1
Bachelor Thesis Chemistry
1,1’,2,2’-Tetracyanocyclopropane Compounds to Study
Tetrel Bonding Interactions
By
Julius J. Roeleveld
6 June 2019
Student Number
10766456
Research Institute
Van 't Hoff Institute for Molecular
Sciences
Research Group
Homogenous, Supramolecular and
Bio-Inspired Catalysis
Responsible Teacher
Dr. T.J. Mooibroek
Daily Supervisor
X. Schaapkens, MSc
Second Reviewer
2
Samenvatting
Een van de sterkere interacties die deel uitmaakt van supramoleculaire, ofwel niet-covalente chemie is de waterstofbrug, vaak afgekort als D–H∙∙∙A (D = donor, A = acceptor). Deze interactie is een resultaat van de aantrekkingskracht tussen een elektronegatief atoom, een acceptor (A) zoals zuurstof in een keton, en het elektropositieve waterstof (H). Waterstof is elektropositiever wanneer het aan een elektron zuigende groep, de donor (D), gebonden is zoals in een alcohol het geval is. Wanneer het waterstof atoom wordt vervangen door een halogeen atoom (Hlg), ontstaat op het halogeen atoom een elektropositief domein. Zo’n domein ligt in het verlengde van de covalente (sigma, σ) binding met het halogeen en wordt een σ-hole genoemd. Het σ-hole kan een interactie aangaan met een elektronegatief atoom om een zogeheten halogeenbrug te vormen, in analogie met waterstof bruggen vaak afgekort als D–Hlg∙∙∙A. Uit verder onderzoek bleek dat naast de halogenen in groep 17 van het periodiek systeem hetzelfde mogelijk was voor groepen 16, 15 en 14, respectievelijk pnictogen, chalcogen en tetrel bonding genoemd. Koolstof is het enige niet-metaal van de tetrel groep en is het meest voorkomend in biologie. Dit onderzoek focust zich op zogenoemde tetrel interacties met koolstof.
Koolstof is vierwaardig en heeft normaliter een tetrahedrale geometrie als het sp3 gehybridiseerd
is (d.w.z.: CR4). In een tetrahedraal koolstofatoom zijn vier mogelijke σ-holes van de vier C–R
bindingen. Interactie met deze σ-holes wordt echter sterisch gehinderd door de drie andere R-groepen. In dit onderzoek is een cyclopropaanring geïmplementeerd om deze R-substituenten van elkaar weg te draaien en zo de σ-holes meer toegankelijk te maken. Om de koolstofatomen van de cyclopropaanring elektron-arm te maken zijn vier cyano substituenten als R-groep gebruikt op de 1,1’,2,2’-posities. Cyanogroepen zijn slecht vertrekkende groepen en relatief eenvoudig om te maken. Dergelijke 1,1’,2,2’-tetracyanocyclopropaan (TCCP) moleculen zijn te synthetiseren uit twee moleculen malononitrile, een keton (of aldehyde) en moleculair broom als terminale oxidator (onder basische condities). De resterende twee substituenten op het TCCP molecuul (de 3,3’-posities) kunnen worden gevarieerd door de substituenten van het gebruikte keton (of aldehyde) te variëren. Producten met te kleine rest groepen lijden onder instabiliteit ten gevolge van ring-openingsactiviteit. Hieruit volgde de hypothese dat grotere restgroepen ring opening tegen zou werken. Om dit te onderzoeken zijn in synthese verschillende ketonen gebruikt, waarvan er drie succesvol zijn geïsoleerd. Dit waren aceton, 3-pentanon en 4H-pyran-4-on. De producten met grotere restgroepen bleken daadwerkelijk verminderde ring-openingsactiviteit te hebben, en waren hierdoor in dit opzicht stabieler. Dit resultaat kan men zien als een stap dichterbij het ontrafelen van de nog relatief onbekende ‘koolstof tetrelbrug’ en eventuele toepassingen daarvan.
3
Abstract
1,1’,2,2’-tetracyanocyclopropane (TCCP) entities have shown to exhibit supramolecular interaction with Lewis acids by means of tetrel bonding. This phenomenon was effectuated by opening up an electropositive domain called a σ-hole generated by the electron withdrawing cyano substituents bound to carbon in a cyclopropane ring due to fixed torsion angle between the regarded bonds. Previous research showed evidence that ring opening might occur by nucleophilic attack at opposite side of the tetrel bonding site. To this end, TCCP entities with bulkier substituents were attempted to be synthesized via a reaction that involved Knoevenagel condensation of a ketone and malononitrile into alkylidenemalononitrile and Michael addition of bromomalononitrile to alkylidenemalononitrile. Besides acetone, products were obtained out of 3-pentanone and 4H-pyran-4-one. These were co-crystallized out of various solvents, of which tetrahydrofuran and 1,4-dioxane resulted in crystals fit for X-ray analysis. Additionally, density function theory and atoms in molecules calculations were carried out to complement X-ray data. Acquired data showed that steric hindrance prevents affinity for the ring opening interaction site of the cyclopropane ring, and demonstrated significant tetrel bonding activity with perpendicular geometry of the Lewis acid backbone with respect to the cyclopropane ring. Furthermore, oxygen in the pyranone product exhibited interaction with the σ-hole of another pyranone product entity. The resulting data call for further exploration of this principle, as many TCCP entities have yet to be studied for signs of tetrel bonding activity.
4
List of abbreviations
AIM Atoms In Molecules
BCP Bond Critical Point
DCM Dichloromethane
DFT Density Functional Theory
D.I. Dispersion Interaction
E.I. Electrostatic Interaction
MEP Molecular Electrostatic Potential
O.I. Orbital Interaction
P.R. Pauli Repulsion
TCCP Tetracyanocyclopropane
THF Tetrahydrofuran
5
Table of Contents
Samenvatting ... 2 Abstract ... 3 List of abbreviations ... 4 Table of Contents ... 5 1. Introduction ... 6 1.1. Supramolecular Chemistry ... 6 1.2. Objectives ... 92. Results & Discussion ... 10
2.1. Synthesis ... 10 2.1.1. Mechanism ... 10 2.1.2. Reactions ... 12 2.2. X-ray Analysis ... 21 2.2.1. General Comments ... 21 2.2.2. 3,3-dimethyl-1,1,2,2-tetracyanocyclopropane ... 22 2.2.3. 3,3-diethyl-1,1,2,2-tetracyanocyclopropane ... 23 2.2.4. 6-oxa-1,1,2,2-tetracyanospiro[2.5]octane ... 24 2.3. Computational Data ... 25
2.3.1. Homo-dimers and Adducts with THF and 1,4-dioxane ... 25
2.3.3. Interactions with Halide Anions ... 28
2.3.4. Effect of Solvent ... 28
2.3.5. Comparative Atoms in Molecules and Energy Decomposition Analyses ... 29
3. Conclusion & Future Prospects ... 34
4. Experimental ... 36 4.1. General ... 36 4.2. Synthesis ... 36 4.2.1. 3,3-dimethyl-1,1,2,2-tetracyanocyclopropane ... 36 4.2.2. 3,3-diethyl-1,1,2,2-tetracyanocyclopropane ... 36 4.2.3. 6-oxa-1,1,2,2-tetracyanospiro[2.5]octane ... 37 4.3. X-ray Data ... 37 4.3.1. 3,3-dimethyl-1,1,2,2-tetracyanocyclopropane from THF ... 37
4.3.2. 3,3-diethyl-1,1,2,2-tetracyanocyclopropane from 1,4-dioxane ... 37
4.3.3. 3,3-diethyl-1,1,2,2-tetracyanocyclopropane from THF ... 38
4.3.4. 6-oxa-1,1,2,2-tetracyanospiro[2.5]octane from 1,4-dioxane ... 38
5. Acknowledgements ... 39
6. Bibliography... 40
7. Appendix ... 41
7.1 Visual clarification of torsion angle ... 41
6
1. Introduction
1.1. Supramolecular Chemistry
Just like elements have the tendency to attract one another by sharing electrons in their orbitals resulting in covalent bonds, molecules among each other are also subjected to forces that lead to structures in which different subunits are oriented in a specific way. This orientation is not necessarily a consequence of covalent bonding of atoms, but rather of non-covalent bonding; the domain of supramolecular chemistry. This branch of science consists of understanding and utilising hydrogen bonding, hydrophobic interactions, metal coordination, π-π interactions, and electrostatic interactions.1 These
kinds of interactions are characterized by adopting a state of equilibrium as opposed to classical reactions, where an irreversible reaction generates truly new molecules. All the knowledge of these interactions put together enables us to understand and explain complex behaviour and structures of both small and macro-molecules.
