Structural and Energetic Properties of Acetonitrile–Group IV (A & B

May 22, 2014 - This plumbing system was equipped with a concentric dual− deposition flange6 that allowed for simultaneous deposition of mutually rea...
2 downloads 0 Views 4MB Size
Download PDF
Article pubs.acs.org/JPCA

Structural and Energetic Properties of Acetonitrile−Group IV (A & B) Halide Complexes Heather M. Helminiak, Robin R. Knauf, Samuel J. Danforth, and James A. Phillips* Department of Chemistry, University of WisconsinEau Claire, 105 Garfield Avenue, Eau Claire, Wisconsin 54702, United States S Supporting Information *

ABSTRACT: We have conducted an extensive computational study of the structural and energetic properties of select acetonitrile−Group IV (A & B) tetrahalide complexes, both CH3CN−MX4 and (CH3CN)2−MX4 (M = Si, Ge, Ti; X = F, Cl). We have also examined the reactivity of CH3CN with SiF4, SiCl4, GeCl4, and TiCl4, and measured low-temperature IR spectra of thin films containing CH3CN with SiF4, GeCl4, or TiCl4. The six 1:1 complexes fall into two general structural classes. CH3CN−TiCl4, CH3CN−TiF4, and CH3CN−GeF4, exhibit relatively short M−N bonds (∼2.3 Å), an intermediate degree of distortion in the MX4 subunit, and binding energies ranging from 11.0 to 13.0 kcal/mol. Conversely, CH3CN−GeCl4, CH3CN− SiF4, and CH3CN−SiCl4, are weakly bonded systems, with long M−N distances (>3.0 Å), little distortion in the MX4 subunit, and binding energies ranging from 3.0 to 4.4 kcal/mol. The structural features of analogous 2:1 systems resemble those of their 1:1 counterparts, whereas the binding energies (relative to three isolated fragments) are roughly twice as large. Calculated M−N potential curves in the gas phase and bulk, dielectric media are reported for all 1:1 complexes, and for two systems, CH3CN− GeF4 and CH3CN−SiF4, these data predict significant condensed-phase structural changes. The effect on the CH3CN−SiF4 potential is extreme; the curve becomes quite flat over a broad range in dielectric media, and at higher ε values, the global minimum shifts inward by about 1.0 Å. In bulk reactivity experiments, no reaction was observed between CH3CN and SiF4, SiCl4, or GeCl4, whereas CH3CN and TiCl4 were found to react immediately upon contact. Also, thin-film IR spectra indicate a strong interaction between CH3CN and TiCl4, yet only weak interactions between CH3CN and GeCl4 or SiF4 in the solid state.

■

INTRODUCTION For some time, we have been interested in the effects of bulk, condensed-phase media on the structural properties of nitrile donor−acceptor complexes. The novel structural chemistry of these systems is best illustrated by some examples from previous studies of nitrile−boron trifluoride complexes.1 For both HCN−BF32 and CH3CN−BF3,3 large differences between the gas-phase and solid-state structures have been observed. Specifically, the B−N distance of HCN−BF3 contracts from 2.47 Å in the gas phase4 to 1.65 Å in the crystal,2 and in the process, the BF3 subunit distorts from a near planar to neartetrahedral geometry. For CH3CN−BF3, the changes are less substantial, but the B−N bond does contract from 2.10 to 1.65 Å in transition from the gas phase to the solid state.1,3 More recently, we have used a combination of computations and Xray crystallography to predict changes for FCH2CN−BF3 and ClCH2CN−BF3 that rival those of HCN−BF3,5 and lowtemperature IR spectra suggest a comparable effect for BrCH2CN−BF3.5 Perhaps even more remarkable is the degree to which inert, bulk condensed phases, even noble gas solids, can affect the structures of these systems. For CH3CN−BF3, frequencies of key, structurally sensitive vibrational modes shift systematically between the gas phase, inert matrix media, and the solid state.6,7 Furthermore, these shifts are even systematic across these inert media and indicate that the B−N bond compresses in response to subtle increases in the charge stabilizing ability of the matrix © 2014 American Chemical Society

environment. Just recently, a similar trend was noted for the singly halogenated analogues of CH3CN−BF3 (e.g., FCH2CN− BF3), but the dynamic range of the shifts was even greater.8 Computations have provided insight into the underlying mechanism of these structural changes. Using continuum models to represent solvation in inert media, we have constructed hybrid B−N potential curves composed of the gas-phase electronic energy and the electrostatic component of the solvation free energy.9 Structural change is reflected when the minima shift inward on these curves as the dielectric constant increases. As such, the key feature of any mediumsensitive system is a flat donor−acceptor potential, in which the energies at short (bonding) and long (nonbonding) distances are comparable, such that a difference in solvation energy can shift the global minimum. In general, complexes tend to have greater solvation energies at shorter distances, due to the increase in polarity that accompanies formation of the donor− acceptor bond.9 Thus, the systems that are most prone to condensed-phase structural effects tend to have potentials with the global minimum at a relatively long donor−acceptor distance, and a gradual energy rise toward shorter distances. Such is the case with FCH2CN−BF3 and ClCH2CN−BF3.8 Received: November 23, 2013 Revised: May 9, 2014 Published: May 22, 2014 4266

dx.doi.org/10.1021/jp4115207 | J. Phys. Chem. A 2014, 118, 4266−4277

The Journal of Physical Chemistry A

Article

In this paper, we report the structural and energetic properties of a series of complexes formed between acetonitrile and Group IV (A & B) tetrahalides (CH3CN−MX4: M = Ti, Ge, Si; X = F, Cl). For the 1:1 complexes. We present calculated structures, frequencies, and M−N potential curves in the gas phase and in bulk, dielectric media. We also present calculated structures and binding energies for the 2:1 (CH3CN:MX4) complexes. We also report observations regarding the direct reactivity of a few of these acids with CH3CN, as well as low-temperature IR spectra of thin-films composed of CH3CN and SiF4, TiCl4, or GeCl4. Overall, there is a broad range of interaction strengths across this series of complexes, and in turn, a great deal of structural variability from system to system. Moreover, at least two of these systems show at least some potential for medium-induced structural change.

