Langmuir 1986, 2, 362-368
362
Surface Photochemistry: On the Mechanism of the Semiconductor- Mediated Isomerization of 4 -Substituted cis -Stilbenes1 Tadashi Hasegawa2and Paul de Mayo* Photochemistry Unit, Department of Chemistry, University of Western Ontario, London, Ontario, N6A 5B7 Canada Receiued October 24, 1985. I n Final Form: February 5, 1986 The cis-stilbenes la-g undergo photocatalyzed cis-trans isomerization in the presence of semiconductors. A Hammet plot (o+) shows, with CdS, a sharp break which is interpreted as indicating a change in rate-determining step. The semiconductor-mediated reaction is quenched with electron donors, indicating the intermediacy of a radical cation. A Stern-Volmer plot has both curved and linear components, and two minimal schemes are proposed, assuming the steady-state hypothesis holds, to describe the observed behavior: they differ as to whether the adsorption of the quencher on the surface competes with that of the reactant (competitive model) or whether they have preferred sites and are adsorbed independently (noncompetitive model). Different expressions for $O/$ are obtained, both of which fit the available data. In both models, which represent the extremes of a possible spectrum of behavior, quenching of the surface radical cation by a quencher in solution is required (Eley-Rideal mechanism). Introduction
The use of semiconductors in connection with solar energy storage, whether in the form of energy-rich materials or as electricity, is one of the most active areas of current photochemical research. Much less attention has been focused on the application of these photocatalysts for more general organic transformations, with the possible exception of ~ x i d a t i o n . ~Among the few transformations reported are c y c l o r e ~ e r s i o ndimerization: ,~~~ the cis-trans isomerization of styrene derivatives,' valence isomerization: and a 1,3-sigmatropic rearrangement.s In all these processes radical cations have been proposed as intermediates, but in few cases have any steps been taken to clarify the detailed mechanism. The effect of substituents represents a probe into the reaction mechanism but as yet has been little used. Recently, Fox and Chen reported one such application in the TiOz sensitized oxidation of 4'-substituted diphenyle t h y l e n e ~ . They ~ obtained a linear Hammett plot (p+ = 4.56) and suggested that the rate-determining step in the reaction was the formation of the respective cation radicals. Recently, we have found that a Hammet plot of the CdS-photocatalyzed reaction of 4-substituted stilbenes la-g (Table I) showed a sharp break and interpreted it as a change in rate-determining step.loa The purpose of the present report is to describe the details of the reactions and present a minimal kinetic analysis on the basis of the steady-state hypothesis.
Experimental Section Materials. Semiconductor powders were obtained from the following sources: CdS (Fisher),WO, (Aldrich),CdSe (Aldrich), (1) Publication No. 356 from the Photochemistry Unit, The University of Western Ontario. (2) Present address: Department of Chemistry, Tokyo Gakugei University, Koganei, Tokyo, Japan. (3) Fox, M. A. Acc. Chem. Res. 1983, 16, 314. (4) Okada, K.; Hisamitsu, K.; Mukai, T. J . Chem. SOC.,Chem. Commun. 1980, 941. (5) Barber, R. A.; de Mayo, P.; Okada, K. J . Chem. SOC.,Chem. Commun. 1982, 1073. (6) Draper, A. M.; Ilyas, M.; de Mayo, P.; Ramamurthy, V. J . Am. Chem. SOC. 1984, 106, 6222. (7) Al-Ekabi, H.; de Mayo, P. J . Chem. SOC.,Chem. Commun. 1984, 1231. ( 8 ) Okada, K.; Hisamitsu, K.; Takahashi, Y.; Hanaoka, Y.; Miyashi, T.; Mukai, T. Tetrahedron Lett. 1984, 25, 5311. (9) Fox, M. A,; Chen, C. C. Tetrahedron Lett. 1983, 24, 547. (10) (a) Hasegawa, T.; de Mayo, P. J . Chem. SOC.,Chem. Commun., in press. (b) Al-Ekabi, H.; de Mayo. P. J . Phys. Chem., in press.
