Temperature-programmed desorption mass spectrometry of butyloxysilyl groups on silica surfaces

ELSEVIER

International Journal of Mass Spectrometry and Ion Processes 148 (1994) 45-54

and Ion Processes

Temperature-programmed desorption mass spectrometry of butyloxysilyl groups on silica surfaces V.M. Gun'ko*, V.A. P o k r o v s k y Institute of Surface Chemistry, Kiev 252022, Ukraine Received 9 January 1995; accepted 3 April 1995

Abstract

The decomposition reactions of the alkoxide groups =-SiO(CH2)3CH3 on dispersed silica (Butosil) surfaces have been studied by temperature-programmed desorption mass spectrometry (TPD) and quantum chemical (AM 1) methods. The rate constants were calculated on the basis of experimental data and by the RRKM theory. The potential energy surface profiles were found by adiabatic and dynamic reaction coordinate methods. The main desorption channel corresponds to the elimination of 1-butene above 600 K. The TPD data suggest that this reaction is of first order and that it can be viewed as "unimolecular". Keywords." Butyloxysilyl group; Dynamic simulation; Silica surface; TPD

1. Introduction Study of the decomposition of different groups bound to silica surfaces by heating or their interaction with particular environments is of interest, as such materials are widely used in the chemical industry and their stability is of great importance for many applications. Chemical transformations at silica surfaces, modified by butanol, were studied by heating to 1000 K and were found to be unlike the adsorption and decomposition of CI-C3 aliphatic alcohols [18] or of butylamine-modified surfaces [9]. In the present work, in which decomposition reaction studies have been continued [3-5], the thermal decomposition mechanism of butyloxysilyl * Corresponding author.

groups on silica surfaces has been investigated by both experimental and theoretical methods.

2. Experimental Dispersed silica samples (Aerosil A-300 with a BET surface area of 300 m 2 g-l) in which O(CH2)3CH3 groups were attached through the reaction =SiOH + CH3(CH2)3OH --~ -SiO(CH2)3CH3 + H20

(1)

at a concentration of 0.5 mmol g-t have been used. A study of the gaseous products released upon decomposition of Butosil surfaces has been carried out by the temperature-

0168-1176/95/$09.50 © 1995 Elsevier Science B.V. All rights reserved SSDI 0168-1 176(95)04174-5

V.M. Gun'ko, V.A. Pokrovsky/International Journal of Mass Spectrometry and Ion Processes 148 (1995) 45 54

46

I ,Qbs,un

[,Abs,un, tO0

tooo-j 750 -" 500 i

18

i

I/ 1'11

i

I

250]

25

0

41

/!

44

...........~,i ,,,F............................................................................. ~!Lh, i.n. .•. . . . .r . . . . ,. J i i i i i i

tO

ttO n/z

r

"

programmed desorption mass spectrometry (TPD) method in analogy with earlier work [3-5,10]. The Butosil samples (0.5-20.0 mg) were placed in a quartz-molybdenum tube and evacuated at 10-1 Pa, and then attached to the inlet system of a MI-1201 (Ukraine) mass spectrometer. The reactor-mass spectrometer interface included a high-vacuum valve with an orifice of diameter 3 m m and a leak-in tube of diameter 5 mm and length 20 cm which were kept at 423 K. The reaction space is open in the ion-source direction, and at the heating rate used (close to 0.1 K s -1) the observed intensity of the ion current (I) may be thought of as proportional to the desorption rate so that diffusion inhibition can be neglected. We assumed stationary conditions when the shape and position of a desorption I,Abs,un, I0001

too 200 300 400 500 600 700 800 gO0 tO00 I,C

0

210

Fig. 1. Survey T P D mass spectrum of Butosil at T = 333 K.

Fig. 3. 1-Butene desorption from Butosil; dependence on temperature. A b u n d a n c e of m/z 41 is shown.

peak did not depend on the temperature of the spectrometer interface, the sample dispersivity or its size. The TPD data were not considered further if these conditions were not fulfilled. In the present experiments the peak position did not change appreciably in the 10-6-10-3 g range of sample weight, i.e. diffusion effects may be neglected. A controlled heating rate of 0.067 K s 1 was used for temperature programming. The desorption mass spectra were recorded in the 10-200 u range and the temperature step was equal to 7.6 K (these procedures have been described in detail previously [10,11]. Analysis of the mass spectra (Figs. 1 4) by comparison with the literature [12] shows that 1-butene (m/z at 27, 41, 55, 56) and water (rn/z at 17, 18) are desorbed. 1-Butene desorption is unequivocally demonstrated from these I,Abs,un 100

18

;~-.. b' L, iI i I i iI

75

t

Ii " ,/';-""-.

