Download as pdf or txt
Download as pdf or txt
You are on page 1of 14

Journal of Colloid and Interface Science 228, 157–170 (2000)

doi:10.1006/jcis.2000.6934, available online at http://www.idealibrary.com on

Mechanism and Kinetics of Hexamethyldisilazane Reaction


with a Fumed Silica Surface
V. M. Gun’ko,∗ M. S. Vedamuthu,† G. L. Henderson,† and J. P. Blitz†,1
∗ Institute of Surface Chemistry, 31 Prospect Nauki, 252022 Kiev, Ukraine; and †Department of Chemistry,
Eastern Illinois University, Charleston Illinois 61920

Received January 26, 2000; accepted April 24, 2000

lanes (5), also act as powerful catalysts. The relevant question


Quantum chemical calculations of the reaction of hexamethyl- then is the general nature of the catalytic effect.
disilazane with silica indicates a two-step mechanism. The first, Two broad classes of reaction mechanisms can be described
rate-determining step results in trimethylsilylation of a surface
to explain the reaction of organic and organosilicon compounds
silanol and formation of trimethylaminosilane. The second step in-
volves reaction of a trimethylaminosilane reactive intermediate to
with surface hydroxyls at moderate temperatures. The first pos-
form a trimethylsilyl surface species and ammonia. This two-step sibility is nucleophilic substitution of OH− or SN i(Si), where the
mechanism was applied to analyze data of the kinetics of hexam- electron-donor property of a reactive atom like N from the adsor-
ethyldisilazane reaction with fumed silica using custom interactive bate results in the formation of a donor–acceptor bond (N → Si
software. Both the extent of surface reaction and the loss of hex- in this case) in the prereaction complex or the transition state, TS.
amethyldisilazane from toluene solution were monitored by FT-IR The acidity of the hydrogen bonded to nitrogen is then impor-
spectroscopy as a function of time. Data analysis indicates that tant as H+ transfer occurs from the adsorbate (HMDS) to oxygen
under the conditions studied the reaction is adsorption rate lim- in the ≡SiOH. This mechanism is counterintuitive in this case
ited. The entire reaction can be explained in terms of reaction with and does not result in the known products trimethylsilyl groups
non-hydrogen-bonded surface silanols, though hydrogen-bonded bonded to the surface.
silanols act as adsorption sites. °C 2000 Academic Press
The second possibility involves electrophilic substitution of
Key Words: mechanism; kinetics; silylation reaction; silica
surfaces.
H+ or an SE i mechanism. In this case the main characteristic of
nitrogen is its proton-acceptor capability (3). Since silica surface
silanols are somewhat acidic (the isoelectric point of SiO2 ≈ 2)
and nitrogen on HMDS is basic, this interaction is highly plau-
INTRODUCTION
sible. The oxygen on the silanol then takes on a greater negative
The reactions of organosilicon compounds in acid–base re- charge due to this interaction rendering it more nucleophilic.
actions at oxides with surface OH groups is of technological It can then react with the electrophilic Si in HMDS, forming
importance in diverse fields (1). Because of its high reactivity the known product. It is thus believed that HMDS interacts with
with silica surface silanols, a system which has been widely used ≡SiOH via the SE i mechanism. Knowledge of the type of mech-
is silica reacted with hexamethyldisilazane (HMDS) to form anism allows for more exact theoretical modeling of the reaction
trimethylsilyl groups (2): pathway and the transition state.
We have undertaken studies to gain a deeper understanding
of the mechanism and kinetics of HMDS reaction with fumed
2≡SiOH + (CH3 )3 –Si–NH–Si–(CH3 )3
silica. The approach taken is to apply quantum chemical cal-
→ 2≡Si–O–Si–(CH3 )3 + NH3 . [1] culations to model the reaction in light of previously reported
data on this system (6). The mechanism found by the quantum
Even though this system is widely employed for the synthesis of chemical methods to be in agreement with previously reported
industrially useful materials, fundamental questions concerning experimental data will then be used to devise a kinetic model for
the chemistry have not yet been answered. this work. With this combined approach and current knowledge
The structural element in HMDS most influential in determin- of the silica surface structure, a greater insight into the details
ing its reactivity with silica surface silanols is the basic nitrogen of this reaction is obtained.
(3). It is well known that aminofunctional silanes are much more
reactive than their hydrocarbon analogs. Basic compounds, such
as amines, when added to either alkoxysilanes (4) or chlorosi- MATERIALS AND METHODS

Materials. Cab-O-Sil HS-5 fumed silica (surface area =


1 To whom correspondence should be addressed. 325 m2 /g), obtained from the Cabot Corp., was evacuated in
157 0021-9797/00 $35.00
Copyright °C 2000 by Academic Press
All rights of reproduction in any form reserved.
158 GUN’KO ET AL.

a 0.01 mm Hg vacuum for a minimum of 2 h prior to use. This ethyldisiloxane), intermediate species (trimethylaminosilane),
dry silica was covered in argon, sealed, and immediately trans- and other unexpected reaction products. None were detected.
ferred to a nitrogen-purged glove bag. Reagent-grade toluene
was dried for a minimum of 24 h by stirring with CaH2 under RESULTS AND DISCUSSION
argon prior to use. Hexamethyldisilazane (Silar laboratories)
was distilled under a nitrogen atmosphere and only opened in Quantum chemical calculations. Quantum chemical calcu-
a nitrogen-purged glove bag. These precautions were taken to lations of the reactivity of HMDS are difficult because each part
avoid atmospheric moisture, which complicates the chemistry of the molecule contributes to calculated parameter values. Only
of this system considerably. several atoms take part in the reaction however, so only their
HMDS reactions with fumed silica. Silica surface reactions contributions correlate with reactivity. Therefore, more local-
and all manipulations were performed in a nitrogen-purged glove ized characteristics of HMDS and adsorbed reaction complexes
bag. Freshly dried silica (1.0 g) was slurried in 25 ml of toluene. derived from HMDS are used to correlate calculated parame-
Hexamethyldisilazane (1.25 mmol) was added to a second flask ters and compound reactivity. This approach has worked well in
containing 25 ml of toluene. The HMDS solution was then added previous studies (3).
to the silica slurry with swirling to initiate reaction. The reaction It is known that HMDS reacts with silica surface silanols
was stopped after a specified period of time by filtrating and as shown in Eq. [1] probably via the SE i mechanism. Two
washing with toluene to remove adsorbed but unreacted HMDS. possible mechanisms can be envisioned within this general
The surface-reacted silica was then dried and cured at 100◦ C for scheme. The first involves the simultaneous reaction via a con-
1 h in a 0.01 mm Hg vacuum. certed mechanism with two neighboring ≡SiOH groups to form
Experiments to determine the adsorption profile were done by 2≡SiOSi(CH3 )3 and NH3 ,
analyzing the loss of HMDS from toluene solution. Dry silica
(4.0 g) was reacted with HMDS (5.0 mmol) in 500 ml of toluene. 2≡SiOH + HN(SiR3 )2 *
) 2≡SiOH· · ·N(H)(SiR3 )2
The reaction was allowed to proceed for 4 h with 0.5-ml aliquots → 2≡SiOSiR3 + NH3 , [2]
of supernatant collected for analysis at various times.
where R = CH3 .
Characterization. Transmission infrared spectra of Cab-
The second possibility involves a two-step reaction:
O-Sil HS5 were obtained using a Perkin–Elmer FT-IR 2000
equipped with a wide-band MCT detector. A small amount of ≡SiOH + HN(SiR3 )2
silica (∼50 mg) was placed between two NaCl plates to ob- H···NHSi0 R3
tain transmission spectra. The small particle size of the silica ) ≡SiOH· · ·N(H)(SiR3 )2 → [≡SiO0 ← Si0 R3 ]
*
gives rise to little scattering and high-quality spectra. A band → ≡SiOSiR3 + H2 NSiR3 . [3]
area ratio of the CH stretching absorbance from surface-bound
trimethylsilyl groups (∼2925 cm−1 ) to an Si–O–Si combination Trimethylaminosilane, the product of reaction [3], can interact
band of silica (∼1860 cm−1 ) was used for quantitative analysis and react with ≡SiOH groups as shown in Eq. [4].
of the surface reaction. All integrations were carried out using
standard Perkin–Elmer software. Silicas with varying amounts ≡SiOH + H2 NSiR3
of trimethylsilyl groups were prepared and analyzed by combus-
tion analysis (Galbraith, Knoxville, TN) and FT-IR spectroscopy ) ≡SiOH· · ·N(H2 )SiR3 → [≡SiO0 ← Si0 R3 ]
*
to obtain a calibration line for subsequent FT-IR analysis. → ≡SiOSiR3 + NH3 . [4]
Non-hydrogen-bonded silanol concentrations were obtained
by a ratio of the ∼3745-cm−1 band to the Si–O–Si combina- Quantum chemical calculations of the reactants and differ-
tion band. Assuming a known initial concentration (vide supra) ent reaction models in a cluster approach for HMDS reaction
(6, 17–21), non-hydrogen-bonded silanol concentrations are of the silica surface via reactions [2]–[4] were performed by
then determined at various times by calculating the amount lost ab intio (HF and DFT) (7, 8) and semiempirical (PM3 (9) and
relative to the initial concentration. AM1 (10)) methods using Gaussian 94 (7) and GAMESS (8)
Quantitative analysis of HMDS in toluene solution was ob- program packages. A few methods of transition state searching,
tained using a Perkin–Elmer FT-IR 2000 spectrometer equipped such as the synchronous transit-guided quasi-Newton method
with a horizontal attenuated total reflectance (HATR) acces- and an eigenvalue-following optimization of the transition state
sory (Spectra Tech, Stamford, CT). Standard HMDS solutions obtained by the quasi-Newton method, were used.
were prepared for calibration purposes. Spectral subtraction of A theoretical comparison of the concerted mechanism
toluene was carried out using standard Perkin–Elmer software. (Eq. [2]) and the stepwise mechanism (Eqs. [3] and [4]) was
GC/MS analysis of the toluene supernatant solution after reac- undertaken. It can be argued that the probability of a concerted
tion was performed using an HP5890 GC/5971 MS. Numerous reaction with two neighboring silanols may be small for a few
temperature profiles were investigated to detect products due to reasons. First, since HMDS has just one nitrogen with an un-
reactions with adventitious water (i.e., trimethylsilanol, hexam- shared electron pair, it can only form one strong hydrogen bond
MECHANISM/KINETICS HMDS REACTION SILICA SURFACE 159