Table 1. Oxidation Potentials of 4-Substituted cis-Stilbenes 1 and Relative Quantum Yield for the Production of the Trans Isomer E1/2", V 6 comod substituent u + ~ vs. SCE on CdS on WOa 1.58 1.00 1.00 la H 0 lb IC
NO2 C02Me
Id
c1
le If 1g
Me Me0 NMe2
0.790
0.484 0.114 -0.311
-0.778 -1.7
1.74 1.69
0.22
1.58
0.58
1.45 1.25 0.61
0.96 0.67
0.40
0.18
0.22 1.18 1.45 0.34
0.07 0.002
"Reference 20. *Incident light intensity is not corrected between CdS and W 0 3 so the values are not quantitatively comparable. All the CdS experiments were performed under the same flux and are comparable among themselves, a s are those with WOV PbO (Fisher),Bi203(Fisher),ZnSe (Aldrich),CdO (Aldrich),Fe203 (Fisher),and V,05 (Fisher). Solvents used were all of spectroscopic grade (Baker or Fisher) and were used without further purification. cis-Stilbene (Aldrich),1,2,4-trimethoxybenzene (Aldrich),and anisole (B.D.H.) were purified by fractional distillation. cis-4Nitrostilbene was prepared by condensation of p-nitrobenzaldehyde and phenylacetic acid and purified by recrystallization from ethanol." cis-4-Carbomethoxy-,cis-4-chloro-,cis-4-methyl-, cis-4-methoxy-,and cis-4-(dimethy1amino)stilbenewere prepared by direct irradiation of the corresponding trans isomer,l2,l3 prepared according to literature methods.13-16 Photocatalyzed Reaction of 1 on a Semiconductor. All reactions were performed in a Pyrex tube with 5 mL of solvent and 45 mg of a semiconductor and were about lo-' M in the cis-stilbene 1. Prior to irradiation the mixture was sonicated for 10 min. The Pyrex tube was placed in an aluminum block with a 1.0 x 1.5 cm window. Irradiation was carried out by using a 1-kW PRA xenon lamp run at 840 W; the output was periodically checked with a Scientech 364 power energy meter. A 3-71 filter (Corning) was used as a 460-nm cutoff except in Table IV and the quantum yield determination. None of the cis-stilbenes underwent direct photoisomerization under these conditions. During irradiation the mixture was constantly stirred. Compressed air was used as coolant, and a water filter removed IR radiation. The tubes were open to the atmosphere via a reflux condenser. (11) Pfeiffer, P.; Sergiewskaja, S . Chem. Ber. 1911, 44, 1107. (12) Sleta, L. A.; Ershov, Y. A,; Kononenko, G. G.; Malkes, L. Y. Teor. Eksp. Khim. 1978, 14, 331. (13) Zechmeister, L.: McNeely, W. H. J. Am. Chem. Soc. 1942, 64, 1919. (14) Wheeler, 0. H.; de Pabon, N. H. B. J. Org. Chem. 1965,30,1473. (15) Fuson, R. C.; Cook, H. G., Jr. J. Am. Chem. SOC.1940,62,1180. (16) Haddow, A.; Harris, R. J. C.; Kon, G. A. R.; Roe, E. M. F. Philos. Trans. R. SOC.London, A 1948, No. 241, 147.
0743-7463/86/2402-0362$01.50/00 1986 American Chemical Society
Langmuir, Vol. 2, No. 3, 1986 363
Surface Photochemistry ? .
-1.1
-1.1
I
I
-1.1
-*.I
I
TIME lhrl
Figure 1. CdS-mediated cis-trans isomerization of the stilbenes: ( 0 ) cis-4-nitrostilbene; (m) trans-4-nitrostilbene; (0) cis-4-
methylstilbene; ( 0 )trans-4-methylstilbene.
In all cases a known amount of a long-chain hydrocarbon, as calibrant, was added to the mixture before irradiation. After irradiation the semiconductor was removed by filtration. The filtrate was analyzed by using a Varian 3700 gas chromatograph equipped with flame ionization detector, which was connected to a Hewlett-Packard 3390 A integrator, except in the case of ci~-4-(dimethylamino)stilbene (le). Analysis was performed on a 2 m X 2 mm column packed with 3% or 10% OV-101 on Chromosorb W(HP) 80-100 mesh. In the case of lg, analysis of the cis and trans isomer was performed using a Varian XL-200 NMR spectrometer: after filtration of the reaction mixture the solvent was removed under reduced pressure below 20 OC; the residue was dissolved in CDCls containing a known amount of hexadecane as internal standard. The amounta of the cis and the trans isomers were determined by comparison of the relative intensities of the peaks of the N-methyl group of the cis and the trans isomers (at 6 2.92 and 2.98 respectively) wi* that of the peak of the standard. Quantum Yield Determination. A methylene chloride solution of la (0.02 M) in a Pyrex tube was irradiated in an aluminum block with a window (1 X 1.0 cm) with a 150-Wxenon lamp via a Corning glass filter 5-58. The light falling on the window was assumed to have been completely absorbed. Potassium ferrioxalate was used as an actinometer. Preparative Photocatalyzed Dimerization of trans -4(Dimethylamin0)stilbene (2g). A mixture of 0.152 g of 2g, 70 mg of CdS, and 5 mL of CHzClzwas placed in a Pyrex tube, sonicated for 10 min, and irradiated for 9 h with a l-kW xenon lamp run at 840 W. After irradiation, the CdS was removed by filtration. The filtrate was concentrated under reduced pressure and the residue chromatographed on silica gel. Elution (ethyl acetate-benzene) gave 0.012 g (conversion,80%)of r-l,t-a-bis(4-(dimethylamino)pheny1)-c-3,t-4-diphenylcyclobutane:mp 110-111OC; 'H NMR (CDClS) 6 2.91 (6 H, 8, NMEZ), 3.55 (2 H, d, J = 9.6 Hz), 3.57 (2 H, d, J = 9.6 Hz), 6.69 (4 H, d, J = 3.0 Hz, Ar), 7.18 (4 H, d, J = 3.0 Hz, Ar), 7.25-7.4 (10 H, m, Ar);maas spectrum (70 eV), m / e (relative intensity) 446 (M+, l),266 (55), 223 (100);exact maea spectrum calcd for C&IuNz 446.2722, found 446.2708. The mass spectrum indicates the dimer is a head-tohead isomer. The stereochemistry of the dimer was deduced by analogy with the chemical shifts of the methine protons in the stilbene dimer r-l,t-2,~-3,t-4-tetraphenylcyclobutane which have been reported to appear at 6 3.63 and those in r-l,c-2,t-3,t-4tetraphenylcyclobutane at 6 4.40.''
