Stöber Synthesis: Monitoring and Kinetic Models

• 29Si-NMR, conductimetry, and photon correlation spectroscopy are used to monitor intermediate concentrations during Stöber synthesis.
• The Stöber synthesis involves the ammonia-catalyzed hydrolysis of tetraethoxysilane in a batch reactor.
• Assessed extreme models of the process by examining how initial composition affects transients.
• Initial composition was examined over a broader range than previously attempted.
• Trends suggest nucleation is rate-limited by the hydrolysis of the singly hydrolyzed monomer.
• The trends are consistent with the aggregation model discussed by G. H. Bogush and C. F. Zukoski and by M. T. Harris.
• Trends are inconsistent with a growth model without aggregation.
• Key words include SiO2 colloids, 29Si-NMR, kinetic model, precipitation, and aggregation.

Introduction

• Monodispersely sized micrometer-scale spherical colloids of metal oxides are important.
• Their synthesis by the hydrolysis of metal alkoxides is of particular interest.
• Tetraethoxysilane (TEOS) is hydrolyzed in a solution of low-molecular-weight alcohol, water, and ammonia to make silica colloids.
• Average particle size can be varied from 2000 to 10 nm by varying the amount of ammonia and of water.
• Only a limited range of the initial solution composition provides self-sharpening growth to relatively large (micrometer-scale) colloids.
• The paper aims to help understand what is special about these compositions.
• 29Si-NMR, conductimetry, and photon correlation spectroscopy data are used to monitor the behavior.
• Reactions systems are used to elucidate the nucleation mechanism.
• The aggregation model of Zukoski/ Harris is tested against experimentally derived profiles of nucleation.
• It is tested against particle size distribution.
• Initial compositions around the optimal compositions are used (0.1-0.5 M TEOS, 0.5-17 M H2O, and up to saturated ammonia concentrations, all in ethanol).
• This is broader than in previous work.
• Data are consistent with the arguments by Zukoski and by Harris, explaining the deterioration of average size and monodispersity as one moves away from the optimal composition.

Process Parameters and Proposed Mechanisms for the Stöber Process

• The reactions occurring during ammonia-catalyzed hydrolysis and condensation of TEOS are of several types:
• NH3 + H2O = NH+ + OH¯: ionization of ammonia.
• =Si-OEt + H2O = =Si-OH + EtOH: hydrolysis of ethoxy groups.
• =Si-OH + OH¯ = =Si-O¯ + H2O: ionization of hydrolyzed monomers.
• =Si-OEt + =Si-OH = =Si-O-Si= + EtOH: alcohol producing condensation.
• =Si-OH + =Si-OH = =Si-O-Si= + H2O: water producing condensation.
• Et is an ethyl group.
• There are a number of distinct =Si-OEt and =Si-OH environments.
• There can be condensation reactions involving deprotonated =Si-O¯ groups.
• It is possible for an intermediate to phase separate from solution.
• Despite the possibility of a huge number of intermediates (as in acid), the batch Stöber process analyzed by a wide variety of techniques.
• Techniques include 29Si-NMR, conductivity, Raman spectroscopy, pH monitoring, gas chromatography, 'H-NMR, and molybdate analysis.
• The unhydrolyzed TEOS is the predominant species in solution.
• Only the singly hydrolyzed monomer (Si (OEt)3(OH)) is evident as a reaction intermediate.
• Any processes subsequent to the first hydrolysis are relatively fast.
• It is not clear whether the first hydrolysis is rate-limiting, or the second hydrolysis/condensations are.
• The nucleation process may involve autoaccelerating polymerization or precipitation of supersaturated intermediates, or both.
• Bailey and Mecartney detected large polymers with cryo-TEM which phase-separated from solution to form dense particles.
• One goal is to measure transients that can help assess the two extreme nucleation hypotheses.
• In ordinary Stöber systems there is only one detectable reaction intermediate.
• An unusual initial condition must be specially prepared to exhibit transients of variously hydrolyzed monomers/dimers.
• This will suggest that autoaccelerating polymerization is an unlikely mechanism for nucleation.
• A second goal is to assess the two extreme models describing how the particle size distribution matures.
• One model is a “growth-only” model based on LaMer's synthesis of monodisperse colloidal particles by homogeneous precipitation.
• The model neglects aggregation and holds that a narrow distribution of colloids can be achieved by a brief burst of nucleation.
• Growth is defined here as the addition of soluble species directly to the particle surface.
• Van Blaaderen et al. proposed a model whereby nucleation is controlled by the aggregation of soluble species.
• Subsequent growth is controlled by the surface reaction.
• In this scenario all particles grow at the same rate.
• This has been shown to be more consistent with the average final particle size and variance of Stöber spheres.
• The other extreme model is an “aggregation-only' model.
• Recent evidence supports continuous nucleation and aggregation of the nuclei to form a narrow final particle size distribution.
• Population balance equations must be solved to predict the particle size distribution.
• Bogush and Zukoski assumed that the nucleation rate was proportional to the concentration of hydrolyzed and deprotonated monomers.
• They showed that the aggregation kernel described the maturing of the average particle size by using Brownian dynamics with classical DLVO interactions.
• Harris assumed that the nucleation rate was proportional to the experimentally measured rate of TEOS consumption.
• Treatments of Bogush and Zukoski and of Harris differ somewhat on these points.
• Both treatments successfully predicted the average size and size distribution over the limited composition range of Stöber synthesis.
• To assess these extreme models, the reactions will also be monitored with PCS.
• Data are not consistent with the growth-only model.
• The understanding of nucleation will allow to show that our data are consistent with the aggregation-only model.
• While it seems reasonable that direct growth should also be at play, this test confirms that aggregation is extremely important.

