Reconciling Electrostatic and n→π* Orbital Contributions in Carbonyl Interactions

Abstract Interactions between carbonyl groups are prevalent in protein structures. Earlier investigations identified dominant electrostatic dipolar interactions, while others implicated lone pair n→π* orbital delocalisation. Here these observations are reconciled. A combined experimental and computational approach confirmed the dominance of electrostatic interactions in a new series of synthetic molecular balances, while also highlighting the distance‐dependent observation of inductive polarisation manifested by n→π* orbital delocalisation. Computational fiSAPT energy decomposition and natural bonding orbital analyses correlated with experimental data to reveal the contexts in which short‐range inductive polarisation augment electrostatic dipolar interactions. Thus, we provide a framework for reconciling the context dependency of the dominance of electrostatic interactions and the occurrence of n→π* orbital delocalisation in C=O⋅⋅⋅C=O interactions.

Pt chemical shifts are reported in parts per million (ppm) from low to high field. 1 H and 13 C values are referenced to the literature values for chemical shifts of residual non-deuterated solvent, with respect to tetramethylsilane. 19 F is referenced externally to CFCl3 at 0 ppm and 195 Pt is referenced externally to K2PtCl6 at 0 ppm. Standard abbreviations indicating multiplicity are used as follows: bs (broad singlet), d (doublet), dd (doublet of doublets), m (multiplet), q (quartet), s (singlet), t (triplet), tt (triplet of triplets), J (coupling constant). All spectra were analysed using MestReNova (Version 11.0.0). NMR tubes, precision glassware and glass syringes were dried under vacuum before use. Deuterated solvents were stood over oven-dried activated 4 Å molecular sieves for a minimum of 24 h prior to use. Non-deuterated anhydrous solvents were used directly as commercially obtained anhydrous solvents, or were redistilled under reduced pressure from analytical-grade solvents. S3

Figure S1
Structures of newly synthesised balance series 1-X. An ortho-substituted control balance series, bearing non-carbonyl containing substituents was also synthesised. Balances S2-OMe and S3-Me were synthesised as previously reported. [S1] Characterisation data was in agreement with literature reported data.

Figure S2
Structures of previously reported balances for studying carbonyl-carbonyl interactions. Balance 2 was previously reported by Diederich [S2] and balance series 3-Y was previously reported by Raines. [S3] S4

Figure S4
Microwave-assisted synthesis of balances 1-H, 1-Me, 1-OMe and 1-NMe2. The synthesis of the starting iodoaryl compounds is outlined in Figure S3. 2-iodoacetophenone used for synthesising balance 1-Me was commercially available. Amide S7 was synthesised as previously reported. [S4] Characterisation data was in agreement with literature reported data.

Figure S5
Synthesis of control balance S1.

2-Iodobenzaldehyde S4
Manganese(IV) oxide (2.62 g, 30.1 mmol) was added to a solution of 2-iodobenzyl alcohol (1.00 g, 4.27 mmol) in CH2Cl2 (50 mL). The reaction mixture was heated to reflux overnight, filtered through celite and concentrated in vacuo to yield the product as a light yellow solid (

Methyl 2-iodobenzoate S5
To a solution of 2-iodobenzoic acid (1.00 g, 4.03 mmol) in MeOH (25 mL), was added concentrated H2SO4 (1 mL) dropwise. The reaction mixture was heated to reflux overnight, then allowed to cool to room temperature. The mixture was poured into water (200 mL) and extracted with CH2Cl2 (3 × 30 mL). The organic layers were washed with saturated aqueous NaHCO3 (3 × 30 mL) followed by brine (3 × 30 mL), dried over MgSO4, filtered and concentrated in vacuo, to yield the product as a light yellow oil (1.023 g, 97%).

2-Iodo-N,N-dimethylbenzamide S6
An aqueous solution of dimethylamine (40%, 40 mL) was added dropwise to 2-iodobenzoyl chloride (3.00 g, 11.26 mmol) at 0 °C. The reaction mixture was stirred at 0 °C for 30 min, quenched with a saturated aqueous solution of K2CO3 (100 mL), and extracted with CH2Cl2 (3 × 30 mL). The combined organics were dried over anhydrous MgSO4, filtered and concentrated in vacuo. The oily residue was dried under high vacuum overnight to yield the product as a light yellow solid (3.010 g, 97%).

