J . Phys. Chem. 1986, 90, 2156-2159
2156
(k22) and the outer monolayer of polymer redox molecules (kll), and Z is the bimolecular collision frequency, typically 10" 12321-23
M-I SKI.It may be seen from the table that while the measured k,'s for [(BPQ2'),],,,f and the series of chemically similar Cr(II1) polypyridyl and 1, IO-phenanthroline complexes vary approximately as K112,as expected from Marcus theory, the absolute values are lower than those calculated from eq 13. The fundamental reason for this discrepancy is not clear. It should be understood that the value of J? in eq 1 1 may be the source of the difficulty, since this value cannot be directly measured. Also, there may be an error in the value of k l l used in the calculation. Values of k , , reflect the electron self-exchange rate constants for the surface-bound ~ ~such, ~ ~ ' the ~ po1yme1-s~~ calculated from polymer Dct v a l ~ e s . As kl I values are bulk properties of the polymer. The bulk k , value may differ somewhat from the surface k l l ,which is the important parameter in these measurements. It seems unlikely, however, that a large difference between bulk and surface k , , is manifested by the data. The two would have to differ by a factor of approximately lo3 to account for the discrepancy between the
,
(21) (a) Swickel, A. M.; Taube, H. Discuss. Furaduy SOC.1960, 29, 42. (b) Meyer, T. J.; Taube, H. Inorg. Chem. 1968, 7, 2369. (22) (a) Ferraudi, G. J.; Endicott, J. F. Inorg. Chim. Acta 1979, 37, 219. (b) Ferraudi, G. J.; Endicott, J. F. J . Am. Chem. SOC.1977, 99, 5812. (23) Brunschwig, B.; Endicott, J. F., quoted in Brunschwig, B.; Sutin, N . J . Am. Chem. SOC.1978, 100, 7568.
calculated and observed rates, and the calculated k,, for reaction according to eq 12 would then be too low. The results of Murray and co-workers,' who have studied electron transfer at polymer/solution interfaces, are consistent with only slight differences between bulk and surface k l l values. Whatever the reason for this systematic trend toward low values of observed k,, relative to calculated values, the differences are modest and the data are roughly in accord with the cross-relation. Such agreements have previously been noted.7 The important finding of this work is that experimental data for thermodynamically favored polymer-solution couple cross-reactions fit closely the theory represented in eq 1 and 11 and that fast interfacial electron-transfer rate constants may be extracted from these data. This work will be extended, in future studies, to problems in electrocatalysis at polymermodified electrodes.
Acknowledgment. We thank the United States Department of Energy, Office of Basic Energy Sciences, Division of Chemical Sciences, for support of this research. Use of facilities in the M.I.T. Center for Materials Science and Engineering supported by the National Science Foundation Materials Research Laboratory Program is also gratefully acknowledged. Registry No. I, 87698-68-8; 11, 74173-49-2; 111, 97551-35-4; Pt, 7440-06-4; SnO,, 18282-10-5; Co(bpy),'+, 19052-39-2; Ru(NH&,+, 18943-33-4; Cr(trpy),'+, 54984-99-5; Cr(~hen),~+, 15276-16-1; Cr(5,6Me,~hen),~+,5 1 261-67-7.
Quenching of Dye-Sensitized Photocurrents at ZnO Semiconductor Electrodes Mark T. Spitled Photoconversion Research Branch, Solar Energy Research Institute, Golden, Colorado 80401 (Received: September 19, 1985; In Final Form: January 10, 1986)
With the use of ZnO single-crystal electrodes in an attenuated total reflection arrangement, it has been confirmed that the low quantum efficiency 9 of photocurrents sensitized by rhodamine B and rose bengal is attributable to a fast back-reaction between the oxidized dye and electrons in surface traps. This conclusion is based upon the observation that is a constant function of electrode surface coverage 0 down to 0 = 0.001 where energy-transfer quenching is negligible. Two different types of surface sites are found, one which leads to an efficient sensitization of photocurrent by the adsorbed dye and another which leads to the rapid back-reaction. These sites are distinguishable by their different heats of adsorption.
