5854
J. Phys. Chem. 1989, 93, 5854-5859
TABLE 11: Hz Evolution Rates in the Photocatalytic Cleavage of Waterz0
solvent .4 A
A A A
B B B B B
activity per gram,” mol X
Pt amount, mol x IO7
H2 formation
rate. mL/h
I 07
2.15 3.00 4.30 6.45 8.60 1.60 3.20 4.00 4.86 6.48
4.2 4.6 4.0 4.1 2.7 2.3 4.5 4.1 3.5 3. I
1.95 1.53 0.93 0.64 0.31 1.44 1.41 1.03 0.72 0.48
“The activity per gram is related to the activity per atom through a constant. terms of activity per Pt atom is shown in Table 11, for the two particle sizes they used; sol A, 16 A; sol B, 1000 A. W e notice that the activity per atom drops with the increase in the total quantity of Pt used, for both particle sizes. This behavior can be understood as indicating first that Pt is aggregated (a conclusion reached by K M themselves) and that the degree of aggregation increases as more Pt is added. Consequently, in the larger aggregates, a larger percentage of the Pt atoms is buried inside the
aggregate and becomes unavailable for reaction, thus lowering the activity per Pt atom. Because of this aggregation, one cannot calculate the total number of active surface atoms based only on the radius of the elementary crystallite. We conclude therefore that the title question in KM’s paper, “Is There a Particle-Size Dependence, etc.”, is yet to be answered. A Concluding Remark. We finally mention that the rutile data in Anpo’s study16 also obeys eq 1, but the data points are scattered (correlation coefficients 0.9 1-0.93). This brings us to a concluding remark: Obviously, it is not that all catalytic processes obey eq 1; this area of chemistry is too vast to be described by a single relation. But so many catalytic and noncatalytic reactions do obey eq 1 2 s 3 that it seemed to us worthwhile drawing attention to that simple correlation, especially in view of the ability of eq 1 to condensate experimental observations, to provide means of comparison between different heterogeneous reactions, to provide a clue as to the distribution and location of active sites, and to allow some prediction as to whether high dispersity or low dispersity should be employed in order to increase the reaction efficiency ( m < 1, DR > D, for the former and m > 1, DR < D,for the latter; see ref 2d for a detailed discussion).
Acknowledgment. This work was supported by the US-Israel Binational Foundation and by the Belfer Foundation. D.F. thanks the Aronberg Foundation and the Pikovski Foundation for grants. Registry No. Water, 7732-18-5.
Luminescence Probe Study of the Conditions Affecting Colloidal Semiconductor Growth in Reverse Micelles and Water-in-Oil Microemulsions Sotirios Modes and Panagiotis Lianos* School of Engineering, University of Patras, 26000 Patras, Greece (Received: June 14, 1988: In Final Form: January 18, 1989)
A series of reverse AOT micelles and w/o microemulsions have been studied by analyzing the luminescence decay of ruthenium tris(2,2’-bipyridine) in the presence of quencher. The analysis was based both on the established model for luminescence decay in micelles and on the recently developed fractal model of microemulsions. Colloidal cadmium sulfide has then been produced in the microemulsions and the conditions for the particle size growth and size polydispersity have been related with the data of the luminescence decay analysis
Introduction
Microemulsions and, in particular, water-in-oil microemulsions as well as reverse micelles are ideal systems for the controlled production and growth of small particles, either organic polymers or inorganic crystallites. The isolation of water-soluble reactants in different water pools limits the extent of interaction to a degree that the growth of the resulting products is highly controlled and it is dependent on the nature of the employed microemulsions. and a few on Several works on microemulsion p~lymerizationl-~ colloidal semiconductor formation in micro emulsion^^^ have been published. However, the details of the reactant distribution and transfer as well as other factors affecting interaction and growth are not well understood. The present work aims to shed some light on this problem. We have thus studied some reverse micelles and water-in-oil microemulsions with luminescence probes by employing both the older and the fractal modelsIw’* of ( I ) Atik, S. S . ; Thomas, J. K. J . Am. Chem. Soc. 1982, 104, 5868. (2) Lianos, P. J . Phys. Chem. 1982, 86, 1935. (3) Candau, F. J . Chim. Phys. 1987, 84, 1095 and references therein. (4) Meyer, M.; Wallberg, C . : Kurihara, K.: Fendler, J. H . J . Chem. Sac., Chem. Commun. 1984, 90. ( 5 ) Lianos, P.; Thomas, J . K . Chem. Phys. Lett. 1986, 125, 299. (6) Lianos, P.;Thomas, J . K . J . Colloid Interface Sci. 1987, 117, 505.
