Mechanism of Formation of Metal Sulfide Ultrafine Particles in Reverse

The mechanism of formation of ultrafine CdS, ZnS, and their composite particles by the injection of H2S into reverse micelles was studied. The particl...
0 downloads 0 Views 191KB Size
Download PDF
92

Ind. Eng. Chem. Res. 1997, 36, 92-100

Mechanism of Formation of Metal Sulfide Ultrafine Particles in Reverse Micelles Using a Gas Injection Method Hiroshi Sato, Yoritaka Tsubaki, Takayuki Hirai,* and Isao Komasawa Department of Chemical Engineering, Faculty of Engineering Science, Osaka University, Toyonaka, Osaka 560, Japan

The mechanism of formation of ultrafine CdS, ZnS, and their composite particles by the injection of H2S into reverse micelles was studied. The particle formation process was followed by the change in UV-visible absorption spectra. The kinetics of the whole process including dissolution of H2S, nucleation, particle growth, and coagulation was analyzed from time-course changes of the size and number of formed particles. The dissolution of H2S was the principal ratedetermining step, and most of the dissolved H2S was consumed for particle growth. The particles formed in the present gas injection method were larger in size than those in the previous solutionmixing method in most cases. A kinetic scheme based on the distribution of the species among the micelles was then proposed, and this successfully explained the particle growth. Composite particles of CdS and ZnS having mixed crystal or core-shell structures were also prepared, and the application of these particles as photocatalysts was investigated. Introduction Ultrafine particles have become increasingly important in recent years. The interest is directed to the size dependent energy structure of these particles and to the preparation and stabilization of uniform ultrafine particles since the functions of ultrafine particles such as the electric and optical properties and catalytic activities are primarily controlled by the energy structure. Reverse micelles consist of self-organized extremely small water droplets stabilized in organic solvents by surfactant molecules. Although the micellar droplets are thermodynamically stable, they do exhibit a dynamic exchange of their contents by fusion of droplets and redispersion of dimer droplets. This micellar exchange process facilitates reaction between reactants dissolved in different droplets. Numerous papers describing preparation methods of ultrafine particles in reverse micelles have been published. Most of these deal with preparation based on mixing two reverse micellar solutions containing reactants separately and using the micellar exchange process (solution-mixing method). Composite ultrafine particles have also been prepared by this method (Ravet et al., 1987; Hirai et al., 1994b; Cizeron and Pileni, 1995). Ultrafine particles may also be prepared by the injection of a reactant gas into reverse micelles containing the other reactants (gas injection method). The preparation of CdS (Meyer et al., 1984), CaCO3 (Kandori et al., 1988), and composite Fe and Cu (Tanori et al., 1995) particles by this method has been reported. This method is important practically since the quantity and the injection rate of the gas can be controlled over the wide range and further since the volume of the solution is not increased by the addition of the gas. A kinetic approach is of great importance in understanding the mechanism of formation of the ultrafine particles and in obtaining a quantitative model for the total process. This can lead to further development in the preparation of particles having special compositions and functions based on a control of the rates of the successive steps in the particle formation process. There have been only a few reports for the kinetic * E-mail: [email protected]. Fax: +81-6-8506273. S0888-5885(96)00141-8 CCC: $14.00

analysis of the particle formation process in reverse micelles, however. For CdS particles produced by the solution-mixing method, a rapid coagulation in the reverse micelles has been found to explain a change in absorbance at the initial stage of the particle formation (Towey et al., 1990). In previous work (Hirai et al., 1994a), the processes in the formation of CdS and ZnS particles by this method were analyzed, and the rapid coagulation model was found to be applicable only for the initial part and a kinetic scheme including the later part of the coagulation process was proposed. The scheme was also found to be applicable for PbS (Hirai et al., 1995) and CdS and ZnS coprecipitated particles (Sato et al., 1995). It is now necessary in the present work to extend the previous kinetic studies for the solution-mixing method to the gas injection method based on the mechanism of formation of metal sulfide particles by the injection of H2S gas into reverse micelles. Here the gas absorption step is possibly the slowest, and the time involved for the gas injection is certainly much greater than that for the conversion of the reactants to the particles in the micelles. The whole process from the dissolution of H2S into the organic phase (gas absorption), the transfer of H2S from the organic phase to the water cores in the reverse micelles, and the formation of particles in the water cores was followed continuously using absorption spectra. A kinetic scheme for the above steps in particle formation was proposed on the basis of the statistical distribution of species and particles among the micelles. The preparation of composite particles of CdS and ZnS having mixed crystal or core-shell structure was also attempted. The photocatalytic activity of the composite particles for the cleavage of water to generate H2 was then investigated. Experimental Section Sodium bis(2-ethylhexyl) sulfosuccinate (AOT), cadmium nitrate, zinc nitrate, and sodium sulfide were supplied by Wako Pure Chemical Industries, hydrogen sulfide was supplied by Sumitomo Seika, and isooctane (2,2,4-trimethylpentane) was supplied by Ishizu Seiyaku. All reagents were used without further purification. Distilled water was filtered using a 0.45-µm © 1997 American Chemical Society

