refuted) and condensed-phase configurations. But now we have a simple philosophical syllogism:
    1. Condensed phase configurations lead to Be and Mg in the p-block.
    2. Be and Mg in the p-block is silly.
    3. An argument that argues for something silly is suspect and not a strong one for anything else, by reductio ad absurdum.
    4. Condensed phase configurations are a suspect argument.
  5. Block homogeneity supports a Lu table. The f-block is about the same either way, but the d-block suffers a big loss of homogeneity when you force La into it. Respectfully, differences between La and the other lanthanides are minutiae.
  6. Broad chemical patterns entirely support a Lu table as it matches the trend of every other transition group. Same for periodic vertical trends. Since Sc and Y are physically perfect d-block metals, and chemically are not far removed from Zr, Nb, Hf, and Ta, the argument of similarity to group 2 dies a natural death. Double sharp (talk) 10:02, 2 February 2020 (UTC)

Lawrencium, with another glance at helium over beryllium

Let's examine what the criteria Sandbh seems to be using for group 3 have to say about the placement of Lr.

  1. Lr shows the configuration [Rn]5f147s27p1, so by those criteria it ends up as a p-element since the 6d orbital has undergone a delayed collapse like for Ac. Therefore we have two possible recurrences after Tl, being Lr and Nh, and unless I've misunderstood it Sandbh's interpretation of the periodic law demands that we take Lr as the first one.
  2. The 234 argument is inconclusive, since neither No nor Cn has maximum oxidation state +2. Therefore, nothing stands in the way of Tl-Lr.
  3. By how Sandbh seems to be interpreting contractions, Lr cannot be considered to be the start of the 6d contraction as it does not have a 6d electron. In fact, it does not even satisfy IUPAC's definition of a transition metal as Lr and Lr3+ both lack 6d electrons. OTOH, some predictions for Nh suggest 6d involvement in its chemistry for higher oxidation states. So the 6d contraction must start at Rf and end at Nh!
  4. Putting Lr into the p-block gets rid of one thorny differentiating electron anomaly. It unfortunately creates one for Nh wherever else we put it, so this is balanced: nihil obstat again.
  5. Inconveniently, Lr does not add the 6d10 core that eka-Tl should following double periodicity literally everywhere else in the table, but we can insist that that just shows that the n+l rule must be discarded – for an isolated anomaly, that is paralleled very well at the start of most blocks in the heavy elements, see Ac and E121, but never mind that(!!).
  6. Inconveniently, Lr does not behave much like Fl through Og, but we can simply insist that one 7p electron fills followed by ten 6d ones with a "premature" closure at Cn and declare it irrelevant, as the situation looks kind of like that going from La to Lu(!!!). We can even draw analogies between Cn-Nh-Fl and Pd-Ag-Cd due to the characteristic oxidation states, with a "prematurely closed shell" occurring at Cn and Pd, and a full shell at Nh and Ag that can be breached, but only with great difficulty.
  7. We can simply insist that we are delimiting the differences in the d-block and make comparisons not to Fl through Og, but the other 6d transition metals, and then we find that Nh actually behaves reasonably similarly to the late 6d transition metals like Rg and Cn, but that Lr does not behave very "transition-y", and therefore that it is actually Lr and not Nh that must be excluded from the d-block(!!!!). And the means of discovery of Nh (bombarding Pb/Bi rather than actinides) is similar to how the late 6d metals were discovered, but not how Lr was discovered, by bombardment of actinides which is more like the means of discovery of Fl through Og(!!!!!).
  8. So the only difficulty that what appear to be Sandbh's current criteria cannot so easily sweep aside is that Lr is a lot more electropositive than Nh (which in the current criteria would be distorted into "ionicity"). But we can invoke Mendeleev, who originally put Tl as a heavier congener of the alkali metals because of its higher electropositivity in the +1 state, which is the more stable one! And therefore Lr as a "trivalent pre-transition metal" could be argued to fit better with this new and improved group 13 element thallium than Nh, for which the +1 state is amphoteric and the behaviour is more similar to Ag and At!

So, it looks like the case for Tl-Lr is holding up reasonably well under these criteria! Only one problem: how on earth do you draw it?

And if the biggest difference between accepting Sc-Y-La-Ac but not Tl-Lr is "but we can't draw the latter", then we have to abandon the line about drawing Nature as She really is. Double sharp (talk) 12:05, 27 January 2020 (UTC)

Easy. Just put Hf-Tl over Th-Lr, but stretch out Hf-Tl so that Tl and Lr are together. Do something like the atomic flower structure. And then fold the table up in such a way that Lu and Lr are still on top of each other as well as Tl and Lr, and also put Lu under Y and also not under Y because Group 3, combined with all of the problems hydrogen has... it's nature's perfect periodic table! ― Дрейгорич / Dreigorich Talk 00:01, 28 January 2020 (UTC)
If it helps, Lr is likely to have a d rather than p electron in the condensed phase. Lr 3+ will be f14, analogous to Lu. Sandbh (talk) 05:22, 28 January 2020 (UTC)
So condensed-phase configurations are OK when they give the result we want (Lr under Lu), but not when they don't (Al over Sc due to partial p-occupancy in Sc, Be and Mg as p-block elements)? Double sharp (talk) 07:48, 28 January 2020 (UTC)
@Double sharp: Sc is predominately (> 85% in this case, noting I haven’t read the article yet) a d metal That’s all that matters. Just as La goes under Y, and Ac goes under La, we’d expect Lr to go under Lu. For sure Lr has a p electron, which looks odd until we consider its condensed phase configuration, which is expected to have a d electron, and we see that Lr 3+ will be f14, analogously to Lu. Sandbh (talk) 10:58, 28 January 2020 (UTC)
In that case He goes over Be, since He is 100% an s-element, and Be is more of an s-element than Ne. And Be and Mg go over Zn, because all three are sp in the condensed phase (see the old chat at our old group 3 submission), which makes it inconclusive with Ca (noting that Ca has some pre-d character), and then the "similarity argument" argues for Be-Mg-Zn because Be and Mg have chemistry that is not very characteristic of Ca through Ra and more like Zn and Cd. Double sharp (talk) 11:29, 28 January 2020 (UTC)
@Double sharp: No. I don't change the goal posts. I observe the ones that are already there, as required, that you overlook. The periodic table layout is more pragmatic than saying He goes over Be since He is 100% an s-element. It's also 100% a noble gas. Be and Mg don't go over Zn for the reason that group 2 is a better choice given the commonality of underlying cores, among other things. Sandbh (talk) 06:28, 1 February 2020 (UTC)
@Sandbh: OTOH helium also 100% doesn't fit in the noble gas trend. Just plot properties, like in the article you linked for me. And I don't mean "doesn't fit" in the sense of O, F, and Ne with the first-row anomaly, I mean literally "He is not even trending in the right direction". And, nota bene, when helium is coaxed into compounds, the theoretically modelled ones are homologous to beryllium and certainly not neon. Which is why I prefer to say: let us stick to something that totally affects chemistry, i.e. chemically active subshells. If it produces He over Be, well, there are good reasons for that, so I don't complain.
As for Be and Mg over Zn: if we argue "commonality of underlying cores", then we cannot have B and Al over Ga (because Ga has the extra 3d10), and we cannot have Ti and Zr over Ce and Th (because Hf and Rf have the extra f14; Th even has the right configuration to not look like an anomaly here). If you want to specify where "commonality of underlying cores" is desirable and where it is not, you basically end up reinventing the Madelung rule to tell you when the underlying core should change, presumably messing with it at La in order to get the result you want. But if we argue "commonality of chemistry", like how you like to argue about group 3 being more like groups 1 and 2 than 4 and 5 (which is in itself not quite true anyway), then Be and Mg go straight over Zn! That's why I think your arguments are weak. The moment you try to apply them as a basis for most of the table, suddenly they all start fighting against each other. Mine doesn't: it just steers pragmatically all the way to the undiscovered element 172. And an attack of sanity should make it work for element 173, as well! Double sharp (talk) 10:22, 1 February 2020 (UTC)
@Double sharp: Ah, well, I was watching the tennis when I typed that. By 100% noble gas I meant it is effectively recognised 100% as a noble gas, nothing more detailed than that. We have B and Al (p elements) over Ga (also a p) because over Sc (a d element)is worse. See also the new section about what Main Groups are. Ti and Zr don't go over Ce and Th in a within same group congener sense, so there's nothing to that observation. I argue commonality of ionic v covalent chemistry, that is all. This is not the same as Be and Mg going over Zn. As I said, see the new section about what Main Groups are. The Be and Mg over Zn argument was lost a long time ago. Remy (1956) says the AEM illustrate the rule that the first element (Be) is apt to constitute a transition to the next Main Group, the second element (Mg) to the Sub-group belonging to the same family, whereas the group character is full developed for the first time in the third element (Ca). That's another pattern that works quite well with the Main groups as they are. I posted about this rule earlier. Sandbh (talk) 11:20, 1 February 2020 (UTC)
Hydrogen is effectively recognised 100% as not an alkali metal, yet it goes at the top of group 1 anyway, so there is more to it than that. "Ti and Zr don't go over Ce and Th in a within same group congener sense, so there's nothing to that observation." – says who? How do you know they're not in the same group? You must have some criteria, or you can take anything as your starting point and defend it after the fact. "Commonality of ionic v covalent chemistry" supports Be and Mg going over Zn because Be and Mg are only slightly electropositive, not highly, which is similar to the Zn group but not the Ca group. A viewpoint being "lost a long time ago" simply suggests that any argument that leads to it is suspect, no matter what else it happens to say. P.S. Remy's rule only really applies much for groups II and III, far from a majority. In what sense is Na a transition to the Cu group, or Si a transition to the Ti group? The character of Si vs. Ge is very similar, as is Na vs. K. Double sharp (talk) 11:35, 1 February 2020 (UTC)
@Дрейгорич: Put helium over beryllium as well as over neon and you have a follower. ^_^ Maybe add a bifurcating group 2 and trifurcating group 3 for that matter! Double sharp (talk) 13:55, 28 January 2020 (UTC)
Why wouldn't I? I was proposing we go all in and wrap the transition metals around the main group elements, in the original Mendeleevian style. ― Дрейгорич / Dreigorich Talk 14:07, 28 January 2020 (UTC)
@Дрейгорич: Perfect! So all the A and B groups make their triumphant reappareance! H and He shall also appear in two spaces, as shall Cu-Ag-Au, Be-Mg, B-Al, and Sc-Y! For good measure, let's toss in all the other secondary-periodicity things from Rayner-Canham's table! What a sight it would be to behold, and how singularly unhelpful for the students taking their first chemistry class! ;) (But a remarkable sight anyway!)
Seriously, this is more or less why I advocate He-Be-Mg with Sc-Y-Lu. There are so many second-order deviations from the ideal n+l arrangement and these are just two of them, so we might as well be consistent and just draw none of them, not just pick and choose only the ones we can draw in a tabular arrangement. Double sharp (talk) 15:06, 28 January 2020 (UTC)
@Double sharp: Ahahahahahaha, yep. This is why there will never be the ideal periodic table, and all attempts must be a compromise, like map projections. There is no ideal map projection. No matter how we slice and dice it to present it to students, there will always be oversimplifications. Different projections may work for different tasks, but as for the one we've got right now, it seems to do its job well at explaining the basic properties of elements, maybe with the exception of a nonmetal being on the "wrong" side of the line (hydrogen). ― Дрейгорич / Dreigorich Talk 15:11, 28 January 2020 (UTC)
@Дрейгорич: Let helium join it; then it won't be alone on the wrong side. ;) Now we can nicely separate the s- and p-blocks, let each one have its characteristic (1s vs. 2p) anomaly, let the Lewis doublet stand apart from the octet, and the trends will be nice and consistent! I recommend particularly the articles of Grochala and of Furtado, De Proft, and Geerlings (the latter does not explicitly call for He over Be, but you can see all the plots showing He simply out of place from the noble gas trend in many basic properties such as electronegativity, hardness and IP; this is distinct from Ne, which shows the right qualitative trend but a quantitative anomaly with a big 2p-3p gap just like what heads groups 13 through 17). Double sharp (talk) 18:41, 28 January 2020 (UTC)
@Double sharp: Nah, helium's physical properties scream "noble gas". How could it join the solid alkaline earths? Sure, it's an s element, but an anomaly from the main group of s elements (along with hydrogen). Maybe in some ways helium is better above the nobles (especially in its normal chemical bonding behavior and physical properties), but in others it fits better above the alkalines (periodic trends, though some trends like boiling point are clearly better suited for He above Ne). As for H, I'm torn (there's so much going on that H deserves its own unique group in the eyes of a beginning chemist), and I'm indifferent on what's below Y. The rest of the table is pretty much settled in my opinion, but then again, stuff.
What if beginning chemistry students are taught the standard 18-column table with a few anomalies, and then they get introduced to the twisty loopy place every element below every other element monstrosity I proposed above in advanced chemistry to try and show that this is how it really is, with branching chemical paths? Now the advanced students hate the beginning students for not initially seeing the whole scheme of things and having misconceptions about the periodic table, and the beginning students hate the advanced students for having an overcomplicated periodic table to fit every minor detail in. Win-win. ― Дрейгорич / Dreigorich Talk 23:55, 28 January 2020 (UTC)
@Дрейгорич: I heartily agree that the average lack of chemistry of He is more like Ne than like Be. :) However, there is one little wrinkle: it's predicted to be possible to coax He into compounds, and theoretical modelling has identified some possibilities (see Grochala's article). Not only do these not have Ne analogues (even predicted; Lewars' Modeling Marvels has a chapter on He compounds that notes that Ne seems to be surprisingly barren), but the bonding in them shows He as a lighter congener of Be, with a strong affinity to oxygen and a He-O bond where oxygen seems to be the more electronegative of the two. ;) So maybe the basic first table should float H and put He in group 18, but after a while when you're ready they should magically float off to groups 1 and 2. ;) (I recommend using He as your balloon when doing this floating, not H. ^_^)
I also agree that H-Li and He-Be are much harder to defend physically than chemically. However: every main group seems to start with a nonmetal, that shielding effects then metallise when you go down the table. But only with H-Li and He-Be; otherwise Be is already a metal, and the s-anomaly for Be is not of the same order as for H. Double sharp (talk) 08:09, 29 January 2020 (UTC)
this is going on for far too long Lawrencium. Back to lawrencium. It's pretty settled that Lu goes above Lr (Tl above Lr would probably be silly, right?), but the question is what's below Y. It doesn't help that La and Ac have a d electron (Th has two and no f electrons!) and the early lanthanides and actinides resemble group 4, 5, etc. which is a solid case for La/Ac. However pure Aufbau suggests Lu/Lr is better, and Lr has an anomalous electron configuration no matter where we place it. And Lr might not be the least of our worries once relativistic effects kick in and might disorganize Nh-Og. Maybe Lr is a warning that the order of the earlier atomic numbers is ending, and to be cautious in how to proceed. Maybe Lr will end up like H, and belong in multiple places, maybe one of them below Tl. For now we don't know. And I'm talking WP:OR without backing anything up, so feel free to dismiss me. ― Дрейгорич / Dreigorich Talk 08:24, 29 January 2020 (UTC)
@Дрейгорич: The early actinides do have some resemblance to transition metals; Ac to Pu have resemblance to Lu to Os. And if you squint you may consider Am and Cm, though especially for the latter this is no longer characteristic. (Nota bene, this is OK as it is a complex of properties, not just one.) For the Ln, not really, apart maybe for Ce a little, but they are anomalous as 4f has no radial nodes; 5f is more like typical f-block behaviour. The fact that the f-block is mostly intermediate between the s- and d-blocks looking holistically at all properties suggests to put it between them (i.e. Lu under Y) and not sandwiched only by the d-block (i.e. La under Y). Lr has an anomalous configuration that seems to mean nothing chemically, so I advocate considering the slow start of 5f in Ac-Th-Pa as the same effect as the slow start of 6d in Lr-Rf-Db and the slow start of 5g in E121+ and not drawing any of them explicitly. Periodicity works almost perfectly up to Rg, probably. It's Cn through Og which are really weird: if Au through At are "super-B" or "C" metals, then Rg is too, and Cn through Og are "super-duper-B" or even "D" metals. Ts and Og are probably actually similar to Ga and Sn respectively, with Cn kind of like Rn and Nh kind of like At, so everything has gone a bit bonkers here. But since the Aufbau principle should be totally valid with no exceptions from Rf through E120(!!), I think it's clear that it's folly to argue about Ac from exceptions, when Lr has an exception with no effect, but Cn through Og have no exceptions but huge effects! Double sharp (talk) 11:34, 29 January 2020 (UTC)
@Double sharp: Fun convo. Learned something new about the superheavies. Best to treat Lr as regular, and not make any assumptions past Rg. It will be very interesting to see where Cn-Og end up in the periodic table in ten, twenty, thirty, fifty years. Will they still be in order like we have now or will it have been better to disorgnanize them to better match their chemical properties? (Rg-X-Ts-Og-Mc-Lv-Nh-Cn/Fl? Jeez that looks weird.) If so, will chemistry teachers just cut the periodic table off at Rg? Will Cn-Og or whatever the last element is at the time be added as asterisks? Who knows. Also, if Deep web is still around, when will that conversation happen? ― Дрейгорич / Dreigorich Talk 00:45, 30 January 2020 (UTC)
@Дрейгорич: If the island of stability does not permit long-enough half-lives for Cn through Og, no one will care, I expect. If it does, then I suspect people will just continue to draw them below Hg through Rn for lack of anything better that doesn't look crazy. The fundamental problem is that you have two strong (6d and 7p1/2) shell closures in a row followed by a weak one (7p3/2), and this is a situation which has no good precedents. (The p-subshell splitting is not so strong for Pb. Po, At, and Rn have some problems getting to high oxidation states, but this is a well-known effect of radioactivity, and it seems likely that just like for Pb, PoVI for instance would be stabilised in organopolonium compounds.) So I suppose that we will have to rationalise it as a super "inert pair effect", that for Mc through Og becomes an "inert quartet effect". Indeed, in period 8 this is expected to be so big that the groups are quite literally staggered by four. A simple Aufbau extrapolation would give E163 and E168 as the p-block elements of the 8th period. But it is expected that E167 to E172 should, because of a super inert quartet effect (on 8s+8p1/2 rather than 7s+7p1/2 this time), become almost perfect members of groups 13 through 18, completely analogous to indium through xenon, even though their atomic numbers are four out! This situation in which E157 through E172 are expected to have a miraculous cancellation of relativistic effects and mimic Y through Xe is really weird. (How I wish we could get there soon! But probably not. T_T) Double sharp (talk) 14:33, 30 January 2020 (UTC)
@Double sharp: Re: "The fact that the f-block is mostly intermediate between the s- and d-blocks looking holistically at all properties suggests to put it between them (i.e. Lu under Y) and not sandwiched only by the d-block (i.e. La under Y)."
The sandwich is only seen in the 32-column form, which is a distraction compared to the 18-column form. Getting rid of the sandwich risks being seen as a false sense of precision or idealisation. Sandbh (talk) 06:41, 1 February 2020 (UTC)
No, even in an 18-column table you have asterisks interrupting the d-block. The periodic table is already an idealisation and everyone who looks deeply at the elements knows it. I claim that a La table, in singling out only one among many second-order anomalies to draw, muddies the waters about how precise it is supposed to be. Double sharp (talk) 09:53, 1 February 2020 (UTC)

@Дрейгорич: When you said earlier, "the early lanthanides and actinides resemble group 4, 5, etc. which is a solid case for La/Ac" is that something you had formed a personal view about? Sandbh (talk) 07:00, 31 January 2020 (UTC)

@Sandbh: Oxidation states. +4 for Ce, +5 for Pr, etc. though this may be a weak argument, given that most lanthanides have a +3 oxidation state or are primarily a +3 oxidation state. Surely Mendeleev would have approved based on typical oxidation states. Too bad of the four elements in controversy, only lanthanum was known. If more were known and the list of lanthanides were more complete, what would Mendeleev have done? ― Дрейгорич / Dreigorich Talk 07:21, 31 January 2020 (UTC)
@Дрейгорич: In fact, we know very well what he would have done, since his last table dates from 1906, the year Lu was discovered. (And also the year before his death.) The 6th period as we have it now stands as Cs, Ba, La, Ce, and then many blanks because the rare earths obviously don't fit the transition-metal pattern – and then Yb in group III (a reasonable mistake), a blank in group IV (later Hf, following his original 1869 prediction), Ta, W, and so on. So, the "redundancy" that Sandbh seems to dislike of La and Ac aligning under Lu and Lr in fact goes back to Mendeleev himself. ^_^
P.S. +5 for Pr is, as I said, just plain silly. We can talk about it again when it is found in something other than matrix isolation. Double sharp (talk) 11:49, 31 January 2020 (UTC)
@Double sharp: Start talking; it's been found in the gas phase. Sandbh (talk) 02:43, 2 February 2020 (UTC)
@Sandbh: Poor choice of words, sorry. I mean "when it is found in something resembling a normal chemical environment". But OK, I'll accept Pr(V), since otherwise He compounds become dodgy (I have referred to them; of course a stronger argument for He atop Be is atomic property trends), and I want to avoid a double standard. (I already dislike seeing it used to accept PrV and ThIII as conclusive and bar TiIII and ZrIII, so I will accept them all for now, plus predicted ArII and maybe even HeII.) I still don't see how it supports a La table, since the same argument I gave holds whether or not we accept it. Double sharp (talk) 10:04, 2 February 2020 (UTC)

Transition character of early f-block elements

@Дрейгорич: @Double sharp: I'd forgotten about the early Ln and An resembling group 4, 5, etc. "which is a solid case for La/Ac", as Дрейгорич said. The resemblances are Ce +4 and Th +4 to Group 4; Pr +5 and Pa +5 to Group 5; and U +6 to Group 6. Here's an La table with the relevant electron configurations:

1 2 3 4 5 6 -----+----+-------+-----------+-----------+--------- 6 Cs | Ba | La d1 | Hf d2s2 | Ta d3s2 | W d4s2 --> Rn -----+----+-------+-----------+-----------+--------- 7 | Rf d2s2 | +-----------+ +-----------+-----------+ 6 | Ce f1d1s2 | Pr f3s2 | --> Lu -----+----+-------+-----------+-----------+--------- 7 Fr | Ra | Ac d1 | Th d2s2 | Pa f2d1s2 | U f3d1s2 --> Lr -----+----+-------+-----------+-----------+---------

In the above table I can see that La and Ac are d elements. The f block starts at Ce, with the first appearance of an f electron. Th is anomalous but we know the condensed form has about ½ of an f electron; this impacts its crystalline structure; and there is an f electron in Th3+. The intrinsic Ln contraction (Ce3+ to Lu3+) is wholly contained within the f-block.

Here's an Lu table:

1 2 3 4 5 6 -----+----+-------+-----------+-----------+--------- 6 Cs | Ba | Lu d1 | Hf d2s2 | Ta d3s2 | W d4s2 -----+----+-------+-----------+-----------+--------- 7 | Lr p1 | Rf d2s2 | +-------+-----------+ +-------+-----------+-----------+ 6 | La d1 | Ce f1d1s2 | Pr f3s2 | --> Yb -----+----+-------+-----------+-----------+--------- 7 Fr | Ra | Ac d1 | Th d2s2 | Pa f2d1s2 | U f3d1s2 --> No -----+----+-------+-----------+-----------+---------

In the above table Lu is regarded as a d block metal, as is Lr. Lawrencium is anomalous but is likely to have a d electron in its condensed phase. It is notable that the low ionisation energy of Lr has been attributed the presence of this p electron. La and Ac, which line up under Lu and Lr, are regarded as starting the f-block but neither have f electrons. Why La and Ac aren't regarded as d metals isn't apparent. The intrinsic Ln contraction starts at the second metal in the f-block (Ce), and finishes in the first Ln in the d-block (Lu).

I can't quite put my finger on it but there is something peculiar or redundant about lining up the first Ln (La) under the last one (Lu). We already know +3 is the most common and stable state across the Ln. It reminds me of a snake eating its own tail. Sandbh (talk) 04:34, 30 January 2020 (UTC)

@Sandbh: Very interesting table, although still having two versions doesn't necessarily solve La/Lu but instead just rewords the problem, trying to offer a compromise by putting both in group 3 while still leaving the spot between Ba and Hf ambiguous. I personally think of the lanthanides like a 14- or 15-element long string below Y that can be taken out of its spot, stretched out and presented if needed. They all sit in one spot for me, in a line. Preparing my brain. Lanthanum lies hidden (pun intended) in its spot below Y, cerium falls out of that spot a bit, trying to get below zirconium but failing, praseodymium follows, and the line loops out of the main table doing its own lanthanide thing for some time before it curls back in with thulium and ytterbium going to head back into the spot below Y, and at the end lutetium sits below Y. It's like a long string that is tied with Ba at one end and Hf on the other, and the entire thing falls out of its designated spot. Both lanthanum and lutetium however claim sole ownership of the "ambassador" of that spot on the periodic table, and in the end no one can agree. The actinides follow the same thing but exactly mirror the lanthanides so there's no disagreement on who's related to who (Ce-Th, Pr-Pa, Nd-U... Yb-No). ― Дрейгорич / Dreigorich Talk 05:38, 30 January 2020 (UTC)
@Sandbh: On the contrary, this line-up is exactly why a Lu table is superior on counting extra homologies. In a Lu table, not only are Lu and Lr in group 3, but La and Ac are also aligned in the footnote below group 3, which is more or less right: La and Ac are useful "extra group 3 members" for comparative chemistry. In a La table, La and Ac are in group 3 of course, but if you want to keep the homologies of Th through Pu, then Lu and Lr appear under the totally irrelevant group 17. And just plot properties along the Ln correlated with the number of f-electrons. Of course you will get a snake eating its own tail as the effect grows on the first half as the f-electrons go in, and shrinks on the second half as they get paired. Nothing new here.
We have beaten the horse about 4f and 5f for La and Ac to death: suffice it to say that they have more such involvement than Lu and Lr. (Do you know the 5f energy in Lr? Consider for a moment how Fm, Md, and No increasingly prefer not to dig into that reserve...) And as for your "intrinsic" Ln contraction (I see that now there is an adjective), you cannot even apply it to the An where there isn't a unifying standard oxidation state, so there's no a priori reason why we should reject an f-block beginning at La and Ac. Otherwise I could equally well say that the d-block cannot begin at group 3 because Sc3+ has no d-electrons.
Also, you referring to Th3+ (not even known in aqueous solution!) and Pr5+ (known in exactly one matrix-isolation compound!!), while calling me out for references to supposedly uncharacteristic Zr3+ and Ti3+(!!!, which should be more on the ionic side) is a double standard if I ever saw one! Double sharp (talk) 08:13, 30 January 2020 (UTC)

@Double sharp: Interesting. I don’t count La and Ac being lined up under Lu Lr as an extra homology due to its redundancy. I also have concerns about the block membership confusion issue; and the associated issue of an intrinsic Ln contraction over two blocks (f, d) rather than one (f).

I recall plots of Ln properties are inconclusive.

In the particular An contraction context I raised, your comment is not relevant. The contraction can be seen in the An as per the Nature article. Like I said, the Ln and An have the +3 oxidation state in common.

The d block can begin at Sc since it has a d electron and this at least impacts its physical properties (per the G & E Group 3 commentary).