0022-3654/89/2093-5854$01.50/0
luminescence decay in the presence of quenchers recently developed by us. In order to test the influence of the nature of microemulsions on the extent of particle growth we have examined the growth of small CdS crystallites inside the droplets since they possess the advantage of a direct relationship between particle size and its absorption and emission characteri~tics.l~-’~ Materials and Methods
UV-visible spectra have been recorded with a Perkin-Elmer Lambda - 15 spectrophotometer. Luminescence decay profiles have been obtained with the photon-counting technique using a home-assembled apparatus, and they have been analyzed with (7) Tachiya, M. Chem. Phys. Lerf. 1975, 33, 289. (8) Dederen, J. C.: Van Auweraer, M.; De Schryver, F. C. J . Phys. Chem. 1981, 85, 1198. (9) Lianos, P.; Zana, R.; Lang, J.; Cazabat, A. M. In Surfactants in Solution; Mittal, K . L., Bothorel, P., Eds..: Plenum Press: New York, 1986: Vol. 6, p 1365. ( I O ) Lianos, P.; Modes, S. J . Phys. Chem. 1987, 91, 6088. ( I I ) Lianos, P. Prog. Colloid Polym. Sci. 1988, 76, 140. (12) (a) Lianos, P. J . Chem. Phys. 1988, 89, 5237. (b) Duportail, G.; Lianos, P. Chem. Phys. Lett. 1988, 149, 73. (13) Brus, L. E. J . Chem. Phys. 1984, 80, 4403. (14) Henglein, A . J . Chim.Phys. 1987, 84, 1043 and references therein.
0 1989 American Chemical Society
Semiconductor Growth in Micelles and Microemulsions TABLE I: Composition of the Microemulsions (weight percent) symbol heptane toluene water AOTO. 5,5 65.8 5.9 11.3 AOT0.5,lO 60.9 21.8 AOT0.5,20 51.3 AOT0.5,32 40.9 33.3 50.9 5.3 SDS0.5,5 48.3 10.1 SDS0.5,10 30.7 31.2 SDS0.5,32 SDSO.07, I O 73.8 1.6 72.7 3.0 SDS0.07,20 71.4 4.8 SDS0.07,32
The Journal of Physical Chemistry, Vol. 93, No.15, 1989 5855
PeOH
AOT
SDS
28.3 27.8 26.9 25.8
[SI”
9
0.5
5 10
0.5
26.7 25.4 22.5 22.1 21.9 21.4
17.1 16.2 15.6 2.5 2.4 2.4
20 32 5
0.5 0.5 0.5 0.5 0.5 0.07 0.07 0.07
IO 32 IO 20 32
[SI is the surfactant concentration in M, and w is the water-over-surfactant molecular ratio. least-squares fits. The quality of the fit was judged by the distribution of the residuals and the autocorrelation function of the residuals.15 Prior to luminescence decay measurements, all samples were deoxygenated by bubbling with N2. Before passing through the sample, N 2 was forced through another identical solution in order to be saturated with the volatile components. In this manner, no evaporation of the solvent occurred during deoxygenation. All products were of the best quality commercially available and they have been used as received: sodium sulfide nonahydrate (Na2S-9H20)and dodecane (Janssen Chimica); sodium dodecyl sulfate (SDS, BDH); bis(2-ethylhexyl) sulfosuccinate sodium salt (AOT), butanol, pentanol, hexanol, and toluene (Fluka); heptane (Ferak Laborat, Berlin); tris(2,2’-bipyridine)ruthenium dichloride hexahydrate ( R ~ ( b p y ) , ~ + G,F S Chemicals); benzyl alcohol and potassium hexacyanoferrate(II1) (Fe(CN)63-, Merck); and cadmium perchlorate hexahydrate (Alfa). The following AOT reverse micelles and water-in-oil (w/o) microemulsions have been studied and subsequently utilized to produce colloidal cadmium sulfide particles. Their compositions are tabulated in Table I and a symbol SURF[S], w has been assigned to each one of them, where S U R F is the name of the surfactant and [SI and w are the surfactant concentration and the water over surfactant molecular ratio. All microemulsions have been prepared by first mixing all components except alcohol and finally adding alcohol to a quantity just above the necessary for transparent solution. The luminescent probe R ~ ( b p y ) , ~and + the quencher Fe(CN)6,- were solubilized by the following procedure. Measured aliquots of concentrated aqueous solutions were introduced into 5-mL flasks. Then the flasks were submerged into a warm bath and the solvent (water) was evaporated by blowing with N,. Finally the microemulsion was added and magnetically stirred until all solid residue was clearly solubilized. The Ru(bpy)t+ concentration was 4 X lV5M while the quencher and 4 X M. concentration varied between 5 X Cadmium sulfide was produced by direct mixing of cadmium perchlorate with Na2S inside the microemulsion droplets. Before reaction the samples were deoxygenated by bubbling with N, in a specially made flask covered with rubber septum. This flask consisted of two communicating compartments for the separation of the reactants before mixing. The final concentration of the M. reactants inside the mixture was in all cases 2 X The reactions were done at room temperature which in our laboratory was around 25 OC. However, the luminescence measurements were performed at controlled 25 “ C . The experimental work was carried out in two stages. First, the microemulsion droplets were characterized by luminescence probing, Le., by the analysis of the luminescence decay profiles of the lumophore in the presence of quencher at various concentrations. Second, the semiconductor particles were produced and characterized by absorption spectrophotometry. Results and Discussion
I . Luminescence Probe Study of Reverse AOT Micelles and w/o SDS Microemulsions. Luminescence probing is a well-established method for studying organized assemblies and it is the (15) Grinvald, A.; Steinberg, I . Z. Anal. Biochem. 1974, 59, 583.