Ind. Eng. Chem. Res., Vol. 36, No. 1, 1997 93

membrane filter, and any dissolved oxygen was purged by Ar gas bubbling, prior to use. The reverse micellar solution was prepared by dissolving 0.1 M (M ) mol/L) AOT in isooctane, followed by filtration using a 0.2-µm membrane filter. The water content (water to surfactant molar ratio, W0 ) [H2O]/[AOT]) was varied in the range 3-10. Aqueous solutions containing Cd(NO3)2 or Zn(NO3)2 were prepared daily. The reverse micellar solutions containing the reactants were prepared by injection of a required amount of aqueous solution and were used within a few minutes. CdS particles were prepared by the injection of H2S into a reverse micellar solution containing Cd(NO3)2, using a gas-tight syringe (Hamilton Co.). The quantity of gas injected was only a few percent greater than the stoichiometric amount. The time duration of the gas injection varied from 2 s (rapid gas injection method) to 13 s (intermittent gas injection method). The instant when the injection was started was defined as t ) 0 s. The temperature of the reverse micellar solution was maintained at 25 °C. Zn(NO3)2 was employed for the preparation of ZnS ultrafine particles instead of Cd(NO3)2. Otherwise, the operations were the same as those for CdS particles. Composite particles expected to have mixed crystal structure were prepared by the coprecipitation of CdS and ZnS using the solution-mixing method (Hirai et al., 1994b; Sato et al., 1995). In the present work, coprecipitated ultrafine particles (Cd1-xZnxS) were prepared by injecting H2S into a reverse micellar solution containing both Cd(NO3)2 and Zn(NO3)2. Two parameters Cf and xf were defined in order to represent the feed composition of each reactant in the reverse micellar solution, where

Cf ) [Cd2+] + [Zn2+]

(1)

xf ) [Zn2+]/([Cd2+] + [Zn2+])

(2)

The preparation of composite ultrafine particles having a core-shell structure was also attempted. ZnScoated CdS particles ((ZnS)0.5(CdS)0.5) were prepared in the previous study (Hirai et al., 1994b) by alternate and repeated injection of small aliquots of two reverse micellar solutions containing Zn2+ ion or S2- ion into a reverse micellar solution containing CdS core particles. This was done to grow ZnS on the core particles but prevent the nucleation of new particles of ZnS and particle coagulation. In the present case, the core CdS particles were prepared by injecting a stoichiometric amount of H2S into a W0 ) 2 reverse micellar solution containing 5 × 10-5 M Cd(NO3)2. An aqueous solution containing Zn(NO3)2 was added to this solution 10 min following initiation. The resulting solution contained 5 × 10-5 M Zn2+ with a W0 value of 3. The required quantity of H2S was then injected to form the ZnS shell. The preparation of CdS-coated ZnS particles ((CdS)0.5(ZnS)0.5) was also carried out but with the reactants reversed. A Hewlett-Packard 8452A diode array spectrophotometer was used to record the spectral change of the micellar solution following the injection of H2S. H2S was injected into a quartz cell containing the reverse micellar solution (3 mL) dissolving Cd(NO3)2, for the measurement of spectra of periods from t ) 1 s to 3 min. Sufficient mixing was supplied throughout the run by a magnetic stirrer, of which the mixing rate was controlled to ensure complete dissolution of the injected

