Geminal silica hydroxyls as adsorbing sites: an ab initio study - The

Alfonso Pedone , Gianluca Malavasi , M. Cristina Menziani , Ulderico Segre , Federico Musso , Marta Corno , Bartolomeo Civalleri and Piero Ugliengo...
17 downloads 0 Views 696KB Size
J . Phys. Chem. 1993,97, 2671-2676

2671

Geminal Silica Hydroxyls as Adsorbing Sites: An ab Initio Study A. M. Ferrari, P. Ugliengo, and E. Garrone' Dipartimento di Chimica Inorganica. Chimica Fisica e Chimica dei Materiali, Universitb di Torino, Via P. Giuria 7 101 25 Torino, Italy Received: September 9, 1992

A b initio techniques are used to study the interaction of H2O and NH3 with the silanediol molecule H2Si(OH)2, taken as representative of geminal hydroxyls occurring a t the surface of silica. Calculations concern fully optimized structures, and other meaningful structures involving interaction with one or both S i O H groups. When one S i O H group only is involved, the presence of the second S i O H group does not markedly affect the acidic properties of the former. For double interaction, three classes of adducts are studied, where the two S i O H groups act: (i) both as H-acceptors; (ii) one as H-donor and the other as H-acceptor; (iii) both as H-donors. Adducts of type i are weakly bound; in case ii, only H20 shows extra stabilization with respect to the interaction with one SiOH; in case iii water is favoured over ammonia, because of the availability of two lone pairs. Most stable adducts are however the fully optimized structures, which envisage interaction with only one hydroxyl with ammonia, and an adduct of type ii with water. Structures and 0-H stretching frequencies of silanediol are in good agreement with those of larger similar model molecules, i.e. orthosilicic acid H202Si(OH)2 and disiloxisilaneiiol (H&0)2Si(OH)2. -

Introduction At the surface of outgassed amorphous silica, besides isolated species, geminal hydroxyls (i.e., two hydroxyls sitting on the same Si atom) are known to occur. Evidence comes from 29SiN M R spectra, showing distinct signals for geminal and isolated hydroxy1s:l4 peak intensities suggest that the population of geminal hydroxyls constitues a substantial fraction (up to 30%) of that of isolated ones. At the surface of crystalline specimens, the presence of geminal species can be even more substantial. Any difference in the adsorptive properties of the two species may therefore affect the overall behavior of silica. As isolated and geminal species seem to have 0-H stretching modes indistinguishable in the IR,54 it is not easy to bring into evidence any such difference, though they have been often hypothesized, in particular towards water, because of the presence, in geminal species, of a potentially active bif~nctionality.~-~ The experimental study of the interaction of water with silica hydroxyls has attracted much a t t e n t i ~ n . Due ~ , ~to~ overlap ~~ of the 0-H modes of SiOH and water in the vibrational spectrum, however, a definite picture of the interaction has not been arrived at. The poverty of experimental evidence imparts particular interest in the present case to the study of computational models. With such covalent solid as Si02, a cluster approach is particularly suitable. The minimal cluster for the description of the isolated hydroxyl group a t the silica surface is the silanol molecule H3SiOH:l3$14the overall evidenceLsis that the silanol molecule (though non-existing as such) is probably somewhat less acidic than the real hydroxyl group of silica, but nonetheless serves quite well as a model.16 The interaction of water with silica hydroxyls has been modelled computationally via the silanol molecule by several authors.t5-t7-21 In the same line, the silanediol molecule H2Si(OH)2 may be assumed as the simplest cluster mimicking geminal SiOH species. Alternatively, orthosilicic acid H4Si04may be considered as a model for both the isolated and geminal types of hydroxyls, by considering one and two 0-H groups, respectively, as active centres, and the remaining ones just as cluster terminators. Silanediol has been studied in a pioneering ab initio work by Hargittai and sei^,^^ bearing no relationship, however, with silica. Orthosilicic acid H4Si04 in various conformations, alone and in interaction with water, has been studied ab initio by Sauer and 0022-365419312097-2671%04.00/0

