567
Langmuir 1990,6, 567-572 photocatalytic decomposition of aqueous sulfide:
Scheme I S2- 2H+ + SH-
+
hu
Rh,O,/Rh/CdS
+ H+
(1)
ecB-(Rh203/Rh/CdS)+ hv~+(Rh203/ Rh/ CdS) (2)
-,
+
-
1/2H, + Rh203/Rh/CdS (3) 2hvB+(Rh20,/Rh/CdS)+ SH- S + Ht + 2Rh,O,/Rh/CdS (4) overall reaction: eci(Rh203/Rh/CdS) H+
-
hu
H2S in alkaline medium:
4
RhpOi-Rh-CdS
H,+S
-szo,2-+ s2-
s + s2- s,2-
(5)
Photoelectrochemical studies show that the photocurrent at a bias of +1 V vs SCE, which is mainly due to the photocorrosion of CdS (CdS + 2hvB+ Cd2+ S)21i22 with Rh/CdS as the photoanode (170 KA) is less than that observed in the case of CdS (245 FA), indicating that the photodeposition of Rh on CdS decreases the photocorrosion of CdS. In conclusion, it can be stated that Rh/CdS is more stable toward photocorrosion and absorbs most of the visible light. Rh, which acts as a sink for the electrons, serves as a good proton reduction center. Hence, Rh/ CdS is a better photocatalyst than CdS. Rh,S,/CdS also functions as a better photocatalyst than CdS. Among the photocatalysts investigated, Rh,S,/Rh/CdS is found to be the most active photocatalyst for the decomposition of aqueous sulfide. The order of activity of the photocatalysts can be given as follows: Rh,S,/Rh/CdS (Rh 1.37 wt % ) > Rh2S3/CdS (Rh 1.37 wt %) > Rh/CdS (Rh 1.37 wt % ) > CdS.
-
+
(6)
in the presence of sulfite ions:
sot- + s22-
(7) In alkaline medium (pH 13), the sulfur formed reacts with the sulfide ions (reactant) to give yellow disulfide ions which hinder the light absorption by CdS. This could be avoided by the addition of sulfite ions (SO,,-) to the reaction mixture, which react with the disulfide ions to give colorless thiosulfate ions (eq 7)., Scheme I holds good in the case of Rh,S,/Rh/CdS also, the only difference being the replacement of Rh203by Rh,S,.
Acknowledgment. We thank the Regional Sophisticated Instrumentation Centre, Indian Institute of Technology, Madras, for providing the facilities for carrying out the XPS studies reported in this paper. Registry No. Rh, 7440-16-6;Rh,S,, 12067-06-0;CdS, 130623-6; Na,S, 1313-82-2;H, 12385-13-6. (21) Rajeshwar, K.; Kaneko, M.; Yamada, A,; Noufi, R. N. J . Phys. Chern. 1985,89,806. (22) Elmorsi, M. A.;Juttner, K. Electrochim. Acta 1986,31,211.
Electron Transfer in Aqueous Colloidal Sn02 Solutions Paul Mulvaney,t Franz Grieser,*.t and Dan Meisel*.$ Department of Physical Chemistry, University of Melbourne, Parkville, Victoria 3052, Australia, and Chemistry Division, Argonne National Laboratory, Argonne, Illinois 60439 Received June 27, 1989. I n Final Form: October 25, 1989 Electron-transfer reactions on transparent colloidal SnO, have been examined by use of radiolytically generated reducing radicals, direct band-gap excitation, and photosensitization using Ru(bpy)32+. Colloidal SnO, was found to be cathodically stable under the reducing conditions examined. Quantitative amounts of H, formed in solution following electron transfer from radiolytically generated isopropyl alcohol radicals. No hydrogen, however, was produced if the sol was illuminated at 300 nm, even in the presence of a hole scavenger. Yet, if the electron acceptor methylviologen was present, the characteristic blue color of the viologen radical formed readily with time of illumination. Charge transfer from excited, adsorbed Ru(bpy),,+ was found to be strongly dependent on pH and electrolyte concentration. This has been scribed to the effect that pH and added electrolyte both control the electrostatic potential on the colloidal SnO,, which in turn influences the adsorption of the ruthenium complex and therefore the extent of quenching.
Tin dioxide is a wide-band-gap (3.5 eV) oxide' which is relatively inert under both cathodic and anodic bias.