H2S bubbles into the solution and not to entrain air bubbles from the open surface. With the reaction times (t) greater than 10 min, the reaction was performed in a beaker-type glass reactor (20 mL) with lid. H2S was injected into 10 mL of the reverse micellar solution in the reactor. When injection was complete, the reactor was sealed and shaken by hand. The solution in the reactor was transferred to a quartz cell in order to measure the absorption spectra as required. The diameter of the formed particles (dp) was estimated from the band gap energy of CdS and ZnS semiconductor particles using the equation proposed by Brus (1984). The energy was determined by the absorption spectra as described previously (Sato et al., 1995). The method of irradiation for the photocatalytic generation of H2 is described in detail in a previous paper (Hirai et al., 1994b). A reverse micellar solution (20 mL) containing the particles was irradiated with a 500 W xenon lamp (Ushio UXL-500D) after the solution had been treated with Ar bubbling for 1 h. The amount of H2 generated was determined by gas chromatography with TCD (Shimadzu GC-14B). Results and Discussion Spectral Change during Particle Formation. UV-visible absorption spectra of the solution measured during preparation of CdS particles under different H2S feeding conditions are shown in Figure 1a-d. Figure 1a shows the change in the spectra observed using the rapid gas injection method, for H2S gas injection in 2 s. An increase in absorption intensity for times up to t ) 5 s and a continuous red-shift in absorption threshold are observed. The former effect indicates the increase in the quantity of the formed particles, and the latter indicates the decrease in band gap energy caused by the increase in particle diameter. The conversion of the precursor ions (Cd2+ and S2-) for both nucleation (formation of a particle) and particle growth (addition of ions to the particle) is seen to be complete by 5 s. The absorbance at a peak wavelength measured after t ) 5 s is therefore expected to indicate the total quantity of ions converted to the particles. There was an adequate quantity of H2S to convert all the metallic ions, as the absorbance increased with increasing quantity of H2S up to the stoichiometric value but was then almost unchanged by further increase. This confirms that the stoichiometric quantity of H2S is almost sufficient to convert all the metallic ions. The period required to obtain complete conversion of the ions to the particles (5 s), for the gas injection method, is much greater than that needed using the solution-mixing method (0.02 s) (Hirai et al., 1994a). Several experiments were carried out; the manner in which H2S was supplied to the micellar solution was varied. Firstly, an adequate quantity of H2S was injected into a reverse micellar solution, and this solution was rapidly mixed with the equal volume of another reverse micellar solution containing Cd(NO3)2. This is the same approach as that used in a previous solution-mixing method using Na2S as a reactant (Hirai et al., 1994a), but here H2S was used as the S2- source. The resulting spectral change is shown in Figure 1b. Here, the absorption of the formed particles is observed 0.02 s after mixing, thus indicating that H2S dissolved in the reverse micelles reacts with the Cd2+ ion very rapidly as in the case of Na2S. Secondly, the kinetics of the transfer of H2S from the organic phase to the water cores was examined. In this case, H2S was

94

Ind. Eng. Chem. Res., Vol. 36, No. 1, 1997

Figure 2. Effect of the preparation method on the absorption spectra of particles: (a) CdS; (b) ZnS.

Figure 1. Absorption spectra measured during preparation of CdS particles under different H2S feeding conditions: (a) rapid gas injection method in 2 s; (b) solution-mixing method using H2S dissolved in reverse micellar solution; (c) solution-mixing method using H2S dissolved in isooctane; (d) intermittent gas injection method over a time of 13 s.

dissolved in pure isooctane rather than reverse micellar solution, and this was mixed with the equal volume of reverse micellar solution containing Cd(NO3)2. The feed reverse micellar solution contained double the quantity

of surfactant to make the resultant surfactant concentration the same as that of the other cases. The spectral change obtained is shown in Figure 1c. Here there is no significant difference between the data shown in parts b and c of Figure 1, thus indicating that the transfer from the organic phase to the micellar water core also proceeds very rapidly. With H2S injection for 13 s, an increase in absorption intensity was observed for times up to 30 s, as shown in Figure 1d. The dissolution of H2S into the organic phase (gas absorption) is, therefore, the dominant rate-determining step in the overall particle growth process. The absorption spectra for the gas injection and solution-mixing methods obtained at 10 min after the initiation are compared in parts a and b of Figure 2. The absorption threshold wavelength for the gas injection method is longer than those for the solution-mixing methods. The size of the particles prepared by the gas injection method is, therefore, greater than that for the other methods. The diameter of CdS and ZnS particles prepared using the solution-mixing method is known to depend on the water content of the micellar solution (Towey et al., 1990; Hirai et al., 1994a). Parts a and b of Figure 3 show the effect of water content on the spectra obtained for the gas injection method with CdS and ZnS particles, respectively. The diameters of the CdS particles (dp) at t ) 10 min for W0 ) 3, 6, and 10 are estimated from the spectra to be 3.3, 4.5, and 5.4 nm, and the values for the ZnS particles are 2.4, 3.3, and 3.6 nm, respectively. The corresponding values for the solution-mixing method have been reported to be 3.2, 3.6, and 3.8 nm for CdS and 3.0, 3.2, and 3.3 nm for ZnS, respectively (Hirai et al., 1994a). The effect of W0 on dp for the gas injection method is more notable than that for the solution-mixing method. Kinetic Analysis of Particle Formation. The particle formation process for the solution-mixing

Ind. Eng. Chem. Res., Vol. 36, No. 1, 1997 95

Figure 4. Comparison of the first-order coagulation rate constants for the solution-mixing methods using H2S and Na2S. The data for the solution-mixing method using Na2S are taken from Sato et al. (1995).

Figure 3. Effect of water content on the absorption spectra of particles prepared by the gas injection method: (a) CdS; (b) ZnS.

method using H2S was investigated and compared with that obtained with Na2S in previous work (Sato et al., 1995). The red-shift observed following t ) 0.02 s and expressing coagulation of the particles was also analyzed in the same manner as in that work. There, when particles had achieved diameters greater than about 2.4 nm, the coagulation rate (rc) was found to follow a firstorder kinetics with respect to the concentration of the micelles containing two or more particles (Cmc), since coagulation was considered to proceed in such micelles. Hence

rc ) -dCp/dt ) kmcCmc

(3)

where Cp and kmc are the molar concentration of the particles and the first-order rate constant for coagulation, respectively. Assuming a Poisson distribution of the particles among the micelles (Atik and Thomas, 1981), Cmc is given by Np

Cmc ) Cm

() Cp

∑ i)2 C

i

exp(-Cp/Cm)

{ ∑( ) m 1

) Cm 1 -

i)0

Cp

Cm

i

i! exp(-Cp/Cm) i!

