NMR kinetic study of ethyl silicate hydrolysis by ...

NMR kinetic study of ethyl silicate hydrolysis by ammonia in alcoholic solvents IDRISSEL BAKALI,ESSAIDEL RHESS,CHRISTOPHE ROUSSELOT, AND RENEMERCIER' Laboratoire d(e'1ectrochimie des solides, Unite Associde au Centre National de la recherche scientijique, no 436, UniversitP de Fratzche-Comte, 25030 Besanqon CEDEX, France AND

MARIE-FRANCE MERCIER Centre de spectrome'trie, Universite' de Franche-Comte',25030 Besanqon CEDEX, France

Can. J. Chem. Downloaded from www.nrcresearchpress.com by 183.91.85.130 on 06/03/13 For personal use only.

Received June 11, 19912 IDRISSEL BAKALI, ESSA~D EL RHESS,CHRISTOPHE ROUSSELOT,RENEMERCIER, and MARIE-FRANCE MERCIER. Can. J. Chem. 70, 1612 (1992). Hydrolysis of Si(OC2H5),by ammonia has been investigated using 'H and I3C NMR measurements in alcoholic media. Use of deuterated CD3CD20D and CD30D solvents proves the absence of transesterification by alkoxy exchange between silicate and alcohols. The rate constant increases when CD30D replaces CD3CDZODas the solvent (k = 2.8 X L mol-' min-' in CD3CD20D;7.5 X L mol-' min-' in CD30D for similar experiments).

IDRISSEL BAKALI, ESSAIDEL RHESS, CHRISTOPHE ROUSSELOT, RE& MERCERet MARIE-FRANCE MERCIER. Can. J. Chem. 70, 1612 (1992). O+rant en milieu alcoolique et utilisant des techniques de RMN du 'H et du 13C,on a h d i &I'hydrolyse du Si(OC2H5),. L'utilisation du CD3CD20Det du CD30D comme solvants permet de demontrer qu'il n'y a pas de transesterification par &changede groupes alcoxyles entre le silicate et les alcools. La constante de vitesse augmente lorsqu'on remplace le CD3CD20Dpar du CD30D comme solvant (k = 2,8 X L mol-I min-' dans le CD3CD20D;7,5 x L mol-' min-' dans le CD30D, pour des conditions experimentales semblables). [Traduit par la redaction]

Introduction Obtaining sol-gel materials has been extensively studied in the last decade. In particular, Stober et al. (1) studied the hydrolysis of tetraethylorthosilicate Si(OC,H,), as reported in a paper on colloidal silica synthesis. He used an ammonia reactant to obtain homodispersed particles in the range of 0.1-1 Fm. He varied the ammonia and water concentrations, as well as the solvents, to obtain particles having final diameters that could be measured. Shimohira and Tomuro (2), Shimoda et al. (3), and Tan et al. (4) studied the effect of temperature on final-diameter size. They found that when temperature increases, the final-diameter size decreases. The influence of all of the above varameters acting on the hvdrolysis and condensation of S~~OC,H,),was revyewed in o;r laboratory by El Rhess ( 5 ) . He studied the kinetics of the evolution of both particle diameter and particle concentration for silica particles having an initial diameter of 10 nm. We wanted to observe the first steps of the hydrolysis and condensation of different alkoxyl silicates. The most efficient technique for this type of study is 2 9 ~Nuclear i Magnetic Resonance (NMR). This technique has already been used by various researchers to follow the hydrolysis and condensation reaction of both methyl and ethyl silicate (in particular, Lin and Basil (6) for Si(OC,H,), and Artaki et al. (7), Balfe and Martinez (8), and Klemperer et al. (9) for Si(OCH,),). 2 9 ~NMR i reveals the identity of hydrolyzed molecules such as Si(OH)(OR),, Si(OH),(OR),, Si(OH),(OR), and Si(OH), by a characteristic resonance line and the appearance of dimeric (Si-0-Si) and oligomeric species at distinct ranges of frequency. The experimental conditions used to study these hydrolyzed molecules do not, in general, coincide with the experimental conditions used in the synthesis of silica. This paper compares the kinetics of the ' ~ u t h o to r whom correspondence may be addressed. 2~evisionreceived December 18, 199 1.