Experimental

• Tetraethoxysilane (TEOS, 98% purity, Aldrich); deionized, distilled, and filtered H₂O; and a standard 4.96 N ammonia solution (Aldrich) were dissolved in ethanol.
• In some samples, anhydrous sodium perchlorate (NaClO4, Mallinckrodt) was added to 4 mM.
• For seeded experiments, the method from Bogush and Zukoski was used.
• For a two-step synthesis, TEOS, water, and 0.1 N hydrochloric acid solution (Johnson Matthey) were dissolved in ethanol.
• initial compositions were (a) 0.5 M TEOS/0.5 M H2O/0.0032 M HCl and (b) 1.43 M TEOS/2.86 M H2O/0.001 M HCl.
• After (a) 150 min and (b) 222 min, 0.1015 N sodium hydroxide solution from Aldrich, ammonium hydroxide, and additional water were added to the initial mixture.
• Composition adjusted to (a) 0.5 M TEOS/6 M H2O/0.0032 M HCl/0.0032 M NaOH/0.02 M NH3, and to (b) 1.43 M TEOS/2.86 M H2O/0.001 M HCl/0.001 M NaOH/0.01 M NH3.
• As soon as the base was added, 29Si-NMR was followed while the solution was homogeneous.
• Photon correlation spectroscopy (PCS), conductimetry, and 29Si-NMR were used to monitor the reactions.
• Samples for PCS were prepared by diluting 0.1 ml of the colloidal sol in 3 ml ethanol.
• PCS measurements were performed on a Coulter N4 SD at a scattering angle of 90°.
• Conductivity measurements were performed with a conductance meter from YSI.
• 29Si-NMR measurements were performed on GE 500 and Varian 500 spectrometers at a spectral frequency of 99.3 MHz.
• To achieve sufficient signal to noise ratios, at least 72 scans were acquired using a 90° pulse angle.
• every sample, 1 wt% Cr(acac)3 in ethanol was added as a paramagnetic relaxation agent.
• A pulse delay of 3 s was used so that analysis would be long enough to ensure complete relaxation between scans.
• Most of the product colloidal sols remained stable for months, though the colloids produced from solutions with [NH3] ≥ 0.2 M settled after ~2 days.
• Surprisingly, the solution at low water and ammonia concentrations (0.5 M TEOS/1.5 M H2O/0.1 M NH3) formed an almost transparent gel after 2 months.
• Apparently, so many nuclei are continuously formed with such a slow growth rate that a gel forms among them, producing little visible light scattering.