The flat band potential is placed very close to the redox level of water., Photoelectrochemical investigations have suggested that reduction of the oxide competes only weakly with hydrogen evolution under a cathodic bias., In order
University of Melbourne. t Argonne National Laboratory. (1) (a) Spence, W.J . Appl. Phys. 1967,38,3767. (b) Jacquemin, J.; Bordure, G. J . Phys. Chem. Solids 1975,36, 1081.
(2) Morrison, S. R. Electrochemistry at Semiconductor and Oridized Metal Electrodes; Plenum Press: New York, 1980. (3) (a) Armstrong, N.;Lin, A.; Fujihira, M.; Kuwana, T. Anal. Chern. 1976,48, 741. (b)Laitinen, H.; Vincent, C.; Bednarski, T. J. Electrochern. SOC.1968,115, 1024.
Introduction
0743-7463/90/2406-0567$02.50/0
0 1990 American Chemical Society
568 Langmuir, Vol. 6 , No. 3, 1990
Mulvaney et al.
to capitalize on this stability, a number of investigations have been made into dye-sensitized electron-transfer reactions to SnO, electrodes, especially by Memming and cow o r k e r ~ .Wrighton ~ et al. have reported water splitting using antimony-doped SIIO,.~ However the redox chemistry of colloidal SnO, and its stability toward cathodic dissolution do not appear to have been investigated to date. In this report, some results are presented on the photoredox behavior of colloidal SnO,.
Experimental Section Colloidal SnO, wm prepared by the hydrolysis of SnCl,. Typically, 3 mL of anhydrous SnC1, (Hopkins and Williams) was added to 1 L of Milli-Q filtered water with stirring. The solution immediately became acidic, and after 24 h the hydrous tin oxide settled out as a fluffy precipitate. The suspension was washed repeatedly with Milli-Q filtered water by decantation until the pH of the supernatant liquid was greater than 5. The suspension was then readily peptized by dilute base or a few drops of ammonia. The resultant sol was dialyzed against 5 L of M NaOH by using an Amicon Hollow Fibre HIP100 cartridge, a t a flow rate of about 600 mL min-'. Very dilute sols (Sn content < 5 X lo-' M) could be prepared without precicpitation by adding the SnCl, to alkaline (NaOH or NH,) solution rather than water, but only if the solution remained alkaline throughout the hydrolysis and the ionic strength remained low. Electron microscopy revealed a particle diameter of 30 f 5 A. From the bulk density of tin dioxide (6.95 g ~ m - and ~) a molecular weight of 150.69, an aggregation number of 390 was obtained. Electron diffraction showed the sol to be amorphous. When the sol was boiled for 5-6 h, the diffraction pattern gradually became less diffuse until it showed the characteristic lines of cassiterite, the rutile form of SnO,. The size of the particles did not increase during heating, and no flocculation occurred. The sol remained perfectly transparent and indefinitely stable in the absence of stabilizers provided the ionic strength was low and the p H remained above 9. When the pH was reduced below about 5, the sol particles slowly coalesced. The mobility could not be measured by laser Doppler electrophoresis because of the small particle size, but the pzc of SnO, suspensions is generally located at pH 4.3.6 Although it is reported that the sol can be peptized below the pzc,' attempts to do this were unsuccessful. Polymeric stabilizers (poly(viny1 alcohol), PVA, and poly(vinylpyrrolidone)),colloidal SiO, (Ludox), and hexametaphosphate were also unable to prevent coalescence in acidic solution. UV and visible irradiations were carried out with a Cathodeon 7833 150-W xenon lamp, using a water jacket and an Oriel 300-nm cutoff filter. Solutions were deaerated with CIG high-purity nitrogen prior to, and during, UV irradiation. Hydrogen was measured by MS using a V.G. Micromass 6 spectrometer. Following irradiation, the solution was frozen in liquid nitrogen and the head-space gas collected in an evacuated sample bulb. The head-space volume was a t least 5 times the sample volume. Blank samples were also analyzed to correct for any hydrogen formed in the absence of the sol. Yields of dihydrogen (expressed as G(H,) in units of molecules per 100 eV) were calculated by comparison with irradiated Milli-Q filtered water containing 0.2 M propan-2-01 and 0.2 M acetone at pH 10.5 for which G(H,) = G, + GH2 was taken to be 1.0.' All irradiations were carried out using a Coeo source a t a dose rate of 18 krad h-'.
200
300
400
500
600
700
Wavelength (nm)
(E)
x13
403
WI
Mn
Wavelength (nml
m
Figure 1. (A) Absorption spectrum of a 0.02 M SnO, sol a t pH 10.5. (B) Emission and excitation spectra of a 0.008 M SnO, sol. For the emission spectrum, excitation was at 300 nm; for the excitation spectrum, the emission was monitored a t 535 nm.
