Henry Rzepa's blog

I have written a few times about the so-called “anomeric effect“, which relates to stereoelectronic interactions in molecules such as sugars bearing a tetrahedral carbon atom with at least two oxygen substituents. The effect can be detected when the two C-O bond lengths in such molecules are inspected, most obviously when one of these bonds has a very different length from the other. The effect originates when one of the lone pair of electrons on one oxygen atom uniquely overlaps with the C-O antibonding σ^* on another oxygen, thus shortening the length of the donating oxygen-carbon length and lengthening the length of accepting C-O bond. Here I take a look at tetra-substituted versions of this (C(OR)₄), where in theory there are up to eight lone pairs, interacting with any of three C-O bonds, giving a total of 24 possible anomeric effects in one molecule.

We start the process with a search of the Cambridge crystal structure database, using the following search query:

This yields 25 hits. We now want to find out what the longest and shortest C-O bonds are, and how large the difference between them is. To do this, we have to resort to applying some functions, using the calculator tool built into the Mercury analysis software. The following functions were used:

1. Greatest('search3'.'DIST1','search3'.'DIST2','search3'.'DIST3','search3'.'DIST4')
2. Least('search3'.'DIST1','search3'.'DIST2','search3'.'DIST3','search3'.'DIST4')
3. Greatest('search3'.'DIST1', 'search3'.'DIST2', 'search3'.'DIST3', 'search3'.'DIST4')-Least('search3'.'DIST1', 'search3'.'DIST2', 'search3'.'DIST3', 'search3'.'DIST4')
   
   

The results can be displayed as below, in which the difference between the two bond lengths is colour coded (red = greatest, blue = least).

1. Here you can see that when the difference between the longest and short C-O bond lengths is small, the colour is blue.
2. Green dots show a difference of about 0.04-0.05Å
3. The red dot has the greatest difference of 0.087Å and corresponds to the entry SILDOH ([1], DataDOI: [2], 10.5517/ccq8lq8.

The next step is to apply a “reality check” using computation, here a MN15L/Def2-TZVPP calculation on the top eight entries as sorted by the largest C-O bond length differences (Δr_C-O > 0.05Å.[3], data DOI: 10.14469/hpc/13925

1. The largest effect occurs for SILDOH, and this is replicated by calculation.
2. The largest discrepancy between measurement and calculation is for KEVFUM,  where calculation predicts almost no C-O bond differences. This will be discussed elsewhere.

Focusing on SILDOH, we look at the NBO E(2) energies for the donor-acceptor interactions of an oxygen lone pair donating into a C-O antibonding σ^* orbital.

Click on the image below for a 3D model of the two interacting orbitals (positive overlap = blue + purple, red + orange)

The interaction of LpO1 to the long bond C5-O4 = 18.0 and LpO2 to C5-O4 = 16.3 kcal/mol, whereas in the reverse directions, LpO4 to C5-O1 is only 6.0 kcal/mol and LpO4 to C5-O2 is 10.7 kcal/mol.  For a “normal” C-O bond however such as  C5-O3,  LpO2 to C5-O3 = 3.1 and LPO1 to C5-O3 = 5.3 kcal/mol. In effect, two oxygens “gang up” on weakening the  long C5-O4 bond, but leave the shorter C5-O3 bond alone. So the individual anomeric effects are no larger than normal, but the cooperative effect of two acting together is what produces the final geometric asymmetry.

The Wiberg bond index mirrors this effect. The bond indices are 0.9882 for O1-C5 and C5-O4 0.8512 (Δ =-0.137) which is a big difference in bond order and accounting for the large (record?) difference in bond length.

In the next post, I will analyse the equivalent molecules B(OR)₄^–.