with a silanol. Furthermore, the distance between two neighbor- TABLE 1


ing ≡SiOH groups must be in a narrow range corresponding Parameters of Electronic Structure of Some Organics and OSC
to the distance between Si atoms in HMDS to form a suitable Used for Modification of Oxide Surfaces
prereaction complex and transition state. Such a reaction mech-
−E HOMO E LUMO
anism may be possible with geminal silanols (=Si(OH)2 ); how- Compound (eV) (eV) µ (D) qSi,C −qO,N qH
ever, the simultaneous transfer of two protons from two silanols
seems unlikely. Related to this problem, the transition state en- ((CH3 )3 Si)2 NHa 9.87 5.04 0.58 1.230 1.030 0.291
tropy (pre-exponential) factor may be very small because the (CH3 )3 SiNH2 a 10.16 5.30 0.89 1.014 0.848 0.277
[(CH3 )3 SiNH3 ]+a 16.61 −0.57 4.56 1.021 0.817 0.373
number of degrees of freedom of two surface silanols and an
((CH3 )3 Si)2 NHc 6.23 1.52 0.51 0.835 0.769 0.254
HMDS molecule in such a transition state are frozen. The theo- (CH3 CH2 O)3 9.74 6.58 0.67 1.540 0.791N 0.293
retically determined activation energy for the concerted mecha- Si(CH2 )3 NH2 b
nism is ∼120 kJ, considerably higher than the 77 kJ experimen- MADESb 10.25 3.25 3.42 1.555 0.728
tally determined by Hertl and Hair (6). Based on these arguments MTESb 10.10 3.37 1.85 1.544 0.734
((CH3 )3 Si)2 Oa 10.90 5.00 0.32 1.134 0.891
and the theoretical activation energy calculation, it is believed
((CH3 )3 Si)2 Ob 10.88 6.14 0.0 1.516 0.807
that the concerted mechanism is unlikely. CH3 OHa 12.04 6.20 1.83 −0.013 0.634 0.335
Theoretical details according to the sequential reactions [3] CH3 (CH2 )3 OHa 11.80 6.26 1.66 0.156 0.675 0.335
and [4] are now described. The relatively high reactivity of (CH3 )3 COHa 11.56 5.65 1.63 0.257 0.651 0.330
HMDS with silanol groups is in part due to the high proton- TMAc 5.56 2.34 0.55 −0.154 0.390
TEAc 5.47 2.52 0.53 0.033 0.400
acceptor capability of N in HMDS. Figure 1 shows the HOMO
localized on the nitrogen atom of HMDS. Data in Tables 1 and 2 Note. Basis sets are a 6-31G(d,p), b 3-21G(d), and c B3LYP/6-31G(d,p)/
indicate that the Si–N bond is both highly polar and polarizable, 6-31G(d,p); µ is the dipole moment; q is atomic charges. MADES is methacry-
which can also contribute to HMDS reactivity. loyoxymethylenemethyl diethoxysilane; MTES is 3-methcryloyloxypropyltri-
Figure 1 illustrates that the LUMO in the prereaction com- methoxysilane.
plex is localized in the same region as the HOMO, and this
localization does not change in the transition state (Fig. 2). The

FIG. 1. Structure of the prereaction complex of HMDS (hydrogen bonding


to ≡SiOH group) in reaction [2] and corresponding electrostatic potential map FIG. 2. The structure of the transition state and corresponding electrostatic
and localization of the boundary MO. potential distribution and localization of the boundary MO in the TS.
160 GUN’KO ET AL.

TABLE 2
Parameters of Compounds with Different Si–X Bonds

−E HOMO E LUMO α µ
Compound (eV) (eV) (a.u.) (D) qSi −qX Basis set

(H3 Si)2 NH 10.62 3.56 0.91 0.918 1.000 6-31G∗∗


(H3 Si)2 NH+2 18.24 −2.22 1.86 0.784 0.919 6-31G∗∗
(H3 Si)2 NH 6.99 0.17 0.89 0.564 0.753 B3LYP/6-31G∗∗ /6-31G∗∗
(H3 Si)2 NH 7.00 0.13 0.90 0.568 0.759 B3LYP/6-31G∗∗
(H3 Si)2 NH 10.60 3.52 46.8 0.90 0.920 1.010 MP2/6-31G∗∗
H3 SiNH2 10.68 3.89 24.1 1.30 0.752 0.845 6-31G∗∗
H3 SiNH2 6.85 0.63 28.5 1.41 0.551 0.729 B3LYP/6-31G∗∗ /6-31G∗∗
H3 SiOH 12.33 3.66 21.9 1.50 0.988 0.754 6-31G∗∗
H3 SiOH 8.21 0.17 24.4 1.46 0.637 0.613 B3LYP/6-31G∗∗ /6-31G∗∗
H3 SiOH 12.37 3.58 23.1 1.58 0.989 0.767 MP2/6-31G∗∗
H3 SiOCH3 7.64 0.45 33.6 0.80 0.449 0.533 B3LYP/6-31G∗∗ /6-31G∗∗
AM1 Calculations
−E HOMO E LUMO α β
Compound (eV) (eV) (a.u.) (a.u.) qSi −qX

(H3 Si)2 NH 9.75 1.43 40.4 74 0.980 0.858


H3 SiNH2 9.65 1.82 22.3 105 1.043 0.725
H3 SiOH 11.29 1.24 20.2 60 1.010 0.575
H3 SiOCH3 10.83 1.56 29.8 66 0.979 0.497

Note. α is the average polarizability; β is the average second-order polarizability along the dipole moment.