Results i. Absorption and Emission Spectra of Colloidal SnO,. The absorption spectrum of colloidal SnO, is shown in Figure 1A. The absorbance of the sol obeyed Beer's law. Despite the small particle size, the absorption onset is consistent with literature values for the optical band gap of macrocrystalline SnO,.' Because there is considerable scatter in the literature for values of the bandgap energy of SnO,? a blue shift due to quantization effects cannot be completely discounted. The optical absorption coefficient, C Y , of a semiconductor close to the band edge obeys" (Y
(cm-') = K(hv - EJn/hu
(1)
(b) Memming, R.; MBllers, F. Ber. Bunsen-Ges. Phys. Chem. 1972, 76, 469. (c) Memming, R.; Mollers, F. Ber. Bunsen-Ges. Phys. Chem. 1972,
where K is a constant, E , is the band gap, and n is an exponent normally equal to 112,312, or 2 depending on the nature of the transition responsible for the absorption. Since the particle size of the stannic oxide sol is small enough that the contribution from both Rayleigh and Mie scattering may be neglected," the absorbance, A , at any wavelength can be related to the absorption coefficient, a , by a (cm-l) = 2303Ap/(M,C), where M, is the molecular weight, C is the sol concentration, and p is the oxide density. The type of transition can thus be determined from plots of (ahu)'ln vs hu for various values of n. The best f i t was obtained for n = 2 , indicating an indirect transition with E = 3.70 V. For highly defective materials such as colloidd particles, the absorption near the band-gap edge may become exponential, due to an Urbach tail.', Tailing of the absorption edge may indeed be observed in the absorption spectrum of Figure 1A. However, plots of In A vs hu were not linear.
( 5 ) Wrighton, M.; Morse, D.; Ellis, A.; Ginley, D.; Abrahamson, H. J. Am. Chem. SOC.1976,98,44. (6) Houchin, M.; Warren, L. J. Colloid Interface Sci. 1974,100, 278. (7) Sharygin, L.; Gonchar, V.; Barybin, V.; Loguntsev, E.; Shtin, A. Kolloid 2. 1981, 43, 192. (8) Sangster, D.; ODonnell, J. Principles of Radiation Chemistry; Edward Arnold: London, 1970.
(9) Jarzebski, Z. J. Electrochem. Soc. 1976, 123,333C. (10) Dare-Edwards, M.; Goodenough, J. B.; Hamnett, A.; Trevellick, P. J . Chem. Soc., Faraday Trans. 1 1979,79,2027. (11) Ramsden, J.; Gratzel, M. J. Chem. Soc., Faraday Trans. 1 1984, 80, 919. (12) Mott, N. F.; Davis, E. A. Electronic Processes in Non-Crystalline Materials; Clarendon: Oxford, 1978.
(4) (a) Memming, R.; Schropel, F. Chem. Phys. Lett. 1979, 62, 207.
76, 609. (d) Ghosh, P. 5543.
K.;Spiro, T. G. J. A m . Chem. SOC.1980, 102,
Langmuir, Vol. 6, No. 3, 1990 569
Electron Transfer in Aqueous Colloidal SnO, Nevertheless, the existence of intra-band-gap states can be inferred from the fluorescence spectrum. The emission and excitation spectra are shown in Figure 1B. The sol was excited at 300 nm, and the emission is strongly red-shifted. The fluorescence is pale green, and the excitation curve closely matches the absorption spectrum, confirming that the fluorescence results from excitation of the colloid. The maximum at 535 nm is about 1.3 eV less than the band-gap energy. The quantum yield of fluorescence was estimated to be 0.026 for amorphous sols and 0.006 for the cassiterite sols, by comparison with the Ru(bpy)?+ emission, for which @ = 0.042 was a~sumed.'~ The fluorescence spectrum did not change with pH. It was also found that the hole scavengers, SO3'and TEOA (see subsection iii) did not quench the fluorescence. ii. y-Irradiation of SnO, Sols. In order to observe the electron transfer from solution species to the sol, pulse radiolysis and y-radiolysis were used to generate reducing equivalents. The 1-hydroxy-1-methylethylradical and the hydrated electron were used to reduce the colloidal stannic oxide by reactions 2 and 3. The generation of these radicals by radiolysis is described e1~ewhere.l~ (Sn02),
+ (CH,),COH
-
(SnO,),-
+ H+ + CH,COCH, (2)
@noz), + e-(aq) -,@no,),-
(3)