most powerful indirect procedure for determining the structure of micelles and microemulsion droplets. In most of the cases, the method consists of the analysis of luminescence decay profiles in the presence of quencher according to the I = A I exp(-A,t A, = ko
+ kJQIkq/A,,
+ A3[exp(-A4t) - 11)
A3 = ( k q / 4 2 [ Q 1 / [ M 1 3
(1)
A4 =
kq + ke[Ml (2) and since in most of the cases kq >> ke[M]
where ko is the luminescence decay rate constant in the absence of quenchers, ke is a second-order rate constant for the exchange of quenchers between droplets, [Q] is the quencher and [MI the micellar (droplet) concentration, and kq is an average first-order-like quenching rate constant. The above equation requires that the quenchers stay resident, within the droplets or at the interface, and they do not exchange between the dispersed and the continuous phase. In the opposite case A2, A3, and A4 correspond to more complicated expressi~n.l~.~’ We have actually analyzed our samples by using eq 1 and 3. The analysis gave the following information. (1) The measured values of A2 allowed us to judge whether exchange of quenchers between droplets exists. An exchange rate constant k, could be measured only in the cases where A, changed appreciably with changing [Q]. (2) A, gave the micellar (droplet) concentration and subsequently the surfactant aggregation number N and the water “aggregation” number N, from the relations N = [ S ] / [ M ] and N, = wN. These expressions imply that the critical micellization concentration (cmc) is considered negligible. Indeed, cmc in MI8 while the reverse micelles is not expected to exceed surfactant concentrations presently used are very high. (3) A4 gave valuable information on the effectiveness of quenching and any type of interaction within the droplet, which is also associated with both the droplet structure and the nature of the water-oil interface. It should be underlined at this point that the application of the above method in studying w/o microemulsions has encountered problems at relatively high quencher concentrations. Thus, for A, values > 1, the droplet concentration seems to increase at the expense of its size.9 This phenomenon has been attributed to the influence of the quencher molecules on the micellar structure.I9 In this work, where the absolute value of the droplet size is not so important as its variation with component concentration, we have adopted A3 values obtained with eq 1 for relatively low [Q]/[M] ratios, Le., around 0.5. Table I1 shows calculated values of the above parameters or other information derived from them. Some of the tabulated values are also presented in Figures 1 and 2. We have chosen (16) Gelade, E.; De Schryver, F. C. J . Photorhem. 1982, 18, 223. (17) Malliaris, A.; Lang, J.; Sturm, J.; Zana, R. J . Phys. Chem. 1987, 91, 1475. (18) Lang,
J.; Jada, A.; Malliaris, A. J . Phys. Chem. 1988, 92, 1946. (19) Pileni, M.-P.; Zemb, T.; Petit, C. Chem. Phys. L e f f .1985, 118, 414.
5856
The Journal of Physical Chemistry, Vol. 93, No. 15, 1989
Modes and Lianos
TABLE 11: Data Obtained from the Analysis of Luminescence Decay Profiles with Eq 1 water area droplet pool per surf. ke. microconcn, diam, polar head, k X108 s-’ emulsion X10-3 M A A2 x108’s-1 M-1 AOTO. 5,5 AOT0.5, I O AOT0.5,20 AOTO. 5,32 SDS0.5,5 SDSO.5,IO SDS0.5.3 2 SDS0.07,lO SDS0.07,20 SDS0.07,32
2.9 1.7 1.4 0.9 2.6 1.9 0.7 5.0 2.5 1.2
37 55 74 102 39 54 107 20 32 48
24.6 32.0 47.6 56.8 23.5 33.5 53.2 83.7 110.9 118.6
3.4 3.1 3.2
1.o
35.9 35.7 29.6 27.4 23.3 17.4
2.9 12.0 1 .o 2.7
0 Figure 2. Variation of the area A per surfactant head group with water-over-surfactant molecular ratio w : ( I ) S D S 0.5 M; (2) AOT 0.5 M; and (3) S D S 0.07 M.
