J. Phys. Chem. 1989, 93, 859-864
859
Photoelectrochemistry in Particulate Systems. 9. Photosensitized Reduction in a Colloidal TiO, System Using Ant hracene-9-carboxylic Acid as the Sensitizer Prashant V. Kamat Radiation Laboratory, University of Notre Dame, Notre Dame, Indiana 46556 (Received: February 16, 1988; I n Final Form: June 6, 1988)
Time-resolved laser flash photolysis and fluorescence quenching studies have been carried out to elucidate the processes of charge injection from the excited anthracene-9-carboxylicacid (9AC) into the conduction band of Ti02 semiconductor colloid and the reaction of sensitized Ti02colloid with an electron acceptor. 9AC adsorbed strongly on colloidal Ti02with an apparent association constant of 6450 M-I, and its fluorescence emission was quenched by Ti02 colloid with an efficiency of 94%. Only the singlet excited state of 9AC was found to participate in the sensitization of TiOz with a rate constant of 4.8 X lo8 s-l for the process of charge injection into the conduction band of the semiconductor. Analysis of the transient absorption spectra confirmed the generation of cation radical (9AC+'), most of which decayed quickly by recombining with the injected charge. The rate constant for the back electron transfer was found to be 5.5 X lo7 s-l. The small fraction of the injected charge ( N 10%) that survived within the particle was used to reduce another substrate, N,N,N',N'-tetraethyloxonine. The quantum yield for such a sensitized reduction process was 0.015. The role of the sensitizer and the semiconductor in promoting the sensitized reduction process is described.
Introduction Photosensitization of a stable large band-gap semiconductor is an interesting and useful phenomenon that is used to extend its absorptive range and thus carry out photoelectrochemical reactions under visible light irradiation. Despite the efforts of several research groups, the photosensitization efficiency in most of the examples was less than 0.1%.1-5 However, recent efforts to achieve efficiency in the range of 30-80% have renewed interest in the photosensitization of large band-gap semiconductor^.^^^ Techniques such as laser flash p h o t o l y s i ~ , ~resonance -~~ Raman spectroscopy,12diffuse refle~tance,'~~ and microwave absorption13b have been demonstrated to be useful in investigating the mechanism of charge injection from the excited state of the sensitizer (1) See for example: (a) Gerischer, H.; Willig, F. Top. Curr. Chem. 1976, 61, 31. (b) Meier, H. Photochem. Phorobiol. 1972, 16,219. (c) Watanabe, T.; Fuhishirna, A.; Honda, K. In Energy Resources through Photochemistry and Catalysis; Gratzel, M., Ed.; Academic: New York, 1983; Chapter 1 1 . (2) See for example: (a) Spitler, M.; Calvin, M. J. Chem. Phys. 1977,67, 5193. (b) Takizawa, T.; Watanabe, R.; Honda, K. J. Phys. Chem. 1980,84, 51. (c) Matsumura, M.; Mitsuda, K.; Yoshizawa, N.; Tsubomura, H. Bull. Chem. SOC.Jpn. 1981, 54, 692. (d) Itoh, K.; Chiyokawa, Y.; Nakao, M.; Honda, K. J. Am. Chem. SOC.1984, 106, 1620. (3) See for example: (a) Jaeger, C. D.; Fan, F. R. F.; Bard, A. J. J . Am. Chem. SOC.1980, 102, 2592. (b) Morishima, Y.; Isono, M.; Itoh, Y.; Nazakura, S. Chem. Lett. 1981, 1149. (c) Bauldreay, J. M.; Archer, M. D. Electrochim. Acta 1983, 28, 1515. (4) (a) Fox, M. A.; Hohman, J. R.; Kamat, P. V. Can. J . Chem. 1983,61, 888. (b) Kamat, P. V.; Fox, M. A. J . Am. Chem. SOC.1984,106, 1191. (c) Kamat, P. V.; Fox, M. A . J . Electrochem. Soc. 1984,131, 1032. (d) Kamat, P. V. J . Electroanal. Chem. Interfacial Electrochem. 1984, 163, 389. (e) Kamat, P. V.; Basheer, R.; Fox, M. A. Macromolecules 1985, 18, 1366. (5) (a) Tributsch, H.; Calvin, M. Photochem. Photobiol. 1971, 14, 95. (b) Fong, F. K.; Winograd, N. J . Am. Chem. SOC.1976,98,2287. (c) Fong, F. K.; Galloway, L. J. Am. Chem. SOC.1976, 100, 3594. (6) Spitler, M.; Parkinson, B. A. Langmuir 1986, 2, 549. (7) (a) DeSilvestro, J.; Gratzel, M.; Kavan, L.; Moser, J. J . Am. Chem. SOC.1985, 107, 2988. (b) Vrachnou, Ersi; Vlachopoulos, N.; Gratzel, M. J . Chem. SOC.,Chem. Commun. 1987, 868. ( 8 ) (a) Moser, J.; Gratzel, M. J . Am. Chem. SOC.1984, 106, 6557. (b) Moser, J.; Gratzel, M.; Sharma, D. K.; Serpone, N. Helu. Chim. Acta 1985, 68, 168. (c) Arbour, C.; Sharma, D. K.; Langford, C. H. J . Chem. SOC., Chem. Commun. 1987, 12, 917. (9) Kamat, P. V.; Fox, M. A. Chem. Phys. Lett. 1983, 102, 379. (10) Kamat, P. V.; Chauvet, J.-P.; Fessenden, R. W. J . Phys. Chem. 1986, 90, 1389. (1 1) Kalyansundaram, K.; Vlachopoulos, N.; Krishnan, V.; Monnier, A.; Griitzel, M. J . Phys. Chem. 1987, 91, 2342. (12) Rossetti, R.; Brus, L. E. J . Am. Chem. SOC.1984, 106, 4336. (13) (a) Kamat, P. V.; Gopidas, K. R.; Weir, D. Chem. Phys. Lett. 1988, 149, 491. (b) Fessenden, R. W.; Kamat, P. V. Chem. Phys. Lett. 1986, 123, 233.