Results and Discussion
The Hammett Plot. Irradiation of a suspension of CdS
in methylene chloride containing cis-stilbene lb-e at A > 460 nm (CdS band gap 2.45 V?O -510 nm) gave, as did
(17) Schechter, H.; Link, W. J.; Tiers, G. V. D. J. Am. Chem. SOC. 1963,85, 1601. (18) Kohl, P. A.; Bard, A. J. J. Am. Chem. SOC.1977,!-W, 7631.
u*
1
I
1.1
1.1
Figure 2. Hammett plot for the production of the trans isomer from the cis isomer la-g on CdS (0, log $O/&: = 0.54cr' + 0.19 (r = 0.9914) and -0.74~' - 0.07 (r = 0.9724)) and W 0 3 ( 0 ,log q5°/$12= 1.54~' (r = 0.9998) and -2.42~' + 1.24) irradiated at X > 460 nm.
I
-1.1
I
-1.1
-
I
1.1
I
-1.1
I
I
I
I
1.5
1.1
U+
Figure 3. Oxidative peak potential (E1 of the stilbens la-g vs. the u+ values. Ellzol= 0.3126 1d33; r = 0.982 (except for
le).
+
the parent la,' a nearly quantitative yield of the trans isomer (Figure 1). Appropriate control experiments indicated that both light an CdS were required. The relative rates for the trans isomers using la as a standard were determined and are listed in Table I together with their oxidation potentials (cyclic voltametry). Prolonged irradiation of cis-4(dimethylamino)stilbene (lg),gave a dimer, and the same dimer was obtained from the trans isomer: at low conversion the trans isomer is the sole product (for details, see Experimental Section). A Hammett plot for the reaction is shown in Figure 2, and has a sharp break at u+ = 0.19. The oxidation peak potentials of the stilbenes correlate linearly with u+ (Ellzol = 0.312~~'+ 1.533;r = 0.982) except for lg (Figure 3). The rate-determining step in the isomerization appears to change in the region of E l I 2 O x = 1.47 V. This is close to the value of the valence band of CdS (band gap 2.45 V; flat band potential -0.85 V in MeCN18). The currently accepted mechanism for the photocatalyzed reaction at a semiconductor surface involves exciton generation and hole/electron separation. Capture of these species by an adsorbed organic molecule may give rise to a cation (or (19) Technic ofElectroorgunicSynthesis; Weinberg,N. L., Ed.; Wiley: New York, 1974; pp 684-687. (20) Brown, H.C.: Okamoto, Y. J. Am. Chem. SOC.1958,80, 4979.
364 Langmuir, Vol. 2, No. 3, 1986
Hasegawa and de Mayo
Table 11. Relative Reactivities for the Photocatalyzed Cis-Trans Isomerization of cis-Stilbene (la) on Different Semiconductors semiband conductor gap CdS 2.45 CdSe 1.7 WOg 2.4-3.65 PbO 2.8 Biz03 2.8 ZnSe 2.7 CdO 2.1 Fez03 2 V205 2.75
flat band potential
ref 18 26 21-24 27 27 26 27 27 27
re1 reactivity'
-0.85 (in CH,CN)" -0.3 to +0.2 (pH 7)b 0 (pH 9)* 0.2 (pH 13)b
0 (pH 13)b -0.1 (pH 13)b 0.9 (pH 7)b
1.0 0.97 0.64 0.02 0.02 0.01
Table IV. Effect of Cutoff Wavelength on the Photocatalyzed Reaction of la on CdS region of light
filter Corning 3-73 Corning 3-71 Corning 3-68 Corning 2-61
X > 403 nm (3.1 V) X > 460 nm (2.7 V) X > 520 nm (2.4 V) X > 611 nm (2.0 V)
reactivity 1.31 1.00 0.65 0.10
" Incident light is not corrected.