My context for referring to Th 3+ and Pr 5+ was not the same as the context for our discussion on Zr 3+ and Ti 3+. Funny :), I recall you tipped me as to the existence of Th 3+ in our IUPAC submission! I was tipped off to Pr 5+ by our list of oxidation states of the elements template. Interestingly, one of the articles on Pr 5+ notes long-standing speculation as to its existence. Sandbh (talk) 11:08, 30 January 2020 (UTC)

@Sandbh: There is no "block membership confusion issue". A Sc-Y-Lu table makes the scientific statement that Lu is more of a d-block element and La is more of an f-block element, between the two. I hope we agree that 4f in Lu, and even more so 5f in Lr, is more or less a core subshell, so there's nothing wrong with that; high coordination numbers can anyway be explained away with bond orders less than unity. (If you needed proof of that, for Lr, just look at the 5f energies going from Es to No, so much so that No has +2 as the preferred oxidation state). OTOH, you cannot "explain away" cubic complexes for La so easily. On symmetry grounds you really need f-orbital involvement from that. I seem to recall Gschneider has been an advocate for 4f character of La, that we did not wholly refute in our submission! To quote it: "These effects largely encompassed thermodynamic properties, and the stability constants of rare earth EDTA complexes. While we found evidence for the former effects to be plausible..." (my new emphasis). I absolutely agree with the conclusion we made: "We think it plausible that the low-lying 4f levels in La may influence some of its properties. It is also conceivable that the filled 4f shell of Lu may influence some its properties but, if so, the scope of this influence is likely to be smaller and more obscure." In other words: Lu has significantly less 4f involvement than La, so if it comes down to that, it is La that must stand as the f-block lanthanide. That is why the f-block may begin at La even without a 4f electron there. Otherwise, the d-block cannot begin at Lr.
(P.S. If you want a block membership confusion issue, try drawing He above Ne. ^_^ You cannot argue that He is anything other than an s-block element, so the difference is whether you draw it there or not. And the placement of He over Ne at least has something more than the Pr5+-level of support: the atomic properties of He very often just do not fit in the noble-gas trend, creating an anomaly on top of a perfectly orthodox first-row anomaly from Ne to Ar.)
Again, you are misunderstanding what I am saying. I am not comparing the Sc-Y-La to the Sc-Y-Lu trend. I am noting the trend across the lanthanide series. In every case where 4f electrons matter and the lanthanides are in the +3 state, the trend will follow the number of 4f electrons from a minimum at La to a maximum somewhere in the middle to a minimum at Lu again. (Of course, the trends go all over the place once the maxim of "constant +3 oxidation state" is violated.) So of course it is like a little circle running in place and the supposed redundancy is not only justified, it also reflects perfectly what we see by giving equal visual weight to the equally plausible trends of Sc-Y-La and Sc-Y-Lu a priori. That is why, while advocating Sc-Y-Lu as a default, I advocate that La and Ac should be included in a comparative chemistry discussion of group 3. But I also think that Al should be included too. ^_^
"The Ln and An have the +3 oxidation state it common": funny, the most common actinides for anybody are thorium and uranium. And for thorium the most common state is +4, and for uranium they are +4 and +6. In fact, throughout most of the first half of the actinide series, +4 might be a better "constant state" to take as the baseline. In the second half +2 gains in importance. So where is this +3 that is had in common? It seems to be about as consequential as the observation that most of the 3d metals have +2 and/or +3 as the significant state. Is that relevant at all for the 4d and 5d ones? The contraction is only significantly visible if everybody involved is in the same oxidation state. And once you force that, it is not special to the lanthanides, it is visible literally everywhere in the periodic table by first-year school chemistry. But then you cannot point to a special oxidation state to force where the contraction "should" begin. What is there to choose between Sr2+ through Pd2+ as your contraction and Y3+ through Ag3+ as your contraction? In terms of choosing common oxidation states they are equally bad, just like +4 vs. +3 vs. +2 for the actinide contraction.
I don't disagree that ThIII is of some relevance, sure. It certainly shows 5f involvement in Th, as does its crystal structure. However, it seems to me that TiIII and ZrIII are even more relevant. You cannot deny the relevance of those two when you argue about group 4's "ionic vs. covalent" behaviour (once you define what that is supposed to mean), and yet accept the relevance of ThIII when arguing about the slow start of 5f activity, when the former are actually more common oxidation states! Ti3+ is stable in water, Zr3+ reduces water, but Th3+ has not even been confirmed in water at all! Even your article refrains from drawing a tripositive thorium ion when showing the An3+ contraction (for which, at least from Th to Pu, we are comparing elements that are unhappy to be in the +3 state, a far cry from what we do when invoking the Ln3+ contraction!)
As for PrV: you have got to be kidding me. There must be thousands of things that have a long-standing speculation of existence that are nearly totally irrelevant. HgIV, for instance. Respectfully, I put it to you that an oxidation state only seen in one Pr compound, that is only stable in matrix isolation, is not at all a demonstration that Pr has crypto-group 5 properties when in every single possible way it will not follow the chemistry of V, Nb, and Ta. Double sharp (talk) 14:17, 30 January 2020 (UTC)
P.S. If the f-elements have some s-character and some d-character, surely this is evidence that the f-block should be drawn between the s- and d-blocks, and not sandwiched inside the d-block, no? Double sharp (talk) 14:26, 30 January 2020 (UTC)
Yes, and perhaps this is why Lu at least theoretically has a better shot of being below Y than La. Maybe the answer to group 3 should be "how does the d block fit between the s/f (depending on if extended periodic table or not) and p blocks, and by analogy, how should this indicate the f block? ― Дрейгорич / Dreigorich Talk 16:33, 30 January 2020 (UTC)
@Дрейгорич: Well, the early d-block groups (3, heavy 4, heavy 5, heavy 6) have similarities to the s-block and f-block (consider Th/Pa/U vs. Hf/Ta/W), and the late ones (12, somewhat 11) have similarities to the p-block, so it works just as well. The groups on the edge of each block have some similarities to the adjacent block: again, this is totally normal in the periodic table, and is just another expression of how it is continuity, not discontinuity, that rules the day. As for the extended table: the 5g series seems to be something like a hexavalent 4f series, which would indicate similarities most of all to the s- and f-blocks (s- in the early groups; later, more similarities to uranium, and the border between 5g and 6f should not be all that clear). That seems to suggest to me that pseudohomology should work more or less all the way up to eka-Pu as E126 (using it in a loose sense; the element below Pu would of course be E148 instead), actually. Maybe at a second order you could see similarities to the d-block through uranium as well. Double sharp (talk) 19:18, 30 January 2020 (UTC)

@Double sharp: Righto.

The block membership confusion issue I was referring to is that in group 3 of the Lu table, Lu is regarded as a d block metal, as is lawrencium. La and Ac, which line up under Lu and Lr, are regarded as starting the f-block but neither have f electrons. Rather, they have d electrons. Why La and Ac aren't regarded as d metals isn't apparent.

It's perfectly apparent from the arrangement that we are claiming them as f metals with the "wrong" electron configuration. Just like thorium. Double sharp (talk) 12:08, 31 January 2020 (UTC)

There is no confusion about He. It's an s-block element placed over a p-block element on stronger resemblance grounds.

And it's the only element that is unarguably in one block and placed visually with elements of the wrong block. Double sharp (talk) 12:11, 31 January 2020 (UTC)

I like Gschneider. As a tipping point argument.

Well, since you yourself claim that the trends of Sc-Y-La and Sc-Y-Lu are inconclusive (thus we are at a tipping point), and you agree that 4f involvement being stronger in La than in Lu works as a tipping point argument, it seems that Lu is immediately recommended. ^_^ Double sharp (talk) 12:11, 31 January 2020 (UTC)

The d-block does not start at Lr; it starts at Sc.

The 6d row starts at Lr by most normal criteria, but not by yours, because it lacks a 6d electron. (Not that it seems to make it very much different from Lu.) So if you want to use the first element as a guideline, we run into delayed collapses everywhere, and the blocks stagger by periods. If that actually corresponded with a natural break in the chemistry of the elements that would be one thing, but it obviously doesn't because Lr is a strong homologue of Lu. Double sharp (talk) 12:08, 31 January 2020 (UTC)

I agree with you about trends across the Ln. These are inconclusive.

The whole point here is that they are inconclusive, which is precisely why it is useful and not "redundancy" to have La and Ac align in the footnote under Lu and Lr, just like Th, Pa, and U align under Hf, Ta, and W. Double sharp (talk) 12:08, 31 January 2020 (UTC)

On the An contraction it seems to me that you overlook my context or roll my argument into different contexts, which I didn't raise. My specific context was that +3 is a state shared by all La and An (Wiberg, p. 1645). Further, in this state, they show an Ln contraction (Ce to Lu) and an An contraction (Th to Lr). Both contractions start with f1 and finish at f14. On contractions generally, the Ln is the most notable. The scandide and boride contractions are of less significance.

If our standards are for a shared +3 state, never mind Th and Pa which are so unhappy in it (from compounds of thorium: "On 1997, reports of amber Th3+ (aq) being generated from thorium tetrachloride and ammonia were published: the ion was supposedly stable for about an hour before it was oxidised by water. However, the reaction was shown the next year to be thermodynamically impossible and the more likely explanation for the signals was azido-chloro complexes of thorium(IV)."), and U which oxidises water in it, then we can equally well say produce shared states across the 3d row. Well, apart from the early elements, +2 is common and stable for everybody, no? So I claim the actinide contraction is equally relevant as the scandide one: not very. The relevance of the Ln contraction is once again, an outlier among all contractions purely because of the common +3 state, which, I reiterate, is something you cannot claim for the An. Double sharp (talk) 12:08, 31 January 2020 (UTC)

You wrote:

" I don't disagree that ThIII is of some relevance, sure. It certainly shows 5f involvement in Th, as does its crystal structure. However, it seems to me that TiIII and ZrIII are even more relevant. You cannot deny the relevance of those two when you argue about group 4's "ionic vs. covalent" behaviour (once you define what that is supposed to mean), and yet accept the relevance of ThIII when arguing about the slow start of 5f activity, when the former are actually more common oxidation states! Ti3+ is stable in water, Zr3+ reduces water, but Th3+ has not even been confirmed in water at all! Even your article refrains from drawing a tripositive thorium ion when showing the An3+ contraction (for which, at least from Th to Pu, we are comparing elements that are unhappy to be in the +3 state, a far cry from what we do when invoking the Ln3+ contraction!)"

This is an example of mixed contexts.

The Th(III) context was the established f involvement in Th, and the An contraction. That was all. [For unstable Th 3+, Wiberg, p. 1719, says it has a deep blue colour in aqueous solution. See also this article for a reference to crystallographically-characterized Th(III) complexes.] This context does not having anything to do with the stand-alone argument that (a) the chemistry of group 3 is predominately ionic whereas that of group 4 is predominately covalent, nor does it (b) have anything to do with a comparison to the status of Ti 3+ and Zr 3+. The fact of the existence of Ti 3+ and Zr 3+ does not usurp my argument that the chemistry of group 4, as per the literature, is predominately covalent.

On Pr(V) it seems to me that you overlooked my context and rolled my argument into a different context, that I didn't raise. The only context is the existence of Pr(V) and its alignment under group 5. The first article on Pr(V) says, "We report the formation of the lanthanide oxide species PrO4 and PrO2+ complexes in the gas phase and in a solid noble‐gas matrix…thus demonstrating that the pentavalent state is viable for lanthanide elements in a suitable coordination environment." The second article notes that, "it has been postulated since the early 1900s that praseodymium, with five valence electrons and the lowest fifth ionization energy, could be oxidizable beyond the +IV oxidation state."

As for the significance of gas phase Pr(V) I can only note the amazement and publicity associated with the isolation of Ir(IX), which was also in the gas phase.

If the possibility of NG compounds had been assigned the insignificance you assign to Pr(V) I surmise we'd still be calling the NG inert gases, and there would be no NG chemistry.

Oh come on. There is such a big difference between a standard reagent in organic chemistry (XeF2) and compounds that are not even known as solid salts (IrO+
and PrO+
). You can bottle krypton and xenon compounds, and you can't do this with Ir(IX) and Pr(V). If Pr(V) is alone to justify Pr as a pseudo-d element (which is about as far from "characteristic behaviour" as you can get), then we can look at the group 4 elements in the +2 oxidation state, even. I bet they will be very much more ionic in that state purely from Fajans' rules. Why is "characteristic behaviour" suddenly important for "ionicity", but not at all important for "resemblances to other blocks", for which we may apparently point to any number of one-hit extreme-condition wonders? If you really force the conditions, potassium is a d-block element with a 3d1 configuration. Are we going to call it relevant like you do for Pr(V), or irrelevant like you do for Ti(III) (which is strangely enough a major oxidation state, despite your dismissal)? I stand by the characterisation of a double standard here. Double sharp (talk) 12:01, 31 January 2020 (UTC)

On the position of the f block, it is the f nature of the f block that is important, as is the case for the s, d, and p nature of the blocks of the same name. So the f-block fits between the s and d blocks. Sandbh (talk) 06:33, 31 January 2020 (UTC)

And it's only there with Lu in group 3. With La in group 3 it's sandwiched between two d-block groups. Double sharp (talk) 12:11, 31 January 2020 (UTC)
@Double sharp: I don't know what I meant to say there. What you said was, "If the f-elements have some s-character and some d-character, surely this is evidence that the f-block should be drawn between the s- and d-blocks, and not sandwiched inside the d-block, no?" My response should've been, no, all of the elements (but for Pd) show some involvement of s electrons, so the question is meaningless. Sandbh (talk) 00:56, 2 February 2020 (UTC)
@Sandbh: Poor choice of words, sorry: s-character means typical characteristics of s-block elements here. Double sharp (talk) 10:06, 2 February 2020 (UTC)

Null hypothesis

@Double sharp: What do you mean by this? Sandbh (talk) 22:56, 30 January 2020 (UTC)

@Sandbh: From the lede of null hypothesis: "In inferential statistics, the null hypothesis is a general statement or default position that there is nothing significantly different happening, like there is no association among groups or variables, or that there is no relationship between two measured phenomena." It is what you start with, and the burden of proof is to marshal evidence to show it cannot hold. If the evidence is not strong enough, we cannot reject the null hypothesis. Since the blocks fall out so well from H to Xe, the null hypothesis should surely be that the n+l pattern keeps going without change in the 6th and 7th periods. Since the Sc-Y-La vs. Sc-Y-Lu trends are inconclusive, and Lu leads to a more homogeneous d-block and better fits the concept of blocks in the first place (4f involvement is significantly stronger in La than in Lu), the strength to overturn the Sc-Y-Lu null hypothesis is just not there. Double sharp (talk) 23:37, 30 January 2020 (UTC)

@Double sharp: If I understand the first part of your edit right, the null hypothesis (H0) is that there are no relationships among the elements.

Like DIM, evidence is then gathered in attempt to show that H0 does not hold.

I don't understand what you mean when you say, "the null hypothesis should surely be that the n+l pattern keeps going…". You seem to be saying that H0 is something (i.e. the n+l pattern) whereas, per our H0 article, "the null hypothesis is a general statement or default position that there is nothing significantly different happening."

I'll pause there and await your response. Sandbh (talk) 02:29, 31 January 2020 (UTC)

@Sandbh: Note the phrase "there is nothing significantly different happening". In other words, we have a clearly defined n+l law that is valid for H through Xe, and the hypothesis we are trying to nullify (the null hypothesis) is that nothing significantly different happens. Previously, indeed, we started with a null hypothesis that there was no correlation between the elements, and H through Xe was enough evidence to refute that and go to the alternate hypothesis that the n+l rule holds. So now past Xe we have a new one that says "well, if it holds so far, it should keep going, right"? Your alternate hypothesis is then that Sc-Y-La is significantly better than Sc-Y-Lu, and it clearly doesn't reach that level of difference. Double sharp (talk) 12:19, 31 January 2020 (UTC)

@Double sharp: OK. The n + l rule is not valid for H through Xe, given the anomalies. And what did you mean by the "two rows at a time hypothesis" up to row 5? Sandbh (talk) 01:59, 2 February 2020 (UTC)

@Sandbh: You focus on the minutiae of anomalies, I focus on the big picture of chemically active subshells noting that mostly the elements are in excited states when chemically bonded, and yet you accuse me of focusing on details instead of your broad-contour philosophy? ^_-☆ "Two rows at a time" just means the shape of the table as it is from H to Xe. Double sharp (talk) 10:10, 2 February 2020 (UTC)

@Double sharp: Why is the Lu table the null hypotheses rather than the La table which is the most common form by a wide margin? Presumably Jensen started with the La table as Ho. Since when did the Lu table become Ho? Sandbh (talk) 04:43, 2 February 2020 (UTC)

@Sandbh: For the same reason the 8-column table is not the null hypothesis? You don't choose your null hypothesis by tradition. You choose it by a simple assessment of what assumes nothing new is going on. If you want to do an experiment on average size of red dragons vs blue dragons, your null hypothesis is that there is no difference. Not even if the received wisdom of the bravest knights in the kingdom reports one. That received wisdom is what you are here to test. Double sharp (talk) 10:10, 2 February 2020 (UTC)

@Double sharp: I so far think the null hypothesis approach is a waste of time. But I will plough on. In the case of red v blue dragons, the null hypothesis is that there is no difference. How does this translate into La table v Lu table? You seem to have started with the Lu table, which is not analogous to a null hypothesis that there is no difference. Or are you saying there is no difference and that symmetry therefore prevails? Yes, I'm confused. Our article on null hypothesis is no clearer. Sandbh (talk) 04:30, 3 February 2020 (UTC)

@Sandbh: The whole point is that a Lu table is saying "there is no difference between elements before and after period 5, the n+l rule still works when examining chemically active subshells". (I have already refuted differentiating electrons too many times to count by now. That if anything is a "waste of time" given its chemical irrelevance.) And that's true. The evidence for a La table (mostly just inconclusive trends, now that everything stronger has been refuted by reduction ad absurdum) is not enough to topple it. Double sharp (talk) 11:50, 3 February 2020 (UTC)

General comments

1. Double sharp, you do not accept predominately ionic v predominately covalent as a relevant distinction. That’s OK. Broad categories such as these are good enough for classification science, as is the concept of predominant or typical behaviour such as acid-base, or metal-nonmetal, in chemistry.

@Sandbh: There is an essential difference between your broad categories: metal-nonmetal is based on a complex of categories together, which I accepted, while acid-base and ionic-covalent are categories that are continuous and are ill-defined without some sort of comparison. Now I have in mind aliens from some hellish planet from our perspective, praising hydrogen fluoride as their water of life, and regarding as self-evident the basic properties of what we call nitric and sulfuric acids. Double sharp (talk) 07:56, 31 January 2020 (UTC)

2. My reference to cats as an example of a natural kind attracted some criticism. There is nothing to argue about. I took this from the literature and will add the source.

Being from the literature does not give you a free pass for a bad argument, you know. ^_^ Double sharp (talk) 07:56, 31 January 2020 (UTC)

3. The composition of Group 2 (Be to Ra) and Group 12 (Zn to Cn) is effectively universally established and, as I see it, there is no need to revisit this.

Precisely, which is why you should look at it for fairness. If you have a new decisive criterion that strenuously pleads for a La table, but applying it to group 2 makes it plead for Be and Mg over Zn, surely this weakens its relevance if you think you know the latter should not happen. It's just like proof by contradiction.
What rule did I use to come up with these four sequences of three numbers each: 1-2-3, 2-4-6, 4-7-10, 8-15-22? Was it "the three numbers are natural numbers in arithmetic progression"? Not a bad guess. But if you keep asking me if such sequences fit the bill, and I keep answering yes, you have no information that you did not have already. You must try to prove yourself wrong. I bet you never would have guessed it, BTW: my rule was "pick any three complex numbers in order of non-decreasing modulus". But of course, would you have thought to test your criterion on something as crazy-looking as i, 4 + i, 17 + 2πi? And do you see why even that would not be enough to figure it out? (I forgot where I heard of this example, unfortunately, but it is a good one.) Double sharp (talk) 07:56, 31 January 2020 (UTC)

4. The "rule" of first row distinctiveness was first extended to 3d and 4f by Jensen AFAIK. Primogenic repulsion has a role to play but does not explain everything. H in Group 1 and He in Group 18 provides the strongest example of distinctiveness across the 1s row (H, He), and the 2s2p row (Li to Ne).

It explains so much in 3d and 4f that denying it is pretty silly, just read Kaupp's paper. He over Ne is not a very distinctive element in the "correct" way that H and the 2p elements are. He over Be stands in the same relation as H over Li. Double sharp (talk) 07:56, 31 January 2020 (UTC)
@Double sharp: When did I say I denied it? Sandbh (talk) 02:39, 2 February 2020 (UTC)
@Sandbh: Poor choice of words, sorry: I mean that you seem to be trying to downplay its relevance, judging from the scare quotes you put around the word "rule". It is not only primogenic repulsion, OK (since 2s shows a sort of "anomaly" that is similar to 2p as well), but that is obviously the most important factor. Double sharp (talk) 16:14, 3 February 2020 (UTC)

5. The differentiating electron criterion drew a lot of interest mainly along the line of the d/e in the d- and f-blocks making no difference to chemistry. I gave examples of where they made a difference. Will the p differentiating electron in Lr make a difference to the chemistry of Lr? This is a complicated question since it looks like condensed Lr will have a d electron. With a p electron, the ionisation energy of Lr is very low. Chemically, our own Lr article includes the following:

  • Lr behaving more like Cm, Fm, and No and much less that of Ru
  • Lr compounds should be similar to those of the other trivalent An
  • LrH2, is predicted to be bent, and the 6d orbital of Lr is not expected to play a role in the bonding, unlike that of LaH2
  • Molecular LrH2 and LrH are expected to resemble the corresponding Tl species (Tl having a 6s26p1 valence configuration in the gas phase, like Lr's 7s27p1) more than the corresponding Ln species
  • Lr may behave similarly to the alkali metals Na and K in some ways
  • Metallic Lr will behave similarly to Cm
  • If Lr has a p electron this is not expected to affect its chemistry in simple compounds.

Yuichiro Nagame , who was a member of the project team that measured the first ionisation energy of Lr, wrote that its place as an f-block actinide, a d-block transition metal or a p element, is yet to be unambiguously settled.

Since the only stable oxidation state is expected to be +3, I consistently don't really care about how the very simplest +1 and +2 compounds look. In that case Lr looks perfectly like a trivalent actinide and a heavier congener of Lu, in which the chemistry is not affected. Not only that, but the trend at the end of the actinide series supports divalency, with Lr as a weird addition (because it is not using a 5f reserve anymore), so the chemistry of Lr perfectly supports a 5f row from Ac to No. Double sharp (talk) 10:26, 1 February 2020 (UTC)

6. There were some perceptions that my article was unbalanced or biased towards La. That is fair enough and I need to consider how to address this in the final draft.

Sandbh (talk) 02:15, 31 January 2020 (UTC)

Default to lutetium and redraw all of the periodic tables?

Reading, er, glancing at this entire conversation (at least 50 thousand words and 100 pages of text so far), maybe we don't have enough evidence to overthrow the Aufbau Sc/Y/Lu/Lr arrangement. I don't know why Deep web was changed to Sc/Y/La/Ac other than "tradition" (I was much younger when that conversation took place). The conversations here are interesting, and I've learned quite a bit from it. Maybe it's time to redraw all of the periodic tables on Deep web? After all of this talk, aren't we overdue for a conclusion? ― Дрейгорич / Dreigorich Talk

While I'm rather receptive to the argument, I find it important to point a few things:
  • first, there is no known overreaching reason as for why Aufbau even is in the first place, which would make foundation of such a move on that basis alone rather wobbly;
  • second, the argument generally goes that group 3 is more like group 2 and 1 than 4 and 5, which makes it reasonable to have a group 3 that would retain trends that match those of groups 2 and 1 rather than 4 and 5.
As for me, I am not particularly impressed by these arguments, but I won't dismiss them, either. Different people have different opinions, and this question, in the end of the day, boils down to what opinion you have. The periodic table is first and foremost a great tool for describing the existing chemical elements but its predictive power is also impressive but not quite as much so, because further research into the periodic table extensions change not only the table but also on which principle the periodic table is based (Mendeleev arranged the elements by their atomic mass, later it was shown that it was the atomic number that mattered; Seaborg fantasized of a 218-element table that continue the current block trends, later calculations put the very notion of blocks at some point under question), and there is no exact agreement on what a periodic trend is (everybody thinks of more or less the same in general, but different people have different views of that). Sandbh used the notion of the periodic trends in this discussion in a way I would not, but while I disagree, I don't want to dismiss his opinion. Given the nature of a debate, I don't think a conclusion is in sight, and it would be great to have an external arbiter of sorts, and that's where the IUPAC taskforce come in. I'm sure that somebody will want to revisit the issue if they rule something different than what we have now, but right now doesn't seem like the time for such a change.--R8R (talk) 18:28, 31 January 2020 (UTC)
The discussion that led to this was at Template_talk:Periodic_table. You can read it to see the novelty of me fighting on the Sc-Y-La-Ac side. ^_^ (I have learnt several more things since then, in Archive 33, that make the arguments that I thought were decisive then look weaker, and that's why I shifted.) IMHO, as long as IUPAC has not come to a decision, Sc-Y-La-Ac should stay on WP. I do not think it is the ideal form myself anymore, but it is the most common form in the literature, and that's what counts for WP. We can and do, of course, discuss the issue where it is relevant. Once IUPAC makes a decision (which I hope will be Sc-Y-Lu-Lr), we should follow it IMHO, whatever it ends up being. (Now, if we can just convince them to start a task force about helium... ^_^)
We started discussing this here because Sandbh wrote an article about it externally, anyway; it didn't really have anything to do with WP. But I do love having a good argument about group 3! ^_^ Double sharp (talk) 19:54, 31 January 2020 (UTC)
P.S. The desired endpoint is to gain insight even if the conclusion is not yet in sight. ;) (With apologies to Richard Hamming.) Double sharp (talk) 11:41, 1 February 2020 (UTC)

P.S. My ideal periodic table at the moment would look something like my userpage. The positions of group 3 and helium have fluctuated a number of times in the past as my views changed. And I absolutely don't rule out changing it again. ^_^ But FWIW, this is how it would look, without those highlights that reflect my WIPs:

(Still not totally decided on whether "alkaline earth metal" should include Be and Mg. Or whether "transition metal" should include group 12. Copernicium makes the exclusion a bit iffy, but since it does follow the weirdness of 7p more than the orthodoxy of 6d, you could still support this arrangement.) Double sharp (talk) 21:37, 31 January 2020 (UTC)

@Double sharp: Be and Mg are not alkaline, so they should not be included into alkaline earths. Droog Andrey (talk) 17:54, 2 February 2020 (UTC)
@Droog Andrey: Well, that's good, since in this table I didn't. ^_^ Double sharp (talk) 18:41, 2 February 2020 (UTC)
P.S. Wulfsberg (see Principles of Descriptive Inorganic Chemistry, pp. 154ff.) breaks "very electropositive metals" as EN < 1.4, "electropositive metals" as 1.4 < EN < 1.9, and "electronegative metals" as EN > 1.9. This is, at least, a better form of "ionic vs. covalent". Though I note that despite him calling the first group groups 1-3 + f-block, Zr and Hf (as usual) satisfy that criterion as well, and in fact Be does not. But a very rough guide would be s+f vs. 3d+4p+5p vs. 4d+5d+6d+6p+7p, noting to put Be and Mg, Al, groups 3, 4 (minus Ti), and 12 with the neighbouring blocks here. Double sharp (talk) 21:52, 31 January 2020 (UTC)
Interesting. My own periodic table would not be divided like that, but go strictly by groups. Hydrogen would float above all of the others in its own color. The alkali and alkaline earth metals form groups 1 and 2 (no H or He), the transition elements (for me synonymous with "d block elements") and the inner transition elements (lanthanides and actinides), with Lu under Y. Lu and Lr are considered to be both inner and outer transition metals, so the categorizations could be merged (non-main group element?). After that would be the icosagens, the tetragens, the pnictogens, the chalcogens, the halogens, and the noble gases (with He). Nh-Og are not out of order. I don't know how you'd draw a periodic table like that, though. But that would be my categorization. ― Дрейгорич / Dreigorich Talk 00:13, 1 February 2020 (UTC)

Main groups, Be, Mg, and Al

@Double sharp: Here’s a further observation on why Be and Mg are in Group 2 and Al is in Group 13.

”The chemical behaviour of the elements of the Main Groups…is determined principally by [their] position with respect to the inert gases…Elements…with atomic numbers which are 1 to 2 units larger, or 5 to 1 units smaller than the atomic number of an inert gas are assigned to the Main Groups.”