0022-3654/89/2093-0859$01.50/0
to the conduction band of the semiconductor. Photoactive compounds such as dyes (for example, erythrosin B? eosin,8chlorophyllin,I0 phthalocyanine^,'^ and R ~ ( b p y ) and ~~+ its a n a l o g ~ e s ' ~with ) , high extinction coefficients in the visible region, are often employed in the modification of semiconductor electrodes and particles. One of the important criteria for an efficient photosensitization is to adsorb these dyes on the semiconductor surface with an electrostatic, hydrophobic, or chemical and Kalyansundaram et interaction. It has been shown by us9*10 al." that photoactive compounds with a carboxylate group as the substituent can interact with the TiOz surface and facilitate the charge injection into the conduction band of the semiconductor. Interaction between the sensitizer and the semiconductor surface can be. probed with the electronic absorption and emission spectra of the sensitizer as the energetics of the ground and excited states are altered. These spectral changes include displacement or broadening of the absorption and emission spectra and changes in the extinction coefficient of absorption. Changes in the excited-state lifetimes also provide important information regarding the kinetics and mechanism of the charge injection process. In this study we have chosen anthracene-9-carboxylic acid (9AC) as the sensitizer, since its association with T i 0 2 colloids 450 nm). 9AC is extended the absorption into the visible (A also a good candidate for probing the charge injection process, as its photophysical properties are similar to those of anthracene itself and its fluorescence emission and transient characterization are intrinsically readily observable.1618 It will be of interest to see how the charge injected into the semiconductor can be utilized to reduce another substrate (Scheme I). In order to demonstrate the feasibility of such an electron-transfer process, an effort has
-
SCHEME I
(14) (a) Fan, F. R. F.; Bard, A. J. J . Am. Chem. SOC.1979, 101, 6139. (b) Giraudeau, A.; Fan, F. R. F.; Bard, A. J. J . Am. Chem. SOC.1980, 102, 5137. (15) (a) Gosh, P. K.; Spiro, T. G. J. Am. Chem. Soc. 1980,102,5543. (b) Borgarello, E.; Kiwi, J.; Pelizetti, E.; Visca, M.; Gratzel, M. J . Am. Chem. Soc. 1981, 103, 6324. (c) Hashimoto, K.; Kawai, T.; Sakata, T. Nouu. J . Chim. 1983, 7, 249. (d) Kiwi, J. Chem. Phys. Lett. 1981, 83, 594. (16) Hirayama, S. J . Chem. SOC.,Faraday Trans. 1 1982, 78, 2411. (17) Masnovi, J. M.; Seddon, E. D.; Kochi, J. K. Can. J . Chem. 1984.62, 2552. (18) Kamat, P. V.; Ford, W. E. Chem. Phys. Lett. 1987, 135, 421.
0 1989 American Chemical Society
860
Kamat
The Journal of Physical Chemistry, Vol. 93, No. 2, 1989 ._
c;-..
- 0.4
0.8
-
- 0.3 -0.2
-
- 0.1 300
350
400
450
D m v)
0
a
m D
z 0
m
500
WAVELENGTH(nm)
Figure 1. Absorption spectra in acetonitrile: (a) 1 mM colloidal Ti02 (reference: CH3CN), (b) 50 pM 9AC (reference: CH3CN),and (e) 50 pM 9AC + 1 mM TiOz (reference: 1 mM Ti02 in CH3CN).
been made to carry out the photosensitized reduction of N,N,N',N'-tetraethyloxonine (0x725) in a colloidal T i 0 2 system. Results of a time-resolved laser flash photolysis study that elucidate various steps of a Ti02-mediated charge-transfer process are described here. An effort has been made for the first time to probe the reaction of a sensitized colloidal semiconductor with an electron acceptor and evaluate the factors that influence the photosensitized reduction process.