S2.3 Conformer assignment using NMR spectroscopy
Conformer assignment for balance series 1 and control balances S1, S2 and S3 was performed by a combination of NMR spectroscopy techniques ( 1 H, 13 C, COSY, 1 H-13 C HSQC, 1 H-13 C HMBC and 1 H NOESY). The main characteristics for distinguishing between the two conformers were: a) 1 H-13 C long-range through-bond correlation between the major formyl proton and the major quaternary carbon in the X-substituted ring, and between the minor formyl proton and the minor quaternary carbon in the F-substituted ring (marked in blue in Figure S24), resulting in respective cross-peaks on the 1 H-13 C HMBC spectrum that allowed for an unambiguous determination of the major and minor conformer. b) Nuclear Overhauser Effect resulting in the through-space coupling between the major formyl proton and the major proton on the F-substituted aromatic ring, and between the minor formyl proton and the minor aromatic proton on the X-substituted ring (marked in pink in Figure S24), resulting in respective cross peaks on the NOESY spectrum. These were treated as a confirmation of the assignment from a), as occasionally the minor cross-peak was too weak to be seen, or multiple cross-peaks would be visible due to the NOESY experiment timescale being comparable to the conformer interconversion timescale.

Figure S24
1 H-1 H and 1 H-13 C couplings used for "O" and "H" conformer assignment (closed and open conformers respectively).

S2.4 Crystal Structures of 1-H and 1-Me
Balances 1-H and 1-Me were crystallised from an Et2O/n-hexane solvent system by slow vapour diffusion technique, and the structures determined by X-ray diffraction are shown in Figure S25 and Figure S26.

Figure S25
The asymmetric unit of balance 1-H, with displacement ellipsoids at the 50% probability level.

Figure S26
The asymmetric unit of balance 1-Me, with displacement ellipsoids at the 50% probability level.
Balance 1-H crystallised as an asymmetric unit containing two distinct polymorphs a and b ( Figure S25). The carbonyl groups of these structures seem to display a sheared parallel orientation. The measured distances between the aldehyde carbon and the formyl oxygen are longer (2.930 Å in structure a and 3.035 Å in structure b, Figure S27A) are shorter than the sum of van der Waals radii of carbon and oxygen (3.22 Å). The interaction angles are larger (123.43° and 131.67°, Figure 29a), than that seen for a typical of a Bürgi-Dunitz trajectory (100-110°). There is also no significant pyramidalisation of any of the carbonyl carbons.
Balance 1-Me crystallised as the "O" (closed) conformer ( Figure S26), with a distance of 2.784 Å between the acetyl carbon and the formyl oxygen ( Figure S27B, C), shorter than the sum of van der Waals radii of carbon and oxygen (3.22 Å). The shortest contact, however, arises between the acetyl  proton (slightly acidic) and the formyl oxygen (2.599 Å). This may constitute an additional attractive interaction stabilising the "O" (closed) conformer of 1-Me. No significant pyramidalisation of any of the carbonyl carbons was seen ( Figure S29).

Figure S27
Interaction distances and angles in A balance 1-H (polymorphs a and b) and B, C balance 1-Me.
The extended crystal structure reveals the presence of weak and long intermolecular CH-O hydrogen bonds in the solid state of 1-Me ( Figure S28).

Figure S28
Intermolecular interactions involving the carbonyl groups in the extended crystal packing of balance 1-Me (dashed lines).

Figure S29
Example of pyramidalisation measurements on balance 1-Me for A the ortho-substituent carbonyl carbon and B the formamide carbonyl carbon.

Table S1
Selected distances and angles of the C=OC=O interactions determined from the crystal structures of 1-H (polymorphs a and b) and 1-Me.

Table S3
Atomic coordinates (× 10 4 ) and equivalent isotropic displacement parameters (Å 2 × 10 3 ) for balance 1-H. U(eq) is defined as one third of the trace of the orthogonalised U ij tensor.

Table S10
Atomic coordinates (× 10 4 ) and equivalent isotropic displacement parameters (Å 2 × 10 3 ) for balance 1-Me. U(eq) is defined as one third of the trace of the orthogonalised U ij tensor.

Table S13
Hydrogen coordinates (× 10 4 ) and isotropic displacement parameters (Å 2 × 10 3 ) for balance 1-Me.    Since rotation around the formamide bond is slow on the NMR timescale, discrete peaks corresponding to the "O" and "H" conformers (closed and open conformers, respectively) are observed. Thus, integration of the conformer peaks provides direct access to the conformational equilibrium constant, K, which can be used to determine the conformational free energy difference,

Figure S31
Plots of Gexp in different solvents vs. ESP over the centre of the substituted ring (Table S24, DFT/B3LYP/6-31G*). Error bars have been omitted for clarity.