The redox potential of the radical derived from propan-2-01has been measured polarographically as E" = 1.05 V at pH 7,15 and electron transfer from this radical, as well as from the hydrated electron, to the particles is thermodynamically feasible. However, the absorption spectrum of the SnO, colloid remained completely unchanged after extensive radiolysis, sufficient to reduce 70% of the Sn(1V) to Sn(I1). If the electrons reduced the oxide or were stored on the particles, changes in the absorption spectrum would be expected. The experimental observations might be explained if all the radicals underwent recombination. From the known recombination rate constant for the 1-hydroxy-1-methylethylradical (7.8 X lo9 M-l s-l)I5 and from the dose rate, it can be estimated that a rate constant of 12.5 X lo3 M-ls-l f or electron transfer to the sol is required for 199% recombination at the colloid concentration used. In view of the exothermicity of the electron transfer from the radicals to the oxide sol, such a low rate constant seems unlikely. It is, therefore, likely that any transferred electrons catalytically reduce water to hydrogen. To test this hypothesis, degassed samples containing 0.008 M SnO, sol and 0.2 M propan-2-01were irradiated at pH 10.5. Excess hydrogen above the radiation chemical yield was found by MS analysis, and G(H,) = 3.7 f 0.4 was determined. This is close to the theoretical yield of G(H,) = GH2 + GH+(Ge-aq + GH + G o H ) / ~ . When either zwitterionic viologen (ZV) or methylviologen (MV") was added to the solutions in addition to the alcohol, the 1-hydroxy-1-methylethylradical was converted into viologen radicals.16 The redox potentials for these radicals are -0.41 and -0.445 V, respectively." The (13) Van Houten, J.; Watts, R. J. J. Am. Chem. SOC. 1975, 97, 3843. (14) Mulvaney, P.;Grieser, F.; Cooper; Meisel, D. Langmuir 1988,4, 1206. (15) Lilie, J.; Beck, G.; Henglein, A. Ber. Bunsen-Ges. Phys. Chem. 1971, 75,458. (16) Meisel, D.; Mulac, W.; Matheson, M. J. Phys. Chem. 1981, 85, 179. (17) (a) Bird, C.; Kuhn, A. Chem. SOC. Reo. 1981, 10, 49. (b) Nahor, G. S.;Rabani, J. Radiat. Phys. Chem. 1987,29, 79.
0.0
4.0
8.0
12.0
16.0
20.0
[SnO,] (xioJ), M
Figure 2. Stern-Volmer plot for quenching of Ru(bpy):+ ed-state luminescence by colloidal SnO, at pH 10.5.
excit-
viologen radicals did not react on a time scale of 1 2 s with the tin dioxide sol at any pH between 3 and 10 as determined by pulse radiolysis. The yield of viologen radicals produced by steady-state radiolysis at pH 7-10 was also identical with that produced in the absence of the sol. From this, it is concluded that viologen radicals are unable to react with the tin dioxide colloid and that the Fermi level for SnO, lies above the viologen redox potential at pH 3. This is confirmed in subsection iv, where it is shown that conduction band electrons generated photochemically will reduce MV2+and ZV in solution. It is worth noting that in the pH range 3-10 the flat band potential' of macroscopic SnO, is more positive than the standard reduction potentials of the viologen species. Therefore, the lack of reactivity between the viologen radicals and the colloidal SnO, may possibly be due quantum size effects since these can shift the flat band potential to more cathodic potentials as the semiconductor size is reduced." iii. R ~ ( b p y ) ~ ' +Quenching * by Colloidal SnO,. Memming and co-workers4have shown that the luminescence of long-chain derivatives of Ru(bpy)?+* is quenched, by an electron transfer reaction, at the surface of SnO, electrodes. It has also been found that luminescence from Ru(bpy)z+* adsorbed on SnO, powder, in vacuo, was quenched by an electron-transfer process from the excited Ru complex to the s e m i c o n d u ~ t o r .Steady-state ~~ illumination of aqueous solutions of Ru(bpy),'+ in the presence of a suitable donor may, therefore, be expected to lead to a buildup of electrons on colloidal SnO,. The SnO, colloid indeed quenches the emission from Ru(bpy)?+*, as can be seen from Figure 2. The linear dependence in Figure 2 leads to a Stern-Volmer constant of Ksv = 510 M-l. Assuming dynamic quenching and a lifetime of 640 ns for the excited state of the ruthenium complex in the absence of a quencher," this value yields a quenching rate constant of 3.2 X 10l1 M-' s-' (in terms of particle concentration). The diffusion controlled limit is kdiff = 4?rRDNA/1000= 2 X lo1' M-' s- If the electrostatic contribution to the mass-transfer limit is included," the collision frequency may be as high as 8 X lo1' M-'s-' . Nevertheless, it is still markedly smaller than the Stern-Volmer-derived rate constant. It is, therefore, more likely that only adsored Ru complex molecules are quenched. The Stern-Volmer plot then mea(18)Nedeljkovic, J. M.; Nenadovic, M. T.; Micic, 0. L.; Nozik, A. J. J.Phys. Chem. 1986,90, 12. (19) Hashimoto, H., Hiramoto, M., Lever, A. B. P.; Sakata, T. J.