Experimental Section Materials. Titanium(1V) 2-propoxide (Alfa), oxazine 725 (N,N,N',N'-tetraethyloxonine, Exciton Laser grade), acetonitrile (Aldrich, gold label), and anthracene-9-carboxylic acid (9AC, Aldrich) were used. 9AC was purified by recrystallization from toluene-ethyl acetate. Suspensions of colloidal Ti02 were prepared by the hydrolysis of titanium(1V) 2 - p r o p o ~ i d e . ~ ,The ' ~ stock solution (5 X lov3M) was diluted with acetonitrile to obtain the desired concentration of Ti02colloid. The average particle diameter, as measured from scanning electron micrographs, was -300 A. The surface area of these spherical Ti02 particles corresponded to 2.8 X m2/particle or 4.2 X lo3 m2/mol of TiO2.I8 Apparatus. Absorption spectra were recorded with a PerkinElmer 3840 diode array spectrophotometer, and emission spectra were recorded with a S L M photon-counting fluorescence spectrometer in a right-angle viewing mode. Flash photolysis experiments were performed with a 355-nm laser pulse (8 mJ, pulse width 6 ns) from a Quanta-Ray Nd:YAG laser system. The details of the flash photolysis setup are described e1~ewhere.l~The experiments were performed in a rectangular quartz cell (6-mm path length along the path of the monitoring light). A typical experiment consisted of a series of 10-20 replicate shots per single measurement, and the average signal was processed with a PDP-11/55 computer. All solutions were deaerated with Ar. Fluorescence lifetime measurements were performed by the time-correlated single-photon-counting techniqueZo using the apparatus that has been described elsewhere2' except that the excitation source was a mode-locked, Q-switched Quantronix 416 Nd:YAG laser system, which provided 80-ps laser pulses of 355 nm with a frequency of 5 kHz and an integrated power of about 10 mW. Interference filters were placed between the monochromator and photomultiplier to filter out scattered light. The instrument response function (excitation profile) was obtained with the monochromator set for detection at 340 nm. Lifetimes were calculated by reiterative least-squares fitting with variable zero time shift between the instrument response function and measured decay curve.21 Steady-state photolysis was performed with a medium-pressure mercury lamp (Bausch and Lomb SP-200),which was coupled with a Bausch and Lomb monochromator (33-86-07) and Corning (19) Das, P. K.; Encinas, M. V.; Small, R. D., Jr.; Scaiano, J. C. J . Am. Chem. SOC.1979, 101, 6965. (20) OConnor, D. V.; Phillips, D. Time Correlofed Single Photon Counting, Academic: London, 1984. (21) Federici, J.; Helman, W. P.; Hug, G . L.; Kane, C.; Patterson, L. K. J . Phys. Chem. 1985, 89, 1202.
WAVELENGTH (nm)
Figure 2. Fluorescence emission spectra of 20 pM 9AC in acetonitrile at various concentrationsof TiOz: (a) 0 M, (b) 0.05 mM, (c) 0.1 mM, (d) 0.25 mM, (e) O S mM, and (f) 1 mM. The excitation wavelength was at 355 nm, and the spectra were corrected for the photomultiplier response. Insert shows the dependence of I/(@ - &,M)on the reciprocal concentration of Ti02 (see text for details).
filters. Photolysis was carried out in a 1-cm cuvette that had the provision for deaeration. Analysis. The quantum yield of 0x725 reduction was measured from the flash photolysis experiment with a 355-nm laser pulse as the excitation source and anthracene triplet as reference.22 The concentration of semireduced 0x725 (Ox-) was determined from the bleaching at 640 nm (t = 1.2 X lo5 M-' cm-' 1.
Results and Discussion Absorption Characteristics of the 9AC-Ti02 System. Interaction with the surface hydroxyl groups of colloidal T i 0 2 led to broadening of the absorption bands of 9AC. The absorption spectra of 9AC in acetonitrile, recorded in the presence and in the absence of colloidal Ti02, are shown in Figure 1. An extended absorption into the visible region (up to X = 450 nm) made 9AC bound to Ti02 a good candidate for the purpose of sensitization. The absorption band in the visible (390-420 nm) which corresponds mainly to 9AC associated with T i 0 2 was used to determine the association constant. The apparent association constant as determined by the Benesi-Hildebrand method was 6000 M-'. Details of the interaction between 9AC and Ti02and its influence on the ground- and excited-state spectra have been discussed earlier.18 Fluorescence Quenching by Colloidal T i 4 . Addition of Ti02 colloids to a solution of 9AC resulted in the quenching of its fluorescence emission. Figure 2 shows the effect of increasing the concentration of T i 0 2 colloids on the fluorescence emission spectrum of 9AC. Nearly 90% of the emission of 20 p M 9AC can be quenched with 1 mM TiO2. This quenching behavior is similar to the previously reported fluorescence quenching of dyes like erythrosin B,9 eosin,8 and chlorophyllin'O and is attributed to the charge injection from the excited singlet of 9AC to the conduction band of Ti02 (reaction 1). The oxidation potential 9AC*(SI) + Ti02 9AC" + Ti02(e),,,, (1)
-
of 9AC*(SI), which is around -1.8 V vs and the energy level of the conduction band of TiO,, which lies around -0.5 V, in acetonitrile provide favorable energetics for such a charge injection process. The decrease of fluorescence yield in colloidal semiconductor suspensions has also been attributed to concentration quenching among the adsorbed dye If such a possibility exists in the present case, the fluorescence emission should have recovered at low surface coverages. The fact that the fluorescence yield (22) Amand, B.; Bensasson, R. Chem. Phys. Lett. 1975, 34, 44. (23) Kamat, P. V . , unpublished results.