W I
,
,
,
,
,
0 1 0 2 0 3 0 Figure 1. Variation of the water pool diameter D with water-over-surfactant molecular ratio w : ( I ) SDS 0.5 M; (2) AOT 0.5 M; and (3) S D S 0.07 M.
to study a relatively high and a relatively low SDS concentration and a high AOT concentration. N o results for lower AOT concentrations are given here. CdS formed in AOT reversed micelles at constant w and varying [SI is shown in ref 6. The droplet concentration and the water pool diameters are tabulated in Table 11. The diameters were calculated by using the abovementioned N , values and by considering the density of water equal to 1 g/cm3. This assumption is closer to reality at high w . In fact the nature of the water in w/o microemulsions and reverse micelles changes with the value of W . ~ O Notice that, at [SI = 0.5 M, small differences are observed in the micellar concentration and size between AOT and SDS. At low SDS concentration, the droplets become more numerous and much smaller. So [SDS] = 0.07 M and w = 10 gives much smaller micelles than [SDS] = 0.5 M and w = 5 . This fact is clearly seen in Figure 1 . The calculated water pool sizes must in all cases be very close to reality at high w but a large error is expected at low w. This is due to the fact that at low w most of the water is taken up for the hydration of the surfactant head groups leaving a small percentage for the formation of the water pool. Subsequently our values for AOT are in very good agreement with values found by other methods for high w but deviate by almost 100% at low w.21,22 At any rate, the trend observed in Figure 1 is expected to be true no matter how steep or not the variation of the size is. The extensive variation of the droplet size with decreasing surfactant concentration in SDS microemulsions is obviously due to the variation of the structure of the interface. The global amount (20) (a) Rosenholm, J. B. Ber. Bunsen-Ges. Phys. Chem. 1987, 91, 106. (b) Kotlarchyk, M.; Huang, J. S.: Chen, S.-H. J . Phys. Chem. 1985,89,4382. (21) Maitra, A . J . Phys. Chem. 1984, 88, 5122. (22) Fletcher, P. D. 1.; Robinson, B. H. Ber. Bunsen-Ges. Phys. Chem. 1981, 85, 863.
of alcohol present in solution does not vary much, but apparently the amount of alcohol present at the interface varies with surfactant concentration. Another quantity of interest tabulated in Table I1 is the area A corresponding to each surfactant head group. Figure 2 shows that this quantity does not differ much from AOT to SDS at high surfactant concentrations, but it is much larger at low SDS concentration. Obviously, in the latter case the structures are more “open” and the empty space between surfactant heads is filled up with alcohol. We then conclude that at low surfactant concentration the interface is rich in alcohol. As w increases, the area per surfactant head group tends to achieve a constant value. This is expected, since the variation of this quantity slows down or completely ceases when all polar heads have been hydrated. Again, the values obtained for low w are most probably larger than the real ones since the amount of water used up for hydration is not taken into account. Nevertheless, the above conclusions are expected to be qualitatively right. The quenching rate constant k, varied with the size of the water pool, becoming smaller in larger droplets. This is expected since quenching is more efficient when the quenchers are forced closer to the lumophore in smaller droplets. Nevertheless, k , was characteristically lower in AOT reverse micelles than in SDS w/o microemulsions. Low k, values in AOT reverse micelles have been observed p r e v i o ~ s l yand , ~ ~they ~ ~ ~are probably due to the particular localization and mobility of the lumophore and the quenchers in each case of surfactant. Apparently for the same reason the k, values are generally smaller when [SI = 0.07 M than when [SI = 0.5 M contrary to the observed micellar sizes. Probably, the higher surfactant population at the interface in the case of high SDS concentration forces the reactants inside the water pool and away from the interface, facilitating reaction. The exchange rate constant was very small,I8 therefore undetectable, in all AOT micelles and in SDS microemulsions containing the smallest quantity of water. k, increased with the water content. It was the smallest when the solution contained 3.0% water and the highest when it contained 31.2%. Even though alcohols are considered responsible for increase in the exchange rates, we notice that the presence of increasing amount of water masks whatever the influence of alcohol on k, is. Keeping, however, in mind that k , must be measured by varying [Q] and that quenchers might influence micellar structure and give erroneous results on the exact k, values, we have proceeded by taking a better look at the phenomenon of solubilizate exchange from a different point of view. We have analyzed our data with decay models developed by us recently, where microemulsionsare treated (23) Atik, S. S.; Thomas, J. K. J . Am. Chem. SOC.1981, 103, 3543. (24) Atik, S. S.: Thomas, J. K. Chem. Phys. Lett. 1981, 79, 351.