References

1. R. Betz, and P. Klüfers, "Norbornane-2,7-diyl 1′,2′-phenylene orthocarbonate", Acta Crystallographica Section E Structure Reports Online, vol. 63, pp. o3933-o3933, 2007. https://doi.org/10.1107/s1600536807042298
2. Betz, R.., and Klufers, P.., "CCDC 663670: Experimental Crystal Structure Determination", 2007. https://doi.org/10.5517/ccq8lq8
3. H. Rzepa, "Detecting anomeric effects in tetrahedral carbon bearing four oxygen substituents.", 2024. https://doi.org/10.14469/hpc/13925

First, a very brief history of scholarly publishing, starting in 1665[1] when scientific journals started to be published by learned societies. This model continued until the 1950s, when commercial publishers such as Pergamon Press started with their USP (unique selling point) of rapid time to publication of ~3 months,[2] compared to typical times for many learned society publishers of 2 years or longer. Fast forward another 50 years or so, and the commercial publishers were now dominating the scene, but the business model was still based on institutional subscriptions, whereby the institution rather than authors paid the costs of publication. As the number of journals expanded, even well-off institutions had to make difficult decisions on which subscriptions to keep and which to cancel. By the late 1990s the delivery model was changing from print to online, but the overall issue was that many scientists around the world no longer had access to many journals.

Enter the APC, or article processing charge, whereby the authors themselves had to reimburse the journals for publishing their papers, although they could often still recover these costs from their institution. The cost of an APC depended on the reputation of the journal; those with the highest “impact factors” often charged the highest APCs, some of which could reach £5000+ for a single “paper” (still called that even in an electronic era). Also, some journals remained “hybrid”, where the costs were split between institutional subscriptions and APC funded. At least the latter could be accessed by anyone (including the “public”) without restriction (Open-Access) often also referred to as GOLD  and even Diamond (also known as platinum) articles which  are  GOLD open access but without author fees. Diamond is typically used by publishers who are keen to emphasise that they do not charge authors to publish open access.

With many APCs ranging from £1000 up to £5000 or more, some started asking why it should cost so much to have this type of publishing infrastructure. Also in the early 2000s, “social media” started up, which at first tended to concentrate on instant publication and hence impact. The longevity of these media was not considered capable or indeed even desirable of rivalling that achieved by journal publishers, which after all had been around for 360 years or so. Things have begun to change however. Enter as an example Rogue Scholar, and its associated blog Front Matter. The aim here is to exploit the underpinning technical infrastructure of a blog host by automatically adding features more commonly associated with learned society or commercial journal publishing.

I wrote[3] about some of the features available last September and now only four months later the functionality continues to expand. This includes:

1. The ability to acquire a JATS XML version (Journal article tag suite), the standard format for scholarly articles
2. I had previously noted that Blog posts are assigned a DOI based on the Crossref registration agency, and hence also acquire a metadata record which becomes useful for searching. All 800+ of the posts on this site have such a DOI for example.^‡
3. One interesting recent use of blogs is to act as a science newsletter associated with a funded grant, as an adjunct to simply publishing the research results in a journal.
4. Indexing is also making big strides with the introduction of an API (application programmer interface), another service offered by scholarly publishers. As part of this, fields of science are being added to the metadata to enable filtering such as eg Chemistry
5. Archiving, in theory for all of posterity, is also starting to be addressed . This requires transformation from HTML, typically used in blogs, to a medium more appropriate for long term archiving.

The cost of the infrastructures described above are certainly very much less than eg the APC charges noted above, in part because they are so highly automated. I expect things will move very rapidly on this front.

^‡It is hoped to automatically include these in the post itself in the future. Meanwhile, it can easily be retrieved by a suitable search.