Remy H 1956, Treatise on inorganic chemistry, vol 1, Elsevier, Amsterdam, p. 5

Sandbh (talk) 10:16, 1 February 2020 (UTC)

The last sentence is excellent for a treatise. OTOH, if you misread it as an argument for the position rather than as a way to remember it, it is hilariously circular because the atomic numbers he gives just define the placement we want. Why not 3 units larger or 6 units smaller, for example? That would include groups 3 and 12, and the latter is mostly main-group. Then the placement of Be, B, Mg, and Al suddenly ambiguous. And if we are playing the game of "look at history and the old treatises", crack open most of them from before WWII, and you will see Be and Mg discussed with the Zn group rather than the Ca group, as Jensen has noted. Double sharp (talk) 10:28, 1 February 2020 (UTC)

@Double sharp: The world has moved on from the days of Be and Mg over Zn. Remy adds: "Defining the main groups as has been done above, the same division of the elements is obtained as can be reached on grounds of [electronic] atomic structure." Precisely. Let's also laugh at the circularity of the n + l rule as well because it has no first principle basis and is based on empirical observations, inaccurate as the rule itself is. Sandbh (talk) 11:47, 1 February 2020 (UTC)

@Sandbh: By that logic I may start by axiomatically defining group 3 as Sc-Y-Lu-Lr. Which is the same division as obtained from electronic structure, due to the core-like f-orbitals of Lu and Lr. The n+l rule is simply a quick way to summarise how chemical relevance of orbitals really goes all the way from H through Og; even delayed collapses do not target chemical relevance (e.g. 4f for La) and never affect the entry into the core (e.g. 5f ends relevance at No still, and 6d at Cn still). As a statement of chemical relevance of orbitals, it is 100% accurate with only the Ca group as mild exceptions (which have to be, since all chemically accessible orbitals have azimuthal quantum number above 0). So I think we can look forward to the day when the world moves on from Sc and Y over La, too. As Droog Andrey has said, it is just a remnant of days when the f-block was unknown, and it is just a matter of time before the final flicker of that obsolete idea dies just like Be and Mg over Zn. Both obsolete Be-Mg-Zn and Sc-Y-La have some sense, but neither had the justification to overthrow electronic structure. Double sharp (talk) 12:18, 1 February 2020 (UTC)

@Double sharp: Nope. The division obtained from electronic structure points to Sc-Y-La-Ac, since the 4f shell does not start filling until Ce. The n + l rule, in this context, is nonsense since it doesn't reflect the situation on the ground. For that matter, when Janet proposed his LSPT in 1928 on the basis of the same value of n + l in each row, he thought that the measured electron configurations of the anomalous elements must have been wrong, so he "corrected" them to obtain a perfect LSPT. Yet another example of drawing the table like we expect it to be rather than how it really is. Sandbh (talk) 01:44, 2 February 2020 (UTC)

@Sandbh: You focus on the minutiae of exact ground-state configurations. I focus on the big picture of chemically active subshells. Chemically bonded elements are not exclusively in the ground state and unionised (mostly they are not), so your criterion is mostly irrelevant anyway for everybody but the noble gases. I acknowledge that in the ground state La has the "wrong" configuration. I don't care because 4f is clearly chemically active and within reach. For Lu it is not. Same for anomalies like Cr, Cu, Nb, Mo, etc. I'm astonished that you argue for the big picture and simultaneously keep talking about this minute thing. Double sharp (talk) 09:39, 2 February 2020 (UTC)

@Double sharp: Well, differentiating electrons only form the foundations of the periodic table so I’m not sure what planet you’re standing on, so to speak. Sandbh (talk) 11:57, 2 February 2020 (UTC)

@Sandbh: I reject that given their general chemical irrelevance. I propose instead Jensen's criteria, that are pretty much the same as the ones I have stated.
  1. Assignment to a major block based on the kinds of available valence electrons (i.e., s, p, d, f, etc.). [In other words, chemically active subshell of highest angular momentum.]
  2. Assignment of the elements within each block to groups based on the total number of available valence electrons. [In other words, Nb vs Ta does not matter.]
  3. Verification of the validity of the resulting block and group assignments through the establishment of consistent patterns in overall block, group, and period property trends. [Which they do, as in a Lu table group 3 follows the rest of the d-block, which is made more homogeneous.]
  4. Verification that the elements are arranged in order of increasing atomic number as required by the periodic law. [Which is why we allow some flexibility in period 8.]
Double sharp (talk) 12:25, 2 February 2020 (UTC)

P.S. This also explains why I believe helium should go in group 2. By its available valence electrons, helium must be an s-block element: there is no room for dispute. And it has to go in the s2 column, thus stuck with Be, Mg, and the alkaline earths. Then helium stands in relation to beryllium in much the same way hydrogen stands in relation to lithium. The trend looks like a typical first-row anomaly for both, so nothing forbids it, and elements remain in order of increasing atomic number. Of course we can talk about it with the noble gases anyway, but putting it above Ne as if it were a p-block element flies in the face of the importance of blocks as a fundamental criterion. It would be the only element clearly in one block placed among elements of another. We understand and agree with all the ways 1s is utterly anomalous for the s-block, and indeed for almost the whole table; but the periodic law should apply to everybody, and these should be second-order corrections to it, even if by far the largest of all of them. Double sharp (talk) 17:24, 2 February 2020 (UTC)

Yet another paper

2019, argues for 15-element f-rows because of inconclusiveness. They find Lu and Lr with core-like f-orbitals following 18-electron rule in the compounds they studied (sticking La/Lu/Ac/Lr into Zintl ion clusters Sn2−
and Pb2−
). La and Ac are also 18-electron here. So we're back to "weak 4f for La vs core 4f for Lu". Since I consider 15-element f-rows unacceptable, we are still in the same situation, which I argue supports Lu. Double sharp (talk) 10:51, 1 February 2020 (UTC)

@Droog Andrey: What do you think of the authors' methodology? Double sharp (talk) 10:52, 1 February 2020 (UTC)

@Double sharp: They found some clear similarities between La-Ac and Lu-Lr, but that is surely not enough to put them both into the f-block. Droog Andrey (talk) 17:58, 2 February 2020 (UTC)
@Droog Andrey: I agree. ^_^ That La-Ac and Lu-Lr would have similarities is pretty much a given due to their electron configurations, but the f-block ought to have only fourteen columns. Double sharp (talk) 19:04, 2 February 2020 (UTC)

Symmetry in chemistry

I've picked up some sense that others feel that symmetry is special, and that we should not break it without a strong reason. In fact there is nothing intrinsically special about symmetry.

What follows is an anonymous and erudite blogger’s account of a public lecture given on Sep 18th, 2015, in conjunction with an exhibition called Periodic Tales: Art of the Elements, which took place at the Compton Verney Art Gallery (Warwickshire, UK) from 3rd October to 11th December, 2015.

I found it to be rather extraordinary.

The public lecture was given at the Inorganic Chemistry Laboratory (Department of Chemistry) of the University of Oxford (one of the first events of the 2015 Alumni Weekend of the University).

Relevant extracts are:

"Later on, the panel of speakers took questions from the audience. The first question addressed allotropes and how they have (or not) changed our conception of the elements. This prompts us to remind Mendeleev’s philosophical stance, underpinned by the distinction between simple substances (graphite and diamond) and elements (carbon)."
"The second question touched on the equivalence symmetry-beauty, a truly broad topic. Georgiana Hedesan pointed out that symmetry was at the heart of the alchemical thought, from the figures used (squares, circles) to the fact that, supposedly, the four elements came in identical amounts. Peter Battle remarked that, although we tend to see symmetry as neat, too much symmetry might hurt (and – I add – sometimes unexpected properties of materials indeed arise from the suppression of long-range regularities and symmetry, the so-called “defects”). All of this makes me think of Italo Calvino’s reflection on literature in his essay Exactitude in his Six Memos for the Next Millennium, in which, discussing 20th-century literature, he pitted the party of the crystal against that of the flame: “Crystal and flame: two forms of perfect beauty that we cannot tear our eyes away from, two modes of growth in time, of expenditure of the matter surrounding them, two moral symbols, two absolutes, two categories for classifying facts and ideas, styles and feelings…”
"Again on the issue of symmetry and beauty, we should remember that symmetry in chemistry is a deeply mathematical concept, based as it is on group theory. Hence, we can recall what the philosopher of chemistry Joachim Schummer wrote in his sweeping paper on the aesthetics of molecules: “apart from early Pythagorean views on beauty in nature, it is difficult to find any source in the whole history of western theory of art that considers mathematical symmetry the essence of beauty. Instead, we have severe criticism of that idea as well as aesthetic theories based either on the alternative concepts of proportion and harmony or on the interplay of symmetry and asymmetry in a broad sense”[8]. Schummer’s remark is at odds with what is generally perceived as a natural relationship between between beauty and harmony of proportions, as in Palladian villas or Leonardo da Vinci’s man. As a final comment on symmetry, progress in the periodic classification of the elements took off when scientists stopped trying to force all elements in neatly symmetrical groups and periods (see for example Gmelin’s V-shaped periodic system, almost perfectly symmetrical), allowing for the existence of separate subgroups."
8. J. Schummer, Aesthetics of Chemical Products, HYLE, 2003, 73-104

I like the contrast between alchemy and chemistry, and between crystal and flame, and the parallels between the two. Sandbh (talk) 04:59, 2 February 2020 (UTC)

@Sandbh: However: symmetry works perfectly from H to Xe, considering the big picture of chemically active subshells and block homogeneity, not the dance floor of differentiating electrons, so widely squashed by the dancing shoes of excited states in chemistry. Therefore, since it works for Cs through Og as well, there is zero reason to reject it. We can talk again when period 8 happens and continuing the symmetry really breaks down. Then I advocate breaking it in the nicest possible way. Because symmetry, which is by definition a reduction of pluralities (because some bits are enough to tell you everything), must be correlated with Occam's razor. Double sharp (talk) 10:37, 2 February 2020 (UTC)

@Double sharp: Well, no, it doesn't work for Cs to Og, as I've pointed out elsewhere, given the ambiguity of La and Lu at the step 1 level. When Occam's razor was formulated, nobody knew about symmetry breaking. Here are a few quotes to this end:

From a recent New Scientist article:

"…symmetries matter, largely because we like to see them broken sometimes: the laws, particles and forces of physics all have their roots in symmetry-breaking. They create what David Gross of the Kavli Institute for Theoretical Physics at the University of California, Santa Barbara, calls the “texture of the world”. These considerations have led Florian Goertz at the Max Planck Institute for Particle and Astroparticle Physics in Heidelberg to propose the existence of a new particle that is single-handedly capable of cleaning up five of the stickiest problems in physics. “Complete symmetry is boring,” says Goertz. “If symmetry is slightly broken, interesting things can happen."

"While the laws of Nature, “are simple, symmetrical, and elegant, the real world isn’t. It’s messy and complicated…The reason is clear. We do not observe the laws of Nature: we observe their outcomes. Since these laws find their most efficient representation as mathematical equations, we might say that we see only the solutions of those equations not the equations themselves. This is the secret which reconciles the complexity observed in Nature with the advertised simplicity of her laws. Outcomes are much more complicated than laws; solutions much more subtle than equations. For, although a law of Nature might possess a certain symmetry, this does not mean that all the outcomes of the law need manifest that same symmetry.” Barrow JD 2008, New theories of everything, Oxford University Press, Oxford, pp. 136–140

"Nature always takes the path of "simplest sufficient complexity"; matter is complex only because it cannot be made any simpler and still come into existence through symmetry-breaking." from here.

Sandbh (talk) 03:47, 3 February 2020 (UTC)

Nope. La has 4f character, Lu doesn't. Symmetry works perfectly applied to chemically relevant subshells rather than chemically irrelevant differentiating electrons. You would do better to argue about group 2. We have a little problem there, and then all the symmetry breaking we want in period 8 when relativity cannot be ignored. (Though even then it breaks in the nicest possible way.) Double sharp (talk) 12:20, 3 February 2020 (UTC)

Greenwood & Earnshaw (2ED)

@Double sharp: I was reading G & E and was surprised to see them discussing the same things as me.

On group 3 they say:

  • they display the gradation in properties that might be expected for elements immediately following the strongly electropositive AEM and preceding the TM proper. (p. 946)
    • Exactly, they're intermediate and there is no big divide like a Sc-Y-La table suggests there is. Nota bene, Be and Mg are not "strongly electropositive". Double sharp (talk) 11:30, 2 February 2020 (UTC)
@Double sharp: The important part here is, "preceding the TM proper".
@Sandbh: So what else is new? The group 3 elements are physically normal transition metals. Only chemically are they not "TM proper", but since the literature widely focuses TM proper on characteristic properties like variable oxidation states, Zr, Hf, Nb, and Ta are also not "TM proper". As usual we have intermediacy and continuity. Double sharp (talk) 23:27, 2 February 2020 (UTC)
@Double sharp: I think you let things like intermediacy and continuity get in the way. Biological taxonomies would never exist but for focussing on the broad contours rather than obsessing about intermediacy and continuity and never making hard-nosed decisions about differences in classes. Classification science usually involves less than sharp boundaries. I choose to be pragmatic about these things. There’s no need to lose sleep about the hard cases. If the boundaries become less than useful then fine, change them. G & E’s “transition metal proper” works well enough. So does E and H’s. As do all the authors who effectively apologise for calling group 3 transition metals. This never happens to group 4. Well, apart from your own view. A case of Double sharp v the world. Sandbh (talk) 00:11, 3 February 2020 (UTC)
@Sandbh: I simply note that many authors who talk about excluding the Sc group unwittingly give criteria that would also exclude Zr, Hf, Nb, and Ta. And I quote their chemistry and even what those authors say about it to justify that. Apparently, you are allowed to scour the literature and criticise it when you see a Lu argument, but I'm not allowed to for this. Double sharp (talk) 07:50, 3 February 2020 (UTC)
@Double sharp: I agree. I'd attribute this to shorthand, and to the fact that transition metal status seems to be attributed to ions containing partly filled d shells. Group 3 doesn't have this, not at a level of mainstream chemistry significance. Group 4 easily does so, with Ti 3+ 2+ etc. If you feel you're not allowed to criticise this because I criticise you for doing it, you're giving away power to me. Your power is your own to use as you see fit unless you let yourself fall into auto-routine mode. Don't give away your power subconsciously. Sandbh (talk) 10:09, 3 February 2020 (UTC)
@Sandbh: Amazing, so M3+ and M2+ (M = Ti, Zr, Hf) according to you are totally relevant for finding ions with partially filled d-shells, but totally irrelevant for finding ionic character in group 4. (I argued heavy group 4 and group 5, so if you're saying "group 4 easily does so" based on one element, it contradicts you saying the covalent Be and Mg don't matter for group 2.) Sorry, but I find myself incredulous. I'm not giving away power to you subconsciously. I'm just saying what I see here: you appear to allow yourself to use properties in favour of La that you don't allow me to use in favour of Lu. Respectfully, that seems like a double standard. Double sharp (talk) 12:10, 3 February 2020 (UTC)
@Double sharp: It doesn't seem amazing to me. I'd say it was pragmatic. C & W 6ed, say, "The main transition group or d block includes the elements that have partially filled d shells only. Thus…Sc…is the lightest member…Ti-Cu all have partly filled 3d shells either in the ground state of the free atom (all expect Cu) or in one or more of their chemically important ions (all except Sc)." (p. 634) So they are saying lower oxidation states in Sc are not chemically important, compared to the situation for Ti. There's no contradiction. They allocate two paragraphs to lower oxidation states of Sc; nine pages to Ti(IV) and 2.5 pages to Ti(III). Sandbh (talk) 01:19, 4 February 2020 (UTC)
@Sandbh: That was published in 1999. Today we know so much better that divalent complexes are known for all stable lanthanides (2013 paper). All three lower oxidation states of Sc are known in organoscandium compounds: 0, +1, +2 (2004 paper). And CsScCl3 (and the analogues with Br and I, as well as Rb analogues for the halogens Cl and Br) was already known in 1980. The borderline between group 3 as not-really-transition and group 4 as "transition proper" has long since stopped working, now that +2 as an oxidation state in group 3 is pretty much as well-characterised as lower oxidation states for Zr, Nb, Hf, and Ta. Double sharp (talk) 19:40, 4 February 2020 (UTC)
  • the chemistry in the main concerns the formation of a predominately ionic +3 oxidation state. (p. 948)
    • So just like Zr, Nb, Hf, and Ta, whose chemistry mostly concerns the formation of the group oxidation state. Page 979: "by contrast most of the chemistries of niobium and tantalum are confined to the group oxidation state +5." Double sharp (talk) 11:30, 2 February 2020 (UTC)
@Double sharp: I left out 'ionic' [now inserted]. So, no, not like Zr, Nb, Hf, and Ta. Sandbh (talk) 22:07, 2 February 2020 (UTC)
@Sandbh: Which we shouldn't read too much into as a foundation for the periodic table, because Be and Mg have a +2 oxidation state that is if anything more covalent. So focusing on ionic vs. covalent, rather than being pragmatic and recognising that oxidation state and atomic size matters as Fajans did, leads to Be and Mg over Zn. And, of course, the fact that Mg is rather covalent suggests that Sc, with a higher charge and a higher electronegativity, should have significant covalence as well. The difference that the literature agrees is fundamental between main-group and transition is variable oxidation states, but then group 3 is just like Zr, Nb, Hf, and Ta. Double sharp (talk) 22:22, 2 February 2020 (UTC)
@Double sharp: Eh? One of the early—foundational—things taught in chemistry is the contrast between ionic and covalent.
Mg, as far as I know, has a mainly ionic chemistry. The EN of Mg to Ra ranges from 1.31 to 0.89. Be is more covalent than ionic, for sure. That said, 5 of the 6 AEM are ionic so yes, they are predominately ionic/strongly electropositive. Your comparison of Mg and Sc a hilarious and baseless distraction. Your fundamental difference is right, and group 4 is the first time this difference is encountered, as per Earnshaw and Harrington. Sandbh (talk) 23:34, 2 February 2020 (UTC)
@Sandbh: Nope, one of the foundational things taught is the EN difference controlling the continuum between ionic and covalent bonding. Everyone knows at that level that there is a difference between nonpolar covalent (e.g. C-C), polar covalent (e.g. C-F), weakly ionic (e.g. Be-F), and strongly ionic (e.g. Na-F). It's not a catastrophic change.
Scandium is more electronegative than magnesium and forms a higher charge, so if anything it should be more covalent. Organomagnesium compounds are mostly covalent (quite polar, sure). Earnshaw and Harrington are just wrong here, since any dismissal of lower oxidation states for group 3 runs into the same problem that Zr, Hf, Nb, and Ta are also very unhappy to be in anything other than the group oxidation state. Double sharp (talk) 23:40, 2 February 2020 (UTC)
@Double sharp: Yes, my mistake. I found this among my documents (source unknown) "The ionic radius for the +2 cation of magnesium is fairly small (0.65 Å). As a consequence the charge density (z/r) is high, which results in a high polarizing power of the Mg2+ ion. Thus, magnesium tends to form polar covalent bonds rather than ionic complexes." That does not make any difference to the predominating ionic behaviour of group 2.
Re continuums, I'm not interested. I'm only interested in predominating behaviour. That is, is the locus of the chemistry more towards the ionic end or the covalent end of the continuum? There is no issue with E & H. Their conclusion is the same as G & E and C & W and everyone else who comments about the atypical behaviour of group 3. No one says this of group 4. I don't know of any genuine lower oxidation state compounds of Sc. Wiberg says there are only a few lower valency cluster halides known in which the formal oxidation states are less than +3. They go on and say in this respect the Sc group metals are not typical transition elements because one would expect them to form the +2 oxidation state by the loss of the two s electrons.
Lower oxidation states for Zr and Hf, while rare, are well established. Sandbh (talk) 04:12, 3 February 2020 (UTC)
CsScCl3. Plus lots of recently discovered Ln(II) complexes that surely have Sc and Y analogues. Double sharp (talk) 07:50, 3 February 2020 (UTC)

On group 4 the most important oxidation state in the chemistry of these elements is +4, which they say is too high to be ionic. Lower oxidation states are rather sparsely represented for Zr and Hf. Whatever arguments may be advanced against describing to Sc, there is no doubt Ti is a “transition metal”. (p. 958)

Which is precisely why I focus on Zr and Hf. By standards of "main group vs. transition" that exclude Sc, the status of Zr/Hf and Nb/Ta suddenly becomes iffy. Indeed, +4 is too high to be ionic, but let's not read too much into it, because otherwise Th and U are incongruous: active metals with predominant states too high to be ionic. Double sharp (talk) 11:30, 2 February 2020 (UTC)
@Double sharp: Ah, well, I observe the overwhelming majority consensus in the literature per the esteemed G & E. More noise about Th and U, I see. Sandbh (talk) 23:45, 2 February 2020 (UTC)
@Sandbh: It's not noise. It's a simple refutation of the relevance of "ionic vs. covalent" distinction. Everyone knows Th and U are active metals, they just have too high oxidation states to be ionic. Everyone knows Tl is a weak post-transition metal, but it has a very low oxidation state and is ionic anyway. Fajans got it right: this is a continuous trend based on oxidation state and ionic radius. That is the broad contour, not the accident of how it ends up looking in the periods. (Of course, we have amphoteric scandium, which then suggests some covalency like for magnesium, which is less electronegative and has a lower charge. ^_^)
The overwhelming majority consensus in the literature has apparently forgotten the vast preference of Zr, Hf, Nb, and Ta for their group oxidation state. Except that I doubt it is the overwhelming majority consensus, since IUPAC has two definitions of a transition metal, and neither actually allows the exclusion of group 3. Funnily enough, my statements about the heavy group 4 and 5 elements are supported by the literature, like Greenwood and Earnshaw. Notice their conspicuous silence on Zr and Hf when defending Ti as a transition metal. As well as their frank admission that Nb and Ta have chemistries mostly confined to the +5 state. Double sharp (talk) 23:57, 2 February 2020 (UTC)
@Double sharp: Zr and Hf are irrelevant to my premise: group 3 = predominately ionic main group-like chemistry; group 4 = predominately covalent main group-like chemistry, as backed up by G & E (not that I need them, as this point is common in the literature). Not forgetting that group 4 is the first group (per Earnshaw and Harrington) in which the really characteristic transitional properties of variable oxidation state, colour and paramagnetism are encountered. Th and U are irrelevant at the broad contour level: On active metals we can say that "They are mostly strongly electropositive, with a few of the light actinides (U to Am) being only moderately electropositive." Sandbh (talk) 22:27, 2 February 2020 (UTC)
@Sandbh: Only because of oxidation state. Which is why Th and U are totally relevant for broad contours because they reveal the real broad contour trend. It's called Fajans' rules: ionic character increases as magnitude of charge goes down, as cations grow, and as anions shrink. The break is later in each period. Be2+ is very covalent, Mg2+ has tendencies, and from Ca2+ onwards we have strongly ionic cations. Al3+ is very covalent, Sc3+ has tendencies (it is an amphoteric cation), and only from Y3+ onwards is it really more ionic. With group 4 we have to wait for Rf4+ before we get something like ionicity, judging from the basicity of that cation. Lumping it together as "group 3 predominantly ionic vs. group 4 predominantly covalent" obscures what is going on. It also fails to do justice to Sc, judging by its amphoterism and diagonal relationship with Mg.
Also, in spite of Earnshaw and Harrington, these "really characteristic transitional properties" are about as weak for Zr, Hf, Nb, and Ta as they are for group 3. Double sharp (talk) 23:27, 2 February 2020 (UTC)
@Double sharp: You're clutching at straws again. The details of what is going on aren't relevant to the broad contours. That's not to say they aren't interesting. I'm not concerned with where the break is in each period, at the broad contour level. I'm only concerned with where the break is at the predominating behaviour within groups level. The weak amphoterism of Sc is irrelevant at this level. So is it's diagonal relationship with Mg. Failing to do justice to Sc. Eh? That would be why Cotton and Wilkison (6 ed) say, "Since the properties of Y are extremely similar to, and those of Sc mainly like, those of the Ln proper, and quite different from those of the regular d-block elements, we treat them also in Chapter 19 [The Group 3 elements and Ln]. I know, of course, C & W must be wrong too. Sandbh (talk) 03:16, 3 February 2020 (UTC)
See, they say Sc is only "mainly like" the Ln. And they're right, because its atomic radius makes it intermediate between the Ln and the later 3d metals starting at Ti. And guess which Ln it is most like, due to its small size? Lu. Double sharp (talk) 07:50, 3 February 2020 (UTC)

On the An they say, “it is clear an ‘actinide contraction’ exists, especially for the +3 state, which is closely similar to the ‘lanthanide contraction”. (p. 1264).

And what do they say about the Ln contraction? Exactly what I said about Y3+ through Ag3+. These contractions are only relevant when the state is always the same, which is not the case for the An. The An contraction is most relevant (1) in the +3 state, i.e. mostly for the second half of the series only and (2) for the transactinides. Looks rather like the 3d contraction with +2 a dominant state only from Mn onwards, and its effect on Ga through Kr. The moment the Ln change oxidation state, suddenly properties stop varying consistently with the contraction. I am very sure I remember them making that point in their chapter about the lanthanides. ^_^ Double sharp (talk) 11:30, 2 February 2020 (UTC)
@Double sharp: The headline is what they say about the status of the An contraction, contrary to your own view. It was relevant enough for them; it's relevant enough for me. The rest of what you said is not relevant to the broad contours.
@Sandbh: Nothing they said contradicts my view. The An contraction exists, I never disputed it. For the +3 state it is closely similar to the Ln contraction, I never disputed that either. The only trouble is that the An are mostly unhappy to be in the +3 state, and I would be astonished if they did not support that. So broad contours say that every other contraction but the Ln is weakened by lack of constant oxidation state, and therefore we should treat that as an exception. Double sharp (talk) 23:27, 2 February 2020 (UTC)
@Double sharp: Who said you disputed its existence? I didn’t. You never said the +3 state is closely similar to the Ln contraction. So why do you need to say you never disputed this? You sought to downplay the relevance of the An contraction. And yet two of our favourite authors, in a calm and non-plussed manner, refer to it and its similarity to the Ln contraction. And there you go again still seeking to downplay its significance, on the basis of irrelevant grounds. The relevance is that the Ln contraction spans the 4f row as does an analogous 5f contraction. This does not work in an Lu table. That’s all. Sandbh (talk) 00:32, 3 February 2020 (UTC)
By your standard it doesn't work for the 5f row either because Th. Oh, but excited states are apparently decisive for Th but cannot be considered for La and Ac. Never mind that chemically bound atoms are usually found in what would be excited states were they alone. You need to start actually looking at what the literature does not say as well as what it says. And maybe read it somewhat critically to note when something said does not make sense or self-contradicts. You're certainly good enough at it whenever someone in the literature supports Lu for not-so-good reasons. Except that then irrelevancies are paraded pretending to be philosophical concerns like the 234 argument, with no evidence in the literature why that is important. Sandbh vs the world, maybe? Double sharp (talk) 07:50, 3 February 2020 (UTC)

I know you like G & E. I do too. Sandbh (talk) 11:04, 2 February 2020 (UTC)

To reduction ad absurdum "predominantly ionic vs. predominantly covalent" further: let's look at anions. Is F predominantly ionic or covalent? Don't be silly, it depends on what it's bonded to. A metal fluoride is probably ionic, a nonmetal fluoride is probably covalent. A higher fluoride is probably covalent, a lower fluoride is probably ionic. (E.g. PtF6 is pretty covalent.) The same is true for the metals. Is Cs predominantly ionic or metallic? It equally well depends on what we're bonding to. Double sharp (talk) 11:58, 3 February 2020 (UTC)

@Double sharp: I've made an exception for this one since it's so close to the intermission lounge:
"Fluorine, the smallest and most electronegative of the halogens, frequently forms predominately ionic compounds." (Bailar JC 1984, Chemistry, p. 829)
Since I'm in the main arena here, I'll add, yes, this is a case of Double sharp v the World, as you put it. Sandbh (talk) 06:31, 6 February 2020 (UTC)
@Sandbh: You said that first, not me. And since the world is well aware of fluoride volatility, and how oxidation state matters here, your statement is highly dubious. Not to mention that here the important oxidation state difference is not the 3 vs. 4 you like to talk about, but 4 vs. 5. ^_^ Double sharp (talk) 12:08, 6 February 2020 (UTC)


@Double sharp:@R8R:@Droog Andrey:

At this point I'd like to spend a day in a library assimilating all that we have written, rather than continuing the thread by replying to each and every post. If I have anymore questions I'll post them. I'll post a little mini-summary of where I think things are up to, shortly. Sandbh (talk) 05:21, 3 February 2020 (UTC)


@Double sharp:@R8R:@Droog Andrey:

Here it is, my pre-library day perspective. All mistakes, omissions, and oversights are my own.

Chemical behaviour

I’m satisfied that the argument about the overall chemical behaviour of group 3 being more like that of groups 1 and 2 than 4 to 12 holds up well enough. Double sharp argues there is a continuum of chemical behaviour and that, effectively, you can’t slice the continuum non-arbitrarily. I think there is a well enough defined hiccough in the continuum between the behaviour of group 3 as predominantly main group ionic metals v the behaviour of groups 4 and 5 as predominantly covalent metals, main group for group 4; TM for group 5. That explains why, in the literature, group 3 metals are frequently referred to as atypical transition metals, whereas this is never the case for groups 4 and 5.