Semiconductor Growth in Micelles and Microemulsions
The Journal of Physical Chemistry, Vol. 93, No. 15, 1989 5857
TABLE 111: Data Obtained from the Analysis of Luminescence Decav Profiles with Ea 4 Fractals k , , X108 k 2 , O X 10’ microemulsion M-I s-! M-I s-f fractal power f SDS0.5,5 SDSO.5,lO SDS0.5,32 2.4 4.4 0.19 0.4 0.14 SDSO.07, I O 1.1 0.13 SDS0.07,20 SDS0.07,32 2.5 0.14 “fpower introduced to take care of the units of time.
Figure 4. Absorption spectra of cadmium sulfide produced in AOT0.5,5 (-); SDS0.5,5 (---); and SDS0.07,lO (..e).
6.0
I
I
2bo
&
I
6b3
1
Sbo ( n s ~’I
Figure 3. Luminescence decay profile of R ~ ( b p y ) , ~in+ the presence of IO-’ M Fe(CN)z- in SDS0.07,IO. Fitted curve according to eq 4. Insert: residuals (R) and autocorrelation function of the residuals (A).
as fractal objects and the motion of the solubilizates as a random walk among solubilization sites.’*12 In fact the analysis of microemulsions with fractals circumvents the problem of the influence of the probe and the quenchers on the micellar structure and concentration and offers structural information from a different point of view. We have thus fitted the data with the following equation1*12
-(ko
+ k l [ Q ] ) t- f
where k , and k 2 are constant and d , is the so-called spectral dimension, showing how the density of statesz5varies on fractal objects. As has been previously discussed,12 for reactions in organized assemblies, the term k l [ Q l t is expected to be nonzero only when the reaction space is large and behaves as pseudocontinuous. This is true for very large aggregates or very large exchange rates k,. The term ( k z [ Q ]/fll/corresponds to relatively small (compact) reaction spaces. The simultaneous presence of all terms of eq 4 is indicative of a polydisperse system. It should be underlined at this point that the application of eq 4 is made under the assumption that the reaction space is not necessarily limited within one droplet but it may extend over a hypothetical cluster (a domain of fractal dimensions) which contains several aggregates (droplets). The quenchers may move within the cluster by exchanging between droplets. Equation 4 is the same as that used for the so-called “target” problem26or the “supertrapping” process2’ in solid-state physics, with the addition of k , [ Q ] appropriate for organized assemblies.Iz The model of eq 4 fitted all the above samples except AOT and SDS0.5,5 and SDSO.5,lO. The results obtained by the analysis are shown in Table 111. An example of the fitting is shown in Figure 3. Notice that only in SDS0.5,32, k , was not zero. This is in accordance with the adove theory. When’the amount of water (25) Alexander, S.;Orbach, R. J . Phys. Lett. 1982, 43, L-625. (26) Klafter, J.; Blumen, A.; Zumofen, G . ;Drake, J. M. J . Lumin. 1987, 38, 113. (27) Evesque, P. J . Phys. 1983, 44, 1217.
is large, k, is large, large aggregates are formed, and polydispersity is favored. k 2 increased with the water content and it was the largest in SDS0.5,32. The analogy to the above-calculated k , values is not a coincidence. When quenchers exchange easily between droplets the interaction within the fractal domain is more efficient and kz increases. The fractal power f was relatively small. We have found such low values in cases where kq is relatively large. Large k, values means that the droplet is structured in such a manner that the quenching is largely facilitated. In fractal modeling this means that the reaction space is structured in such a manner that the exploration space of the random walker is very compact, Le., the number S ( t ) of sites visited within a time t is much smaller than the number of the random steps.28 Thus S ( t ) t d S l 2 , where d, is a relatively small number. The existence of such low values forf(and ds),much lower than the value of the percolation threshold (d, = 4/3),253’ is a matter of some concern. For this reason we have undertaken a separate where the luminescence quenching data were reproduced by Monte Carlo simulation on clusters formed on a 300 X 300 square lattice. We have indeed found that for cluster sizes equal to 25 sites the value o f f drops down to 0.21 and it tends to even smaller values at smaller cluster sizes. This verifies that interaction in very small clusters leads to lowfvalues. However, the values offshown in Table 111 do not vary with increasing water content and micellar (droplet) size, contrary to our computer simulation29and other previous results.30 This is another matter of concern and it is studied further in our laboratory. Nevertheless, it seems that for relatively small droplets as those formed with 0.07 M SDS, which do not vary much with increasing w (cf. values of Figure l ) , the fvalue is not affected by w. Furthermore, as we have already said, the fractal domain extends beyond a single droplet and, apparently, its spectral dimension for 0.07 M SDS is not affected by changing w. Finally, eq 4 did not apply to AOT or to SDS0.5,5 and SDSO.5,lO. We have in fact found”,12that the application of eq 4 is related with the existence of exchange of solubilizates between microemulsion droplets. In this respect, SDS0.07,lO and SDSO.5,lO constitute an exception. Application of eq 4 did not follow the rule here. Apparently these are typical cases falling into the limits of uncertainty. The alternative study of such systems with the help of both eq 1 and 4 facilitates better understanding of their actual behavior. Our fractal analysis offers a scale of efficiency of interaction between the reacting species in the fractal domain, which for the SDS microemulsions studied increases in the following order: SDS0.5,5 and SDSO.5,lO; SDS0.07,lO; SDS0.07,20; SDS0.07,32; and SDS0.5,32. 2. Formation of CdS. It has been established that for relatively small semiconductor crystallites ( D < 100 A),31the absorption spectrum is a direct measure of the size of the crystallite^.^^*^^^^^-^^
-
(28) (29) (30) (31)
Rammal, R.; Toulouse, G. J . Phys. Lett. 1983, 44, L-13. Lianos, P.; Argyrakis, P. Phys. Reu. A 1989, 39, 4170. Unpublished results. Rossetti, R.; Ellison, J. L.; Gibson, J. M.; Brus, L. E. J . Chem. Phys. 1984, 80, 4464. (32) Brus, L. E. J . Chem. Phys. 1983, 79, 5566.