map of electrostatic potential in the prereaction complex and the atoms participating in the reaction (Figs. 1 and 2). E HOMO
the TS shows a strong electrostatic interaction between reac- is 0.8 eV higher and atomic charges of O, H, and Si are greater,
tants. This suggests a large electrostatic contribution for both but qN is lower than those for the prereaction complex. These
hydrogen bonding and the TS. This is not surprising since it is results are generally consistent with the ab initio calculations
found that 80% of the hydrogen bond strength between two wa- (Table 3), except that qN was calculated to be larger in the TS
ter molecules is from electrostatic contributions as calculated by than in the prereaction complex.
the Kitaura–Morokuma method (11). Additional evidence that The transition states of reactions [3] and [4] were modeled
electrostatic contributions play a large role in the HMDS reac- using ab initio calculations. (H3 Si)2 NH and H3 SiNH2 were
tion is given in Table 3. The atomic charges in TS-1 are higher used instead of hexamethyldisilazane and trimethylaminosilane
than those in the prereaction complex. Using PM3, E HOMO and for simplicity. A comparison of the energy of formation of
E LUMO in both the prereaction complex and the transition state hydrogen-bonded prereaction complexes and E ‡ values for reac-
of the HMDS reaction with a silanol (TS-1) are localized on tions [3] and [4] shows that trimethylaminosilane may be more

TABLE 3
Energy of the Boundary MO and Atomic Charges (q) in Prereaction Complex and TS

E HOMO E LUMO
Complex (eV) (eV) −qN −qO qSi qH Basis set

(H3 Si)2 N(H)· · ·HOSiH3 11.15 3.02 0.954 0.761 0.769 0.386 6-31G∗∗
(H3 Si)2 N(H)· · ·HOSiH3 7.57 −0.29 0.803 0.642 0.573 0.336 B3LYP/6-31G∗∗ /6-31G∗∗
(H3 Si)2 N(H)· · ·HOSiH3 11.24 2.96 0.852 0.690 0.639 0.371 MP2/6-31G∗∗
TS-1 9.98 3.17 1.145 0.909 1.093 0.484 6-31G∗∗
TS-1 6.63 −0.23 0.885 0.677 0.689 0.390 B3LYP/6-31G∗∗ /6-31G∗∗
TS-1 10.55 3.08 1.073 1.031 1.032 0.493 MP2/6-31G∗∗
H3 SiNH2 · · ·HOSiH3 11.52 3.29 0.894 0.764 0.762 0.405 6-31G∗∗
H3 SiNH2 · · ·HOSiH3 7.66 0.05 0.769 0.642 0.567 0.348 B3LYP/6-31G∗∗ /6-31G∗∗
H3 SiNH2 · · ·HOSiH3 11.60 3.37 0.959 0.820 0.908 0.434 MP2/6-31G∗∗
TS-2 10.38 3.31 0.878 1.033 0.984 0.471 6-31G∗∗
TS-2 6.49 −0.14 0.721 0.791 0.614 0.386 B3LYP/6-31G∗∗ /6-31G∗∗
TS-2 10.67 3.52 0.913 1.032 1.001 0.484 MP2/6-31G∗∗

Note. The second basis set (if it is shown) was used for the geometry optimization.
MECHANISM/KINETICS HMDS REACTION SILICA SURFACE 161

TABLE 4 These calculations include the reactants and transition state


Energy of Formation of Prereaction Complex (Hydrogen Bonding) energies in consideration of solvent effects. The solvation energy
and Activation Energy of Reactions [1] (TS-1) and [2] (TS-2) of a reagent includes the solvent free energy, solvent internal en-
ergy, electrostatic interaction, cavitation energy, dispersion free
Complex 1E t (kJ/mol) Basis set
energy, and repulsion free energy. The stabilization of TS-1 may
(H3 Si)2 N(H)· · ·HOSiH3 −22 B3LYP/6-31G(d,p)/6-31G(d,p) be higher than the model molecules studied (H instead of CH3 ),
(H3 Si)2 N(H)· · ·HOSiH3 −27 MP2/6-31G(d,p)/6-31G(d,p) but TS-1 stabilization is relatively low, even for water (Table 5).
TS-1a 152 PM3 Furthermore, TS-1 destabilization is observed for other solvents,
TS-1 (small model) 135 PM3
resulting in increased E ‡ values for TS-1.
TS-1 161 6-31G(d,p)
TS-1 113 B3LYP/6-31G(d,p)/PM3 For TS-2, reaction of trimethylaminosilane with silanol, sta-
TS-1 93 B3LYP/6-31G(d,p)/6-31G(d,p) bilization is observed and 1E ‡ increases with growing solvent
TS-1 91 MP2/6-31G(d,p)/6-31G(d,p) polarity. Thus, the first stage (reaction-rate-determining TS-1) is
TS-1 67 MP2/6-31G(d,p) inhibited in polar solvents due to a lower stabilization of the TS
H3 SiNH2 · · ·HOSiH3 −19 6-31G(d,p)
than the prereaction complex (reactants), the opposite occurs in
H3 SiNH2 · · ·HOSiH3 −34 B3LYP/6-31G(d,p)/6-31G(d,p)
H3 SiNH2 · · ·HOSiH3 −34 MP2/6-31G(d,p)/6-31G(d,p) the second stage (TS-2) in the stepwise mechanism. These find-
H3 SiNH2 · · ·HOSiH3 −42 MP2/6-31G(d,p) ings are qualitatively consistent with a recent report by Jones
TS-2 88 6-31G(d,p) et al. (12) who report that HMDS reactions with silica per-
TS-2 57 B3LYP/6-31G(d,p)/6-31G(d,p) formed in more polar solvents are inhibited compared to the
TS-2 50 MP2/6-31G(d,p)/6-31G(d,p)
same reaction in less polar solvents. This lends credence to the
TS-2 47 MP2/6-31G(d,p)/MP2/6-31G(d,p)
calculations on solvent effects.
a R = CH3 , SiO2 cluster with 12 polyhedra. The influence of solvents on silylation can be linked to more
than the free energy change of the TS relative to the prereaction
reactive than HMDS (Table 4). Thus, reaction [3] may be the complex (Table 5). A decrease in reagent molecule diffusion
rate limiting step in the reaction of HMDS with the silica sur- can result with an increasing solvent dielectric constant (polar-
face. A model of reaction [3], (using (H3 Si)2 NH and H3 SiOH) ity) due to strong intermolecular solvent/reagent interactions. A
indicates it is exothermic (1E t = −46 kJ/mol using the 6-31 polar solvent may also enable the formation of a charged reactant
G(d,p) basis set). to be formed, e.g., due to proton transfer:
The ab initio calculations of activation energy show a rela- ≡SiOH + [solvent] + HN(Si(CH3 )3 )2
tively small difference between reaction [3] and the reaction of
→ ≡SiO− + [solvent] + H+ NH(Si(CH3 )3 )2 . [5]
a model methoxyfunctional silane (H3 SiOCH3 ), ∼140 kJ/mol
by 6-31 G(d,p)). Since the calculated activation energy of reac- In this case TS stabilization may be lower than the charged re-
tion [3] or that of more simple model compounds by the PM3 actants (one or two), as the charge can be more delocalized in
method agree with the more rigorous calculations (Table 4), the the TS. The probability of the process depicted in Eq. [5] is
adequacy of the simpler models in this case is demonstrated. increased with increasing solvent polarity, resulting in a lower-
Including electron correlation effects on the TS (optimized by ing of the reaction rate in polar solvents. These two effects can
the synchronous transit-guided quasi-Newton method and an be more important than the activation energy change for reac-
eigenvalue-following optimization of the TS structure correction tion between intact reactants due to TS stabilization by solvent
using the 6-31 G(d,p) basis set or PM3) utilizing the B3LYP/ molecules since HMDS is a fairly reactive silylating reagent and
6-31G(d,p) basis set in DFT or TS searching using the MP2/6-31 the reaction can be diffusion limited.
G(d,p) basis set in DFT or TS searching using MP2/6-31G(d,p)
gradients in HF calculations gives a decrease in activation en-
ergy in very close agreement with Hertl and Hair’s experimental TABLE 5
value of 77 kJ/mol (Table 4). Theoretically determined acti- The Changes in the Activation Energy (kJ/mol) of Silylation
vation energies for reactions [3] and [4] are ∼70 and ∼40 kJ, re- / − E =/ ) Due to Solvation (6-31G∗∗ )
(∆E =/ = E =sol gas
spectively. The stepwise mechanism will thus be used to model
the experimentally derived kinetic data of the HMDS reaction Solvent TS-1a TS-1b TS-2b εd
with fumed silica in a later section.
Cyclohexane 22 6 −3 2.02
Previously reported kinetics studies of HMDS reaction with Benzene 21 5 −6 2.28
fumed silica (6) have been in the vapor phase, so the theoretical Toluene 22 −5 2.38
treatment has thus far not included solvation effects. The ex- Acetone 24 20.7
perimental data reported herein involve reactions in an organic Acetonitrile 33 15 −10 36.2
(toluene) solvent slurry. Thus, a theoretical study of the interac- Water −16c 78.5
tion between HMDS and silica was studied in different solvents Note. The geometry was calculated using a MP2/6-31G∗∗ and b 6-31G∗∗ .
utilizing capabilities provided by a new version (May, 1998) of c TS-1 was calculated using AM1-SM1 R = CH3 , SiO2 cluster with 12 poly-
the GAMESS program package. hedra. d The dielectric constants (ε) from Ref. (14).
162 GUN’KO ET AL.