0 0 0
" Vs. SCE. Vs. NHE. Irradiation was carried out by using a Corning 3-73 (X > 403 nm, 3.1 V) filter. The values are not corrected to constant incident flux. Table 111. Relative Efficiency of the Reaction of cis-Stilbene in Different Solvents solvent n-pentane cc1, CH& CH,COCH, CH30H CH,CN
" Reference
ea
1.84 2.238 9.08 20.70 32.70 37.5
re1 quantum yield 1.37 0.70 (1.00) 0.03 0 0
28.
anion) radical which can lead to net chemical change. The capture is generally exergonic. Electron transfer to the hole in the CdS valence band must be endergonic in la-d, and this process may well be the rate-determining step in the isomerization of these compounds. This is consistent with the negative p+ value (4.74); unfortunately we are unaware of any values of p+ for endergonic reaction of cation radicals in the literature. The rate-determining step of the reaction of le-g cannot be so rationalized because of the In positive p+ value (0.57), the reaction being exerg~nic.~ this situation the rate-determining step is most probably the rate of bond rotation. A similar Hammett plot for the cis-stilbenes over WO, is shown in Figure 2: here the break occurs near a+ = 0.31, the rate-determining step can be considered to change -El,zox = 1.63 V, and ita interpretation may be similar to that of CdS. No information about the location of the flat band potential of WO,in pure organic solvents is available, and different values of the band gap energy and the flat band potential in aqueous electrolyte solution have been reported, ranging from 2.5 to 3.65 V and -0.3 to +0.2 V vs. SCE (corrected to pH 7), respectively.21-24 It has been noted that the scattering is probably to be attributed to a strong influence of the preparative technique on the electrical and optical properties,22and it has also been pointed out that the capacitance depends strongly on the sample history and the electrolyte and that linear MottSchottky plots are hard to obtain.25 Although our value is certainly not consistent with the reported values determined in aqueous electrolyte solution, it may represent (21) Butler, M. A. J. Appl. Phys. 1977,48, 1914. (22) Gissler, W.; Memming, R. J. Electrochem. SOC.1977, 124, 1710. (23) Hardee, K. L.; Bard, A. J. J.Electrochem. SOC.1977, 124, 215. (24) NenadoviE, M. T.; Rajh, T.; MiEiE, 0. I.; Nozik, A. J. J. Phys. Chem. 1984,88,5827. (25) Butler, M. A.; Nasby, R. D.; Quinn, R. K. Solid State Commun. 1976,19, 1011. (26) Krylov, 0. V. Catalysis by Nonmetals; Academic Press: New York, 1970. (27) Morrison, S. R. Electrochemistry at Semiconductor and Oxidized Metal Electrodes; Plenum Press: New York, 1980; p 183. (28) Lunge's Handbook of Chemistry, 13th ad.; Dean, J. A,, Ed.; McGraw-Hill: New York, 1985.
\
\
L
\
\
600
700
nm
Figure 4. Photoacoustic spectrum of CdS.
the level of the valence band of WO, in methylene chloride. Efficiency of the Reaction. The efficiency of the isomerization depends strongly on the semiconductor: the relative quantum yields for cis-stilbene are listed in Table 11,the value for CdS being 0.02. The quantum yield does not depend on the valence band level and probably depends on the lifetime of the photogenerated hole which may vary over several orders of magnitude.29 The efficiency is also solvent-dependent (Table 111), but the results do not admit of a simple interpretation. Probably polarity plays a part, not only in stabilizing charges but, as in an elutropic series,30in varying the amount of substrate adsorbed on the surface. As expected, the light wavelength affects the efficiency (Table IV), but it should be noted that reaction still occurred with undetermined efficiency with light of energy lower than the band gap. The photoacoustic spectrum (Figure 4) shows tailing beyond the band gap. This absorption is most likely caused by transition from the valence band to states below the conduction band. These interband states may be distributed over a relatively large energy range: thermal excitation may transfer some of these trapped electrons to the conduction band.22 Kinetic Analysis for the Hammett Plot. A minimal simple scheme for the photocatalyzed reaction of the stilbenes can be written as in Scheme I, where c- and &Dad are adsorbed cis- and trans-stilbene, respectively, SC and SC* are the ground and the excited state of the semicon(29) Smith, R. A. Semiconductors, 2nd ed.; Cambridge University Press: Cambridge, 1978. (30) Thin-Layer Chromatography, 2nd ed.; Stahl E., Ed.; SpringerVerlag: New York, 1969.