Figure S32
Correlation of Gexp vs. E (30). Grey data points correspond to measurements in methanol and ethanol (outliers).

Figure S36
Correlation of Gexp vs. s parameter, where s and s are solvent H-bond donor and acceptor parameters, respectively.

Figure S40
Correlation of Gexp vs. Hansen solubility parameters P + HB (dipole interactions and H-bond interactions).

S3.3 Application of the Hunter solvation model
The data can be further analysis based on the Hunter solvation model [S1, S7a, S7b] (Figure S41 and Equation S1), where two parameters describe solvent polarity (s and s, describing H-bond donor and acceptor ability of the solvent respectively).

Figure S41
Hunter's solvation model adapted for the formamide balances. [S1, S7a, S7b] EH and EO correspond to intramolecular interactions in the "H" and "O" conformer (or open/closed conformers respectively). s and s correspond to H-bond donor and acceptor constants of the solvent respectively. O and O correspond to to H-bond donor and acceptor constants of the O conformer respectively, and H and H correspond to H-bond donor and acceptor constants of the H conformer respectively.

Equation S1
describes the Hunter solvation model as adapted for formamide balances. [S1, S7a, S7b] Eexp is the intramolecular interaction energy, and  and  are the differences in the H-bond donor and acceptor constants between the "O" and "H" conformers (also referred to as closed and open conformers respectively) (understood globally for each conformer, i.e. Boltzmann-averaged).

Calculated conformational free energy (Gcalc) errors
Calculated conformational free energy (Gcalc) errors were calculated using the following equation: where E,  and  are standard multilinear regression fitting errors, as output by Origin v9.0.

Figure S42
Correlation of Gexp (E in gas phase calculations (  Pyr, values of Gexp (E) and Eexp from literature data. [S1] Gexp errors have been omitted for clarity.

Figure S43
Plot of Gexp vs. Gcalc for all studied balances in up to thirteen solvents each. Error bars have been omitted for clarity.

Figure S44
Plot of conformational energy differences between full molecular balances and molecular balance series 1-X. Calculated using B3LYP/6-31G*.

S4.1 Geometry minimisation and calculated conformational free energies
Full molecular balance structures shown in Figures S1 and S2 were minimised in both the open and closed conformations using either the B3LYP or B97X-D methods and basis sets 6-31G* in Spartan '14. Frequency calculations were performed on all minimised structures, which confirmed no imaginary frequencies. The resulting gas-phase energies and corresponding energy differences, Ecalc in each conformer are reported.

S59
Only minimal differences were observed between minimised structures using either B3LYP or B97X-D methods ( Figure S46).

Figure S47
Minimised geometries of the open and closed conformers of molecular balances in series 3-Y.

Table S20
Distance and interaction angles between carbonyl donor and acceptor groups in the closed conformer for molecular balances in series 1-X, 2 and 3-Y (Figures S1 and S2). Calculated using Spartan '14 DFT/B3LYP/6-31G* except for balance 2, which is the X-ray geometry.

Figure S48
Summary of the O···C distances observed in the minimised structures of balances 1-X, 3-Y and the X-ray structure of 2.

Figure S49
Minimised structures of molecular balances A 1-H, B 1-Me, C 1-OMe and D 1-NMe2 showing dipoles for the donor (teal) and acceptor (pink) carbonyls. The orientation of the ortho-carbonyl acceptor group is different in 1-H than the other molecular balances resulting in repulsive local dipoles. Structures minimised using Spartan '14 DFT/B3LYP/6-31G*.

S4.2 Electrostatic Surface Potentials
Electrostatic Surface Potentials were obtained using Spartan '14 software on minimised balance structures (DFT/B3LYP/6-31G*). The ESP errors were determined as standard deviation from four ESP measurements (two measurements on the position of interest on both faces of the aromatic ring).

Table S24
ESP calculated for X-substituted rings. Values measured over the substituted carbon or over the centre of the ring. a very large error due to asymmetric geometry and large ESP differences between both sides of the ring

S4.3 fiSAPT analysis
The PSI4 software [S13] was used to perform functional group intramolecular symmetry adapted perturbation theory (fiSAPT) calculations using the SAPT0 methodology. [S14] Geometries from minimisations performed in Spartan '14 were used as an input for fiSAPT calculations. The 6-311G*, jun-cc-pVDZ or aug-cc-pVQZ basis sets were used. The output gave the energetic contributions of electrostatics, induction, exchange-repulsion and dispersion to give a total SAPT interaction energy prediction.