hydrolysis of Si(OC,H,), in two alcoholic media, CD,OD and CD3CD20D,using 'H NMR spectroscopy. Our experimental conditions are the same as those required to obtain silica colloids by adding ammonia to alcoholic media. Assink and Kay (10-12) studied Si(OC,H,), and Si(OCH,), using 'H NMR, first in an acid medium and then in a basic (NH,OH) medium. We also wanted to determine if silicate and alcohol exchange alkoxy groups and thus cause transesterification Si(OC2HS), + 4 CD30D + Si(OCD3), + 4 CrH50D

This reaction could occur in competition with the hydrolysis reaction.

It is not possible, using solely 'H NMR, to determine if there is competition because deuterical species such as Si(0-CD,) do not produce signals. Therefore, in certain experiments, we used I3cNMR because it allowed us to distinguish Si-OCD, from SiOCH, using proton coupling techniques, which produce singlets for ',cH, and multiplets for ',cD,.

Experimental conditions We used wholly deuterated anhydrous alcohols (CD30D and C2D50Dfrom Aldrich) as solvents. 'The silicate and the ammonia solution (NH3, H20) are in their protonated forms. Freshly distilled ethyl silicate was used to avoid atmospheric hydrolysis. Si(OR),, R'OD, and concentrated ammonia were successively added to NMR tubes at 20°C, for example: 500 pL C2D50D + 15 pL Si(OEt)4 + 90 pL NH40H (13.5 M). The different experiments are summarized in Table 1. Immediately after adding NH,OH, the tubes were stirred for 15 s and then placed in the NMR instrument (Bruker Spectroscopic AC 200 spectrometer linked to an ASPECT 3000 cornputer). The 'H spectra were recorded at 200 MHz. The spectrum produced by the results of 16 pulses was immediately recorded,

TABLE 1. Designation of characteristics of the 'H NMR experiments


Code

Solvent

Silicate

[Silicate] (M L - I )

giving a spectral record of 2 min. The I3C spectra were recorded at 50 MHz by using 'H broad-band decoupling: 256 pulses were recorded with a 5 s delay between each pulse. It took 30 min to record the I3Cspectrum. The NMR apparatus was calibrated for D resonances for the alcohol chosen, by using anhydrous mixtures (silicate + alcohol). In Table 1, the four experiments are for the hydrolysis of Si(OC2H,),. The concentration of the solvent is about 14 M L-' (C2D50D)and 20 M L-' (CD30D). Chemical shifts, 6, are expressed in ppm (for I3C and 'H NMR), with respect to the zero value of TMS (1%). Using 'H NMR, we can easily follow the disappearance of silicate species and the appearance of alcohol species formed during hydrolysis. Si(OR), + 2 H20+ SiO,

+ 4 ROH

At the beginning of the kinetic study, 'H NMR signals correspond to Si-0-CH2-CH, or Si-0-CH, protons and to only one broad peak for both NH3 and H20 protons. During hydrolysis, this last peak encompasses all hydroxyl groups (R-OH, Si(0H)). Precise intensity measurements enable one to determine the exact [0-H]/[Si-0-R] ratio, as well as the evolution of [Si-0-R] groups as a function of time. The ETET 2 mixture corresponds to experimental conditions required for the synthesis of 300-nm silica particles at 20°C (12 times more H,O than (OC2H5)in concentration).

Results NMR ETET 1 With C2D,0D, the solvent is characterized by a quintuplet at 56.8 pprn (CD,) and a heptaplet at 17.5 pprn (CD,) . Si(OC,H,), presents singlets at 59.74 and 18.39 ppm. After hydrolysis, there are CH,-CH20H singlets at 57.68 pprn (CH,) and 18.32 pprn (CH,). If transesterification were to occur, one should note the appearance of a Si(0-CD2-CD,) quintuplet at about 59.7 pprn at the beginning of the experiment. We never observed quintuplets in this case. MEET I By using CD,OD as alcohol and Si(OC2H5),, a fast transesterification could be expected. CD,OD is characterized by a heptaplet at 49.0 pprn and Si(OC,H,), by singlets at 60.20 pprn (-0CH2-) and 18.40 pprn (-0-CH2-CH,). After 8 days, without adding NH,OH, the spectrum did not change. This indicates that no substitution was detected. Adding NH,OH does not enhance substitution; only the hydrolysis of Si-OCH,CH, is detected by 'H NMR. When 80% of Si(OC2H5), is hydrolyzed, the ',c NMR spectrum shows only the appearance of a CH,CH,OH peak at 58.26 pprn and the residue of a Si(-OCH,CH,) peak at 60.20 ppm. These experiments demonstrate that alkoxy groups do not substitute from the solvent to the silicate. The hydrolysis and condensation of Si(OR), in R'OH as alcohol correspond to the following equation: l3c