But the literature (or rather the small subset of it that doesn't like group 3 to be transition) has overlooked that Zr, Hf, Nb, and Ta have equally weak credentials as transition metals proper, being stuck mostly in their group oxidation state. (They have lower oxidation states, but at the same order of those for Sc: well-characterised, but uncommon.) Meanwhile, "ionic" vs "covalent" has not stopped depending on what is the other member of the bond. Is fluorine predominantly ionic or covalent? That's a silly question, because it depends on the EN difference (i.e. what it's bonded to and in what oxidation state that other element is). Sodium? Of course, NaF is strongly ionic. Uranium? Er, harder to say: UF3 is certainly ionic, judging from its high melting point, but UF4 is starting to be iffy despite also having a high melting point (it reacts with water in a way that covalent halides like BeCl2 does and MgCl2 slightly does). UF5 is polymeric, and UF6 is molecular. Carbon? Of course, polar covalent. Another fluorine atom? Of course, nonpolar covalent. Double sharp (talk) 12:03, 3 February 2020 (UTC)

@Double sharp: I think there is an order of magnitude difference between lower oxidation Sc and Ti3+. Yes, the dependence of ionic v covalent depends of the other member of the bond. That said, the literature is quite clear that group 3 is predominately ionic whereas group 4 is predominately covalent; it isn't necessary to examine every possible other member of the bond to conclude this is so. Sandbh (talk) 00:21, 5 February 2020 (UTC)

@Sandbh: Yes, it is because of that word "predominately". Just because the literature says something does not mean you can accept it without question, especially when it is something this nonsensical, just look at fluorine. Sure, there is some difference between lower oxidation states of Sc and Ti. That is just because transition properties proper flower later in each period. Sc is not so different from Zr and Hf, here. Double sharp (talk) 17:05, 5 February 2020 (UTC)

@Double sharp: Our own entry on fluorine says, "Fluorine's high electron affinity results in a preference for ionic bonding; when it forms covalent bonds, these are polar, and almost always single." Sandbh (talk) 06:21, 6 February 2020 (UTC)


Double sharp also argues that Lu is more homogenous with the individual physical and chemical properties of the rest of the d-block elements than is the case for La. Since Lu is more like an Ln than a TM and since this is also the case for La, acknowledging that Lu is more TM-like than La, I argue that periodic vertical trends, and overall group behaviour is more important than homogeneity with the individual physical and chemical properties of the rest of the d-block, given this block already spans a heterogenous range of such properties.

Overall group behaviour is more similar with Lu in group 3, and you yourself claim that periodic vertical trends between Sc-Y-La and Sc-Y-Lu are inconclusive. Double sharp (talk) 12:03, 3 February 2020 (UTC)

@Double sharp: I agree periodic vertical trends between Sc-Y-La and Sc-Y-Lu are inconclusive, as per the six graphs. That's why I say, "Since the chemical behaviour of Group 3 generally resembles that of Groups 1−2 rather than that of Groups 4−11 it follows that Group 3 is better placed next to Groups 1−2, with the result that elements of like chemistry are more closely grouped together." And the only way to properly show that is La under Y, with a split d block in the 32-column form. Sandbh (talk) 06:51, 4 February 2020 (UTC)

@Sandbh: The problem is, they don't. Zr, Nb, Hf, Ta are also basically pre-transition. And Sc should not even be fully ionic, judging by applying Fajans' rules to compare it with Mg. Double sharp (talk) 07:54, 4 February 2020 (UTC)

@Double sharp: Group 3 is mostly pre-transition ionic (per groups 1 and 2); groups 4 and 5 are mostly pre-transition covalent. Sc atoms don't care what you think they should be :) Remy (1956, p. 32): "Sc and its homologues have only a very slight tendency to form covalent compounds…". Sandbh (talk) 00:01, 5 February 2020 (UTC)

@Sandbh: Then they're wrong, just look at organoscandium compounds with that small EN difference. I see we finally agree that by the usual standards of main group vs. transition that authors who exclude group 3 as transition use, group 4 and 5 are mostly pre-transition. Now, does it make any sense to call an element mostly ionic or covalent? I've already answered why it is in fact a silly notion: just look at fluorine. Fluorides are more ionic or more covalent depending on how electropositive the counter-cation is and what oxidation state it is in. Same for the metals, just look at the counter-anion. So all this is is "look at EN and oxidation state", in which case Sc must definitely have as much covalent character as Mg. That's not me thinking, that's simply Fajans' rules: calculate charge over radius for Sc3+, it is more than for Mg2+. Double sharp (talk) 00:08, 5 February 2020 (UTC)

@Double sharp: Per G&E: "In the main, the chemistry of these elements concerns the formation of a predominantly ionic +3 oxidation state arising from the loss of all 3 valence electrons and giving a well-defined cationic aqueous chemistry. Because of this, although each member of this group is the first member of a transition series, its chemistry is largely atypical of the transition elements. The variable oxidation states and the marked ability to form coordination compounds with a wide variety of ligands are barely hinted at in this group although materials containing the metals in low oxidation states can be prepared (see p. 949) and a limited organometallic (predominantly cyclopentadienyl) chemistry has developed."

Yes, per the literature, it is useful to call an element mostly ionic or covalent; the group 1, 2 and 3 metals, for example, as well the Ln, are mostly ionic. "For chemists…the most important feature of an element is its pattern of chemical behaviour, in particular, its tendency toward covalent bond formation (or its preference for cation formation)" (Rayner-Canham Overton 2010, p. 29). Sandbh (talk) 00:43, 5 February 2020 (UTC)

@Sandbh: The literature saying this has evidently never taken a moment to think about what they are saying. Exercise: based purely on the chemistry of these four elements, decide whether their chemistries are predominantly ionic or covalent.
  1. Fluorine;
  2. Caesium;
  3. Uranium;
  4. Thallium.
  5. [optional] Carbon.
You are allowed to refer to the literature to see what the chemistry is like, but not just quote what they say about "predominantly ionic". You have to use the chemistry to decide for yourself if such a characterisation is warranted for any of these four elements. (Hint: it is not, as it varies too much on what the elements are bonded to, in what oxidation state they are in, and in what oxidation state the other element they are bonded to is in.) Double sharp (talk) 17:03, 5 February 2020 (UTC)h

@Double sharp: Here you are. I've indicated the most common oxidation state/s and whether the element in that state is predominately ionic or covalent. For added interest I added some commentary from the literature after each entry.

F (−1) ionic
"Fluorine's high electron affinity results in a preference for ionic bonding; when it forms covalent bonds, these are polar, and almost always single."
Cs (+1) ionic
"Cesium is the most electropositive and most alkaline element, and thus, more easily than all other elements, it loses its single valence electron and forms ionic bonds with nearly all the inorganic and organic anions"
U = (+6) covalent
"Because of the actinide contraction, uranium's chemistry is quite similar to that of molybdenum."
"The uranyl oxygen bonds are covalent in nature."
Tl (+1) ionic
As previously quoted
C = (±4) covalent
"In most organometallic compounds, the metal-carbon bond has predominantly covalent character…"
"Carbon chemistry is overwhelmingly covalent…"

Sandbh (talk) 06:19, 6 February 2020 (UTC)

Differentiating electrons

I’m satisfied that block membership as a global property, in which the focus is on predominant differentiating electrons, provides a purely quantitative way of distinguishing between an La table and an Lu table. Double sharp has proposed an alternative which is not fully quantitative. Double sharp also argues that anomalous differentiating electrons don’t make a difference to the actual chemistry of the elements. I’ve given examples of where they do make a difference. Double sharp argues that the anomalous presence of a p electron in Lr won’t make any difference to its chemistry. I’ve said that we can’t pick and choose as to which differentiating electrons do or do not count, since block membership is a global property based on predominant differentiating electrons. I have further noted that we don’t yet know enough about the chemistry of Lr.

Nope. Predominant differentiating electrons are predicated on ground states that the elements, in compounds, simply are not in. It is also predicated on "add a proton and an electron, that's our next element", which never happens chemically. Whereas examining chemically active subshells, like Jensen and I do, is something global that deeply controls chemistry. Double sharp (talk) 12:05, 3 February 2020 (UTC)

@Double sharp: I rest my case on the periodic table as the organising icon of chemistry, and the role of differentiating electrons (per Bohr) in determining its structure and shaping periodic trends. Sandbh (talk) 22:12, 4 February 2020 (UTC)

@Sandbh: So all we have to go on is "tradition". And sometimes you advocate looking at condensed-phase configurations for this, and sometimes gas-phase. I rest my case on the simple fact that in chemically bonded environments, atoms don't show their ground-state gas-phase configurations for the most part, and therefore considering differentiating electrons alone is folly. Whereas chemically active subshells are actually based on chemistry and not an irrelevant idealisation like isolated gas-phase atoms, which is in favour of them since the periodic table is the organising icon of chemistry. Differentiating electrons don't shape periodic trends, not in the d- and f-blocks with so many excited states so close to each other. Double sharp (talk) 00:06, 5 February 2020 (UTC)

@Double sharp: What we go on is the periodic table which represents an accumulation of scientific thought. I start with gas-phase and if need to, then consider condensed-phase configurations, and other relevant factors. I don't do this arbitrarily. Differenting electrons give order and shape to the periodic table, per the idealised n + l approximation, and Jensen's sorting hierarchy. From them we are able to discern the s, p, d, and f blocks, and the periodic trends occurring in each of them. Sandbh (talk) 01:46, 5 February 2020 (UTC)

@Sandbh: So you start with something irrelevant to most chemical environments, and then go on only if needed(!) to something that only reflects one particular chemical environment with no other elements around! The wonder is how much of the ideal n+l rule, that works perfectly for chemically active subshells considered holistically (rather than one or two special cases), still manages to survive this transformation with its huge kernel. It must surely be fundamental with how hardy it is under such trying conditions! Double sharp (talk) 17:09, 5 February 2020 (UTC)

@Double sharp: If you're confident the n + l rule works "perfectly" for "chemically active sub-shells" considered holistically you should be writing this up for a journal. Differentiating electrons are plain as day; chemically active sub-shells will be quite another thing, and convincing the chemistry community that such things are chemically significant for e.g. La while they are not for e.g. the heavy alkaline earths will be a challenge. What does "active" mean, anyway?Sandbh (talk) 05:37, 6 February 2020 (UTC)

@Sandbh: I don't need to write it up, as anything I write about this is going to be a rehash of Jensen. (Which is good, because I am not that free. ^_^) There is nothing new at all about it that requires the chemistry community to be convinced, because it already uses such considerations. From the thorium article: "A thorium atom has 90 electrons, of which four are valence electrons. Three atomic orbitals are theoretically available for the valence electrons to occupy: 5f, 6d, and 7s." (With a citation, that includes 7p as well, even!) And it's so well-known that the TM's are not dns2 but dn+2 in complexes that G & E's tables in their chapters on those groups even explicitly write out those electronic configurations (like we have at iron and silver). "Active" simply means that it may be occupied by the valence electrons in chemical environments, which includes all low-excitation-energy configurations.
I also recommend to some extent Schwarz's article, even if he's not quite right once we reach very heavy elements. He goes beyond me in stating an order of energy levels for chemically active subshells, which I don't; but, as usual, the exception is the s-block. But I think it is better not to do that. The first trouble is that you never see a non-delayed f-block collapse because by the time it starts we are already at medium-to-high Z; you see a mildly delayed one and then a significant delayed one. (Nota bene, if you insist on delaying the start of the f-block, then it extends to Lu and Lr, maybe Rf if you are consistent, where the f-electrons are totally inactive. And I can much better stomach having the pre-f character of La in the f-block than the non-f character of Lu.) And then by the time of the 5d elements it is no longer completely true that 5d lies below 6s, and it's even more wrong for the 6d elements. So I prefer to say that the n+l rule gives chemically active subshells, though their energy level order may change. And for almost all the elements it does it perfectly.
And this brings me to my last point. The perfection is marred by the Ca group, indeed. It is also marred by Cs, which is starting to show some honorary d-character too; it is also probably marred by Mc through Og with that small 7p-8s gap. But note:
  1. Even counting all of these, I have just nine anomalies. That's less than any differentiating electron scheme can come up with.
  2. Chemically active subshells even in defeat retain relevance. OK, the Ca group mars the perfection, and they actually show it in some honorary transition-metal character. That's why I say, OK, calcium and company are strictly speaking d-elements, but we call them s-elements because of greater commonalities at broad strokes level with the alkali metals. Now, does anyone seriously think that Nb d4s1 vs Ta d3s2 actually changes anything in their chemistries? In a real chemical environment both are d5−n in oxidation state n anyway.
  3. This seems to imply that Be-Mg-Zn is actually a harder case than Sc-Y-La. Which is borne out by Be and Mg siding Zn in most occasions rather than Ca, so by homogeneity of group trends it's a shoo-in. Note that the only reason why Be-Mg-Zn gives one more anomaly in differentiating electrons is because Rg has for once in group 11 produced a d9s2 configuration, so counting anomalies once again means silliness like deciding for Sc-Y-La purely based on that 7p electron in Lr (which I can fight just by quoting Jensen), since we are here apparently deciding for Be-Mg-Ca purely based on that configuration of Rg over a hundred elements away from Be(!). Whereas considering periodic trends here at least focuses on something actually relevant to the elements involved, and it goes entirely on Be-Mg-Ca so that the s-block is harmonised (as the d-block is a bit inconclusive, though physically speaking the Zn group is a much better fit). Similarly, it goes entirely on Sc-Y-Lu so that the d-block is harmonised (as the f-block is a bit inconclusive). Double sharp (talk) 12:06, 6 February 2020 (UTC)
P.S. I missed your examples where differentiating electrons supposedly made a difference. So now I have posted some refutations. ^_^ Double sharp (talk) 16:29, 3 February 2020 (UTC)
Homogeneity hypocrisy

Double sharp regards my position on the homogeneity of differentiating electrons and the chemical behaviour of groups 1-3 as being inconsistent because I have criticised him for his arguments to do with best fit homogeneity for the element under Y in terms of individual physical and chemical properties of the rest of the d-block. I regard differentiating electrons as a more fundamental property, that shapes the periodic table. I regard the overall chemical behaviour of groups as being more important than the panoply of individual chemical and physical properties of elements within a block. Of course, individual blocks show their own characteristics but group 3 does not show typical d-block behaviour. Nor does group 12, for that matter.

Overall chemical behaviour of group 3 is more homogeneous with Lu in it. Y is like a late lanthanide (which Lu is), and Sc is intermediate between late lanthanides and early 3d metals (closer to Lu than La due to atomic size). Double sharp (talk) 12:05, 3 February 2020 (UTC)

@Double sharp: Homogeneity, absent of any consideration of periodic trends, does not imply group kinship.

From our IUPAC submission:

A comparison of ionic data by Atkins et al. (2006, p. 34) concludes that Sc-Y-La is preferred over Sc-Y-Lu. Their comparison is expressed as a problem and answer, in the context that ionic radii generally increase down a group (pp. 89–90):

Problem 1.14

At various times the following two sequences have been proposed for the elements to be included in Group 3: (a) Sc, Y, La, Ac; (b) Sc, Y, Lu, Lr. Because ionic radii strongly influence the chemical properties of the metallic elements, it might be thought that ionic radii could be employed as one criterion for the periodic arrangement of the elements. Use this criterion to describe which of the sequences is preferred.

Answer The common ionic state for the group 3 elements is +3, so the electron configurations for the elements in each sequence are:

Sequence (a) Sc3+: [Ar] Y3+: [Kr] La3+: [Xe] Ac3+: [Rn]

Sequence (b) Sc3+: [Ar] Y3+: [Kr] Lu3+: [Xe]4f14 Lr3+: [Rn]5f14

The electron configurations in sequence (a) are all rare gas configurations so the ionic radii should increase slowly as the principal quantum number, n, increases. In sequence (b), Lu3+ and Lr3+ also have filled f subshells. Since f electrons shield the nuclear charge so poorly, Z* is expected to be much larger for Lu3+ and Lr3+, thereby reducing the ionic radius. Thus, sequence (a) is preferred based on ionic radii. The measured ionic radii bear this conclusion out. For six coordinate radii, the values found are 0.885 Å for Sc3+, 1.040 Å for Y3+, 1.172 Å for La3+, and 1.001 Å for Lu3+.

Sandbh (talk) 22:05, 4 February 2020 (UTC)

@Sandbh: This logic alone suggests B and Al over Sc. In sequence B3+, Al3+, Sc3+ etc., everyone has a rare gas configuration. In sequence B3+, Al3+, Ga3+ etc., we get a d-block contraction and hence a reduction of ionic radius. Therefore B-Al-Sc is preferred based on ionic radii. Ionic radii do not generally increase down a group because of contractions. Therefore, Atkins et al.'s argument is flawed because it presupposes that the group 1 and 2 s-block trend should work for everybody. Sc, Y, and Lu are d-elements, and should follow the trend set by all the other d-block groups. Which they do, and Sc-Y-La doesn't. Case closed. Double sharp (talk) 23:59, 4 February 2020 (UTC)

@Double sharp: Al is a p block element, which is the more important consideration, per the literature.

@Sandbh: And I've already conclusively demonstrated that La is an f-block element just like Ac and Th, so this argument is, indeed, needless. But I carry on because it confirms what we already knew. Double sharp (talk) 16:41, 5 February 2020 (UTC)

As per Atkins et al. (pp. 89–90): "The general trends for ionic radii are the same as for atomic radii. Thus: ionic radii increase down a group (The Ln contraction restricts the increase between the 4d- and 5- series metal ions)."

@Sandbh: That's exactly why the increase should be "restricted" here. It always happens just after a new block is inserted. Look at 1s-2s (pure) vs. 2s-3s (restricted); 2p-3p (pure) vs. 3p-4p (restricted); 3d-4d (pure) vs. 4d-5d (restricted); 4f-5f (pure) vs. 5f-6f (surely restricted by that enormous superactinide contraction that's looming). Regularity of the general trend demands double periodicity that is not fulfilled by Sc-Y-La, only by Sc-Y-Lu. Double sharp (talk) 16:41, 5 February 2020 (UTC)

From our article on ionic radius: "The ionic radius is not a fixed property of a given ion, but varies with coordination number, spin state and other parameters. Nevertheless, ionic radius values are sufficiently transferable to allow periodic trends to be recognized. As with other types of atomic radius, ionic radii increase on descending a group.

I looked at the crystalline ionic radii in our ionic radius article. The general rule works for 14 of 17 groups (not counting group 3). Sandbh (talk) 02:46, 5 February 2020 (UTC)

@Sandbh: You neglect the generality of the result of contractions, just like from 3p to 4p and from 4d to 5d. And as usual you fixate purely on reducing a number to its sign (does it increase or not), forgetting the important thing: the magnitude of the increase is, as usual, dampened significantly by a contraction. Double sharp (talk) 16:58, 5 February 2020 (UTC)


H+-Li+ He2+-Be2+
90 59
Li+-Na+ Be2+-Mg2+
26 27


B3+-Al3+ C4+-Si4+ N5+-P5+ O2−-S2− F-Cl
26.5 24 25 44 48
Al3+-Ga3+ Si4+-Ge4+ P5+-As5+ S2−-Se2− Cl-Br
8.5 13 8 14 15

d-block (just the first half, to avoid arguments about choosing oxidation states; low-spin values chosen):

Sc3+-Y3+ Ti4+-Zr4+ V5+-Nb5+ Cr6+-Mo6+ Mn7+-Tc7+
15.5 11.5 10 15 10
Y3+-Lu3+ Zr4+-Hf4+ Nb5+-Ta5+ Mo6+-W6+ Tc7+-Re7+ Ru8+-Os8+
−3.9 −1 0 1 −3 3

The same drop is encountered and the s >> p > d (> f) motif is faithfully reproduced. That is, unless you insist on La under Y and creating a one-time-only exception to the wider periodicity. It's just that for 3d-4d, the increase was already so small that for 4d-5d the increase hovers around zero. With the variation between high- and low-spin complexes for the 3d metals being over 10 pm in some cases, let's not pretend that the sign differences are some important cleft given to us by Nature herself. Double sharp (talk) 16:58, 5 February 2020 (UTC)

@Double sharp: I'm relaxed and comfortable (since we're in the intermission lounge) with my original post that mapped the broad contours of the situation. The same can be said for Atkins et al.; Jensen 1982 in his trigger paper; and Scerri and Parsons 2018. Wulfsberg (2000) has a nice Lu periodic table of crystal ionic radii for 6 CN for multiple oxidation states. Ionic radii increase going down 14 of 18 groups.
When we prepared our IUPAC submission referencing Atkins, I remember checking what they said about the general pattern of ionic radii going down groups, and I found they were right. Otherwise I would've taken it out. Same thing happened when I rechecked it this time. Sandbh (talk) 05:24, 6 February 2020 (UTC)
@Sandbh: As you can see, the differences from the 4d row to the 5d row are minuscule and hovering around zero. As I noted already, the difference in ionic radius for the same 3d metal, in the same oxidation state, between high-spin and low-spin complexes, can vary by around 10 pm. This suggests a similar sort of variation for the 4d and 5d metals in such chemical environments. Therefore I put it to you that your analysis is seriously flawed because the possible changes in ionic radius for a single element are larger than the difference between increasing and decreasing for the 4d-5d comparison! Nothing significant can be drawn there for a difference between whether the radii increase or decrease here. All we know is that the increase from 4d to 5d is mostly cancelled out, which supports the trend where after the first row, increase drops. Lu in group 3 confirms this trend, La in group 3 spits in its face. Double sharp (talk) 12:35, 6 February 2020 (UTC)
@Double sharp: I'll continue to rely on the literature, which notes the general trends involved, based on comparable data, in comparable conditions, and same oxidations state; and same CN; with no mixing of different spin environments, as did Jensen, and Chistyakov, whom Jensen relied on. C&W: "There are many important trends and correlations to be found among these results [for ionic radii]." Shriver & Atkins: "The sizes of ions, ionic radii, generally increase down a group, decrease across a period, increase with coordination number, and decrease with increasing oxidation number." etc. It's like extracting trends from data despite the 20% differences, in order to make useful generalisations (acknowledging the need for caution) rather than highlighting the 20% differences in the data and concluding that no useful or general trends can be observed. Sandbh (talk) 00:12, 7 February 2020 (UTC)
@Sandbh: How can you extract a trend from data where the range of variation exceeds the magnitude of the trend you are looking for? Anything you find will just be noise. Do you have values for the 4d and 5d metals that distinguish low-spin from high-spin? Our article ionic radius only gives such for 3d, where they differ significantly. Double sharp (talk) 00:01, 8 February 2020 (UTC)
@Double sharp: By using comparable data. I don’t know if I have that data. I hope the literature is not so uniformly “stupid” as to use uncomparable data. Sandbh (talk) 10:18, 8 February 2020 (UTC)
@Sandbh: OK, I looked it up and see why now: the 4d and 5d metals are usually low-spin in complexes, so I made the right choice above. But still: notice that the magnitude of the increase in size is not constant, but constantly wiggles a bit. If it is wiggling around zero, then I put it to you that the wiggle is not significant, and the important takeaway is that the increase has been cancelled out pretty much by a contraction. Which is what the literature says for 4d vs. 5d. By looking only at the sign you are still observing noise. As is everybody who counts TM groups showing an increase (or not), rather than getting that key takeaway.
For the series X-Y-Z I could just as well have plotted (Y-X increase) minus (Z-Y increase) to bring home the point about the increase dropping sharply after the 1st row. Then putting La in group 3 creates a big outlier. Double sharp (talk) 10:23, 8 February 2020 (UTC)
Periodic law (revisited)

I’m satisfied that the periodic law, combined with our understanding of the actual electron configuration filling sequence, provides a robust argument for La, as the first d element after Y, going under Y (whereas Lu is the third element in which a d electron appears). Double sharp is more concerned with the n + l or Madelung rule, which is only an approximation. According to the n + l rule, La should have an 4f electron (only it doesn’t).

It has a low-lying 4f state that contributes. That's good enough for chemistry, in which elements are often in what would be excited-state configurations were they alone. According to your periodic law Th is the second 6d element because Th has the advantage of incumbency over Rf when it comes to the 6d2 configuration. I know what will follow: a double standard in which Th as an f-block element is defended on the grounds of its ions and excited-state configurations that contribute in the metal, but La and Ac with similar excited-state credentials are blocked. While meanwhile Lu and Lr are let in with absolutely core-like, inactive f-orbitals in an unprecedented step. Double sharp (talk) 12:02, 3 February 2020 (UTC)

@Double sharp: My understanding of the situation follows.

La is the first element with a 4d electron so it goes under Y according to the periodic law, and the aufbau principle as manifested in real life rather than the idealised n + l rule. I’d regard the presence of a low-lying 4f state as a tipping point argument, in the absence of more fundamental arguments.

La does not form a cation having an 4f electron; OTOH the Lu3+ cation has the configuration [Xe]4f14. The poor shielding of the 14 f electrons results in a large contraction of the size of the Lu cation, and this impacts it chemistry, making it the least basic of the lanthanides.

Ce is the first 4f element so it starts the f-block. The other block starting elements are H, B, and Sc.

Th as the second 6d element would normally go under Hf, as occurred historically.

Subsequently (as proposed by Seaborg) it was realised that Th 6d2, Pa 5f26d1, and U 5f36d1 in fact represented the start of new series analogous to the Ln. Thorium thereby came to be relocated under Ce.

It was then determined that the presence of f character in Th influenced its crystalline structure and that Th could form a +3 cation having an [Rn]5f1 configuration. The f-electron count for condensed thorium is thought to be up to 0.5 due to a 5f–6d overlap.

The presence of any f-character in La is not considered to be of comparable significance: "...its 4f character, if there is one, is in any case very small (B. Coqblin 1977, The electronic structure of rare-earth metals and alloys, Academic Press, p. v).

A few authors have referred to some properties of Lu being influenced by the presence of its filled 4f shell: Langley 1981; Tibbetts and Harmon 1982; Clavaguéra, Dognon and Pyykkö 2006; Furet et al. 2008; Xu et al. 2013; Ji et al. 2015. The most surprising of these is likely to have been Clavaguéra and colleagues, who reported a pronounced 4f hybridisation in LuF3 on the basis of three different relativistic calculations. Their findings were questioned by Roos et al. (2008) and Ramakrishnan, Matveev and Rösch (2009).

Citations for all but Furet et al. can be found in our IUPAC submission. Furet E, Costuas, K, Rabiller, P & Maury O 2008, "On the sensitivity of f electrons to their chemical environment, Journal of the American Chemical Society, vol. 130, no. 7, 2180–2183

Sandbh (talk) 21:46, 4 February 2020 (UTC)

@Sandbh: My understanding follows instead:
Lutetium utterly lacks 4f character. All but one study conclude that those orbitals are core-like, and that one is questioned. OTOH, lanthanum certainly has 4f character, since 4f is a low-lying configuration well within the range that chemical bonding can excite an atom to. (It is actually lower in energy than d9s2 for silver.) Otherwise cubic La complexes would be terribly difficult to explain on symmetry grounds. This is utterly decisive: lanthanum is the beginning of the f-block. This is exactly like all the heavy element delayed collapses: 5f doesn't start in the gas phase until Pa, 6d until Rf, 5g until E125. Not the slightest problem, they are all the same. Do you not realise how much of a double standard it is to keep insisting on differentiating electrons, but backpedaling and saying "whoops, actually this 5f excited state is important for thorium only"? When is the differentiating electron just an inconvenience like for thorium and lawrencium, and when is it prohibitive like you seem to want it to be for lanthanum and actinium? Where is the consistency? Double sharp (talk) 00:06, 5 February 2020 (UTC)

@Double sharp: I agree the 4f electrons in Lu are core like. That does not mean the presence of those fourteen f electrons in Lu3+ has zero impact on the chemistry of Lu. Quite the contrary. Lanthanum, OTOH, has zero 4f electrons to start with. There is no back pedalling in the case of Th, as I will explain shortly. Differentiating electrons are the first order sorting mechanism, per Jensen. They do not resolve La and Lu. For La the presence of any f-character is minimal. For Lu, its fourteen f electrons have a sizeable impact on its chemistry. I raised other considerations in my first response. Thorium causes some problems which, on closer examination and consideration of other factors, go away. Sandbh (talk) 01:34, 5 February 2020 (UTC)

@Sandbh: The 14 f electrons in Lu3+ have as much impact on Lu chemistry as the 14 f electrons in Hf4+ have on Hf chemistry. Or the 14 f electrons in Ta5+ on Ta chemistry. As do the 10 d electrons in Ga3+ chemistry. Every single one of them are the same thing: the effect of a filled core-like subshell in providing incomplete screening. That happens once a block is finished, which is exactly why Lu needs to go into the d-block. And it is totally different from that of the 14 f electrons in Yb2+ because then the 4f subshell is still available for chemical reactions, proving that at Yb we are in the f-block and at Lu we have left it.
Jensen did not say that differentiating electrons were the first-order sorting mechanism. He wrote of the "kinds of available valence electrons". And given the way d- and f-block elements move freely between configurations, for La through Yb that includes 4f, 5d, 6s, and 6p due to having so many configurations so close to each other: chemical bond energy is sufficient for such excitations. The amount of energy required for La to display 4f1 is well within that of chemical bond range. Now, does anyone actually have a figure for how much energy Lu requires to breach the 4f shell? I certainly cannot find that figure listed in the NIST tables. (Not saying it can't happen, since the table is not so complete: here we are told that "New and more complete observations of Lu I and Lu II are needed". But I bet it is far more than the energy required for La to put something in its 4f shell, since now we are breaking open a full subshell.) I put it to you that these two considerations are decisive: La has direct 4f involvement as a valence subshell, just like Th for 5f, that just happens to not be occupied in whatever configuration happened to be the ground state. (Same for 5s in Pd.) 4f in Lu acts like a core subshell, just like for Hf, Ta, W, and so on. There is no difference between my approaches for La and Th.
The one time I have to appeal to other factors is for a very different situation altogether: group 2 vs 12, where there is a genuine ambiguity. And I try not to sweep it under the rug as best as I can: I am happy to call Ca through Ra strictly also d-elements, just putting them in the s-block since Zn, Cd, and Hg are more d-element-like and the d-block can't hold eleven columns. When those higher oxidation states of Cs are discovered I am willing to call Cs strictly also a p-element as well but still display it in the s-block as eka-Rb. The s-block is weird anyway, so it can stomach these odd cases better than any other one. Double sharp (talk) 16:33, 5 February 2020 (UTC)

@Double sharp: Bear in mind my focus here is only on the composition of group 3. In the case of the f-block, its start is delayed until Ce. It therefore finishes at Lu. This can be seen in the configurations of the applicable trivalent cations, from Ce f1 to Yb f13, and Lu f14.