At this point, we are extending this work to the acetonitrile complexes of the Group IV (A & B) halides (CH3CN−MX4: M = Ti, Ge, Si; X = F, Cl), and our interest is 2-fold. For one, we seek additional systems akin to the nitrile-BF3 complexes that exhibit medium-dependent structural properties, so that we can further understand the physical origins of this behavior. However, nitrile−MX4 complexes are of particular interest because of the potential for nanotechnology applications, because both component molecules have bonds that are opposite of and parallel to the donor−acceptor linkage. This configuration would facilitate the incorporation of these structural motifs (i.e., −CN−MX3−) into larger structures (e.g., a polymer or macromolecular assembly), and ultimately, if these M−N bonds were tunable, they could be manipulated to exert a force within the larger structure, parallel to the donor− acceptor bond. Previous computational studies of other nitrogen donor− MX4 complexes10,11 have revealed systematic trends in the stability of the complexes with respect to the metal and the halogen. Specifically, donor−acceptor bond strength tends to increase with size of the metal (Sn > Ge > Si), and decrease with the size of the halogen (F > Cl > Br).11 In addition, there seems to be a preference for the amines to coordinate in an axial configuration,10,11 whereas an equatorial geometry is more stable for the imines.11 Furthermore, in a few of these systems, the donor−acceptor potentials exhibit unusual features. For example, two equilibrium bond distances have been identified for H3N−SiCl4, with lengths of 2.181 Å (global minimum) and 3.534 Å (secondary minimum).11 Also, the N−Si potential energy curve of pyridine−SiF4 is broad and flat, with no bonded minimum.11 However, the N−Si potential of bipyridine−SiF4 exhibits a barrier at intermediate distances, between the longer, secondary minimum (∼3.5 Å) and the shorter, global minimum (∼2.1 Å). In addition, the potentials of the GeX4 complexes are similar to their silicon analogues.11 Furthermore, there have been several matrix-isolation IR studies of nitrogen donor−MX4 complexes.12−17 For example, vibrational frequencies of H3N−SiF4 were observed in argon and nitrogen matrices,12 and the measured frequencies of the Si−F asymmetric stretching (vaMX) bands, at 952 and 854 cm−1, do indicate a reasonably strong donor−acceptor bond in the complex. This vaMX mode is triply degenerate in free MX4 compounds18 but splits and shifts to lower frequencies when a Lewis base coordinates to the MX4 subunit. As such, the analogous bands in the methyl-substituted amine−SiF413 complexes exhibit slightly larger shifts than H3N−SiF4,12 which reflects their greater base strength. Nonetheless, for all the amine−SiF4 complexes, and the pyridine complex,15 the vaMX mode is shifted substantially from the frequency in free SiF4 (1022 cm−1).18 Nitrile complexes, including CH3CN−SiF4 and CH3CN−GeF4, have also been observed via matrix-isolation IR spectroscopy, and it was inferred via product yields that CH3CN−SiF4 was a weaker complex than CH3CN−GeF4.16 a Moreover, the shift of the observed vMX frequency for −1 16 CH3CN−SiF4 (974 cm ) is much less than that for the analogous amine or pyridine complexes, and thus indicates a correspondingly weaker donor−acceptor interaction, in accord with base strength. A goal of the work described in this manuscript is to provide gas-phase and solid-state frequencies to establish a context for the previous matrix-IR work, and in turn, illustrate any structural changes that may be induced by bulk condensed-phase media.

■

METHODS Computational Methods. Computations were performed using Gaussian 0319 (Revision E.01) and Gaussian 0920 (Revisions A.02, B.01, and D.01). For equilibrium structure calculations, geometry optimizations were usually conducted within the point group with the maximum possible symmetry (exceptions are noted in the text below), employed an ultrafine grid, and convergence criteria were set according to the “opt=tight” option. We did explore the performance of numerous computational methods in an attempt to identify a preferred computational method suitable for all six complexes. The effort focused initially on CH3CN−TiCl4 but was extended to the CH3CN−SiF4 in the later stages of the work. To determine a suitable method for predicting IR spectra, we first predicted the frequencies of TiCl4 with X3LYP, B3PW91, BPW91, B3LYP, BLYP, mPWPW91, mPW1PW91, B97-2, B98, and MP2, using the 6-311+G (2df,p) basis set.21 Of these, X3LYP was determined to be our preferred model, mainly because it was far superior in terms of agreement experimental frequencies of TiCl418,22 (RMS error = 1 or 4 cm−1). We note that the most significant frequency shifts we observed were in the CH 3 CN/TiCl4 experiments, so accurate frequency predictions for CH3CN−TiCl4 and (CH3CN)2−TiCl4 are of the utmost importance. We also note that the experimental frequency data for TiCl4 seem to be the more reliable among the MX4 compounds studied here. For example, the RMS difference between the two sets of data18,22 for SiF4 is 12 cm−1. Nonetheless, we subsequently extended the scope of our validation study after gaining access to a few more modern density functionals available via Guassian09 (M06,23 ω-B97X,24 and ω-B97X-D25). None of these methods predicted the TiCl4 frequencies as well as X3LYP, however, as RMS errors ranged from 11 to 15 cm−1. As a secondary validation measure, we did compare the DFT and MP2 structure results for CH3CN−TiCl4 as well as CH3CN−SiF4, using a handful of the functionals noted above (X3LYP, B3PW91, mPW1PW91, M06, ω-B97X, and ω-B97XD, of which the first four have performed well in our previous work on related systems5,8), again with the 6-311+G(2df,p) basis set. For CH3CN−TiCl4, all perform reasonable well, most yielding agreement with about 0.01 Å for the Ti−N distance (X3LYP, B3PW91, mPW1PW91, and M06), whereas the others were within about 0.02 Å (ω-B97X and ω-B97X-D). For CH3CN−SiF4, a weaker complex (see below), there was somewhat less consistency; mPW1PW91, M06, ω-B97X, and ω-B97X-D predicted Si−N distances within about 0.05 Å of the MP2 result, whereas X3LYP and B3PW91 predicted values 4267

dx.doi.org/10.1021/jp4115207 | J. Phys. Chem. A 2014, 118, 4266−4277

The Journal of Physical Chemistry A

Article

Bulk-Phase Reactivity. Bulk-phase reactions were carried out between four pairs of the substances: CH3CN/SiF4, CH3CN/SiCl4, CH3CN/GeCl4, and CH3CN/TiCl4. For CH3CN and SiF4, about 2 mL of liquid CH3CN was added to a dry Schlenk tube (with a Teflon stopcock side arm), which was then purged with a steady flow of nitrogen gas to displace ambient air. Next, SiF4 was flowed through the tube, over the CH3CN, resulting in no temperature change and no other signs of reaction. Subsequently, the tube was sealed and cooled to facilitate the formation of a solid product by storing it for several days in a series of freezers with temperatures ranging from 0 to −77 °C. Analogous experiments involving CH3CN with TiCl4, SiCl4, and GeCl4 differed somewhat, because both components were liquids. In these experiments, about 1 mL of liquid CH3CN was added to a dry Schlenk tube, and the acid was added in a dropwise manner through a septum using a glass syringe. Low-Temperature Infrared Spectra. All spectra were recorded at 1 or 2 cm−1 resolution, using a Nicolet Avatar 360 FTIR, and 200 scans were averaged for each spectrum. Samples were prepared using a previously described matrix-isolation IR apparatus6 based on a Cryomech ST15 optical cryostat evacuated by a Balzers TPU 240 turbo molecular pump. Temperature was measured and controlled using a Scientific Instruments 9600-1 temperature controller and a single silicon diode located at the end of the second refrigeration stage. For these experiments, the cold head of cryostat apparatus resided on a moveable cart, which was wheeled into the spectrometer for each measurement. This plumbing system was equipped with a concentric dual− deposition flange6 that allowed for simultaneous deposition of mutually reactive gas mixtures (e.g., CH3CN and TiCl4). This design kept the gas mixtures separate until approximately 1 in. from the cold KBr window, and we used it even in cases when the component gases were not reactive (e.g., CH3CN and GeCl4). As such, we prepared separate mixtures containing either CH3CN or MX4 (SiF4, GeCl4, or TiCl4) in nitrogen for each experiment. Gas mixtures were prepared in 2 L bulbs on a glass manifold evacuated with glass diffusion pump, and concentrations were approximately 0.25% for GeCl4 and TiCl4 experiments and 0.25% or 1% for SiF4 experiments, and the CH3CN concentration either was equal to that of the acid or was in 2-fold excess. Nitrogen was used exclusively as a carrier gas. Thin-film spectra were obtained by codepositing the two samples onto the cold (∼60 K) KBr sample window, at flow rates ranging from 2 to 6 mmol/h, which were controlled using Granville-Phillips 203 variable leak values. At this temperature, the nitrogen carrier did not accumulate on the sample window, and only a thin film of MX4 and CH3CN was deposited. In some instances, the relative composition of the film was varied by adjusting the flow rates. Each experiment involved several successive depositions lasting five to 30 min, and spectra were recorded after each deposition.