,/

50

1,5

,/ ,,,"

250

"""..............."""..........."..........

25

o io

[

11o

21o

Fig. 2. Survey T P D mass spectrum of Butosil at T = 808 K (ions characterize l-butene as the main desorption product).

0

'

~

,

,

r

i

,

'

1?

[

'

~

'

]

I

r

,

r

]

I00 200 300 400 500 600 700 800 900 I000 T,C Fig. 4. Water desorption during Butosil thermolysis. A b u n d a n c e

o f m / z 17 and 18 is shown.

V.M. Gun'ko, V.A. Pokrovsky/International Journal of Mass Spectrometry and Ion Processes 148 (1995) 45-54

mass spectra [12]. The abundances of signals due to other molecules and their fragments is insignificant (Figs. 1 and 2). The highest intensity is seen for butene lines, i.e. decomposition occurs mainly according to the scheme = Si_O(CH2)3CH 3 T --SiOH + CH2CHCH2CH 3

(2) In the 750-880 K region, water desorption is observed (Fig. 4). These water molecules desorb from interstitial sites or by diffusion from bulk centers or they arise as a consequence of migration and condensation of hydroxyls [13], possibly formed via Reaction (2). However, water loss is much less important than that o f butene, by at least one order of magnitude (Figs. 2-4). Therefore water has only a small influence on chemical transformations o f bound groups and we do not consider water desorption further in detail. The kinetic parameters for butene desorption were found for the most intense ions, m / z 41 (Fig. 3), using methods described earlier [11]. The characteristic asymmetrical form of this curve corresponds to a first-order reaction in which butene elimination occurs by a process that is in keeping with that shown in Eq. (2). To interpret the mass spectra we assume that a desorption equation can be written as

dO~dr = -kO n

47

where q~(t) =

It 0

kdt

(8)

We now assume that the magnitude of the ion current is directly proportional to the desorption rate. This assumption is valid if the desorption rate is high in comparison to the heating rate. F r o m Eqs. (5)-(7) one obtains

dO(t)/dt = - k e x p [ - e ~ ( t ) ]

n= 1

(9)

dO(t)/dt = -k/[1 + ~(t)] 2

n= 2

(10)

dO(t)/dt = -k/{1 + 2~I,(t)] 's

n= 3

(11)

If k 0 = k B T / h ~ 10 ~3 s -t and E d = 2 0 0 kJ mo1-1 then, according to Eq. (10), the peak shape has a nearly symmetric form, but for Eq. (9) the dO/dt function decreases faster on the right-hand side and for Eq. (11) the converse is true. Hence, the first-order reaction is characterized through greater steepness of the peak shape in the high temperature region. The O(t) dependence on I may be written as

O(t) = kv(t)/S

(12)

or

dO/dt = - I / S

13)

where

(3)

where 0 is surface coverage, n is the reaction order (it is considered an unknown parameter), and the reaction rate constant k is given by k = k0 exp (-Ed/k BT)

(4)

where E d is the activation energy of desorption. At the initial condition O(t = 0) = 1 we have the following solutions of Eq. (3) for different n:

O(t) = exp[-dg(t)]

n= 1

(5)

O(t) = 1/[1 + e~(t)]

n= 2

(6)

O(t) = 1/[1 + 2q)(t)] °5

n= 3

(7)

S =

Idt

tg(t) =

V

(14)

Idt

(15)

t

Using Eqs. (3)-(7) and (13) one obtains Ink = ln[-~(t)/I]

n= 1

(16)

l n k = ln[-~2(t)/SI]

n = 2

(17)

l n k = ln[t~(t)3/IS 2]

n = 3

(18)

If all assumptions are valid for n = 1, the function l n k = f ( 1 / T ) is linear over the entire temperature range and any deviation may be

V.M. Gun'ko, V.A. Pokrovsky/International Journal of Mass Spectrometry and Ion Processes 148 (1995) 45-54

48

I% = 9 0 3 ~01ss I

k . = 1 74 l O " s

Ea = 2 5 3 T9 kJ/mo~

rnk

~,abs u

I a~s u

4

900.