References

1. H. Oldenburg, "Epistle dedicatory", Philosophical Transactions of the Royal Society of London, vol. 1, pp. i-ii, 1665. https://doi.org/10.1098/rstl.1665.0001
2. D. Ginsburg, and W.J. Rosenfelder, "Alicyclic studies—X", Tetrahedron, vol. 1, pp. 3-8, 1957. https://doi.org/10.1016/0040-4020(57)85003-0
3. H. Rzepa, "Improving the Science blog – The Rogue Scholar service.", 2023. https://doi.org/10.59350/8m2d8-47b52

In an earlier post on this topic,[1]^‡ I described how the curly-arrows describing the mechanism of a nucleophilic addition at a carbonyl group choreograph in two distinct ways, as seen in red or blue below. The arrows in red can be described as firstly addition to the carbonyl group to form either a transient intermediate (a two-step process) or instead a formal transition state state as a concerted single-step mechanism. The blue arrows do the reverse; firstly elimination and then followed by addition. I will use the shorthand AE for the first type and EA for the second type. Here I explore some more nucleophiles to see which of these two mechanisms they follow. Data for these results can be found at 10.14469/hpc/13171
N- carbon ylid: This is a very facile (low-barrier) reaction with a C-O bond length response that initially increases steeply, followed by a more modest decline and hence corresponds to an AE mechanism.

P carbon-Ylid:  Essentially identical to the previous example, and again an AE mechanism.

S carbon-ylid: Again, an AE mechanism.

S-nucleophile:  This one is different, showing a larger barrier and initial small decrease in the C-O length followed by a larger increase. This one is an EA mechanism.

As I noted previously, it would be useful to have two double headed curly arrows available in palletes of these; <—> (AE) and >—< (EA) to illustrate the difference between the two mechanistic types.

^‡This is the first instance where I cite a blog using a CrossRef DOI generated for it. Previous such citations used a DataCite DOI, which the bibliographic software used here to add them to the post (Kcite) does not support.

References

1. H. Rzepa, "The "double-headed" curly arrow as used in mechanistic representations.", 2023. https://doi.org/10.59350/f00wf-5tq46

What’s a Journal For? This debate has been raging ever since preprint servers were introduced as far back as 1991! Indeed, during my recent submission of a journal article, one of the questions asked was whether the article was already deposited in such a preprint server (in a positive sense, and not one excluding further submission progress). Since my previous comment on this theme was made more than three years ago, I thought I might update it.

I might start with the observation that some think the concept of a journal really comprises three separate components (up to eight have been suggested); the story or narrative being told, the data on which that story is based and the citations or bibliography which set the context of the story. The latter two components have both developed their own publishing models; the data in a repository and accompanied by rich metadata which provides at least some of the context and citations which have their own model. Article metadata also includes its own citations helping to place the data into a wider context or “bigger picture” as expressed by a knowledge graph,[1] which even CAS Scifinder will now reveal based on your specific search!^‡. Such metadata is also now generally a component of the overall metadata associated with journal articles. The data component is being accompanied by extensive work to enhance the accompanying metadata models[2] and we might expect rapid progress to be made here in the near future.

So again to ask “what’s a journal for” if two of its essential components have their own publishing models? The story will always have an important role to play and peer review of that story^† will always be an important aspect of the journal – indeed perhaps the most important aspect. So should we focus in our attention on this?  I noted that in 2017, a brave new experiment claiming “Open access • Publication charge free • Public peer review • Wikipedia-integrated” of which public peer review was an important component, has accumulated relatively few publications since. I also noted an article[3] in which the reviewers (but not their reviews) are clearly indicated. This concept too has not made much headway. Will things change in the future? Some think that they have too, or the entire concept of scientific publishing will indeed fragment into many different models and no longer fully serve its purpose.

^‡I cannot resist including my own knowledge graph here, which reveals nicely the impacts of some of the work I have been associated with, as represented by the fans on the outside of the central graph.

^†Although a major component of many peer reviews has the focus on the data and (missing) citations.