}

(4)

where Np and Cm are the total number of particles in the solution and the concentration of the micelles, respectively. The first-order coagulation rate constant, kmc, was thus calculated from the time-course variation of Cp. The value of Cp was calculated by the total number of ions converted to the particles and the number of ions in a particle, as in the previous work (Sato et al., 1995). Figure 4 shows kmc as a function of particle size for the solution-mixing method using H2S as reactant. The

results obtained using Na2S (Sato et al., 1995) are also plotted for comparison. The coagulation rate constant decreases with increasing particle diameter. This is caused by the increase in the electrostatic repulsion between particles (Hirai et al., 1994a). While some deviations are observed, the values of kmc for both mixing methods seem to be almost identical. Since coagulation is the principal step in the solution-mixing method (Towey et al., 1990; Hirai et al., 1994a), the diameters of the formed particles obtained by the solution-mixing methods using either H2S and Na2S are identical. The mechanism of dissolution of H2S is important since this is possibly the slowest step in the overall particle formation process using the gas injection method. However, only a few data for the analysis were obtained using the rapid gas injection method in which time duration of gas injection was 2 s. This is because the rate of increase in the absorption intensity was very fast and was complete in as short a time as 5 s. The intermittent gas injection method was then employed. In this, small aliquots (0.5 µL) of H2S were injected repeatedly (14 times) into the W0 ) 3 reverse micellar solution (3 mL) containing 1 × 10-4 M Cd(NO3)2 in a quartz cell using a syringe with a repeating dispenser and time interval of 1 s. Thus, a total of 7 µL of H2S was injected over a time period of 13 s. Figure 1d shows the spectral change during this procedure. An increase in the absorption intensity in the initial stage and a continuous red-shift in the absorption threshold are observed as in the case of the rapid gas injection method in 2 s (Figure 1a). The conversion of the precursor ions to particles (Xp) was calculated using the absorbance at the peak or shoulder wavelength which was 400430 nm for the CdS in this case. The time-course changes in Xp, particle concentration (Cp), and size of particles (dp) were calculated using the spectra shown in Figure 1d. These results are shown in Figure 5a, and the results obtained similarly for ZnS particles are shown in Figure 5b. The changes can be separated into two stages. In the first stage which occurs up to about t ) 30 s for CdS and t ) 50 s for ZnS, both Xp and Cp increase, indicating the progress in particle growth and nucleation, respectively. In the case of CdS following this first stage, a slow increase in dp and a decrease in Cp occur; the changes are attributable to coagulation or Ostwald ripening (growth of larger particles ac-

96

Ind. Eng. Chem. Res., Vol. 36, No. 1, 1997

rn + rg ) rdiss

(8)

The concentration of H2S, not dissolved in the water cores, is almost identical to CS,t. Thus, rdiss is expressed by a first-order kinetics with respect to CS,t. Hence

rdiss ) rn + rg ) kdissCS,t

(9)

where kdiss is the rate constant for the dissolution of H2S. Nucleation and particle growth proceed competitively, and sum of these rates is controlled by the rate of slower dissolution as shown from eq 8. The rate of the process slower than the micellar exchange can be expressed by the first-order kinetics with respect to the concentration of micelles containing all required reactants (Hirai et al., 1993, 1994a). Both the rates of nucleation (rn) and particle growth (rg) are thus assumed to be expressed by first-order kinetics, i.e. for nucleation with respect to the concentration of micelles containing both Cd2+ ion and H2S (Cmn) and for particle growth with respect to that of micelles containing Cd2+ ion, H2S, and particles (Cmg), respectively. Hence

Figure 5. Time-course variations of the conversion of precursor ions to particles (Xp), the particle diameter (dp), and the concentration of particles (Cp) during particle formation for the intermittent gas injection method over 13 s. Comparison of the observed data with calculated ones: (a) CdS; (b) ZnS.

companied by the dissolution of smaller particles caused by the difference in solubility). Model for Particle Formation Process. A kinetic scheme for the simultaneous nucleation, particle growth, and coagulation during the first stage of the gas injection method was formulated, having particular emphasis on the rate of dissolution of H2S into the micellar solution. In general, balance equations for the concentrations of the particles (Cp) and the Cd2+ ion (CCd) may be expressed in terms of the rates of nucleation (rn), particle growth (rg), and coagulation (rc). Hence

dCp/dt ) rn - rc

(5)

dCCd/dt ) -rn - rg

(6)