300

' "

-05~

ooi

//

, 300.

-05

_ ~._-~

4t9 .......

;00

)

/

•

A

300 -

1 16 10 J

I'l'l'l

I'l'l

1 3103

, 1 4 4 ~0 ~

lO0

.

/'\

500 -

1071 -175

• l'l

f,c ......

E• = 2 2 9 17 kdh~ol

'

'

J

700.

'

fnk

/ /

-107

~.1

lrr, K'

i

- 17

~

/

5

\. i

f,c

i .i

i.i

Sb~

'",1

,,6,o,

i.i

i

i.r

~n

Fig. 5. Rate constant dependence on temperature and determination of the Ed and k 0 values over the low temperature region for l-butene desorption (m/z 41). Broken lines indicate the chosen region with linear dependence of In k(T) on T for determination of the rate constant•

caused by model inadequacy or by errors in the experimental data. The TPD data processing for the butyloxysilyl group decomposition according to Eqs. (16)-(18) gives the best results for n = 1. Yet, in the high temperature region, k(T) has a smaller slope than the model curve for n = 1. This effect may result from the dependence of Ed on 0 at low coverages. The present method of data handling, of course, does not give complete and unambiguous determination of decomposition processes, but it allows one to understand their mechanism. The low temperature part of Fig. 5 gives a linear dependence of the logarithm of the rate constant for reaction (2) over four orders of magnitude of the k(T) values, and this is in favor of the first-order reaction model; however, in the high temperature region of a deviation from linear dependence is already seen within one order of magnitude of k(T). These deviations are typical for TPD data and they may be explained by the desorption energy changes caused by heterogeneity of the surfaces and their coverages [14,15]. The values of the activation energy (Ed) and the pre-exponential factor (k0) of the rate constant for reaction (2) were found from the low temperature region (Fig. 5) at coverages

J

,

" ~

,31o,

~,4,o'

K'

Fig. 6. Rate constant dependence on temperature and determination of the Ed and k 0 values over the entire temperature interval indicated by broken lines for l-butene desorption (m/z 41)•

0 >_. 0.3; Ed = 253 kJ mol-1;

k0 = 9.0 × 1015 S-1 (19)

The Ed and k0 values calculated for the total temperature range of butene desorption (Fig. 6) are as follows: E d = 229 kJ mol-1;

k0 = 1.7 x 1014 s -1 (20)

These data are averaged values as some deviation from the linear dependence In k = f(1 / T) is seen in Fig. 6. However, the differences between the Ed and k0 values for these two temperature intervals do not exceed 10%. It seems reasonable to state that for the best kinetic description, the values obtained in the low temperature decomposition region should be used in preference to the values for the total temperature interval.

3. T h e o r y and d i s c u s s i o n

Quantum chemical calculations of potential energy surface profiles (PESP) of reaction (2) have been carried out by considering a cluster structure using the A M I method [16], in analogy with earlier work [3]. In the silica cluster (O~SiO)3SiO(CH2)3CH3(I) (0" is a

V.M. Gun 'ko, V.A. Pokrovsky/International Journal of Mass Spectrometry and Ion Processes 148 (1995) 45-54

r

250

..,

49

-

200 H , ,"1

t''

,',

"~.i~.,-

~5' 7' , >' n .

*~

:* '"~

L~ o~

~"' o

~

. ,~

"

: .

,

~

150-

~a I 0 0 ~ taJ

d, ~ 50-

0.095

rsio,nm roc,nm roH,nm rcc,rlm rcH,nrn ZSiOC, ° LOCC, ° /CCC, °

0.17470 0.14589 0.12470 0.14982 0.13617 120.3 ° 90.1 ° 1171 °

monovalent oxygen pseudoatom) the bond length (rou) between O and H atoms of the /3 = C H 2 group, as shown in Fig. 7,

\ /

=

Si --

HH

/ Cot\

~0

/

\ / /ct\

Cp\

H+ H

H

/H C s -- H

\

H

"--~ ~-~ SiOH + C4H8

(21)