Surface Photochemistry
Langmuir, Vol. 2, No. 3, 1986 365
ductor, and SC(h+,e-) is the charge-separated semiconductor.
Scheme I
sc 2s c * k3
I
-2.1
SC-*/C-D,d+.
k8 +
SC-./t-D,,+.
+
t-D,d
+ sc
(g)
Of the many suggested kinetic equations for the description of catalytic systems, the best approximations are related to the Langmuir-Hinshelwood treatment.6~31-34 These assume that there is a fast adsorption-desorption equilibria of the components of the reaction mixture on the catalyst surface and that the rate of reaction be proportional to the surface coverage. Lateral interactions are ignored. Application of the steady-state hypothesis to Scheme I leads to the quantum yield expression in eq 1, 40
k4 =-
k3
k6°r0
k8
(1)
+ k4 k5 + k&,O k, + k8
where 0: is the fraction of the surface covered by the reactant. An expression for 0: can be obtained by application of the steady-state hypothesis but is too complex to manipulate. If it be assumed that adsorption and desorption of the reactant (and quencher) are faster than reaction and that the surface coverage of the intermediate and product can be neglected, then if solvent and reactant do not compete for the same sites (non-competitivemodel) 0: may be written as eq 2A, where Kr = kl/k2. If the
solvent does compete for the reactive sites, then we have eq 2B, where Ka = Ks,ad/Ka,dea
In this approximation the real surface coverage of the reactant should be lower than indicated by these equations which have the form of the Langmuir-Hinshelwood model. The Hammett equation can be written by using eq 1and 2: k6°:
4O 41,
log 7 = log
where
k8
+ k,O,O + log k7 + k8 kgla k5 + kslaOl> k71a + k ala k,
kg"812
(3)
and 4" are the quantum yields for production
(31)Laidler, K.J. In Catalysis; Emett, P. E., Ed.; Reinhold New York, 1956;Vol. 2,p 119. (32)Cerveny, L.; Puzicka, V. Adu. Catal. 1981, 30, 335. (33)Shih, Y.S.;Yang, S. S. J. Catal. 1983, 79,132. (34)Hsiao, C. Y.;Lee, C. L.; Ollis, D. F. J. Catal. 1983, 82, 418.
-1.5
1 -1.1
I
I
-1.5
u '
I
1.1
1 1.1
Figure 5. Hammett plot for the production of the trans isomer from the cis isomer la-g on CdSe irradiated at h > 460 nm.
of the trans isomer from cis-stilbene itself and substituted stilbenes, respectively. The first term in eq 3 is related to the electron-transfer process from the reactant to the hole on the semiconductor surface, and the second term can be considered to depend on the rate of the bond rotation of the cis cation radical. Both of the first and the second terms should depend on the definic *-electron density (bond order), the rate of bond rotation increasing, and the rate of electron transfer decreasing with decrease of the electron density. It seems reasonable that both the first and the second terms should correlate linearly with U+.
When the rate of bond rotation ((g) in Scheme I), k,[SC-./c-Dd+.], is much faster than that of the back electron transfer from the semiconductor to the cation radical ((f) in Scheme I), k7[Sc--/c-Dad-'], k7 can be neglected compared with k8 and the second term in eq 3 can be regarded as a constant. In this case the Hammett plot should decrease with increase of the oxidation potential of the reactant. This seems to be the case for the reaction of la-d on CdS and that of l b and IC on W03. When the rate of the electron transfer ((e) in Scheme I), k,O:[SC(h+,e-)], is much faster than that of the hole-electron recombination ((d) in Scheme I), k5[SC(h+,e-)],k5 can be neglected compared with k,O,O, and the first term can be considered to be constant. This seems to be the case for that reaction of le-g on CdS and la,d-lg on W03. The first requirement depends on the 7-electron density of the reactant and the interaction with the semiconductor surface. The second requirement depends on the reductive power of the reactant and the position of the valence band of the semiconductor; only an exergonic electron-transfer process fulfills this requirement. When these requirements are not fulfilled a linear Hammett plot is not to be expected. Such appears to be the case of CdSe (Figure 5). The sharp break in the CdS Hamett plot thus indicates a change in rate-determining step, and the oxidation potential at that break may give the level of the valence band of the semiconductor provided the intrinsic barrier35for the electron transfer from the stilbenes to the semiconductor hole is near zero. If the intrinsic barrier has a positive value then a larger value for the valence band level than that indicated by the break can be expected. This may be the case for W03. The Hammett plot for the reaction on CdS, a t least, can be rationalized in terms of (35)Scandola, F.;Balzani, V.; Schuster, G. B. J.Am. Chem. SOC.1981, 103, 2519.
366 Langmuir, Vol. 2, No. 3, 1986
Hasegawa and de Mayo
I
I
s
t
IJ
I S ([Oll*)I
ll
1
I
IS
I
I
5
It ([PIII).