[NH,] (M L-I)

[H,Ol (M L-')

Si(OR), + 2 H,O

[HzO]/[-0-C,H5]

R'OH +

Si0,

+ 4 ROH

The solvent acts only as a solvent and not as a reactant, because the alkoxy group is not substituted.

' H NMR The solvents used for these experiments were C2D50Dand CD30D. Because the solvents are deuterated, they are not detected by NMR. Therefore, the NMR spectra are very clear (Fig. 1) and they present the Si-0-CH,-CH, quadruplet (J = 7 Hz; 3.870, 3.835, 3.800, 3.765 ppm) and the Si-0-CH2CH, triplet (1.242, 1.207, and 1.172 ppm). To determine if water vapor contaminated the preparation, test tubes without NH, were monitored during several weeks. After 1 month, only 4-6% of the solution was hydrolyzed, depending on the solvent. This contamination was not taken into account in the following kinetic studies, when adding NH,. In these four experiments, the evolution of the spectrum is identical. After adding NH,, the mobile 0 - H species are characterized by a line at 4.8, 4.9 pprn and do not interfere with other 'H signals. The quadruplet for CH, silicate does not shift, except for a slight shoulder on each quadruplet line for experiments 3 and 4. The same phenomenon occurs for the -OCH,CH, triplet. The CH,CH,O H molecules appear as a quadruplet (3.63-3.53 ppm) and a triplet (1.201.13 ppm). The two quadruplets are in separate ranges. This simplifies the kinetic study. Intensities are integrated for each quadruplet, and the sum total is kept constant (+ /- 1%) during the reaction. To discuss the kinetic laws, a and b designate respectively the initial concentrations in Si(OC,H,), and in H20. Many equations can be used to follow both the hydrolysis and the condensation. The simplest way is to consider the overall reaction.

If x is the molar concentration in SiO,, one can imagine a kinetic law, in which u is the rate of the reaction:

Integration of this differential equation gives rise to: In (b/2) a-x

=

2(b/2

-

a) kt

+ ln(b/2a)

If the water concentration is much greater than that of the silicate, we find the classical law:


CAN. J. CHEM. VOL. 70. 1992

310

3:9

3.7

3.6

35

PPM

FIG.1 . 'H NMR of Si(OC2H,), hydrolysis in C,D,OD (ETET 1) after (a) 2 min; (6) 66 rnin; (c) 276 min; (4456 rnin.

a In - b'kt a-x

-

or a-x In -

- -btkt

with

b'

-b

As an example, Fig. 2 shows plots of the evolution of the silicate concentration (in In units) as a function of time for the two experiments ETET 1 and ETET 2. In each experiment, more than 80% of the initial silicate has disappeared. Two straight lines are obtained, which give rise to apparent rate constants kt = 0.0033 min-I (ETET 1) and k' = 0.043 rnin-' (ETET 2). The ratio of 12 between them should correspond to a mean ratio of 12 between the "mean" b' values. The initial b values are in the ratio 3.6 and the value 4.7 is obtained from the (b-a) ratio. For ETET 2 only, this approximation can be used (b/a 49) and the linearity of the plot is a proof that the law is indeed of order 1 versus the silicate concentration. In Fig. 2, we have plotted the comparison of the experiments ETET 1 and MEET 1 in which the concentrations of both silicate and water (and NH,) are the same, the solvent being the only distinct parameter. The slopes of the lines show that, in CD,OD, the apparent rate constant kt is three times greater than for C2D50D(MEET 1, k' = 9 X min-'; ETET 1, kt = 3.3 X min-I). The k' values for all experiments are listed in Table 2.