H

was chosen as a reaction coordinate A. The PESP along the reaction pathway was obtained using steps AA = 0.005-0.010 nm by the adiabatic reaction coordinate (ARC) method (Fig. 8). For each small step along A other geometric parameters of the system were fully optimized. If A is set equal to ronco, i.e. when a simultaneous transfer of the second H from the/3-CH 2 ground to Ca occurs, H

HH

H

H+/\ / \ / C~ C,t\ Si--

0/

H+~e l / /

H

H C 8/- H ' - ~

\

\

H

H

~SiOH+C4H

0.195

0.245

Fig. 8. Total energy change dependence on reaction coordinate A(rOH).

Fig. 7. Structures of the ground and transition states for reaction (21).

H

O. 1 4 5

r(OH), nrn

T r a n s i t i o n State

G r o u n d State 0.17388 0.13974 0.26479 0.15226 0.11203 121.8 113.2 110.8

8

(22)

then E $ is > 350 kJ mo1-1, i.e. E~2 - E~l >~ 90 kJ mol -l . Thus, any H transfer not originating

from the /3-CH 2 group requires a second H transfer that leads to E* increasing in comparison with Eq. (21). The calculations for Eq. (21) with restricted configuration interaction (3 x 3) [17] give a further decrease in E $ by only 5 kJ tool -1, and the changes in transition state (TS) geometry are small (Table 1), i.e. the decomposition of butyloxysilyl groups may occur in the ground electronic states but in vibrationally excited levels. Reaction (21) occurs rapidly at T >/750K (Fig. 3) when the surface groups are vibrationally excited and this has an influence on the PESP character. These effects lead to pathway variations especially near the TS [18]. The A R C method gives OEp/OXij = 0 (Ep is potential energy, x~ is atom coordinate, x~j ¢ A) for system motion along the reaction pathway, but all atoms take part in thermal vibrations with periods in the 8.9 x I0 -15 S (O-H) to 3.5 x 10 -14 S ( S i - O ) range, which are smaller than the time needed for a chemical transformation of the bound groups. Therefore, as the system proceeds along A there are random, small deviations from the equilibrium positions of the atoms, which no longer fulfil the condition OEp/Oxij = 0. It is probable that atom tunneling may contribute to this process [18]. We modeled a random preparation of the activation barrier as follows. The barrier changes which occur through random atom vibrations

50

V.M. Gun'ko, V.A. Pokrovsky/International Journal of Mass Spectrometry and Ion Processes 148 (1995) 45 54

Table 1 The parameters of the reactants (GS), TS, and products of Reaction (21) Parameter

=SiO(CH2)3CH 3 GS

TS

Products

TS (CI (3 x 3))

AEt --EHoMO ELUMO Qsi -qo qc,, qc;~ qHc~ roc,~ rc,,c~ rc,~H #

-10.50 0.47 1.911 0.606 0.038 -0.182 0.083 0.13974 0.15226 0.11203 0.99

260 8.62 0.40 1.844 0.566 0.029 -0.366 0.267 0.14589 0.14982 0.13617 1.50

103 9.53 0.53 1.920 0.662 -0.198 -0.174 0.238 (OH) 0.31293 0.13314 0.35551 1.52

255 8.61 0.40 1.842 0.565 0.029 -0.366 0.266 0.14591 0.14983 0.13627 1.48

q is the atom charge, r the bond length (nm), EHOMO and ELUMO (eV) are boundary orbital energies, AEt is change of total energy (kJ mol-l), # is dipole moment (D) as calculated by the AM1 method.

of the active subsystem (=SiO)3SiOC4H 9 were simulated at A ~> 0.14 nm, considering bond vibration amplitudes of up to 0.02 nm and deformation vibrations up to 10°. For more accurate determination of the reaction pathway in the TS region, only those atom vibrations which correspond to the geometry changes in the product direction were considered. These are elongations of the O - C a and C g - H bonds and decreases of the C~-C 9 and O - H c bonds owing to the valent (Uoc, /.JCH,/.JCC) and deformation (6sioc, 6occ, 6ccH) vibrations. After random changes of geometry, a total energy (Et) minimization was carried out by varying all parameters (except A), and an ascent to the TS was controlled, i.e. the A¢fluct value was calculated. The E ~ value (Table 1, AEt) was obtained by taking into account the modeling barrier random fluctuations and at a slow rate of ascent to TS without random vibrations AA* = A~uct= A* >~ 0.005 nm and Enuct~< E ~. The decomposition (Eq. (21)) can be considered a unimolecular reaction [3-5] and the rate constant (ku~i(T)) was therefore calculated by the RRKM theory [19]. An application of this theory to the decomposition of bound groups to a surface rests on the following assumptions.

