CHOOSE fails to calculate requested structure

CHOOSE fails to calculate requested structure

Postby Tobias Kraemer » Thu May 03, 2018 5:57 am

Dear all,

I have a question which concerns calculating a NBO Lewis structure in a transition metal complex, which seems to cause some problems. I used NBO calculations quite a bit in the past, and I do appreciate the fact that the analysis of transition metal complexes can be challenging at times. However, perhaps here I am facing a methodological misconception, and I'd appreciate your input on this. So I am dealing with a square-planar palladium(II) complex, with Sb-donor, P-donor and Cl ligands coordinated. It seems odd to me that the natural choice in NBO for the Lewis structure involves bonds between Pd and some of the ligands, but lone pairs are assigned to others (say 4 on Cl-). I hence played around with the CHOOSE keyword, in order to guide the localisation in a different way. Naturally, I would introduce 4 bonds between the metal centre and all 4 donor atoms (removing lone pairs accordingly). This however produces the following error message:

CHOOSE failed to calculate requested structure; try DETAIL for details

indicative that the suggested Lewis structure is not appropriate. I then went to a much simpler system, tetrachloridoplatinate(II) and tried my luck here. It turns out that there are similar issues. The canonical NBO Lewis structure (with 99.3% of the total electron density) assigns two Pd-Cl bonds to the structure, plus two Cl with 4 lone pairs each. The bonds are cis to each other. What concerns me is the fact that this obviously breaks the symmetry of the system, since the Pd-Cl bonds have a polarisation of 80% Cl/20% Pd while the other "bonds" are defined through the lone pairs on Cl (86% Cl and 14% Pd). So there are two different types of bonding involved here, which appears odd in a chemical sense. When I try to assign 4 equal bonds between Pd-Cl I get the same error as above, it is obviously not possible to create a Lewis structure that would fit such a description (although it seems intuitive to me to assume such a structure).

Coming back to the original example above, the assignment of different Lewis structures also implies different polarisations of the Pd-Sb bond (the situation seems complicated by the fact that Sb in this case is hypervalent). The default Lewis structure yields a 50%Pd/50%Sb bond, while when I assign Pd-Cl, Pd-Sb and Pd-Sb plus Pd |Sb the polarisation of this respective bond becomes 60% Sb/38% Pd. I should also mention that the total electron density is more than 99% in all these cases. The fact that the polarisation is so different is interesting (and in fact the crux of the whole analysis).

I have performed a Pipek-Mezey localisation as well in order to check the results. It appears that a more polarised Pd-Sb bond is in better agreement with the PM localisation analysis, but I am looking for some vindication for my NBO analysis. I am using the most recent version of the code, NBO6.0.18a (11-Mar-2018).

If anyone could give me some insight into some conceptual details here, I would very much appreciate it.

Thanks in advance.

Tobias Kraemer
Posts: 4
Joined: Wed May 02, 2018 12:12 pm

Re: CHOOSE fails to calculate requested structure

Postby ericg » Thu May 03, 2018 6:47 pm


CHOOSE fails in these cases because it is unable to orthogonalize four bonds to square-planar d8 Pd(II) and Pt(II).

Orthogonalization of these four bonds, together with the four lone pairs of the metal center, requires eight atomic hybrids, which in the tetrachloridoplatinate case should include the 6s, the five 5d's, and two of the 6p's. The 6p AOs are the problem. Natural population analysis shows that only 0.02 electrons occupy the 6p subshell. Because the 6p is essentially unoccupied, calculations of the one-center (LP) and two-center (BD) eigenvectors of the density matrix yields strongly overlapping bonds and lone pairs, exhibiting 6s/5d character only. CHOOSE fails to orthogonalize these functions and terminates with the error message you report.

The Lewis structure that you report for tetrachloridoplatinate is exactly what I get: four Pt lone pairs, two cis Pt-Cl bonds (20% Pt and 80% Cl), and two chlorides. This structure satisfies the 12-electron rule that NBO analysis generally supports for transition metal complexes (corresponding to six hybrids from the valence s and d AOs). Note that the Lewis structure is one of four equivalent resonance structures that you could write for this complex. In fact, this is the representation suggested by natural resonance theory (NRT), essentially 25% weights for each of these four structures, giving Pt-Cl bond orders of 0.5.

Posts: 317
Joined: Sat Dec 29, 2012 9:31 am

Re: CHOOSE fails to calculate requested structure

Postby Tobias Kraemer » Tue May 08, 2018 7:35 am


thanks for the detailed answer and refreshing my mind on this. Yes, this makes total sense. After studying the output more carefully, this started to make much more sense, and I started to remember the nuts and bolts of NBO analysis. I appreciate your detailed answer regarding the orthogonalisation problem, which has helped me to understand this situation even more. I find it always very instructive to hear more of the methodological details, which helps to gain more insight into the chemical system and the program. So, thanks very much again.

I am also glad to hear that my results appear to be reliable and hence can be interpreted. In the meantime I have analysed all these results in full, and it does make sense.

Thanks again.

Kind regards,

Tobias Kraemer
Posts: 4
Joined: Wed May 02, 2018 12:12 pm

Return to General NBO Discussion

Who is online

Users browsing this forum: No registered users and 3 guests