The Journal of Physical Chemistry, Vol. 93, No. 2, 1989 861
Photosensitized Reduction in a TiOz System
IO01
0
1
'
'
IO
'
I
20
'
I
30
'
'
40
1
TIME (ns) Figure 3. Fluorescence decay and normalized instrument response curves for a deaerated sample of 10 p M 9AC in 1 m M Ti02/CH3CN suspension at 298 K. Excitation was at 355 nm, and the emission was recorded at 460 nm. Solid line is a calculated decay curve derived from nonlinear least-squares fit to a two-exponential decay law with the following pa= 9.13 ns, and reduced rameters: al = 0.033, ll = 1.71 ns, u2 = 0.020, i2 xz = 1.08.
decreases continuously with increasing concentration of T i 0 2 colloid suggests a negligible contribution from concentration quenching to the observed fluorescence quenching of 9AC*(SI). The participation of T i 0 2 in the quenching process was further analyzed by considering the equilibrium between adsorbed and unadsorbed molecules of sensitizer with an apparent association constant of Kapp(reaction 2). As shown earlier,I0 the observed 9AC
Kw + Ti02e [9AC-- -TiOz]
(2) quantum yield, &(obsd), of the sensitizer in a colloidal T i 0 2 suspension can be related to the fluoresence yields of unadsorbed (&") and adsorbed (4;) molecules of the sensitizer by the equation
+
&(obsd) = (1 - a)@ a&' (3) where a is the degree of association between T i 0 2 and 9AC. At relatively high TiOz concentrations a can be equated to (Ka ,[TiOZ])/(1 + Kapp[Ti02]).Equation 3 could then be simplified to 1 1 =- l + (4) 4? - +dobsd) +? Kapp(+? - + [Ti021 i)
+;
If the observed quenching is due to the association of 9AC with T i 0 2 colloid, one would expect a linear dependence of 1/(@ &(obsd)) on the reciprocal concentration of Ti02colloid with an intercept equal to I/(&" - 4;) and a slope equal to l/(4? q$')Kaw Indeed the linearity of the double reciprocal plot shown in the insert of Figure 2 confirms this behavior. The values of Kappand 4; as determined from this plot were 6450 M-I and 0.01, respectively. This value of Kappmatched well with the value of 6000 M-I, determined independently by the Benesi-Hildebrand method. The good agreement between these values of Kapp highlighted the validity of the assumption proposed for the association between 9AC and TiOz colloid. The large value of Kapp indicates a strong interaction between the sensitizer and the semiconductor colloid, which is necessary for observing an efficient charge injection. The fluorescence quantum yield of 9AC*(SI) adsorbed on Ti02 colloid (&' = 0.01) was considerably lower than that of 9AC*(S1) in acetonitrile (4f' = 0.26). Since no marked increase in the triplet yield of 9AC was observed in the adsorbed state,I8 one can attribute the singlet-state quenching as due to the charge injection process. If we express the net quenching efficiency (a) as equal t o (@ - &')/&", one can obtain an upper limit for the sensitization
Figure 4. Transient absorption spectra observed upon 355-nm laser pulse excitation of 0.2 mM 9AC in acetonitrile containing 1.5 mM T i 0 2 colloid. The spectra were recorded l a ) (0)immediately and (b) (e) 200 ns after the laser flash excitation. Insert shows the absorption profiles at 720 nm recorded at two different time intervals.
efficiency. An upper limit of 94% was obtained for the sensitization of T i 0 2 in these experiments. Fluorescence Lifetime Measurements. It has been shown earlier9J0 that the sensitizer molecules adsorbed on the T i 0 2 surface had a significantly shorter excited singlet lifetime than in homogeneous solution and this decrease in lifetime could be correlated with the charge injection process. In neat acetonitrile, the fluorescence of 9AC had a single-exponential decay with a lifetime of 8 ns. However, in a TiOz suspension, the fluorescence emission of 9AC deviated from a single-exponential decay. The fluorescence decay for 9AC in 1 mM T i 0 2 suspension and the calculated decay curve (fitted to a two-exponential decay) are shown in Figure 3. The fluorescence lifetimes of the two components attributed to 9AC adsorbed on Ti02 colloid and unadsorbed 9AC present in acetonitrile were 1.7 f 0.1 and 9.1 f 0.2 ns, respectively. If we assume the observed decrease in fluorescence lifetime is entirely due to the electron injection process (reaction 1) and the other radiative and nonradiative decay processes of 9AC associated with Ti02 colloid occur a t the same rate as in neat solvent, one could correlate the observed lifetimes by the following expression.I0 where r and radsare the lifetimes of the sensitizer in acetonitrile and adsorbed on to the TiOz surface and k,, is the specific rate of the charge injection process. The value of ke:4 obtained upon substitution of the values of r (9.1 ns) and l a d s (1.7 ns) in eq 5 was 4.8 X lo8 s-l. This value of k,, is close to the value reported for the eosin Y-colloidal T i 0 2 system in water (ket = 8.5 X lo8 but is an order of magnitude smaller than the value reported for the chlorophyllin-colloidal Ti02 system in acetonitrile. The variation in the solvent environment and the energetics of the excited sensitizer could influence the specific rate of the charge injection process. Laser Flash Photolysis Studies To Probe the Charge Injection Process. Time-resolved laser flash photolysis is very useful in the investigation of the interfacial charge-transfer processes in colloidal semiconductor systems."" The role of 9 A C * in sensitizing Ti02 colloids was further elucidated by recording transient absorption at different time intervals. Figure 4 shows the transient absorption (24) A word of caution regarding the estimation of ke,: The decrease in the fluorescence lifetime is only by a factor of 5 while the decrease in the fluorescence quantum yield of 9AC is by a factor of 25 when adsorbed on Ti02 colloid. This indicates that a faster quenching process may be present in the deactivation of the excited singlet state. The present experimental setup does not permit determination of lifetimes shorter than 500 ps.