SchrMer21 at low level of treatment (STO-3G, 3-21G, 4-31G basis sets). Indications are that the stretching modes of geminal and isolated species are indeed indistinguishable, and that the bifunctional nature of the former does not impart any particular affinity for the water molecule. Very recently,22two conformations of orthosilicic acid have been studied ab-initio in interaction with water molecule. A higher order model envisages the d o x y group H3SiO as a substituent in thecoordinationsphereofthesi atom. For example, Sauer and SchrMer21have considered as a model for the isolated hydroxyl (H3Si0)3SiOH. Owing to its relevance, we have resumed the computational study of the adsorptive properties of geminal SiOH groups. We have tried to adopt a level of treatment as high as possible (large basis sets, inclusion of basis set superposition error and electron correlation): this has led to theassumption of silanediol as model molecule in the interactions. On the other hand, to have an idea how much the results were model-dependent, we have compared conformations and vibrational properties of silanediol in the absence of interactions with larger model molecules, namely orthosilicic acid H202Si(OH)2 and disiloxysilanediol (H3SiO)ZSi(OH)2. Beside water, we have considered as adsorbate also the ammonia molecule, which is a stronger base with much less tendency to act as proton donor, and is therefore more apt to establish the purely acidic behavior of SiOH groups of different nature. Methods All the calculations have been performed at ab initio level, using the GAUSSIAN88 suite of programs.24 Basis set adopted is of double zeta quality (DZP) of the type (12s 8p ld)/[6s 4p Id] for Si, (10s 5p ld)/[3s 5p Id] for 0 and N, and (4s lp)/[2s lp] for H respectively. GTO primitive sets are taken from H u ~ i n a g a .The ~ ~ exponents of the polarization functions are 0.45 (d on Si), 0.8(d on 0 and N ) and 1.1 (p on H). The d sets consist of six functions only. Some calculations have also been run with the lesser-quality basis set known as MINI-1 .26 The isolated silanediol molecule has been studied first in its absolute-minimum configuration of C2 symmetry (Figure l), referred to hereafter as GG (short for gauche-gauche, such being the conformation of the two hydroxyls arrived at). Other three possible configurations have been investigated, though corre0 1993 American Chemical Society

2612

Ferrari et al.

The Journal of Physical Chemistry, Vol. 97, No. 11, 1993

€5 5 3

55

144

EE 165

Figure 1. OptimizedSCF/DZPgeometry of varioussilanediolconformers. GG: fully optimized; others: for constraints see text. Angles in degrees, lengths in angstroms; energy difference with respect to GG conformer in kJ mol I . h

%*

Figure 3. Chart of the relative stabilitiesof silanediol various conformers. Ordinate scale: Energy difference at MPZ/DZP//SCF/DZP level, including .SCF/DZP//SCF/DZP ZPE correction and BSSE when appropriate.

;C96

u Figure 2. Optimized SCF/DZP geometry of orthosilicic acid and disiloxysilanediol: for constraints, see text. Angles in degrees, lengths in angstroms.

sponding to transition states, because they may play a role in adduct formation. These are also shown in Figure 1, and referred to hereafter, following the nomenclature adopted in ref 21, as EE, ES and SS, respectively (E standing for eclipsed, S for staggered configuration of one hydroxyl group with respect to the other Si-0 bond). As to orthosilicic acid and disiloxysilanediol, we have adopted a conformation of the substituent groups thought to be suitable for modelling geminals at the surface of silica, instead of absolute minimum-energy configurations. As to H202Si(OH)2, the two hydroxyls not considered as interaction centres have been given a downward asset, so to increase the average distance from the pair of SiOH groups active in the interactions (Figure 2). Similarly, the two Si-0 bonds in H&O groups of disiloxysilanediol (Figure 3) have been constrained to be coplanar, and the Si-0-Si

angle has been given a standard value of 143'. In both cases the two S i - 0 bonds in the substituent groups and in the SiOH groups lie in two planes normal to each other. Given these constraints, optimization has been carried out for the conformer of type GG: additional constraints are imposed in the optimization of structures of type ES, EE and SS. Adducts with water (W) and ammonia (A) have been studied with both GG (yielding configurations of absolute minimum energy), and the three conformers. In the latter case, interaction may take place either with only one 0-Hgroup (we refer to such complexes as monodentate, and label them EE-1-A, ES-1-W, etc.) or with two hydroxyl groups (bidentate adducts, labelled accordingly EE-2-A, ES-2-W, etc.). Geometry optimization both of model-molecule conformers and of the various adducts of silanediol has been carried out at the self-consistent field (SCF) level using analytical gradient techniques. From case to case, a C, or C2, symmetry has been imposed: only the search for adducts with absolute minimum energy has been carried out without any constraint. S C F harmonic normal-mode frequencies have been computed from analytical second energy derivatives and solving the equations of nuclear motion by standard methods. Electronic correlation has been evaluated using perturbative Moller-Plesset te~hnique,~' truncated at second order (MP2) on the geometries optimized at the SCF level. Basis set superposition error (BSSE) has been evaluated using the full counterpoise method2*for both S C F and MP2 calculations.