5858
The Journal of Physical Chemistry, Vol. 93, No. 15, 1989
00 300 3% 403 450 500 5% Figure 5. Absorption spectra of cadmium sulfide produced in: AOT0.5.32 (-): SDS0.5,32 (---); and SDS0.07,32 (...) .
We have thus recorded some absorption spectra of CdS formed in the above water pools and we have adopted that when the absorption shifts to lower wavelengths the particles decrease in size. Molecular CdS is not soluble even in the above water pools and it readily forms colloidal particles in suspension inside the water pools. These particles have semiconductor properties and they absorb photons with wavelengths smaller than about 530 nm. Figures 4 and 5 show absorption spectra of colloidal CdS formed in the above water pools. These spectra possess a number of distinguishable features. (1) The first is an absorption onset which varies much from one microemulsion to the other. Figures 4 and 5 show that the absorption onset is the smallest in the case of AOT micelles with the lowest water content and the largest in the case of S D S microemulsions with the highest water content. The absorption onset is an index of the particle size since larger particles have smaller energy gaps and they start absorbing at larger wavelength^.^^ (2) A second feature is the position and the height of the main shoulder in the absorption spectrum which is characteristic again of the size of the semiconductor particles. Thus larger particles absorb mainly at larger wavelengths. Figures 4 and 5 show that the smallest particles are produced in AOT0.5,5 and SDS0.5,5 and the largest in all samples where w = 32. Other cases such as S D S 0.07,lO can be characterized as intermediate. (3) A third feature is the difference in nanometers between the position of the onset and the mean position of the shoulder. Since both onset and shoulder are related with particle size, the difference between their positions characterizes the size polydispersity. Thus Figures 4 and 5 reveal that the most polydisperse particles are obtained in droplets where w = 32. Figure 4 shows that the lowest polydispersity is expected in AOT0.5,5 and second lowest in SDS0.5,5. Among the three types shown in Figure 5 , AOT0.5,32 produces particles with the lowest polydispersity. (4) At wavelengths approaching 300 nm all spectra show a large increase in absorption which is characteristic of all cases examined without marked differences. It must be underlined at this point that repeated experiments under the same conditions do not exactly reproduce the spectra, as it happens with molecular solutions. Nevertheless, the above features and differences are reproducible with certainty, no matter whether small variations in the amount of absorption might be observed. 3. Comparison of the Results Obtained with Luminescence Probing and Colloidal Semiconductor Formation. This work allows for the discussion of two aspects of the colloidal particles formed in the above water pools, Le., the particle size and its polydispersity. We have found that the largest particles are formed in droplets with the highest water content. Figure 5 shows that the particles formed in reverse AOT micelles and S D S w/o microemulsions with w = 32 are of about the same average size independent of the system used. However, in the three systems giving the absorption curves of Figure 5, there is a multitude of factors which differ much from one system to the other. For example, the droplet size changes much from SDS0.5,32 to ( 3 3 ) Fojtik, A.; Weller, H.; Koch, U.; Henglein, A. Ber. Bumen-Ges. Phys. Chem. 1984, 88. 969.