Phys. Chem. 1988,92,1016. (20) Willner, I.; Yang, J.-M.; Laane, C.; Otvos, J. W.; Calvin, M. J. Phys. Chem. 1981,85,3277. (21) Mulvaney, P.;Grieser, F.; Swayambunathan, V.; Meisel, D., submitted to Langmuir.
570 Langmuir, Vol. 6, No. 3, 1990
Muluaney et al.
1.2,
I
I
I
O . 1
L
0.0
I
I
0.00
0.04 0.08 NaClO4/M
0 11
Figure 3. Luminescence intensity of 2.8 X M Ru(bpy) '+* in the presence of 8 X lo-, M SnO, as a function of added NaCIO,. Excitation at 450 nm. Intensity calculated as in Figure 1.
0.9
m /
\
with X > 300-nm light.
I
0.3
0 . 0 1 , I 5 6
,
I
7
,
I
,
8
I
9
,
I
,
I
, I
1 0 1 1 1 2
PH
Figure 4. Effect of pH on luminescence quenching of 2.8
I )me imm\ i
Figure 5. Absorption of solutions at 395 nm containing 8 X lo-, M SnO,, 0.10 M TEOA, and (a) 1.2 X M MV2+, (b) 4 X lo-, M MV2+, and (c) 4 X M ZV vs time of illumination
X
M Ru(bpy),2+by 8 X M SnO,. Curve A: pH lowered from 10.2 to 6.4;curve B: pH raised to pH 11;curve C: pH lowered again to pH 7.5. Arrows indicate direction of pH change.
sures the adsorption isotherm of Ru(bpy)gP+onto the sol particles, and the Stern-Volmer constant can then be used to calculate the adsorption constant, Kad = 2.0 X M-l. Willner et a1." found that adsorption of Ru(bpy),'+ onto SiO, is dominated by electrostatic interactions. Consequently, by screening the charge on the sol the adsorption should then decrease. Indeed, the fluorescence from Ru(bpy):+* increased dramatically as NaC10, was added a t constant sol concentration. The results are shown in Figure 3. (It should be noted that reference solutions containing only Ru(bpy),2+ (5 X M) and NaClO, (up to 0.1 M) showed no change in the emission intensity from the Ru complex with increasing amounts of NaClO,.) Above 0.050 M NaClO,, the fluorescence approached the value in the absence of the sol. This further supports the contention that the sol primarily quenches the emission of the adsorbed Ru(bpy),'+. In Figure 4,the effect of the solution pH on the quenching is shown as the pH of the colloid was cycled between about pH 10 and pH 6. The addition of NaClO, to a sample a t pH 11 (after cycles A and B) did not significantly affect the fluorescence further, suggesting that the effect of raising the pH to 11 was irreversible. This hystersis can be explained in terms of two competing effects on the adsorption of the complex-the effect of pH and the effect of ionic strength-and is discussed in more detail below. When EDTA (0.05 M) or oxalate (0.05 M) was added to 0.008 M SnO, and 2.8 X M Ru(bpy),'+, the sensitizer emission increased. This again is attributable to
the increases in the ionic strength of the solution, causing the ruthenium complex to desorb. This leads to decreased quenching by the sol, at additive concentrations where direct quenching by the additive is still insignificant. iv. Photochemical Generation of MV+ by Excitation of SnO,. In the presence of a suitable hole scavenger, electrons would be expected to accumulate in the conduction band of the colloid particles. That this occurs can be demonstrated by observing the formation of viologen radicals (MV+ or ZV-) from photogenerated conduction band electrons under UV irradiation. A number of common sacrificial donors were investigated, but they were generally inefficient. PVA (0.1%),EDTA (0.050 M), and propan-2-01 (1.0M) did not scavenge photogenerated holes, (0.02 M) was a poor scavenger. The quenchand SO-: ing efficiency by sulfite improved with increased concentration (>0.05 M) but also induced coagulation of the sol in less than 1 h. Triethanolamine (TEOA) (0.1 M) was found to scavenge valence band holes and did not coagulate the sol under UV irradiation. When the sol was irradiated under nitrogen in the presence of TEOA and methylviologen, the blue color of the viologen radical appeared in the solution and increased steadily over 20 min. Blank solutions containing no SnO, did not turn blue when irradiated with X > 300 nm light. When the neutral zwitterionic viologen replaced the cationic methylviologen, the rate of formation of the radical was observed to decrease as shown in Figure 5. As might be expected, the specific adsorption of methylviologen onto the negatively charged sol enhances electron transfer. The rate of radical formation was found to increase with