Defining CS,t as the total concentration of the unconverted H2S in the solution, including S2- ion in the water cores, H2S in the organic phase, and H2S in small bubbles in the solution, the balance equation is

dCS,t/dt ) -rn - rg + rinj

(7)

where rinj is the rate of injection of H2S into the reverse micellar solution. The rate of consumption of precursor ions by nucleation and particle growth in the gas injection method is determined by the rate of ratedetermining dissolution of H2S into the solution (rdiss). The concentration of H2S dissolved in the water cores is negligible, and the rate of consumption of H2S is, thus, expressed as

rn ) kmnCmn

(10)

rg ) kmgCmg

(11)

where kmn and kmg are first-order rate constants for nucleation and particle growth, respectively. The values of Cmn and Cmg can be estimated from the probability of containment of the reactants in a micelle. Since the concentrations of micelles are in the range 4.31 × 10-4 to 5.89 × 10-3 M (Hirai et al., 1994a), and these are much larger than the concentration of the reactants, the probability is thus identical to the average number of reactants per micelle. The values of Cmn and Cmg are therefore obtained as follows

Cmn ) Cm(CCd/Cm)(CS/Cm)

(12)

Cmg ) Cm(CCd/Cm)(CS/Cm)(Cp/Cm)

(13)

Finally, the ratio of the respective rates is obtained from eqs 10-13.

rn/rg ) (kmnCmn)/(kmgCmg) ) (kmn/kmg)(Cm/Cp) (14) The rate of coagulation (rc) for the intermittent gas injection method was investigated. However, the diameter of particles formed in the W0 ) 6 reverse micelles was too large to analyze the coagulation kinetics precisely. For W0 ) 3, the coagulation rate constant (kmc) for CdS particles obtained following completion of particle growth (30 s) and the data obtained for the solution-mixing method using Na2S (Sato et al., 1995) are shown to correlate well together as shown in Figure 6. This indicates that the coagulation kinetics for the intermittent gas injection method, in this period, follows the same first-order kinetics as expressed by eq 3. Thus the coagulation rate constant for W0 ) 6 may be estimated from the data obtained from the previous solution-mixing method (Sato et al., 1995) and shown in Figure 4 in the present work. Assuming a linear relationship between the logarithm of kmc and dp

ln(kmc/s-1) ) 58.4-15.9dp/nm

(15)

However, as proposed by Towey et al. (1990), the coagulation rate cannot exceed that of the rapid coagu-

Ind. Eng. Chem. Res., Vol. 36, No. 1, 1997 97

Figure 6. Comparison of the coagulation rate constants for the intermittent gas injection method and for the solution-mixing method using Na2S. The data for the solution mixing method are taken from Sato et al. (1995).

lation in reverse micelles. Thus,

rc e kexCp2

(16)

where kex is the exchange rate constant of reverse micelles and is estimated as 5.6 × 107 M-1 s-1 (Hirai et al., 1994a). Application of the Model for Simulation of Particle Formation. The proposed model contains four rate constants (kdiss, kmn, kmg, and kmc) to be determined. These are estimated from the time-course changes of the observed values of Cp, Xp, and dp shown in Figure 5a. The values of Cp, dp, dCp/dt, and dCCd/dt obtained at t ) 8 s (the shortest reaction time at which data are obtained) are 4.2 × 10-8 M, 4.2 nm, 3.5 × 10-9 M/s, and -4.1 × 10-6 M/s, respectively. The rates of nucleation (rn), particle growth (rg), and coagulation (rc) are then estimated. Values for kmc and Cmc are estimated as 6.9 × 103 s-1 and 1.0 × 10-12 M using eqs 15 and 4, respectively. Thus, rc at time t ) 8 s is calculated to be 5.0 × 10-16 M/s using eq 3. The rates rn and rg are then estimated to be 3.5 × 10-9 and 4.1 × 10-6 M/s, respectively, from eqs 5 and 6. The value of rg obtained is found to be greater than both rn and rc, and it is almost identical to the rate of conversion of Cd2+ ion (-dCCd/dt). This indicates that most of the precursor ions are consumed by particle growth. The rate constant for dissolution of H2S, kdiss, was estimated from eqs 6 and 9. In this calculation, the value of CS,t at t ) 8 s is needed. CS,t is estimated as 3.0 × 10-5 M, since 6.2 × 10-5 M of H2S is injected in 8 s and 3.2 × 10-5 M of H2S is converted to particles as shown by Xp in Figure 5a. The rate constant, kdiss, is now obtained as 0.14 s-1. Two rate constants, kmn and kmg, have not been determined. Since the rates of these steps are controlled by the slower dissolution step, the exact values of these rate constants cannot be obtained by analysis based on the gas injection method. Thus, the ratio of rate constants (kmn/kmg) is estimated from rn, rg, Cp, and Cm using eq 14. Since the value of Cm for W0 ) 6 reverse micelles has been reported as 9.27 × 10-4 M (Hirai et al., 1994a), the value of kmn/kmg is finally calculated to be 3.8 × 10-8. The complete process in particle formation was simulated using eqs 3-7, 9, and 14-16 using the obtained values of kmc, kdiss, and kmn/kmg. The rate of injection of H2S (rinj) was 7.6 × 10-6 M/s since 1 × 10-4 M H2S was