TABLE 6 where Cim iv


is the MO coefficients, v is the orbital type of atom
Atomic Charges for Reactants and TS with Consideration i (m ∈ v), El is the orbital energy, l is the orbital number, and
of Solvation (6-31G∗∗ ) σ 2 is the dispersion of the state populations estimated from the
Solvent −qN −qH −qO −qSi State
ESCA or Roentgen emission spectra as a half-width. The ρiv (E)
values are more informative and appropriate to analyze the sys-
Cyclohexane 0.9043 0.3628 0.7361 0.7762 Reactants-1 tem states than the separated eigenvalue (El ) and eigenvector
Benzene 0.9053 0.3642 0.7381 0.7771 Reactants-1 iv
(Cim ) values.
Toluene 0.9201 0.3673 0.7546 0.7807 Reactants-1
The local density of electronic states (LDS) calculations of
Acetonitrile 0.9126 0.3815 0.7756 0.7796 Reactants-1
0.9996 0.3530 0.7544 0.9191 Reactants-1 atoms in the prereaction complexes (hydrogen bonding) and
Cyclohexane 1.0696 0.4828 0.8602 0.9605 TS-1 TS-1 in different solvents show that differences in the LDS of
Benzene 1.0704 0.4833 0.8605 0.9615 TS-1 N, O, H, and Si are less dependent on solvent than that between
Toluene 1.0011 0.4903 0.9888 0.9219 TS-1 the prereaction and the transition states (Fig. 3). One might ex-
Acetonitrile 1.0696 0.4862 0.8591 0.9576 TS-1
pect to detect a shift of N and Si LDS to the valence zone top
1.1447 0.4843 0.9087 1.09331 TS-1
Cyclohexane 0.8623 0.3628 0.7361 0.7613 Reactants-2 (E v ) in TS-1 compared with that of the prereaction complex.
Benzene 0.8641 0.3642 0.7381 0.7622 Reactants-2 The LDS intensity of O and H in TS-1 increases near E v , but
Toluene 0.8648 0.3648 0.7389 0.7621 Reactants-2 a shift of LDS peaks occurs at the bottom of the valence band
Acetonitrile 0.8797 0.3815 0.7756 0.7638 Reactants-2 corresponding to changes in the atomic charges (Table 6). These
0.8452 0.3530 0.7544 0.7521 Reactants-2
LDS changes of reactive atoms can influence the activation en-
Cyclohexane 0.8933 0.5176 0.9410 0.7773 TS-2
Benzene 0.8940 0.5171 0.9416 0.7786 TS-2 ergy as silylation occurs not only under charge control (electro-
Toluene 0.8942 0.5167 0.9416 0.7788 TS-2 static contribution to the interaction energy is large (Table 7))
Acetonitrile 0.8974 0.5078 0.9457 0.7863 TS-2 but also under orbital control. Charge transfer and the charac-
0.8779 0.4708 1.0329 0.9839 TS-2 teristics of the Si ← O bond in the TS depend on MO structure.
The formation of this bond in the TS causes the shoulder in the
Si–LDS near the top of the valence band (Fig. 3). This effect can
Additional data shown in Table 6 complement the previous be considered as evidence of at least partial orbital control of the
discussion concerning the role of solvents in the stepwise reactions studied. By and large, the relatively low influence of
mechanism. The differences in the atomic charges between the solvents studied on the electron density of reactive bonds can
reactants and TS-1 for solvents studied are frequently greater be indirect evidence that a diffusion impact inhibiting reaction
(with the exception of qH ) than for those states without solvent. increases with increasing solvent polarity.
However, the polarity of reactive bonds is higher for the gas In terms of the kinetic modeling detailed in the next section,
phase (Table 6). It seems likely that the changes in qH can the major conclusions of the theoretical study can be summa-
be more important with respect to the influence of solvents rized. The mechanism of reaction between (R3 Si2 )NH, where
on reaction, as proton transfer is the reaction-limiting stage. R = CH3 or H, and silica surface silanols shows that the reaction
Therefore, a reduction in qH in TS-1 and bond polarity due occurs through two stages according to the stepwise mechanism
to the solvent can correspond to a decreased reaction rate of depicted in Eqs. [3] and [4]. The first corresponds to reaction
Eq. [3] in these solvents. At the same time, the reaction rate of of (R3 Si2 )NH with ≡SiOH and the second is interaction be-
Eq. [4] can increase due to solvent influence. tween aminosilane R3 SiNH2 and another ≡SiOH group. The
The influence of solvents on TS-2 (Table 6) cause an increase first stage is reaction rate determining, as its activation energy is
in the charges of H and N, but the qO and qSi values are larger greater than E ‡ for the second stage. Also, polar solvents inhibit
without solvent. These changes are favorable for H+ transfer
from O → N due to a decrease in the electrostatic interaction
between H and O and an increase in the attractive interaction TABLE 7
between H and N. Analysis of the energy components in the hy- The Energy Interaction Components (kJ/mol) for Hydrogen
Bonding SiOH···NH2 SiH3 and TS-2 According to the Kitaura–
drogen complex and TS-2 using the Kitaura–Morokuma method
Morokuma Method (6-31G∗∗ )
(Table 7) shows that the main portion of the activation energy
of silylation is linked with the elongation of the SiO–H bond Energy components Hydrogen bonding TS-2
and strong electron–electron interaction in the deformed bond
system in the TS. These findings together suggest highly favor- Electrostatic −41 −233
Exchange 28 407
able conditions for the reaction of trimethylaminosilane with
Polarization −5 −110
silanol. Local density of the electron states (LDS) of atoms (and Charge transfer −9 −248
orbital LDS) can be calculated on the basis of the eigenvalues High-order coupling −0.3 71
and eigenvectors of the Hamiltonian of the system as follows, 1E −27 −112
1E BSSE −22 −100
X¡ ¢ Deformation 13 (amine)
ρiv (E) = (2πσ 2 )−0.5 iv 2
Cim exp[−(E − El )/2σ 2 ] 198 (silanol)
l,m
MECHANISM/KINETICS HMDS REACTION SILICA SURFACE 163

FIG. 3. LDS (valence zone of N (a), H (b), O (c), and Si (d) atoms) and total density of electronic states (e) for the TS-1 and reactants (R) in the gas phase
(TS) and different solvents: acetonitrile (A), benzene (B) cyclohexane (C) calculated at the 6-31G(d,p) level.

the first-stage (rate-limiting) reaction and accelerate the second A different approach is to analyze the amount of organosilane
stage-reaction. reacted with the silica surface over time. The reaction kinet-
ics of trimethylchlorosilane (14) and recently aminopropyltri-
Kinetics analysis. Previous workers have experimentally in- ethoxysilane (15) with silica have been studied in this manner.
vestigated the kinetics of organosilane reactions with silica sur- In general, one detects a rapid initial uptake followed by a slower
faces. The classic work of Hair and Hertl (6, 13) utilized infrared reaction over time. Shimizu et al. (15) in analyzing their kinetic
spectroscopy to study vapor-phase reactions by monitoring the data postulated three separate reaction processes assuming three
consumption of the non-hydrogen-bonded silanol absorbance at first-order reactions with respect to aminosilane.
∼3745 cm−1 . These workers determined activation energies for Previous reports have not explicity addressed silica sur-
a variety of surface silylation reactions and linearized their data face heterogeneity. There exists on silica surfaces both non-
using standard methods to determine reaction order with respect hydrogen-bonded and hydrogen-bonded silanols, the former
to non-hydrogen-bonded silanols. giving rise to an infrared absorbance at ∼3745 cm−1 and the
164 GUN’KO ET AL.