within about 0.1 Å. Nonetheless, given that the frequency predictions are key for interpreting our experimental results and that we sought a single method to make comparisons across this series of complexes, we emphasize X3LYP/6-311+G(2df,p) values for structure and frequency predictions. However, to offer some indication of uncertainty, and because they are presumably more reliable for the weaker complexes, we also report MP2/6-311+G(2df,p) results for key structural parameters and discuss them at various points in the text (see below). An extended set of binding energies was also calculated for CH3CN−TiCl4 using MP2, MP3//MP2, MP4 (DQ)//MP2, MP4 (SDQ)//MP2, CCD//MP2, and CCSD//MP2 (using the 6-311+G (2df,p) basis set in each case). On the basis of the performance of MP2 relative to higher-level post-Hartree−Fock methods, we concluded that MP2 energies would suffice for the other complexes (MP2 overestimates the value by 0.7 kcal/mol relative to CCSD, whereas the other post-HF methods underestimate by 0.3−0.8 kcal/mol). Potential energy curves along the M−N distance were computed in a pointwise manner, in 0.1 Å steps ranging from 1.6−3.5 Å for all complexes, though for the weaker systems, CH3CN−SiCl4, CH3CN−SiF4, and CH3CN−GeCl4, the curves were extended to 4.0 Å. All degrees of freedom besides the M− N distance were allowed to optimize (within the C3v symmetry constraint) at each point. For CH3CN−TiCl4, we examined the potentials for all DFT and post-HF methods noted above, and initially we chose mPW1PW91 as our preferred method. However, after subsequently considering the M06, ω-B97X, and ω-B97X-D functionals, and applying them to the CH3CN− SiF4 potential, we chose ω-B97X-D as our preferred method. This was based in its performance along the interior wall of the potentials, and good agreement with the CCSD/6-311+G(2df,p) energies along the CH3CN−SiF4 potential, and we will examine these data in more detail below. We also examined the effects of dielectric media on the potentials using the polarized continuum model (PCM),26 with various ε-values of 2.0, 5.0, and 20.0. To construct these curves, we added the electrostatic component of the free energy of solvation to the gas-phase electronic energies at each point along the potential.9 When these calculations were performed in Gaussian 03, we set the “OFAC” parameter to 0.55, and the “RMIN” parameter to 0.85.19 All degrees of freedom aside from the M−N distance were optimized for the PCM/ω-B97X-D/6-311+G(2df,p) potentials, though we note that in several instances (perhaps 20% of the points in total), the symmetry constraint was abandoned during the optimization, but the resulting structures exhibited only trivial differences from the C3v point group, and the curves exhibit no significant discontinuities that result from this. Materials. For thin-film IR experiments: CH3CN (99.5%, VWR) was transferred to a sample tube via an in-house solvent purification system and was subjected to several freeze−pump− thaw cycles immediately prior to use. In some instances, a small amount of P2O5 was added to the CH3CN sample to remove water. SiF4 was obtained from Matheson and used without further purification. GeCl4 and TiCl4 were obtained from Sigma-Aldrich; both were purified via freeze−pump−thaw cycles before use. Nitrogen gas (99.9999%, Praxair) was used (as a carrier gas) without further purification. For bulk reactivity experiments, CH3CN (Aldrich), SiF4 (Matheson), TiCl4 (Aldrich), GeCl4 (Aldrich), and nitrogen (standard purity, Praxair) were used without further purification.

■

RESULTS AND DISCUSSION Bulk-Phase Reactivity. In the CH3CN/SiF4 reactivity experiments, no solid reaction product was observed under any conditions, and no noticeable temperature change or other sign of reaction was observed when the reactants were first exposed to one another. In CH3CN/GeCl4 and CH3CN/SiCl4 experiments, not only was there no sign of reaction upon adding GeCl4 or SiCl4 to the 4268

dx.doi.org/10.1021/jp4115207 | J. Phys. Chem. A 2014, 118, 4266−4277

The Journal of Physical Chemistry A

Article

included in Figure 1 in parentheses, are somewhat shorter, by as much as 0.15 Å (CH3CN−TiF4). Nonetheless, the Ti−N bond lengths values are near the sum of the corresponding covalent radii27 (2.31 Å), though the Ge−N distance is 0.3−0.4 Å longer than the analogous sum for Ge and N (1.91 Å). However, the Ge−N distance in CH3CN−GeF4 is much shorter than the sum of the Ge and N van der Waals radii (Ge− N = 3.36 Å).28 Furthermore, the N−M−X angles of CH3CN− TiCl4, CH3CN−TiF4, and CH3CN−GeF4 were 82.5°, 81.2°, and 80.9° respectively (via X3LYP), and these values lie between those corresponding to the tetrahedral (70.5°) and trigonal bipyramidal (90°) limits and thus indicate a moderate degree of distortion in the MX4 subunit. Conversely, the equilibrium structures of CH3CN−SiF4, CH3CN−GeCl4, and CH3CN−SiCl4 exhibit relatively long M− N distances of 3.032, 3.622, and 3.858 Å, respectively (via X3LYP/6-311+G(2df,p)). In these cases, the MP2/6-311+G(2df,p) distances are notably shorter, by about 0.1 Å in the case of CH3CN−SiF4, and by several tenths of an Ångstrom for the SiCl4 and GeCl4 systems, which is an indication that X3LYP is underestimating the strength of the weak interactions. Regardless, the N−Si bond distance in CH3CN−SiF4 is significantly longer than the sum of the Si and N covalent radii (1.83 Å),27 though still about 0.3 Å shorter than the sum of the Si and N van der Waals radii (3.35 Å).28 In the case of CH3CN−GeCl4 and CH3CN−SiCl4, the M−N distances (via MP2) compare favorably to the sums of the corresponding and van der Waals radii (Ge−N = 3.36 Å, Si−N = 3.35 Å).28 Consequently, the MX4 subunits of all three of these complexes are only slightly distorted, with the N−M−X angles of 73.1°, 72.1°, and 71.3°, for CH3CN−SiF4, CH3CN−GeCl4, and CH3CN−SiCl4, respectively. These structural results indicate weak interactions in these systems and are consistent with observations made in the bulk reactivity experiments, in which we found no stable product formed from SiF4 and CH3CN, and that both GeCl4 and SiCl4 were immiscible with CH3CN. Overall, there are some clear trends in structural properties across this series of complexes. Both titanium complexes have short M−N bonds and exhibit a moderate degree of angular distortion in the acid subunit. The germanium complexes differ; the GeF4 complex resembles the TiX4 complexes, whereas the GeCl4 complex has a long Ge−N distance and a neartetrahedral MX4 subunit. As a whole, the fluorine-containing complexes tend to have shorter bonds than their chlorine-based counterparts, but this difference is most extreme for the GeX4containing complexes. The SiF4 and SiCl4 complexes also exhibit long M−N distances, much like the GeCl4 complex. Thus, the overall trend in the M−N distances as a function of the M atom is Ti < Ge < Si. Binding Energies. We report binding energies for these complexes from MP2/6-311+G(2df,p) calculations. In our past work, we have found that MP2 binding energies exceed those from higher-level post-Hartree−Fock methods by 2−3 kcal/ mol29 but, in this case, however, the difference is much less, about 0.7 kcal/mol, thus it appears that MP2 provides only a slight overestimate. The binding energies for the short-bond complexes, CH3CN−TiF4, CH3CN−GeF4, and CH3CN− TiCl4, were 13.0, 11.4, and 11.0 kcal/mol, respectively, and indicate that these complexes are somewhat intermediate in terms of strength. These are a few kcal/mol larger than that for CH3CN−BF3 (8.7 kcal/mol),29 yet much less than that for CH3CN−BH3 (22.6 kcal/mol).29 Also, they are about half as large as the H3N−BF3 complex,29,30 a benchmark example of a