The highly exoergic photooxidation of organic dyes such as rhodamine B and rose bengal at semiconductor electrodes proceeds through the injection of an electron by the excited dye into the semiconductor conduction band.' At single-crystal surfaces of ZnO, Ti02, and S r T i 0 3 electrodes, however, the quantum efficiency Q, for current production by these dyes through this process remains low, about where Q, is defined as the number of electrons measured as current per photon absorbed by the adsorbed dye.2-6 This low efficiency is surprising since electron-transfer theory would predict a Q, approaching unity for the highly exoergic photooxidation of the Two causes for this low Q, have been discussed in the literat ~ r e . ~ , 'One , ~ attributes the low yield of the current to a fast back-reaction at the surface between the injected electron and the oxidized parent molecule before the charge penetrates to the semiconductor bulk. This is given in the following equations for the excited dye D*: D*
-
D+
+ e-
oxidation
(1)
Present address: The Polaroid Corporation, 103 Fourth Avenue, Waltham. MA 02254.
0022-3654/86/2090-2156$01.50/0
D+ + e-
-
D
recombination
(2)
The other supposes a fast quenching of the excited dye at the surface through a rapid energy transfer between the adsorbed molecules. As long as the neighboring dyes are within 50-60 A, the excitation finds its way to a molecule where it is quickly dissipated by a quencher Q at a rate which is fast in relation to electron transfer to the electrode of eq 1: D*
+D
-
D
+ D*
energy transfer
(3)
(I)Gerischer, H.; Willig, F. In Topics in Current Chemisfry, Vol. 61, Davison, A., Ed.; Springer: New York, 1976; p 31. (2) Kavassalis, C.; Spitler, M. T. J . Phys. Chem. 1983, 87, 3166. (3) Spitler, M. T.; Calvin, M. J . Chem. Phys. 1977, 66, 4294. (4) Spitler, M. T.;Calvin, M. J . Chem. Phys. 1977, 67, 5193. ( 5 ) Spitler, M. T.; Lubke, M.; Gerischer, H. Ber. Bunsenges. Phys. Chem. 1979, 83, 663. (6) Sonntag, L. P.; Spitler, M. T. J . Phys. Chem. 1985, 85, 1453. (7) Gerischer, H.; Spitler, M. T.; Willig, F. In Electrode Processes 1979, Bruckenstein, S., Ed.; Electrochemical Society: Princeton, NJ, 1980; p 1 1 5 . (8) Arden, W.; Fromherz, P.J . Electrochem. SOC.1980, 127, 370. (9) Willig, F. In Advances in EIectrochemical Engineering, Vol. 12, Gerischer, H., Tobias, C., Eds.; Wiley: New York, 1981, p 1.