Results and Discussion Sianediol and Related Model Molecules. Computed geometric assets of silanediol are shown in Figure 1 for all four configurations. The Si-0-H angle is always close to 120O. This implies that the two hydroxyl groups interact to a negligible extent: in ES, the distance H-0 between the acidic proton of one hydroxyl and the oxygen in the second is 2.666 A, which does not allow H-bonding; in EE the two 0 - H groups are nearly parallel. The 0-Si-0 angle is comprised between 1 0 5 O and 113 O , and follows the changes in the total energies (Table I). The order of stability is GG > ES

Geminal Silica Hyroxyls as Adsorbing Sites

The Journal of Physical Chemistry, Vol. 97, No. 11, 1993 2613

TABLE I: Total Energy hartrees) of the Various Systems under Study at SCF/DZP)/SCF/DZP and MPYDZP// SCF/DZP Level: Dipole Moments (debye) and Zero-Point Energy ZPE (kJ/mol) Calculated at SCF/DZP//SCF/DZP Level system

GG

ES

ss

EE W A GG- 1-A ES- 1-A SS-I-A ES-2-A EE-2-A GG-2-W ES-I-W

ss-I-w

ES-2-W ss-2-w EE-2-W

SC F

MP2

P

ZPE

-441 . I 12997 -441 . I 1 1548 -441.108587 -441.107591 -76.046643 -56.209531 -497.335671 -497.333855 -497.330627 -497.333640 -497.329523 -5 17.17 I482 -517.168431 -517.165335 -517.169824 -517.163403 -517.167786

-441.569027 - 4 4 1.566541 -441.563522 -441.562648 -76.239350 -56.394422 -497.981256 -497.977952 -497.91467 3 -497.978860 -497.973429 -5 17.824550 -517.818688 -517.81 5632 -517.821574 -517.81 3635 -51 7.8 I8980

0.918 1.917 2.227 2.668 2.186 1.805 3.627 4.740 3.785 3.452 5.067 2.432 4.944 3.265 3.747 5.057 5.525

128.2 125.9 125.5 124.8 61.1 95.6 234.5 23 1.7 231.5 233.2 231.0 199.6 194.4 194.4 196.6 193.9 195.9

TABLE 11: Total Energies (hartrees) of the Lar er Model Molecules X2Si(OH)2(X = OH, H3SiO) at SCF/!DZP// SCF/DZP and MPZ/DZP//SCF/DZP Level, and SCF/ DZP//SCF/DZP 0-H Harmonic Stretching Frequencies (Wavenumbers) X system SCF MP2 YOH(I) YOH(2) 4248 4250 -591.802798 OH GG -591.004950 ES

SS OSiHj

EE GG ES

SS EE

-591.005805 -59 1.003120 -591.000786 -1171.260566 -1171.259072 -1171.256042 -1171.254725

-591.804031 -591.801 5 18 -591.79861 1 -1 172.425109 -1 172.423974 -1 172.421424 -1 172.419054

4255 4268 429 1 4249 4256 427 1 4285

4262 4268 429 1 4252 4264 427 1 4294

> SS > EE: differences in energy with respect to the GG global minimum are 3.8, 11.6 and 14.2 kJ mol-', respectively, a t S C F level and 6.5, 14.4 and 16.5 kJ mol-' at MP2 level: these latter values are reported in the charts of Figures 3 and 7 . Such a sequence is not accounted for by simple considerations, involving for instance repulsion among the hydrogen atoms: the changes in the 0-Si-0 angle seem to suggest instead that the whole structure responds to a change in geometry. By the same token, the larger stability of ES over SS could be in principle ascribed to some H bonding interaction in ES: this is clearly denied, however, by the above considerations and the value of the corresponding v(0H) mode, which is only marginally smaller than that corresponding to the free hydroxyl (see below). Such considerations suggest that the relative stability of the various assets of the two hydroxyls may depend on the rest of the model molecule, Le. on the groups saturating the Si coordination sphere. Results obtained for the larger cluster models are reported in Table 11: an overall illustration is given in Figure 3. It comes out that the relative stability of the conformers is basically the same for the three model molecules. Note that, whereas the orthosilicic acid shows one exception (ES configuration more stable than GG), the energy differences between GG, ES, EE and SS conformers of the larger cluster (HSi0)2Si(OH)z nearly coincide with those for silanediol, at least at S C F level. This suggests that results for the silanediol molecule may be regarded with confidence. The good agreement of results for H3SiOH and (HjSiO)2Si(OH)2,suggesting a close role for the H atom and the H3SiO group in determining the structural properties of silanol derivatives, also lends support to adopting H3SiOH as a model for the isolated silica hydroxyl. Some selected vibrational features of the various silanediol conformers are reported in the upper part of Table 111, together with those of silanol (SIL). Of the low-lying modes, only the

