Posts Tagged ‘Reactive intermediates’

What is the (calculated) structure of a norbornyl cation anion-pair in water?

Saturday, April 1st, 2017

In a comment appended to an earlier post, I mused about the magnitude of the force constant relating to the interconversion between a classical and a non-classical structure for the norbornyl cation. Most calculations indicate the force constant for an “isolated” symmetrical cation is +ve, which means it is a true minimum and not a transition state for a [1,2] shift. The latter would have been required if the species equilibrated between two classical carbocations. I then pondered what might happen to both the magnitude and the sign of this force constant if various layers of solvation and eventually a counter-ion were to be applied to the molecule, so that a bridge of sorts between the different states of solid crystals, superacid and aqueous solutions might be built.

I augmented the model in stages. The results are summarised in the table below.

  • Firstly, adding a self-consistent-reaction-field (SCRF) continuum model for water.
  • Then adding to that four explicit water molecules symmetrically arranged around the four C-H groups mostly likely to be solvated via hydrogen bonds.
  • The final model added a chloride anion to complete the ion pair and a further three water molecules to act as its solvation sphere. A search of the Cambridge structure database for any instances of a molecule with a designated C+ and a nucleophilic halide– with zero coordination number (a free halide anion) reveals no hits; such ion-pairs are clearly very unstable towards covalent bond formation, existing if at all only as transient species or when the counter-ion is non-nucleophilic such as R4B–.
Calculated geometries, Def2-TZVPP/SCRF=water

Model

Apical C-C

distance,Å

Basal C-C

distance,Å

ν [1,2]

cm-1

 DataDOI
Vacuum, cation
B3LYP+D3BJ		 1.888 1.388 +140 10.14469/hpc/2410
Vacuum, cation
ωB97XD		 1.830 1.388 +235 10.14469/hpc/2409
Vacuum, cation
B2PLYPD3		 1.872 1.390 +194 10.14469/hpc/2238
SCRF, cation
ωB97XD		 1.819 1.387 +236 10.14469/hpc/2413
SCRF, cation
B2PLYPD3		 1.858 1.388 +202 10.14469/hpc/2243
SCRF+4H2O, cation
B2PLYPD3		 1.838 1.390 +254 10.14469/hpc/2246
SCRF+7H2O+Cl– ion pair
B3LYP+D3BJ		 1.593, 2.485 1.510 – 10.14469/hpc/2408
SCRF+7H2O+Cl– ion pair
ωB97XD		 1.795, 1.817 1.385 +249 10.14469/hpc/2411

As the solvation and environment of the cationic model improves, the apical distance shortens significantly. But the crunch comes when a chloride counter-anion is added to desymmetrise this environment. Using the veritable B3LYP functional, but with an added dispersion term (D3BJ) and starting from a partially optimised ion-pair geometry, this geometry optimisation (shown animated below) rapidly quenches the ion-pair to form a covalent norbornyl chloride. It is noteworthy that the magnitude of the [1,2] vibration force constant (140 cm-1) is rather smaller using B3LYP than the other methods explored. 

The next method tried was ωB97XD, which contains a built-in dispersion term (D2) and also reveals a larger force constant for the gas phase [1,2] shift (≡235 cm-1). Starting from the same initial geometry as the B3LYP calculation, optimisation of the ion-pair proceeds remarkably slowly‡ (even using the recalcfc=5 keyword to recompute the force constant matrix/search direction every five cycles to improve behaviour), suggesting that the potential energy surface is very flat indeed. The final geometry retains the ion-pair character (dipole moment 23D) but reveals distinct asymmetry in the resulting bridged structure, for which the [1,2] shift is ν 249 cm-1.

It is clear that the structure of the norbornyl ion-pair is balanced on a knife-edge. Perturbations such as change of density functional (e.g. B3LYP+D3BJ) can topple it over that edge. Weaker asymmetry can also be induced by the presence of the contact-anion and water molecules. I have selected just one solvation model, which includes seven water molecules and an explicit anion. Clearly a more statistical and dynamical approach to the number of waters and their orientation around the norbornyl ring system would sample a much larger set of models. It may be that some of them do again topple the symmetric bridge structure off its delicate perch whilst others retain it. Perhaps this is why the results from the enormous range of solvolysis mechanisms are so difficult to always reconcile. A crystal structure may also be a relatively large perturbation to the solution structure of this species!