0 1986 American Chemical Society
The Journal of Physical Chemistry, Vol. 90, No. 10, 1986 2157
Quenching of Dye-Sensitized Photocurrents D*
+Q
-
D
+Q
quenching
Surface coverage 8
(4)
In this work the role of the energy transfer of eq 3 and 4 between molecules in the adsorbed dye layer has been determined at single-crystal surfaces by measuring @ as a function of the surface coverage of the electrode by the adsorbed dye. As the surface concentration of the dye becomes more dilute, the intermolecular distance should become large enough to preclude energy transfer and its role in reducing @ should be revealed. This experiment was performed with the attenuated total reflection of a laser beam (laser ATR) within a ZnO single crystal which serves as both the internal reflection element and the semiconductor electr~de.~” This technique is known to be very sensitivelo and has been shown in this case to be capable of measuring the dye down to 0.1% of a monolayer. As this corresponds to an intermolecular separation of greater than 350 A the role of energy transfer in determining 0 should become apparent. Materials and Methods The attenuated total reflection technique used in these experiments has been described in detail previously.2 In brief this method permits the measurement of absorption of light by the adsorbed dye layer through the use of internal reflection of a laser beam within an specially shaped semiconductor electrode. In this case ZnO single crystals were ground and polished to allow five intersections of the laser beam with the electroactive surface. The attenuation of the light intensity through absorption by the adsorbed dye gave the number of photons absorbed; when these data were compared with the current produced by this absorption, @ could be determined. The ZnO single crystals used in this work were obtained from Dr. R. Helbig of the University of Nuremberg-Erlengen in the form of needle prisms with cross sections of 0.5 to 1.O mm. The entrance and exit surfaces for the laser beam were cut at a 4 5 O angle to the (1010) prismatic face selected for use as the electrode surface. The quality of the prismatic faces was such that they could be used for these internal reflection experiments as they were grown. Between experiments the electrodes were rinsed in boiling methanol to remove the adsorbed dye. After about ten measurements, the crystals became too contaminated for use. For all of these experiments s-polarized light at 560 nm was used. The light intensity was kept below 5 mW with a spot size on the crystal of 150 pm in diameter. The electrochemical setup involved the use of a three-electrode cell with the ZnO serving as the anode. A bias was maintained during the experiments of +0.25 V vs. S C E where the potential dependence of the photocurrent reaches a plateau. The cell was constructed so that new electrolyte and dye-containing electrolyte could be introduced in a continual flow across the electrode surface. Both rhodamine B and rose bengal were used in this work. Laser grade rhodamine B obtained was used as purchased. The rose bengal was obtained from Aldrich and checked for purity by TLC. Results @ was determined as a function of the amount of adsorbed dye through the use of the flow cell arrangement described above. Low concentration solutions of the dyes in 0.1 M KCl were introduced to the cell containing only the 0.1 M KCl electrolyte. Measurements of current and light absorption were made as this dye slowly diffused to the electrode and adsorbed to the surface. The average experiment required about 15 min. A result typical for rose bengal is shown in Figure 1 for a 3 X lo-’ M solution of the dye. The calculated @ is given as a function of the percent of light absorption per monolayer of adsorbed dye up to 0.5%. These results indicate that there is no dependence of 0 on the surface concentration of the dye down to the coverage represented by a 0.01% absorption. The error bars represent the uncertainty in @ caused by the uncertainty in the (IO) Harrick, N. J. Internal Reflection Spectroscopy; Interscience: New York, 1967.
I
0.0
Rose Bengal 3 x 107M 0.1 M KCI
-$
0.01 0.0
I
I
I
I
I
0.1
0.2
0.3
0.4
0.5
Absorption (%)
,
Figure 1. The quantum efficiency for current production @ by rose bengal at ZnO is shown to be a constant function of the amount of light absorbed by the adsorbed dye down to a surface coverage 0 of 0.001. In this experiment, current and absorption data were recorded over a IO-min period as dye adsorbed from a 3 X lo-’ M solution of the dye.
Surface coverage 8 5,li
.0,2
.;4
Rhodamine B 1.2 x 10-6 M 0.1 M KCI
-
0.0 I 0.00
I
I
0.25
0.50
I
0.75
1 1.00
Absorption ( O h ) Figure 2. In an experiment similar to that in Figure 1,
for rhodamine B at ZnO was found to remain relatively constant as a function of 0 to 0 = 0.0025.