Hydrogen bonding is undoubtedly a well-studied concept in chemistry. A concept that is rather easy to understand. An electronegative atom is covalently bound to hydrogen, resulting in decreased electron density at the hydrogen. This resulting absence of electrons allows a Lewis base to interact with the hydrogen, forming a hydrogen bond in the process. But this phenomenon is not limited to hydrogen alone. Research has shown that the same applies for halogens for instance.2 At first sight it would seem
illogical for an electronegative moiety like this to display the ability to interact with another electronegative atom, instead of repelling it. But it becomes clear that an electropositive domain is present on the halogen when it is bound to an electron withdrawing atom or molecule. Such a domain is called a σ-hole and is located at the end of the vector of a covalent bond. More specifically, where the antibonding σ*-orbital of the bond is located. Besides with halogen bonding, σ-holes are also present with chalcogen, pnictogen or the recently disclosed tetrel elements. Interactions with the latter three elements are less well known,3 which is the incentive for studying tetrel bonding as documented in this
7 Halogen and tetrel bonding are fairly
comparable. However, the difference in valence gives rise to different characteristics. A higher valence gives the ability to bind a larger amount of electron withdrawing groups, which might lead to stronger polarization and thus stronger interaction with Lewis bases. However, a higher valence also means that a larger amount of atoms is bound to the regarded atom, making the σ-hole more sterically hindered. This becomes
clear when comparing the electrostatic potential maps (MEPs) of several small molecules (see Figure 1). BrF for instance generates a σ-hole that is relatively more accessible than those of SiF4. When two
tetrel atoms with electron withdrawing substituents (X) are bound together like in R2XC–CXR2, their
σ-holes may be sterically inaccessible due to steric and/or electronic repulsion of the other R and X substituents.
To make such an R2XC–CXR2 moiety sterically accessible they can be taken up in a small
cycloalkane. As can be seen in Figure 2, the X–C–C–X torsion angle decreases when a cycloalkane is smaller (see also appendix 7.1). This notion was substantiated for X = F and CN by high-level ab initio computations and inspection of the Cambridge Structure Database.4 The larger carbon rings labelled C5
and C6 in Figure 2 have more freedom of orientation and will consequently assume a conformation in which the substituents are more distant from each other characterized by a relatively big torsion angle. Smaller rings on the other hand are more strained. Cyclopropane (C3) does not allow for any contortion
Figure 1: MEP surfaces of a range of different molecules at the DFT-B3LYP/6-31+G* level of theory with energies in kcal·mol-1.2
Figure 2: Representations of the lowest energy conformers of tetracyanocycloalkanes (X=CN) with MEP surfaces and smallest torsion angle with potential energies at the σ-hole indicated in kcal mol-1.4
Table 1: Energies in kcal mol-1 and equilibrium distances
in Å of interactions between cycloalkanes and Lewis bases.4
8 of the structure whatsoever, which leads to a torsion angle of 0°. Cyclobutane (C4) however does have does have some degree of torsion.
Nevertheless, MEP surfaces show that in both these smaller carbon rings the σ-holes join together forming a bowl of electropositivity. As previously stated, this area of electropositivity is caused by electron withdrawing substituents, of which only two were studied: –F and –CN. This is because of their synthetical accessibility and the fact that they are poor leaving groups.4 When a strong leaving
group is used as substituent, an SN2 reaction could occur, which is undesirable. The desired
supramolecular interaction of a Lewis bases was found to occur mainly with tetracyanocyclopropane (TCCP) and -butane (see Table 1). The cyclopentane and cyclohexane rings showed no complex formation at all. Of the fluorinated alkanes only cyclopropane showed energetically favourable interactions and even those are not very strong. The strongest interactions have an energy of -6.5 kcal/mol (with HO– or F–) and this is about 6 times less than their tetracyano counterparts. This proves that in this case the cyano substituent is better than fluoride in terms of ability to interact with Lewis bases, and with that, its importance in exploring tetrel bonding.
The carbon ring with the spatially more accessible σ-hole, cyclopropane, was chosen to be subjected to chemical research. Previous research showed that the ring had the tendency to open when trying to effectuate tetrel bonding with several nucleophiles due to SN2 activity.5 Two different entities
exhibited this tendency. The first one, 3,3-dimethyl-1,1’,2,2’-tetracyanocyclopropane seemed to be prone to nucleophilic attack of chloride directly on the cyclopropane ring where the (CN)2C–C(CN)2
bond between the two quaternary carbon atoms is broken (see Scheme 1). The resulting carbanion decomposes into another carbanion and isopropylidenemalononitrile. The second entity,
3-phenyl-3-
trifluoromethyl-1,1’,2,2’-tetracyanocyclopropane, was found to exhibit ring opening activity by nucleophilic attack of chloride on the para-carbon of the phenyl substituent (see Scheme 2). The conjugated ring distributes the negative charge to the cyclopropane ring where one of its bonds is broken. The resulting carbanion decomposes into another carbanion and a malononitrile
Scheme 1: Potential ring opening mechanism of 3,3-dimethyl-1,1,2,2-TCCP.5
Scheme 2: Potential ring opening mechanism of 3-phenyl-3-trifluoromethyl-1,1,2,2-TCCP.5
9 derivative again. These two examples attest to the instability TCCP derivatives can experience.
To this end, the decision was made to attempt to synthesize TCCP derivatives with bulky R-groups (see Scheme 3), so the ring is sterically more hindered. The synthesis was executed on the basis of the method of Vereshchagin et al.6 that employs a basic aqueous bromine solution containing NaOAc
to transform two molecules of malononitrile and one molecule of ketone into TCCP derivatives with various R-groups.
1.2. Objectives
Six different ketones were chosen to form six different TCCP derivatives (see Figure 3). These derivatives contain different alkyl R-groups, except
1c, which is a cyclic ether with its O-atom opposite
the (CN)2C–C(CN)2 moiety. It was hypothesized
that due to this O-atom 1c has affinity for itself, because it might form a tetrel bond between the σ-hole region and the oxygen (due to its lone pair). The different alkyl groups were regarded for the effect that their bulkiness was going to have on any ring opening activity that might occur.
The TCCP products and their purity were analysed by 1H and 13C NMR, MS, and IR
spectroscopy. Their ability to form tetrel bonded adducts with solvents that have Lewis basic properties was investigated by X-ray crystallography.
In addition to spectroscopy and crystallography, density functional theory (DFT) calculations at the B3LYP-D3/def2-TZVP level of theory was conducted of different adducts. Geometry optimized adducts were further inspected by an ‘atoms in molecules’ (AIM) and an energy decomposition analysis (B3LYP-D3/TZ2P). Interactions of Lewis bases with water and halogens were studied in a similar fashion for comparison purposes.
Scheme 3: Direct formation of TCCP through cascade reaction of ketone with malononitrile using basic bromine solution.
Figure 3: Selection of TCCP derivatives aimed at in this work.