862 The Journal of Physical Chemistry, Vol. 93, No. 2, 1989
Kamat
spectra obtained upon 355-nm laser pulse excitation of 0.2 mM 9AC in a 1.5 mM Ti02 suspension. The absorption of Ti02 colloid at 355 nm was negligible, and it did not interfere with the process of 9AC excitation. The transient absorption spectrum recorded immediately after the laser pulse excitation exhibited maxima at 428 and -720 nm. The absorption at 428 nm was due to the triplet excited state of 9AC.18 The absorption band at 720 nm was similar to the absorption of the radical cation of anthracene derivatives, which were generated by electrochemical oxidation" and by y-irradiationZs methods. Deactivation of the singlet and triplet excited states follows several pathways, as presented by reactions 6-1 1. Both singlet 9AC*(SI)
-
9AC(So)
9AC*(SI) 9AC*(SI)
+ Ti02
9AC*(TI) 9AC*(TI)
+ Ti02
9AC*(TI)
--
9AC*(Si)
+
+h~'
9AC(So)
9AC+'
+ Ti02(e),,,
9AC(So)
9AC"
+ Ti02(e),,,,
(6) (7) (8) (9) (10) (1 1)
and triplet excited states (reactions 9 and 11, respectively) can participate in the charge injection process, to yield 9AC". However, the participation of 9AC*(TI) in the charge injection can be ruled out on the basis of the following observations: (i) The formation of 9AC+' was prompt and was completed within the duration of the pulse. A slower growth that could correspond to the decay of 9AC*(TI) could not be detected. (ii) The lifetime of 9AC*(T1) (T = 200 ps) was not very much affected by the presence of Ti02 (T = 142 ps). The small decrease in triplet lifetime could be due to the environmental effects as a result of adsorption on the T i 0 2 surface. Changes observed in the triplet yield were also marginal.I8 Hence, 9AC+' must have originated from the excited singlet state (reaction 9). The contribution of excited-state redox processes such as excited-state annihilation and photoionization was also checked. A negligible amount of 9AC+' could be seen when an acetonitrile solution containing 9AC alone was excited with a 355-nm laser pulse. The inability of the triplet excited state to participate in the charge injection process is not unusual. Sensitizers such as eosin,8 erythrosin B,9 chlorophyllin,I0 and zinc porphyrins" are known to inject charge into T i 0 2 colloid only from the singlet excited state. The lack of a driving force for 9AC*(T1) (AI5 5 0.1 v) as compared to that for 9AC*(SI) ( A E i= 1.3 V) to inject an electron into the conduction band of the semiconductor could explain such a difference in the reactivity of the excited states in the present experiments. An important aspect of the sensitization process is the recombination between the injected charge and the cation radical of the sensitizer, which often limits the efficiency of sensitization. As can be seen from the insert in Figure 4, nearly 90% of 9AC" decayed quickly (7 = 18 ns or k = 5.5 X lo7 s-l) by a recombination process while the rest exhibited a longer lifetime. Similar interparticle and intraparticle recombination processes have been described in an earlier study.8a However, it is worth noting that about 10% of the injected charge is able to survive, possibly by being trapped within the semiconductor particle. The obvious question would then be whether such a charge would be accessible for charge transfer to another substrate. Sensitized Photoreduction of N,N,N',N'- Tetraethyloxonine. N,N,N'JV'-Tetraethyloxonine (0x725) is a good electron acceptor (l?o/o-. = -0.02 V vs N H E ) and can be used to probe the interfacial charge-transfer process in colloidal semiconductor syst e m ~ . This ~ ~ -dye ~ ~adsorbs strongly onto the surface of colloidal T i 0 2 with an apparent association constant of 50000 M-I. Since 0x725 has no absorption at 355 nm, it does not interfere with (25) Hiratsuka, H.; Tanizaki, Y . J. Phys. Chem. 1979, 83, 2501. (26) (a) Kamat, P. V. J . Chem. Soc., Faraday Trans. I , 1985, 28, 513. (b) Kamat, P. V. Langmuir, 1985, I, 608.
WAVELENGTH (nrn)
Figure 5. Transient absorption spectra observed upon the laser pulse (355 nm) excitation of 20 p M 9AC and 5 pM 0x725 (a) ( 0 )in acetonitrile and (b) (0) in acetonitrile containing 2 mM Ti02. The spectra were recorded 110 ps after the laser flash excitation. Inserts are the absorption profiles recorded at 395 and 640 nm indicating the formation of Ox-' and depletion of 0x725, respectively.