16'
Figure 6. Quenching for the production of the trans isomer from la with 1,2,4-trimethoxybenzene (e)and anisole (0). Curve a was obtained from eq 14 by using a = 239 A, b = 4047 A, and c = 728 A. Curve b and line c are obtained from the partial Stern-Volmer equation of (1 + b[Q])/(l + c[Q]) and 1 + a[&],
respectively.
I
I
I5
21
102
Figure 8. Quenching for the production of the trans isomer from I f with 1,2,4-trimethoxybenzene( 0 )and anisole (0). Curve (a) was obtained from eq 14 by using a = 130 A, b = 1290 A, and c = 530 A. Curve b and linear line c are obtained from the partial Stern-Volmer equation (1 + b[Q])/(l + c[Q]) and 1 + a[Q], respectively.
be noted that all the Stern-Volmer plots have both a linear and a curved section. The efficiency of the quenching of the reaction should depend on the difference between the oxidation potential of the reactant and the quencher. This is well reflected in the slope of the Stern-Volmer plots of the stilbenes a t near zero concentration of the quencher. On the other hand, anisole, which has a higher oxidation potential (Ell2" = 1.76 vs. SCE)= than la and lf, did not quench their reaction.
Scheme I1
I
I
s
I
I
II I S ( ( 1 , l . ~ . r r i m ~ ~ ~ o . " a ~ nI U~ )~ In *I] cl
I
I
24
15
Figure 7. Quenching for the production of the trans isomer from l b with 1,2,4-trimethoxybenzene.Curve a was obtained from eq 14 by using a = 135A, b = 8336 A, and c = 1303A. Curve b and line c are obtained from the partial Stern-Volmer equation of (1 b[Q])/(l + c[Q]) and 1 + a[&], respectively.
+
both an exergonic and endergonic photocatalyzed process with different rate-determining steps. The partial quantum yields represented in eq 1 can be approximated from the Hammett plot. Thus, the intercepts a t u+ = 0 indicate the values of log + k81a)/ka1a and log (ke + k61a61>)/k61a61> are 0 and 0.19,respectively, which give partial quantum yields of unity for the bond rotation of stilbene and 0.65 for the hole quenching by la. The quantum yield for the production of the trans isomer was determined to be 0.02. From these values the partial quantum yield, k3/(k3+ k4), can be estimated to be -0.03 for CdS. This small value of k3/(k3+ k4) seems to support the idea that only the longer lived holes participate in the photocatalyzed reaction. Quenching of the Photocatalyzed Reaction. A photocatalyzed reaction on a semiconductor involving a cation radical should be quenched by electron donors. Indeed, the reactions of la, lb, and If were efficiently quenched with 1,2,4-trimethoxybenzene ( E l I 2 O x = 1.12 vs. SCE).36 The results are shown in Figures 6-8 and it will (36)Zweig, A.; Hcdgson, W.G.; Jura, W. H.J. Am. Chem. SOC.1964, 86, 4124.
k13
SC-*/c-Dad+'
+
Qsol
Eley-Rideal-type quenching*
Qad+* C-Dad (k)
SC-.
-sc + k14
SC-*/Qad+*
Qad
(1)
Application of the steady-state hypothesis to the system now including Scheme I1 (indicating the quenching processes incorporated) gives eq 4. Here, eq is the fraction
4= k3 k66r k3 + k4 k5 + h 6 r + k116q
k8
k7
k~ + k12eq + K13[Qsoll
(4)
of the surface covered by the quencher. Dividing eq 1 by eq 4 gives eq 5, where u," and ur are the surface coverage
4O -_
4
(k5
+ k&, + kllBq)B," (k5 + k6&)6,"
k120q
1+-+k7 k8
+
K13
k7
+ ks [&sol1
)
(5)
Surface Photochemistry
Langmuir, Vol. 2, No. 3, 1986 367
of the reactant in the abence and the presence of the quencher,respectively It will be noted that quenching by interaction with a molecule in solution is included in the quenching processes (Eley-Rideal model). In a previous treatment of the quenching of cis-stilbene itselflob only surface quenching was considered. The second term in eq 5 can be regarded as encompassing cation radical and Eley-Rideal-type quenching. The first term in eq 5 includes two factors: hole quenching and the effect of displacement of the reactant on the surface by the quencher (the surface occupation quenching). If we use the steady-state hypothesis and assumptions that we used to derive 8: in the absence of quencher, we may use the Langmuir-Hinshelwood model. The surface coverage, 8, may be expressed in two extreme ways: either adsorption is site-competitive, or it is not. When the quencher is adsorbed on the surface and competes with the reactant (the competitive model), Br and 8, can be written as follows:
.