The title of one of the last articles published (posthumously) with Paul Schleyer as a co-author[1] is “Norbornyl Cation Isomers Still Fascinate“. True indeed.

‡This renders refinement using the B2PLYPD3 double-hybrid method[2] an exceptionally slow process, since computing the force constant matrix using this method is very computationally intensive at the selected triple-ζ level.

References

  1. P.V.R. Schleyer, V.V. Mainz, and E.T. Strom, "Norbornyl Cation Isomers Still Fascinate", ACS Symposium Series, pp. 139-168, 2015. https://doi.org/10.1021/bk-2015-1209.ch007
  2. L. Goerigk, and S. Grimme, "Efficient and Accurate Double-Hybrid-Meta-GGA Density Functionals—Evaluation with the Extended GMTKN30 Database for General Main Group Thermochemistry, Kinetics, and Noncovalent Interactions", Journal of Chemical Theory and Computation, vol. 7, pp. 291-309, 2010. https://doi.org/10.1021/ct100466k

Tags:, , , , , , , , , ,
Posted in crystal_structure_mining, Interesting chemistry, reaction mechanism | 8 Comments »

Silyl cations?

Thursday, March 23rd, 2017

It is not only the non-classical norbornyl cation that has proved controversial in the past. A colleague mentioned at lunch (thanks Paul!) that tri-coordinate group 14 cations such as R3Si+ have also had an interesting history.[1] Here I take a brief look at some of these systems.

Their initial characterisations, as with the carbon analogues, was by 29Si NMR. The first (of around 25) crystal structures appeared in 1994 (below) and they continue to fascinate to this day. I decided to focus on searching the Cambridge structure database (CSD), using the search query shown below (NM = non-metal). For a planar system the three angles subtended at the Si would of course total to 360°.

The first such structure, published in 1994[2] is shown in 2D representation below

However, the three angles subtended at the Si are 113, 115 and 114°. Could it be that these types of cation are not planar but pyramidal (a ωB97XD/Def2-TZVPP calculation of SiH3+ certainly gives it as planar). Below is a plot of the three angles:

Ringed in red are two systems where all three angles are ~120° (the ones with red dots). The blue circle contains examples where all three angles are <110°. So I took a closer look at the first of these published[2] and known by the code HAGCIB10 (angles of 113, 115 and 114°). The Si appears to be connected to a toluene present in the crystals via an Si-C bond (blue arrow). If correct, that would account for the angles around Si being <120° and indeed closer to tetrahedral, but it would also mean that the species was actually an arenium cation, otherwise known as a “Wheland intermediate”. That extra bond means that it is not a tri-coordinate silicon, but a four-coordinate silicon and that perhaps the indexing in the CSD needs correcting (as was done here).

Looking further, quite a few of the 25 examples contain so-called “N-heterocyclic carbene” ligands, as below (DOI for 3D model: 10.5517/CC12FWM0[3]).

Again one might question the location of the formal +ve charge. Perhaps it might instead reside on the nitrogen as per below, in which case we again do not have a true tri-coordinate silicon cation for systems with such ligands.

This cannot be the whole story, although I would note that Si=C bonds can contain pyramidalised Si. The bonding clearly needs more investigation! 

Very probably each of the 25 examples identified by this search as a silylium or silyl cation has its own story to tell. But in unravelling these stories, one should always perhaps take the 2D representations shown in both the CSD and the original publications with a pinch of salt until other possibly better representations such as the one above are excluded.