10
2. Results & Discussion
2.1. Synthesis
2.1.1. Mechanism
As is shown in Scheme 4, the reaction process consists of four different reaction steps: (I) Knoevenagel condensation of the carbonyl compound and malononitrile into alkylidenemalononitrile, (II) bromination of malononitrile, (III) Michael addition of bromomalononitrile to alkylidenemalononitrile, and (IV) cyclization of substituted 1-bromo-1,1,3,3-tetracyanopropane carbanion into the corresponding 1,1,2,2-tetracyanocyclopropane.6 The tendency of malononitrile to deprotonate is a result of the stability
of the resulting carbanion. The electron can delocalize over both cyano triple bonds. Fortunately, the anion is not stable enough to prevent it from attacking a ketone which allows the eventual formation of alkylidenemalononitrile to occur.
Previous research showed that utilizing an excess of three equivalents of bromine to the mixture as opposed to two resulted in a higher yield.5 Hence this deviation from the literature method was
implemented. To ensure smooth progression of the reaction, the mixture was refluxed for half an hour prior to adding bromine. This way the alkylidenemalononitrile could be formed without being exposed to bromine activity that might result in unnecessary by-products. The bromine was dissolved in water before adding it to the rest of the mixture as a way to reduce its reactivity and thereby make the reaction more selective.6 Also the product seemed to precipitate in a water mixture, which shifts the reaction to
the product.
The more bulky the alkylic side groups on the ring were, the more the resulting products seemed to exhibit apolar characteristics, consequently making them harder to obtain by precipitation. The larger products did not precipitate at all, but rather formed as an oily substance. This excluded filtration of the mixture as a work-up method to obtain the product. With molecules 1d-f no pure compounds were obtained, which directed the focus toward products 1a-c, which all precipitated from the reaction mixture.
11
Scheme 4: Reaction mechanism of formation of TCCP molecules out of malononitrile and ketone with assignation of different steps.
12
2.1.2. Reactions
Initially, synthesis of the methyl and ethyl products 1a and 1b seemed to yield impure products. To this end sublimation of the products was executed to try to purify the products. This resulted in very low yields (below 20%). Later on it became clear that more thorough washing of the filtration residue was the only impediment to obtain pure products. This mostly resolved the need to sublimate the products and allowed the method of A.N. Vereshchagin et al.6 to be implemented, with use of the excess of
bromine alteration, that is.
2.1.2.1.
3,3-dimethyl-1,1,2,2-tetracyanocyclopropane (1a)
The thorough washing of the residue might be the reason for a lower yield of 25% than the yield of 83% achieved by previous research.5 As is shown in Figure 4, presence of the product was confirmed by 13C
NMR. This spectrum is rather pure, in contrast to those obtained without sufficiently washing the solid. The 1H NMR spectrum of 1a is shown in Figure 5, revealing a large singlet for the methyl groups.
Besides a large solvent (DMSO) and (solvent) water peak, two small peaks are present around 2.06 and 2.38 ppm. These were assigned to acetone and isopropylidenemalononitrile respectively. The presence of isopropylidenemalononitrile impurity was further confirmed by presence of a peak around 1642.75 cm-1 in the IR-spectrum shown in Figure 6. Such a signal is in the range of alkene bond stretch. And
besides the expected signal at 2257.16 cm-1 indicating the situation of nitrile groups in the molecule,
another signal was measured at 2215.46 cm-1. The conjugation of the doubly bound carbon with the
triple bond of the cyano groups may cause the cyano stretching mode to vibrate at a lower frequency compared to those for regular alkylnitriles.7 Together with the fact that this intermediate logically forms
as seen by the mechanism (see Scheme 4), this strongly indicates a slight contamination with isopropylidenemalononitrile. It cannot be present in an excessive amount because else it would be measured by 13C NMR. Indeed, the amount isopropylidenemalononitrile is about 5.8% based on
integration of the 1H NMR spectrum. For MS measurements FD+(eiFi) ionization method was
implemented, where no separation takes place beforehand. Electronic as opposed to chemical ionization is utilized and few to no fragmentation occurs, resulting in the clean spectrum shown in Figure 7. The main peak in the mass spectrum (170.0532) corresponds with the monoisotopic mass of the desired product (170.0592), attesting to successful synthesis. No isopropylidenemalononitrile was observed in the mass spectrum. The origin of the peak at 185.0996 is unknown. The mass of tetracyanoethylene, which could result from in situ fragmentation/recombination of 1a, comes close (with H2O and K+ =
13
Figure 4: 13C NMR spectrum of 1a.
14
Figure 6: IR spectrum of 1a.
15
2.1.2.2.
3,3-diethyl-1,1,2,2-tetracyanocyclopropane (1b)
A higher yield was obtained for 1b (34%) than for 1a (25%). But just like the latter, 34% is not notably high when compared to the original method out of literature, where a yield of 55% was achieved for 1a, and 48% for 1b,6 while addition of 3 equivalents malononitrile should significantly increase the yield.
The purity of the product was confirmed by spectroscopy. A very clear 1H NMR spectrum, void of any
contamination peaks, was obtained (see Figure 8). The triplet is logically ascribed to the CH3 group
because it has a relative integral value of 3, and the group neighbours a CH2 group as seen by the
quadruplicity of the signal of that group, which in its turn has a relative integral value of 2. The 13C
NMR spectrum shown in Figure 9 reveals peaks belonging solely to the solvent and the product just like the one belonging to the 1a (see Figure 4). Another similarity is the cyano IR peak presence (2253.56 cm-1) and the carbonyl peak absence (see Figure 10), ensuring that all of the ketone has either reacted,
or has been removed. In this case, no alkene nor conjugated nitrile peak is observed, making 1b more pure than 1a as no isopropylidenemalononitrile is left in the product. The main peak in the mass spectrum (199.0909) corresponds to the monoisotopic mass of the product (198.0905) and no significant contamination peaks are shown (see Figure 11).
16
Figure 9: 13C NMR spectrum of 1b.
17
Figure 11: Mass spectrum of 1b + calculated monoisotopic mass.
2.1.2.3.
6-oxa-1,1,2,2-tetracyanospiro[2.5]octane (1c)
Unlike the previous smaller products, 1c was obtained with a significantly high yield (58%) compared with those of the methyl and ethyl equivalents in literature, which are close to that value. No comparison can be made to the same product because this reaction has not been done before. The only previously executed synthesis is by means of electrolysis. The yield of this procedure was 69%,8 which is also fairly
close to the achieved 58%. The initial 1H NMR spectrum showed that the product contained ethanol
contamination. After vacuum drying, another 1H NMR spectrum was measured that proved the ethanol
was removed (see Figure 12). Where the initial spectrum showed clear multiplicity of the product and solvent peaks, unfortunately the spectrum of the dry product does not due to faulty NMR analysis settings. Nevertheless, the peaks in the pure spectrum were easily assigned to the hydrogen atoms in the product. The identity of 1c was further verified by the 13C NMR spectrum shown in Figure 13, clearly
revealing only resonances of the carbon nuclei in 1c and the solvent. The IR-spectrum shown in Figure 14 clearly contains a cyano signal at 2254.58 cm-1. The peak at 1092.98 cm-1 is attributed to the C-O
stretch of the cyclic ether in 1c. The spectrum is void of ketone and alkene vibrations, signifying the absence of starting ketone and possible isopropylidenemalononitrile impurity respectively. The mass spectrum of 1c, shown in Figure 15, reveals a peak at 212.0702, which is consistent with the monoisotopic mass of 1c (212.0698).
18
19
Figure 13: 13C NMR spectrum of 1c.