latter a broader absorbance at lower frequency. Furthermore, the surface Si), hydrogen-bonded or vicinal silanols (OH· · ·), and
non-hydrogen-bonded silanol peak has contributions from (1) geminal silanols HOOH.
completely noninteracting single silanols, (2) very weakly in- These surface sites can react with HMDS as explained by the
teracting silanols giving rise to a low-frequency shoulder (16), stepwise mechanism: (a) HMDS adsorbs onto a free or isolated
and (3) weakly and noninteracting geminal or disilanols (17). silanol to form an ((CH3 )3 Si)2 NH· · ·OH complex in which the
These surface groups may behave differently with respect to nitrogen atom of the HMDS molecule hydrogen bonds with the
HMDS adsorption and subsequent surface silylation. hydrogen atom of the hydroxyl group on the surface. The rate
To perform a kinetic analysis of the HMDS silylation reaction of adsorption is described by the rate constant ka and the rate of
cognizant of silica surface structure, one must obtain the total desorption by kd :
number of silanol sites available for adsorption/reaction and the ka
relative percentages of the different types of surface species. Us- ((CH3 )3 Si)2 NH + OH *
) ((CH3 )3 Si)2 NH· · ·OH. [6]
kd
ing solid-state 1 H NMR, Liu and Maciel (18) reported a value of
2.8 ± 0.3 SiOH/nm2 on Cab-O-Sil HS-5, consistent with values The complex can then react to form the stable surface product,
obtained by other methods (20). These workers (18) also found ≡Si–O–Si(CH3 )3 and trimethylaminosilane:
that approximately one-third of the silanols are not susceptible to k1
D2 O exchange and thus are not of interest as potential adsorp- ((CH3 )3 Si)2 NH· · ·OH → (CH3 )3 SiO + (CH3 )3 SiNH2 . [7]
tion/reaction sites. Application of this data for this 325 m2 /g If the availability of reactive non-hydrogen-bonded silanols is
silica suggests the total silanol content available for adsorption low, (CH3 )3 SiNH2 may either adsorb onto a hydrogen-bonded
and reaction is approximately 3.5 µmol/m2 . silanol or react with another (CH3 )3 SiNH2 to reform HMDS
The relative percentages of non-hydrogen-bonded and liberating NH3 (Eq. [8]).
hydrogen-bonded silanols were determined experimentally. An
HMDS reaction with silica carried out at room temperature in the k2
2(CH3 )3 SiNH2 → ((CH3 )3 Si)2 NH + NH3 . [8]
absence of moisture is believed to react primarily if not exclu-
sively with non-hydrogen-bonded silanols (6, 20). Quantitative (b) HMDS can adsorb onto a geminal silanol site with the
characterization of the number of trimethylsilyl groups bonded equilibrium process being similar to that in Eq. [6] for single
to the surface after HMDS reaction for 4 h was obtained by FT-IR silanols:
spectroscopy. Ninety percent of the non-hydrogen-bonded kag1
silanols had reacted under these conditions. Using the total num- ((CH3 )3 Si)2 NH + HOOH ((CH3 )3 Si)2 NH· · ·HOOH. [9]
ber of available silanols as 3.5 µmol/m2 , plus the experimen- kdg1

tally determined number of trimethylsilyl groups reacting with Although the single N atom in HMDS can strongly hydro-
90% of the silanols, it was calculated that 45% of the available gen bond with only one hydroxyl group, the presence of
silanols (1.56 µmol/m2 ) exists as non-hydrogen-bonded silanols ((CH3 )3 Si)2 NH adsorbed onto one hydroxyl group may steri-
and the remaining 55% (1.94 µmol/m2 ) exists as hydrogen- cally hinder the second hydroxyl group from readily interacting
bonded silanols. This result is in general agreement with others with another ((CH3 )3 Si)2 NH or a (CH3 )3 SiNH2 intermediate.
(21) who suggest that fumed silica consists of approximately The silanol that is interacting with the N atom of HMDS is
equal numbers of non-hydrogen-bonded and hydrogen-bonded first converted to a product and trimethylaminosilane, the re-
silanols. Of the 1.56 µmol/m2 non-hydrogen-bonded silanol active intermediate, is produced. The second hydroxyl remains
content, previously reported 29 Si solid-state NMR of this silica unattached at this point as indicated by Eq. [10]:
(18) has shown that approximately 30% exists as geminal silanol
k1
pairs. These values, along with the initial amount of HMDS ((CH3 )3 Si)2 NH· · ·HOOH → (CH3 )3 SiOOH + (CH3 )3 SiNH2 .
added, are used as the starting conditions in the kinetic model, [10]
which will now be discussed. These values are summarized in
Table 8. The rate constant for the above reaction should be similar to
HMDS, ((CH3 )3 Si)2 NH, may adsorb onto three types of sur- that of Eq. [7] since the adsorbed species is hydrogen bonded to
face silanols: free or isolated silanols (referred to as OH, the only a single silanol site. The intermediate (CH3 )3 SiNH2 may
oxygen atom is underlined, indicating that it is attached to the subsequently react with the hydroxyl group released from the
geminal site or find another unattached (CH3 )3 SiNH2 to form
((CH3 )3 Si)2 NH according to Eq. [8]. Reaction of the 2nd in the
TABLE 8
geminal silanol pair may more readily occur with the less bulky
Summary of Silica Surface Starting Conditions
trimethylaminosilane intermediate than with HMDS (Eq. [11]).
in the Model of Kinetic Data
(CH3 )3 SiNH2 + (CH3 )3 SiOOH
Total silanol content 3.5 µmol/m2
kag2
Non-hydrogen-bonded silanol content 1.56 µmol/m2 (CH3 )3 SiOOH· · ·H2 NSi(CH3 )3 . [11]
Hydrogen-bonded silanol content 1.94 µmol/m2 kdg2
Geminal or disilanol content 0.47 µmol/m2
Adsorption is followed by reaction analogous to Eq. [7], leading
MECHANISM/KINETICS HMDS REACTION SILICA SURFACE 165

to product formation: d(AOOH· · ·H2 NA)


= kag2 [ANH2 ](AOOH)
k1 dt
(CH3 )3 SiOOH· · ·H2 NA → (CH3 )3 SiOOSi(CH3 )3 + NH3 . − kdg2 (AOOH· · ·H2 NA)
[12]
− k1 (AOOH· · ·H2 NA), [21]
The total product formed is a sum of the amounts of (CH3 )3 SiO,
(CH3 )3 SiOOH, and (CH3 )3 SiOOSi(CH3 )3 produced at any time d(OH· · ·)
= −kah [A2 NH](OH· · ·) + kdh (A2 NH· · ·OH· · ·),
during the reaction. It is not possible to distinguish the dif- dt
ferent product species; experimental data refer to total product [22]
formation.
(c) HMDS can adsorb onto hydrogen-bonded silanols to form d(A2 NH· · ·OH· · ·)
the adsorption complex ((CH3 )3 Si)2 NH· · ·OH· · ·. Rate con- = kah [A2 NH](OH· · ·)
dt
stants for the forward and reverse adsorption processes are kah − kdh (A2 NH· · ·OH· · ·), [23]
and kdh , respectively:
kah d(AOOH)
((CH3 )3 Si)2 NH + OH· · · ((CH3 )3 Si)2 NH· · ·OH· · ·. [13] = k1 (A2 NH· · ·HOOH) − kag2 [ANH2 ](AOOH)
kdh dt
The rate of disappearance or appearance of various species in- + kdg2 (AOOH· · ·H2 NA), [24]
volved in the above mechanism (Eqs. [6]–[13]) can be rep-
resented as follows in terms of the corresponding rate equa- d(AOOA)
= k1 (AOOH· · ·H2 NA), [25]
tions. For mathematical convenience the trimethylsilyl group, dt
Si(CH3 )3 , has been replaced with A in these equations:
where [X ] is the solution concentration of X (mol/L) at time t
d[A2 NH] and (Y ) is the number of moles of Y on the surface at time t.
= −ka [A2 NH](OH) + kd (A2 NH· · ·OH) Because the differential equations describing the various reac-
dt
tants, products, and transient species are coupled, they cannot be
+ k2 [ANH2 ]2 − kag1 [A2 NH](HOOH)
solved independently or by any simple analytical method. How-
+ kdg1 (A2 NH· · ·HOOH)− kah [A2 NH](OH· · ·) ever, the various concentration profiles can be accurately calcu-
lated from known initial conditions and specified rate constants
+ kdh (A2 NH· · ·OH· · ·), [14]
by numerically integrating the corresponding finite difference
d(OH) equations. For example, the differential change in concentration
= −ka [A2 NH](OH) + kd (A2 NH· · ·OH), [15] of A2 NH can be approximated as
dt