Schlenk tube containing the CH3CN, but also the substances failed to mix: distinct layers were apparent. When TiCl4 was added to CH3CN, the substances reacted immediately to form solids (yellow and white in color), and the tubes became noticeably warm. The 1:1 (CH3CN:TiCl4) mixture developed a small amount of yellowish solid at the bottom of the tube and a white solid near the top. The 2:1 (CH3CN:TiCl4) reaction mixture had slightly less yellow solid and more white solid than the 1:1 sample, whereas the 1:2 (CH3CN:TiCl4) reaction mixture produced almost no white solid and a larger amount of the yellow solid. These observations suggest that the yellow-colored product contains less CH3CN than the white product, perhaps indicating that the yellow solid is CH3CN−TiCl4 and that the white solid is (CH3CN)2−TiCl4. However, we were unable to determine the specific composition of either product, because we could not isolate a single crystal suitable for an X-ray structure, and standard chemical analyses were not possible because the samples decomposed quickly when exposed to room air. Equilibrium Structures. As a starting point for the computational studies, we considered four possible isomeric structures for each complex: axial-eclipsed (C3v), axial-staggered (C3v), and two possible equatorial configurations (Cs). The use of the terms “axial” or “equatorial” implies a trigonal bipyramidal geometry about the M atom (i.e., the result of a fully formed M−N bond), which is not always the case, but we use these labels regardless of the extent of the distortion within the MX4 subunit. As such, the axial orientation has the nitrile C−N bond oriented 180° from an M−X bond, and the two corresponding C3v geometries have the methyl C−H bonds either eclipsed or staggered relative to other (potentially equatorial) M−X bonds. In the equatorial orientation, the nitrile bisects the X−M−X angle of the would-be axial X’s, and in the two possible Cs-symmetry orientations, a C−H bond is either eclipsed or staggered relative these M−X bonds. According to MP2/6-311+G(2df,p) energies, the eclipsed conformer was the minimum energy structure for the axial configuration, albeit by less than 0.1 kcal/mol in each instance, indicating that the methyl groups are essentially free rotors in these systems. Furthermore, with the exception of CH3CN− SiCl4, no imaginary frequencies were obtained for any of the axial-eclipsed structures, whereas for all the axial-staggered structures, the methyl torsional frequencies were imaginary. In the case of CH3CN−SiCl4, the torsional frequency was imaginary for both axial structures (via X3LYP). However, the MP2 value for the torsional frequency of the axial-eclipsed form was found to be real. As for the equatorial geometries, all stable structures were found to be several kcal/mol higher in energy. Furthermore, in some instances, optimizations initiated from these geometries would not retain that overall geometry during the optimization and were rearranged to either the axialeclipsed isomer or a halogen-bonded configuration in which an M−X bond was directed toward the nitrile with an M−X−N angle near 180°. As such, we based our subsequent computational analyses on the axial-eclipsed structures and report structural parameters for those immediately below. Minimum-energy, equilibrium structures of all six complexes obtained via X3LYP/6-311+G(2df,p) calculations are displayed in Figure 1, and we note that, to some extent, there are two distinct structural types. The M−N bonds in CH3CN−GeF4, CH3CN−TiCl4, and CH3CN−TiF4 are relatively short, with X3LYP/6-311+G(2df,p) distances of 2.274, 2.312, and 2.348 Å, respectively. We do note that the MP2 values, which are 4269

dx.doi.org/10.1021/jp4115207 | J. Phys. Chem. A 2014, 118, 4266−4277

The Journal of Physical Chemistry A

Article

Figure 1. X3LYP/6-311+G(2df,p) structures for CH3CN−MX4 (M = Si, Ge, Ti; X = F, Cl) complexes with MP2/6-311+G(2df,p) binding energies. MP2/6-311+G(2df,p) values for the M−N distances and X−M−N angles are included in parentheses. Differences between MP2 and X3LYP are not significant for the other structural parameters. See text for discussion.

relatively soft inner wall in comparison to most other methods, for both potentials, which could unduly favor the prediction of medium effects8,9 (though the difference is rather slight for CH3CN−TiCl4). Ultimately, we chose ω-B97X-D as a preferred method for M−X potentials, due to its performance for both test complexes. For CH3CN−TiCl4, the ω-B97X-D curve was similar to the mPW1PW91 curve, but somewhat higher in energy along the inner wall. For CH3CN−SiF4, we found that ω-B97X-D predicted a firmer inner wall than mPW1PW91, and tracked very well with the MP2 and MP2// CCSD curves in terms of overall shape (Figure S1B, Supporting Information). Moreover, the points on the ω-B97X-D/6311+G(2df,p) curve agree quite well with those on the CCSD/ 6-311+G(2df,p)//MP2/6-311+G(2df,p) curve, in terms of overall energy relative to the SiF4 and CH3CN fragments (Figure 2F). Figure 2 displays the M−N bond potentials for all six complexes, using our three preferred computational methods, including ω-B97X-D/6-311+G(2df,p) as well as the corresponding X3LYP and MP2 curves, which were included because those methods were used for equilibrium structure predictions. For CH3CN−TiCl4 (Figure 2A) and CH3CN− SiF4 (Figure 2F), the CCSD//MP2 data points are also

fairly strong donor−acceptor system. The binding energies of the “long-bond” complexes, CH3CN−SiF4, CH3CN−GeCl4, and CH 3 CN−SiCl 4, were 4.4, 4.0, and 3.0 kcal/mol, respectively, comparable to a moderately strong hydrogen bond or one of the weaker nitrile−BF3 complexes (e.g., HCN− BF3).31 N−M Bond Potentials. As noted above, to determine a preferred, across-the-board method to model the M−N bond potentials in these systems, we examined both DFT and postHF curves for CH3CN−TiCl4 and made an extensive performance comparison. A selection of these curves for both complexes, plotted with their respective minima set to zero to facilitate a comparison of shape, is included as Supporting Information (Figure S1). Initially, due to its performance on CH3CN−TiCl4, we chose mPW1PW91, because the energies were intermediate between the values from post-HF methods, and other, older DFT methods. However, once we had noted the peculiar features in the curves for the weaker systems and had gained access to more recently developed functionals via Gaussian 09, we broadened our consideration of methods to include M06,23 ω-B97X,24 and ω -B97X-D25 and also extended this comparison study to the CH3CN−SiF4 complex. At that point, we discovered that mPW1PW91 tended to predict a 4270

dx.doi.org/10.1021/jp4115207 | J. Phys. Chem. A 2014, 118, 4266−4277

The Journal of Physical Chemistry A

Article

Figure 2. M−N distance potentials obtained via X3LYP, ω-B97X-D, and MP2 with the 6-311+G(2df,p) basis set: (A) CH3CN−TiCl4, (B) CH3CN− TiF4, (C) CH3CN−GeCl4, (D) CH3CN−GeF4, (E) CH3CN−SiCl4, and (F) CH3CN−SiF4. CCSD//MP2 points are included in (A) and (F). The corresponding energy of the CH3CN and MX4 fragments is set to zero in each graph.