Yes, my mistake. Jensen said, "Assignment to a major block based on the kinds of available valence electrons (i.e., s, p, d, f, etc.)." In La this is s and d. In Lu this is s and d. Thus, for example: "…lanthanum and lutetium, both of which have just three (sd) valence electrons." (Freeman et al. 1984, Handbook on the lanthanides and actinides, p. 166). So Jensen does not work here. Everything else I wrote still stands. I think you need to be a bit careful saying the s-block is weird, since it is supposed to be the epitome of regularity and periodicity. Sandbh (talk) 04:53, 6 February 2020 (UTC)

@Sandbh: False premise. La has f-involvement just like Th does: it's only that you implicitly admit chemically active unoccupied subshells for Th and Lr by looking at the condensed phase, and refuse to consider them for La and Ac, creating an effective double standard. Jensen noted much the same thing while criticising Lavelle: you seem to be making Lavelle's mistake, characterised by Jensen as "he ignores the evidence for irregular configurations and the loose correlation between these configurations and chemical behavior and instead relies solely on their ground-state valence configurations coupled with apparently arbitrary criteria for when they are or are not of significance when it comes to assigning the elements in question to either the d- or the f-blocks." So in La the valence orbital types include f, but in Lu they do not.
The s-block is weird precisely because it looks like the epitome of regularity and periodicity from the schoolkids' perspective. This is actually very weird because it means that:
  1. Incomplete screening effects are minimised because of the hard noble gas core;
  2. Secondary periodicity does not arise for the same reason;
This is weird because every other element has this. You like to refer to predominance, so consider this: in school we like to look at elements like group 1 and 2, maybe the first row, and we see that our usual tricks like simple cations and stable octets work perfectly there. And then we go elsewhere and it all stops working. Since most elements have this stop working, perhaps we might consider that the elements we know and love are, in fact, the weird ones? And primogenic repulsion supports that for period 2, while absence of screening effects and secondary periodicity supports that for the s-block. Double sharp (talk) 12:32, 6 February 2020 (UTC)

@Double sharp: I observe Occam's razor and only consider secondary criteria when primary criteria are inconclusive. La is a d-element, as is Th and Lu. La, Th and Lu fit well where they are in a convention table for all of the other secondary reasons I've previously listed. Sandbh (talk) 23:28, 6 February 2020 (UTC)

@Sandbh: Precisely. And not only do I do the same thing, I also make sure my primary criterion (chemically active subshells) is actually relevant to the elements' chemistry, which yours (differentiating electrons) isn't. Jensen and Schwarz have already debunked the latter. Double sharp (talk) 00:10, 7 February 2020 (UTC)
Debunking differentiating electrons

@Double sharp: Where did J & S debunk differentiating electrons? Sandbh (talk) 07:38, 7 February 2020 (UTC)

Jensen, quoting Jørgensen: 'There is not the slightest doubt that no simple relation exists between the electron configuration of the ground state of the neutral atom and the chemistry of the element under consideration. Thus iron and ruthenium differ much more from each other chemically than do nickel, palladium, and platinum, though the configurations are analogous in the former case but differ in the latter. The most spectacular discrepancy between the spectroscopic and chemical versions of the periodic table is that helium is an alkaline earth element from the standpoint of spectroscopy, since its configuration does not terminate with with np6 like the other noble gases. Hence, it is not too surprising that the almost invariant trivalency of the lanthanum series has little to do with the ground states of the neutral atoms.'
Schwarz: 'The second reason for differences between chemically bound transition-metal atoms and free atoms in vacuum is that the electronic motions in free atoms are not disturbed by adjacent atoms. Most free atoms have open valence shells, where the electrons can arrange differently. The orbit−orbit and spin−orbit angular-momenta couplings result in a large number of different electronic states with different energies. For instance, the 3d54s1 configuration of a free Cr0 or Mo0 atom comprises 504 different states with 74 different degenerate energy levels, scattered over several hundred kJ/mol. ...
'The qualitative behavior of chemical elements can be rationalized with the help of the dominant electronic valence configurations of the atoms embedded in a molecular or crystal environment. These may be correctly called the “electronic configurations of the chemical elements”. However, what is listed in respective tables of chemical textbooks under this headline is something else, namely, what physicists call “the configurations from which the J-level ground states of free unbound atoms in vacuum derive”. ...
'The third exception concerns the free neutral transition-metal atoms in vacuum, including the f block. Their ground-state configurations depend in an involved manner on the often-discussed averaged d−d and d−s Coulomb-repulsion energies and also on the individual orbit−orbit (term) and spin−orbit splittings, even if the latter are small. The correct quantitative explanation is vital for the interpretation of atomic vacuum spectra, but exceeds the scope of general chemical education. There are only a few special topics in chemistry that require the correct understanding of free atoms in vacuum (e.g., atom-molecular gas-phase reactions) or of orbit−orbit and spin−orbit couplings of bonded open-shell atoms (e.g., the chemistry of the transition, lanthanoid, and actinoid metals; spin-flip enhanced reaction mechanisms; so-called spin-forbidden processes). [Nota bene, bonded open-shell TM atoms show different configurations from ground-state free ones.]
'Finally, it is misleading to present free atoms as prototypes for the microscopic description of chemical elements in compounds. The common qualitative textbook explanations of the atomic ground states (correctly: J levels) are incorrect. Therefore, we plead for teaching the correct atomic-orbital order (sequence 6) together with the regular exception, sequence 8, for the s block. One need no longer apologize for irregularities.' Double sharp (talk) 00:07, 8 February 2020 (UTC)
I have added some bolding. Double sharp (talk) 19:48, 8 February 2020 (UTC)

@Double sharp: Jensen undermines his own quote when his #1 criteria for block assignment is, “Assignment of the element to a major block based on the kinds of available valence electrons and/or valence vacancies (i.e., s, p, d, f, etc.).” Jorgensen did not imply there was no relationship between electron configuration and the chemistry of an element. Instead, he said there was no simple relationship. Nor did he say anything about differentiating electrons, and their relationship to blocks. His point was that you could not necessarily tell all of the oxidation states that any particular element could manifest. That is all.

Schwarz similarly did not have anything to say about differentiating electrons, nor their connection to the aufbau principle and the n+l rule. Sandbh (talk) 10:57, 8 February 2020 (UTC)

  1. Jørgensen is obviously not only talking about common oxidation states. These differ in group 10 just as they do in group 8, taking his example. What he is talking about is the totality of chemistry, which is something you have so far in these arguments been totally unable to grapple with, respectfully. It is a simple fact that group 10 is more homogeneous in chemistry than group 8, but ground-state electron configurations match better in group 8 than 10!
  2. Jensen has mentioned "valence vacancies", not just electrons. Surely that includes 4f for La. In fact, considering valence electrons and vacancies basically means considering chemically active subshells, as Droog Andrey and I advocate.
  3. What on earth are you talking about wrt Schwarz? He already explained very well why ground-state gas-phase configurations are irrelevant chemically. The fact that he doesn't use the words "differentiating electrons" is irrelevant since those depend on ground-state gas-phase configurations. And if you actually read the article (I gave a link) you would see him discussing the energy level order of subshells as well. Double sharp (talk) 12:17, 8 February 2020 (UTC)

@Sandbh: And here is Glenn T. Seaborg himself on my side that the Ln contraction is exceptional and that we should consider chemically active subshells (i.e. those that are involved for ions and compounds) instead of just ground-state gas-phase electron configurations. I have added some bolding:

'It is important to realize that the electronic structures listed in Table 6 are those of the neutral (unionized) gaseous atoms, whereas it is the electronic structure of the ions and compounds that we are chiefly concerned with in chemistry. The relationship of the electronic structure of the gaseous atom of an element to that of its compounds can be rather complicated. For example, in the case of the actinide and lanthanide elements, one would not necessarily predict the predominance of the III oxidation state from the electronic structures of the gaseous atoms; there are usually only two so-called "valence electrons," the 7s or 6s electrons, which might indicate a preference for the II oxidation state.

Apparently, specific factors in the crystal structure of, and the aquation (hydration) energies of, the compounds and ions are important in determining the stability of the III oxidation state. Thus, the characteristic tripositive oxidation state of the lanthanide elements is not related directly to the number of "valence electrons" outside the 4f subshell, but is the somewhat accidental result of a nearly constant small difference between large energy terms (ionization potentials on the one hand, and hydration and crystal energies on the other) which persists over an interval of fourteen atomic numbers. Therefore, if we could somehow have a very extended Periodic Table of Elements containing numerous "f" transition series, we might expect that the 5f, rather than the 4f, elements would be regarded as more nearly representative of such f series.'

Respectfully, your stance is Sandbh vs the world. Double sharp (talk) 19:47, 8 February 2020 (UTC)

@Double sharp: I could not find anything in Seaborg saying the Ln contraction is exceptional. I liked where he said, "The relationship of the electronic structure of the gaseous atom of an element to that of its compounds can be rather complicated." This is much more useful than Jensen, quoting Jørgensen: 'There is not the slightest doubt that no simple relation exists between the electron configuration of the ground state of the neutral atom and the chemistry of the element under consideration." I liked Seaborg's two tables showing the Ln as 58 to 71, and the other two showing the f-block as Ce to Lu, etc. How did you find these tables? Sandbh (talk) 06:12, 10 February 2020 (UTC)
@Sandbh: Regarding the exceptional nature of the 4f contraction, see the last bit that I already quoted: "Thus, the characteristic tripositive oxidation state of the lanthanide elements is not related directly to the number of "valence electrons" outside the 4f subshell, but is the somewhat accidental result of a nearly constant small difference between large energy terms (ionization potentials on the one hand, and hydration and crystal energies on the other) which persists over an interval of fourteen atomic numbers. Therefore, if we could somehow have a very extended Periodic Table of Elements containing numerous "f" transition series, we might expect that the 5f, rather than the 4f, elements would be regarded as more nearly representative of such f series." In other words, the main reason why we care so much more about the Ln contraction than any other one (constant +3 oxidation state) is an exception. We all know that the s, p, and d series do not show this. And as Seaborg notes, the 5f series doesn't show this, and higher f-series that we might expect for the extended periodic table don't show it either.
The whole point is that there is no simple relationship between ground-state electron configurations and chemistry. DE's are based only on ground-state electron configurations, and therefore the relationship between them and chemistry is foggy. As Seaborg wrote, 'it is the electronic structure of the ions and compounds that we are chiefly concerned with in chemistry'. As Schwarz wrote, 'The qualitative behavior of chemical elements can be rationalized with the help of the dominant electronic valence configurations of the atoms embedded in a molecular or crystal environment. These may be correctly called the “electronic configurations of the chemical elements”. However, what is listed in respective tables of chemical textbooks under this headline is something else, namely, what physicists call “the configurations from which the J-level ground states of free unbound atoms in vacuum derive”'. In other words, look at electron configurations over all chemically relevant environments, which do not include ground-state gas-phase ones and your favourite DE's that are based on them.
I found the La tables there understandably mistaken, given the era. Same as all the Zn tables from the immediately previous era. As we can see from figures 2 and 4, the old mistaken idea of the f-block as degenerate members of the d-block had not died yet, and so what came below Y was instead La and all the succeeding elements, not only La, and this placement no doubt influenced the mistaken idea that La should go up there as the placeholder for all of them. We are past this, I hope. (Oh, and he duplicates Al in group 3 and group 13 in figure 2. ^_^) Double sharp (talk) 11:34, 10 February 2020 (UTC)
PS: Another interesting passage in Seaborg was, "Some spatial classifications of the elements appeared in which the heaviest elements, starting with thorium as the homolog of cerium, are listed as the chemical homologs of the rare-earth elements, but the reason in these cases appears to be mainly connected with the symmetry of and the ease of making such an arrangement (Cjounkovsky and Kavos (1944): Talpain (1945)."
@Double sharp: Well, consider this passage: "The excitation energies of the free atoms and ions correlate with many properties of their compounds: redox potentials, energy of cohesion of the metals, and thermodynamic stability…" from here. Surely it is reasonable to observe that there are good correlations to be drawn between the electron configurations of the elements in their gas phase, and their properties and those of their compounds on the ground, noting that such correlations can sometimes be complex, obscure, tricky or buffeted by irregularities? Sandbh (talk) 00:52, 11 February 2020 (UTC)
@Sandbh: But the author doesn't refer to the ground-state electron configurations of the gas-phase atoms that you restrict your consideration to. She refers to the excitation energies, which implies that excited states are being considered and an interplay of configurations. As indeed she does on p. 41. Double sharp (talk) 23:11, 11 February 2020 (UTC)
@Double sharp: The abstract sums it up well enough for me:
"A number of properties of d-elements, lanthanides, and actinides depend on the initial dq-, fq- and final dq–1-, fq–1-electron configurations. These properties include the ionisation potentials of the free atoms and ions, electron affinity, redox potentials, excitation energies, and enthalpies of the decomposition and disproportionation of the halides, oxides, chalcogenides, and pnictides of the lanthanides and actinides, the oxidation state in compounds with variable oxidation states, etc."
Sandbh (talk) 23:26, 11 February 2020 (UTC)
@Sandbh: It sums it up in my favour. She's already mentioned two electron-configurations in it: fn and fn-1. They can't both be the ground state, so it can't be supporting your stand. This is clearly pretty much what I mentioned about the interplay between fn and fn+1 configurations in the Ln and An. And guess what, La through Yb show it, Lu mathematically cannot. Double sharp (talk) 23:31, 11 February 2020 (UTC)
@Sandbh: You're guilty of that yourself. You exclude La as an f-element because it lacks a 4f electron, and demand that the f-block should start at cerium. But then you notice that thorium also lacks an f-electron. Since your own logic for La would demand that thorium also be excluded as an f-element, you quickly make special exemptions for it on the grounds of chemically more relevant configurations where 5f appears in it. (Meanwhile, those configurations for lanthanum and actinium are quietly ignored.) If you were being consistent, rather than being concerned with the symmetry that demands that each block begins as a complete vertical column, you would be forced to begin the 5f block at protactinium by your logic. And this is what Seaborg's forerunners did not rule out, suggesting elements all the way up to element 99 as possible starting points for the 5f series (yes, a protactinide series was considered possible among other things), to be confirmed or disproved by experimental investigation of their chemistries! So you are the one demanding symmetry that each block starts in a vertical column, that those pioneers did not demand, but simply treated as a hypothesis to stand or fall on later discoveries! Which is why your reaction to La and Ac is "yes, that is prohibitive", whereas your reaction to Th, Lr, and presumably E121 when we discover it is to scramble to find other reasons to avoid exposing the inconsistent use of DE's. Just like Lavelle's argument that Jensen criticise, respectfully. If you were really consistent about it, you would be noting that your logic demands that Th is not an f-block element (and hence a protactinide series), and the Lr is not a d-block element. But you won't do that, because "there is no other place [Lr] can practically go" as you said. Well, so what happened to drawing Nature as She was rather than how we would like Her to be?
When Seaborg formulated his actinide concept, not only was the electron configuration of the Ln and known An not really known for sure (Ce was thought to be 4f26s2), but he also explicitly noted that the important thing was chemistry and that it should trump the exact ground state configurations. To quote him: "It may be, of course, that there are no 5f electrons in thorium and protactinium and that the entry into a rare-earth like series begins at uranium, with three electrons in the 5f shell. It would still seem logical to refer to this as an actinide series." And why? Because of the predictions for Am and Cm that don't make sense without an actinide concept, with Cm refusing mostly to go past the IV state and Am having difficulty as well. The chemistry rules, not the little waves here and there of which among the dozens of close configurations happens to luck out and barely become the lowest. That's exactly why a Lu table is so much superior to a La one holistically. Double sharp (talk) 11:42, 10 February 2020 (UTC)
@Double sharp: I have nothing to feel guilty about! :) As we wrote in out IUPAC submission, a block starts upon the appearance of the first applicable electron. Everything else falls into place after that, guided by the periodic law and the n+l approximation. I don't have to justify Th's place under Ce, although it is interesting to comment upon the marked 4f nature of Th. Sandbh (talk) 00:58, 11 February 2020 (UTC)
@Sandbh: As I wrote above: you're still assuming symmetry to force a block to begin in a vertical column, even if your first criterion of DE's says it doesn't. So why is this symmetry inviolable when you seem to consider it fine to violate the symmetry of rectangular blocks? Double sharp (talk) 23:11, 11 February 2020 (UTC)
I'm grooving with the Seaborg vibe :) Sandbh (talk) 06:17, 10 February 2020 (UTC)
@Sandbh: It appears that the both of us can read the same article and both come away with completely different impressions of whose arguments they are supporting. Which maybe supports R8R's contention on your talk page: "I also fear, and I hope to be wrong at that, that the discussion with Double sharp is not helping you much---not because DS is not giving you reasonable arguments but rather because, as it appears to me from the sidelines of your discussion, you two cannot truly hear each other. I would be inclined to think that a complete consideration would include his arguments too, and then try to compare each one hand-by-hand by the same metrics. If I were to summarize in one sentence how your paper could be improved, it's that---it could use a comparison of both sides of each argument by the same metrics. I don't know if that would yield the same result or whether it would yield a definitive result at all, but if a result was to be gained this way, it would stand a much greater chance to persuade me, even if into the option I don't usually fancy. The same, I believe, is also to be said of your intended readers." Double sharp (talk) 11:42, 10 February 2020 (UTC)

I’m satisfied with the rare earth argument and the regularity of its horizontal and vertical trends, consistent with the other 18 groups and the actinides, but I’m not satisfied with how I’ve explained its relevance.

Because it doesn't have any. It is predicated on a category we made up that is not wholly based on chemistry (partly natural occurrence) and blurry in its boundaries. The logic is moreover not applicable to any other category we made up other than the group-based ones. Double sharp (talk) 12:02, 3 February 2020 (UTC)

@Double sharp: We have an article on the rare earth metals. Books have been written on the chemistry of the rare earths e.g. Topp NE 1965, The chemistry of the rare-earth elements, Elsevier, Amsterdam. As with classification schemes generally, there is some variation and overlapping of properties within and across each category. That is to be expected. Sandbh (talk) 07:01, 4 February 2020 (UTC)

We also have articles on transition metals, platinum group metals, refractory metals, noble metals, etc. Every one has wobbly boundaries (noting what you said about Au as an honorary PGM), and not a single one of them can be stretched out into a straight line. At the very least that questions the relevance of your own argument about the stretching by your own standards. Double sharp (talk) 19:32, 4 February 2020 (UTC)

@Double sharp: All those categories can be stretched out into lines, of increasing Z, consistent with their appearance in Z order in the periodic table. PGM = 44 45 46 77 78 79; etc. In an Lu table, the REM appear as 21 39 71; 57 to 70. In an La table they appear as 21 39 57 to 71. Sandbh (talk) 01:16, 5 February 2020 (UTC)

@Sandbh: Ah, so now we are allowed to connect Pd to Os even if they're not next to each other. That's already backpedalling from your statement in #Rare earth series: 'The first option can be "bent", in ascending numerical order, to read Sc 21 to Lu 71.' Now, since you have returned to reasonably allowing people to read the periodic table in the order they read English(!), note that in a Lu table, the REM appear as 21, 39, (*) 71 with an asterisk leading to 57 through 70. So, that is just 21, 39, 57-71 by the simple rule of reading footnotes. There's always an asterisk in an 18-column form (as long as it actually takes a stand on group 3) that lets you reconstruct a 32-column form from it, and people know how to read asterisks and find footnotes and do the gluing. And at this rate every category can be stretched out into lines of increasing Z just because the periodic table goes in increasing Z, even a stupid one like Be-O-P-S-V-Mo (atomic numbers from the Lost numbers sequence), so this still ends up saying nothing. Double sharp (talk) 16:38, 5 February 2020 (UTC)

@Double sharp: I had to go back to my article to see what I'd written [lines and arrows added]:

"The rare earth series (Sc, Y and the lanthanides La–Lu) appear listed in order of their atomic numbers in a 32-column periodic table with Group 3 as Sc-Y-La-Ac. If Group 3 is shown as Sc-Y-Lu-Lr, the minority of the rare earths appear in order of their atomic number whereas the majority appear in a backwards order. Thus, in the first instance they appear as…

| Sc21
| Y39
| La57 Ce58 Pr59 Nd60 Pm61 Sm62 Eu63 Gd64 Tb65 Dy66 Ho67 Er68 Tm69 Yb70 Lu71

…whereas in the second instance, as follows:

                                                                      Sc21 |
                                                                       Y39 |
La57 Ce58 Pr59 Nd60 Pm61 Sm62 Eu63 Gd64 Tb65 Dy66 Ho67 Er68 Tm69 Yb70 Lu71 | 

The second option is awkward, or highly anomalous at best, since the horizontal, vertical, and diagonal trends that characterise the periodic table are based on an increasing sequence of atomic numbers."

The PGM look like:

Ru44 Rh45 Pd46
Os76 Ir77 Pt78

Here, the majority do not appear in backwards order. In fact there is no backwards order really, since the two horizontal triads are aligned.

Of course, I can word the last line of my extract better. Sandbh (talk) 04:20, 6 February 2020 (UTC)

@Sandbh: There's nothing backwards about the REM even in a Lu table, even if I still think the argument is silly. It goes like this:
                                                                      Sc21 |
                                                                       Y39 |
La57 Ce58 Pr59 Nd60 Pm61 Sm62 Eu63 Gd64 Tb65 Dy66 Ho67 Er68 Tm69 Yb70 Lu71 |
Why do you allow yourself to do that for the PGM, but not for the REM? Just take each line and read it, left to right. If not, any category which starts earlier in each period is suspect. Good riddance to the post-transition metals, then. Double sharp (talk) 12:27, 6 February 2020 (UTC)

@Double sharp: In the PTM the majority do not appear in "backwards" order, or uniquely –x spaces behind the first member of the series, if you will.

@Sandbh:. Yes they do. The first member is Al in group 13, in all future periods they start in group 12. Double sharp (talk) 00:36, 8 February 2020 (UTC)
@Double sharp: There are 12 PTM. How many are in negative spaces behind Al? Sandbh (talk) 10:04, 8 February 2020 (UTC)
@Sandbh: Ah, so that's what you mean: just the first element in each row, not the number of early-starting rows. Fine, not in this case, then. But I can look at the definition that starts in group 11, and suddenly it is inconclusive. At the extreme and only considering true metals (i.e. not Bi) we may end up with {Cu, Ag, Au, Zn, Cd, Hg} on the left, and {Al, Ga, In, Tl, Sn, Pb} on the right. Now where are we? This choice is just as arbitrary as your choice between REM definitions. (You do know that Wulfsberg treats Bi, Po, At and sometimes even Sn as nonmetals in his tables, right?)
Anyway: you are shifting the goalposts. First it is about stretching in order. Once we point out that it doesn't work for the PGM, you back up and say it's actually about vertical placements. When are we going to allow readers of the periodic table to use their common sense that they exercise when reading English? OK, the REM start much earlier in period 6. Big deal, it's just like reading indented paragraphs. This is a pure graphic-design argument that has nothing to do with chemistry. Double sharp (talk) 10:14, 8 February 2020 (UTC)

Tempted to add: try "spectroscopically s2 elements". Since spectroscopic stuff is one of the only times your favourite DE's become important. Well, in your table, they start with He, and everyone else is suddenly 30 columns to the left. ^_-☆ Double sharp (talk) 10:17, 8 February 2020 (UTC)

Here's another angle for your consideration:


The rare earth metals, in a conventional periodic table, effectively appear as follows:

Y 39
La57 Ce58 Pr59 Nd60 Pm61 Sm62 Eu63 Gd64 Tb65 Dy66 Ho67 Er68 Tm69 Yb70 Lu71

Whereas In a table with Lu in group 3 they appear as follows:

                                                                      Y 39
La57 Ce58 Pr59 Nd60 Pm61 Sm62 Eu63 Gd64 Tb65 Dy66 Ho67 Er68 Tm69 Yb70 Lu71


Option A
The REM appear in order of Z i.e. as Group 3 (21-39-57) and 58-71 (i.e. the old school definition of lanthanides). Chemically, in terms of their trivalent cations, this can be explained as group 3 having NG cores, and Ce to Lu having [Xe]f1 to f14 cores.

Alternatively, you can treat them as just 21-71, and leave the group 3 concept out. Chemically, in terms of the trivalent cations, this can be explained as Sc, Y and La having NG cores, and Ce to Lu having [Xe]f1 to f14 cores.

Option B
The REM appear either as:

  • Group 3 (21-39-71) and 57-70; or
  • 57-70 and group 3 (21-39-71).

Chemically, in trivalent cation terms, this can be either explained as:

  • Sc and Y having [NG] cores; Lu having an [NG]f14 core; La having an [NG] core and Ce to Yb having [Xe]f1 to f13 cores; or
  • La having an [NG] core, Ce to Yb having [Xe]f1 to f13 cores; Sc and Y having [NG] cores; and Lu having an [NG]f14 core.

Alternatively, you can treat them as 21-39, and 57 to 71, and leave the group 3 concept out. Chemically, in terms of the trivalent cations, this can also be explained as Sc, Y and (reaching back 14 places) La having NG cores, and Ce to Lu having [Xe]f1 to f14 cores.

I think I’ve answered my own question, after my head stopped spinning, but which of these two options is easier to present? Sandbh (talk) 23:20, 6 February 2020 (UTC)

@Sandbh: You're making it too hard. In a Lu table, the REM are just the 4f metals plus group 3. (Yes, La is a 4f metal, as I have demonstrated far too many times here already.) Double sharp (talk) 00:09, 7 February 2020 (UTC)

@Double sharp: In terms of differentiating electrons it’s not. Differentiating electrons are mainstream; everything else is noise by comparison, IMO of course.

In archive 40, you wrote, “I think saying that a block starts when its characteristic electron appears in the ground state only is too coarse."

I responded, “It’s consistent with all my responses above. It's the simplest rubric I can think of that produces consistent results viz. "The best education is found in gaining the utmost information from the simplest apparatus". (Whitehead AN 1929, The aims of education and other essays, The Free Press, New York, p. 37).

You replied, “But if the cost is that a block then has to extend to the point where its characteristic electrons are core electrons, then I'd rather have weakly involved excited states at the start instead.”

As I explained earlier, while Lu was originally thought to have an f differentiating electron, the fact that it didn’t, and that the f block then extended to the point where it’s characteristic electron become a core electron, made no difference to the chemistry of interest.

If you want to prioritise block details in the terms you’ve described, when it makes no difference to the chemistry of interest, that’s fine. I’ll stick to the chemistry. We can agree to disagree. Sandbh (talk) 09:16, 7 February 2020 (UTC)

@Sandbh: I trust the quotes I have given from Jørgensen and Schwarz have explained why it is that differentiating electrons have very little to do with actual chemistry. The fact that La and Lu both lack a 4f differentiating electron is completely inconsequential. We have to look at chemical activity of subshells, because that is at least something actually relevant to chemistry. And when we do that, we find 4f character for La and not for Lu. Double sharp (talk) 00:09, 8 February 2020 (UTC)

@Double sharp: You’ll see from my response that J & S were of no value. The universal impact of the 14 f electrons in the Lu3+ cation on the chemistry of Lu dwarfs any marginal f character in La. Sandbh (talk) 11:05, 8 February 2020 (UTC)

@Sandbh: This is a total canard. J & S have addressed the heart of the issue, but you refuse to see it even when I spell it out for you. And the universal impact of the 14 f-electrons in Lu is the same as that in Hf, Ta, W, etc., all the way to Rn thanks to the Ln contraction. They are core electrons! And it's not even the first time I've said this.
I'm tired of this endless double standard. You throw out a barrage of arguments that look like they might support the La table, and flip non-stop between them the moment one of them looks like it is in trouble. And then you selectively read the literature in order to not see refutations. And you never consider whether any of your criteria are truly fundamental by putting them to the test of the rest of the table, cloaking this masterpiece of inconsistency under the garment of minimising change. Which is more or less like taking the La table as an axiom and throwing out whatever you want to save it. All I can say is: this is not science. There is no way we can proceed if you refuse to follow scientific principles, actually put your theories to the test, and at least compare all possibly relevant criteria and no others equally regardless of whether they seem to favour La or Lu. If you don't do that, then I say: go ahead and get what you want published. But don't thank me for the critique in the acknowledgements, because I disagree with so much of your paper that I don't want it to look like I supported the final product. Double sharp (talk) 12:26, 8 February 2020 (UTC)

I’ll add the +2+3+4 maximum oxidation states pattern as something else observed by the La form but not by the Lu form.