I
u)
>
>
Eo(S/Sf)
Y
w
TIO~
2.0
VB
3.0 Figure 6. Schematic diagram describing the conduction and valence band energy levels for Ti02, electron-donating energy levels for 9AC, and electron-accepting energy level for 0x725.
9AC excitation. The transient absorption spectrum recorded upon excitation (355 nm) of 9AC-Ti02 in the presence of 0x725 is shown by trace b in Figure 5. The transient spectrum recorded 110 ~s after the flash exhibited a maximum at 420 and 860 nm, which matched well with the absorption characteristics of semireduced 0 ~ 7 2 5 (reaction ,~ 12). When 9AC was excluded from TiOz(e),,,,
+ Ox
-
Ti02 + Ox-'
the system, no such reduction was seen. The laser pulse at 355 nm is not capable of inducing direct charge separation within the TiO, particle, and hence the possibility of direct participation of the TiOz colloid in the reduction process is considered to be negligible. When a solution of 9AC and 0x725 was excited with a 355-nm laser pulse, a small but different transient absorption was seen (spectrum (a) in Figure 5). This transient absorption is due to 0x725*(T1), which is formed as a result of the triplet-triplet energy-transfer process (reaction 13). Such a process can also 9AC*(TI)
+ OX
+
9AC
+ Ox*(TI)
(13)
occur on the surface of TiOz particles. But it has been shown in our previous study that the T-T energy transfer efficiency was at a maximum only when these molecules were present as a closely packed monolayer on the Ti02 particle but decreased rapidly as the coverage was decreased. The experimental conditions for recording spectrum b in Figure 5 involved a submonolayer coverage of 0x725 so that the T-T energy transfer between 9AC*(T1) and 0x725 was minimal. (Submonolayer coverages can be ob-
(27) Kamat, P. V.;Dimitrijevic,N. M.; Fessenden, R. W. J . Phys. Chem. 1987, 91, 396.
(12)
(28) Kamat, P. V.; Lichtin, N. N. Zsr. J . Chem. 1982, 22, 113.
The Journal of Physical Chemistry, Vol. 93, No. 2, 1989 863
Photosensitized Reduction in a Ti02 System 0.04
-
O
---JFGa
0
L
PA
0.8
n C
-f
II4
I
I
0 d
.- - ; - - - -
(D
Y
3,
a
Q
O
I
O
400 L
500
I
600
t. I
700
WAVELENGTH (nm)
0.0oL
0
=
I
-
-1
I
I
I
2
3
4
[TiOl]
b
,mM
Figure 7. (a) Dependence of the yield of sensitized reduction of (Ox-') on the concentration of TiOz in acetonitrile. The yield of Ox-' was monitored by the bleaching of 0x725 at 640 nm. The suspension contained (a) (0)20 p M 9AC and 5 pM 0x725 and (b) ( 0 )5 p M 0x725 (in the absence of sensitizer). Excitation was at 355 nm.
tained at Ti02 concentrations greater than 0.5 mM29.) All these experimental observations support reaction 12 as the process responsible for the reduction of 0x725. The mechanism of the sensitized reduction of 0x725 can be understood on the basis of the energy diagram in Figure 6. The energy difference between the singlet excited state of the sensitizer and the conduction band of the semiconductor (AEi= 1.3 V) and the energy difference between the conduction band of the semiconductor and the reduction potential of 0x725 (AI2 = 0.5 V) promote such an electron transfer. Alternatively, one can envisage the Ti02colloid as the mediator in promoting charge transfer from 9AC*(S1) to 0x725. Dependence of Sensitizer Reduction on T i 4 Concentration. The mediating role of the TiO, colloid in the reduction of 0x725 was further established by observing the effect of TiOz concentration on the production of Ox-'. The bleaching at 640 nm (after attaining a plateau, Le. 100-150 I.LS after the laser flash) was taken as a measure of efficiency of the photosensitized reduction process. The dependence of the absorbance change at 640 nm versus the concentration of T i 0 2is shown in Figure 7. At low concentrations of Ti02, the photosensitized reduction efficiency increased steeply as an increasing proportion of 9AC and 0x725 molecules in the system became adsorbed onto the particles. Based on the apparent association constants mentioned previously, the amount of these compounds that was adsorbed was in excess of the amounts required for monolayer coverage at Ti02 concentrations less than 0.5-1.0 mM for particles having diameters of 200-500 A, if the monolayer is assumed to be close-packed with a thickness of 10 8, and density of 1 g ~ m - ~As . most of the 9AC and 0x725 molecules became adsorbed on T i 0 2 particles at higher Ti02 concentrations, the sensitized photoreduction efficiency attained a plateau; the reduction under these experimental conditions was limited by the total amount of sensitizer present. If the observed reduction process occurred via a direct electron transfer between 9AC* and 0x725 on a T i 0 2 particle, a decrease in Ox-' should have been observed at submonolayer coverages as observed earlier for an intermolecular electron-transfer reaction on Si02/A1203 particles.3o Increasing the average distance separating the donor and acceptor molecules on Ti02particles having submonolayer coverages makes the intermolecular electron transfer less efficient. When the above experiment was repeated in the absence of 9AC, a negligibly small amount of 0x725 was reduced (trace b in Figure 7). The difference in the yield of Ox-' in these two (29) A detailed discussion on the determination of surface coverage of 9AC and 0x725 on colloidal Ti02 particles and its effect on T-T energy transfer efficiency can be found in ref 18. (30) Kamat, P. V.;Ford, W.E. J . Phys. Chem., in press.