20
21
2.2. X-ray Analysis
2.2.1. General Comments
The successfully obtained products were crystallized out of different solvents containing electron rich atoms: acetone, dichloromethane (DCM), tetrahydrofuran (THF), acetonitrile and 1,4-dioxane. When crystallizing a product out of these solvents, the solvent could join packing in the crystal because of their affinity for the σ-hole in the TCCP derivatives. Analysing such a co-crystal with X-ray crystallography could adeptly depict this. Besides the mentioned solvents, crystallisation with tetrabutyl ammonium bromine and chloride was attempted as well, but this was not successful because the salts did not dissolve well enough in a solution of product in the mentioned solvents that were volatile enough. Because of better solvation, crystallization out of methanol was also attempted, but this also ended in failure as crystals did not grow big enough and the ones that did contained no ions. Fortunately, enough crystals were analysed correctly to allow for meaningful characterization of the products. The successful analyses will further be discussed.
When examining the model structures and their orientations, the existence of tetrel bonding seems evident. But more accurate determination of its occurrence was conducted by examining the distance between the atoms that took part in the alleged tetrel bond, and comparing this with the sum of their van der Waals (vdW) radii. One might define the vdW radius in terms of the distance between two atoms at which their repulsion just balances their attraction forces.9 Various atoms have distinct values
of this radius. If two atoms are closer to each other than their vdW radii justify, it can be said that an interaction occurs. An intermolecular interatomic distance significantly shorter (~0.1 Å or more) than the sum of the vdW radii on the atoms involved can be seen as evidence for an attractive interaction.
Interactions between carbon and oxygen were studied mainly. Hydrogen and nitrogen are less important, but their vdW radii will be used to get a broader understanding of the interactions anyhow. The only type of O-atoms present in all of the crystals were ethers (THF, 1,4-dioxane and 1c). The two types of C-atoms involved in tetrel bonding interactions are the sp3-hybridized C(CN)
2 carbon atoms of
the cyclopropane ring and the sp-hybridized C-atoms of the nitriles. The vdW radii of carbon, oxygen, hydrogen and nitrogen are 1.70 Å, 1.52 Å, 1.55 Å and 1.2 Å respectively.9 Experimental details can be
22
2.2.2. 3,3-dimethyl-1,1,2,2-tetracyanocyclopropane
When looking at the outcome of the X-ray analysis of 1a, its tendency to bind to THF becomes clear (see Figure 16). The packing shows the oxygen atoms are oriented towards the pocket of the TCCP unit. On the other side at the methyl groups the THF molecules are logically faced with their carbon backbone towards the unit.
Orientation aside, the distances between the oxygen atom and six carbon atoms, four cyano and two quaternary, were measured (see Figure 17). The sum of the vdW radii of C+O is 3.22 Å, which is up to 0.213 Å larger than the measured C∙∙∙O distances (between 3.007 and 3.099 Å). It is noteworthy that the interatomic O∙∙∙H distance in a weak (<12 kJ/mol) hydrogen bond ranges from 2.2 to 3.2 Å according to Jeffrey.10 The latter
distance is 0.48 Å longer than the sum of the vdW radii of O (1.52) and H (1.20) of 2.72 Å. If the vdW radii of O and H overlapped by 0.213 Å, the interatomic O∙∙∙H distance would be 2.507 Å. Thus, a tetrel bond interaction like this may be compared to a weak hydrogen bond.
This promising result begged the question whether having two oxygen atoms in the co-crystallized solvent molecule would lead to ditopic binding to TCCP units with the two oxygen atoms in two σ-holes. Seemingly, implementation of 1,4-dioxane solvation resulted in exactly this, although not involving the (CN)2C–C(CN)2 pocket as with THF.
Instead, 1,4-dioxane is positioned in the two
Figure 16: Spatial model of crystal packing from X-ray analysis of 1a in THF.
Figure 17: Distance measurements between the oxygen atom of THF and carbon atoms of 1a.
Figure 18: Spatial model of crystal packing from X-ray analysis of 1a in 1,4-dioxane.
Figure 19: Distance measurements between the oxygen atom of 1,4-dioxane and carbon atoms of 1a.
23 (CH3)2C–C(CN)2 pockets forming infinite alternating rows of 1a–1,4-dioxane (see Figure 18).
The interatomic C∙∙∙O distances in this structure range from 3.044 to 3.194 Å as is shown in Figure 19. While these distances are longer that the THF co-crystal, they are still well within the van der Waals benchmark of 3.22 Å.
2.2.3. 3,3-diethyl-1,1,2,2-tetracyanocyclopropane
A crystal of 1b suitable for X-ray analysis was obtained out of a solution of THF. When examining the resolved crystal structure, no THF was present, as is illustrated in Figure 20. This might be due to overly rapid crystallization as a result of the solvent evaporating too quickly. Within this structure, the (CN)2C–
C(CN)2 pockets are stacked on each other by means of nitrile-nitrile interactions, forming an infinite 1D
chain throughout the crystal (see Figure 21). The C∙∙∙N distances vary between 3.162 and 3.196 Å, which is within the sum of the van der Waals radii of C and N (i.e. 1.70 + 1.55 = 3.25 Å).
When 1b was recrystallized from 1,4-dioxane, a solvent co-crystal was obtained, as is depicted in Figure 22. From this structure it appears that 1,4-dioxane binds to the (CN)2C–C(CN)2 pocket as
observed in 1a∙THF, but not in a ditopic fashion as observed in 1a∙1,4-dioxane. This latter observation might well be due to the increased steric constraint of the ethyl vs the methyl groups. The C∙∙∙O distances
Figure 23: Distance measurements between the oxygen atom of 1,4-dioxane and carbon atoms of 1b.
Figure 22: Spatial model of crystal packing from X-ray analysis of 1b in 1,4-dioxane.
Figure 20: Distance measurements between the cyano nitrogen atoms and the cyano and cyclopropane carbon atoms of adjacent 1b molecules.
Figure 21: Spatial model of crystal packing from X-ray analysis of 1b in THF.
24 observed are as short as 2.908 Å, which is 0.312 Å within the van der Waals benchmark (see Figure 23).
2.2.4. 6-oxa-1,1,2,2-tetracyanospiro[2.5]octane
When crystallizing 1c from 1,4-dioxane, a solvent co-crystal was obtained as is shown in Figures 24–26. As hypothesized, 1c exhibits affinity for its own (CN)2C–
C(CN)2 pocket and stacks into 1-D infinite chains (see
Figure 24). The C∙∙∙O distances observed are as short as 2.955 Å, which is 0.265 Å within the van der Waals benchmark (see Figure 25). Interestingly, the
co-crystallized 1,4-dioxane separates these infinite chains of 1c by interacting with both (C4H6O)C–C(CN)2
sides of each 1c. These interactions are also characterized by short intermolecular C∙∙∙O distances ranging from 2.954 – 3.195 Å, as is depicted in Figure 26. Interestingly, these distances can be split into a short set involving the sterically most accessible side of 1c (2.954 – 3.114 Å) and a longer set involving the more congested side of 1c (3.022 – 3.195 Å). These data show the importance of different substituents on the cyclopropane ring, as they are all accounted by the orientation of the pyran moieties on both sides. It bends away from 1,4-dioxane on one side and towards it on the other side. Bending towards 1,4-dioxane results in steric hindrance of the quaternary carbon and thus a longer distance. Bending away logically results in the opposite. The relatively short distance between oxygen and the sterically unhindered quaternary carbon (2.954 Å) might also partly be attributed by push of the pyran moiety of the opposite 1c molecule, but this is not certain.
Figure 25: Distance measurements between the oxygen atom of 1c and carbon atoms of neighbouring 1c.
Figure 26: Distance measurements between the oxygen atom of 1,4-dioxane and carbon atoms of 1c.
Figure 24: Spatial model of crystal packing from X-ray analysis of 1c in 1,4-dioxane.