0.5 kcal/mol at distances ±0.3 Å from the minimum, which occurs near 3.6 Å. This curve also exhibits a shelf-like feature along the inner wall, but there is no secondary minimum. It is also worth noting that the minimum energy points on the ωB97X-D curves lie between the MP2 and X3LYP values for the M−N distances of CH3CN−SiCl4 and CH3CN−GeCl4 noted in Figure 1. The curve for CH3CN−SiF4 indicates that it would be most prone to condensed-phase structural change. Near the minimum, the shape of the potential resembles those for CH3CN−SiCl4 and CH3CN−GeCl4, with an energy rise of about 0.4 kcal/mol at distances ±0.3 Å from the global minimum, near 3.0 Å. However, the energy increases by only about only about 5 kcal/mol at 2.0 Å, just beyond a typical N− Si single bond distance.27 On the CH3CN−SiCl4 and CH3CN−

included for reference as well, but the energy values discussed below refer to the ω-B97X-D values. For the TiX4 systems, the energy rises somewhat abruptly about the minima on the respective curves. Specifically, for CH 3 CN−TiCl4 and CH3CN−TiF4, the energy rises 6−7 kcal/mol upon moving 0.3 Å inward from the from minima (∼2.3 Å) and 1.5−2 kcal/ mol upon moving 0.3 Å outward. The CH3CN−GeF4 potential is considerably flatter, especially toward longer Ge−N distances; the energy rises about 2.5 kcal/mol upon moving inward 0.3 Å from the minimum (∼2.3 Å), and a mere 1.5 kcal/ mol at a distance of 0.3 Å outward. The CH3CN−GeCl4 potential is a bit flatter than the CH3CN−GeF4 curve, with the energy rising only 0.2−0.5 kcal/mol at distances ±0.3 Å from the minimum energy point of 3.4 Å. The potential of CH3CN− SiCl4 is comparably flat, and the energy also increases by 0.2− 4271

dx.doi.org/10.1021/jp4115207 | J. Phys. Chem. A 2014, 118, 4266−4277

The Journal of Physical Chemistry A

Article

Figure 3. M−N potentials for CH3CN−MX4 complexes in bulk dielectric media: (A) CH3CN−TiCl4, (B) CH3CN−TiF4, (C) CH3CN−GeCl4, (D) CH3CN−GeF4, (E) CH3CN−SiCl4, and (F) CH3CN−SiF4. The gas-phase curves are ω-B97X-D/6-311+G(2df,p) energies. The potentials in dielectric media (ε > 1) are hybrid curves; the sum of the gas-phase energy and electrostatic component of the solvation free energy (PCM/ω-B97XD/6-311+G(2df,p)). See text for discussion. The energy of the isolated, gas-phase fragments is set to zero.

0.1 Å. In the case of CH3CN−GeF4, the effect is more notable, as the minimum shifts inward by about 0.2 Å, from 2.3 Å in the gas phase to 2.1 Å with an ε value of 5.0 or greater. In the curves of the weaker complexes, CH3CN−GeCl4, CH3CN−SiCl4, and CH3CN−SiF4, the preferential stabilization along the inner wall is quite notable and is due to greater solvation energies is at shorter M−N distances.9 However, the shallow, global minima on these curves tend to shift outward by 0.1−0.2 Å in the dielectric media. Regardless, on the CH3CN− GeCl4 curves with ε ≥ 5.0, a shelf-like feature akin to that noted above for CH3CN−SiCl4 appears between 2.0 and 2.5 Å. In addition, the feature in the CH3CN−SiCl4 curve becomes more distinct with increasing ε, and there is actually a local minimum on the ε = 20.0 curve near 2.1 Å. For CH3CN−SiF4, the curves

GeCl4 curves, the corresponding energy change is about +10 kcal/mol. Effects of Dielectric Media. As in our previous work,8,9 medium effects were explored by constructing hybrid M−N bond potentials summing the gas-phase electronic energy (ωB97X-D/6-311+G(2df,p)) with the electrostatic component of the solvation free energy (PCM/ω-B97X-D/6-311+G(2df,p)). Curves for each complex in dielectric media with ε-values of 2.0, 5.0, and 20.0 are displayed in Figure 3, and each gas-phase potential is included for reference. In general, the energies decrease as the dielectric constant increases, and in some cases, the global minimum shifts inward to some extent via interactions with the bulk medium. This effect is minor for the TiX4 complexes, in which the maximum shift is only about 4272

dx.doi.org/10.1021/jp4115207 | J. Phys. Chem. A 2014, 118, 4266−4277

The Journal of Physical Chemistry A

Article

Figure 4. Infrared spectra of low-temperature (60 K) thin films containing (A) CH3CN and/or GeCl4, (B) CH3CN and/or SiCl4, (C) CH3CN and/ or TiCl4.

become incredibly flat over a broad region in the dielectric media, and moreover, a second minimum occurs near 2.0 Å for ε =5.0 and ε = 20.0. This is the global minimum on the ε = 20.0 curve, and though the specific implications for structural change are unclear with such a flat potential, a shift of this magnitude is unprecedented our work. Infrared Spectra and Frequencies. We measured infrared spectra of thin films containing CH3CN and GeCl4, SiF4, or TiCl4, at 60 K, and focused our attention primarily on the

region of the triply degenerate asymmetric stretching modes (vaMX) of the MX4 subunit. For the CH3CN/SiF4 film, this mode occurs at or below about 1000 cm−1, and for the GeCl4and TiCl4-containing films, it lies at or below about 500 cm−1. As such, spectra for these regions are displayed in Figure 4. Above, we illustrated via a comparison of frequencies for nitrogen donor−SiF4 complexes12,16 that the frequency of the vaMX mode shifts systematically to lower energy and splits as a Lewis base coordinates to the MX4 subunit. Thus, it is an 4273

dx.doi.org/10.1021/jp4115207 | J. Phys. Chem. A 2014, 118, 4266−4277

The Journal of Physical Chemistry A

Article

Figure 5. Equilibrium structures (X3LYP/6-311+G(2df,p)) and binding energies (MP2/6-311+G(2df,p)) for the cis isomers of (CH3CN)2−MX4 complexes (M = Ti, Ge, Si; X = F,Cl). Some parameters were omitted for clarity. H−C−H angles across all complexes ranged from 109.4° to 110.2°. N−M−X angles (X3LYP) were as follows: N−Ti−Cl angles were 81.5° and 88.4°, N−Ti−F angles were 78.8° and 85.5°, N−Ge−Cl angles were 71.7° and 70.9°, N−Ge−F angles were 80.1° and 85.6°, N−Si−Cl angles were 71.8° and 69.4°, and N−Si−F angles were 71.9° and 71.7°.

the solid-state environment does not promote or enhance the donor−acceptor interaction between CH3CN and GeCl4. Moreover, the computational data are quite consistent with this; the shape of the computed Ge−N potential of CH3CN− GeCl4 does not change in dielectric media, at least in the region near the global minimum. The spectra of the CH3CN/SiF4 films (Figure 4B) are similar to those for CH3CN/GeCl4 films. The lone, prominent feature is the vaMX of the SiF4 moiety, which is centered at about 991 cm−1 in both the CH3CN/SiF4 and pure SiF4 spectra. Also, the spectrum in the CN stretching region for the CH3CN/SiF4 film nearly matches that for the pure CH3CN film. Though we do see an additional, weak absorption feature in the CN-stretching region at 2273 cm−1, between the components of the CNstretching doublet (2294 and 2253 cm−1), the doublet frequencies themselves are much stronger (>10-fold) and do not shift whatsoever in the presence of SiF4. Thus, the spectra seem to indicate that the medium does not affect the interaction between CH3CN and SiF4, but this notion is somewhat at odds with the predicted response to a dielectric continuum discussed above, in which the potential flattens significantly toward shorter N−Si distances, and substantial shift in the global minimum occurs at high ε-values.