As mentioned above and in Table 1, the ratio b/a is not very great for ETET 1, MEET 1, and MEET 2 and it could be four times smaller if we consider the ratio (H20/(Si-0C2H5 group). We have imagined four reactions and the most probable kinetic laws: [1I

k Si(OC2H5), + H 2 0+ SiOH(OC2H5)3+ C2HsOH a b

v = k(4a-x) (b-x) = &/dt

-

(total hydrolysis)

hydrolysis

+ condensation


E L BAKALI ET AL

- 2.5

0

50

100

150

200

250

300

350

LOO

L50

500

Time (min) FIG. 2. Evolution of the ethyl silicate concentration during NH3-H20 hydrolysis in C2D50D:[H20]/[Si-0C2H5]= 0, 12.3; A, 1.025.

- 2.5

0

50

100

150

200

250 300 Time(min1

350

LOO

450

500

Evolution of the ethyl silicate concentration during hydrolysis by NH3-H20 in A , CD,OD and 0, C2D50D:[H20]/[Si-0C2H51

CAN. J. CHEM. VOL. 70, 1992

TABLE 2. Kinetic results of Si(OC,H,), hydrolysis Experiment

Solvent

Si(OC,H,) (M L-')

[H,O]/[Si-OR]

kt (min-I)

k L mol-I min-'

C2DSOD C2DSOD CD30D CD30D

0.40 0.12 0.40 0.20

1.025 12.3 1.025 2.15

0.0033 0.0430 0.0096 0.0134

0.0028 0.0075 0.0075 0.0079

2b0

300

-


ETET 1 ETET2 MEET1 MEET2

0l

0""'""

loo

l

-

'

.

'

l

LOO

-

'

sdo

-

*

w

Ti me (min) Frc. 4. Plots of ln[(b/2-x)/(a/x)] as a function of time for the four experiments corresponding to the global equation: Si(OC2H5),+ 2H20+ SiO, + 4C2H50H.u = k[Si(OC2H5)41[H,Ol; a = [Si(OCZHS)dlini,,al; b = [HZOlini,ial.

We have plotted, for these five kinetic laws, the evolution of a term that is linear versus time (example: case [I]: ln[(b-x)/(a-x)] = (6-a)kt + ln(b/a). The best linearity for the four experiments was obtained by using eq. [4] and the kinetic law v = k(a-x) (b-2r) = dx/dt = -d(a-x)/dt. The four plots are collected in Fig. 4, in which ln[(b/2-x)/ (a-x)] is plotted as a function of time; the mean slope of each fitted straight line corresponds to 2(b/2-a)k. These values of k are listed in the last column of Table 2. The experimental equation v = k(a-x) (b-2x) followed by NMR measurements proves that the overall reaction [4] is not the elementary mechanism. It corresponds surely to a second-order hydrolysis reaction such as [I]. The decrease in silicate concentration, as measured by NMR, is slow compared to further condensation steps that release water.

Discussion The solvent plays a considerable role. Hydrolysis rate increases by a factor of 3 when CD30D is used instead of CzD50D. We would expect two separate steps for Si-OCH-2-CH3: (i) hydrolysis leading to monomeric or oligomeric Si-OH species; (ii) condensation of these hydrolyzed species leading to siloxane Si-0-Si bridges and to the growth of silica particles. But at mid-stage, the liquid is already milky. Therefore, during the synthesis of homodispersed silica particles (Si(OC2H5), + C,D50D), particles grow during the hydrolysis step. The presence of large particles (I+ 50 nm) leading to a milky aspect is accompanied by nonhydrolyzed Si-0-CH2CH3 groups. The Ostwald (13) ripening model, described by Fischmeister and Grimwall (14), implies that nuclei are

-

1617


EL BAKALI ET AL.