(1) Vibrational states of bound groups and solid surface atoms correspond to the normal distribution. Effective exchange of energy occurs between the matrix and bound groups. (2) Active vibrations are vibrations of atoms from bound groups and neighboring fragments of a solid surface; the rest of the solid is an adiabatic subsystem, i.e. it does not make a contribution to a change in the ratio of statistical sums at the decomposition of the bound groups. (3) Rotational degrees of freedom are active if they are related to the bond changed when reaction occurs; otherwise they are adiabatic. (4) The deactivation rate of activated bound groups is proportional to the gaseous product pressure; yet these products are ionized and removed from the reaction zone (surfaces) by external fields so the deactivation efficiency is taken to be small ( <~ 0.1). (5) The averaged distance between neighboring bound groups is equal to 0.65-0.70 nm and lateral interactions between them at coverages 0 ~< 0.3 are disregarded. (6) The decomposition reaction is taken to be direct since stable prereaction states, which are different from the initial state, are absent, i.e. the decomposition time corresponds

Q~/Q

V.M. Gun'ko, V.A. Pokrovsky/International Journal of Mass Spectrometry and Ion Processes 148 (1995) 45 54 Table 2 Frequencies for the ground state and TS (in cm l) Vibration

GS

TS

Vibration

GS

TS

Usio+ 6CCH+

949 974 1139 1143 1155 1157 1164

902 952 981 1009 1077 1089 1133

1171 1180

1144 1152

us~o+ 6HCH

1215 1219 1247

1161 1163 1185

UCC+ 6HCH

1290 1334 1345

1209 1228 1270

1346 1347 1356 1364 1385

1342 1345 1350 1360 1369

1398 1449 1438

1401 1497 2037

2898 2908 2934 2955 2956

2897 2954 2959 2970 2977

2958 2976 2996

3015 3041 3053

Xcccc+ 6siosi÷ XOCCC+ XSiOCC

47 59 68 88 145

40 51 63 86 110

OCCH+ ~SiOSi+

156 164

150 163

6ccc+ 6occ

195 220 229 246 291 413 462 574

182 212 227 234 266 373 464 570

575 580 584 592

575 578 580 592

596 606 645

595 605 642

658 732

649 676

USiO+

808

805

6CCH+ 6SiOC

809 817 824 825

808 809 823 824

830 869

829 835

UCH

3041

closely to 10 -13 s. Therefore the decomposition occurs faster than energy exchange between the activated subsystem and the solid atoms connected to the adiabatic subsystem. (7) Statistical sums Qe and Qt (electronic and translational degrees of freedom) of the reagents and TS were considered as constants. The atom vibration frequencies were found for the ground and transition states. Moreover, for the Qv and Q~v calculations, only those active vibrations were considered which concern the (SiO)3SiOC4H 9 fragment (the number of vibrations is n = 3N = 63 for this fragment in the ground state and 62 for TS,

51

where N is the number of atoms in this fragment (Table 2); total no = 3 N 0 - 6 = 8 4 , where N o = 30 is the number of atoms in the cluster studied). We assume that the Q*v/Qv ratio for other vibration variables is equal to 1 as the reaction proceeds without formation of the prereaction complex, i.e. it is a direct reaction with a short time of transformation. Therefore the vibrations of remote solid atoms do not make a contribution to the change in Q*v/Qv value for the bound group. Besides, we used the atom vibration frequencies without normalized factors, assuming that their influences on Q*v and Qv are similar. In the independent rotator approach [19] the rotations around the Si-O, O - C ~ and C ~ - C ~ bonds were regarded as active rotators with the ratio Q~,ac/Qr,ac,.~ 0.6, but the C~-~)-C~ and C.~-~Ce rotators were adiabatic at Q{,ad/Qr,ad ~ 1.205. The density of vibration-rotational states was calculated using the method of Whitten and Rabinovitch described in Ref. [19]. The microscopic rate constant k(E +) [19] was integrated between the limits E + = E - E ~ = 0 and E+ax = 200 kJ mo1-1 as the k(E +) contribution to kuni(T) at E+>~ E+ax is low, because the k(E +) function for high E + ~> 200 kJ tool -1 shows an exponential drop (Fig. 9). The deactivation probability of activated bound groups ~6.0 -~