Results

• To help discern between two competing nucleation mechanisms such as autoaccelerating polymerization or precipitation, we first consider the specially prepared system.
• TEOS prehydrolyzed at low pH as to provide both monomers and dimers, based on the work of Ng and McCormick.
• The protocol was carefully chosen to provide only these species This will allow a straightforward interpretation of the results.
• Base added to achieve Stöber conditions.
• 29Si-NMR spectra were collected after the base was added.
• The conventional Qnotation is used (Q denotes a quadrafunctional Si site, subscript denotes the number of siloxane bridges attached to the Si atom, superscript represents the number of silanol bonds on the Si atom).
• The monomers and dimers which are hydrolyzed more than once and the singly hydrolyzed ring all disappear as soon as the base is added.
• The poorly hydrolyzed monomers and dimers remain, reacting only very slowly. In the discussion section, two possibilities are tested (autoaccelerating polymerization or precipitation).
• Next, examine the ordinary single step Stöber system over a range of initial solution composition.
• Only one reaction intermediate (singly hydrolyzed monomer) is visible by 29Si-NMR.
• The reactions have to be slow enough that dynamics can be captured.
• Strategy is to select batch reactor compositions that exhibit relatively slow kinetics.
• TEOS decays monotonically and the singly hydrolyzed monomer exhibits a maximum with time..
• The fractional yield is measured as the fraction of the Si nuclei which vanishes from the solution NMR spectra because of the inefficient rotational averaging.
• Mass balance calculations on our system following centrifugation confirm that most of the lost Si (>90%) goes to make particles that are detectable with PCS.
• If the aggregation model of Zukoski/Harris (1–3) is correct, the nucleation profile can be deduced from the singly hydrolyzed monomer concentration profile.
• The “vanished'' Si is largely incorporated into particles large enough to be detected by PCS.
• The measured yield and the measured size can be used to calculate a number density.

Effect of ammonia concentration

• Increasing the initial ammonia concentration (up to 1 M [NH3]) increases the particle size.
• Relatively low ammonia concentration is used to slow the reaction for study using 29Si-NMR.
• With higher ammonia concentration particles also have larger average and a narrower distribution.
• The profile of the TEOS concentration (from 29Si-NMR) for 0.15 M TEOS/8.8 M H2O/0.05-0.3 M NH3 is showed
• TEOS decays faster at higher ammonia concentration.
• Profiles of the singly hydrolyzed monomer concentration (from 29Si-NMR), the conductivity, the average particle size (from PCS), and the nominal number density (calculated from the yield and average particle size from the PCS).
• As more ammonia is used, the singly hydrolyzed monomer appears for shorter periods of time.
• The maximum concentration of the singly hydrolyzed monomer is affected very little by the ammonia concentration.
• The conductivity is strongly influenced by the charge of the singly hydrolyzed monomer.
• The time to maximum conductivity coincides with the time to maximum singly hydrolyzed monomer concentration.
• These trends were also confirmed at higher TEOS concentration, which allows the formation of more Si nuclei
• The NMR spectra also allow to calculate free water concentration.
• It remains about the same as the initial water concentration.
• It appears that there is little need for the models to consider transients in dielectric constant or in water effects on activity coefficients.

Effect of water concentration

• The expected effect of water on the particle size distribution is confirmed in Fig. 3: there is an optimal water concentration for the maximum particle size (here the optimum [H₂O] is ~9 M).
• The maximum size, though, does not always correspond to the narrowest distribution.
• The TEOS concentration profiles at different water concentrations show that the TEOS consumption becomes faster with increasing water concentration.
• As the water concentration increases, the singly hydrolyzed monomer and conductivity appear more briefly.
• The maximum singly hydrolyzed monomer concentration does not change much.
• As expected, more water speeds the first hydrolysis.
• More water serves to increase the rate of consumption of the singly hydrolyzed monomer.
• The conductivity profiles suggest that more water increases the deprotonation of the monomers.