25
2.3. Computational Data
2.3.1. Homo-dimers and Adducts with THF and 1,4-dioxane
To probe the binding energies of interactions with TCCP derivatives, density functional theory (DFT) calculations at the B3LYP-D3/def2-TZVP level of theory were carried out of different adducts of 1a,
1b and 1c with themselves, with THF, or with 1,4-dioxane (see Table 2), calculations in the gas phase
at 25 °C). The orientation of the Lewis bases relative to the TCCP derivative is first described as ‘front’ or ‘side’ to denote if the Lewis base is oriented towards the (CN)2C–C(CN)2 or one of the (R)2C–C(CN)2
pockets respectively (with R ≠ CN). Secondly, the Lewis bases are defined as ‘horizontal’ when their ring-plane is co-planar to the cyclopropane ring-plane and as ‘vertical’ when perpendicular to the cyclopropane ring-plane. This is visually clarified in appendix 7.2. Homo-dimers where the TCCP derivative face each other with their (CN)2C–C(CN)2 pockets are labelled as ‘π-π’.
Table 2: DFT calculations (gas phase) at the B3LYP-D3/def2-TZVP level of theory of different adduct conformations.
Lewis base Orientation 1 Orientation 2 Energy (kcal/mol)
1a 1b 1c 1a/1b/1c (homo dimers) Front π-π -8.31 -8.30 -8.39 Front Horizontal -10.27 Front Vertical -11.35 THF Front Horizontal -9.97 Front Vertical -11.17 Side Horizontal -9.12 Side Vertical -10.43 1,4-dioxane Front Horizontal -9.34 Front Vertical -11.27 -11.05 Side Horizontal -7.92 -9.14 Side Vertical-1 -10.48 -9.27 -9.71 Side Vertical-2 -9.19
The energy calculations shown in Table 2 reveal that all the conformations of a Lewis base with a TCCP tetrel bond donor considered are favoured with binding energies as large as -11.35 kcal/mol for the 1c homo-dimer that has its O-atom pointing towards the (CN)2C–C(CN)2 pocket. When regarding the
ability of the three TCCP entities to form homo-dimers in a ‘π-π’ stacking orientation it becomes clear that these dimers are all favourable by about -8.3 kcal/mol. This is higher in energy than all of the other
26 adduct conformation energies except the 'horizontal’ orientation of 1,4-dioxane at the ‘side’ position. Positioning the Lewis base at the ‘front’ is always energetically more favourable than positioning the Lewis base at the ‘side’ of the TCCP derivative by about 1 kcal/mol. This can be rationalized by the diminished steric constraints of a (CN)2C–C(CN)2 versus a (R)2C–C(CN)2 pocket, and/or a more potent
σ-hole region of the ‘front’ position. The differences in energy values between the ‘front’ and ‘side’ positions of 1,4-dioxane with respect to 1a and 1b in the ‘vertical’ orientation are respectively: -10.48 – (-11.27) = 0.79 kcal/mol and -9.27 – (-11.05) = 1.78 kcal/mol. The higher value for 1b suggests that 1,4-dioxane has more difficulty occupying the ‘side’ position when interacting with 1b than with 1a. This was expected, because the larger ethyl groups sterically hinder the ‘side’ position more than the small methyl groups. The ‘vertical’ positioning of both THF and 1,4-dioxane is also about 1 kcal/mol more stable than the ‘horizontal’ orientation. This is also the case for the homo-dimer of 1c where its O-atom is pointing towards the (CN)2C–C(CN)2 pocket. That a ‘vertical’ orientation is preferred could be
explained by orbital overlap. When a ‘vertical’ orientation is adopted, the lone pairs of the sp3 hybridized
oxygen point directly towards the σ*-orbitals of the quaternary carbon atoms of the cyclopropane ring. Interaction of 1c with 1,4-dioxane at the ‘side’ position is more than 1 kcal/mol weaker in energy than the 1c homo-dimer at ‘front’ position, both in ‘vertical’ and horizontal’ orientation. This weaker interaction with ‘vertical’ orientation is 0.52 kcal/mol stronger at the sterically more accessible ‘side’ position than at the more congested one, which suggests that bulkier side groups on the cyclopropane ring might counteract nucleophilic ring opening activity.
Interestingly, in the crystal structure of 1a∙THF the THF is horizontally oriented (see Figure 13). This deviation from computational data is unexpected, but not very surprising, because with the interaction energy calculations a gaseous state was the case and the X-ray analysis counts on molecules being packed in a crystal. Molecules in crystals are significantly affected by surrounding molecules. The carbon backbone of the horizontally oriented THF approximates vertically oriented methyl groups of another TCCP unit, partly explaining the observed orientation of THF in the co-crystal. Besides, the energy difference between the ‘horizontal’ and ‘vertical’ orientations is merely 1.2 kcal/mol for 1a∙THF. However, the calculated preferred ‘vertical’ orientation of 1,4-dioxane molecules was actually found in the co-crystals 1b∙1,4-dioxane and 1c∙1,4-dioxane (see Figures 22 and 24). 1,4-dioxane also adopts ‘vertical’ orientation in the co-crystal 1a∙1,4-dioxane, but at the ‘side’ position as opposed to the ‘front’ position, which results in the energetically unfavourable π-π stacking orientation of the (CN)2C–C(CN)2
moieties. This might be because it has the same defect as THF involving surrounding molecules, which would have resulted in ‘horizontal’ orientation. But 1,4-dioxane is readily able to bind in a ditopic fashion when in the ‘side’ position, making that the energetically most favourable option in crystal form. Having the same defect as THF is a plausible possibility, because if the 1,4-dioxane interaction was strong enough it would assume ditopic binding in ‘vertical’ geometry at the ‘front’ position.
27 This inability of 1,4-dioxane is further substantiated by the structure of co-crystal 1b∙1,4-dioxane, where no ditopic binding is observed either. But in this crystal 1,4-dioxane actually is situated in the ‘front’ position in the ‘vertical’ orientation, which accords to the energetically most favourable adduct orientation. The lack of ditopic binding means that one side of 1,4-dioxane is adjacent to ethyl groups of another TCCP unit. In co-crystal 1a∙THF, THF adopts ‘horizontal’ orientation to avoid close proximity with adjacent methyl groups situated in ‘vertical’ orientation, whereas with 1b, the terminal CH3 extensions of the ethyl groups bend off in ‘horizontal’ direction, consequently favouring the
‘vertical’ orientation of 1,4-dioxane. 1,4-dioxane is not situated at the ‘side’ position of 1b like it does in co-crystal 1a∙1,4-dioxane. The explanation that this is due to 1,4-dioxane experiencing more steric hindrance at 1b because of the larger ethyl groups, corresponds with the previously calculated energy differences between ‘front’ and ‘side’ positioning of 1,4-dioxane with respect to 1a and 1b.
In the structure of co-crystal 1c∙1,4-dioxane only ‘vertical’ orientation of the molecules is found. This is justified by the interaction energies at ‘vertical’ orientation being higher than the ones at ‘horizontal’ orientation. The 1c homo-dimer interaction with ‘vertical’ orientation at the ‘front’ position has the strongest energy of all calculated interactions (-11.35 kcal/mol), which explains that this specific orientation is observed in the co-crystal. Furthermore, the fact that the interaction is stronger at the sterically more accessible ‘side’ position corresponds with the intermolecular C∙∙∙O distances being shorter at that position, further substantiating the suggestion that bulkier side groups on the cyclopropane ring might counteract nucleophilic ring opening activity.