References

  1. J.B. Lambert, Y. Zhao, H. Wu, W.C. Tse, and B. Kuhlmann, "The Allyl Leaving Group Approach to Tricoordinate Silyl, Germyl, and Stannyl Cations", Journal of the American Chemical Society, vol. 121, pp. 5001-5008, 1999. https://doi.org/10.1021/ja990389u
  2. J.B. Lambert, S. Zhang, and S.M. Ciro, "Silyl Cations in the Solid and in Solution", Organometallics, vol. 13, pp. 2430-2443, 1994. https://doi.org/10.1021/om00018a041
  3. T. Agou, N. Hayakawa, T. Sasamori, T. Matsuo, D. Hashizume, and N. Tokitoh, "Reactions of Diaryldibromodisilenes with N‐Heterocyclic Carbenes: Formation of Formal Bis‐NHC Adducts of Silyliumylidene Cations", Chemistry – A European Journal, vol. 20, pp. 9246-9249, 2014. https://doi.org/10.1002/chem.201403083

Tags:, , , , , , , ,
Posted in crystal_structure_mining | 8 Comments »

George Olah and the norbornyl cation.

Friday, March 10th, 2017

George Olah passed away on March 8th. He was part of the generation of scientists in the post-war 1950s who had access to chemical instrumentation that truly revolutionised chemistry. In particular he showed how the then newly available NMR spectroscopy illuminated structures of cations in solvents such “Magic acid“. The obituaries will probably mention his famous “feud” with H. C. Brown over the structure of the norbornyl cation (X=CH2+), implicated in the mechanism of many a solvolysis reaction that characterised the golden period of physical organic chemistry just before and after WWII. 

The dispute between Olah and Brown was not played on a pitch using quite the same goal posts. Olah did much of his work in magic acid and Brown did his in aqueous solutions. I was involved in a tiny way when the discussion about the precise character of the norbornyl cation was reaching its peak in the mid 1970s. At the time, I was working with Michael Dewar, who was himself not shy in joining in the fun and sometimes very acrimonious disputes at conferences. We contributed by calculating the so-called core-electron carbon ESCA spectrum.[1] History records that we came down on the wrong side,‡ by suggesting that this form of spectroscopy supported Brown rather than Winstein/Olah on the basis of a 6:1 spectral deconvolution (classical) rather than 5:2 (non-classical). More recently of course the crystal structure of the parent cation itself has been shown to be non-classical[2] (there are other crystal structures which differ in respect to having one or more additional methyl groups[3]). For a 3D model of norbornyl cation, see DOI: 10.5517/CCZ21LN. This still leaves the issue (very slightly) open for the structure of the solvated cation when formed in water! 

When I started to teach a course in molecular modelling, I touched briefly on how modelling could contribute and whilst updating the notes in the 1990s, wondered why the boron analogue had never been so studied (X=BH2). Unlike the crystallographically difficult norbornyl ion-pair, the iso-electronic boron species would be neutral and not need a counter-ion. Perhaps it might be a more manageable molecule? Checking the Cambridge structural database, such a species has never been reported!† So here as my homage to Olah, I report its calculated structure (b2plypd3/Def2-TZVPP, DOI: 10.14469/hpc/2236).

The norbornyl cation has symmetrical C-C bridging distances of ~1.80±0.02Å and a basal C-C distance of ~1.39±0.02Å. The calculated values for the boron equivalent are 2.16Å and 1.36Å respectively, with all positive force constants. B-C bonds are normally 1.66-1.72Å, significantly longer than C-C bonds, which makes the longer B-C lengths in this example unsurprising. More interestingly, the species has one vibrational normal mode (ν 203 cm-1) which corresponds to the [1,2] shift of the BH2 group across the basal C-C. For a classical species, this vibrational motion would correspond to a transition state (an imaginary vibration) but for a non-classical species it is of course real. In this sense it is analogous to the so-called real Kekulé mode in non-classical benzene, which “equilibrates” the two classical Kekulé structures. The corresponding calculated vibration for the norbornyl cation itself is ν 194 cm-1 (DOI: 10.14469/hpc/2238).

Of course, the entire controversy over the structure of this species is littered with comparisons between not quite similar systems, differing in a methyl group more or less. So morphing a C+ to a B might be seen as quite a large change. But perhaps if it had been crystallised in say the 1960s, would the subsequent debates have taken a different turn?

‡ We were also wrong about the symmetry of the Diels-Alder cyclisation, which is nowadays accepted to be synchronous rather than asynchronous for simple  Diels-Alder reactions. But that is another story.