Effect of ionic strength

• In the solutions discussed so far, the ionic strength is free to vary over the course of the reaction.
• The initial ionic strength changes as change the initial composition.
• For these reasons need to assess whether these ionic strength variations significantly affect the kinetics.
• NaClO4 is provided at a high enough concentration (4 mM) that the ionic strength is virtually constant
• The ionic strength does, of course, influence the DLVO contribution to the aggregation kernel.
• The final size distribution clearly shows that with added salt larger particles are produced and that

Discussion

• The two-step experiment creates unusual initial conditions for base-catalyzed reactions where variously hydrolyzed monomers and dimers are present.
• The results allow some observations on nucleation mechanism.
• It is unlikely that autoaccelerating polymerization can account for the data.
• The dimers (Q and Q) obviously do not condense faster than the monomer.
• The two-step experiment also shows that both unhydrolyzed and singly hydrolyzed dimers (Q and Q1, respectively) are soluble without precipitation.
• If species which are larger than a monomer were formed in the Stöber synthesis, 29Si-NMR should be able to detect them, but it does not.
• if the autoaccelerating polymerization idea is correct, then the doubly hydrolyzed monomer (Q3) must condense at high rate.
• The condensation rate constant between Q's must be greater than 104 liters/mol-h.
• It is noted that such a huge rate constant has never been reported from any silanol condensation reactions.
• Under the autoaccelerating polymerization scheme, the Q produced by the condensation reaction between Qos should disappear quickly.
• In initial Stöber process, nucleation involves the precipitation of the first insoluble species formed.
• It is proposed that the precipitating species is the doubly hydrolyzed monomer.
• We have not yet considered whether, after the onset of nucleation, the particle size distribution matures by growth, by aggregation.
• Now the conditions for the Stöber process must be identified
• In this case, nucleation limited by the hydrolysis of the singly hydrolyzed monomer
• Not yet considered whether, after the onset of nucleation, the particle size distribution matures by growth, by aggregation (with continuous nucleation), or by a combination.
• If the particle size distribution matures primarily by aggregation with continuous nucleation, then the following reaction scheme describes the Stöber process.
• Q + H2O → Q + EtOH.
• Q + H2O → Q + EtOH → SiO1.8(OH)0.3(OEt)0.1 + 2.9 EtOH + 0.1 H2O.
• The reverse hydrolysis is expected to be very slow (15); in fact, we considered such a model, but the data were insensitive to the reverse hydrolysis rate constant, meaning that reverse rate is negligible here.
• • (i) particle yield rate is equal to the second hydrolysis rate
• • • evaluated from 1 and 2 transients should be:

independent of presence of particles (if ionic strength remains constant), independent of water concentration, and dependent (increasing) on the ammonia concentration.

Assessing Extreme Views of PSD Maturation

• It can be inferred from the change in the number concentration of detectable particles whether the LaMer theory is correct, i.e., is single nuclei formation.
• It is argues that the soluble silica concentrations were high enough to support continuous nucleation until late in the reaction. Suggesting that these species induce detectable particle concentrations above 0.01 M
• Suggesting seeded experiment
• If the aggregation model is complete, then the only way Si could disappear from the NMR spectra is by nucleation
• has the rate-limiting step of profiles is directly proportional to linear function of yield
• As the concentration increases, the rates are so as the maintain []

Conclusions

• the final particle size and its relation to the nucleation rate indicate high sensitivity to the underlying charged structures
• By increasing one factor, say ammonia, the water concentration has its effects balanced
• this is tested using alternative ionic concentrations as an effect. These results lead to the conclusion the system is able to accurately predict when (or less thand when it is not possible to see)