d(A2 NH· · ·OH) d[A2 NH] [A2 NH]i+1 − [A2 NH]i


= ka [A2 NH](OH) − kd (A2 NH· · ·OH) ≈ [26]
dt dt δt
− k1 (A2 NH· · ·OH), [16]
and Eq. [14] can then be replaced by
d[ANH2 ]
= k1 (A2 NH· · ·OH) − 2k2 [ANH2 ]2 [A2 NH]i+1
dt
©
+ k1 (A2 NH· · ·HOOH) − kag2 [ANH2 ](AOOH) ≈ [A2 NH]i + −ka [A2 NH]i (OH)i + kd (A2 NH· · ·OH)i
+ kdg2 (AOOH· · ·H2 NA), [17] + k2 [ANH2 ]i2 − kag1 [A2 NH]i (HOOH)i

d(HOOH) + kdg1 (A2 NH· · ·HOOH)i − kah [A2 NH]i (OH· · ·)i
= −kag1 [A2 NH](HOOH) + kdg1 (A2 NH· · ·HOOH), ª
dt + kdh (A2 NH· · ·OH· · ·)i δt. [27]
[18]
Similar finite difference expressions can be written for
d(A2 NH· · ·HOOH) Eqs. [15]–[25]. We find these numerical solutions converge to
= kag1 [A2 NH](HOOH) three significant figure precision by dividing the observed time
dt
− kdg1 (A2 NH· · ·HOOH) domain into 3000 equally spaced intervals of δt = 4.8 s. We have
developed interactive LabView software (National Instruments)
− k1 (A2 NH· · ·HOOH), [19] that uses the finite difference equations and the initial conditions
(Table 8) to numerically integrate the rate equations and display
d(AO) both the calculated and experimentally measured concentration
= k1 (A2 NH· · ·OH), [20]
dt profiles along with standard deviations of the corresponding fits.
166 GUN’KO ET AL.

TABLE 9 is a plot of the data for the “total intermediate” which contains
Experimental Data for the HMDS Reaction with Fumed Silica contributions from both adsorbed HMDS and trimethylaminosi-
lane. The curves shown in Figs. 4 and 5 result from the kinetic
Reaction HMDS solution Trimethylsilyl (TMS) Total
time concentration groups grafted intermediate
model used to fit the experimental data. The data were generated
(s) (mol/L) (µmol/m2 ) (µmol/m2 ) from infrared spectra such as those shown in Figs. 6 and 7.
The characteristics of the data must be accounted for by the
0 0.0262 0 0 kinetic model. As has been seen in previous work with organosi-
31 0.0230 0.49 (14s) 0.24 lane reactions, a rapid initial reaction is followed by a slower
60 0.0215 0.51 0.46
300 0.0185 0.59 0.90
reaction with the surface (15). With an initial rapid loss (ap-
600 0.0140 0.89 1.43 proximately 38% after 5 min of reaction), followed by a level-
1200 0.0119 1.02 1.69 ing off and longer reaction times, a small increase in the amount
2400 No data 1.24 No data of HMDS in solution is found. The quantity of HMDS unac-
3600 0.0108 1.32 1.71 counted for as either surface product or solution species (total
7200 0.0122 1.33 1.48
14,400 0.0128 1.37 1.37
intermediate) exhibits a rapid initial increase followed by a slow
162,000 1.55 decay at longer reaction times. Spectra in Fig. 6 show a rapid
initial reduction of the non-hydrogen-bonded silanol peak with
concomitant increase in Si(CH3 )3 groups attached to the surface.
The program provides a graphics user interface equipped with After 4 h of reaction, unreacted non-hydrogen-bonded silanols
interactive controls that allows the user to independently adjust are still detected, but these are completely consumed after 45 h
and optimize the value of the unknown rate constants to achieve (Fig. 7). The rapid initial loss of non-hydrogen-bonded silanols
minimum standard deviations for each concentration profile. corresponds with the rapid initial uptake of TMS groups (Figs. 4
Experimental data for the HMDS reaction is shown in Table 9. and 6); the slower uptake of TMS groups at longer reaction times
The amount of HMDS remaining in solution as a function of time corresponds with the loss of remaining non-hydrogen-bonded
was obtained directly from solution infrared data; TMS groups silanols (Figs. 4 and 7). The spectra thus support the contention
grafted were obtained directly from infrared spectroscopy of that even the slow reaction occurring at longer times should
the silica after drying. The column termed total intermediate be explained in terms of reaction with non-hydrogen-bonded
was obtained by determining how much HMDS was lost from silanols. The kinetic model must not only fit the experimental
solution at a given time (calculated by the difference [HMDS data described but also provide reasonable insight into the nature
(t = 0) − HMDS (t = t)]) and subtracting from that number half of these reactions consistent with the spectral information.
the amount of TMS groups found bonded to the surface. That Reasonable fits to the experimental kinetic data for both
difference is the amount of HMDS unaccounted for in solution HMDS adsorption and reaction (Fig. 4) could only be obtained
or on the surface. Data for the amount of HMDS in solution and using rate constants dictating that the reaction is adsorption rate
TMS groups bonded to the surface are plotted in Fig. 4; Fig. 5 limited. This is in agreement with work reported by Procter and

FIG. 4. Experimental data and results of kinetic fits. (A) Fractional coverage of all available silanols (θ) with trimethylsilyl groups. (B) HMDS solution
concentration in mol/L.
MECHANISM/KINETICS HMDS REACTION SILICA SURFACE 167

FIG. 5. Experimental data and results of kinetic fits. (A) Fractional coverage of all available silanols (θ) with intermediate species comprised of (B) HMDS
adsorbed on hydrogen-bonded silanols and (C) trimethylaminosilane.

Blitz (22), showing that the vapor-phase reaction of HMDS with (Eq. [10]) and trimethylaminosilane reaction with the 2nd of a
silica gel is considerably faster than the same reaction carried geminal silanol pair (Eq. [12]) are all set equal to k1 . This sig-
out in an organic solvent slurry. The rate constant k1 for HMDS nificantly minimizes the number of adjustable parameters in the
reaction is much larger than the adsorption rate constants. Thus, model restricting its’ ability to fit the experimental data. Since
Eqs. [6], [9], and 11 may be considered irreversible. Further- excellent fits were obtained with these restrictions (Figs. 4 and
more, since adsorption is rate limiting, the rate constants for 5), this lends credence to the kinetic model and interpretation.
reaction of HMDS with a single non-hydrogen-bonded silanol Spectra in Fig. 6 clearly illustrate the rapid initial reaction
(Eq. [7]), HMDS reaction with the 1st of a geminal silanol pair of non-hydrogen-bonded silanols. Since the HMDS adsorption

FIG. 6. Transmission infrared spectra of unreacted (A), 12-s reacted (B) and 1-h reacted (C) Cab-O-Sil HS-5 with HMDS.
168 GUN’KO ET AL.