matical treatment. From eq 10 and 11with squared terms of [QmJ neglected, we obtain eq 12 and 13, both of which Competitive Model
4O.-
--1+ 4
6O --1+_ dJ
h 3
k,
+ ka [Qsoil +
can be written in the form of eq 14. The data in Figures
where Ks = k,/kl,. When the quencher is adsorbed on the surface independently of the reactant (noncompetitive model), 0, and 8, can be written as follows:
Bq =
Kq[Qs0il
1 + Kq[Qsoil
Both seta of equations are useful and have been used in surface ~ h e m i s t r y . ~However, ~ they lead the different expressions for $O/$. Equations 10 and 11can be obtained
Conclusions
Competitive Model
hole
k7
+
+ 1
k,
surface occupation quenching
+
K,ESl
+
cation radical
+
KqCQe01l
KrCc-DsoI1
+
+
KqCQsOlI
Eley-Rideal- type quenching
Noncompetitive Model
c[, + d k,
kll(l
+
6-8 fit curves obtained with the values indicated in the captions.37 The results obtained, therefore, seem compatible with the models used, which include an Eley-Rideal process. In view of the assumption made in the derivation of eq 6 and 8-that the adsorption and desorption of the reactant and quencher (and solvent) are faster than the reaction and that the surface coverage of the intermediate and product can be neglected-we do not feel that the significance of the numerical evaluation should be pressed. While 8, can, in principle, be derived from the values of c we doubt the value of such a derivation but consider the simultaneous operation of Langmuir-Hinshelwood and Eley-Rideal processes as rendered likely by this analysis.
+
(ks
Kr[~-Dsoll) k6)K,[C-Da,11
+
KqCQaol’
1
+
KqCQso~l
The Hamett plot for the photocatalyzed reaction of the cis-stilbenes 1 on CdS and W03 has a sharp break, which can be explained in terms of a change in the rate-determining step. The Stern-Volmer plot of the reaction involves a linear and a curved part. Simple kinetic analysis based upon the steady-state hypothesis has been successfully applied. This procedure can rationalize the Hammett plot for the reaction of 1 on the semiconductor surface and quencher behavior. In kinetic analysis of quenching both a competitive and a noncompetitive model have been used. Different expressions for $ O / $ have been obtained by using these models; however, the experimental results can be rationalized by either of them. The models must be considered as limiting cases for the photocatalyzed reaction on a semiconductor surface, and both include simultaneous surface quenching (Langmuir-Hinshelwood) and quenching from solution (Eley-Rideal).
hole quenching
cation radical
+
Eley-Rideal quenching
by assuming a competitive and a noncompetitive model, respectively. The form of the first term in eq 10 and 11 differs because of the definition of 8 used. The real situation must lie between these limits but does not lend itself t o mathe-
Acknowledgment. We are grateful to Professor J. P. Guthrie (Chemistry, U.W.0,) for assistance with the nonlinear least-squares calculations. We thank Dr. H. AlEkabi for fruitful discussion and A. M. Draper for the oxidation potential determinations. We are indebted to the Natural Sciences and Engineering Research Council of Canada for financial support. (37) T h e curve fitting was achieved by a nonlinear least-squares me-
thod.
368
Langmuir 1986, 2, 368-372
Registry No. cis-la, 645-49-8; trans-la, 103-30-0; cis-lb, 6624-53-9; trans-lb, 1694-20-8; cis-lc, 46925-32-0; trans-lc, 1149-18-4;cis-ld, 1657-49-4;trans-ld, 1657-50-7;cis-le, 1657-45-0; trans-le, 1860-17-9;cis-lf, 1657-53-0;trans-lf, 1694-19-5;cis-lg, 14301-11-2;trans-lg, 838-95-9;CdS, 1306-23-6;CdSe, 1306-24-7;
W03, 1314-35-8;PbO, 1317-36-8;Bi203,130476-3; ZnSe, 1315-09-9; CdO, 1306-19-0;FezO3,1309-37-1;V20,,11099-11-9;CCh, 56-23-5; CH2C12,75-09-2;CH3COCH3,67-64-1; CH,OH, 67-56-1; CH3CN, 75-05-8 n-pentane, 109-66-0;1,2,4-trimethoxybenzene,135-77-3; anisole, 100-66-3.