imaginary frequencies are given in Table 111: these show that ES, SS and EE are indeed transition states (EE is so with respect to two degrees of freedom) and GG an absolute-minimum configuration. As to v(OH) modes, we note that the GG values, where perturbation from the surrounding atoms is least, are close to the value for silanol: a small splitting between the two modes is observed. With SS,a degenerate single frequency is calculated, higher than that of silanol, because of the repulsive action exerted by the SiH2 protons; with ES, the lower-frequency value is associated with the hydroxyl in eclipsed configuration, apparently available for an H-bond, but indeed more free than that in staggered configuration. With EE, two high-frequency modes are calculated, because of the repulsion exerted by hydroxylic protons on each other: this is also the cause of the splitting observed. Similar data reported in the lower part of Table I1 for larger cluster models show exactly the same features: indeed the effect of changing substituent groups in the coordination sphere of the Si atom on the absolute values of u(0H) is rather limited. In order to establish which configuration is probably assumed in the real system, one may note that the differences in energies among the various conformers are of the order of some HT (RT = 2.5 kJ mol-I for T = 300 K), Le. large enough to suggest that, if one is allowed to neglect the surrounding solid in the real system, the configuration assumed a t moderate temperatures would be GG. Though such assumption could be incorrect, Le. the surrounding solid could in principle alter the relative stability of the conformers, two observations support the above conclusion. First, the larger model molecules, which somehow take into account the presence of a medium into which the geminal hydroxyls are embedded, invariably indicate GG as the most stable conformer. A second type of support comes froin the IR features of both isolated and geminal hydroxyls. The experiment indicates that v(0H) modes of isolated and geminal species are indistinguishable within few wavenumbers, and both correspond to the 3747 cm-I band: this is in good agreement with the closeness of the values of 4250,4244 and 4246 cm-' calculated for SIL and GG respectively (as usual with S C F harmonic vibrational data, a scaling factor around 0.90 is needed to relate calculated and experimental vibrational frequencies). No other conformer reproduces such feature. Adducts with Water and Ammonia. Fully optimized structures for the adduct of silanediol with ammonia and water respectively are reported in Figure 4. It is noteworthy that in the former case interaction only takes place with one SiOH group, whereas in the latter the water molecule acts as proton-acceptor with one hydroxyl, and as proton-donor to the other. The two structures will therefore be denoted as GG- 1-A and GG-2-W respectively. As to the other conformers, interaction with only one hydroxyl group may take place with ES and EE. The four structures ES(EE)-1-W(A) are illustrated in Figure 5. Twofold interaction may take place with all the three conformers of silanediol, as shown in Figure 6. With ES, interaction takes place with one O H acting as H donor, and the other as H-acceptor; with SS both O H groups act as H acceptors (this type of adduct with ammonia has not been studied in detail, being very labile); with EE both act as H donors. Geometrical details are reported in Figures 5 and 6 for ES, SS and EE adducts, absolute energies in Table I, vibrational features in Table 3, and binding energies at various levels of treatment in Table IV. The chart in Figure 7 yields an overall view of the relative stability of the various adducts, and a comparison of the similar data for silanol. The changes in the geometry of the two moieties brought about by adduct formation are small. In H-donor molecules some increase in the length of the 0 - H (or N-H) bond is observed: in H-acceptor molecules opposite variations are seen. These features monitor the tiny charge transfer and intermolecular polarization that take place upon adduct formation, and are in accord with the so called Gutmann rules.29

2674

The Journal of Physical Chemistry, Vol. 97, No. 11, 1993

Ferrari et al.