increasing concentrations of both TEOA and viologen. However, the effects were not linear. For example, increasing the TEOA concentration from 0.02 to 0.2 M led to only a threefold increase in the rate of viologen radical formation, and increasing the MV2+concentration from 4X to 1.2 x lo-, M led to a 30% increase in the rate of MV+ formation. The rate of radical formation invariably slowed down after approximately 20 min under virtually all experimental conditions. This may be due to an increasing rate of back-electron transfer from the viologen radical to valence band holes and/or to the consumption of viologen molecules. When the irradiation of the sols was carried out at lower pHs, the rate of viologen formation gradually decreased. A t pH 7 it was 12% of the rate at pH 10.5. Below pH 3, no viologen was detected following irradiation. However in acidic solution, photolysis could only be main-
Langmuir, Vol. 6, No. 3, 1990 571
Electron Transfer in Aqueous Colloidal SnO, tained for a short time ( 300 nm) of a solution containing 8 X M SnO, and 0.20 M TEOA under nitrogen, the solution was frozen in liquid nitrogen and the head-space gas removed. Hydrogen was not found in the photoirradiated solutions at pH 10.5. The absorption spectrum of the solution, however, remained unchanged. In the absence of a suitable acceptor, electron-hole recombination appears to dominate.
Discussion Of the three methods used to create excess electrons on the colloidal stannic oxide, the use of radiolytically produced radicals is the most effective in hydrogen production. Hole scavenging appears to be inefficient due to rapid recombination of electron-hole pairs. The fluorescence spectrum (Figure 1B) provides direct evidence for extensive intra-band-gap recombination centers. These have been variously attributed to intercalated C1- when chloride was present during the oxide preparation and to oxygen v a c a n ~ i e s . ~The . ~ ~low scavenging efficiencies found for numerous reductants despite the very positive valence band potential suggests that most holes are trapped very rapidly by surface states, so that the effective oxidation potential of the holes is much less positive (by up to 1.3 V) than that of the valence band holes. This is indirectly supported by the fact that valence band holes on TiO,, which have a similar redox potential to those in SnO,, oxidize propan-2-01 under UV i r r a d i a t i ~ n while ,~ the holes in stannic oxide do not. Charge transfer from Ru(bpy),,+* to colloidal SnO, is clearly dominated by electrostatic effects on the adsorption of the Ru(bpy),,+. Added electrolyte decreases the {potential of the sol particles, thereby leading to decreased adsorption of the ruthenium complex and hence to smaller quenching yields. The effect of pH (Figure 4) is more complicated. Changes in pH alter the energy of the conduction band edge at the surface by -0.059 V/pH. Since the excited state redox potential of Ru(bpy),,+* is pH independent, the free energy of electron transfer increases as the solution becomes acidic. An increase in the quenching efficiency would then be expected. In fact, the opposite was observed: the efficiency of quenching decreased on decreasing the pH. This again can be ascribed to the electrostatic origin of the adsorption of the ruthenium complex on SnO,. The isoelectric point (iep) of SnO, is located a t about pH 4.3. As the pH is lowered from higher values than the iep, the { potential becomes less negative, and hence the fraction of adsorbed complex decreases and the fluorescence intensity increases. As the pH is increased again, the { potential becomes more negative, and initially the quenching efficiency again increases. However, due to the high sol concentration (0.008 M), a large amount of base is needed to change the pH, which leads to a significant increase in the ionic strength. Once the ionic strength is above 5 X M, adsorption of the ruthenium complex decreases rapidly (Figure 3), and regardless of the pH, the { potential is decreased by the added electrolyte. This reduces the amount of adsorbed complex, and the fluorescence intensity recovers. By raising (22) Kim, H.;Laitinen, H. J . Electrochem. SOC.1975, 122, 53. (23) Henglein, A. Ber. Bunsen-Ges. Phys. Chem. 1982,86, 241.