†GAXLIA is perhaps the closest analogue.[4],

References

  1. M.J.S. Dewar, R.C. Haddon, A. Komornicki, and H. Rzepa, "Ground states of molecules. 34. MINDO/3 calculations for nonclassical ions", Journal of the American Chemical Society, vol. 99, pp. 377-385, 1977. https://doi.org/10.1021/ja00444a012
  2. F. Scholz, D. Himmel, F.W. Heinemann, P.V.R. Schleyer, K. Meyer, and I. Krossing, "Crystal Structure Determination of the Nonclassical 2-Norbornyl Cation", Science, vol. 341, pp. 62-64, 2013. http://dx.doi.org/10.1126/science.1238849
  3. T. Laube, "Redetermination of the Crystal Structure of the 1,2,4,7‐<i>anti</i>‐tetramethylbicyclo[2.2.1]heptan‐2‐yl cation at 110 K", Helvetica Chimica Acta, vol. 77, pp. 943-956, 1994. https://doi.org/10.1002/hlca.19940770407
  4. P.J. Fagan, E.G. Burns, and J.C. Calabrese, "Synthesis of boroles and their use in low-temperature Diels-Alder reactions with unactivated alkenes", Journal of the American Chemical Society, vol. 110, pp. 2979-2981, 1988. https://doi.org/10.1021/ja00217a053

Tags:, , , , , , , , , , , , , ,
Posted in Interesting chemistry | 18 Comments »

What’s in a name? Carbenes: a reality check.

Sunday, September 11th, 2016

To quote from Wikipedia: in chemistry, a carbene is a molecule containing a neutral carbon atom with a valence of two and two unshared valence electrons. The most ubiquitous type of carbene of recent times is the one shown below as 1, often referred to as a resonance stabilised or persistent carbene. This type is of interest because of its ability to act as a ligand to an astonishingly wide variety of metals, with many of the resulting complexes being important catalysts. The Wiki page on persistent carbenes shows them throughout in form 1 below, thus reinforcing the belief that they have a valence of two and by implication six (2×2 shared + 2 unshared) electrons in the valence shell of carbon. Here I consider whether this name is really appropriate.

carbenes

Let us start by counting the electrons in the 2p atomic orbitals on the ring atoms of 1, forming what we call a π-system. There are six; two from the carbons shown connected by a double bond, C=C and a further four from the two nitrogen lone pairs. Now in benzene, we also have six π-electrons in a ring and this molecule is of course famously aromatic due to the diatropic ring current created by the circulation of these six electrons. Moreover, all the C-C bonds are equal in length, ~1.4Å long (although the reasons for this equality are subtle).

So does 1 behave similarly? A ωB97XD/Def2-TZVPP calculation[1] shows the following calculated structure, in which all the bonds are clearly intermediate between single and double. The N-C(“carbene”) length of 1.357Å in particular is much shorter than a C-N single bond (~1.44AÅ), which tends to suggest that the resonance form 2 is a better representation than 1. This form is also pretty similar to pyrrole, itself a well-known hetero-aromatic species.nhc1

An alternative reality check is crystal structures. There are 42 examples (no errors, no disorder, R < 0.05) in the Cambridge structure database (CSD) and the distribution of C-N bond lengths below is indeed quite similar to the calculation shown above for the unsubstituted parent, with the lhs “hot-spot” almost exactly coincident. The C-C length similarly corresponds.

nhc2

nhc3

Let us try a technique for explicitly counting electrons, the ELF (electron localisation method), which works directly on a function of the electron density to identify the centroids of localized “basins” containing the integrated density. The three surrounding the “carbene” atom sum to 7.54e (with small seepage also into the carbon 1s core; 2.08e). A “normal” carbon on the C=C bond is 7.65e. The localization below turns out to closely resemble resonance structure 2 above.

nhc4

Further in-silico experiments can be carried out with species 3 and 4, in which a carbon atom replaces each of the nitrogens. This reduces the total electron count by two and now this poor molecule has a difficult choice to make. Should it be the π-system that sacrifices these two electrons, or could it be the σ-lone-pair found on the two-coordinate carbon? We will let the quantum mechanical solution decide[2] (with a constraint that the molecule be planar). The electrons arrange themselves to resemble the resonance form 4, choosing to retain the six π-electrons and sacrifice the carbene “unshared pair”. The 2-coordinate carbon as a vinyl cation now does have ~6 valence electrons (ELF indicates 5.23e). nhc5