Figure 8. Absorption spectra of 5 p M 0x725 in acetonitrile containing 40 pM 9AC and 2 mM TiOz: (a) before photolysis and (b) after 55 min of photolysis at 385 nm. (The reference solution was 40 pM 9AC and 2 mM TiOz in CH,CN.)
experiments (traces a and b) highlights the role of 9AC in the sensitized reduction of 0x725. The maximum yield of Ox-' corresponded to a quantum efficiency of 1.5%. This indicated that only 6% of the injected charge from 9AC*(S,) (& = 0.26) was utilized in the sensitization process. As discussed earlier,1q2*'0the major limiting factor for such a low efficiency is the recombination of injected charge with the radical-cation of the sensitizer. If one could scavenge these radical cations quickly from the surface of TiOz with a suitable electron donor, Pt should be possible to enhance the efficiency of the reduction process. Alternatively, one could also modify the surface of the TiO, colloid or tailor the charges on the sensitizer molecule, such that the cation radical is quickly repelled from the TiO, surface by electrostatic effects. Such a possibility has been demonstrated for the reduction of viologen compounds in S O 2 and Si02/Ti02 particle^.^'^^^ Currently, efforts are being made to employ similar approaches to improve the efficiency of sensitization. Steady-State Photolysis. If indeed, 9AC-Ti02 is responsible for the sensitized reduction of 0x725, it should be possible to observe the leucoform of the dye with continuous irradiation. The semireduced forms of oxazine dyes (Ox-') are known to undergo a disproportionation reaction to yield leuco dye (Ox2-) and the parent dye26,27(reaction 14). Ox2- is stable in an inert atmosphere such as N2 or Ar.
-
+
20x-* Ox" ox (14) The changes in the absorption spectra recorded after irradiating 9AC in a Ti02 suspension containing 0x725 with a monochromatic light (385 nm) are shown in Figure 8. The bleaching a t 640 nm and simultaneous formation of a product with an absorption at 330 nm confirmed the conversion of 0x725 to its leucoform. It was confirmed in a separate experiment that the leucoform of 0x725 had an absorption maximum around 330 nm. (0x725 can be reduced to its leucoform by bubbling H2 through a solution of 0x725 containing platinum sponge). Regeneration of 0x725 could be seen when the photolyzed product was exposed to air. The total conversion of 0x725 to its leucoform is affected by the reaction between 9AC" and Ox-', which competes with reaction 14. If a suitable sacrificial donor is employed in the present system, it should be possible to scavenge 9AC" and improve the yields of the photosensitized reduction process.33 Absorption of light by 9AC adsorbed on T i 0 2 initiated the reduction of 0x725, as the other two components, 0x725 and Ti02,exhibited negligibly small absorptions at 385 nm. Moreover, no changes in the absorption spectrum could be seen when one of the components of 9AC-Ti02-0x725 system was deliberately excluded. Sensitized photoelectrochemical reduction of 0x725 under steady-state irradiation should be of considerable interest (31) Laane, C.; Willner, I.; Otvos, J. W.; Calvin, M. Proc. Natl. Acad. Sci. U.S.A. 1981, 78, 5928. (32) Frank, A. J.; Willner, J.; Goren, Z.; Degani, Y. J. Am. Chem. SOC. 1987, 109, 3568. (33) Minimal success was achieved when EDTA was employed as a sa-
crificial donor. Because of its participation in other photochemical processes, it was not possible to carry out a quantitative study.
864
J . Phys. Chem. 1989, 93, 864-867
in the direct conversion and storage of visible light.26 Conclusions. The reduction of 0x725 has been performed on the surface of colloidal TiOz with 9AC as the sensitizer. Processes responsible for both the sensitization of T i 0 2 colloid and the reduction of an electron acceptor have been elucidated with time-resolved laser flash photolysis. The dependence of photosensitized reduction efficiency on the concentration of Ti02 clearly highlighted the role of the semiconductor colloid and the sensitizer in promoting the electron transfer in microheterogeneous systems. Current efforts are being directed toward enhancing the efficiency of sensitization by controlling electrostatic effects and using
suitable redox couples as supersensitizers.
Acknowledgment. I thank Drs. Richard W. Fessenden and William E. Ford for many helpful discussions and Dr. Maria Bohorquez for her assistance in fluorescence lifetime measurements. The research described herein was supported by the Office of Basic Energy Sciences of the Department of Energy. This is Document No. NDRL-3068 from the Notre Dame Radiation Laboratory. Registry NO. 9AC, 81503-67-5; 9AC+', 117559-18-9; OX,47367-75-9; Ox-', 62671-94-7; TiO,, 13463-67-1.