Figure 7. Comparison of the observed and calculated particle diameters for the rapid gas injection method at 1 min. The calculated values are obtained using the parameter values in the present intermittent gas injection method and in the previous solution-mixing method.

injected in 13 s. The calculated values for dp, Xp, and Cp from the model are indicated by the respective lines shown in Figure 5a. These calculated values agree well with the observed data, but the increase in dp and thus decrease in Cp observed after 40 s are not shown in the simulation. A possible explanation of this deviation may be attributable to Ostwald ripening due to the very large size of the particles in this case. The simulation for ZnS particles was also carried out in the same manner from the time-course variations shown in Figure 5b and the following relationship between kmc and dp obtained on the basis of the solution-mixing method data (Sato et al., 1995).

ln(kmc/s-1) ) 32.3-9.03dp/nm

(17)

The estimated values of kdiss and kmn/kmg are 0.15 s-1 and 2.0 × 10-7, respectively, and the simulated timecourse variations are shown in Figure 5b. These show good agreement with the observed data. The values of the rate constant of dissolution step, kdiss, agree well for both CdS and ZnS particles and are thus almost independent of the material of the particles, as expected. The present results were also applied to the case of the rapid injection method by changing the time duration and the rate of the injection, rinj, to 2 s and 5 × 10-5 M/s ()10-4 M/2 s), respectively. The simulation was also applied to other W0 cases using reported values of Cm (Hirai et al., 1994a) and previously obtained relationships between kmc and dp for each W0 (Sato et al., 1995). Other parameters were assumed to be independent of W0. The calculated and observed diameters for CdS and ZnS particles for t ) 1 min are shown in Figure 7. The calculated values agree well with the observed data. The proposed model is therefore shown to explain successfully the diameter of formed particles both in the rapid gas injection method and in the intermittent gas injection method. Preparation of Composite Particles Using the Gas Injection Method. Composite particles of CdS and ZnS having mixed crystal or core-shell structures using the gas injection method were prepared. The solid lines in Figure 8a show the absorption spectra for coprecipitated CdS and ZnS particles (Cd1-xZnxS) prepared by the gas injection method for various xf values (i.e. molar fraction of zinc for all the metallic ions in the feed solution) at time t ) 1 h. Only one threshold in the absorption spectrum between that of CdS (xf ) 0) and that of ZnS (xf ) 1) is shown by the coprecipitated

98

Ind. Eng. Chem. Res., Vol. 36, No. 1, 1997

Figure 8. Absorption spectra for composite particles of CdS and ZnS: (a) effect of feed composition (xf) of the coprecipitated particles; (b) effect of composite structure of the particles.

particles. The wavelength of the threshold also decreases with increasing xf and agrees with the results of the solution-mixing method as shown by the dotted lines, thus indicating the formation of mixed crystal particles of CdS and ZnS (Hirai et al., 1994b; Cizeron and Pileni, 1995). The wavelengths of the absorption thresholds for the gas injection method are, however, longer than those for the solution-mixing method, indicating that the coprecipitated particles prepared by the gas injection method are larger than those obtained by the solution-mixing method as observed in the case of single-component particles. Figure 8b shows the absorption spectra of ZnS-coated CdS ((ZnS)0.5(CdS)0.5) and CdS-coated ZnS ((CdS)0.5(ZnS)0.5) particles, as prepared by the two-step injection of both metallic ions and H2S and measured at time t ) 1 h after the second H2S injection. The absorption spectra of the single-component CdS and ZnS particles and the coprecipitated Cd0.5Zn0.5S particles at t ) 1 h are also shown for comparison. The absorption spectrum of (ZnS)0.5(CdS)0.5 particles is significantly different from those of (CdS)0.5(ZnS)0.5 and Cd0.5Zn0.5S particles. An increase in absorbance is observed at wavelengths less than 360 nm compared with the spectrum of CdS particles and can be attributed to the formation of a ZnS shell (Hirai et al., 1994b). On the other hand, (CdS)0.5(ZnS)0.5 particles exhibit two obvious absorption thresholds close to those of CdS and ZnS, indicating the existence of two parts with different compositions. These spectrum features suggest the formation of core-shell structure particles in both cases. In the previous case of an alternate solution injection method, addition of an excess quantity of Na2S was necessary to prepare core-shell structure particles (Hirai at al., 1994b). In the present case, the spectrum of the prepared particles is practically unchanged by an excess

Figure 9. Photocatalytic activity of composite CdS and ZnS particles: (a) effect of the feed composition for the coprecipitated particles; (b) comparison of ZnS-coated CdS particles ((ZnS)0.5(CdS)0.5) with a mixture of single-component CdS and ZnS particles (CdS + ZnS).