TABLE 111: SCF/DZP//SCF/DZP Harmonic Vibrational Features of the Various Systems under Study in cm-I (X = 0, N for Water and Ammonia, Respectively) system

GG ES

ss

EE

W A SIL-A GG-I-A ES- 1-A SS-1-A ES-2-A EE-2-A SIL-w GG-2-W ES-I-W

ss-I-w ES-2- W ss-2-w EE-2-W

YEH

YX-H

A&

,,

AWH

4244,4246 4244,4256 4263,4263 4275(B2), 4281(Ai) 3971 3958,4240 3966,4245 3972,4261 3923,4245 4184(B2), 4202(Ai) 41 17 4152,4227 4105,4244 4105,4261 4108,4235 4247,4247 4189(B2), 4207(Ai)

low-lying modes 177i 214i 425i, 2651

4164,4291 3729,3878,3878 3719,3861,3861 3721,3863,3864 3721,3863,3863 3720,3862, 3862 3720,3863,3869 3712,3844,3852 4161,4283 4 128,4266 4161,4282 4160,4280 4132,4274 41 66,4264 4149.4269

-10, -17, -17 -8, -15, -14 -8,-15,-15 -9, -16, -16 -9, -1 5, -9 -1 7, -34, -26 -3, -8 -36, -25 -3, -9 -4, -1 1 -32, -17 2, -27 -1 5. -22

-279 -287, -5 -290, 1 -291,-2 -321, -1 1 -91, -?9 -133 -93, -18 -151,O -158,-2 -136, -21 -16, -16 -86, -74

ES-1 - A

€5-1-w

41,66, 76, 181, 240, 310 23,41, 71, 186, 232, 261 117i, 21i, 29,60, 165, 234 116i, 17,4S, 60, 166,228 46i, 65, 97, 136, 192, 216 71i, 33i, 32, 127, 141, 169 40, 71, 97, 104, 176, 191 78, 109, 145, 167, 231,318 122i, 34,46,77, 148, 182 123i, 42,47,77, 150, 186 164i, 50i, 130. 142, 166,366 187i, 29,48, 115, 167,207 IOli, 71i, 81, 104, 142, 300

55-1-A

ss-1-w

Figure 5. Optimized S C F / D Z P geometry of ES and EE monodentate adducts with water and ammonia: angles indegrees, lengths in angstroms.

l&m 0,942 Figure 4. Fully optimized S C F / D Z P geometry of silanediol adducts with water and ammonia: angles in degrees, lengths in angstroms.

As to the relative location of the two moieties, we note that in monodentate adducts the SiO-H group is nearly aligned with the acceptor atom (0and N respectively) in the basic molecule and its lone pair, as expected on the basis of electrostatic description of H-bonding.30 The distance between the acceptor 0 (or N ) atom and the H atom which is being donated is about the sum of the corresponding van der Waals radii, and slightly depending on the binding energy (Table IV): for instance, in the GG-1-A adduct the N--H distance is 1.997 A, the shortest in the present set, this being the most stable ammonia adduct. Inclusion of electron correlation in the geometry optimization is expected to shorten thisdistance byO.l-0.15 A (becauseofattractivedispersive contributions not accounted for by the S C F approach) and to bring it slightly below the sum of van der Waals radii. In bidentate adducts, as well as in fully optimized structures, water and ammonia behave quite differently, as the latter shows little tendency to interact with the second SiOH group. In ES2-A, only a slight bending of the NH3 moiety is seen towards the H-acceptor hydroxyl (see Figure 6), whereas a substantial bending

ES-2-W

€E-2-W

Figure6. OptimizedSCF/DZPgeometryofES andSS bidentateadducts with water and ammonia: angles in degrees, lengths in angstroms.

is seen in the case of water in ES-2-W. The geometry of this structure is close to the structure reported by Pelmenschikov and coworkers**at SCF/6-3 lG*//SCF/6-31G* level, for the inte-

The Journal of Physical Chemistry, Vol. 97, No. 11, 1993 2675

Geminal Silica Hyroxyls as Adsorbing Sites

TABLE I V Binding Energies of the Various Adducts at Different Levels of Treatment. system

ASCF

ASCFC

ASC F M I h I. I

AMP2

AMP2C

AZPE

AHo(0)