indicator of the extent of the N−M donor−acceptor bonding in these films. However, these spectra exhibit very broad absorption features, and in systems that are prone to only weak interactions, and lack distinct band shifts, the extent of association between the acid and base can be unclear.32 In the spectra of the CH3CN/GeCl4 films (Figure 4A), the lone, prominent feature is the vaMX band of the GeCl4 moiety, which is centered at about 450 cm−1. These bands are essentially coincident in the pure GeCl4 and CH3CN/GeCl4 films, and the only difference in the latter case is a slight broadening and blue shift (from 453 to 456 cm−1), which suggests no significant interaction between the acid and base components in the film. Furthermore, we note that in the CN stretching region, the CH3CN/GeCl4 spectrum looks nearly identical to that of the pure CH3CN film. These observations are quite consistent with the lack of reactivity noted above, as well as the computational results that predicted a weak interaction between GeCl4 and CH3CN. According to X3LYP calculations, the predicted vaMX frequency in GeF4 is 444 cm−1, and only a slight shift/splitting (to 446 and 431 cm−1) is predicted for the analogous mode in gas-phase CH3CN−GeCl4. These predicted frequencies lie within the line width of the observed vaMX band for the film; thus, the spectra indicate that 4274

dx.doi.org/10.1021/jp4115207 | J. Phys. Chem. A 2014, 118, 4266−4277

The Journal of Physical Chemistry A

Article

of the CN-stretching doublet in CH3CN. The intensities of the 2319 and 2265 cm −1 peaks do vary somewhat with concentration, as the former is favored by excess CH3CN, whereas the latter is favored by excess TiCl4. Properties of the 2:1 Complexes (CH3CN)2−MX4. Given the evidence for different coordination ratios in the CH3CN/ TiCl4 film spectra and bulk reactivity experiments, as well as an interest in how the explicit inclusion of a second CH3CN would affect the coordinate bonding in these systems, we calculated X3LYP/6-311+G(2df,p) and MP2/6-311+G(2df,p) structures of the 2:1 CH3CN:MX4 complexes, i.e., (CH3CN)2−MX4. These results are displayed in Figure 5, with MP2/6311+G(2df,p) binding energies, i.e., the energy difference between the complex and the three isolated fragments. We initially considered both cis and trans (C2h) isomers of these complexes, initially with the highest possible symmetry (C2v or C2h), and ultimately determined in each instance that the cis (C2v) forms were lower in energy. For the weaker complexes, (CH3CN)2−GeCl4, (CH3CN)2−SiCl4, and (CH3CN)2−SiF4, we considered two general types of starting structures, a highly symmetric (C2h) structure with a square planar MX4 unit, and a lower-symmetry (Cs) structure with a tetrahedral MX4 unit, in which each opposing CH3CN unit coordinated along a line bisecting two X−M−X angles. In the former cases, we obtained metastable structures with relatively short M−N bonds (∼2.0 Å), and energies 3−7 kcal/mol above that of energy the separated fragments. In the latter cases, the CH3CN subunits assumed a bent geometry after the optimization. For the stronger complexes, we focused almost exclusively on the C2h structures, which were ultimately the most stable. To check these results, however, we did rerun the optimizations without the symmetry constraint and also restart them with a small degree of distortion in the MX4 subunit (while retaining Cs symmetry), and these yielded structures with at most trivial deviations from C2h. A more detailed consideration of the structural properties for (CH3CN)2−GeCl4, (CH3CN)2−SiCl4, and (CH3CN)2−SiF4 illustrates that the addition of a second CH3CN molecule had little effect; the coordination of the second CH3CN unit was weak compared to that of the first and did not act to strengthen any of the M−N coordinate bonds. Furthermore, as was observed for their 1:1 analogues, the M−N distances via MP2 are notably shorter than those via X3LYP. For (CH3CN)2−SiF4, the Si−N distance is about 0.1 Å shorter via MP2, and in the case of (CH3 CN) 2 −GeCl 4 and (CH3CN)2−SiCl4, the corresponding M−N distances are between 0.4 and 0.5 Å shorter. Nonetheless, we note for all three of these weaker systems, the 2:1 complexes exhibit a slight lengthening of their M−N bonds relative to their 1:1 counterparts: 0.06 Å in the case of (CH3CN)2−SiF4, 0.30 Å for (CH3CN)2−SiCl4, and 0.11 Å for (CH3CN)2−GeCl4 (via MP2). As such, these data indicate that further coordination by a second CH3CN subunit, albeit within a constrained orientation, does not enhance the M−N coordinate bonds relative to the 1:1 systems or lead to any additional distortion of the MX4 subunit. Although there are, at least in principle, a myriad of possible aggregation geometries in the solid film samples, the absence of further distortion within the MX4 subunit by an additional CH3CN is consistent with the lack of complex-induced shifts in the GeCl4/CH3CN and SiF4/ CH3CN spectra. For (CH3CN)2−TiCl4 and (CH3CN)2−TiF4, the energetic preference for the cis (C2v) seems to be consistent with a

A more detailed comparison of frequency data does not yield much additional consistency. First, it is worth noting, however, that the vaMX band in the pure SiF4 film (centered at 991 cm−1), is shifted from its gas-phase frequency (1022 cm−1),18 and this suggests that the self-interactions in the SiF4 film are comparable to those in the CH3CN/SiF4 film. Calculations (X3LYP) predict only a small shift/splitting for the vaMX band in the gas-phase complex: 1011 cm−1 for free SiF4, 1017/968 cm−1 for CH3CN−SiF4. As noted above, the vaMX frequencies for the 1:1 CH3CN−SiF4 complex in solid argon were observed at 974 cm−1,16 which is reasonably consistent with the prediction (for the lower frequency component). We note that the observed absorption band extends to about 960 cm−1. Nonetheless, if the film were to promote the N−Si donor−acceptor interaction, one would expect a rather large shift in the vaMX mode, comparable to that observed for H3N−SiF4, for which the frequencies are observed at 952 and 854 cm−1.12 However, we observed no significant peaks anywhere between 800 and 950 cm−1. Thus, we can conclude that the interactions in the CH3CN/SiF4 film have not promoted or enhanced the Si−N donor−acceptor bond to any great extent. Again, this appears to conflict with computational results, but we note the calculated potentials in the dielectric media are incredibly flat, varying by less than ±0.4 kcal/mol over the entire 2.0−3.0 Å range for ε = 5.0. As such, the predictions do not clearly portray any specific, definitive structural change. Furthermore, the value of the dielectric constant of the SiF4/CH3CN film is somewhat uncertain. It is also possible that the film is not homogeneous and lacks 1:1 complex-like associations of CH3CN and SiF4. Indeed, we observed no increase in temperature when these two substances were combined, and it is possible that the SiF4 simply did not dissolve. In contrast to the SiF4- and GeCl4-containing films, the spectra of the CH3CN/TiCl4 films (Figure 4C) reflect a strong interaction between the acidic and basic components. In this case, the vaMX band in the pure TiCl4 spectrum (centered at 493 cm−1) is shifted to lower frequencies, and a very broad absorption feature results, ranging from 375 to 475 cm−1. This extends below the specified range of the spectrometer, so the trace becomes quite noisy in the low-frequency limit. Nonetheless, as we vary the composition, an excess of CH3CN tends to enhance the low-frequency portion of this broad feature (i.e., in the 2:1/CH3CN:TiCl4 trace), whereas an excess of TiCl4 favors the higher-frequency side (i.e., in the 1:2/ CH3CN:TiCl4 trace). There is also clear evidence of unreacted TiCl4 in the 1:2/CH3CN:TiCl4 trace, which is also present to a small degree in the 1:1 trace. These observations suggest the occurrence of different coordination ratios in the sample, possibly CH3CN−TiCl4 as well as (CH3CN)2−TiCl4. Indeed, X3LYP predictions indicate that the vaMX mode shifts from 497 cm−1 in free TiCl4 to 464 and 430 cm−1 in the 1:1 gas-phase complex, and we note that the observed bands for H3N−TiCl4 in solid argon occur at 456 and 440 cm−1.17 These frequencies correspond to the high-frequency portion of the board absorption feature in Figure 4C. This, together with the fact that the Ti−N potential curves indicate that the structure of CH3CN−TiCl4 is largely insensitive to bulk media, suggests that the high-frequency side of the broad absorption feature results from a 1:1 complex interaction the film. We will discuss predictions for the (CH3CN)2−TiCl4 complex below. We also note that three absorption features are observed in the CNstretching region (at 2265, 2293, and 2319 cm−1), of which the central band is coincident with the high-frequency component 4275