o

15.5 I

/ "/'/~

4

~

/

140

b

13.0 ] 50o

i

60o

i

700 T,

,

i

i

t

8oo

900

Iooo

K

Fig. 9. Microscopic rate constant dependence on E + = E - E ~ at T = 800 K calculated by the R R K M method as described in the text.

52

V.M. Gun'ko, V.A. Pokrovsky/International Journal o f Mass Spectrometry and Ion Processes 148 (1995) 45-54 -2-

-4-

-5~

-62 --

- 7 -

- 1 0

,

,

,

,

,

50

,

,

,

,

,

,

I00

,

,

150

~

,

,

i

,

200

E + , kJ/mol

Fig. 10. Dependence of the pre-exponential factor of the rate constant on temperature: curve a, with allowance for low frequency vibrations at n = 63; curve b, considering only u i >~ 200 cm -1 a t n = 5 5 .

by an interaction with gas-phase molecules was regarded as being minor and the deactivation efficiency was set at ~5= 0.1 as the pressure during the TPD reaction was low. In this case, kuni(T ) is practically independent of pressure and the decomposition is caused by vibrational energy (a population of excited vibrationalrotational states), which passes from the matrix to bound groups. The logarithm of the effective prefactor k0,ef f = k u n i

exp(E*/kB T)

(23)

depends on temperature nearly linearly (Fig. 10(a), n = 63). The large values of k0,eff obtained both by the R R K M theory and from the experimental data are determined mainly by the contributions of low frequency vibrations, 50 ~ u~ ~< 200 cm -l, which give larger Q*v than Qv for the ground state on account of the bond elongations and decreases of the frequencies u~ < /]i (except U~c~c~> Uc~c~ and a few others for high frequency uCH, which give lower contributions to Qv than low frequency vibrations (Table 2)). If the low frequency vibrations ui <<.200 cm -1 are excluded from the vibrational partition function Qv then k0,eff (n = 55) decreases (Fig. 10(b)) and kuni is lower by one order of magnitude at 800 K. Hence the low frequency

vibrations (with large amplitudes at T > 700 K) play a significant role in the reaction in Eq. (21). These were simulated in the PESP calculations in the TS region as random vibrations. Experimental k0 values depend on T weakly as the deviation from a linear dependence lnk(T) ~ - 1 / T is small (Figs. 5 and 6). A stronger dependence of k0,e~- (n = 63) on T (Fig. 10(a)) is conditioned by ln(Q*v/Qv) ~ T. At the same time, k0,eff (n = 55) depends on T weakly and its change with T corresponds to the experimental data accurately, but its value is lower by one order. Reaction (21) is strongly endothermic, AAH(21) ~ 100 kJ mol -l, as are the decompositions of alkoxides -SiOCH3(II) and - S i O C R ( R = H (III), CH3(IV)) on the silica II O surfaces [3-5]. Such processes occur through high vibrational energy contributions [15] and ko,eff depends strongly on Q*v/Qv. Only the formation of 1-butene in the decomposition of the -SiO(CH2)3CH 3 group is governed by the H + transfer from the/3-CH 2 group. The E~21) value is higher than for III and IV group decompositions by 60 and 50 kJmo1-1 owing to a lower CH acidity of the I group, e.g. qH,In is 0.1 higher than for I. The peak temperature of reaction (2) is higher by 50 and 60 K, respectively, than that for the III and IV decompositions, but is lower than for II by 150 K although E~ ~ E~I. Moreover, k0,i is higher by one order than the corresponding prefactors for II, III and IV. This is due to the absence of any strong barrier to active rotations in TS(21) unlike the decay of the II group where a new bond forms between H and matrix Si and Q[/Qr < 0.1 [3] (in reaction (21) Q[/Qr ~ 0.72) and Q*v/Qv is higher than in groups II, III and IV because nI > nll,lXl,l v at u~ < ui for active bonds. Quantum chemical computation of PESP and statistical calculations of k(T) give an incomplete view of the reactions if the PESP