I can come up with an infinite number of things that are only displayed in a periodic table if we put some set of elements into a vertical column. And none of those will be important, fundamental considerations worth paying attention to until and unless I actually demonstrate some reason why it should be fundamental. Given that I think everyone would agree that the possible existence of HgIV doesn't weaken Tl's placement in group 13, I'm not seeing it here. Double sharp (talk) 12:00, 3 February 2020 (UTC)

@Double sharp: Henry Bent was the source of this one (in his Fresh energy for the periodic law). He asked the question whether it would be possible to capture the LSPT from chemical data without knowledge of group membership for any elements whatsoever. He started with DIM's line and the elements' maximum oxidation states. He does get to the LSPT using this method and quotes DIM, "the forms of oxides and…atomic weights…give us the means to erect an unarbitrary system as complete as possible."

Bent does manage to extract some quasi-regularities in the LSPT, on this basis, but he missed the most regular one of all i.e. +2 +3, +4 because of Lu under Y!

The DIM heritage, quote, and the Lu irony is what makes this one important. You alerted me to the fact that Hg(IV) has not been reproduced. Sandbh (talk) 06:34, 4 February 2020 (UTC)

I know Hg(IV) has not been reproduced. But my point is: if it was, does it really impact the placement of Tl in group 13? And why is 234 so important anyway given that you will not find regularities throughout the table for 012, 123, 345, 456, 567, or 678? Why does it become fundamental and not just a neat coincidence? The 012 one being weak affects the alkali metals, that paragon of great group trends. ^_^ Double sharp (talk) 07:52, 4 February 2020 (UTC)

@Double sharp: I thought it was significant since it shows the La form is more regular than the Lu form, in this particular context. I'd describe its occurence as a manifestation of the periodic law, rather than a coincidence. It's particularly interesting to consider why this pattern arises, too. A pattern does not need to be reproduced elsewhere, in order for it to be relevant. It does raise an interesting question, as you said, why it doesn't occur so well for the 012, which leads into questions about the structure of the periodic table. Sandbh (talk) 01:08, 5 February 2020 (UTC)

@Sandbh: Certainly, it leads to the question why this matters at all. We are all agreed that the alkali metals are one of the poster children of great group trends, and I hope we agree that that is, in fact, a good manifestation of the periodic law. And yet group I shows this so badly: there isn't such a pattern recurrence for group IB (for which +1 is never the maximum oxidation state), and the 012 pattern goes away when the noble gases from Kr onwards become chemically active. Still hoping for the day the ArF+ salts are synthesised in the solid phase and I can start saying "Ar onwards", which will lower the pattern to literally two occurrences in the whole table. And once the helium compounds are discovered we can bump it down to one, at which point it is not even a pattern anymore. That gives me two questions about this whole exercise:
  1. What kind of a fundamental pattern is this when the poster child of group trends is on the verge of not showing it at all? I can come up with a billion regularities that are conditional on one form of the periodic table or another: for example, the outward form looks more symmetrical in Scerri's old version with H-F-Cl and He-Ne-Ar and four columns on each side flanking the transition elements. They all stand or fall depending on whether they are fundamental. And if they are actually fundamental, it is not too much to ask for their repercussions on the whole table.
  2. Why on earth are the indiscretions of the heavier noble gases (snooping around with that upstart commoner fluorine, who just latches on to everything and never lets go, oh my) relevant when contemplating the trend of the alkali metals?
Double sharp (talk) 16:22, 5 February 2020 (UTC)

@Double sharp: I don't have answers at hand to many of your questions. For me, the only relevance of the 234 pattern is to the group 3 question. This cuts across all tables. It doesn't work for the Scerri table you referred to me, for example. That the failure of the 234 pattern originated in Bent's successful construction of the LSPT—supposedly the epitome of regularity and symmetry—from only DIM's line and the elements' maximum oxidation states, is priceless. Sandbh (talk) 01:28, 6 February 2020 (UTC)

@Sandbh: I can come up with infinitely many patterns that are only relevant to one particular question in the periodic table, and none of them will mean anything unless you give some reason why it is important. The LSPT has many patterns that fail in the current table or Scerri's old table. The same can be said of the current table vs. the other two, or Scerri's old table vs. the other two. What is missing is a consideration of why the pattern is significant, and without them, this argument doesn't have anything to stand on. Double sharp (talk) 12:25, 6 February 2020 (UTC)

@Double sharp: I doubt you could come up with a pattern as significant i.e. with a direct link to the very same property used by DIM to construct his PT, or as precise. Sandbh (talk) 23:09, 6 February 2020 (UTC)

@Sandbh: And this isn't quite one of them either, since DIM evidently had sanity prevail when it came to putting O and F to head their groups. And he quite evidently cared only about the maximum valence of the actual elements under consideration, not their neighbours, or else putting the Zn group in group II becomes questionable. Double sharp (talk) 00:08, 7 February 2020 (UTC)

@Double sharp: I agree. A few exceptions in the margins don’t detract from the direct link. Sandbh (talk) 07:50, 7 February 2020 (UTC)

@Sandbh: Yes they do when the exceptions are in the groups we universally regard as the epitome of great group trends. Group I works terribly badly under your scheme (which prioritises looking at irrelevant neighbouring groups, as Mendeleev never did); group II still badly because of their group IB neighbours; group VII badly because (1) the first and only halogen to centre a 678 pattern is iodine and (2) manganese ends up missing from group VIIB, which gets rid of the one member of that group that Mendeleev actually was aware of. Double sharp (talk) 00:13, 8 February 2020 (UTC)

@Double sharp: Those exceptions don’t detract from the headlines. Group 1 functions quite nicely as a neighbour of group 2. Group 11 is a neighbour of groups 10 and 12. Group 17 is irrelevant to the fact of the 234 pattern existence. Mn is irrelevant too. The 234 link is significant since it shows the La form is more regular than the Lu form, in this particular context—maximum oxidation number—which was a primary focus of DIM. Sandbh (talk) 09:53, 8 February 2020 (UTC)

@Sandbh: For the last time: the maximum oxidation states, without looking at neighbours like you push in, were important to Mendeleev precisely because those exhibited periodicity throughout the whole table without much exception! If they were only relevant for one group, like your criterion is, he would either have found some other basis for periodicity, or not found one at all! So the moment you draw a line and say "everything is irrelevant for this criterion apart from what it says about group 3", it means "this means nothing for the periodic table". Double sharp (talk) 10:01, 8 February 2020 (UTC)
Ln contraction

I’ve argued that the lanthanide contraction fits naturally within the f-block, in an La table, whereas it runs over two blocks in an Lu table. I’ll add this as a congruency based argument.

See Droog Andrey's reply. Double sharp (talk) 12:02, 3 February 2020 (UTC)
He, Be, Mg, Sc, Al etc

We have had some subsidiary discussion in which Double sharp has argued that my arguments would require He over Be; Al over Sc; or Be, Mg over Zn. My position is that there are several other arguments for He over Ne, and Be and Mg in group 2, and Al in group 3 and that despite the popularity of the La form over the Lu form by a wide margin, arguments for such changes have never gotten up. Sandbh (talk) 06:51, 3 February 2020 (UTC)

So we can now take any argument, oblivious to what else it would demand, and advocate in isolation for any placement. And as long as the literature is united, it must be right. Then why are we here? Double sharp (talk) 07:54, 3 February 2020 (UTC)
@Double sharp: I hope the situation turns out to be more sophisticated than that. I recall while we were talking about He, Be, Mg, Sc, Al etc you were arguing that one or more of my arguments for La would imply the merits of one or more of the other movements in question. My response was that there were other arguments for the current situation of He over Ne, Be and Mg in group 2, and Al in group 13, related and unrelated to my case for La, that would work against these movements. But I won't know for sure until my library day, and I have an opportunity to unpack what we discussed.
One impression I did form is that changes to the PT should be minimal to achieve the intended outcome. Presumably Lu in group 3 would suggest He over Be, but that would breach my suggested principle. Each change proposal stands on its own. So, if there is a case for one or more of these moves on the basis of either La or Lu in group 3, that's fine but don't drag me into to it so to speak or suggest that if either La or Lu go in group 3 it would follow that one of the other moves must follow, therefore this speaks against La or Lu in the first place. It's analogous to arguing that if we do something, then it will follow that the sky will down which it won't of course; the PT is much more resilient than that.
@Sandbh: That's why I'm consistent. I argue for Lu in group 3 based on blocks and recognise that this suggests He in group 2. I then have two logically consistent options:
  1. Decide that blocks are not so good an argument after all (modus tollens);
  2. Decide that He should be in group 2 indeed (modus ponens).
This is just basic philosophy and logic. Since blocks are pretty fundamental (since chemically active subshells rather than differentiating electrons control chemistry), I choose option 2, and shunt helium over to group 2. You seem to want to have your cake and eat it too: have your arguments, claim they are fundamental considerations, and not look at where they lead for the rest of the periodic table. With respect, that is trying to pick the actions and then the consequences. You don't get to do that. Double sharp (talk) 11:53, 3 February 2020 (UTC)
Of course, as a respected colleague, you are free to argue as you see fit.
The literature is generally right, in my experience. When I was doing my Masters in Human Resource Management, I was told my opinion didn't count---if I wanted to at least pass---so I had to support all my opinions and arguments with citations. When I write for academic journals, since I don't have any science qualifications, I generally always support my arguments with citations. In a way I'm pleased I don't have any science qualifications because I can ask questions without preconceived notions of what can, can't or shouldn't work wrt to the periodic table. Having said that if I'm going to challenge something or advance a position I do generally seek to support my argument with a citations. Some of my other colleagues interpreted this as preaching to them, until I explained my background to them. The citations I do provide have helped others too, even professional chemists.
In saying the literature is generally right, that doesn't mean there is no room to question, challenge, reinterpret or develop the literature. And it is sometimes wrong, contradictory, silent or confusing (I blame a lot of that on "publish or perish"). Take the notion that elemental metals reduce their electrical conductivity as the temperature increases. Wrong! Pu increases its electrical conductivity when heated in the temperature range of around –175 to +125 °C. Or that H behaves like a metal. Wrong! Hydrogen generally always finds a way to complete its valence shell, in this way behaving as a nonmetal. Or that the Ln contraction runs from La to Lu. Wrong! It starts at Ce and finishes in Lu. Or that there are metals, metalloids, and nonmetals. Wrong! This notion of an intermediate class has been the source of decades of unnecessary confusion, and mumbo jumbo like all metalloids are semiconductors. Wrong! Metalloids are much more easily thought of as chemically weak nonmetals etc (this will be elaborated in my forthcoming FoC article, "Organising the metals and nonmetals). Sandbh (talk) 09:56, 3 February 2020 (UTC)
In fact, the contraction goes all the way from Cs to Hg. You can take a part of it (say, from La to Lu) to explain something, but the boundaries are to be selected arbitrarily. Droog Andrey (talk) 10:36, 3 February 2020 (UTC)
@Droog Andrey: +1 Double sharp (talk) 11:44, 3 February 2020 (UTC)
@Sandbh: And equally: take the notion that only group 3 among the early transition metal groups lacks characteristic transition metal properties of multiple common oxidation states. Wrong! The same is true for Zr, Hf, Nb, and Ta. Double sharp (talk) 11:56, 3 February 2020 (UTC)
@Droog Andrey:@Double sharp: I understand. In my view, it's useful to be able to divide the contraction by (a) magnitude, thus (1) Ce to Lu; (2) La; (3) Ba; and (b) knock-on effects (Hf +). I don't regard that as being arbitrary. Sandbh (talk) 00:58, 4 February 2020 (UTC)
@Sandbh: Where is the big difference in magnitude? We just see a monotonic decrease all the way, and the difference from La to Ce is no different qualitatively and quantitatively from the difference from any other lanthanide to its Ln neighbour. Double sharp (talk) 19:49, 4 February 2020 (UTC)
@Double sharp: Thank you. If I read you correctly, that's an inaccurate description of the notion. The notion is, "multiple oxidation states", rather than "common" such states. The literature distinguishes group 3 from group 4 due to the well-established chemistry of Ti, Zr and Hf in lower oxidation states, especially for Ti, noting these states are nevertheless uncommon. The literature further distinguishes group 3 from group 4 according to the predominant ionic chemistry of the former v the predominant covalent chemistry of the latter. Could you please see my supporting quotes, including the one by Rayner-Canham and Overton, re the importance of ionic v covalent, that I posted a few minutes ago. Sandbh (talk) 00:58, 4 February 2020 (UTC)
@Sandbh: Then the literature you quote has certainly never heard of CsScCl3, not to mention the recent explosion of work done on Ln(II) compounds. Nothing wrong with that; G & E aren't aware of Cm(VI) in their book either. Presumably the burgeoning on Ln(II) chemistry came after the books you cite were published. But it significantly weakens your point. Knowledge about group 3 has moved on from "only a +3 state". (On my phone right now, will link later.) Edit: links appear in section on G&E 2nd ed. Double sharp (talk) 07:47, 4 February 2020 (UTC)
@Double sharp: This does not change the fact that groups 1-3 are predominately ionic and groups 4 to 12 predominately covalent. Sandbh (talk) 00:51, 5 February 2020 (UTC)
@Sandbh: See our replies below. Respectfully, this distinction is nonsensical. Double sharp (talk) 16:07, 5 February 2020 (UTC)

On a lighter note: wow, we passed 500k! ^_^ Double sharp (talk) 17:27, 5 February 2020 (UTC)

And on a side note: why is the notion of lower oxidation states suddenly dropped when it is obvious that it no longer works as a La argument? It would be fairer to continue to consider both, noting that one of them supports Lu and the other (as much as it means anything at all) is somewhat of a La argument (although it can still be blown out of the water by simply focusing on one set of compounds or another, as I did by looking at fluorides instead of the chlorides we considered for the IUPAC submission). Double sharp (talk) 00:54, 8 February 2020 (UTC)

@Double sharp: I didn’t see the announcements in the world’s chemistry journals that +3 is no longer the predominant oxidation state in La and Lu, and that it is now +2. I don’t look at one set of compounds, I look at the chemistry of the element or group as a whole. Sandbh (talk) 09:39, 8 February 2020 (UTC)
  1. And yet you have no problem referring above to well-established lower oxidation states for the Ti group above, even outright saying the point was "multiple" oxidation states rather than multiple "common" oxidation states. Why not for the Sc group, then?
  2. "I don’t look at one set of compounds, I look at the chemistry of the element or group as a whole." – and yet you only look at one oxidation state, one item on the continuum from ionic to covalent, and one differentiating electron (until it gets to difficulties with Th and Lr when you backpedal). What is this if not a double standard? You want one of everything, so one basic principle for the PT might be a good start, respectfully. Double sharp (talk) 10:32, 8 February 2020 (UTC)
Predominately ionic

What the hell is going on with this? We know that groups 3 and 12 are much less "transition" than, say, 6 to 10. That's pretty normal. Why should we make 3 vs. 4 a special case? Yes, a lot of philosophy could be cited about the sharp edges and meanings of classifications, but what's the matter? Droog Andrey (talk) 10:30, 5 February 2020 (UTC)

+1 Exactly, there is not a big difference between group 3 and 4 here. +4 is too high for ionicity, OK, so look at the same elements in the +2 or +3 states and they suddenly become more ionic like uranium by Fajans' rules. Everything is continuous. Meanwhile, "predominantly ionic" continues to be complete nonsense, as can be seen by asking: is fluorine predominantly ionic? Is caesium predominantly ionic? Is uranium predominantly ionic? Is thallium predominantly ionic? None of these questions have a good answer, it depends too much on the chemical environment. Just because some literature has used unfortunate terminology that literally means nonsense does not give it a free pass from being nonsense. Double sharp (talk) 15:57, 5 February 2020 (UTC)

@Droog Andrey:@Double sharp: I don't know. I'm not saying any more than the chemistry of groups 1 to 3 is predominantly ionic, whereas the chemistry of groups 4 to 12 is predominately covalent. It's like saying in my street that, predominately, the houses are made of brick (say forty brick and ten wood). Whereas in the next street down from us the houses are predominantly made of wood (say forty wood, and ten brick). The most common oxidation state of Tl is +1 and the compounds of Tl in this oxidation state may be covalent, as for example in thallium acetylacetonate, but more frequently they are ionic (Durrant & Durrant 1962, Introduction to advanced inorganic chemistry, p. 558). Sandbh (talk) 00:42, 6 February 2020 (UTC)

@Sandbh: So, two follow-up questions:
  1. By this logic, is not the placement of Tl as a p-block element weakened? Because due to typically higher oxidation states, the p-block elements are mostly pretty weak in their ionicity, see for example Ga and Sn. By these standards, Mendeleev's initial placement of Tl as eka-Cs seems to look better.
  2. So you admit that oxidation state matters, as by itself the +3 state of thallium is assuredly not so ionic. (It is much more electronegative.) So why not do as Allred did, and consider different oxidation states separately? Remember, organothallium chemistry has the +3 state much more prominent, so saying the most common oxidation state of Tl is +1 demands a caveat about what distribution of compounds we are talking about! ^_^ And from there we can stop trying to crush everything down to a yes-or-no binary answer to "what predominates", and realise, of course, that there was a continuous trend all along that Fajans formulated in 1923. Double sharp (talk) 12:23, 6 February 2020 (UTC)

@Droog Andrey:@Double sharp: Actually I do know, but not the reason why it matters to Double sharp. For example, the chemistry of group 1 is predominantly ionic. Double sharp says this is meaningless in that the chemistry of group 1 depends on the environment. I agree and I'm sure we could find some degree of covalency in Group 1 if the conditions were extreme enough. That does not negate, however, the statement the chemistry of group 1 is predominately (i.e. not exclusively) ionic. Here's another example. Group 1 predominately form +1 cations. In some circumstances some of them can form -1 anions. Nice. Interesting. It nevertheless remains true that group 1 metals predominately form +1 cations. I hope this helps. Another example. The population of Australia is predominately urban. Etc. Sandbh (talk) 00:52, 6 February 2020 (UTC)

Sandbh (talk) 00:52, 6 February 2020 (UTC)

@Sandb: This is still wrong, respectfully. Well, take Cs compounds with every other element. Most of them are metals, up to group 12, and the bonds of Cs with those metals will be metallic. So "group 1 metals predominately form +1 cations" is only true when they are bonding with the more nonmetallic elements. Which is a wider class than for anybody else, admittedly, because with the group 13 elements already they form M+ (some Zintl ion), and it takes an incredible amount of electropositivity to make the arsenides ionic (even lanthanum apparently cannot do it). But most elements are not in the p-block. Double sharp (talk) 12:23, 6 February 2020 (UTC)

@Double sharp: I see. The compounds that Cs forms with metals are electrically conducting alloys, comprised of a mixture of Cs cations and the other metal's cations. Like C & W say, "the chemistry of these elements is principally that of their +1 ions...The chemistry of these elements is mainly that of ionic salts in the solid state and solvated cations" etc. Sandbh (talk) 23:05, 6 February 2020 (UTC)

@Sandbh: That's not ionic bonding. That's metallic bonding. Double sharp (talk) 23:55, 6 February 2020 (UTC)

@Double sharp: Would you count them as chemical compounds? Sandbh (talk) 07:46, 7 February 2020 (UTC)

@Sandbh: The world certainly does, see intermetallic compound. Double sharp (talk) 00:35, 8 February 2020 (UTC)
@Sandbh:@Double sharp: There's not a problem that caesium compounds are predominantly ionic. The problem is: why that specific predominance (among a heavy lot of others) becomes so important? Why we don't rip apart Groups 13 and 14 on the basis that, for example, Group 14 compounds are predominantly coloured while for Group 13 they are not? Droog Andrey (talk) 17:23, 7 February 2020 (UTC)

@Droog Andrey:@Double sharp: "For chemists…the most important feature of an element is its pattern of chemical behaviour, in particular, its tendency toward covalent bond formation (or its preference for cation formation)." (Rayner-Canham Overton 2010, p. 29) Sandbh (talk) 09:16, 8 February 2020 (UTC)

@Sandbh: So what happened to the maximum oxidation states you like to talk about when it comes to supporting the 234 bad argument? Or do we have yet another double standard where you pick and choose the criterion most likely to save the La table in each context? Double sharp (talk) 10:29, 8 February 2020 (UTC)

@Double sharp: What has the 234 argument got to do the with the ionic chemistry of Cs? Sandbh (talk) 12:01, 8 February 2020 (UTC)

@Sandbh: Sorry, this is poorly phrased and placed, and I apologise for it. What I meant to say is something that still makes me uncomfortable about this. In the 234 argument you seem to be focusing on maximum oxidation states if I read you rightly, and here you take ionic vs covalent. In case they contradict each other, as for Be-Mg-Zn, who wins? (The 123 pattern works better with Be and Mg in group 2, but they are more covalent like group 12.) Or will you jump once again to ground-state gas-phase configurations for that one? Or decree that this is off-limits because it's not strictly about group 3, never mind that a fundamental basis for the table has to be relevant for everything (that's why in our submission we argued "well, what about the s-block" against considering carbonyl valencies)? That's why I want to ask: please set down your criteria, a precedence order for them, and reasons why each one is important. That's the only way you can get a consistent basis for the PT. Double sharp (talk) 12:28, 8 February 2020 (UTC)

@Double sharp: Hmmm. The 123 horizontal triad fails at He, in an La table and an Lu table, so that doesn't help with the group 3 question. There is a 123 pattern for Li-Be-B, and Na-Mg-Al, and K, Ca, Sc. This does not change in either table, so does not help. I wouldn't say horizontal triads are necessarily fundamental, noting however their use by DIM and Newlands, and Dias' opinion about them connecting the whole table. I'd say they were more of a piece of the puzzle.

Nelson examined periodicity in carbonyls and on this basis supported Lu in group 3. He argued that the number of outer electrons possessed by an atom, and the number required for it to achieve an inert gas configuration exhibit an almost exact periodicity, which it didn't, as we argued. We further queried what happens to the s block if one takes this approach. We said we're not aware of s-block carbonyls, but this would seem to suggest that Mg (which needs 6 electrons to achieve the [Ar] configuration) cannot be placed above Ca (which needs 16 to achieve the [Kr] configuration), and that Sr (which needs 16 to get to [Xe]) cannot be placed above Ba (which needs 30 to get to [Rn]).

I don't see any issues with our logic here, but happy to discuss further.

More to follow re criteria etc. Sandbh (talk) 07:03, 9 February 2020 (UTC)

@Sandbh: This is an exact demonstration of why I feel your logic is biased. When examining a Lu argument, namely that of Nelson, you are willing to apply modus tollens. The logic goes like:
  1. The carbonyl approach predicts Be-Mg-Zn (since they all need 6 electrons to get to the next noble gas configuration), i.e. P implies Q.
  2. Be-Mg-Zn is not good, for other reasons, i.e. not Q.
  3. Therefore the carbonyl approach is not good, i.e. not P.
But you seem unwilling to apply modus tollens to a La argument such as isodiagonality, where I say:
  1. Isodiagonality works better with Al in group 3, so that Al is really diagonally adjacent to Be and Ti, i.e. P implies Q.
  2. But Al over Sc is not good, for other reasons, i.e. not Q.
  3. Therefore the isodiagonality approach is not good, i.e. not P.
Or when I apply it to horizontal triads for oxidation state, where I say;
  1. Such triads work the worst for groups I, II, and VII, so they should not be considered homogeneous, i.e. P implies Q.
  2. And yet those groups are widely considered the most homogeneous, i.e. not Q.
  3. Therefore those triads are not particularly important in the grand scheme of things, i.e. not P.
Why is it that me referring to what happens outside group 3 is irrelevant when using modus tollens for a La argument, but you can do that with impunity when attacking a Lu argument like that of Nelson? Double sharp (talk) 21:26, 9 February 2020 (UTC)
Classification science

A small, relaxed contribution:

“At any given time, during the historical development of a scientific discipline, classification of available evidence offers itself as the explanandum that asks for a theory (or alternative theories) able to explain it. But this is just one segment in a potentially unending chain of recursive relationships between classification and theory. Theory and classification indeed change over time. As a consequence, the theory that provides explanation for the data organized in a classification at a given time can influence subsequent classificatory effort, and so on. “By means of this a discipline advances: each new pattern raises questions that call for explanations, and each verified phenomenon or fact gives a new pattern” (p. 163). What counts as a fact or a theory is a matter of temporal relativity. The authors’ “concern is that we do not replace observation with theory and think that we have made some progress. Science is founded upon empirical observations, no matter how these are tied up with local and cross-disciplinary theoretical commitments or stances. Once we abandon this aspect of science…science becomes little more than a matter of worldviews and epistemic statements of faith” (p. 163).”

Minelli, A.: The nature of classification: Relationships and kinds in the natural sciences—By John S. Wilkins and Malte C. Ebach. Systematic Biology. 63 (5), 2014, pp. 844–846

Sandbh (talk) 11:30, 3 February 2020 (UTC)

"Perhaps in time the so-called Dark Ages will be thought of as including our own" – Georg Christoph Lichtenberg.

Exactly, that's why arguments based on the predominant form in today's literature, or today's categories, are not useful. You have to look at fundamental properties anew. Double sharp (talk) 12:07, 3 February 2020 (UTC)
@Double sharp: I agree with the sentiment, but not your interpretation of it. Arguments based on the predominant form in today's literature, or today's categories, may or may not be useful, rather than being not useful. What's important is the merit of any new perspective. Sandbh (talk) 23:50, 3 February 2020 (UTC)
@Sandbh: Well, that's your perspective. My perspective is that if we are seeking a fundamentally best periodic table, it's not enough to just appeal to tradition. To read the tradition, a thousand times yes; to understand why the tradition made their choices, ten thousand times yes, indeed! But to use something just because it keeps some tradition in some way means nothing. You must analyse the tradition and draw your own conclusion independent of it, even if it is based on all you learned from it. It even means less than nothing when it is not clear what the tradition itself is, such as for the boundary of the REM we use; and when it is not clear why the way in which the tradition is claimed to be kept means anything, as for your argument about unspooling the REM as a continuous line. Double sharp (talk) 19:56, 4 February 2020 (UTC)
@Double sharp: Rather than keeping tradition I'm upholding what represents an accumulation of scientific thought. Like Jones said, "Though classification is an essential feature of all branches of science, there are always hard cases at the boundaries. The boundary of a class is rarely sharp…Scientists should not lose sleep over the hard cases. As long as a classification system is beneficial to economy of description, to structuring knowledge and to our understanding, and hard cases constitute a small minority, then keep it. If the system becomes less than useful, then scrap it and replace it with a system based on different shared characteristics." I don't see sufficient merit in changing the status quo, as an accumulation of scientific thought. Please see my response in the REM subsection. Sandbh (talk) 02:59, 5 February 2020 (UTC)
@Sandbh: Would we ever have gotten rid of Be-Mg-Zn if we had done that? Double sharp (talk) 17:13, 5 February 2020 (UTC)
@Double sharp: Yes, it was gotten rid of with the development of electronic periodic tables. Sandbh (talk) 00:19, 6 February 2020 (UTC)
@Sandbh: But if we followed your quote's advice, then since Be-Mg-Zn is mostly an OK classification, and the hard cases are a small minority, it should have been kept. Double sharp (talk) 12:13, 6 February 2020 (UTC)
@Double sharp: You have to consider how the quote works in real life. There is no drama over Be-Mg-Zn since Be and Mg are s metals, whereas Zn is a d10s2 metal. We've discussed this before, in the arena. Sandbh (talk) 22:58, 6 February 2020 (UTC)
@Sandbh: The little problem with that is that at the time Be-Mg-Zn was popular, there was enough drama for authors to apologise when they moved Be and Mg over Ca. And it was certainly "beneficial to economy of description, to structuring knowledge and to our understanding", because everything else is the same, and Be and Mg are indeed more like Zn than like Ca. So the system did not become "less than useful", and yet we still went and scrapped it. Why? Because there are other factors, like figuring out something deeper that is driving the chemistry we see. All right, then; we have already figured out that chemistry has a lot more to do with generally considering chemically active subshells rather than focusing only on ground-state configurations, so there is enough merit to change to Sc-Y-Lu. Simply put: how do you know we're not in a Be-Mg-Zn situation? Right now we see only the Ca table, so if anyone suggests a move back to the Zn one, we will hear "not enough merit for the sea change needed". Well, I put it to you that if the Lu table were to win, nobody would be suggesting a move back to the La one for exactly the same reason! But as long as the La one is around, you will hear constant arguments to scrap it. Double sharp (talk) 00:00, 7 February 2020 (UTC)
@Double sharp: It's interesting to see Jensen's paper has 65 citations. Judging from Google Scholar, none of these 65 citations were about supporting Be-Mg over Zn. Jensen is in the wilderness on this one. Sandbh (talk) 06:47, 9 February 2020 (UTC)
@Sandbh: You're missing his point. He's not supporting Be-Mg-Zn. He's saying that this was the more common classification historically.