References

1. H. Cousijn, R. Braukmann, M. Fenner, C. Ferguson, R. van Horik, R. Lammey, A. Meadows, and S. Lambert, "Connected Research: The Potential of the PID Graph", Patterns, vol. 2, pp. 100180, 2021. https://doi.org/10.1016/j.patter.2020.100180
2. R.M. Hanson, D. Jeannerat, M. Archibald, I.J. Bruno, S.J. Chalk, A.N. Davies, R.J. Lancashire, J. Lang, and H.S. Rzepa, "IUPAC specification for the FAIR management of spectroscopic data in chemistry (IUPAC FAIRSpec) – guiding principles", Pure and Applied Chemistry, vol. 94, pp. 623-636, 2022. https://doi.org/10.1515/pac-2021-2009
3. L. Li, M. Lei, Y. Xie, H.F. Schaefer, B. Chen, and R. Hoffmann, "Stabilizing a different cyclooctatetraene stereoisomer", Proceedings of the National Academy of Sciences, vol. 114, pp. 9803-9808, 2017. https://doi.org/10.1073/pnas.1709586114

Four-coordinate carbon normally adopts a tetrahedral shape, where the four angles at the carbon are all 109.47°. But how large can that angle get, and can it even get to be 180°?

A search of the CSD (crystal structure database) reveals a spiropentane as having the largest such angle, VAJHAP with 164°[1]

Because crystal structures might have artefacts such as disorder etc, it is always good to check this with a calculation; hence ωB97XD/Def2-TZVPP (FAIR data DOI: 10.14469/hpc/11148) for which a calculated angle of 163.8° is reassuring. The smallest angle in this system by the way is 58°, pretty normal for three-membered rings.

The localised orbitals show the C-C region defining the large angle to be very “bent” (a banana bond) but otherwise fairly normal.

So can one “engineer” an even larger angle? Replacing the C=C of the benzo group with a shorter CH₂ group produces the following, which is now almost linear (almost a “hemispherical” carbon).

What about the smallest angle at 4-coordinate carbon? Could it be significantly smaller than the 57° noted above for a three membered ring? Searching the CSD reveals XAQHIH[2] with an angle of 43° but the calculation above now does not confirm this, the angle changing from 43° to 59° during optimisation. A reminder that when exploring extreme geometric values, always check with a calculation!

The next candidate is CAZFUE[3] with an apparent measured angle of 48°. This appears to have C₂ symmetry, and a calculation with this gives a value of 46.6°. But all is not what it seems. This is a classic example of a semibullvalene [3,3] Cope rearrangement, caught in the “middle” so to speak (See this post here). In fact this geometry is actually a transition state, and the crystal structure is the thermal average of two positions, making it appear symmetrical. The ground state for this structure as calculated is different! The two angles now emerge as 40 and 57° (average 48.6°). At the “transition state”, one of the four “bonds” to carbon is unusually long (2.07Å), which is the direct cause of the small 48° angle. If this is not allowed as a “bond”, the angle at the other true 4-coordinate carbon emerges as normal at 57°

So the answer to the smallest angle does seem to be around 57°, but it could be as small as 47° if one allows bonds of 2.07Å in one’s definition of 4-coordinate. The candidate for the largest bond angle, of almost 180°, seems a reasonable synthetic target!

References

1. R. Boese, D. Blaeser, K. Gomann, and U.H. Brinker, "Spiropentane as a tensile spring", Journal of the American Chemical Society, vol. 111, pp. 1501-1503, 1989. https://doi.org/10.1021/ja00186a058
2. M.V. Roux, J.Z. Dávalos, P. Jiménez, R. Notario, O. Castaño, J.S. Chickos, W. Hanshaw, H. Zhao, N. Rath, J.F. Liebman, B.S. Farivar, and A. Bashir-Hashemi, "Cubane, Cuneane, and Their Carboxylates:  A Calorimetric, Crystallographic, Calculational, and Conceptual Coinvestigation", The Journal of Organic Chemistry, vol. 70, pp. 5461-5470, 2005. https://doi.org/10.1021/jo050471+
3. H. Quast, Y. Görlach, J. Christ, E. Peters, K. Peters, H.G. von Schnering, L.M. Jackman, A. Ibar, and A.J. Freyer, "Crystal and molecular structure and the cope activation barriers of some dicyano-1,5-dimethylsemibullvalenes", Tetrahedron Letters, vol. 24, pp. 5595-5598, 1983. https://doi.org/10.1016/s0040-4039(00)94150-9