V.M. Gun'ko, V.A. Pokrovsky/International Journal of Mass Spectrometry and Ion Processes 148 (1995) 45-54

is obtained by the A R C method, in which motion along the reaction pathway is independent of time. The PESP calculations by the dynamic reaction coordinate (DRC) method give additional information about the reaction mechanism. In this case the starting point (t = 0) on the PESP is the TS found by the A R C method. The atom velocities and coordinate changes on the j + 1 time step can be written as --/kvk+l

:

Axkij+, =

~.Atj + [d(Yjikj)/dt](Atj)2/2

products

J

250

TS

-

h, 0 200-

<.

.

/i

_

-~ 150-

5.,a i o o - " /

500

Ill./'-" i,,i,,,i,,,i,,,i,,,i,,,i,,,i,,,

-50-40-30-20-10

Atj(VEpj)ki/mi + [m[l(Aty)2/2]d(V£p/)~/dt

reagents

300 -

(24) (25)

where k = l, 2, 3; At is time interval (~< 10 -16 s); i is the atom number; Epj is the potential energy for the j t h step; V = 0 / 0 x k. After the At/ step, Epj+l is calculated by quantum chemical methods, then new xij+l and vi/+l are calculated. The decomposition reactions occur on the solid surfaces and energy exchanges between bound groups and the matrix should be taken into consideration, e.g. dissipation of vibrational energy (which can be accumulated on bound groups) or kinetic energy of the desorbed molecule. In the present work the dissipation time of the E k (energy of vibrationally excited states of the bound group) half value was changed in the 10-13 ~< % ~< 3 × 10 -14 s range (this time interval corresponds to the reaction time ( ~ 10 -13 s) or vibrational period of the S i - O bonds), i.e. simulated energy dissipation on the solid was fast. Three cases were considered: Ek = 0, 42 or 126 kJ mo1-1 at TS (the starting point). The Ep dependence on time had an asymmetrical shape near TS (Fig. 11, I) which is caused by a difference between the system motions for the processes TS ~ reagents and TS ~ products. In the latter case the C 9, C7 and C6 atom velocity vectors are directed to the surface (in TS the ZOCC angle is decreased strongly)

53

f,,,i,,ll

0 10 20 t, f s

30 40

50

Fig. 11. P o t e n t i a l e n e r g y (Ep) d e p e n d e n c e s o n t i m e w h e n E k in TS = 0 a n d To = 3 x l 0 13 S ( I ) ; E k = 126 kJ mo1-1 a n d To = 3 x 10 -13 s (II); E k = 42 k J m o l - t a n d To - 3 x 10 -14 s (III).

which leads to the C and H atoms drawing near to the matrix atoms (Fig. 7) and Ep drops more slowly than in the process TS ~ reagents (bound group in a ground state), as the extent of geometry reorganization in the decomposition TS ~ products is larger and the integrated pathway (Fig. 12) changes faster. The first minimum of PESP in the process TS ~ reagents (t ~ ( 3 - 6 ) x 10 -15 s) corresponds to r c ~ u ~ 0 . 1 1 nm (Arc~H ~ 0.025 nm), but the ZOCC change is small (LOCC ,,~ 90°), i.e. the rate of increase of this angle is slow and / O C C ~ 110 ° after At ~ 4 x 10 -14 S. In general the Ep of fast reagents

0.20

0.15

products

"

i'-

c

0.10

0.05

0.00

~

,

l

I

I

-50 -40

,

I

l

[

l

'

I

l

-30 -20

l

l

l

-10

I,

l

I

0

l

I

l

l

10

'

'

'

'

20

'

I

30

fs

Fig. 12. I n t e g r a t e d p a t h w a y ( S 0 d e p e n d e n c e s o n t i m e a t the s a m e values o f E k a n d 7-o as in Fig. 11.