dx.doi.org/10.1021/jp4115207 | J. Phys. Chem. A 2014, 118, 4266−4277

The Journal of Physical Chemistry A

Article

preference for a “cooperative” donor−acceptor interaction, with the CH3CN ligands involved in a σ-donor interaction with different orbitals on the acceptor, as opposed to an “anticooperative” interaction, in which they would be donating to the same orbital.33 In the trans configuration, each CH3CN would interact with opposite lobes of the one of the two LUMO orbitals on the TiX4 (which are primarily 3dz2 or 3dx2−y2 in character, and degenerate in the Td point group). In a cis, perpendicular, arrangement, each CH3CN could interact with a different acceptor orbital on the TiX4, which is a cooperative interaction. For the cis forms of both (CH3CN)2−TiCl4 and (CH3CN)2−TiF4, the binding energies are nearly twice that of the analogous 1:1 systems, indicating that the second CH3CN unit is bound with nearly equal energy as the first. For the trans isomers, in which each CH3CN would be interacting with the opposite lobe of the same acceptor orbital (anticooperative), the binding energies are much less; with values of −8.9 and −13.4 kcal/mol for the TiF4 and TiCl4 systems, respectively. This illustrates that the second CH3CN is bonded much more weakly within that configuration. For (CH3CN)2−GeF4, the LUMO orbital is largely 4s in character, and the cis isomer lies only 0.5 kcal/mol lower in energy, with an overall binding energy of −18.4 kcal/mol. In contrast to their weaker counterparts, the MP2 and X3LYP structural predictions are reasonably consistent for (CH3CN)2−TiCl4, (CH3CN)2−TiF4, and (CH3CN)2−GeF4; the M−N distances via MP2 are a few hundredths of an Ångstrom shorter for the TiX4 systems, though about 0.1 Å for (CH3CN)2−GeF4. Nonetheless, as is the case with the weaker systems, the structural properties of the GeF4 and TiX4 systems resemble those of their 1:1 analogues, for the most part. Specifically, the Ti−N bonds in (CH3CN)2−TiCl4 are slightly shorter (0.03−0.04 Å) and those in (CH3CN)2−TiF4 are slightly longer (0.01−0.02 Å), than in the corresponding 1:1 systems. For (CH3CN)2−GeF4, the Ge−N bonds are 0.003 Å longer than in the 1:1 complex according to X3LYP, but 0.17 Å shorter via MP2. Finally, we note that X3LYP/6-311+(2df,p) frequency predictions for (CH3CN)2−TiCl4 do indicate that the vaMX bands of this 2:1 complex are red-shifted to a greater extent than those for CH3 CN−TiCl 4 . Indeed, the predicted frequencies of the vaMX bands are 412 and 397 cm−1, and these lie within the low-frequency side of the broad (375−475 cm−1) absorption feature in the observed spectrum. This, together with the frequency predictions for the 1:1 complex and the composition dependence of the absorption band, seems to indicate that both 1:1 and 2:1 complexes are present in the CH3CN/TiCl4 film. Also, the rough agreement between measured solid-state spectra and calculated gas-phase frequencies further supports the conclusion that bulk interactions in the film do not affect the structures of these complexes to any great extent. Moreover, this is consistent with the computational results; i.e., the shape Ti−N potential curve for CH3CNTiCl4 changes only slightly in dielectric media.

strong, and although the M−N distances are relatively short (∼2.3 Å), there is an intermediate degree of distortion in the MX4 subunit, and their binding energies range from 11 to 13 kcal/mol. The SiX4 and GeCl4 complexes are weak, nonbonded systems with long M−N distances (3.0−3.8 Å), and binding energies that range from 3.0 to 4.4 kcal/mol. We also examined the structural and energetic properties of the analogous 2:1 complexes, (CH3CN)2−MX4. In these systems, the observed M−N distances were similar to those in their 1:1 counterparts, and binding energies indicate that the second CH3CN is bound with nearly equal strength as the first. The computed M−N bond potentials of the stronger CH3CN−MX4 complexes exhibit no remarkably peculiar features, but the curve for CH3CN−GeF4 is rather flat, and the minimum on the potential shifts about 0.2 Å inward in dielectric media. However, no significant condensed-phase structural changes are predicted for CH3CN−TiCl4 or CH3CN−TiF4 on this basis. The M−N potentials of the weakly bonded bonded complexes, CH3CN−GeCl4, CH3CN− SiCl4, and CH3CN−SiF4, are considerably flat in the direction of shorter M−N distances, and curves in dielectric media exhibit a preferential solvent stabilization in their interior regions. However, CH3CN−SiF4 is only one of these weaker systems for which we predict condensed-phase structural changes. In this case, the potential becomes quite flat over an incredibly broad range in dielectric media, and furthermore, global minimum shifts to 2.0 Å at ε = 20.0; a distance that is 1.0 Å shorter than the equilibrium gas-phase N−Si distance. IR spectra of low-temperature thin films composed of CH3CN and GeCl4, SiF4, or TiCl4 and bulk reactivity experiments involving these substances are largely consistent with the results of the computational study, with the exception of the CH3CN/SiF4 system. Both theory and experiment indicate a very weak interaction between GeCl4 and CH3CN, which is not enhanced by interactions in bulk media. On the other hand, both theory and experiment indicate a strong interaction between TiCl4 and CH3CN. Both the 1:1 and 2:1 complexes are predicted to be stable, and there is evidence that both persist to a varying degree in the solid state. However, because a short, relatively strong Ti−N donor−acceptor bond is predicted for even the gas-phase complex, this system is not prone to condensed-phase structural effects. The situation for the CH3CN/SiF4 system is less clear, however. Although computations indicate that the N−Si potential changes dramatically in dielectric media, there is no evidence of any enhanced interaction between these substances either in the bulk reactivity experiments or in the thin-film IR spectra.

■

ASSOCIATED CONTENT

S Supporting Information *

M−N potentials of CH3CN−TiCl4 and CH3CN−SiF4 (Figure S1) are available free of charge via the Internet at http://pubs. acs.org.

■

CONCLUSIONS We have conducted an extensive computational study of the structural and energetic properties of acetonitrile−Group IV (A & B) tetrahalide complexes (CH3CN−MX4: M = Si, Ge, Ti; X = F, Cl) and also examined the structures of the analogous complexes containing two acetonitrile subunits ((CH3CN)2− MX4). The 1:1 complexes fall, for the most part, into two structural types. The TiX4 and GeF4 complexes are moderately

