Noncovalent interactions play a significant role in a wide variety of biological processes and bio-inspired species. It is, therefore, important to have at hand suitable computational methods for their investigation. In this paper, we report on the contribution of dispersion and hydrogen bonds in both stacked and T-shaped catechol dimers, with the aim of delineating the respective role of these classes of interactions in determining the most stable structure. By using second-order Møller–Plesset (MP2) calculations with a small basis set, specifically optimized for these species, we have explored a number of significant sections of the interaction potential energy surface and found the most stable structures for the dimer, in good agreement with the highly accurate, but computationally more expensive coupled cluster single and double excitation and the perturbative triples (CCSD(T))/CBS) method.

Nowadays, there is a general consensus about the primary role played by noncovalent interactions, in particular those involving aromatic rings, in molecular, life, and materials sciences. In addition to being responsible for key biological processes that range from base stacking in deoxyribonucleic acid (DNA) [

In the past few years, catechol has attracted increasing attention as a precursor of bio-inspired materials [^{mod}, has been automated and extended to the optimization of the orbital exponents of

Here, the MP2^{mod} method is applied to the catechol dimer in the gas phase. First, MP2^{mod} accuracy is validated against high-quality CCSD(T)/CBS predictions, purposely carried out for a number of selected geometries of catechol dimers. Next, MP2^{mod} is employed in the exploration of the catechol’s interaction potential energy surface (IPES), with the aim of finding the optimal structure of the dimer by a comparison of different possible arrangements. This allows us to investigate the different roles played by HB and π-stacking interactions in the dimer formation. Incidentally, it might also be of interest, following Wheeler group’s suggestions [

The catechol dimer has also been studied at the DFT level by Estévez et al. [

The full geometry optimization of the catechol monomer has been performed by DFT, at the B3LYP/

As far as the intermolecular energy is concerned, reference CCSD(T)/CBS calculations have been carried out on catechol dimers following the protocol adopted in previous works [

The difference Δ_{CC-MP2}

The MP2 energy in the CBS limit,

Finally, the CCSD(T)/CBS interaction energy,

All energies were corrected for the basis set superposition error (BSSE) with the standard counterpoise (CP) correction [

The MP2^{mod} exponent optimization was performed by means of the _{geom}^{mod} calculations were carried out with the 6-31G** basis set, and the exponents of the ^{mod} calculations, the CP correction was applied to take care of the basis set superposition error.

Finally, to better compare with the results reported by Estévez et al. [

All CCSD(T), MP2, MP2^{mod} and DFT calculations were carried out with the Gaussian09 software package [

After geometry optimization, the catechol monomer is planar with the two hydroxyl hydrogens pointing in the same direction (see

Based on the results recently achieved for several heteroaromatic dimers, where stacked and T-shaped (TS) conformers where found to be the most stable, four starting arrangements have been set up by placing the two monomers at different distances and relative orientation. Namely, the face-to-face (FF, _{1}, _{2}, ^{m}^{od} best exponents were determined as follows: starting from each of the four selected conformations, a set of dimer arrangements was created by displacing one monomer along a selected coordinate R, defined as the line connecting the centers of the two rings, as shown in the insets of ^{m}^{od} level, employing the basis set recently optimized by us for quinhydrone [^{mod} calculations. The starting exponents of the standard 6-31G** basis set are 0.80 for

The resulting MP2^{mod} curves are displayed in ^{mod} method to the study of the catechol dimer. According to both CCSD(T)/CBS and MP2^{mod} results, the most stable structure is the TS_{2} one (around −5.0 kcal/mol), with the minimum at a slightly smaller value of R (5.4 Å), with respect to the similar TS_{1} conformer (5.6 Å), which is in turn almost as stable (≈−4.0 kcal/mol) as the antiparallel stacked conformer (AFF, −3.8 kcal/mol). Among the two stacked conformations, FF and AFF, the second one is more stable, in agreement with the repulsive interaction between the OH dipoles in the FF form.

Due to its importance, the stacked arrangement has been studied with some care as a function of the ring–ring distance

In order to gain a deeper insight into the orientation dependence of the stacking forces in the catechol dimer, taking advantage from the low computational cost of the MP2^{mod} method, we can explore different sections of the catechol IPES. For instance, in

This simple picture is consistent with the minimum of −5.2 kcal/mol (^{mod} interaction energy is −4.7 kcal/mol. This subtle difference can find a rationale at a closer look of the molecular structure, embracing Wheeler’s idea that unexpected substituents effects can be explained by considering their direct interaction with the neighboring cloud of the other ring [^{mod} energy, which is repulsive in both cases, increases by 1 kcal/mol, in going from

As shown in _{1} and TS_{2} geometries.