FIG. 7. Transmission infrared spectra of 4-h reacted (A) and 45-h reacted (B) Cab-O-Sil HS-5 with HMDS.

rate constant (ka = 0.102 L/mol · s) is much less than the rate group. Previous workers have shown that geminal silanols are
constant value for reaction (k1 = 0.32 s−1 ), very little adsorbed highly reactive to silylating reagents (23). Work reported by
but unreacted ((CH3 )3 Si)2 NH· · ·OH ever exists, even though Sindorf and Maciel on HMDS reaction with a silica gel sur-
the HMDS/non-hydrogen-bonded silanol ratio is approximately face using solid-state 29 Si NMR spectroscopy has shown that
4/1. In this early stage of the reaction the concentration of geminal silanol sites are more reactive than single-silanol sites
trimethylaminosilane intermediate increases since it is rapidly (24). This fast reaction may be due to reaction of the first of
formed as the product of HMDS reaction with a non-hydrogen- the geminal silanol pair since this would result in loss of the
29
bonded silanol. After approximately 20 min of reaction a Si Q2 resonance. The slow reaction at longer times is due
large proportion of the non-hydrogen-bonded silanols are con- to the second, initially unreacted geminal silanol. Once HMDS
sumed. Given the fewer number of adsorption sites available for is adsorbed onto a geminal silanol, both hydroxyl groups be-
HMDS and (CH3 )3 SiNH2 , as well as the increasing amount of come unavailable for further reaction. One silanol leads to
(CH3 )3 SiNH2 present, the (CH3 )3 SiNH2 concentration begins to product formation (k1 = 0.32 s−1 ), contributing to the initial
slowly decrease as recombination of the intermediate to reform fast reaction. The second silanol remains sterically hindered by
HMDS and NH3 (Eq. [8] k2 = 6.8 mol−1 L−1 s−1 ) becomes sig- first the presence of adsorbed HMDS ((CH3 )3 Si)2 NH· · ·HOOH)
nificant. This explains the initial phase of the reaction as well as and then by product ((CH3 )3 SiOOH). The second unreacted
the initial increase and subsequent decrease in intermediate con- silanol can then adsorb either (CH3 )3 SiNH2 reactive inter-
centration (Fig. 5). The slow reaction at longer reaction times, mediate ((CH3 )3 SiOOH· · ·H2 NSi(CH3 )3 ) prior to reaction to
also detected by others for similar reactions (5b, 15), is now form (CH3 )3 SiOOSi(CH3 )3 (Eqs. [11] and [12] respectively) or
addressed. HMDS prior to reaction. The latter is unlikely due to steric con-
Infrared spectra indicate that the slow reaction at longer times straints. The rate constant for the adsorption of ANH2 onto the
is also due to a reaction with non-hydrogen-bonded silanols unreacted hydroxyl of the disilanol pair is 0.028 L/(mol s). This
as previously described. This slow reaction involves geminal kag2 value is approximately 10 times slower than the kag1 value
silanols. The adsorption rate constant of HMDS onto geminal for HMDS adsorption onto the first disilanol. The ramping por-
silanol sites (kag1 ) is found to be 1.4 × 10−3 m2 /(µmol s), ap- tion of the product concentration profile (Fig. 4) is controlled by
proximately 3 times faster than the adsorption of HMDS on kag2 , the adsorption of reactive intermediate (CH3 )3 SiNH2 onto
single-silanol sites. This may be due to the presence of two a geminal silanol site that already has one hydroxyl reacted to
adsorption sites per geminal silanol and/or enhanced electro- form a trimethylsilyl group.
static attraction resulting from hydrogen bonding between the Infrared spectra (Fig. 7) show that reaction of non-hydrogen-
HMDS nitrogen and one silanol plus an attractive interaction bonded silanols are responsible for the slow reaction at longer
between the oxygen atom of the second hydroxyl group and the times. Data in Table 9 shows that from 4 to 45 h of reaction
silicon atom of the trimethylsilyl group closer to the hydroxyl the product concentration increases from 1.37 to 1.55 µmol/m2 .
MECHANISM/KINETICS HMDS REACTION SILICA SURFACE 169

After 4 h of reaction quantitative analysis of the infrared data in- Inspection of Figs. 4 and 5 suggest that approximately 45% of
dicates that 90% of the non-hydrogen-bonded silanols has been available silanols is consumed during the course of the reaction.
reacted, evidenced by the small peak at ∼3745 cm−1 . After 45 h This is not coincidentally the same percentage of silanols pre-
the 3745-cm−1 peak has disappeared, indicating that this slow sumed to exist on the surface as non-hydrogen-bonded silanols.
ramping portion of the curve corresponds to reaction with non- Another 45% of available silanols is involved in HMDS ad-
hydrogen-bonded silanols. The number of hydroxyls that react sorption. Since all silanols involved with HMDS adsorption are
corresponding to the longer reaction time is qualitatively consis- hydrogen-bonded silanols, a significant percentage (≈25%) of
tent with the number of geminal silanols that exist on this fumed these groups does not adsorb HMDS. The number of hydrogen-
silica (18). The 1.55 µmol/m2 of TMS bonded to the surface bonded silanols accessible to HMDS is less than the number
after 45 h is in remarkable agreement with the 1.56 µmol/m2 of accessible to the smaller D2 O. Since hydrogen bonds between
non-hydrogen-bonded silanols that are assumed to exist. Since silanols at the surface of a single particle should be weak due
no more HMDS than this reacts in 45 h, these results are con- to surface topology, hydrogen-bonded silanols exist primarily
sistent with the contention that hydrogen-bonded silanols are between aggregates of neighboring primary particles. This ef-
unreactive under these conditions. Hydrogen-bonded silanols fect corresponds to large steric effects for hydrogen-bonded
still play a role as adsorption sites, however. silanols due to interparticle contact. This is evidenced by the
Experimental data in Table 9 shows that more HMDS is lost fact that the BET surface area of fumed silica is largely depen-
from solution than can be accounted for solely by reaction to dent on adsorbate size (25). Therefore, the low reactivity (6)
form trimethylsilyl groups. Figure 5 shows a plot of HMDS of hydrogen-bonded silanols can in part be ascribed to a steric
unaccounted for in solution or in the amount consumed by effect.
surface reaction, with fits for reactive intermediate concentra- Since adsorption of HMDS onto hydrogen-bonded silanols is
tion (previously discussed) and adsorbed HMDS concentration. not followed by fast reaction, as is the case for non-hydrogen-
The amount of adsorbed HMDS far exceeds the non-hydrogen- bonded silanols, desorption of HMDS can also occur. The des-
bonded silanol content, particularly at long reaction times where orption rate constant from hydrogen-bonded silanols (kdh ) helps
nearly all of these groups have been consumed. The HMDS to explain the data of HMDS solution concentration in Table 9.
is adsorbed onto non-hydrogen-bonded silanols, surface silox- The HMDS solution concentration exhibits a rapid initial de-
anes, trimethylsilyl groups, or some combination of the three. crease in the first hour of reaction followed by a gradual increase.
To determine whether siloxanes and surface-bound trimethylsi- This experimental observation, which is fitted well by the model,
lyl groups act as adsorption sites, a silica was prepared con- is explained by kdh and k2 , the rate constant for the recombina-
taining trimethylsilyl groups without silanols. The silica was tion of (CH3 )3 SiNH2 to form ((CH3 )3 Si)2 NH (HMDS). These
first heat treated to 600◦ C in air for 6 h to condense hydrogen- two rate processes lead to an increase in HMDS solution con-
bonded silanols and then reacted with HMDS for 48 h to react centration after the fast process involving product formation is
all non-hydrogen bonded silanols. The infrared spectrum of the complete.
dried product indicated a complete lack of silanols. This ma-
terial was then exposed to HMDS in toluene to determine the
extent of HMDS adsorption by solution infrared spectroscopy. SUMMARY
Only 2% of the HMDS was adsorbed onto the surface, indicat-
Theoretical investigations of the mechanism of reaction be-
ing that neither trimethylsilyl groups or siloxanes are adsorption
tween hexamethyldisilazane and silica suggest a two-step reac-
sites for HMDS. Thus, practically all of the HMDS unaccounted
tion. The first is reaction of hexamethyldisilazane with a silica
for must be adsorbed onto hydrogen-bonded silanols. The rate
surface silanol to form a surface-bound trimethylsilyl and a re-
constant for HMDS adsorption onto hydrogen-bonded silanols
active intermediate trimethylaminosilane. The second step in-
(kah , Eq. [13]) is similar to the adsorption rate constant for non-
hydrogen-bonded silanols (ka , Eq. [6]), both being second-order volves reaction of trimethylaminosilane with a surface silanol
rate constants. These values are given in Table 10. to form a trimethylsilyl surface species and ammonia. The cal-
culated activation energy of the first, rate-determining step is
in good agreement with the apparent experimental activation
TABLE 10 energy reported by Hertl and Hair (6).
Rate Constants for Various Processes in HMDS Reaction The two-step mechanism was used to build a model to describe
with Fumed Silica the kinetics of HMDS reaction with fumed silica derived from
experimental data. Since the mechanism necessitates solving
ka 0.102 (±0.008) L mol−1 s−1
kag1 0.22 (±0.06) L mol−1 s−1
coupled differential rate equations that cannot be solved inde-
kah 0.080 (±0.006) L mol−1 s−1 pendently or by analytical methods, finite difference equations
kdh 4.4 (±0.6) × 10−4 s−1 were used to numerically integrate the rate equations. Values of
k1 0.32 (±0.03) s−1 the unknown rate constants were adjusted to fit the integrated
k2 6.8 (±2.5) mol−1 L−1 s−1 concentrations to observed profiles by a least-squares criteria.
kag2 0.028 (±0.015) L mol−1 s−1
Calculated and experimentally measured concentration profiles
170 GUN’KO ET AL.