■

AUTHOR INFORMATION

Corresponding Author

*J. A. Phillips: phone, 715-836-5399; e-mail, [email protected]. Notes

The authors declare no competing financial interest. 4276

dx.doi.org/10.1021/jp4115207 | J. Phys. Chem. A 2014, 118, 4266−4277

The Journal of Physical Chemistry A

■

Article

(18) Herzberg, G. Molecular Spectra And Molecular Structure: Infrared and Raman Spectra of Polyatmoic Molecules; Krieger Publishing: Boca Raton, FL, 1989. (19) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Montgomery, J. A., Jr.; Vreven, T.; Kudin, K. N.; Burant, J. C.; et al. Gaussian 03, Revision E.01; Gaussian, Inc.: Wallingford, CT, 2004. (20) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Scalmani, G.; Barone, V.; Mennucci, B.; Petersson, G. A.; et al. Gaussian 09, Revisions A.02, B.01, and D.01; Gaussian, Inc.: Wallingford, CT, 2009−2013. (21) For a recent overview of the basis sets and older computational methods referred to herein, see: Cramer, C. J. Essentials of Computational Chemistry, 2nd ed.; John Wiley and Sons: Chichester, U.K., 2004; and references therein. (22) NIST Chemistry Webbook. http://webbook.nist.gov. (23) Zhao, Y.; Truhlar, D. G. The M06 Ssuite of Density Functionals for main Group Thermochemistry, Thermochemical Kinetics, Noncovalent Interactions, Excited States, and Transition Elements: Two New Functionals and Systematic Testing of Four M06-Class Functionals and 12 Other Functionals. Theor. Chem. Acc. 2008, 120, 215−241. (24) Chai, J. D.; Head-Gordon, M. Optimal Operators for HartreeFock Exchange from Long-Range Corrected Hybrid Density Functionals. Chem. Phys. Lett. 2008, 467, 176−178. (25) Chai, J. D.; Head-Gordon, M. Long-Range Corrected Hybrid Density Functionals with Damped Atom-Atom Dispersion Corrections. Phys. Chem. Chem. Phys. 2008, 10, 6615−6620. (26) Tomasi, J. Thirty Years of Continuum Solvation Chemistry: A Review, and Prospects for the Near Future. Theor. Chem. Acc. 2004, 112, 184−203. (27) Cordero, B.; Gomez, V.; Platero-Prats, A. E.; Reves, M.; Echeverria, J.; Cremades, E.; Barragan, F.; Alvarez, S. Covalent Radii Revisited. Dalton Trans. 2008, 2832−2838. (28) Mantina, M.; Chamberlin, A. C.; Valero, R.; Cramer, C. J.; Truhlar, D. G. Consistent van der Waals Radii for the Whole Main Group. J. Phys. Chem. A 2009, 113, 5806−5812. (29) Smith, E. L.; Sadowsky, D.; Cramer, C. J.; Phillips, J. A. Structure, Bonding, and Energetic Properties of Nitrile-Borane Complexes: RCN-BH3. J. Phys. Chem. A 2011, 115, 1955−1963. (30) Jonas, V.; Frenking, G.; Reetz, M. T. Comparative Theoretical Study of Lewis Acid - Base Complexes of BH3, BF3, BCl3, AlCl3, and SO2. J. Am. Chem. Soc. 1994, 116, 8741−8753. (31) Phillips, J. A.; Cramer, C. J. Quantum Chemical Characterization of the Structural and Energetic Properties of HCN−BF3. J. Chem. Theor. Comput. 2005, 1, 827−833. (32) Eigner, A. A.; Wrass, J. P.; Smith, E. L.; Knutson, C. C.; Phillips, J. A. Structural Properties of CH3CN−SO2 in the Gas Phase and Condensed-Phase Media via Density Functional Theory and Infrared Spectroscopy. J. Mol. Struct. 2009, 919, 312−320. (33) Weinhold, F.; Landis, C. R. Discovering chemistry with natural bond orbitals; Wiley: Hoboken, NJ, 2012.

ACKNOWLEDGMENTS The work was supported by the National Science Foundation grants CHE-0718164 (J.A.P.), CHE-1152820 (J.A.P.), and CHE-1229354 (MERCURY Consortium). Additional support was obtained from the Petroleum Research Fund (46291-B6 and 53066-UR6), administered by the American Chemical Society, as well as UWEC’s Office of Research and Sponsored Programs. J.A.P. also acknowledges a Henry Dreyfus TeacherScholar Award from the Camille and Henry Dreyfus Foundation. We also thank the reviewers for several insightful and constructive comments.

■

REFERENCES

(1) Leopold, K. R.; Canagaratna, M.; Phillips, J. A. Partially Bonded Molecules From the Solid State to the Stratosphere. Acc. Chem. Res. 1997, 30, 57−64. (2) Burns, W. A.; Leopold, K. R. Unusually Large Gas-Solid Structure Differences in HCN−BF3. J. Am. Chem. Soc. 1993, 115, 11622−11623. (3) Dvorak, M. A.; Ford, R. S.; Suenram, R. D.; Lovas, F. J.; Leopold, K. R. van der Waals vs. Covalent Bonding: Microwave Characterization of a Structurally Intermediate Case. J. Am. Chem. Soc. 1992, 114, 108−115. (4) Reeve, S. W.; Burns, W. A.; Lovas, F. J.; Suenram, R. D.; Leopold, K. R. Microwave Spectra and Structure of HCN−BF3: an Almost Weakly Bound Complex. J. Phys. Chem. 1993, 97, 10630−10637. (5) Phillips, J. A.; Halfen, J. A.; Wrass, J. P.; Knutson, C. C.; Cramer, C. J. Large Gas-Solid Structural Differences in Complexes of Haloacetonitriles with Boron Trifluoride. Inorg. Chem. 2006, 45, 722−731. (6) Wells, N. P.; Phillips, J. A. Infrared Spectrum of CH3CN-BF3 in Solid Argon. J. Phys. Chem. A 2002, 106, 1518−1523. (7) Eigner, A. A.; Rohde, J. A.; Knutson, C. C.; Phillips, J. A. IR Spectrum of CH3CN-BF3 in Solid Neon: Matrix Effects on the Structure of a Lewis Acid-Base Complex. J. Phys. Chem. B 2007, 111, 1402−1407. (8) Buchberger, A. R.; Danforth, S. J.; Bloomgren, K. M.; Rohde, J. A.; Smith, E. L.; Gardener, C. C. A.; Phillips, J. A. Condensed-Phase Effects on the Structural Properties of FCH 2 CN−BF 3 and ClCH2CN−BF3: A Matrix-Isolation and Computational Study. J. Phys. Chem. B 2013, 117, 11687−11696. (9) Phillips, J. A.; Cramer, C. J. B-N Distance Potential of CH3CNBF3 Revisited: Resolving the Experiment-Theory Structure Discrepancy and Modeling the Effects of Low-Dielectric Environments. J. Phys. Chem. B 2007, 111, 1408−1415. (10) Alkorta, I.; Rozas, I.; Elguero, J. Molecular Complexes Between Silicon Derivatives and Electron-Rich Groups. J. Phys. Chem. A 2001, 105, 743−749. (11) Davydova, E. I.; Timoshkin, A. Y.; Sevastianova, T. N.; Suvorov, A. V.; Frenking, G. Lewis Acidity of Group 14 Tetrahalides in Gas Phase. J. Mol. Struct.-THEOCHEM 2006, 767, 103−111. (12) Ault, B. S. Matrix-Isolation Studies of Lewis Acid-Base Interactions: Infrared Spectra of the 1:1 Adduct SiF4-NH3. Inorg. Chem. 1981, 20, 2817−2822. (13) Lorenz, T. J.; Ault, B. S. Matrix-Isolation Studies of Lewis AcidBase Interactions 2:1/1 Adducts of SiF4 with Methyl-Substituted Amines. Inorg. Chem. 1982, 21, 1758−1761. (14) McNair, A. M.; Ault, B. S. Matrix-Isolation Studies of Lewis Acid-Base Interactions 3: Infrared Spectra of 1/1 Tetrafluorogermane Amine Complexes. Inorg. Chem. 1982, 21, 1762−1765. (15) Ault, B. S. Matrix-Isolation Studies of Lewis Acid Complexes of Pyridine. J. Mol. Struct. 1985, 129, 287−298. (16) Ault, B. S. Matrix Isolations Investigation of the Complexes of Acetonitrile and Hydrogen Cyanide with SiF4 and GeF4. J. Mol. Struct. 1985, 130, 215−226. (17) Everhart, J. B.; Ault, B. S. Infrared Spectrscopic Study of the Cryogeninc Thin-Film and Matrix-Isolated Complexes of TiCl4 with NH3 and (CH3)3N. Inorg. Chem. 1995, 34, 4379−4384. 4277

dx.doi.org/10.1021/jp4115207 | J. Phys. Chem. A 2014, 118, 4266−4277