SIL-A GG-I-A ES-I-A S S - 1-A ES-2-A EE-2-A SIL-w GG-2-W ES-I-W

32.5 34.5 33.5 32.8 33.0 32.6 25.8 31.1 26.9 26.5 30.5 21.5 35.6

30.1 31.9 31.3 30.3 29.8 30.1 24.3 28.0 25.3 24.8 28.0 19.1 33.4

30.1

43.5 46.8 44.6 43.9 47.0 42.9 32.0 42.5 33.6 33.5 41.2 28.3 44.6

36.2 39.3 36.6 36.7 38.0 36.0 27.2 34.9 28.7 28.4 33.5 22.0 37.8

8.6 10.7 10.2 10.4 11.7 10.6 6.9 10.3 7.3 7.8 9.6 7.3 10.1

27.6 28.6 26.4 26.3 26.3 25.4 20.3 24.6 21.4 20.6 23.9 14.7 27.7

ss-1-w ES-2-W ss-2-w EE-2-W

32.4 31.8 28.6 22.5 24.6 24.2 28.3 28.5

C stands for BSSE-corrected value; AZPE zero point correction energy for the complexes; AHo(0) enthalpy of interaction at 0 K. MP2 data are relative to S C F / D Z P optimized geometries. All data in kJ/mol.

I

I

I

Figure 7. Chart of the relative stabilities of silanol, silanediol various conformers and of their adducts with ammonia and water. Silanol and GG conformer are given arbitrary nil energy. Ordinate scale: Energy difference at MPZ/DZP//SCF/DZP level, including SCF/DZP//SCF/ DZP ZPE correction and BSSE when appropriate.

action of water with the orthosilicic acid conformer with the two active hydroxyls in an arrangement similar to ES. It is also worth of note that in GG-2-W, the water molecule is lying on the silanediol species so to fill almost exactly one half of the available space: it so appears that the silanediol molecule may accomodate another water molecule with ease. In EE-2-A and EE-ZW, being the interaction constrained to be the same with both hydroxyls (the overall symmetry is imposed to C, or C2r,respectively), the distances 0.-H or Nq-H are both respectively larger than the corresponding value in monodentate adducts. It is noteworthy that the relative unstability of EE-2-A with respect to EE-2-W is revealed by a N-H distance which has become much larger than 0.-H, whereas they are very close to each other in monodentate adducts. The geometrical features of the SS-2-W adduct, not described as a figure because of the instability of such complex, are as expected (Cz1symmetry, 0.-H distance is 2.365 A, 0.-H-W angle of 1 4 1 O ) . Data in Table IV show that the binding energies of monodentate complexes with either water or ammonia are all close to each

other, and close to the values obtained respectively with silanol: some extra stabilization is only shown by ammonia with GG. This is clear evidence that the acidic properties of a single 0-H group in silanediol are not much affected by the presence of the other hydroxyl group by some sort of inductive effect. With bidentate adducts, ammonia shows no further stabilization in ES-2-A, because of the negligible tendency of ammonia to act as H-donor; in EE-2-A a decrease in the binding energy isobserved. Water shows the opposite behavior, being somewhat stabilized both in GG and in ES, the two structure where water also acts as proton donor. Extra-stabilization of a different kind comes from the symmetrical interaction with the second SiOH acting as H-donor as in EE-2-W. These results deservesomecomments. Theinteraction of water or methanol with the isolated hydroxyl of silica has been sometimes described as involving both the acidic hydroxyl and the oxygen atom of a nearby siloxane bridge:3' the present results seem to indicate that the role of the latter is far from being vital, if it is taken into account that the siloxane oxygen atom is far less basic than that of the SiOH group here considered. The peculiar stabilization of water interacting with EE seems to be related to the occurrence of two lone pairs in the oxygen atom: we have preliminary computational evidence that a relatively large stabilization is also observed for the adduct with formaldehyde. Table IV also reports binding energies, after BSSE correction, calculated via the minimal basis set MINI-1 known to be particularly suited for H-bonding interactions. These results are indeed rather close to those calculated via the more sophisticated double-zeta quality basis set, in full agreement with the pioneering work on this subject by Sauer and H0bza.3~ It is obviously necessary to distinguish clearly between binding energies of the same molecule with different conformers, and the absolute energies of the adducts. These latter only dictate which species is actually formed on the surface, and do differ sizably from binding energies, because, as illustrated in Figure 3, the absolute energy of the various conformers is substantially different. Figure 7 reports the absolute energies for all adducts studied in the present paper, and the corresponding ones involving silanol. A common zero energy level has been assumed for silanol and silanediol (representative of the isolated and geminal hydroxyls, respectively). As to silanediol adducts, it is seen that the energy gain in adduct formation does in no case compensate for the energy increase in passing from GG to any other conformer. As a consequence, GG adducts for both water and ammonia are the most stable. Comparing now SIL and GG, we note that, whereas the energy gain upon ammonia adsorption is the same for the two surface structure, there is a definite energy gain for the interaction of water with GG. The vibrational modes of all adducts are gathered in the lower part of Table 111. The low-lying modes show invariably at least the same number of imaginary frequencies as the involved silanediol conformer. Only adducts with GG are real energy