are simultaneously displayed along with interactive control of Challacombe, M., Peng, C. Y., Ayala, P. Y., Chen, W., Wong, M. W., Andres,
unknown rate constants. J. L., Replogle, E. S., Gomperts, R., Martin, R. L., Fox, D. J., Binkley, J. S.,
The entire reaction can be explained by reaction of non- Defrees, D. J., Baker, J., Stewart, J. P., Head-Gordon, M., Gonzalez, C.,
and Pople, J. A., “Gaussian 94, Revision.” Gaussian, Inc., Pittsburgh, PA,
hydrogen-bonded silanols, which is adsorption rate limited un- 1995.
der the conditions studied. The reaction is initially fast as non- 8. Schmidt, M. W., Baldridge, K. K., Boatz, J. A., Elbert, S. T., Gordon,
hydrogen-bonded single and disilanols are consumed. The slow M. S., Jensen, J. J., Koseki, S., Matsunaga, N., Nguyen, K. A., Su, S.,
reaction at longer times results from hindered adsorption of Windus, T. L., Dupuis, M., and Montgomery, J. A., J. Comput. Chem. 14,
1347 (1993).
a silylating reagent onto geminal silanols that already have
9. Stewart, J. J. P., J. Comput. Chem. 10, 209 (1989). (b) Stewart, J. J. P.,
one hydroxyl group silylated. While approximately half of the J. Comput. Chem. 10, 221 (1989).
hydrogen-bonded silanols available for D2 O adsorption serve 10. Dewar, M. J. S., Zoebisch, E. G., Healy, E. E., and Stewart, J. J., J. Am.
as HMDS adsorption sites, no detectable reaction occurs with Chem. Soc. 107, 3902 (1985).
these groups. 11. (a) Morokuma, K., J. Chem. Phys. 55, 1236 (1971). (b) Kitaura, K., and
Morokuma, K., Int. J. Quantum Chem. 10, 325 (1976). (c) Morokuma,
K., and Kitaura, K., in “Chemical Applications of Electrostatic Potentials”
ACKNOWLEDGMENTS (P. Politzer and D. G. Truhlar, Eds.), p. 215. Plenum, New York, 1982.
(d) Stevens, W. J., and Fink, W. H., Chem. Phys. Lett. 139, 15 (1987).
Thanks to Dr. Charles B. Little and Cabot Corp., Cab-O-Sil Division, for
(e) Cammi, R., Bonaccorsi, R., and Tomasi, J., Theor. Chim. Acta 68, 271
partial support of this work.
(1985). (f) Chen, W., and Gordon, M. S., J. Phys. Chem. 100, 14316 (1996).
12. Jones, F. R., Vedamuthu, M. S., and Blitz, J. P., in “Fundamental and Applied
REFERENCES Aspects of Chemically Modified Surfaces” (J. P. Blitz and C. B. Little, Eds.),
p. 173. Royal Society of Chemistry, Cambridge, U.K., 1999.
1. (a) Leyden, D. E., “Silanes, Surfaces and Interfaces.” Gordon and Breach, 13. (a) Hertl, W., J. Phys. Chem. 72, 1248 (1968). (b) Hertl, W., J. Phys. Chem.
New York, 1985. (b) Leyden, D. E., and Collins, W. T., “Chemically Mod- 72, 3993 (1968). (c) Hair, M. L., and Hertl, W., J. Phys. Chem. 73, 2372
ified Oxide Surfaces.” Gordon and Breach, New York, 1989. (1969).
2. Vansant, E. F., Van Der Voort, P., and Vrancken, K. C., “Characterization 14. (a) Evans, B., and White, T. E., J. Catal. 11, 336 (1968). (b) Nadiye-
and Chemical Modification of the Silica Surface.” Elsevier, Amsterdam, Tabbiruka, M. S., and Haynes, J. M., Colloid Polym. Sci. 272, 1602 (1994).
1995. 15. (a) Yoshino, A., Okabayasha, I. I., Shimizu, I., and O’Connor, C. J., Colloid
3. (a) Dabrowski, A., and Tertykh, V. A., “Adsorption on New and Modi- Polym. Sci. 275, 672 (1997). (b) Shimizu, I., Yoshino, A., Okabayasha, H.,
fied Inorganic Sorbents,” Studies in Surface Science and Catalysis, Vol. 99. Nishio, E., and O’Connor, C. J., Trans. Faraday Soc. 93, 1971 (1997).
Elsevier, Amsterdam, 1996. (b) Gun’ko, V. M., Voronin, E. F., Pakhlov, 16. (a) McFarlan, A. J., and Morrow, B. A., J. Phys. Chem. 95, 5388 (1991).
E. M., and Chuiko, A. A., Langmuir 9, 716 (1993). (c) Gun’ko, V. M., (b) Morrow, B. A., and McFarlan, A. J., J. Phys. Chem. 96, 1395 (1992).
Voronin, E. F., and Chuiko, A. A., Zh. Obsch. Khim. 61, 552 (1991). 17. Hoffman, P., Surf. Sci. 188, 181 (1987).
(d) Gun’ko, V. M., Voronin, E. F., and Chuiko, A. A., Zh. Neorg. Khim. 36, 18. Liu, C. C., and Maciel, G. E., J. Am. Chem. Soc. 118, 5103 (1996).
320 (1991). (e) Gun’ko, V. M., Zh. Fiz. Khim. 65, 398 (1991). (f) Bogillo, 19. (a) Zhuravlev, L. T., Colloids Surf. A 74, 71 (1993). (b) Morrow, B. A., and
V. I., and Gun’ko, V. M., Langmuir 12, 115 (1996). (g) Gun’ko, V. M., McFarlan, A. J., Langmuir 7, 1695 (1991).
Kinet. Katal. 34, 716 (1993). 20. (a) Blitz, J. P., Colloids Surf. A 63, 11 (1992). (b) Haukka, S., Lakomaa,
4. (a) Blitz, J. P., Murthy, R. S. S., and Leyden, D. E., J. Am. Chem. Soc. 109, E. L., and Root, A., J. Phys. Chem. 97, 5085 (1993).
7141 (1987). (b) Blitz, J. P., Murthy, R. S. S., and Leyden, D. E., J. Colloid 21. Parfitt, G. D., and Sing, K. S. W. in “Characterization of Powder Surfaces,”
Interface Sci. 121, 63 (1988). p. 403. Academic Press, New York, 1976.
5. (a) Berendsen, G. E., Dikaart, K. A., and de Galan, J., J. Liquid Chromatogr. 22. Proctor, K. G., and Blitz, J. P. in “Chemically Modified Surfaces” (H. A.
3, 1437 (1980). (b) Kinkel, J. N., and Unger, K. K., J. Chromatogr. 316, Mottola and J. R. Steinmetz, Eds.), p. 209, Plenum, New York, 1992.
193 (1984). 23. (a) Buszweski, B., Schmid, J., Albert, K., and Bayer, E., J. Chromatogr.
6. Hertl, W., and Hair, M. L., J. Phys. Chem. 75, 2181 (1971). 552, 418 (1991). (b) Fyfe, C. A., Govvi, G. C., and Kennedy, G. C., J. Phys.
7. Frisch, M. J., Trucks, G. W., Schlegel, H. B., Gill, P. M. W., Johnson, B. G., Chem. 89, 277 (1985). (c) Scholten, A. B., de Hann, J. W., Claessens, H. A.,
Robb, M. A., Cheeseman, J. R., Keith, T., Petersson, G. A., Montgomery, van de Ven, L. J. M., and Cramers, C. A., Langmuir 12, 4741 (1996).
J. A., Raghavachari, K., Al-Laham, M. A., Zakrzewski, V. G., Ortiz, 24. Sindorf, D. W., and Maciel, G. E., J. Phys. Chem. 86, 5208 (1982).
J. V., Foresman, J. B., Cioslowski, J., Stefanov, B. B., Nanayakkara, A., 25. Barthell, H., Colloids Surf. A 101, 217 (1995).

You might also like