Figure S50
Defining the geometry for intramolecular fiSAPT calculations for balance series 1-X and 3-Y. Colours define the interacting functional groups, while black bonds define the linking groups. Balance 2 was treated with a comparable intermolecular SAPT analysis (see Figure S54).

S67
A further fiSAPT analysis was performed on the B97X-D/6-31G* minimised structures of the closed conformations of balance series 1-X that included dispersion correction. This resulted in only minor changes to the components and total SAPT interaction energy compared to the B3LYP-minimised structures (Figure S52, c.f. Figure S51).

Table S26
Energetic contributions of the interactions seen in the closed conformer of molecular balance 1-H. Calculated using PSI4 SAPT0 with various basis sets using geometry minimised balances from Spartan '14 DFT/B3LYP/6-31G*.

Table S27
Energetic contributions of the interactions seen in the closed conformer of molecular balance 1-H. Calculated using PSI4 SAPT0 with various basis sets using the geometry from the crystal structure.

Figure S54
A Structure of the two interacting fragments used in the SAPT analysis of carbonyl···carbonyl interactions isolated from Diederich's molecular balance 2. The fragment geometry was identical to the known crystal structure of balance 2. B Energetic contributions of the interaction terms in the depicted molecular fragment. Calculated using PSI4 SAPT0/6-311G*.

Table S32
Energetic contributions of the fragment depicted in Figure S54 dervied from the known X-ray structure of balance 2. Calculated using PSI4 SAPT0/6-311G*.

Table S34
Energetic contributions of the interactions seen in the closed conformer of molecular balance 3-NO2. Calculated using PSI4 SAPT0 with various basis sets using geometry minimised balances from Spartan '14 DFT/B3LYP with various basis sets.

Figure S56
Energetic contributions of the interactions seen in the isolated fragments of balances A 1-H, B 2 and C 3-H as the O···C distance was varied (insets). The C=O···C angle was maintained at as determined from the minimised structure of balances 1-X and 3-Y and the X-ray structure of 2, A 1-Me 100.79°, B 2 109.33°, C 3-H 99.78°. Calculated using PSI4 SAPT0/6-31G*, raw data in Tables S35-S37).

Table S35
Energetic contributions of the interactions seen in the isolated components of balance 1-Me as the O···C distance was varied ( Figure S56). The O···C=O angle was maintained at 100.79° as determined from the minimised structure of the whole balance. Calculated using PSI4 SAPT0/6-31G*.

Table S36
Energetic contributions of the interactions seen in the isolated components of balance 2 as the O···C distance was varied ( Figure S56). The O···C=O angle was maintained at 109.33° as determined from the X-ray structure of the whole balance. The interaction is decomposed into electrostatics, exchangerepulsion, induction and dispersion to give a total SAPT predicted interaction energy. Calculated using PSI4 SAPT0/6-31G*.

Table S37
Energetic contributions of the interactions seen in the isolated components of balance 3-H as the O···C distance was varied ( Figure S56). The O···C=O angle was maintained at 99.78° as determined from the minimised structure of the whole balance. The interaction is decomposed into electrostatics, exchangerepulsion, induction and dispersion to give a total SAPT predicted interaction energy. Calculated using PSI4 SAPT0/6-31G*.

S4.4 Natural Bond Orbital analysis
Natural bond orbital (NBO) analyses were performed using a single point energy calculation using Gaussian 09 Revision E.01 [S15] at DFT/B3LYP/6-31G* using the geometry from the already minimised full molecular balance from Spartan '14 (Section S4.1) to generate an NBO output. The NBO output was then inputted to NBO 6.0 [S16] to obtain second-order perturbation theory output energies, images created using JmolNbo Visualization Helper Version 2.0 and Jmol or Chemcraft.