2676

The Journal of Physical Chemistry, Vol. 97, No. 11, 1993

minima: all others are transition states with respect to at least one degree of freedom. Among the intramolecular modes, the shift in the SIO-H frequency of silanediol is the most significant in characterizing the interaction and allows direct comparison with experiment. The extent oftheshift in the H-donor hydroxylgroupissubstantial, and parallels the interaction energies, as expected for H-bonded adducts.14 16 With ES-2-W(A) adducts, thestretching frequency of the acceptor SiOH undergoes a small red-shift both with ammonia and water. That of the donor SiOH is close to those observed with monodentate adducts in the case of water, and somewhat larger with ammonia. This is surprising, because in the ES-2-A adduct the interaction of ammonia with the second hydroxyl is limited, and one would expect a value for Au(0H) coinciding with what observed for monodentate adducts. Moreover, such an increase is not accompanied by a parallel change in binding energy, and is therefore probably amenable to a delicate balance between electrostatic and charge-transfer effects. The pair of u(0H) modes of EE-2-A (or W) are both shifted to lower frequency, though to a lower extent than when the interaction takes place with one H-donor S O H group: the splitting between modes is also increased. As to the comparison with experiment, the shift of the Au(0H) band at 3747 cm-I, representative of both isolated and geminal SiOH species caused by the interaction with wateri5is not available for practical reasons. In the case of ammonia, the experimental value is known to be 650 cm-I, whereas the computed one are around 290 cm-' (see Table 111): the latter is underestimated by a factor around 0.5. Previous workI5 has shown that this is due both to the neglect of electron correlation and anharmonicity, and to the probable less acidic nature of the model molecules with respect to the real surface hydroxyls. The v(0H) mode of the hydroxyl group not involved in H-bonding is only marginally perturbed.

Conclusions Silanediol provides a satisfactory model for geminal hydroxyls, as good as the much larger cluster (H3Si0)2Si(OH)z. Evidence provided in this paper confirms that geminal silanols are likely to be indistinguishable from isolated ones in the IR. It also suggests that geminal hydroxyls may show adsorption properties different from those of isolated species as far as water is concerned, whereas ammonia is probably insensitive. Indeed, considering the most stable structures (those fully optimized) it results, on the one hand, that adsorption of water on geminal hydroxyls is some 8 kJ mol-' more energetic than silanol species; on the other hand, the structure arrived at may accomodate easily two water molecules per geminal species, so affecting the maximum coverage value. Being the population of geminal species rather relevent on amorphous silica (and the more so on crystalline modifications), this could be an important source of surface heterogeneity. The other structures studied in the present paper seem to play a minor role: it has to be noted, however, that on the actual

Ferrari et al. surface the relative stability of structures could be somewhat altered, and consequently those structures, like ES-2-W, the energy of which is not far from the absolute minimum could come into play.