In order to verify this assumption, the MP2^{mod} computational feasibility has been exploited once again to explore an additional IPES section, related to the TS conformers and shown in _{1} in _{2} arrangement shown in the right bottom panel of

The above described competition between stacked and TS geometries misses although another player, which could significantly alter the delicate balance between them. In fact, apart for a small contribution to the stability of the ^{mod} level, starting from four different conformations (see _{d} and TS_{u}, respectively). Notice that the latter is very similar to that taken from crystallographic data and investigated by Estévez et al. [

Dimer formation does not result in large changes in the internal geometry of each catechol monomer. Bond lengths within each monomer change by less than 0.03 Å and the backbone remains planar. For each ring, only one hydroxyl hydrogen moves out of plane, establishing OH–O or OH–π interactions, while the other O–H bond remains nearly coplanar with the ring, due to the formation of an intramolecular OH–O HB with the closest oxygen atom (in _{d}, which becomes II, while OH–π weak HBs guide the hydroxyl rotation and are prevalent in AFFD, which becomes I. Although less stable, the last optimized conformer III, is characterized by a single hydroxyl rotation, which allows the insurgence of a HB (green dashed line in

The interaction energies for the four final structures are reported in ^{mod} values and their CCSD(T)/CBS counterparts is very good, especially considering that these latter geometries are outside the MP2^{mod} training set, while the computational advantage of using MP2^{mod} with small basis sets is apparent from the last three columns. Surprisingly, the MPW1B95 functional severely underestimates the reference CCSD(T)/CBS interaction energies, yielding, in the present case, only a qualitative correct description, at least according to the protocol provided in [

Finally, it is interesting to investigate the different HB contributions in the two most stable conformations I and II. This can be done by performing a rigid scan of the rotation angle

In this paper, we have reported our study of the intermolecular landscape of a catechol dimer with a two-fold interest. On the one hand, noncovalent interactions, and especially those involving aromatic rings, govern many biological processes and it is, therefore, of basic importance to reach a good comprehension of the different role that the various forces play in specific systems. On the other hand, noncovalent interactions are still a challenging benchmark for standard computational methods, hence, it can be significant to exploit dedicated approaches.

Catechol is well known to be a precursor of many bioinspired materials and it is, therefore, a good candidate to investigate on the interplay between dispersion interactions, essentially due to aromaticity, and strong (OH–O) or weak (OH–π) HBs, settled by the hydroxyl substituents. The employed MP2^{mod} computational route consists in MP2 calculations with a small 6-31G** basis set, in which the exponents of the polarization functions are suitably modified. This has been done through a validation procedure based on the comparison with the highly accurate CCSD(T)/CBS calculations, resulting in new exponents for polarization functions on carbon (0.27), hydrogen (0.36), and oxygen (0.34).

Within the IPES sections explored, two minima were identified, held together by a network of stacking, OH–O, and OH–π interactions, whose relative weight has been analyzed in some detail. The two catechol units tend to aggregate in stacked conformation, which eventually result more stable than the TS ones, thanks to their ability to establish strong and weak HBs.

A final remark should be made concerning the effects that solvation can have in these systems. Despite most computational approaches designed for noncovalent interactions only focus on two isolated molecules, we are aware that water might affect the results and change the picture that we report here (see, for instance, [

The present article has been conceived by the four authors during a nice dinner. Computations have been shared to be more efficient as it was with the writing. All the authors have therefore equally contributed to the final work.

The authors declare no conflict of interest.

_{2}S

(_{1} and (_{2}. C: Cyan; H: White; O: Red.

Comparison between the ‘best exponent’ and CCSD(T)/CBS for the interaction energy profiles obtained by displacement of the four structures shown in

MP2^{mod} results for the stacked configurations. (

(^{mod} level. (

Stacked displaced (

(^{mod} level. The IPES section was sampled by varying the angle _{1} geometry shown in the right top panel of

MP2^{mod} geometry optimization starting from the displaced AFF (AFFD), TS_{d}, and TS_{u} conformations (top panels). The corresponding optimized structures, I, II, III, and IV, are displayed in the bottom row. The rotated hydroxyl groups are evidenced in the top panel with a blue arrow, while the atoms involved in OH–O and OH–π interactions are connected in the bottom panels by green and orange dashed lines, respectively.

MP2^{mod} scans of the HOCC dihedral (

Interaction energies, in kcal/mol, for the four optimized conformations shown in ^{mod}, CCSD(T)/CBS and MPW1B95/6-311++G(2^{®} Xeon CPU are also given for an evaluation of the computational cost of the different methods.

Geometry | Energies (kcal/mol) | CPU Time (min) | ||||
---|---|---|---|---|---|---|

MP2^{mod} |
CCSD(T)/CBS | MPW1B95 | MP2^{mod} |
CCSD(T)/CBS | MPW1B95 | |

I | −10.7 | −11.1 | −8.1 | 27 | 50,640 | 145 |

II | −12.4 | −12.6 | −8.3 | 25 | 49,740 | 180 |

III | −5.3 | −5.7 | −2.2 | 27 | 50,820 | 79 |

IV | −6.1 | −7.3 | −5.7 | 18 | 51,720 | 142 |