What about the other choice? By promoting two electrons from HOMO to LUMO one can also calculate 3 (again constrained to planarity)[3] which finally does correspond to the classical description of a carbene.

nhc6

The arrow connecting 3 and 4 in the scheme at the top is NOT in this case an electronic resonance, but a a real equilibrium between two different species separated by an energy barrier. With only four π-electrons in a cycle it is also antiaromatic, and so the two localised alkene bonds avoid any conjugation with each other. This form has a free energy some 5.7 kcal/ml higher than the aromatic form. In fact, the molecule is very keen to avoid all antiaromaticity and hence if the planar constraint is lifted, it will distort with no activation to a non-planar diene (just as cyclo-octatetraene does to a non-planar tetra-ene). And to complete the tale, even though 4 is aromatic, it too distorts without activation to an odd-looking non-planar form with no symmetry[4],[5],[6] (but that is another story).

The final word should be that the naming of these types of persistent carbene does need a reality check; they should not be called this at all! They are really dipolar species or carbon-ylides as shown in 2. As it happens, a very closely related species in which one sulfur replaces one nitrogen is a very familiar compound, vitamin B1 or thiamine. The only example of a stable deprotonated thiamine derivative is referred to as a carbene[7], perhaps because with an acid catalyst it can dimerise in the manner expected of a real carbene. Significantly however, without acid catalyst this does not happen; a true carbene would not require such a catalyst.

References

  1. H. Rzepa, "NHC wfn", 2016. https://doi.org/10.14469/hpc/1473
  2. H. Rzepa, "butadiene carbene aromatic -192.700746", 2016. https://doi.org/10.14469/hpc/1581
  3. H. Rzepa, "butadiene carbene antiaromatic guess=alter -192.691607", 2016. https://doi.org/10.14469/hpc/1582
  4. H. Rzepa, "C5H4 non-planar, Cs symmetry", 2016. https://doi.org/10.14469/hpc/1583
  5. H. Rzepa, "C5H4 non-planar, C2 symmetry", 2016. https://doi.org/10.14469/hpc/1584
  6. H. Rzepa, "C5H4 non-planar, no symmetry", 2016. https://doi.org/10.14469/hpc/1585
  7. A.J. Arduengo, J.R. Goerlich, and W.J. Marshall, "A Stable Thiazol‐2‐ylidene and Its Dimer", Liebigs Annalen, vol. 1997, pp. 365-374, 1997. https://doi.org/10.1002/jlac.199719970213

Tags:, , , , , , , , , , , , ,
Posted in crystal_structure_mining, General | No Comments »

What's in a name? Carbenes: a reality check.

Sunday, September 11th, 2016

To quote from Wikipedia: in chemistry, a carbene is a molecule containing a neutral carbon atom with a valence of two and two unshared valence electrons. The most ubiquitous type of carbene of recent times is the one shown below as 1, often referred to as a resonance stabilised or persistent carbene. This type is of interest because of its ability to act as a ligand to an astonishingly wide variety of metals, with many of the resulting complexes being important catalysts. The Wiki page on persistent carbenes shows them throughout in form 1 below, thus reinforcing the belief that they have a valence of two and by implication six (2×2 shared + 2 unshared) electrons in the valence shell of carbon. Here I consider whether this name is really appropriate.

carbenes

Let us start by counting the electrons in the 2p atomic orbitals on the ring atoms of 1, forming what we call a π-system. There are six; two from the carbons shown connected by a double bond, C=C and a further four from the two nitrogen lone pairs. Now in benzene, we also have six π-electrons in a ring and this molecule is of course famously aromatic due to the diatropic ring current created by the circulation of these six electrons. Moreover, all the C-C bonds are equal in length, ~1.4Å long (although the reasons for this equality are subtle).