Modes and Lianos SDS0.07,32. The structure of the interface is different from AOT to S D S and from SDS0.5,32 to SDS0.07,32. Thus, in the last case it is richer in alcohol. The exchange rate k,, the intradroplet quenching rate kq, and the interaction rate k2 in the fractal domain differ much from one system to the other. The only common characteristic which can be associated with the same average particle size is the value w = 32 itself. This amount of water is large enough to overpass the quantity necessary for the hydration of the polar head groups and to form a water pool of substantial size. In this water pool the semiconductor particles grow in relatively large sizes independent of the particular structural characteristics of the water pool itself. Even though the average particle sizes are the same in the three systems with w = 32, the same situation does not hold with size polydispersity. S D S microemulsions form more size-polydisperse particles than AOT micelles. Association of this fact with the luminescence data suggests that exchange of reactants between water pools should be the reason of particle size polydispersity. We believe that Ostwald ripening is responsible for the extensive growth of some particles. Ostwald ripening proceeds by interaction of an already formed colloidal particle with isolated ions. This process is obviously favored when the ion exchange between droplets is relatively large. However, we should not think of a linear relationship between exchange rate and size polydispersity. This relationship should rather be very complicated since other factors might locally influence particle growth, such as local reactant concentration and structure of the water-oil interface. Thus the exchange rate k, and the related reaction rate constant k2 in the fractal domain is larger in SDS0.5,32 than in SDS0.07,32. However, the CdS particles do not grow accordingly (see Figure 5). Furthermore, if in the solution SDS0.07,32 we substitute toluene by dodecane, k, increases by almost an order of magnitude; Le., it is comparable with k , measured in SDS0.5,32. CdS formed in dodecaneSDS0.07,32 grows rapidly into large particles and precipitates out of the solution. This anomalous behavior indicates that not k, alone but k, in association with the nature of the water-oil interface will decide the particle growth and the polydispersity of their sizes. However, when the system is modified progressively, without drastic changes in its constitution, increase of k, clearly results in increase of size polydispersity. Thus if k, is increased in reverse AOT micelles, CdS particles start absorbing at higher wavelengths. We have verified this using AOT0.5,32 containing 0.5 M benzyl alcohol which is known to reduce the rigidity of the water-oil interface in AOT micelles and allow exchange of sol~ b i l i z a t e s .Furthermore, ~~ we have substituted pentanol by butanol in SDS0.5,32 thus increasing k , by a factor of 3. CdS formed in this last system was of the same average size as in all other w = 32 cases, but it started absorbing at 550 nm; Le., it was even more polydisperse than the one obtained in SDS-pentanol. To conclude this paragraph we may say that, when w = 32, CdS particles are formed in relatively large sizes. The average size is the same in both AOT and S D S microemulsions but the distribution of sizes is more narrow there where the exchange of reactants between water pools is negligible. Contrary to the importance of w , k,, and k2 in the present discussion, k , seems to play a rather unimportant role in deciding the nature of the CdS particles formed. Accordingly, the fractal power,f, remained practically constant in all S D S microemulsions studied. Finally, in a pure CdS crystal the energy gap corresponds to 2.42 eV (51 3 nm). The fact that we have recorded smaller energy gaps, apparently, indicates that size-polydisperseparticles possess low-lying energy states, due to structural defects. At low w ratios we have obtained smaller CdS particles (cf. Figures 4 and 5 ) absorbing at smaller wavelengths. The limited growth of CdS in this case is not due to one reason only but rather to the cooperative action of various factors. Thus we cannot say that it is due to low local reactant concentration; otherwise particles would become smaller in SDS0.07,lO where micelles are more numerous and the corresponding number of reactants per micelle is the ~ m a l l e s t (Table ~ . ~ 11). We cannot say that it is solely due to the size of the water pool since again the smallest sizes are observed with SDS0.07.10. A review of our luminescence data
J . Phys. Chem. 1989, 93, 5859-5865
5859
inside the water pools of reverse AOT micelles and SDS w/o reveals that when w is small particle sizes are small but differences microemulsions, and the nature of the employed water pools. Our between them give much importance to the nature of the interface analysis is based on two main aspects of the colloidal particles, in each case. Thus the smallest particles are obtained in AOT their size and polydispersity, both associated with their absorption with the most compact water-oil interface. The compactness of spectra. the interface is higher in AOT which contains no alcohol, which The microemulsions have been studied by analyzing the ludoes not allow exchange of solubilizates between droplets, which minescence decay profiles of Ru(bpy),Z+ in the presence of varying has very small area A per surfactant head group and which, finally, quencher, Fe(CN)63-, concentration. We have used two models does not allow analysis of the data with fractals. Reaction in AOT for the