54

V.M. Gun'ko, V.A. Pokrovsky/International Journal of Mass Spectrometry and Ion Processes 148 (1995) 45 54

reogenfs

0.14

tion may be explained by the lower CH acidity of the butyloxysilyl groups. The increase in peak temperature for reaction (21) relative to decomposition ofthose groups leads togrowth of the pre-exponential factor k 0 and an increase in E*. The low frequency (u ~< 200 cm-l) vibrati°ns (valent, def°rmati°n, and libration) in surface-bound groups are the major factors causing k0 to increase: k0 >>

TS

,~

ir

0.13

I

E

i

\

C0.12

A,

-r

T

L-

0.10

0.09

,

E',. /,i'

,

,

-SO

-40

r

,

,

,

~ ', ,i ~"i i /H,, '/' ~['./ ~ ' " if 'J i

- 5 0

,

,

,

i

,

20

,

,

p -10

,

,

i 0

,

,

, q

kBT/h.

10

~, fs Fig. 13. Bond length (rcH) dependences on time at the same values of Ek and TOas in Fig. 11.

oscillations are conditioned by the H atom vibrations (t,~ 10 -14 S) (Figs. 11 and 13). The Sr(T) function calculated as a sum of mass-weighted coordinate changes of all atoms has a nearly symmetrical shape (Fig. 12). The Sr change rate depends strongly on Ek and To. Hence the decomposition (21) occurs through vibrational energy accumulation on the O-Ca(v), O C9(6 ), C~-C~(v) and C~-C9-H(6 ) bonds.

4. Conclusion Experimental studies of the butyloxysilyl group decomposition have shown that the main channel corresponds to the elimination of 1-butene with the peak desorption temperature around 800 K and this process corresponds to a first-order reaction. The quantum chemical simulation has shown that butyloxysilyl group decomposition occurs via H + transfer from the/3-CH 2 group to the O atom. Theoretically obtained E* and k 0 data correspond to experimental data quite satisfactorily. Higher E* values in reaction (21) than for - S i O ~ R (R = H, CH3) decomposiO ii

References [1] C. Morterra and M.J.D. Low, J. Phys. Chem., 73 (1969) 321. [2] K.S. Kim, M.A. Barteau and W.E. Farneth, Langmuir, 4 (1988) 533. [3] V.V. Brie, V. M. Gun'ko, V.V. Dudnik and A.A. Chuiko, Langmuir, 8 (1992) 1968. [4] V.V. Brei, V.M. Gun'ko, V.D. Khavryuchenko and A.A. Chuiko, Kinet. Katal., 31 (1990) 1164. [5] V.M. Gun'ko, V.V. Brei and A.A. Chuiko, Kinet. Katal., 32 (1991) 103. [6] V.V. Brei and A.A. Chuiko, Teor. Eksp. Khim., 21 (1985) 635. [7] V.V. Brei, V.A. Pokrovsky and A.A. Chuiko, Teor. Eksp. Khim., 23 (1987) 501. [8] H. Noller and G. Ritter, J. Chem. Soc., Faraday Trans. 1, 80 (1984) 275. [9] R. Sokoll, H. Hobert and I. Schmuck, J. Chem. Soc., Faraday Trans. 1, 82 (1986) 3391. [10] V.A. Pokrovsky, K.B. Yatsimirsky and V.A. Nazarenko, Teor. Eksp. Khim., 23 (1987) 377. [11] V.A. Pokrovsky, Yu.I. Bratushko, K.B. Yatsimirsky and E.N. Korol, Teor. Eksp. Khim., 14 (1978) 754. [12] A. Cornu and R. Massot, Compilation of Mass Spectral Data, Heyden, London, 1966. [13] A.A. Chuiko, V.A. Sobolev and V.A. Tertykh, Ukr. Khim. Zh., 37 (1971) 1242. [14] S.U. Payne and H.J. Kreuzer, Surf. Sci., 222 (1989) 404. [15] V.P. Zhdanov, Elementary Physicochemical Processes at Surface, Nauka, Novosibirsk, 1988. [16] M.J.S. Dewar, E.G. Zoebisch, E.E. Healy and J.J. Stewart, J. Am. Chem. Soc., 107 (1985) 3902. [17] W. Thiel, J. Am. Chem. Soc., 103 (1981) 1413. [18] V.A. Bendersky, V.I. Goldansky and D.E. Makarov, Dokl. Akad. Nauk SSSR., 311 (1990) 626. [19] P.I. Robinson and K.A. Holbrook, Unimolecular Reactions, Wiley Interscience, London, 1972.