the pH to 11, virtually all the Ru complex is liberated from the surface, and further pH changes have little effect on the quenching rate (Figure 4, curve C). As can be perceived from the above discussion, a full quantitative analysis of the effect of pH on the sensitized electron transfer to SnO, is difficult. Electrochemical studies have established that tin dioxide is quite inert under both cathodic and anodic bias, with both hydrogen and oxygen evolution occurring without significant lattice destruction, although some metallic tin has been r e p ~ r t e d . ~ Although '~~ macroscopic tin oxides exhibit significant hydrogen overpotentials, the G(H,) values obtained in this study indicate that both the 1-hydroxy-1-methylethylradical and the hydrated electron are quantitatively scavenged by the tin dioxide sol at the concentrations used and that electron transfer results in quantitative hydrogen formation. It is particularly interesting that no change in the spectrum of the sol could be found even after absorbing large radiation doses, which suggests that trapping of electrons as Sn3+centers within the lattice is minimal. By contrast, TiO, sols turn blue in the presence of accumulated electrons, due to trapping as Ti3+ sites near the surface of the colloidal particle^.^^.^^ The instability of Sn3+by compar'ison seems to prevent permanent trapping of electrons on the particle. Hydrogen formation is preferable to permanent reduction of the lattice, even in the absence of a metallic catalyst. Thus, although the overall reduction of SnO, to Sn(OH), may be thermodynamically possible, the first electron reduction step leads to a relatively shallow trapping of the electron, which in turn can efficiently proceed to afford the two-electron reduction of water. The thermodynamic feasibility of cathodic dissolution of SnO, may be further explored by considering the decomposition potential of this material.25 Comparing the two cathodic processes possible at the colloid surface a t pH 10.5 (reactions 4 and 5) shows that for both pathways the free energy of the reaction will have the same pH dependence (-0.059 V/pH).
-
2H+(aq) + 2e-(cb)
+
+
H,(g)
(4)
SnOp 2Ht 2e- Sn(OH), (5) The position of the conduction band edge will also have the same pH dependence. Thus the free energies for both processes are independent of pH and are determined by
mSn02/Sn(OH)z
= @SnOZ/Sn(OH), - @cb
The standard electrochemical potential for the reduction of hydrous SnO, to Sn(OH), is 0.075 V,26 so that lattice reduction is slightly favored over proton reduction. Literature values for the conduction band edge of tin dioxide have been reviewed by Morrison.' Most studies yield Echo < 0; both reactions are thus thermodynamically feasible. On the other hand, the Sn(II1) state has not been identified in any stable compound to date.27 The instability of the Sn(II1) seems to explain the resistance of the oxide to cathodic decomposition a t low cathodic bias. The actual cathodic polarization induced (24) (a) Kolle, U.; Moser, J.; Gratzel, M. Inorg. Chem. 1985,24,2253. (b) Duonghong, D.; Ramsden, J.; Gratzel, M. J . Am. Chem. SOC.1982, 104, 2977. (25) (a) Gerischer, H. J. Electroanul. Chem. 1977,82,133. (b)Gerischer, H.Pure Appl. Chem. 1980,52, 2649. (26) Standard Electrode Potentials In Aqueous Solution; Bard, A., Parsons, R., Jordan, J., Eds.; Marcel Dekker: New York, 1985. (27) Abel, E. W. In Comprehensive Inorganic Chemistry; Pergamon Press: Oxford, 1973; Vol. 14.
572
Langmuir 1990, 6, 572-578
by the accumulated conduction band electrons is diffithe fact that overall free energy changes suggest that cult to calculate, since it depends upon the unknown cathodic decomposition is slightly more exoergonic than steady-state concentration of electrons on the particle and proton reduction. Although its wide band gap precludes their distribution within the particle. It should be noted its use as a sensitizer in solar energy conversion, it may that low doped oxides have donor densities < 1017~ m - ~ . provide a useful, inert redox carrier for photocatalysis. In tin dioxide, such a low donor density corresponds to about one ionized site in 2 X lo5. Sincethe average parAcknowledgment. The assistance of S. Bigger with ticle aggregation number is 390, only one colloidal partithe determination of the hydrogen yields is greatly apprecle in 500 will have one ionized impurity donor. The addiciated. This work was supported by grants from the Austion of just one electron to a particle of aggregation numtralian Institute of Nuclear Science and Engineering. Work ber Nagg= 390 corresponds to an impurity concentration at Argonne National Laboratory is performed under the of 9 X lof9 cm-3 and may therefore substantially raise auspices of the office of Basic Energy Sciences, Division the quasi-Fermi level of the parti~le.'~The overpotenof Chemical Sciences, US-DOE under contract no. W-31tial for both H,discharge and lattice decomposition may 109-ENG-38. P.M. acknowledges the receipt of a Combe easily reached in this case. monwealth Postgraduate Research Award.