analysis, i.e., eq 1 and 4. Equation 1 gave information is obviously a mainly intradroplet phenomenon and a major reon the size of the water pools, the nature of the w a t e r 4 interface, action space, if it exists, is negligible. The opposite happens with and the solubilizate exchange rate between droplets. Equation SDS0.07,IO which gives the largest of the small particles. Indeed, 4 investigated the conditions under which a microemulsion can in this case we have a particularly flexible interface, with high be treated as fractal object, a condition which is associated with alcohol content and large area per surfactant head group, and it the ability to exchange solubilizates between droplets. Equation does allow reaction in a major (fractal) domain. I n conclusion, 4 offered a scale of reactivity between reactants in the fractal small particles are formed when w is small but the structure of domain. As long as eq 4 was valid, it gave a more realistic picture the interface becomes then very important, affecting size variation of the reaction since it avoids localizing it inside the water pools. or size polydispersity. Some concluding remarks are necessary before closing this Comparison of the absorption spectra of the semiconductors discussion. A very recent pubIicationls presents a systematic study with the luminescencedata clearly distinguished three main factors affecting particle growth. (1) The water over surfactant ratio w: of the structure of reverse AOT micelles. The data presented there are in agreement with our results. Also in our previous publiWhen it is large, Le., w = 32, a substantial water pool is formed, c a t i o n ~we ~ , have ~ shown formation of CdS in reverse AOT miCdS particles become large, and the absorption onset extends celles and w/o microemulsions. We have examined there mainly approaching or even going beyond the E, of CdS crystals. ( 2 ) The extension of the reaction space into a domain which involves the importance of the local reactant concentration inside the water several microemulsions droplets: This is possible when exchange pool. Larger particles are formed there where the concentration of solubilizates between droplets is allowed, and it is responsible of the reacting species is high. However, as long as concentration for size polydispersity. (3) The structure and the composition of does not drastically change, the above-discussed factors are the water-oil interface which becomes very important at low w paramount in defining particle growth. Finally, we must discuss the fact that our conclusions are based on the reactivity of ions values where the formation of a water pool is doubtful: The with a large organic core. Meanwhile, the ions forming CdS are smallest particles are obtained when the interface is very compact. simple atoms. Can one draw conclusions by comparing molecules Nevertheless, large-scale differences between local reactant differing so much in size? The pair R U ( ~ ~ ~ ) , ~ + - F ~ ( Chas N ) , ~ - concentrations or drastic changes imposed by variations in the been systematically tested in several luminescence quenching solvent might also induce large-scale differences in particle sizes. measurements and it gives more reliable results than quenching with atomic ions (e.g., copper). We have made our choice with Acknowledgment. We acknowledge financial aid from the the belief that our conclusions are to a sufficient degree appropriate Greek General Secretariat of Research and Technology. in describing the behavior of both small and large ions. Registry No. AOT, 577-1 1-7; SDS, 151-21-3;CdS, 1306-23-6;RuSummary (bpy)?, 15158-62-0;Fe(CN),'-, 13408-62-3;benzyl alcohol, 100-51-6; In this work we have detected some relationships between the pentyl alcohol, 71-41-0; butyl alcohol, 71-36-3;dodecane, 112-40-3;hexyl size of light-absorbing colloidal semiconductor particles, grown alcohol, 11 1-27-3; toluene, 108-88-3; heptane, 142-82-5.
NMR Studies of Model Hydrodenitrogenation Catalysis: Acetonitrile Hydrogenation on
Grant W. Haddix,+,% Alexis T. Bell,t and Jeffrey A. Reimer*%$ Center for Advanced Materials. Lawrence Berkeley Laboratory, and Department of Chemical Engineering, University of California, Berkeley, California 94720 (Received: July 29, 1988; I n Final Form: March 8, 1989)
Temperature-programmed reduction of CH3CN adsorbed on unsupported, high surface area 7-Mo2N is studied in situ by 'H and I3C NMR spectroscopy. CH3CN adsorbs molecularly at 298 K, with a fraction of the adsorbate in a highly mobile state. Upon reduction in flowing H2, some of the adsorbed CH3CN appears to convert to adsorbed CH3CH,NH,. Increased reduction temperature leads to cleavage of the C-N bond, leaving only NH, fragments on the catalyst surface.
and ortho/para conversion of Hz.6 More recently, it has been Introduction demonstrated that y-MozN is active for hydrodenitrogenation of Molybdenum nitride (7-Mo2N) is known to be an active catalyst for a variety of reactions. These include C O hydrogenation to ( 1 ) Boudart, M.; Oyama, S. T.; LeClerq, L. In Proceedings, 7fh Internahydrocarbon^,'-^ ethane hydrogen~lysis,~ ammonia ~ y n t h e s i s , ~ . ~ 'Lawrence Berkeley Laboratory. *Department of Chemical Engineering. Present address: Shell Development Co., Houston, TX 77001.
0022-3654/89/2093-5859$01.50/0
tional Congress on Catalysis, Tokyo, 1980; Seiyama, T., Tanabe, K., Eds.; Elsevier: Amsterdam, 1981; p 578. (2) Saito, M.; Anderson, R. B. J . Catai. 1980, 63, 438. (3) Ranhotra, G. S.; Bell, A. T.; Reimer, J. A. J . Caral. 1987, 108, 10. (4) Volpe, L.; Boudart, M. J . Phys. Chem. 1986, 90, 4874.
0 1989 American Chemical Society