So does 1 behave similarly? A ωB97XD/Def2-TZVPP calculation[1] shows the following calculated structure, in which all the bonds are clearly intermediate between single and double. The N-C(“carbene”) length of 1.357Å in particular is much shorter than a C-N single bond (~1.44AÅ), which tends to suggest that the resonance form 2 is a better representation than 1. This form is also pretty similar to pyrrole, itself a well-known hetero-aromatic species.nhc1

An alternative reality check is crystal structures. There are 42 examples (no errors, no disorder, R < 0.05) in the Cambridge structure database (CSD) and the distribution of C-N bond lengths below is indeed quite similar to the calculation shown above for the unsubstituted parent, with the lhs “hot-spot” almost exactly coincident. The C-C length similarly corresponds.

nhc2

nhc3

Let us try a technique for explicitly counting electrons, the ELF (electron localisation method), which works directly on a function of the electron density to identify the centroids of localized “basins” containing the integrated density. The three surrounding the “carbene” atom sum to 7.54e (with small seepage also into the carbon 1s core; 2.08e). A “normal” carbon on the C=C bond is 7.65e. The localization below turns out to closely resemble resonance structure 2 above.

nhc4

Further in-silico experiments can be carried out with species 3 and 4, in which a carbon atom replaces each of the nitrogens. This reduces the total electron count by two and now this poor molecule has a difficult choice to make. Should it be the π-system that sacrifices these two electrons, or could it be the σ-lone-pair found on the two-coordinate carbon? We will let the quantum mechanical solution decide[2] (with a constraint that the molecule be planar). The electrons arrange themselves to resemble the resonance form 4, choosing to retain the six π-electrons and sacrifice the carbene “unshared pair”. The 2-coordinate carbon as a vinyl cation now does have ~6 valence electrons (ELF indicates 5.23e). nhc5

What about the other choice? By promoting two electrons from HOMO to LUMO one can also calculate 3 (again constrained to planarity)[3] which finally does correspond to the classical description of a carbene.

nhc6

The arrow connecting 3 and 4 in the scheme at the top is NOT in this case an electronic resonance, but a a real equilibrium between two different species separated by an energy barrier. With only four π-electrons in a cycle it is also antiaromatic, and so the two localised alkene bonds avoid any conjugation with each other. This form has a free energy some 5.7 kcal/ml higher than the aromatic form. In fact, the molecule is very keen to avoid all antiaromaticity and hence if the planar constraint is lifted, it will distort with no activation to a non-planar diene (just as cyclo-octatetraene does to a non-planar tetra-ene). And to complete the tale, even though 4 is aromatic, it too distorts without activation to an odd-looking non-planar form with no symmetry[4],[5],[6] (but that is another story).

The final word should be that the naming of these types of persistent carbene does need a reality check; they should not be called this at all! They are really dipolar species or carbon-ylides as shown in 2. As it happens, a very closely related species in which one sulfur replaces one nitrogen is a very familiar compound, vitamin B1 or thiamine. The only example of a stable deprotonated thiamine derivative is referred to as a carbene[7], perhaps because with an acid catalyst it can dimerise in the manner expected of a real carbene. Significantly however, without acid catalyst this does not happen; a true carbene would not require such a catalyst.

References

  1. H. Rzepa, "NHC wfn", 2016. https://doi.org/10.14469/hpc/1473
  2. H. Rzepa, "butadiene carbene aromatic -192.700746", 2016. https://doi.org/10.14469/hpc/1581
  3. H. Rzepa, "butadiene carbene antiaromatic guess=alter -192.691607", 2016. https://doi.org/10.14469/hpc/1582
  4. H. Rzepa, "C5H4 non-planar, Cs symmetry", 2016. https://doi.org/10.14469/hpc/1583
  5. H. Rzepa, "C5H4 non-planar, C2 symmetry", 2016. https://doi.org/10.14469/hpc/1584
  6. H. Rzepa, "C5H4 non-planar, no symmetry", 2016. https://doi.org/10.14469/hpc/1585
  7. A.J. Arduengo, J.R. Goerlich, and W.J. Marshall, "A Stable Thiazol‐2‐ylidene and Its Dimer", Liebigs Annalen, vol. 1997, pp. 365-374, 1997. https://doi.org/10.1002/jlac.199719970213

Tags:, , , , , , , , , , , , ,
Posted in crystal_structure_mining, General | No Comments »