Table S388
NBO second-order perturbation energies observed in minimised structure for balance 1-Me and balance series 3-Y. Other balances in series 1-X and balance 2 did not show NBOs. The sum total lone pair n→* interaction energy is also given. Calculated using single point energy calculation using Gaussian 09 Revision E.013 at DFT/B3LYP/6-31G*. Second-order perturbation theory output energies calculated using NBO 6.0 Table S39 NBO second-order perturbation energies observed in minimised structure for balance 3-NO2. Calculated using single point energy calculation using Gaussian 09 Revision E.013 at DFT/B3LYP with different basis sets. Second-order perturbation theory output energies calculated using NBO 6.0

Figure S577
Diagram of predicted electron delocalisation from the lone pair of the oxygen formyl into the antibonding * bond of the C=O group of compound 1-Me. 1-Me has the shortest O···C distance of the balances in series 1-X (Table S20). Calculated from the crystal structure geometry (distanceO···C = 2.784 Å) using a using Gaussian '09 and NBO 6.0. NBO analysis on the B97X-D/6-31G* minimised structures also revealed n→* electron delocalisation to occur 1-Me worth 6.3 kJ mol −1 .

Figure S588
A, NBO corresponding to n→* electron delocalisation from both lone pairs of the amide donor in molecular balances from series 3-Y. B Plot of predicted stabilisation energies from NBO second-order energy perturbations from n→* electron delocalisation against experimentally measured conformational free energy difference determined by 1 H NMR spectroscopy (CDCl3, 400 MHz, 298 K). Experimental values taken from reference S3. NBO calculations performed using Gaussian '09 and NBO 6.0 B3LYP/6-31G*.

Table S40
Distance dependence on NBO second-order perturbation energies in molecular balance 1-Me. The O···C=O angle was maintained at 100.8°, as seen in the minimised structure (Table S20)

Table S43
Angle dependence on NBO second-order perturbation energies in molecular balance 2. The O···C distance was maintained at 3.333 Å, as found in the known X-ray structure (

Figure S59
O···C=O angle dependence of the sums of the 2 nd order perturbation energies corresponding to n→* electron delocalisation of both lone pairs in carbonyl-carbonyl dimers modelling the equivalent interactions hosted within molecular balances A 1-Me B 2, and C 3-H. Energies were calculated using NBO6.0 (see Tables S41, S43, S45).

Figure S60
Distance dependence of the sums of the 2 nd order perturbation energies corresponding to n→* electron delocalisation of both lone pairs in carbonyl-carbonyl dimers modelling the equivalent interactions hosted within molecular balance series 3-Y and balance 1-Me. Simulated trends relate to data presented in Table S40 and S44, where the O···C distance was artificially varied starting from the B3LYP/6-31G* minimised geometry of balance 1-Me and 3-H respectively (all other angles/distances kept constant). Energies were calculated using NBO6.0).

Figure S60
compares the sums of the 2nd order perturbation energies corresponding to n→* electron delocalisation of both lone pairs in carbonyl-carbonyl dimers across balance series 3-Y and balance 1-Me. The simulated trends relate to data presented in Table S40 and S44, where the O···C distance was artificially varied starting from the B3LYP/6-31G* minimised geometry of balance 1-Me and 3-H respectively (all other angles/distances kept constant).
The deviation of the data points for balance series 3-Y (purple) from the simulated trend reflect electronic/steric differences arising from the different substituents. Likewise, the deviation of the data point for balance 1-Me (orange) from the simulated data reflect differences between the crystal structure geometry and simulated geometry based on B3LYP/6-31G* minimisation. Figure S59 demonstrates that the O···C=O angle has minimal effect on the 2nd order perturbation energies.
Given this variation as a result of precise structure/substitution, the NBO analysis of balance 1-Me fits remarkably well with the trend seen for balance series 3-Y, with the O···C distance being the dominant factor for determining the 2nd order perturbation energies.

S4.5 Molecular orbital analysis
A detailed molecular orbital analysis was performed in this study in which the orbital energies of open and closed conformers of the molecular balances were compared. To enable identification and pairing of molecular orbitals found in the open and closed conformers it was necessary to avoid orbital splitting arising from the canonical resonance forms of the aromatic electrons (that were not involved in the interactions of interest).
The minimised full molecular balance structures of series were subsequently used to generate simplified balance structures from series 1-X of the type shown in Figure S60, in which the 4-fluorophenyl moiety was replaced with a proton with a N-H bond length of 1.012 Å. Fragment series frag 1-X was found to reflect the full balance series 1-X energies well ( Figure S62).
For molecular balance series 3-Y, the full molecular balance structures were used for the molecular orbital analyses.

Figure S61
Simplified fragment of molecular balance series 1-X used for molecular orbital analyses to generate fragment series frag 1-X.