Dissociative and Molecular Bromine Adsorption on an Fe( 100) Surface P. A. Dowben" and M. Grunzef Fritz-Haber-Institut der Max-Planck-Gesellschaft, 1000 Berlin 33, FRG Received August 21, 1985. I n Final Form: December 16, 1985 We have studied bromine adsorption on an Fe(100) single-crystal surface between 110 and 550 K by X-ray and ultraviolet photoelectron spectroscopy. For the whole temperature range, the initial adsorption was found to be dissociative and lead to formation of a chemisorbed overlayer. At T 5 140 K, we observed the formation of solid bromine growing in a layer by layer mechanism. No evidence for iron halide formation was found under our experimental conditions. Introduction
I n this paper we describe the adsorption behavior of bromine on an Fe(100) surface between 110 and 550 K. The results presented here complement our previous reports on X-ray photoemission experiments for bromine and iodine adsorption on an iron(100) surface.' Initial adsorption of bromine and iodine at T > 110 K is always dissociative and is followed by molecular halogen adsorption at substrate temperatures of T I 140 K for Br, and T I 230 K for I,, respectively. Angle-resolved photoemission studies have identified two molecular bromine adsorption states on Ni(100), and Fe(110)2 adsorbed onto the chemisorbed atomic bromine overlayers at low temperatures. The spectral features of the molecularly adsorbed bromine have been interpreted as arising from Br, molecules having different orientation of their molecular axis with respect to the surface normal. Similar behavior has been observed with molecular I, adsorption on Ni( Fe(l10),4 and iron i ~ d i d e .In ~ a recent paper, Benndorf e t ala5studied bromine adsorption on an Ag(ll0) surface. The formation of AgBr following the initial dissociative chemisorption of bromine was indicated for T 1 130 K. Only at 130 K upon the AgBr corrosion layer was a molecular bromine state observed. As detailed in this paper, we found no evidence for iron bromide formation on Fe(100) under the low-pressure conditions applied in this study. The results for halogen adsorption on solid surfaces in general have been summarized and reviewed6r7 and a comprehensive survey of the literature may be found there. Experimental Section
The photoemission experiments were performed in a commercial stainless steel UHV chamber (Leybold-Heraeus),described previously.s The chamber was equipped with a hemispherical electron energy analyzer, an A1 K a X-ray source (Ekln= 1486.6 eV), a differentiallypumped rare-gasdischarge lamp, a quadrupole *Address correspondence to this author at Department of Physics, Syracuse University, Syracuse, NY 13244-1130. Present address: Department of Physics and Laboratory for Surface Science and Technology, University of Maine, Orono, ME 04469.
mass spectrometer, an electron gun, and an Ar+ ion gun for cleaning the specimen. The angle of incidence of the probing photon and electron sources was 40' off normal and the emitted electrons were collected normal to the surface. The spectra were recorded by direct pulse counting into a Nicolett instrument computer to improve the signal to noise ratio and then plotted on an n-y recorder. Binding energies in this paper refer to energies below EF of the clean iron substrate. Adsorption curves were determined from integrated XPS core level intensities or the XPS core level intensity maximum. The methods of cleaning the Fe(100) surface and estimating the concentration of the residual contaminents have been described elsewhere.' The oxygen residue on the surface in the experiments was estimated to 3% of a monolayer and slight, if any, increase was observed after an experimental run. The work function change measurements were obtained from the cutoff of the secondary electron photoemission background in He1 photoemission experiments. Earlier retarding potential diode work function change measurementsgwere also repeated. Coverages throughout this paper will be denoted in terms of atoms per surface unit cell of the clean Fe(100) surface (1.214 X 1019atoms m-2),denoted by r, or, alternatively, with respect to the saturation coverage of the bromine chemisorbed overlayer at room temperature and denoted as 6'/6',. Exposures given here are not corrected for the ionization gauge sensitivity of bromine unless otherwise stated. Gaseous bromine was admitted to the chamber via a standard leak valve. The base pressure was generally 2 x mbar, and some bromine adsorption on the surface was found to occur during photoelectron spectra acquisition on the previously clean surface. This bromine adsorption was not found to alter our results significantly and remained less than 3% overall. Results
Bromine adsorption was followed as a function of ex(1) Dowben, P. A.; Grunze, M.; Tomanek,D. Phys. Scr. 1983, T4,106. (2) Dowben, P. A.; Mueller, D.; Rhodin, T. N.; Sakisaka, Y. Surf. Sci., in press. (3) McConville, C. F.; Woodruff, D. P. Surf. Sci. 1985, 152/153,434. (4) Mueller, D.; Rhodii, T. N.; Dowben, P. A. Surf. Sci. 1985,164,271. (5) Benndorf, C.; Kruger, B. Surf. Sci. 1985, 151, 271. (6) Grunze, M.;Dowben, P. A. Appl. Surf. Sci. 1982, 10, 209. (7) Farrell, H. H. In The Chemical Physics of Solid Surfaces and Heterogeneous Catalysis. King, D. A., Woodruff, D. P., Eds.; Elsevier: New York, 1984, Vol. 3b. (8) Grunze, M. Surf. Sci. 1979, 81, 603. (9) Dowben, P. A.; Jones, R. G. Surf. Sci. 1979, 88, 348.
0743-7463/86/2402-0368$01.50/063 1986 American Chemical Society