Conclusions Transparent colloidal SnO, has been shown to be cathodically stable under reducing conditions. This is despite
Registry No. EDTA, 60-00-4;TEOA, 102-71-6;MV+, 2523955-8; MV+,4685-14-7; Ru(bpy),2+,15158-62-0; SnO,, 1828210-5; H,, 1333-74-0;propan-2-01, 67-63-0; oxalate, 144-62-7.
Characterization of the Electrical Double Layer of Montmorillonite S. E. Miller and P. F. Low*>+ Department of Agronomy, Purdue University Agricultural Experiment Station,$ West Lafayette, Indiana 47907 Received July 27, 1989 and {, the The primary objective of this paper was to determine the relative magnitudes of $o, electrostatic potentials in the plane of the clay-water interface, the outer Helmholtz plane, and the plane of shear, respectively. Another objective was to determine the relative magnitudes of uo and 0 6 , the charge densities in the first two planes, respectively. Four different methods were used to achieve these objectives. All of them gave the same results. It was found that was much smaller than $o but was equal to C also, u6 was much smaller than uo. Moreover, it was found that $a was independent of electrolyte concentration, pH, and uo but dependent on the nature of the exchangeable cations. On the other hand, a6 varied linearly with the square root of the electrolyte concentration. Nevertheless, it remained much smaller than u,,. These results were interpreted to mean that most of the exchangeable cations are condensed on the montmorillonite surface in a Stern layer and that adjustments in the ionic occupancy and/or thickness of this layer maintain $a at a constant, critical value that cannot be exceeded.
Introduction Electrical double-layer theory had its inception with the work of Gouy' and Chapman.' It was extended by L a n g r n ~ i rDerjaguin? ,~ and others and was refined and integrated by Verwey and Overbeek.' Especially since the latter work, it has become the fundamental theory of colloid chemistry. Although modern statistical mechanical methods have been used to further extend and refine Former graduate assistant and Professor of soil chemistry, respectively. Contribution No. 12088. (1) Gouy, G. J.Phys. 1910, 9, 457. (2) Chapman, D.L. Philos. Mag. 1913,25, 475. (3) Langmuir, I. J. Chem. Phys. 1938, 6, 873. (4) Derjaguin, B. V. Trans. Faraday SOC.1940,36, 203. (5) Verwey, E. J. W.; Overbeek, J. Th. G. Theory of the Stability of Lyophobic Colloids;Elsevier: New York, 1948.
*
0743-7463/90/2406-0572$02.50 f 0
the theory: its essential features have remained unchanged. The application of electrical double-layer theory to clays was initiated primarily by Sch~field,~?' but many have used the theory to explain their observations, e.g., Bolt,g Bolt and Miller," Warkentin et al.,ll Norrish,12 Kemper,13Van Olphen,14 Q ~ i r k , Friend '~ and Hunter," ~
(6) Carnie, S.L.; Torrie, G. M. Adu. Chem. Phys. 1984, 56, 141. (7) Schofield, R. K. Trans. Faraday SOC.B 1946, 42,219. (8) Schofield, R.K. Nature 1947, 160, 408. (9) Bolt, G. H.J. Colloid Sci. 1955, I O , 206. (IO) Bolt, G. H.; Miller, R. D. Soil Sei. SOC.Am. Proc. 1955, 19, 285. (11)Warkentin, B. P.; Bolt, G. H.; Miller, R. D. Soil Sci. SOC.Am.
Proc. 1957,21, 495. (12) Norrish, K. Faraday SOC.Discuss. 1954, 18, 120. (13) Kemper, W. D. Soil Sci. SOC.Am. h o c . 1960,24, 10. (14) Van Olphen, H.An Introduction to Clay Colloid Chemistry;Interscience Publishers: New York, 1963. (15) Quirk, J. P. Isr. J. Chem. 1968, 6, 213.
0 1990 American Chemical Society