formed as they grow. As particles smaller than a critical size dissolve, they are surrounded by a zone of excess solute that finds its way to particles larger than r*. This model does not seem in accordance with our experimental results in which large particles exist with Si-OR groups. Condensation and particle growth occur at the same time as hydrolysis. One can suppose that condensation reactions leading to dimers, polymers, and particles occur very rapidly after SiOH groups are formed by hydrolysis. The hydrolysis rate is therefore the slow step, but the water concentration does not follow eq. [3] for hydrolysis because condensation gives rise to water. If the condensation rate was very slow, the kinetic law should follow reaction [3]. This last model gives rise to curves instead of straight lines, which suggests that condensation is the fast step. A similar point of view has been developed (8) in the case of the hydrolysis of Si(OCH,), in alkaline media (KOH 1 M) because the 2 9 ~NMR i signals do not display any intermediate molecules which apparent Si-OH group. Even if the final state in basic media when Si(OCH,), is used is a gel, which we have also seen in kinetic experiments with much greater rates than for Si(OC2H,),, and even if homodispersed Si02 particles are obtained when Si(OC,H,), is used as the starting material, this experiment suggests that, in both cases, condensation is the fast step, which enables one to give evidence of nonhydrolyzed Si0C,H5 groups in silica particles. Even after a long time, such groups are still present and are only broken by a thermal treatment at 400°C ( 5 ) . Increase of the rate constant k by increasing the ratio H,O/ silicate in ETET 2 could result from differences in growth kinetics of the silica microparticles, because it has been proven3 that smaller concentrations of NH, and H,O give rise to smaller particles at the end of the experiments. The heterogeneity of these systems may therefore contribute to different rate constants. Nevertheless, we have definitely established that the role of the solvent (substitution of CD,CD20D by CD,OD) strongly modifies the kinetics law although we have proved that ethyl silicate does not lead to methyl silicate in CD30D by transesterification. The role of the solvent should involve the ionization of OH- in these ,E. El Rhess, I. El Bakali, and R. Mercier. Manuscript submitted for publication.

NH,-H20 solutions. The speciation of the protonic groups (OH-, NH,, H20) can be different in C2D,0D and CD,OD, arising from different dielectric constants (E methanol 33; E ethanol 24); the viscosity of the solvent could be another factor explaining this behaviour difference (20°C; Pa s). qcH30H 6 X lo-, Pa s; qCH3cH20H 12 X

-

-

-

-

Conclusion The solvent affects the kinetics of hydrolysis. CD,OD is more effective than C2D50D. ',c NMR demonstrates that no transesterification occurs between the silicate and the alcohol. Si(OC,H,), produces stable colloidal systems with SiO, particles in the 0.1-0.5 km range. The relative rates of hydrolysis and condensation differ for both systems; the condensation rate is much faster than the second-order hydrolysis i study in alkaline media reaction rate. A complete 2 9 ~NMR should provide further information about hydrolysis and condensation. 1. W. Stober, A. Fink, and E. Bohn. J. Colloid Interface Sci. 26, 62 (1968). 2. T. Shimohira and N. Tomuro. Funtai Oyali Funmatsuyakin, 4, 19 (1976). 3. K. Shimada, Y. Fukushige, Y. Hirata, and K. Nishimuta. Kogoshina Daigku Kogakubu Kenkya Mokoku, 26, 53 (1984). 4. G. G. Tan, B. D. Bowen, and N. Epstein. J. Colloid Interface Sci. 118, 290 (1987). 5. E. El Rhess. Doctoral Thesis, Universite de Franche-ComtC. 1988. 6. C. C. Lin and J. D. Basil. Mater. Res. Soc. Symp. Proc. 73, 585 (1986). 7. I. Artaki, M. Bradley, T. W. Zerda, and J. Jonas. J. Phys. Chem. 89,4399 (1985). 8. C. A. Balfe and S. L. Martinez. Mater. Res. Soc. Symp. Proc. 73, 261 (1986). 9. W. G. Klemperer, V . V . Mainz, and D. M. Millar. Mater. Res. Soc. Symp. Proc. 73, 3 (1986). 10. R. A. Assink and B. D. Kay. Mater. Res. Soc. Symp. Proc. 32, 301 (1984). 11. R. A. Assink and B. D. Kay. J. Non-Cryst. Solids, 99, 359 (1988). 12. B. D. Kay and R. A. Assink. Mater. Res. Soc. Symp. Proc. 73, 157 (1986). 13. W. Ostwald. Z. Phys. Chem. 34,495 (1900). 14. M. Fischmeister and G. Grimvall. Mater. Sci. Res. 6, 119 (1973).