Indeed, prior to the introduction of electronic periodic tables, the similarity between Be and Mg and Zn and Cd was often considered to be greater than the similarity between Be and Mg and the rest of the alkaline earth metals (Ca–Ra). Many inorganic texts written before the Second World War placed their discussion of the chemistry of Be and Mg in the chapter dealing with the Zn subgroup rather than in the chapter dealing with the Ca subgroup, and the same is true of many older periodic tables, including those originally proposed by Mendeleev (34, 35). Even as late as 1950, N. V. Sidgwick, in his classic two-volume survey of The Chemical Elements and Their Compounds, felt that it was necessary to justify his departure from this scheme in the case of Mg (36).

And that's all I am saying. The system was useful, with hard cases a tiny minority, and yet we scrapped it. Why? Because our understanding progressed and the consensus changed. Eventually that may well happen with the triumph of Sc-Y-Lu, and looking at the chemistry and the actually important things such as chemically active subshells, that would be a most excellent outcome IMHO. Double sharp (talk) 21:29, 9 February 2020 (UTC)
Ionic vs. covalent

One more contribution:

"For chemists…the most important feature of an element is its pattern of chemical behaviour, in particular, its tendency toward covalent bond formation (or its preference for cation formation)."

Rayner-Canham G and Overton T, In Descriptive inorganic chemistry (2010, p. 29)
I wonder how they deal with fluorine, say? Depending on what it's bonded to it either forms (more or less polar) covalent bonds, or anions. Double sharp (talk) 00:10, 4 February 2020 (UTC)
C & W at least say that most metal halides are predominantly ionic. Sandbh (talk) 01:37, 4 February 2020 (UTC)
Then they've forgotten the salient point that said ionicity depends on the oxidation state involved, see the uranium fluorides from UF3 going up to UF6. And even then they've noticed that ionicity vs covalency of a halide depends on metallicity of the counter-cation, i.e. electropositivity, though the point about ionic radius and hence polarisability is not in your quote. I am not even saying anything new here; Kazimierz Fajans codified this in his rules all these years ago back in 1923. Double sharp (talk) 07:42, 4 February 2020 (UTC)
@Double sharp: I'm sure they took that into account, consciously or subconsciously, when they made their general statement. Such a statement does not need to detail its supporting considerations. Sandbh (talk) 03:05, 5 February 2020 (UTC)
@Sandbh: I'm sure they didn't, because that's not something supporting their statement, it's something refuting it. If they had thought about that they might have chosen their words a little more carefully and said "metal halides are predominantly ionic unless the metal is in a high oxidation state". Think WF6 and ReF7. And in case you want to call this "Double sharp v the world" again, the world has a term for this: fluoride volatility. This is used when separating radionuclides, and here the important gap in oxidation state is not the 3 vs. 4 you like to focus on, but 4 vs. 5. So much for discontinuities: in different contexts they move and exemplify a larger continuity. Double sharp (talk) 16:03, 5 February 2020 (UTC)
@Double sharp: The expression "predominantly ionic" needs no further qualification. If I have six snooker balls, 4 red, 1 yellow, and 1 blue, I can say they are predominantly red. I don't have to add, "unless the ball is yellow or blue". Or the chemical elements are predominately metals, etc. Sandbh (talk) 00:25, 6 February 2020 (UTC)
@Sandbh: But you don't have that. You have a lot of snooker balls, with a correlation that the bigger they are, the bluer they are. So they go along with from some small red ones (e.g. CsF) through some medium-sized purple ones (e.g. BeF2, TiF4, SnF4) all the way to the large blue ones (e.g. WF6). Not to mention the other related factor of electronegativity (e.g. TiF4 vs. ZrF4, GeF4 vs. SnF4). As a result, saying "most are red" is (1) not the full story, because there is some other factor that is clearly at work controlling the colour, and (2) is not even clearly true; just think about how many metals show an oxidation state over 3 or 4, and how many show multiple oxidation states whose fluorides behave differently (UFn, n = 3, 4, 5, 6). It's not for nothing that compounds of fluorine divides low-oxidation state from high-oxidation state metal fluorides. Double sharp (talk) 12:12, 6 February 2020 (UTC)
@Double sharp: We've discussed this before in the arena. 4 out of 6 = a predominance. It's equivalent to saying of all Ln compounds, most are ionic in nature, taking everything into account, including e.g. all possible oxidation states, and all possible EN differences. Sandbh (talk) 22:58, 6 February 2020 (UTC)
@Sandbh: It's not much of a predominance if it fractures so well along EN and oxidation state lines. That tends to mean you have two different classes here and arguing about predominance is just the tyranny of the majority group (which in this case is not even much of a majority, segueing into my next point). Also, 4 out of 6 (even if it was relevant) is not much of a predominance either. If a predominance is not skewed so far as to at least 90% or so you cannot neglect the other one. Like you do for U and Tl, only considering the +6 and +1 states respectively, and forgetting how important the +4 and +3 states are too respectively. I look forward to seeing this approach used on Tc, where it is extremely hard to say what the most common oxidation state is because it likes converting between them so much. That in itself tells me that it's reducing the complexity of the situation far too much. You have to at least look at all common oxidation states and notice the pattern between them. And once you do that, the whole misapprehension about "predominately ionic" is blown out of the water. Double sharp (talk) 00:04, 7 February 2020 (UTC)
@Double sharp: It's a good enough, useful generalisation. I'll rely on the fact that 4 > 2, and the wide use of the concept of "mainly" etc in the scientific literature to make useful generalisations and characterisations, analogous to e.g. most elements are metals. In the specific case of the Ln I'll continue to rely on the 100% agreement in the literature that the chemistry of the Ln is mainly (i.e. not exclusively) ionic. Mainly is good enough to make a useful generalisations, acknowledging the exceptions. Sandbh (talk) 00:32, 7 February 2020 (UTC)
@Sandbh: One more time: something like 51% vs. 49% is totally not predominant behaviour. (This is the sort of situation we are in for technetium with a bunch of equally common oxidation states.) Something like 70% vs. 30% is still not it because the minor oxidation state can be a big part of chemistry. Crack open any inorganic book about the p-block elements, it'll tell you that for the heavy ones the important one is the interplay between the group oxidation state and the one two oxidation units lower. See, both are considered. You appear to want to reduce the chemistry of the elements to something tiny: one oxidation state, one typical EN difference, one differentiating electron, one of everything controlling everything. What a magnificent situation this will be:
(I think I like your country's literature, Droog Andrey! ^_^ Maybe not surprising, as I am a fan of E. T. A. Hoffmann.) Double sharp (talk) 00:20, 8 February 2020 (UTC)

@Double sharp: Yes, of course, 51 > 49. Is 51 predominately greater than 49? No, I’d say it's barely greater. I’d say something like 70% vs. 30%, at a 2:1 ratio, would represent a predominant majority. The 30% could still be important, of course. For example, weeds take up 70% of my lawn, and therefore represent the dominant form of growth in my lawn. The remaining 30% occupied by grass is important, but not predominant. Group 13 is mainly known in the +3 oxidation state but the +1 oxidation state becomes more important for thallium. I can tell this from our oxidation state article. Yes, I do seek to reduce the chemistry of the elements to their most common oxidation states, in order to map the broad contours of the situation; to typical EN differences for the same reason; and to differentiating electrons in order to delineate the s, p, d and f blocks. As another example, the fact that there are the four seasons of spring, summer, autumn and winter represents a controlling insight into the weather (at least where I live). Sandbh (talk) 09:08, 8 February 2020 (UTC)

@Sandbh: But you're not even consistent about it. You do it for La and Ac and claim that placing them in the f-block is verboten. Then you look at Th, think "ah, that's a bit inconvenient", and save yourself by going to an uncommon oxidation state (+3) and retreating away from differentiating electrons. And while you like talking about how Nature is supposedly like and not how easy She is to draw, you then insist that Lr must go in the f-block under Lu because there is nowhere else that it can practically go(!). IMHO, you are making exactly the same mistake as Lavelle: inconsistency. At least when my criteria have a little problem (the Ca group), I accept the inconsistency, and it actually points to something real in the chemistry. Now, remind me, exactly how much does the DE anomaly between Nb and Ta mean? Absolutely nothing. Is Pb categorically different from its lighter congeners because of its different major oxidation state? No, Ge shows it already, Sn more so, so it is a continuum. And the stability of Pb(IV) anyway depends on ligands, so the only sane thing to do is to mark both 2 and 4 as important and consider their relative importance down group 14. As everyone does. Except you, who seem to want to reduce everything to one dimension. Sandbh vs the World, respectfully. Double sharp (talk) 09:56, 8 February 2020 (UTC)

@Double sharp: I don’t claim La and Ac in the f-block are verboten. I argue they are better placed in the d block, in a chemical table. I do so on the basis of shared chemistry; differentiating electrons; the periodic law; a pattern inconsistency in the REM; a unique 234 pattern; and isodiagonality. These are linked arguments. An Lu table is less regular in all these aspects. Th is as anomalous in an La table as in an Lu table. Lr is superficially a “problem” but this goes away in the context of e.g the differentiating electron argument. Yes, Ge (most stable oxidation state +4) is categorically different from Pb (+2), in that context. The d/e anomaly between No and Ta means as much for an La table as for an Lu table. Sandbh (talk) 11:36, 8 February 2020 (UTC)

@Sandbh: Shared chemistry supports Lu, DE's are chemically irrelevant, you misread the periodic law, your pattern is only graphic design, the 234 pattern is ungeneralisable, and isodiagonality supports Al in group 3. Your case for La convinces me about 0%. All it is is a bunch of one-off arguments glued together, with no hint as to which ones are the more important ones, and no heed to what nonsense they have to say for the rest of the periodic table. The only commonality with them is that they all support the La table in what IMHO are just increasingly desperate ways having less and less to do with actual chemistry. Since R8R and Droog Andrey have echoed some of my points, not to mention Jensen and Schwarz, I think the world would agree. At least I critically analyse and accept the tiny sparks of nonsense my single highest criterion appears to throw up (He in group 2, the ambiguous position of some s-block elements), and see that it is in fact not nonsense and has some chemical repercussions in the real world. So all is well. Double sharp (talk) 12:38, 8 February 2020 (UTC)

If you claim Ge is categorically different from Pb, why do you put them in the same column? You have DIM to appeal to if you want to put Pb under Ba. But I bet you won't do it, in order to not rock the boat too much. So this is just like Lavelle: La must go under Y, ultimately because that is the way the majority draws it. Respectfully, we would never have gotten rid of Be-Mg-Zn with that logic. Double sharp (talk) 12:41, 8 February 2020 (UTC)

@Double sharp: Err, the periodic law and the real world aufbau principle (the latter which explains why we got rid of Be-Mg-Zn)?
@Sandbh: Precisely, our understanding progressed, and we went for a new basis that pointed to Be-Mg-Ca. As it will progress from just looking at chemically irrelevant gas-phase differentiating electrons to configurations in all chemically relevant environments, which will end up meaning looking at chemically relevant subshells and Sc-Y-Lu. Not only Jensen, Jørgensen, and Schwarz, but also Seaborg mentioned this. Double sharp (talk) 21:31, 9 February 2020 (UTC)

Isodiagonality and group 3

Rayner-Canham has written extensively on isodiagonality.

The three examples of Li and Mg, Be and Al, and B and Si, are commonly cited.

La table
The following isodiagonal relationship can be observed:
2 3                  4
Ca Sc.................Ti
Sr Y..................Zr
Ba La Ce Pr Nd etc

  • all three elements (Ca-Y-Ce) are strongly basic
  • similarities between Ca and the lanthanides (inc. Ce) are well known;
  • Y is a member of the rare earths, as are the lanthanides;
  • all three elements exhibit predominantly ionic chemistry.

Lu table
The following diagonal is observed:

2 3 4
Ca Sc Ti
Sr Y  Zr
Ba Lu Hf

  • Ca and Y are strongly basic; Hf is amphoteric;
  • Ca and Y exhibit a predominately ionic chemistry whereas Hf exhibits a predominately covalent chemistry;
  • Y and Hf can form intermetallic compounds with Pt, noting this capacity does not extend to Ca.

In a chemical table, from an isodiagonality perspective, La seems to be preferred. An Lu table introduces another irregularity.

I was surprised when I discovered this. I posted it here as I couldn't get the courier-text trick to work in the main thread. Sandbh (talk) 06:55, 8 February 2020 (UTC)

@Sandbh: Actually we have Na-Ca-Y-Hf in both La and Lu tables. In 32-column PT there's no diagonal after Na-Ca, but there is Rb-Ba-La-Th, again in both La and Lu versions. But if we compare La-Rf and Lu-Rf, the latter will stand much better. So the diagonal argument supports Lu table. Droog Andrey (talk) 23:36, 11 February 2020 (UTC)

@Droog Andrey: Yes, Na-Ca-Y-Hf is in both 18-col tables. In the 32-column La table, Na-Ca-Y-Hf disappears, and we see Na-Ca-Y-Ce. In the Lu table Na-Ca-Y-Hf remains. In both forms of the 32-column table there is no Rb-Ba-La-Th; you may have wanted to refer to K-Sr-La-Th. I don't know why you refer to La-Rf or Lu-Rf; these are not contiguous diagonal relationships. Sandbh (talk) 02:29, 12 February 2020 (UTC)

@Sandbh: Sorry, I thought about K-Sr-La-Th indeed. As for Lu-Rf vs. La-Rf: just look closer at the 18-column table. Droog Andrey (talk) 05:32, 12 February 2020 (UTC)

@Droog Andrey: No matter. Before I posted my first answer I printed two 32-col tables, one La and one Lu. After examining them I folded them into 18-col tables. In every case of isodiagonality it's possible to trace a zig-zag path. For example, with Li3-Mg12 the zz path is Li3-Na11-Mg12. With Be4-Al13, the zz path is Be4-Mg12-Al13.

In an La or Lu table it's not possible to zz La-Rf. In both tables it's possible to zz Lu-Rf. I don't see your point. Sandbh (talk) 23:16, 12 February 2020 (UTC)

@Sandbh: In an 18-column La table, the zigzag is 57La-89Ac-104Rf. Double sharp (talk) 23:23, 12 February 2020 (UTC)

@Double sharp: That's not a zz path since there is no contiguity with Z. There is a chasm of 14 elements between Ac and Rf. The best that can be done is 57La-89Ac-90Th. Sandbh (talk) 01:14, 13 February 2020 (UTC)

@Sandbh: Diagonality means "one period lower, one column to the right". So we have Na-Ca-In-Pb in 8-column, but Na-Ca-Y-Hf in 18-column, and Na-Ca-Y-Ce in 32-column. We also have:
K-Sr-Tl-Fl in 8-column,
K-Sr-La-Rf in 18-column La,
K-Sr-Lu-Rf in 18-column Lu,
K-Sr-La-Th in 32-column.
Droog Andrey (talk) 12:34, 13 February 2020 (UTC)
@Sandbh: I don't see why "contiguity with Z" recommends itself above other criteria for relevance of diagonal relationships. The name literally has "diagonal" in it, so elements in it should actually appear along a diagonal in the periodic table layout you're analysing. Why do you allow Ca to Y in a Lu 32-column table, even though there is a big chasm between Sr and Y there? Why is the placement of Al not a strike against the Be-Al diagonal? Double sharp (talk) 13:48, 13 February 2020 (UTC)

@Double sharp:@Droog Andrey: Oh dear. This looks like an example of extrapolating my argument beyond its boundaries.

Na-Ca-In-Pb is invalid since Ca and In are not next to each other in Z contiguity terms
Na-Ca-Y-Hf is invalid since Y and Hf are not next to each other in Z contiguity terms valid Sandbh (talk) 02:41, 15 February 2020 (UTC)
Na-Ca-Y-Ce is good
K-Sr-Tl-Fl is invalid
K-Sr-La-Rf is invalid
K-Sr-Lu-Rf is invalid
K-Sr-La-Th is valid.

I recognise Ca-Y (says he, quickly getting out his 32-col table to check :) in an Lu 32 table since there is contiguity in Z.

I recognise Be-Al (as does the entirety of the literature) since there is Z contiguity via Mg.

Are there are any more "mind-bending" counter-arguments? You guys must have missed something! I must be wrong somehow! Sandbh (talk) 04:45, 14 February 2020 (UTC)

@Sandbh: If you insist that Y-Hf is invalid, why do you refer to it in your argument? Droog Andrey (talk) 18:34, 14 February 2020 (UTC)
@Droog Andrey: My bad ×2. Now corrected. Sandbh (talk) 02:41, 15 February 2020 (UTC)
@Sandbh: This is yet another of those arguments that ends up supporting Al in group 3. Read his paper; he focuses more on Al-Ti than Al-Ge (mentioning the latter only as a minority view based on only some synthesis attempts), and his periodic table figure even has Al in group 3 to show this diagonal! Double sharp (talk) 09:48, 8 February 2020 (UTC)

@Double sharp: Al in group 3 is a non-starter for other reasons we’ve discussed, including its p differentiating electron. I must be onto something if the best you can do is attempt to distract the audience :) Sandbh (talk) 11:47, 8 February 2020 (UTC)

@Sandbh: Only for you is modus tollens irrelevant. So much for a philosophical perspective. Double sharp (talk) 12:39, 8 February 2020 (UTC)
I've moved this to the main thread, as the Courier text seems to be working fine. Double sharp (talk) 20:18, 8 February 2020 (UTC)

@Double sharp: Thank you.

I've added a space in front of this sentence. It shows as courier new in preview mode but won't in published change mode.

I wanted to add that this was a nice example of something Jensen referred to: arguing against "the relevance of such evidence using the time honored technique of first setting up a nonexistent straw man…and then attacking it".

It's not a straw man. It's just modus tollens.
  1. Sandbh claims that isodiagonality is relevant for element placement. (P)
  2. Isodiagonality supports Al over Sc. (Rayner-Canham may not have argued it explicitly, but he explicitly devotes more time to Al-Ti as isodiagonality than Al-Ge, and he shows an Al over Sc table.) (P implies Q.)
  3. Sandbh seems to agree that Al over Sc is not correct. (Not Q.)
  4. Conclusion: isodiagonality is not so relevant after all. (Therefore Not P.)
Double sharp (talk) 22:21, 8 February 2020 (UTC)

As well, Rayner-Canham did not argue for Al over Sc. Habashi (a metallurgist) did, but not on isodiagonality grounds per se. Being a metallurgist, many of his arguments ignore periodic trends going down a group, as well as the fundamental nature—from a classification science perspective—of the s-p-d, and f-classification. There may be a case for Al over Sc in a metallurgical periodic table.

It occurs to me there are now at least nine chemical arguments for La:

  1. Liu et al.;
  2. isodiagonality;
  3. horizontal triads;
  4. Atkins et al. Q & A;
  5. trends going down Group 3;
  6. monocations of Sc, Y, La and Lu;
  7. Laing's comparing the pairs Ca-Sc and Sr-Y with Ba-La;
  8. the Ln contraction starts in Ce and ends in Lu, noting your objections; and
  9. Restrepo's stoichometry findings, noting your concerns and that I've asked him about these;

I recall you disputed Laing so I'll have to go back and look at what you said; library day will be tomorrow :) Sandbh (talk) 22:04, 8 February 2020 (UTC)

  1. A table that claims La is closer to Au and Hg than the other 4f metals is just plain silly and discountable;
  2. Most of the elements don't display this, so what? And some elements display it better if you put them in "wrong" places.
  3. Not generalisable to the rest of the periodic table, therefore irrelevant. Most of the elements don't display this.
  4. Already debunked; at the 4d-5d scale, your increases are just noise.
  5. Supports Lu, because group 3 is a d-block group and should follow the d-block trend. If you dispute this, off to Be-Mg-Zn we go logically.
  6. Not sure why this is relevant since Sc, Y, La, and Lu are almost never in the +1 state.
  7. It's obviously irrelevant, just compare Be-B and Mg-Al with Ca-Sc, it's much closer than Ca-Ga.
  8. You have noted my objections, though actually addressing them would be a good start rather than just always saying "this is a tipping point argument" and continuing to argue from something that I have already demonstrated is quite irrelevant chemically (DE's);
  9. You have noted them here too.
Scorecard: 0/9 according to any sensible criterion that demands that properties trotted out as fundamental concerns for element placement on the PT be actually relevant to all or at least almost all the elements, not just "picking and choosing" for an arrangement you like. Double sharp (talk) 22:15, 8 February 2020 (UTC)

@Double sharp: Comments from teacher :)…

  1. La, Au, Hg. You know this. [A] All three metals are in the same period; [B] La shows similarities to main group metals in groups 1 and 2; remind me again of the emerging interest in the +2 oxidation state of the Ln, and compare this to Hg, which looks like it is one of the very few 5d metals with +2 as the predominant oxidation state; [C] Like La, Au has a +3 common oxidation state; Au has the capacity to act as a main-group metal in its common +1 oxidation state; it also shows similarities to the halogens, which are main group metals; and [D] Hg is a post-transition metal, i.e. main group metal, with +2 as the common oxidation state cf La.
    Are you kidding? Au and Hg, in their main-group-like states, are strongly post-transition. La is strongly pre-transition. Au is one of the worst metals in the periodic table and Hg is not far behind, whereas La is chemically strong. Just like Ce through Lu, in fact. Double sharp (talk) 21:39, 9 February 2020 (UTC)
  2. Diagonal relationships in the periodic table were recognized by both Mendeleev and Newlands. Most of the elements don't display the properties of the noble gases. Therefore the NG don't mean anything, just like the conclusion at the end of your scorecard sentence. Per Rayner-Canham, "Isodiagonality is, in some ways, a general attribute of the properties of the chemical elements. For example, the metal-nonmetal divide forms an almost diagonal demarkation (Edwards and Sienko 1983). Similarly, the elements often considered to be semimetals fall on a roughly diagonal border between the metals and nonmetals (Hawkes 2001)."
    And the Be-Al relationship shows up much better in their 8-column tables than in our 18-column form, because then it is actually a diagonal. So what? Just because something is common somewhere in the periodic table does not make it a fundamental law for it. Relativistic effects are today recognised by every single authority on the heavy elements from late period 6 onwards. That doesn't make them fundamental for the periodic table because they are totally irrelevant for the first 5. Double sharp (talk) 21:48, 9 February 2020 (UTC)
  3. Mendeleev used horizontal triads when he predicted the properties of the then undiscovered elements Al, Ga, and Ge. He discussed his technique using the horizontal triad As, Se, Br to estimate the atomic weight of selenium. Dias (1989; 1990; 1991) asserts that, “A periodic table is defined as a partially ordered set forming a two-dimensional array which complies with the triad principle where any central element has some metric property that is the arithmetic mean of two flanking [i.e. horizontal] member elements."
    But he never used them like you do when considering maximum valences. Otherwise he would have noticed that periodicity in those values simply did not work for what were otherwise the most homogeneous of his groups. Double sharp (talk) 21:48, 9 February 2020 (UTC)
  4. Ionic radii. This stands up on the basis of comparable properties. You argue against "the relevance of such evidence using the time honored technique of first setting up a nonexistent straw man…and then attacking it". The ionic radii in question are those of Shannon (1976), with 50,522 citations. All the people behind these citations are wrong, of course!
    If they say that ionic radii generally increase down the table, like Atkins does, they are absolutely right. That is generally what happens in the absence of a contraction. And indeed, it happens in the presence of most contractions as well, albeit diminished, and I'm sure they all know that effect of a contraction. In particular, I'm sure they all know that the lanthanide contraction, being the longest contraction thus far in the periodic table, almost exactly wipes the 4d to 5d increase out. And I'm sure they know that the increase wobbles about over a period. And I'm sure they learnt back in high school about significance, and realised that it doesn't really matter if the increase is actually a slight decrease going down from a few 4d elements to their 5d congeners, because the wobble around zero means that the sign is not significant. Therefore there is a very clear-cut reason why the general trend has been overthrown for this specific case that I'm sure all these authors are aware of. Therefore the failure of the ionic radius to increase for Y3+ to Lu3+ is not a mark against it because the same happens for the early 5d elements, where we have the exact same failure from Zr4+ to Hf4+, and for Nb5+ to Ta5+, for the exact same reason. So the problem is not that those authors are wrong, but that you seem to be wrongly taking their general statement which permits itself exceptions under extreme circumstances (one of which before us), and elevating it to a law. Double sharp (talk) 21:48, 9 February 2020 (UTC)
  5. Trends going down Group 3: Another paper tiger.
    Not a rebuttal. Double sharp (talk) 21:48, 9 February 2020 (UTC)
    To be read in Monty Python's voices: "That's not a rebuttal." "No, it's a figure of speech though." "I didn't think it was a very good argument." "No, it's the end of the thread, they must be running out of ideas." ;) Double sharp (talk) 22:01, 9 February 2020 (UTC)
  6. The +1 state is used to measure the ionisation energies of the elements, from which general trends are derived.
    The only problem with that is that this argument is about the chemistry of the +1 state, not just abstractly taking away that first electron. Whatever happened to sticking to characteristic oxidation states and properties? Double sharp (talk) 21:48, 9 February 2020 (UTC)
  7. Laing's pairs. There's no argument here since B and Al are p metals while Sc is a d metal.
    See my earlier post about your selective and biased use of modus tollens from my perspective. Double sharp (talk) 21:48, 9 February 2020 (UTC)
  8. Ln contraction. While I note your objections, there is nothing about them needing to be addressed, since the fact of the Ln contraction running from Ce to Lu, and the congruence of this with a Ce to Lu f-block, is not in question.
    It is obviously in question. You look only at the +3 cations, forgetting that it works just as well for atomic radii as for ionic radii, when all three electrons are still there. For no other contraction is there a characteristic oxidation state like that. Double sharp (talk) 21:48, 9 February 2020 (UTC)
  9. Restrepo. Interim comments by me:
@Double sharp:
Horovitz and Sârbu (2005), on the basis of 18 mainly physical properties, argued that Sc and Y are rather dissimilar to the lanthanides. They found that Yb, Lu, Eu and La were outliers. Eu and Yb are understandable given their low values of density and melting point compared with those of their immediate neighbors (indicating a 2+ oxidation state). That left La and Lu. They plumbed for Lu under Y given it was more of an outlier than La. We queried this conclusion noting the absence of any consideration or comparison of the merits of La.
In two earlier studies by Horovitz and Sârbu, based on physical, chemical, and structural data, there was more support for La under Y. And La was next door to clusters including Hg and Au =:o
I will read the paper later. But note: "next door" means different clusters. (A classification that puts La closer to Hg and Au than the other lanthanides, and Sc and Y, is by the consensus of just about everybody who ever used "group 3" or "rare earths" as a category nonsensical.) Also, I notice that Zr gets in as a main-group metal in one cluster (p. 481, hence probably Hf would as well but for some edge effects given their great similarity), supporting my point that main-group properties don't suddenly stop at group 3, but continue to be seen in 4 and to some extent 5 and even 6. Double sharp (talk) 12:15, 10 February 2020 (UTC)
Liu et al. found that, "Except La, all other lanthanide elements (inc. Lu) are classified into one cluster colored in pink, which reflects the strong similarities between lanthanide elements." Y has its own colour.
A classification that puts La closer to Hg and Au than the other lanthanides, and Sc and Y, is by the consensus of just about everybody who ever used "group 3" or "rare earths" as a category nonsensical. Double sharp (talk) 12:15, 10 February 2020 (UTC)
In the introduction to Scerri's Red Book, he wrote: "…there are in fact two big ideas in chemistry. They are chemical periodicity and chemical bonding, and they are deeply interconnected (p. xiii)." I contend this adds support to Restrepo's approach.
But Restrepo's approach makes it impossible to see relationships between elements with different characteristic valences. Since Mendeleev noted diagonal relationships as part of periodicity and so do you, it supports the idea that his approach is not complete. ^_^ Double sharp (talk) 12:17, 10 February 2020 (UTC)

Scorecard: 0/8. Very disappointing; careless, could do better. Your homework is to write a 5,000 word assignment explaining why the authors of the 50,522 articles citing Shannon were wrong :) Sandbh (talk) 05:23, 9 February 2020 (UTC)

DS' simple case for Lu

Whereas I have actually strong ones for Lu:

  1. Lu lacks 4f involvement but La has it, so chemically active subshells imply Sc-Y-Lu without any further ado.
  2. The resulting trends are more homogeneous, be they the trends in the d-block, the shape of the periodic table, the general changes of electron configurations (in all chemically relevant states), the general addition of a contraction before the first element in the next block, etc. etc. etc.