28
2.3.3. Interactions with Halide Anions
Besides intermolecular interactions with neutral guests, calculations of interactions with halide anions with 1a in gas phase were conducted. The anions were oriented at the ‘front’ of the TCCP molecules. The resulting energies listed in Table 3 are all large and negative. From fluor to iodine the interaction energy decreases from -46.36 to -15.20 kcal/mol. This trend follows the ionic radii, which increase from 133 pm of
F– to 220 pm of I–. This implied that when the ionic radius increases, the interaction energy decreases. Interaction energies diminish when the anionic charge is smeared out over a larger region. Nonetheless, the energies of these anion containing adducts are significantly bigger than adducts with THF or 1,4-dioxane. This is no surprise, because the actual negative charge of these entities logically has a great effect on the electropositive domain of the σ-hole compared to the regarded neutral oxygen atoms. On top of that comes the fact that oxygen in the Lewis basic molecules as well as in 1c are bound to methylene on two sides, and these methylene groups will always have some degree of steric hinderance.
2.3.4. Effect of Solvent
Table 4: DFT calculations at the B3LYP-D3/def2-TZVP level of theory of interaction of Lewis bases with 1a in solution.
Lewis base Explicit model of solvation
Geometry Energy (kcal/mol)
1,4-dioxane THF (ε = 7.43) Vertical -7.76 DMF (ε = 37.22) -7.31 H2O (ε = 78.30) -7.16 Cl- THF (ε = 7.43) -6.47 DMF (ε = 37.22) -5.20 H2O (ε = 78.30) -4.98 Br- THF (ε = 7.43) -4.97 DMF (ε = 37.22) -4.12 H2O (ε = 78.30) -3.95
Another point of interest was TCCP characteristics in solution, because little can be gained out of tetrel bonding activity when merely present in gaseous or crystalline state. To this end, geometry optimizations
Table 3: DFT calculations (gas phase) at the B3LYP-D3/def2-TZVP level of theory of monoatomic anionic interaction with 1a.
Anion Ionic radius11
(pm) Energy (kcal/mol) F- 133 -46.36 Cl- 181 -25.90 Br- 196 -20.41 I- 220 -15.20
29 of selected adducts were repeated by using an explicit solvent model. Three solvents were chosen with markedly different dielectric constants (ε): THF as apolar solvent (ε = 7.43), N,N-dimethylformamide (DMF) as polar aprotic solvent (ε = 37.22) and water as polar protic solvent (ε = 78.30). The resulting energies are listed in Table 4. Apparently, applying this model for solvation leads to less stable adducts by about 4 kcal/mol for 1,4-dioxane, 20 kcal/mol for Cl- and 15 kcal/mol for Br- (versus in vacuo
calculations). The energies are smaller with increased dielectric constant, although these differences are fairly small (~1-2 kcal/mol). The above implies that the kind of solvent is not an important factor in determining the binding enthalpy, but solvation itself is.
2.3.5. Comparative Atoms in Molecules and Energy Decomposition Analyses
To further scrutinize the geometry optimized adducts with TCCP derivatives, an ‘Atoms In Molecules’ (AIM) and ‘Energy Decomposition’ (ED) analysis was conducted of several adducts. To contextualize the results with TCCP molecules, similar analyses were performed with BMe3 as traditional Lewis acid,
water as hydrogen bond donor, Br2 as halogen bond donor and iodomethane as halogen and as carbon
bond donor. The Lewis bases considered were Cl– and 1,4-dioxane. The AIM and ED analyses were carried out after single point calculation at the B3LYP-D3/TZ2P level of theory (no frozen cores). The resulting data are collected in Table 5. E1 is the interaction energy after geometry optimization at the
DFT-B3LYP-D3/TZVP level of theory and E2 is the energy after a single point DFT-B3LYP-D3/TZ2P
calculation. The constituent components of E2 are Pauli repulsion (P.R.) and electrostatic, orbital, and
dispersion interaction (respectively E.I., O.I., and D.I.).
The Pauli repulsion is added up to all of the attractive components. The resulting total energy is compared to the energies obtained from calculations at the B3LYP-D3/def2-TZVP level of theory to ensure congruence, which in most instances turned out to be the case. In addition to energetic properties, bond critical point densities were calculated. The bond critical point (BCP) is a point situated along the bond path of an interaction in which the electron density reaches a minimum. Its value can be taken as a measure of the electron density in the bonding region and therefore of the covalent character of the bond.12 No entirely covalent bonds were studied however, but supramolecular interactions. That’s why
a stronger covalent character, and thus a high electron density in the BCP (ρb), is a good measure of the
strength of an interaction.
Table 5: Data of ED and AIM analyses. E1 = interaction energy after geometry optimization at the DFT-B3LYP-D3/TZVP level
of theory and E2 is the energy after a single point DFT-B3LYP-D3/TZ2P calculation. E2 has been decomposed into: P.R. =
Pauli Repulsion, E.I. = Electrostatic interactions, O.I. = Orbital Interactions, and D.I. = Dispersion Interaction. Energies in kcal/mol. ρbdenotes the BCP density (in 102 a.u.) in the BCP obtained after AIM analysis.
Adduct with: E1 E2 P.R. E.I. O.I. D.I. ρ
b
30 H2O -18.32 -17.07 13.26 -19.72 -9.91 -0.70 3.10 Br2 -40.90 -47.87 74.00 -52.42 -69.21 -0.24 6.17 BMe3 -26.31 -42.07 94.32 -64.57 -68.95 -2.87 7.65 CH3I @ C -15.91 -15.29 13.14 -13.89 -12.89 -1.65 1.62 CH3I @ I -13.17 -11.47 28.65 -18.17 -21.65 -0.30 2.81 1a -25.90 -24.10 24.6 -26.1 -18.53 -4.07 1.48
Lewis Base = 1,4-dioxane
H2O -6.93 -6.65 9.09 -9.33 -4.64 -1.77 3.04 Br2 -7.07 -6.81 16.39 -12.63 -8.11 -2.46 2.82 BMe3 -6.36 -8.53 21.31 -14.46 -9.01 -6.37 2.37 CH3I @ C -2.76 -2.36 2.80 -1.73 -0.93 -2.50 0.75 CH3I @ I -0.81 -1.83 20.27 -13.41 -5.64 -3.05 2.66 1a -11.27 -11.22 11.99 -11.62 -3.42 -8.17 1.02 Lewis acid = 1a F- -46.36 -42.52 36.43 -42.24 -34.20 -2.51 2.68 Cl- -25.90 -24.10 24.6 -26.1 -18.53 -4.07 1.48 Br- -20.41 -20.77 23.75 -23.83 -15.89 -4.80 1.23 I- -15.20 -16.28 23.45 -21.44 -12.96 -5.33 1.02
2.3.5.1.
Chloride Ion
Covalent character is determined by orbital overlap. Higher ρb
corresponds to stronger O.I. energies in most cases. It should be noted however, that ρb and O.I. energy are not entirely proportional to each
other. They are merely comparable, so a trend is not observed in all cases. The strongest calculated O.I.’s of chloride with Br2 and BMe3 also have
the highest values of ρb (see Figure 27). A deviation from this is
interaction of chloride with water. This hydrogen bonding interaction demonstrates relatively high ρb compared to the O.I. energy (3.10∙102 a.u.
and -9.91 kcal/mol respectively). Interaction of chloride with 1a is in the
order of hydrogen bonding with an energy of -17.07 – (-24.10) = 7,03 kcal/mol lower than that of hydrogen bonding, which indicates that tetrel bonding is stronger in this case. Interestingly, ρb of this
tetrel bond is lower than that of the hydrogen bond (see Figure 28). Of this interaction, E.I. turns out to be its main driver, followed by O.I., which is similar to hydrogen bonding.
Figure 27: Electron densities at BCP on bond paths in 100 x a.u. for interaction of chloride anion with Br2 and BMe3.