amount of H2S. Thus core-shell structure particles can be prepared by the gas injection method in a much more convenient procedure than by the alternate solution injection method. This is because particle growth is the principal mechanism of the gas injection method while coagulation is that of the solution-mixing method. Photocatalytic Activity of Composite Particles of CdS and ZnS. The photocatalytic activity of composite CdS and ZnS particles prepared by the gas injection method was investigated, according to the same procedure employed previously for the solutionmixing method (Hirai et al., 1994b). To avoid photocorrosion of the particles, it is necessary to add a hole scavenger to the solution before light irradiation. Since a large excess amount of Na2S was used as scavenger in the previous study, a 10-fold excess quantity of H2S was therefore injected at the preparation stage of the particles for the present study. The quantity of H2 (nH2) generated by a 20 h irradiation of the coprecipitated particles is shown in Figure 9a. For wavelengths longer than 320 nm, H2 production increased linearly with increasing xf as shown by the open circles. This is the similar to the result obtained previously for the solutionmixing method and is explained by the increase in the reducing ability of the photoexcited electron as a function of increasing zinc content (Hirai et al., 1994b). When the cutoff wavelength was changed to 340 nm, H2 production decreased greatly in the range of xf from 0.75 to 1 as shown by the closed circles. Since the wavelength of the absorption threshold decreases with increasing xf as shown in Figure 8a, the photoexcitation of electron by wavelengths longer than 340 nm becomes more difficult with increasing xf. ZnS (xf ) 1) particles

Ind. Eng. Chem. Res., Vol. 36, No. 1, 1997 99

especially cannot absorb 340 nm light. This difficulty of photoexcitation in this xf range causes the decrease in the photocatalytic activity. The photocatalytic activity of ZnS-coated CdS particles was also investigated. Figure 9b shows the amount of H2 generated by (ZnS)0.5(CdS)0.5 particles with a cutoff wavelength of 340 nm for various irradiation times (closed key) and that for a mixture of separately prepared single-component CdS and ZnS particles (open key). The coated particles show a higher H2 photogeneration activity compared to the mixture of single-component particles, but the activity is smaller than that for the coprecipitated particles shown in Figure 9a. This is in line with the previous result for the solution-mixing method (Hirai et al., 1994b). Conclusion The mechanism of formation of CdS and ZnS ultrafine particles in reverse micelles by injection of H2S was studied. The particle formation process was followed by UV-visible absorption spectra, and the following results were obtained: 1. When H2S gas is injected for 2 s, the conversion of Cd2+ ion and H2S to the particles due to nucleation and particle growth continues for about a further 5 s, which is much slower than that for the solution-mixing method using H2S (0.02 s). The rate-determining step is due to the transfer of H2S from the gas phase to the liquid phase (gas absorption). The principal step in the particle formation is particle growth, and this continues for a longer time than that for the solution-mixing method. 2. A kinetic model is proposed for the complete process in particle formation for the gas injection method, based on the average number of reactant molecules per micelle. The rates for the various steps in the process were estimated from an analysis of the absorption spectra. The particle coagulation rates for both the gas injection method and the solution-mixing method using H2S follow the same kinetics as reported for the solution-mixing method using Na2S. Values of the rate constants for these steps were also estimated. A simulation model based on these results explains successfully observed time-course variations in the diameter and in the number of particles during particle formation. 3. The preparation of composite particles was carried out. Mixed crystal or core-shell structured particles of CdS and ZnS can be prepared by the simultaneous precipitation of both materials or by the growth of the shell material on the core particles using the two-step injection of metallic ions and H2S. The generation of H2 by the cleavage of water due to the photoirradiation of the particles was also investigated. Here the activity of composite particles was found to be greater than that for the mixture of the single-component particles and depends on the composite structure of the particles as observed previously in studies of the solution-mixing method. Acknowledgment The authors are grateful to the Department of Chemical Engineering, Osaka University for the scientific support from the “Gas-Hydrate Analyzing System (GHAS)” constructed by a supplementary budget in 1995 and to financial support from a Grant-in-Aid for Cooperative Research (No. 07305034) for I.K. from the