That's it. I would've said that this is unarguably chemically relevant and not minutiae, but it appears that you are keen to argue it even if it is self-evident that looking at all chemically relevant configurations must surely be more relevant than the minutiae of which one happens to be the gas-phase ground state. These are the ones that convinced me finally that Sc-Y-Lu was the better form: (1) implies that Lu as an f-block element is suspect, and looks at all configurations: gas-phase, condensed-phase, you name it. It's holistic. And (2) confirms that the result is OK. If (1) ordered something and (2) turned out not so good, that would be bad: but it works even for He over Be, so I'm logically consistent. Which, with all due respect, I don't see you being. Double sharp (talk) 22:24, 8 February 2020 (UTC)

@Double sharp:
1. Any 4f involvement in La is marginal. The 14 f electrons in the Lu3+ cation, are manifestly more significant in terms of their impact on the chemistry of Lu.
@Sandbh: Not this canard again. The 14 f electrons in Lu are important, yes. In the same way as those of Hf are important. And those of Ta are important. And so on and so on and so on. As core electrons. Blocks are supposed to be about valence electrons and vacancies. Otherwise, the 3d block must extend until Kr as the repercussions of the 3d contraction are still being felt till then. Double sharp (talk) 21:37, 9 February 2020 (UTC)
2. That the trends of the d-block are less homogenous is due to the delayed start of the filling of the 4f shell. The shape of the periodic table is an outcome of the real-world aufbau filling sequence (and the associated differentiating electrons) not the imaginary or idealised n+l rule. All chemically relevant states are not as relevant as the characteristic or predominant states.
@Sandbh: They are for sure more relevant than the gas-phase electronic configurations which are neither characteristic nor predominant in actual compounds. Double sharp (talk) 21:38, 9 February 2020 (UTC)
I haven't based my arguments exclusively on gas-phase configuration, as you know.
@Sandbh: No, but when you use them, you inconsistently flip between when they are decisive (for La and Ac) and when other factors can overthrow them (like for Th and Lr). As Jensen commented about Lavelle: "When it comes to the question of why La and Ac should remain in the d-block rather than being reassigned to the f-block, Lavelle offers no new chemical or physical evidence other than his constant reiteration of the fact that both elements contain d-electrons in their ground-state valence configurations, but no f-electrons. Yet in the cases of both Lu and Th, for which this is equally true, he proceeds to inconsistently argue that this fact is of no consequence when it comes to assigning them to the f-block. As with the case of the revised configuration for Lr, which counts when it comes to not placing this element in the d-block but is irrelevant when it comes to placing it in the f-block, this arbitrary and naive use of electron configurations, to the exclusion of all other evidence, is logically inconsistent and leaves one with the impression that the only true argument that Lavelle has for the major premise of his diatribe is that La and Ac should remain in the d-block because that is where IUPAC places them in its official periodic table and therefore all rational discussion of other possibilities is strictly forbidden." You've made a better case than Lavelle by bringing in other properties as well, but (1) they are often weak arguments and (2) we still don't know which of your criteria you consider to be the most important and fundamental. Without such an order, how can we distinguish it from a scattergun firing of arguments? Respectfully, I think Jensen's strategy (which I have pretty much adopted) of first looking at valence subshells (either with electrons or vacancies), then confirming our assignment with periodic trends, comes the closest yet to the Holy Grail of finding one fundamental principle behind periodic table drawing. Yours doesn't seem to be anywhere near that. Double sharp (talk) 21:37, 9 February 2020 (UTC)
The La and Lu question is better viewed as follows:
In terms of Scerri's continuum of periodic tables the three versions appear broadly as follows:
Towards the Platonic end
  • Tetrahedral ADOMAH in a cube
  • LSPT
  • Lu table
Somewhere near the middle
  • La table
Towards the chemistry end
The continuum puts things into perspective; each form has its purpose. Each one is worth exploring. Frex, the LST prompted surprising research on the He over B question. In this spirit, I say give people a toolbox, not a hammer. Sandbh (talk) 06:10, 9 February 2020 (UTC)
@Sandbh: I say that whatever minuscule advantage the La form has does not overthrow the decisive advantages of the Lu one for harmony of blocks and chemically active subshells that control them. The La form is an unsatisfactory compromise between showing no secondary effects (He-Be, Sc-Y-Lu) and getting drowned in them (Rayner-Canham), with one particular secondary effect (delayed collapses, and only for La/Ac, because the Lr one gets ignored) blown out of proportion above all others. (Maybe because it is the only one that can easily be drawn.) Teach students the Sc-Y-La trend alongside the Sc-Y-Lu trend, for sure; it makes sense for comparative chemistry. And do the same for Be-Mg-Ca vs. Be-Mg-Zn, which is a real ambiguity in electronic structure. But there is no advantage significant enough to overthrow the bald facts that 4f is an active, valence subshell in La, but is stuck in the core for Lu just as much as it is for the rest of the 6th period. Double sharp (talk) 21:37, 9 February 2020 (UTC)

DS' current reassessment of the old IUPAC submission

Here are my revised views of the arguments from our old submission, now that I learnt more in 2018 from Droog Andrey about the issue and about Lu's total lack of 4f involvement:

Lu arguments:

Argument Summary Type Analysis Status
1958: Electron configurations 4f shell complete in Lu Commonality Strong, since 4f is core-like in Lu. 3d is not a valid comparison as it still contributes to bonding in Zn. Substantiated
Core argument
1965: Similarity of Lu with Sc, Y Lu is closer to Sc, Y than is La Commonality True, but not the full story. This does not address relevance of similarity v. linearity and requires more fundamental considerations to deal with. Since there is no big discontinuity from groups 1 to 6, just a continuous easing into more and more transition-like behaviour we instead note that group 3 appears to be a standard d-block group, in which case the group 4+ trend ought to be followed, and this becomes correct. This is something that affects many of the other arguments: Sc-Y-La follows a group 1 and 2-like trend, Sc-Y-Lu a group 4-to-8-like trend. The reason why the latter is preferred is for the increased d-block homogeneity. Substantiated (though incomplete)
Core argument (once completed)
1967: f-character of La Literature suggests this could explain some properties Commonality Indeed, significant for La and nonexistent for Lu Substantiated
Core argument
1982: Separation groups Y, less so Sc, chemically closer to Lu than to La Commonality True, but not the full story. We have to address the continuity of the group 2 to 4 trend to show that this is the expected periodic trend that follows group 4. Substantiated (though incomplete)
Irregular electron configurations Gas phase electron configs support Lu under Y Commonality Gas phase configurations are usually chemically irrelevant Irrelevant
Electron configuration analogy Consistent atomic number difference of 32 between heavy d block metals Trend similarity True; breaking this results in an inconsistency of cores and a breaking of the post-contraction pattern Substantiated
Core argument
Ionization potentials IE 1 + IE 2 for Sc-Y-Lu similar to groups 4–8, unlike Sc-Y-La Trend similarity Correct, but why first two IE? Inconclusive
Atomic radii Periodic trends favour Sc-Y-Lu Trend similarity Correct. But strictly speaking this comes after the core arguments, when we confirm that the periodic trends it gives make sense. Substantiated
Ionic radii Trends favour Sc-Y-Lu Trend similarity No data nor references given, but easy to supply by looking at the d-block trend and how it becomes more homogeneous this way. But strictly speaking this comes after the core arguments, when we confirm that the periodic trends it gives make sense. Substantiated
Redox potentials Comparison of periodic trends favour Sc-Y-Lu Trend similarity Ditto Substantiated
Electronegativities Allred-Rochow trends favour Sc-Y-Lu Trend similarity Pauling EN favours Sc-Y-La. Corrected "chemical" electronegativty (Kulsha-Kolevich) favours Sc-Y-Lu as more consistent with d-block trends. But strictly speaking this comes after the core arguments, when we confirm that the periodic trends it gives make sense. Inconclusive, though somewhat suggestive
Melting points Sc-Y-Lu trend fits better with groups 4–10 Trend similarity Correct. But strictly speaking this comes after the core arguments, when we confirm that the periodic trends it gives make sense. Substantiated
Crystal structures (elements) Sc, Y, Lu are HCP but La is double-HCP Commonality Structures in other d-block groups vary, too Meaningless
Crystal structures (oxides etc) M2O3 etc for Sc, Y, Lu are = whereas La different Commonality Structures in other d-block groups vary, too Meaningless
Excited state spectra Spectra for Sc, Y, Lu differ from La Commonality Empty d-bands in Ca, Sr and Ba precede d block proper and differ from spectra for Be, Mg. But does illustrate the point that 4f is a valence vacancy for La, so has some chemical relevance. Be-Mg-Zn is a true ambiguity that has to be resolved at the 2nd order by trends for periodic table placement, and the Ca group should be considered in comparative chemistry discussing the TMs. Reflection of a stronger argument
Superconductivity Sc, Y, Lu incapable of superconductivity in bulk at normal pressure whereas La is Commonality Lu may or may not also be capable, but most recent work suggests that it is not for reasons related to band structure (chemically relevant) Inconclusive but suggestive
Conduction band structures Sc, Y, Lu have conductivity bands with d block like structure but La does not Commonality X-ray isochromats of Gd to Lu all show similar structures. But it does offer an illustration of Lu displaying more d-elementishness than La. Weak but suggestive
1985: Seven properties Density triads, and some mechanical and thermodynamic properties support Lu under Y Periodic trends/Commonality Density triads OK, but only because we have already established where the d-block goes, so only confirmatory. The other properties are inconclusive Inconclusive
2005: Physical properties Sc, Y dissimilar to Ln; as Lu is also an Ln outlier it should be in group 3 Commonality Lu is hardly an outlier, and Y is more or less like an Ln Groundless
2006: Characteristic electron configurations of ions Configurations of ions are more important than those of gas phase atoms Other Incomplete, even if going in the right direction towards all relevant subshells; on the face of it, looking at these alone looks like it should support -La-Ac Argument supports -La-Ac, and is incomplete
2011: Triads Y-Lu-Lr is a triad; Y-La-Ac is not Other Sc-Y-La form triad but not Sc-Y-Lu Inconsequential
2012: Split d block Sc-Y-La-Ac yields symmetry deficit in 32-column version Other In itself this is not a criticism. It only becomes so because a comparison of the two options shows that the symmetry deficit of the La form does not increase explanatory power appreciably, and in fact decreases it in many cases. Incomplete
2013: Carbonyls Periodicity in carbonyl valencies supports Lu under Y Commonality Not representative; breaks down in s block Unsubstantiated
2014: Lu as a TM Lu shares many properties and similarities with period 6 TMs Commonality Correct; Lu is more d-elementish than La, and makes the d-block more homogeneous Substantiated
Name matches position La, Ac in f block is consistent with names Commonality This is not about chemistry Baseless and irrelevant
2015: Dimer spectroscopy Spectra of Lu M2 dimers is more like those of other period 6 transition metals, so Lu is better fit under Y than is case for La. Commonality Correct, but this is not a form the Ln are usually in chemically, so relevance is weak Weak but suggestive
Relativistic contraction of 6s shell Lu falls on period 6 TM trend line and off the lanthanide trend line Commonality Pt and Au fall off the period 6 TM trend line; the other lanthanides with a d electron (La, Ce, Gd) also fall off the lanthanide trend line; but it is true that Lu falls off far more, and fits better with the period 6 TM trend line than even the "corrected" one-d-electron trendline leading on from La, Ce, and Gd Weak but suggestive
Aluminide dimers AlLa dimer different bond compared with AlSc, AlY, AlLu Commonality Bonds of other groups' compounds vary, too Meaningless, incomplete
Heat of vaporization Sc-Y-Lu trend fits better with early d block than does Sc-Y-La Trend similarity True, though this is just one property among many others and might well be combined with them Substantiated
IE of Lr Experimental determination of Lr IE has no implication Other Agree with Jensen Meaningless

La arguments:

Argument Summary Type Analysis Status
1970: Six properties Basicity increase down a group is a general principle Trend similarity La is more basic than Y; Lu is less basic than Y. As far as it goes this is true, but the contraction before 4f is especially strong. Contractions usually weaken basicity (note Al vs. Ga) for the elements coming after them and there is no reason why this should not be so here. Since we have previously determined the 4f block through chemical activity of the f-subshell, the violation of this principle is justified. Incomplete
1974: Integrity of the f block Double periodicity in the f block is apparent Periodicity True, but misleading (4f is active at La but not at Lu), and incomplete since based on gas-phase configurations Irrelevant
2004: La-Ac remains the most popular form Popularity of La-Ac calls into question the merits of Lu-Lr Popularity True Inconclusive, given lack of critical examination in literature
2008: A pair out of place Lu-Lr would mean La-Ac become only case of a vertical pair with no outer electrons in common with their block Commonality True, but based on gas-phase configurations and hence not relevant. La-Ac makes Lu-Lr the only case of a vertical pair in a block where their characteristic electrons are core electrons, which is even worse since the table is supposed to be based on valence electrons. Irrelevant
2016: Condensed-phase v gas-phase electron configurations Condensed phase configurations are more pertinent to the chemistry of the elements, and support Sc-Y-La-Ac Commonality True, but their invocation means playing with fire for Be-Mg-Zn. Also incomplete: you want to look at all chemically relevant configurations, not just condensed-phase only or gas-phase only. Incomplete
Blocks Concept of periodic table blocks supports Sc-Y-La-Ac Commonality False, because a block's members should have its characteristic electrons as valence electrons or at least valence vacancies. This is true for La-Ac in the f-block but not for Lu-Lr in the f-block. Baseless
Chemical behaviour Group 3 chemical behaviour is like groups 1–2; Sc-Y-La trends fit better with groups 1–2; Sc-Y-Lu trends fit better with groups 4–10; ergo Sc-Y-La preferred Trend similarity, commonality Group 3 chemical behaviour is in actuality inconclusive between group 2 and 4; it is quite intermediate if you look at all aspects of chemistry. Therefore this argument is not strong enough to overturn the delineation of blocks by 4f involvement of La. Groundless

Double sharp (talk) 22:53, 8 February 2020 (UTC)

(And so we pass 600K!) Double sharp (talk) 00:18, 9 February 2020 (UTC)

Sandbh's criteria

@Double sharp: Earlier, you posted:

"That's why I want to ask: please set down your criteria, a precedence order for them, and reasons why each one is important. That's the only way you can get a consistent basis for the PT."

It looks something like this, with # denoting an IUPAC argument.

Important arguments/criteria

  • D/e’s are important, per giving shape to the blocks
    • And since this is where I start disagreeing with you, having quoted many papers in support of DE's being generally chemically irrelevant, it is not a surprise that my disagreement with your logic runs to almost the whole of it. Double sharp (talk) 11:27, 10 February 2020 (UTC)
-- Our criterion that a block starts when the first electron of its name enters the applicable sub-shell #
-- One less d/e anomaly in an La table
-- The Ln contraction proper starts in Ce and ends in Lu
  • The periodic law and the actual afbau process
  • The ionic v covalent distinction, per Rayner-Canham.
-- Ionic radii trends going down Sc-Y-La and Sc-Y-Lu compared to groups 1 and 2, and 4+ #
  • Internal regularity

Supporting arguments/criteria

  • One more 234 triad in an La table (regularity)
  • Isodiagonality seen in Ca-Y-Ce but not Ca-Y-Lu (importance per Rayner-Canham)
  • Laing's comparing the pairs Ca-Sc and Sr-Y with Ba-La # (DIM extrapolation)
  • Restrepo's stoichometry findings (importance re connection with chemistry)
  • Monocations of Sc, Y, La and Lu (noting many correlations made to 1st IE)
  • Rare earth metals (consistency, regularity)

Sandbh (talk) 05:48, 10 February 2020 (UTC)

The metallic elements

@Double sharp:

These extracts are from: Parish RV 1977, The metallic elements, Longman, London

I was looking through Parish to see if I could get a better appreciation of the homogeneity issue i.e. are the vertical trends down group 3 more relevant than the horizontal trends along the 5d row, etc. No real joy there.

And I'm not surprised, given his considering 4d and 5d metals are something different from the 3d metals, rather than going group-by-group like Greenwood and Earnshaw. Double sharp (talk) 12:07, 10 February 2020 (UTC)

I did discover a treasure trove instead.

"The chemistry of the s-block metals is the simplest of any group of elements in the Periodic Table, because each element displays only one oxidation state: +1 for the alkali metals (lithium to caesium), and +2 for beryllium, magnesium and the alkaline earth metals. A great deal of their chemistry is explicable in terms of simple ionic bonding and the number of covalent compounds is small, lithium, magnesium and, particularly, beryllium providing the most examples." (p. 34)

Note use of the expression, "A great deal of their chemistry…etc".

"…of the 3d -metals, scandium, (along with yttrium and lanthanum) bears a much closer resemblance to the 4f-metals and all these metals are more conveniently treated with the latter in Chapter 6." (p. 48)

"The metals of the 4d- and 5d-series form a strong contrast with those of the 3d-series. The metals themselves are generally very hard, high-melting and unreactive, and are often used for these properties. In their compounds a wide range of oxidation states is shown, but all are characterised by the formation of bonds of high covalent character, and there are no simple ionic compounds. Only in a few cases are even the simple aquated cations known, other ligands, often anions, being bound in preference to water. Within each Group, the 4d — and 5d—metals are often closely similar, especially in the earlier Groups, and this gives rise to difficulties in separation and identification. It is, for example, not easy to obtain samples of hafnium or zirconium uncontaminated by the other. These similarities are the result of the ‘lanthanide contraction’. In Group III, yttrium and lanthanum show the differences in their chemistry which would be anticipated from the difference in ionic radii, ionisation energies. etc., but the interpolation of the fourteen lanthanide elements between lanthanum and zirconium, with the resultant increase in effective nuclear charge, reduces the radii and increases the ionisation energies to values close to those for hafnium. These effects are seen to build up through the 4f series…" (p. 112)

Note reference to Y and La showing the expected trends, and the reference to the interpolation of the Ln.

This is circular if you apply it to the group 3 question. By saying that the Ln are interpolated, you are implicitly assuming that they come between La and Hf. Which means that this is useless for actually determining whether the 4f block should start at La or Ce. Well, if you think it starts at La, then the trend at Lu is totally expected because the 4f series is over. Well, if you think it starts at Ce, then the trends in group 3 and 4 are totally expected because La comes before the 4f series, but Hf comes after it. Either way, it is perfectly rationalisable. So you have to appeal to some other factors, hopefully chemically relevant ones (whence chemically active subshells and certainly not those tiresomely irrelevant DE's), that determine the start of a block. And this is where we part, because you insist on chemically irrelevant DE's (which you then massage away to get rid of the annoying problem that according to them, 4f begins at Ce but 5f begins at Pa), and I insist on chemically relevant active valence subshells (which you then misread all the time by dragging in the canard about the core 4f subshells at Lu). After this fundamental difference there is not much common ground to be had. Double sharp (talk) 12:01, 10 February 2020 (UTC)

"The f-block consists of two series of elements in which the 4f- and 5f-orbitals are filled. However, the chemistry and properties of these two series are so very different that, as with the d-block, it is more convenient to treat them separately. The 4f-elements are therefore considered in this Chapter, and the 5f-elements in the next. The 4f elements (alias lanthanons, lanthanides or lanthanoids) have a very simple chemistry since with but few exceptions only one oxidation state is displayed, viz. +3. In this respect they resemble the s-block elements, and indeed there are pronounced similarities to calcium, which often occurs in lanthanide minerals, and to strontium and barium, particularly in the complicated solid—state structures found." (p. 142)

Note reference to resemblance to s-block elements and pronounced similarities to calcium, and to Sr and Ba.

For minerals the important broad-strokes classification is Goldschmidt's. In which, indeed, the Ln and An occur as lithophiles, just like groups 1 and 2. Unfortunately for the prospects of this as a La argument, so do the elements of groups 3, 4, and 5. Double sharp (talk) 12:02, 10 February 2020 (UTC)

"Two features are particularly striking in the chemistry of the 4f—elements, Viz. (a) the uniformity of the +3 oxidation state and the small number of other oxidation states, and (b) the irregularity of structures and occurrence of high coordination numbers. In the compounds described here, coordination numbers of six, seven, eight, and nine have been mentioned, and in other compounds ten- and even twelve-coordinate metal ions are found, eg. in La2(SO4)3.9H20 and (NH4)2Ce(NO3)6. It is tempting to think that the f-orbitals must be involved in the bonding, since the maximum number of hybrid orbitals which can be constructed from an s-p-d -set is nine, and to obtain eight bonding orbitals directed towards the corners of a cube (as in C602) requires at least one f-orbital. There is, however, no evidence to suggest that the f-orbitals are involved at all in the bonding, and even ligand-field effects are extremely small. All the compounds appear to be essentially ionic, with very little covalency involving even the 6s- or 6p-orbitals. The curious structures and high coordination numbers are similar to those found with other large cations (e.g. salts of Ba2+ or Pb2+) and are presumably a result of the optimisation of electrostatic forces. The large internuclear distances necessitated by the radii of the cations will cause the electrostatic energy per pair of ions to be relatively small, despite the high cationic charge, and many such pairs must be formed to achieve a sufliciently large lattice energy. Similarly, the large radii diminish considerably the polarising effect of the +3 charge which would otherwise be expected to lead to considerable covalency." (p. 151)

Note no evidence for f orbital involvement, including in La. Sandbh (talk) 11:25, 10 February 2020 (UTC)

Despite my strong scepticism about this whole idea (especially given the 1977 date of the book), I'll bite: if there is no 4f involvement in any lanthanide, then there isn't any in Lu either, and so we are back to square one. The whole point is that 4f is a reserve area for electron; it prevails in indirectly bonding MO's, but electrons may appear in it in chemically relevant configurations and jump out from it to involve themselves in the bonding. Again, this is something that happens for La and not Lu. Double sharp (talk) 12:04, 10 February 2020 (UTC)

More about the transition metals: a case study for why DS chose to adopt Jensen's criteria

This will probably get called fogging, but I will put it up anyway because it well illustrates my policy: before elevating something to a principle of the periodic table, we work backwards from it and see what more fundamental stuff we can derive it from to get to the bottom of it. (Which is why my approach accepts only chemically active valence subshells confirmed by periodic trends.) We start to understand more clearly why the TMs in later and later periods start showing characteristic transition properties more and more sluggishly if we step backwards and look at the more fundamental reasons why this is happening. (Which makes sense, because we want to go to the most fundamental reason possible for the periodic table.) This surely beats the alternative of fogging (to use Sandbh's word) about how the gap is supposedly between groups 3 and 4 all the time, when it is plain as day that by slightly shifting our focus it can move as far as the line before group 2, or the line after group 7 even. (Yes, seriously. Ca-Sr-Ba have significant pre-transition character. But Tc-Re-Bh have +7 as the most major oxidation state, although Tc admittedly likes interconverting between them a lot; we may draw parallels with what happens with a lot of p-block main-group elements, with one characteristic group oxidation state, but another common lower one!)

Generally speaking, if something happens in one element of the 3d transition series, it will only happen one or two elements later in 4d and 5d (e.g. the melting-point trend: cf. the positions of Cr vs. Tc/Re, which Greenwood and Earnshaw note IIRC). This is, in fact, a diagonal relationship (note compensations for charge/radius, per Fajans' rules); it becomes a straight-down relationship from 4d to 5d because of the 4f contraction, of course. (From La to Yb, naturally. ^_^)

The end result is: in the 3d row, titanium is the first one with bona fide low oxidation states in water. (They are still readily oxidised to the +4 state and are not the predominant ones, of course; for more stability and dominance you really need to wait till vanadium with its +4 state.) But in the 4d and 5d rows: lower oxidation states of zirconium and hafnium reduce water outright. For niobium and tantalum, the lower oxidation state is readily oxidised, and the group state dominates. The "crooked diagonals" here are Sc to Zr-Hf; Ti to Nb-Ta; and V to Mo-W (this set is the one with stronger lower oxidation states, although for tungsten this is not so stable.)

If we dare to go to the 6d metals, we get a double whammy. Firstly, the 5f contraction's action is not so strong as the 4f one, and so we still have some increase in atomic radius. But secondly and more importantly, 6d is relativistically strongly destabilised. The end result is absolutely predictable: even at seaborgium, Sg(IV) is more unstable than W(IV) and should be readily oxidised to Sg(VI). The first time we get a lower oxidation state well-defined enough to exist in water and display acidic/amphoteric/basic properties is probably for bohrium (+6), which should be readily oxidised to the group state of Bh(VII). Hassium is the first one where the group oxidation state is not among the most stable ones! (As we should have guessed, from the dominance of +7 for technetium and rhenium, even though lower oxidation states are well-defined there.)

So, now that we understand what is going on, we pencil in: "due to the increased activity of d-subshells relative to s-subshells down the table going from 3d to 4d (avoiding primogenic repulsion) and 5d to 6d (relativistic destabilisation), main-group character persists for longer and longer at the start of each transition series, but incipient transition behaviour still begins at about the same time". And now we understand what is going on, instead of endlessly talking about a dichotomy, raising it to a principle of the periodic table, when in fact it is the job of the principles behind the periodic table to explain that dichotomy. And, lo and behold, we have returned to the true fundamental basis: chemically active valence subshells. And the number of electrons in them in total, which gives group assignments and controls valency. Notice, of course, that the little surface fluctuations of ground-state DE's never came into it. In fact, I did not need to reference DE's even once here to explain this. By Occam's razor, they fly away.

While looking up stuff for this comment, which will no doubt end up being referred to as fogging anyway, I came across my descriptive chemistry of the d-block from archive 27, and I found this extremely revealing comment: 'Suppose I talk to you about an element M which is usually tetravalent. Now I tell you that MCl4 is a volatile molecular liquid that hydrolyses readily and completely to MO2 even when in contact with moist air; that oxoacid salts of MIV do not exist, only hydrolysed species being formed; that supposed "MO2+" ions sometimes alluded to in the literature are really polymeric –M–O–M–O– species; and that it has no simple cationic chemistry. Looking at how you are interpreting the literature to make the case for the metalloids as nonmetals, I am sure you would say that M should be classified as a nonmetal, although perhaps a weak nonmetal, analogous to silicon and germanium. Except that M is, of course, titanium.' Mendeleev was right: valence is very important. And it controls ionic vs covalent and all such dichotomies so much that it even transcends distinctions of elements that are metallic vs. nonmetallic as simple substances. The "breakpoint" simply occurs at a particular z/r ratio per Fajans. But that simply shows that "ionic vs covalent" is not the basic dichotomy to use here, as it is a consequence of valence and Fajans' rules. Instead we must look back at the recurring pattern of valences and atomic radii. And then look at what causes it, which yields chemically active subshells and electron configurations in all chemically reasonable environments. Only with such a regression are we ever going to get to something like a sound basis for the periodic table. Everything else can be traced back to that and is secondary. That's why looking at periodic trends is only a confirmation that we have not been barking up the wrong tree.

TL;DR: go back to the most basic relevant phenomenon you can think of for chemistry and use it as the basis for drawing your PT. That's why we go to chemically active subshells, with control everything that happens next, incl. ionic vs covalent, metallishness, valences, diagonal relationships, contractions, trends, etc.: everything should be as simple as possible, so we go back to the first basic relevant cause. But not further to the chemically irrelevant ground states: simple as possible, but not simpler. Double sharp (talk) 17:46, 10 February 2020 (UTC)

Jensen's criteria, recapitulated:

  1. Assignment of the element to a major block based on the kinds of available valence electrons and/or valence vacancies (i.e., s, p, d, f, etc.).
  2. Assignment of the element within a given block to a particular group based on the total number of available valence electrons.
  3. Verification of the validity of the resulting block and group assignments through the establishment of consistent patterns in overall block, group and period property trends.
  4. Verification that the elements are arranged in order of increasing atomic number as required by the periodic law.

Frankly, the only place I differ with him on these criteria is where he says here that criteria 1 and 2 alone do not resolve the group 3 question. As he himself points out here, "[La and Ac] have low-lying empty f orbitals, which is more than can be said for Lu and Lr". The debate should never even have gotten past step 1. It's only for the problem of Be and Mg that there is a serious issue that requires going down to step 2, noting that the d-electrons of group 12 at least contribute meaningfully to bonding. (Yes, I disagree with his stance on group 12, but that doesn't forbid me from agreeing with him on group 3.) Double sharp (