31 A different composition of attractive energies is seen in the interactions with Br2 and BMe3, where O.I. accounts for a larger part of E2,
instead of E.I. being the main contributor. E.I. here is not necessarily weak (Br2 = -52.42 kcal/mol, BMe3 = -64.57 kcal/mol), but the O.I. is just very
strong (Br2 = -69.21 kcal/mol, BMe3 = -68.95 kcal/mol). BMe3 is sp2
-hybridized because it has only five valence electrons, and a p-orbital is higher in energy than an sp2-orbital. Because of this the empty, electron
accepting p-orbital is very accessible, allowing it to overlap with a chloride orbital easily. Bromine has also got excellent orbital overlap with the chloride anion. In this case, chloride interacts with the σ*-orbital of bromine, which can also be characterized as a σ-hole, with its lone pair without any steric hindrance or other difficulties. Moreover, boron can be hypervalent like in BF4-, and trihalide anions are stable as
well, meaning an actual bond can be formed in these cases. So high O.I. energies may be expected here. Interaction with iodine in methyl iodide has stronger O.I. than E.I. as well. The chemical similarity of the halogen atoms allows them to readily align orbitals
resulting in a stable interaction. But when chloride tries to attack the methyl moiety in the same way, the attempt to overlap its lone pair orbital with carbons σ*-orbital is halted due to steric hindrance of the hydrocarbon bonds. This may explain why the interaction at the methyl moiety unlike the interaction at the iodine moiety has lower O.I. than E.I. energy (-12.89 kcal/mol < -13.89 kcal/mol). This lower energy corresponds with the lower value for ρb (see Figure 29).
2.3.5.2.
1,4-dioxane
Most of the listed interactions have weaker D.I. than E.I. and O.I., as can be expected from considerably electronically driven interactions. But interaction of 1a with 1,4-dioxane generates a higher D.I. than O.I. energy. The reason for this is the fact that London dispersion depends on the polarization volume of an entity.13 Such a volume is larger for a part of an polyatomic molecule like 1,4-dioxane than for a
monoatomic chloride anion, and this is confirmed by all of the D.I. energies for chloride interaction being lower than their 1,4-dioxane counterparts. Interaction between 1,4-dioxane and 1a generates the highest D.I. energy (-8.17 kcal/mol), as these two are the largest of the examined entities. Just like with chloride, interaction of 1,4-dioxane with 1a is in the order of hydrogen bonding with an energy of -6.65 – (-11.22) = 4.57 kcal/mol lower than the interaction of 1,4-dioxane with water, which indicates that tetrel bonding is more stable in this case too.
Figure 28: Electron densities at BCP on bond paths in 100 x a.u. for interaction of chloride anion with 1a and H2O.
Figure 29: Electron densities at BCP on bond paths in 100 x a.u. for interaction of chloride anion with CH3I at both sides.
32 Similarity between these two interactions ends at the D.I. energy. D.I. energy of 1a with 1,4-dioxane is high compared to O.I. (-3.42 kcal/mol), whereas D.I. energy of hydrogen bonding remains low compared to O.I. (-1.77 kcal/mol and -4.64 kcal/mol respectively). This is not surprising as hydrogen’s polarization volume is smaller than that of the (CN)2C–C(CN)2 moiety of 1a. Furthermore,
O.I. energy of 1a with 1,4-dioxane is also low compared to 1a with chloride (-3.42 kcal/mol and -18.53 kcal/mol respectively). This could be explained by orbital overlap in congruence with the favoured ‘vertical’ orientation of a Lewis acid with respect to a TCCP unit. A chloride anion has all of its lone pairs available in multiple directions as it is not bound to any other atoms,
while oxygen in 1,4-dioxane is bound to the (CH2)2-O-(CH2)2 backbone,
consequently presenting just two lone pairs with limited freedom of movement and orientation. And this limitation can make overlapping oxygen’s lone pair orbitals with the σ*-orbitals of the quaternary carbon atoms of the cyclopropane ring more difficult. Just like with chloride, the ρb value of the hydrogen bonding interaction between water and
1,4-dioxane is high compared to the O.I. energy (3.04∙102 a.u. and -4.64
kcal/mol respectively), and the ρb value of tetrel bonding between 1a and
1,4-dioxane is lower than that of hydrogen bonding (see Figure 30). Both phenomena can be explained in the same way as their chloride equivalents.
Besides 1a, methyl iodide too experiences substantially more D.I. than O.I. when interacting with 1,4-dioxane on the methyl side (-2.50 kcal/mol and -0.93 kcal/mol respectively), because four atoms (CH3) are considered here that contribute to the polarization volume. And just like with chloride,
the hydrocarbon bonds of the methyl moiety sterically hinder O.I.. Even more so for 1,4-dioxane interaction, as the interacting oxygen is bound to two atoms, which explains the low O.I. energy. No steric hindrance takes place at the iodide moiety and a higher O.I. energy is the result (-5.64 kcal/mol). The O.I. energies correspond with the ρb values (see Figure 31). Interaction with the iodide moiety
generates stronger D.I. than with the methyl moiety, but this is insignificant because the rest of the energy components of the interaction with the iodide moiety are substantially higher than those of the interaction with the methyl moiety with a factor of at least six. The total energy of the iodide interaction has a remarkably low value of -1.83 kcal/mol. The same interaction at the iodide moiety of chloride was driven by the halogens’ orbital compatibility, but this is not applicable when interacting with 1,4-dioxane. The same goes for interaction of 1,4-dioxane with bromine, where no high O.I. energy is observed either.
Figure 30: Electron densities at BCP on bond paths in 100 x a.u. for interaction of 1,4-dioxane with 1b and H2O.
Figure 31: Electron densities at BCP on bond paths in 100 x a.u. for interaction of 1,4-dioxane with CH3I at both sides.
33 Interaction of BMe3 with 1,4-dioxane generates E2 closest to
that of interaction of 1a with 1,4-dioxane (-8.53 kcal/mol and -11.22 kcal/mol respectively). The former exhibits the strongest attractive interaction energies of all interactions with 1,4-dioxane apart from the D.I. energy of the interaction with 1a (-8.17 kcal/mol). The high D.I. energy of BMe3 with 1,4-dioxane is a result of the large size of BMe3,
and the readily accessible empty p-orbital shows similar characteristics interacting with 1,4-dioxane as with chloride, seen by the high values of ρb and O.I. energy (2.37∙102 a.u. and -9.01 kcal/mol respectively).
The high E.I. energy may be attributed to boron’s low electronegativity. The P.R. is the strongest of all energy components of the interactions with 1,4-dioxane, but it is compensated by strong attractive interactions. So E2 of this interaction is still second highest.
2.3.5.3.
Dimethyl TCCP with Monoatomic Anions
Lastly, interactions of 1a with different halide ions were characterized. As previously stated, from fluor to iodine the ionic radius increases from 133 to 220 pm. This trend follows the D.I. energy, which increases from -2.51 kcal/mol for F- to -8.17 kcal/mol for I-; the O.I. energy, which
decreases from -34.20 kcal/mol for F- to -12.96
kcal/mol for I-; and the E.I. energy, which
decreases from -42.24 kcal/mol for F- to -21.44
kcal/mol for I-. A larger ionic radius signifies a
larger polarization volume, and this inevitably leads to stronger D.I.. Like previously calculated,
interaction energy diminishes when the anionic charge is smeared out over a larger region. Bigger atoms have outer electrons that are spread wider, which means they have less electrostatic effect on the electropositive domain of the σ-hole. Similarly, the orbitals of bigger atoms occupy more space reaching farther out. Consequently, lesser ability to overlap with the smaller orbitals of the quaternary carbon atoms of the cyclopropane ring results in weaker O.I. energy, and also in lower ρb (see Figure 33).
Because D.I. energy only marginally grows stronger, E2 decreases when anionic radius increases. Figure 32: Electron densities at BCP
on bond paths in 100 x a.u. for interaction of 1,4-dioxane with Br2 and BMe3.
Figure 33: Electron densities at BCP on bond paths in 100 x a.u. for interaction of 1a with halide ions from upper left to lower right: fluoride, chloride, bromide, iodide.