Ministry of Education, Science, Sports and Culture, Japan. H.S. gratefully acknowledges financial support from the Research Fellowships of the Japan Society for the Promotion of Science for Young Scientists. Nomenclature CCd ) molar concentration of cadmium ion in the reverse micellar solution, M (mol/L) Cf ) [Cd2+] + [Zn2+] ) molar concentration of all metallic ions in the feed solution, M Cm ) molar concentration of the reverse micelles, M Cmc ) molar concentration of the reverse micelles containing two or more particles, M Cmg ) molar concentration of the reverse micelles containing cadmium ion, hydrogen sulfide, and particles, M Cmn ) molar concentration of the reverse micelles containing cadmium ion and hydrogen sulfide, M Cp ) molar concentration of the formed particles, M CS,t ) total molar concentration of hydrogen sulfide in the reverse micellar solution, M CS ) molar concentration of hydrogen sulfide dissolved in the water cores in the reverse micellar solution, M dp ) average diameter of the particles, nm kex ) second-order rate constant for the exchange process of reverse micelles, M-1 s-1 kmc ) first-order rate constant for coagulation, s-1 kmg ) first-order rate constant for particle growth, s-1 kmn ) first-order rate constant for nucleation, s-1 kdiss ) first-order rate constant for dissolution of hydrogen sulfide into the liquid phase, s-1 nH2 ) production of hydrogen, mol Np ) number of formed particles in the solution, L-1 rinj ) rate of injection of hydrogen sulfide, M/s rc ) rate of coagulation, M/s rdiss ) rate of dissolution of hydrogen sulfide into the liquid phase, M/s rg ) rate of particle growth, M/s rn ) rate for nucleation, M/s t ) time from the start of the injection of hydrogen sulfide, s W0 ) [H2O]/[AOT] ) water to surfactant molar ratio (water content) x ) molar fraction of zinc in all the metallic ions in the particles xf ) molar fraction of zinc in all the metallic ions in the feed solution Xp ) conversion of precursor ions to the particles [ ] ) molar concentration of species in the brackets, M

Literature Cited Atik, S. S.; Thomas, J. K. Transport of Photoproduced Ions in Water in Oil Microemulsions: Movement of Ions from One Water Pool to Another. J. Am. Chem. Soc. 1981, 103 (12), 35433550. Brus, L. E. Electron-electron and electron-hole interactions in small semiconductor crystallite: The size dependence of the lowest excited electronic state. J. Chem. Phys. 1984, 80 (9), 4403-4409. Cizeron, J.; Pileni, M. P. Solid Solution of CdyZn1-yS Nanosize Particles Made in Reverse Micelles. J. Phys. Chem. 1995, 99 (48), 17410-17416. Hirai, T.; Sato, H.; Komasawa, I. Mechanism of Formation of Titanium Dioxide Ultrafine Particles in Reverse Micelles by Hydrolysis of Titanium Tetrabutoxide. Ind. Eng. Chem. Res. 1993, 32 (12), 3014-3019. Hirai, T.; Sato, H.; Komasawa, I. Mechanism of Formation of CdS and ZnS Ultrafine Particles in Reverse Micelles. Ind. Eng. Chem. Res. 1994a, 33 (12), 3262-3266. Hirai, T.; Shiojiri, S.; Komasawa, I. Preparation of Metal Sulfide Composite Ultrafine Particles in Reverse Micellar Systems and Their Photocatalytic Property. J. Chem. Eng. Jpn. 1994b, 27 (5), 589-596.

100 Ind. Eng. Chem. Res., Vol. 36, No. 1, 1997 Hirai, T.; Tsubaki, Y.; Sato, H.; Komasawa, I. Mechanism of Formation of Lead Sulfide Ultrafine Particles in Reverse Micellar Systems. J. Chem. Eng. Jpn. 1995, 28 (4), 468-473. Kandori, K.; Kon-no, K.; Kitahara, A. Formation of Ionic Water/ oil Microemulsions and Their Application in the Preparation of Calcium Carbonate Particles. J. Colloid Interface Sci. 1988, 122 (1), 78-82. Meyer, M.; Wallberg, C.; Kurihara, K.; Fendler, J. H. Photosensitized Charge Separation and Hydrogen Production in Reversed Micelle Entrapped Platinized Cadmium Sulphide. J. Chem. Soc., Chem. Commun. 1984, 90, 90-91. Ravet, I.; Nagy, J. B.; Derouane, E. G. On the Mechanism of Formation of Colloidal Monodisperse Metal Boride Particles from Reversed Micelles Composed of CTAB-1-Hexanol-Water. Stud. Surf. Sci. Catal. 1987, 31, 505-516. Sato, H.; Hirai, T.; Komasawa, I. Mechanism of Formation of Composite CdS-ZnS Ultrafine Particles in Reverse Micelles. Ind. Eng. Chem. Res. 1995, 34 (7), 2493-2498.

Tanori, J.; Duxin, N.; Petit, C.; Lisiecki, I.; Veillet, P.; Pileni, M. P. Synthesis of Nanosize Metallic and Alloyed Particles in Ordered Phases. Colloid Polym. Sci. 1995, 273, 886-892. Towey, T. F.; Khan-Lodhi, A.; Robinson, B. H. Kinetics and Mechanism of Formation of Quantum-sized Cadmium Sulphide Particles in Water-Aerosol-OT-Oil Microemulsions. J. Chem. Soc., Faraday Trans. 1990, 86 (22), 3757-3762.

Received for review March 11, 1996 Revised manuscript received August 29, 1996 Accepted November 5, 1996X IE9601413

X Abstract published in Advance ACS Abstracts, December 15, 1996.