Change of Cu(I1) Cation Coordination in H-ZSM-5 Channels upon the Sorption of n-Hexane and Xenon: ESR Spectroscopic Evidence A. V. Kucherov* and A. A. Slinkin N . D.Zelinsky Institute of Organic Chemistry, USSR Academy of Sciences, Moscow. USSR (Received: February 18, 1988; In Final Form: June 27, 1988)
Physical sorption of n-hexane or xenon, at 20 " C , inside CuH-ZSM-5 channels is accompanied by noticeable changes in the hyperfine structure of the Cu2+ESR signal from isolated coordinatively unsaturated Cu(I1) cations. Such influence of nonlocalized physical adsorption of inert gas atoms or alkane molecules may be explained by a slight geometrical displacement of the five-coordinated Cu(I1) ion due to dispersion forces. Some changes in the Fe3+ESR spectrum upon the interaction of Fe(II1) cations and n-hexane molecules inside H-ZSM-5 channels also may be indicative of a slight displacement of Fe(lI1) ion located in a strong crystal field of low symmetry.
Introduction The location and reactivity of Cu(II) cations in H-ZSM-5 and H-mordenite have been stuhied previously' by the electron spin resonance (ESR) technique. ESR spectroscopy is of great value for the investigation of copper(I1)-containing zeolites, since the ESR parameters of Cu2+ions depend strongly on the local crystal field and allow the coordination of isolated cations to be determined. We showed that CuH-ZSM-5 calcined at 800 OC is a very interesting and quite simple system containing two discrete types of isolated coordinatively unsaturated Cu(I1) cations. The coordination of these ions may be treated as a square pyramid distorted to a planar square, the distortion for both types of sites being significant. All these cations are accessible for gas-phase molecules which can enter into zeolite channels (elliptic channels with a cross section 5.2 X 5.8 A, intersected with crisscross channels having 5.5 8, cross section). Oxygen sorption leads to the sharp broadening of hyperfine structure (hfs) lines in ESR spectra, caused by dipole-dipole interaction between O2 molecules and Cu(I1) cations.' It is not surprising that the interaction of CuH-ZSM-5 with NH3, pyridine, H,O, and olefins changes drastically the hfs of Cu(I1) ESR spectra.2 The above molecules may be treated as additional ligands for coordinatively unsaturated Cu(I1) cations. However, the sorption of such saturated nonpolar compounds as n-hexane also provokes a noticeable change in the hfs of the Cu(I1) ESR spectrum. The possibility of alkane molecule interaction with coordinatively unsaturated Cu( 11) ion is of interest. This paper describes a detailed ESR study of n-hexane, argon, and xenon interaction with CuH-ZSM-5. In addition, the interaction of the above substances with coordina(1) Kucherov, A. V.; Slinkin, A. A.; Kondratyev, D. A,; Bondarenko, T. N.; Rubinstein, A. M.; Minachev, Kh. M. Zeolites 1985, 5, 320; Kinetic. Katal. 1985, 26, 409. Kucherov, A. V.; Slinkin, A. A. Zeolites 1986, 6, 175. (2) Slinkin, A. A,; Kucherov, A. V.; Nikishenko, S . B., to be published.
tively unsaturated Cr(V) and Fe(II1) cations inside H-ZSM-5 is studied. Experimental Section The ESR spectra, at 20 and -196 OC, were taken on a reflecting spectrometer (A = 3.2 cm), equipped with a magnetometer, with diphenylpicrylhydrazyl (DPPH) as a standard. The sample CuH-ZSM-5, containing 0.6 wt % Cu (-20% of H+ exchanged for Cu2+),was prepared by a threefold exchange of NH,-ZSM-5 (Si02/A1203= 69) with 1 N Cu(N03)*solutions and calcined in an air stream at 520-550 "C for 5 h. The quartz ampule, with the sample, was calcined in air at 800 O C for 2 h, placed in the spectrometer probe, and soldered to an adsorption system which permitted the evacuation of the sample and the introduction of the substances from the gas phase at temperatures from 20 to 400 "C. Prior to the inlet of adsorbates the following sample treatment in the spectrometer probe was carried out to remove water: (1) heating to 400 "C and evacuation for 30 min; (2) introduction of pure O2at 400 "C, heating for I O min and cooling to 20 "C; (3) evacuation at 20 "C for 30 min. Xenon and n-hexane (chemically pure grade) were purified to eliminate traces of moisture and oxygen by repeated freezing in liquid nitrogen and evacuation to 10" mmHg. Samples of H-ZSM-5 containing isolated Cu(II), Cr(V), and Fe(II1) cations were prepared by a solid-state reaction between H-ZSM-5 and CuO, Cr03, or FeC1, according to ref 1, 3, and 4. Results and Discussion Figure 1 shows the perpendicular components of Cu(I1) ESR spectrum of parent CuH-ZSM-5 calcined at 800 "C and evac(3) Kucherov, A. V.; Slinkin, A. A. Zeolites 1987, 7, 3 8 . (4) Kucherov, A. V.; Slinkin, A. A. Zeolites 1988, 8, 110.
0022-365418912093-0864$0 1.5010 0 1989 American Chemical Society