34
3. Conclusion & Future Prospects
Three TCCP molecules were successfully synthesized as reported by spectral data. 1a was obtained with a yield of 25%. By using three equivalents of malononitrile, a yield higher than in literature6 should have
been obtained.5 Unfortunately, this was not the case. This was most likely due to the thorough washing
of the residue. The same applies to 1b with an improvable yield of 34%. Despite thorough washing, 1a was contaminated with isopropylidenemalononitrile. This was apparent from 1H NMR and IR analysis.
Fortunately, this impurity posed no problem to the successful growth of analysable crystals of 1a. The less widely studied product, 1c, was obtained NMR-pure with a satisfying yield of 58%. Four successful co-crystals were grown: Of the two containing 1a, one contained THF and one 1,4-dioxane, another contained 1b with 1,4-dioxane and the last one 1c with 1,4-dioxane.
Firstly, it will be noted that a crystal of 1b out of a solution in THF was successfully obtained, but it contained no THF. Analysing the possibility of such an adduct would require careful crystallization. X-ray data showed distances between the σ-hole generating quaternary carbon atoms and the Lewis basic oxygen atoms in different adducts that indicated tetrel bond activity. Distances between
1a and THF were significantly shorter than the VdW radius sum. The ‘horizontal’ geometry THF
adopted in the crystal did not correspond to the energies resulting from the DFT calculations that favoured ‘vertical’ geometry. However, these calculations employed a gaseous environment. When packed in a crystal, several neighbouring molecules could disturb adopting the most favourable adduct orientation.
It was hypothesized that 1,4-dioxane could possibly bind to TCCP in a ditopic fashion, interacting with two σ-holes simultaneously. But instead, data showed that 1,4-dioxane adopted a ‘side-to-side’ orientation. This is another example of an orientation that does not accord to the gaseous interaction energies, which favour ‘vertical’ geometry at the ‘front’ position. Additionally, AIM calculations demonstrated that this targeted adduct conformation has an interaction energy more stable than that of its respective hydrogen bonding interaction. The orientation and its relatively short interatomic distances however, is an indication that ring breaking activity is a plausible possibility.
The question whether such an unwanted reaction/orientation could be prevented was answered by examining the interaction between 1b and 1,4-dioxane. This adduct showed proper σ-hole binding of the Lewis base 1,4-dioxane. Nevertheless, it still did not bind in a ditopic manner. The energy of the interaction between these two molecules in ‘vertical’ orientation at ‘front’ position had an even lower energy than that of the same interaction with 1a. And this resulted in ‘side-to-side’ binding. The fact that the ethyl groups sterically hinder the side positions is the bigger reason for binding to the σ-hole. At 1c only one side position is sterically hindered. This is because of the substituent ring bending off to one side. 1c formed an adduct with 1,4-dioxane, and also exhibited ‘side-to-side’ binding. Logically, the hindered side was subject to longer, and the unhindered side to shorter interatomic distances. This
35 corresponded with energy values out of DFT calculations of the interaction, producing two different values for ‘vertical’ orientation at ‘side’ position. Another hypothesis was that the oxygen in the substituent ring would exhibit affinity for a neighbouring molecule’s σ-hole. And this turned out to be correct according to the X-ray analysis of the crystal containing 1c and 1,4-dioxane. Infinite 1-D chains of 1c molecules are present in the crystal. The fact that the calculated energy of this interaction is the highest of the polyatomic DFT calculations further justifies this orientation.
Several attempts were made to grow crystals of the three tetrel donors with halide ions using tetrabutylammonium chloride and bromide without success. No crystals were obtained that contained ions. To this end, DFT and AIM energy calculations of such adducts were carried out. In gaseous phase, very strong interaction energies were calculated up to -46,36 kcal/mol for the interaction between 1a and fluorine, which amounted to 4 times the value of the strongest molecular interaction between 1a and 1,4-dioxane. But when the same was done employing an environment in which adducts were dissolved, way smaller interaction energies resulted, independent of solvent polarity. Especially for chloride and bromide. 1,4-dioxane was less affected by this change of surroundings. This explains the difficulty of growing TCCP-halogen crystals.
Aspiring to create stable tetrel bonding by means of a TCCP unit requires the right steric hindrance to prevent nucleophilic attack on the side, and a readily dissolvable Lewis base with strong affinity for the σ-hole. A lot of options have not been explored yet. Various bulky side groups have yet to be tried out, but this can be fulfilled in research to come. Ditopic binding also has yet to be achieved. Different di- or maybe even trivalent species can be reviewed as possible candidates for polytopic tetrel bonding. Further investigation of tetrel bonding and TCCP units might reveal possible applications such as Lewis acid catalysis, as TCCP units act as electron pair acceptors; or enzyme activation, as the relatively small TCCP units might fit in enzyme pockets. Future research holds the answers.
36
4. Experimental
4.1. General
All solvents and chemicals were purchased from commercial suppliers. Malononitrile was purified by distillation before use. NMR spectra were recorded on a Bruker DRX 500 (125.72 MHz for 13C), Bruker AMX 400 (100.62 MHz for 13C), Bruker DRX 300 (75.48 MHz for 13C) or on a Varian Mercury 300 (75.48 MHz for 13C) spectrometer at room temperature. The residual solvent was used as internal standard. Chemical shifts (δ) are given in part per million (p.p.m.). Mass spectra were recorded on an Advion (T)LC-MS expression LCMS mass spectrometer (with a TLC plate express and isocratic pump), using the Field Desorption (FD+(eiFi)) method. X-ray intensities were measured on a Bruker D8 Quest Eco diffractometer equipped with a Triumph monochromator ( = 0.71073 Å) and a CMOS Photon 50 detector at a temperature of 150(2) K. Intensity data were integrated with the Bruker APEX2 software.14
Absorption correction and scaling was performed with SADABS.15 The structures were solved using
intrinsic phasing with the program SHELXT.14 Least-squares refinement was performed with
SHELXL-201316 against F2 of all reflections. Non-hydrogen atoms were refined with anisotropic displacement
parameters. The H atoms were placed at calculated positions using the instructions AFIX 13, AFIX 43 or AFIX 137 with isotropic displacement parameters having values 1.2 or 1.5 times Ueq of the attached C atoms.
4.2. Synthesis
4.2.1. 3,3-dimethyl-1,1,2,2-tetracyanocyclopropane
To a 500 mL round-bottom flask ethanol (60 mL), malononitrile (5.9463 g, 90 mmol), NaOAc (0.7384 g, 9 mmol) and acetone (2.2 mL, 30 mmol) were added respectively. The mixture was colorless. It was heated under reflux at 78 °C for 30 minutes. The mixture turned orange. Then a 0.2 M solution of bromine (1.56 mL, 30 mmol) in water (150 mL) was added dropwise to the mixture using a dropping funnel. Precipitate formed and the mixture turned brown. The mixture was heated under reflux at 40 °C overnight. The dark brown mixture was cooled to room temperature, vacuum filtrated and thoroughly washed with water and ice cold ethanol respectively. The residue was a white powder.
White powder. Yield 1.2887 g (25%); δH (400 MHz; DMSO) 1.578 (s, 6 H, CH3); IR 2257.16 cm-1
(medium, CN); HR-FD-MS observed (calcd.) m/z for [C9H6N4]+ = 170.0532 (170.0592).
4.2.2. 3,3-diethyl-1,1,2,2-tetracyanocyclopropane
To a 500 mL round-bottom flask ethanol (60 mL), malononitrile (5.9514 g, 90 mmol), NaOAc (0.7378 g, 9 mmol) and 3-pentanone (3.2 mL, 30 mmol) were added respectively. The mixture was colorless. It was heated under reflux at 78 °C for 30 minutes. The mixture turned brown. Then a 0.2 M solution of