Acknowledgment. The authors are grateful to CSI Piemonte for allowance of computer resources. References and Notes ( I ) Bronnimann. C. E.; Zeigler, R. C.; Maciel, G. E. J. Am. Chem. Soc. 1988, 110, 2023. (2) Morrow. B. A.; Gay, I . D. J. P h p . Chem. 1988. 92, 5569. (3) Legrand, A. P.; Hommel, H.; Taibi, H.; Miquel, J. L.; Tougne, P. Colloid Sur/. 1990, 45, 391. (4) Leonardelli. S.; Facchini, L.; Fretigny, C.; Tougne. P.; Legrand, A. P. J. Am. Chem.Soc. 1992, 114, 6412. ( 5 ) Hoffman, P.; Knozinger, E. Surface Sci. 1987, 188, 181. (6) McFarlan, A. J.; Morrow, B. A. J. Phys. Chem. 1991, 95, 5388. (7) Knozinger, H. In The hydrogen bond Schuster, P., Zundel, G.. Sandorfy, C. Eds.; North-Holland: Amsterdam 1976; Vol 111, Chapter 27, p 1263, and references therein. (8) Her, R. K. I n Thechemisrryofsilica;Wiley-Interscience: New York 1979; Chapter 6. (9) Kiselev,A. V.; Lygin, V. I . I n lnfraredspectraofsurfacecompounds; Wiley: New York, 1975. (IO) Anderson, J. H.; Wickersheim, K. A. Surf, Sci. 1964, 2. 252. ( I 1 ) Fubini, B.; Bolis, V.;Giamello, E. Inorg. Chem. Acta 1987,138. 193. (12) Klier, K. J . Chem. Phys. 1973, 58, 737. (13) Sauer, J. J. Phys. Chem. 1987, 91, 2315. (14) Ugliengo, P.; Garrone, E. J. Mol. Catal. 1989, 54, 439. (15) Ugliengo, P.; Saunders, V. R.;Garrone, E . J . Phys. Chem. 1989, 93, 521O;SurJ Sei. 1989,224,498; J. Phys. Chem. 1990,94,2260; Chem. Phys. Lett. 1990, 169, 501. (16) Garrone, E.; Ugliengo, P. Mater Chem. Phys. 1991,29,287; Garrone. E.; Ugliengo, P.; Ferrari, A. M. In New Trends in Physical Chemistry. edited by the Council of Scientific Research Integration, Trivandrum (India), in press. (17) Hobza. P.; Sauer, J.; Morgeneyer, C.; Hurych, J.; Zahradnik. R. J. Phys. Chem. 1981, 85, 4061. (18) Sauer, J.: Zahradnik, R. Int. J. Quanrum Chem. 1984, 26, 793. (19) Zhidomirov, G. M.; Kazansky, V. B. Ado. Carol. 1986, 34, 131. (20) Chakoumakos, B. C.; Gibbs, G. V. J. Phys. Chem. 1986, 85, 996. (21) Sauer, J.; Schrader, K. P. 2. Phys. Chem. Leipzig 1985, 266, 379. (22) Pelmenschikov, A. G.; Morosi, G.; Gamba, A. J. Phys. Chem. 1992, 94. 7422. (23) Hargittai, 1.; Seip, H. M. Acta Chem. Scand. 1976, A30, 153. (24) GAUSSIAN88, Frisch, M. J.; Head-Gordon, M.; Schlegel, H. B.; Raghavachari, K.; Binkley, J. S.; Gonzalez. C.; Defrees, D. J.; Fox, D. J.; Whiteside, R. A,; Seeger, R.; Melius, C. F.; Baker, J.; Martin, R.; Kahn, L. R.; Stewart, J. J. P.; Fluder, E. M.; Topiol, S.; Pople, J. A. Gaussian, Inc.: Pittsburgh, PA, 1988. (25) Huzinaaa. s.Aooroximalearomicwaoelunctions: Deot ofchemistrv Report, Universcyof Aiberta, Edmonton, Alberia,Canada, 1671. Huzinagi, S. J. Chem. Phys. 1965, 42, 1293. (26) Tatewaki, H.; Huzinaga, S. J. Commr. Chem. 1980. I. 205. (27) M d e r , C.; Plesset, M. S. Phys. Re;. 1934, 46, 618. (28) Boys, S. F.; Bernardi, F. Mol. Phys. 1970. 19, 553. (29) Gutmann, V . I n The Donor-Acceptor Approach to Molecular 1nteractions;Plenum Press: New York, 1978. Gutmann,V.; Resch.G.;Linert, W. Coord. Chem. Reo. 1982. 43, 133. (30) Legon, A. C.; Millen, D. J. J . Chem. Reo. 1986,86,635; Arc. Chem. Res. 1987, 20,39. Legon, A. C.; Millen, D. J. J . Am. Chem. Soc. 1987. 109. 356; Legon, A. C. Chem. Sor. Reo. 1990, 19. 197. (31) Pelmenschikov. A. G.; Morosi, G.; Gamba. A. J . Phys. Chem. 1992, 96. 2241. (32) Hobza, P.: Sauer, J . Theor. C'him. Acta 1984, 65, 279