absorption measurement and can be seen to decline as this measurement becomes more accurate. A similar experiment for rhodamine B is given in Figure 2 where M solution was used. The results are not much a 1.2 X different from those for rose bengal. @ remains constant down to a dye coverage which only absorbs 0.05% of the incident light. It was observed in the experiments of Figures 1 and 2 that rose bengal was found to adsorb to ZnO much more strongly than rhodamine B with almost five times more rhodamine B required than rose bengal to obtain the same surface coverage. As is evident in the data of Figure 3 for rose bengal, the current declines sharply when the interface is pulsed with high-intensity light; at the same time the absorption of the dye layer changes to a much lesser extent. The decay of the absorption signal does correlate 1:l with the photocurrent decay if one considers only the absorption signal above the dashed line of Figure 3 and amplifies this by a factor of 7.5, that is, one concerns oneself with the top 13.3% of the absorption signal. This dashed line divides the absorption signal, and thereby the dye population, into two categories, one which is responsible for the photocurrent, and one which is not. The location of the dashed line on the ordinate of the absorption signal has been selected to give the best fit of the amplified absorption decay above the line to the decay of the photocurrent. The photocurrent decay is a complicated function of the light intensity, @, and the adsorption kinetics of the dye.4 Therefore it is unlikely that this linear relationship between the photocurrent decay and the absorption would be coincidental. This
2158
The Journal of Physical Chemistry, Vol. 90, No. 10, 1986 . RO.
8.np.l
1 a I lo-'
*.O
M
I.
0 1 M KCI 0 3 V
Spitler
I
AO/AOCI
X?
8.4%
0.0%
PHOTOCURRENT
ABSORPTION
Figure 3. Photocurrent and absorption signals are shown here which have been simultaneously recorded with the ATR arrangement. With 10 mW of illumination at 560 nm, the photocurrent is seen to decay much faster than the absorption signal. However, if only that portion of the absorption signal above the dashed line is considered, it correlates one-toone with the photocurrent when amplified by a factor of 7.5 (see inset).
correlation is the simplest interpretation of these data and it implies that only 13.3% of the adsorbed dye is responsible for the photocurrent. The remainder appears to produce little or no current. Similar experiments for rhodamine B gave a figure of 9.l(f0.2)% for the fraction of the adsorbed dye yielding photocurrent. If these two different dye populations were adsorbed at different sites on the surface, and these two sites had different heats of adsorption, there would be a correlation between the heats of adsorption, AHads,at these sites and the CP for the two forms of the adsorbed dye. Given an electrode with a full surface complement of the dye, those molecules with the lowest AHadswill wash off first when the surface is rinsed with electrolyte containing no dye. If CP is measured as the dye desorbs during such a wash, a qualitative relation between CP and AHadscould be established to determine which form of the dye, efficient or inefficient, is more strongly adsorbed to the surface. Such an experiment was performed. Dye was quickly adsorbed onto the surface with the introduction of a concentrated dye solution into the cell and then washed from the surface with a steady flow of electrolyte containing no dye. During the desorption process, CP was measured. Typical results are given in Figure 4 where CP for rhodamine B falls quickly as the surface concentration declines and CP for rose bengal declines less rapidly. The extent and degree of this decline is dependent on the age and quality of the sample used. Unfortunately, there was too little electrode material availble for a more detailed examination of this behavior.
Discussion For rhodamine B it has been estimated that 5 5 A is the critical intermolecular separation in a two-dimensional system at which the rate of energy transfer between dye molecules falls below the rate of radiative decay of the individual This corresponds to a surface coverage of 0.06. If energy transfer were to play a role in the quenching of the photocurrent, @ would be expected to increase once 6 fell below this value. In order to compare this estimate with the results of Figures 1 and 2, the absorption signal of the adsorbed dye must be transformed into the surface coverage 6. Two additional data are required. One is the measurement that a monolayer of rhodamine B adsorbed on ZnO single crystals absorbs 2.6% of the incident light in a direct transmission m e a s ~ r e m e n t .The ~ other is a factor of 4.4 which relates the greater absorption of such a layer in an internal reflection arrangement at ZnO than in a direct transmission measurement.13 Therefore, in the ATR measurement a monolayer (1 1) Nakashima, N.; Yoshihara, K.; Willig, F. J . Chem. Phys. 1980, 73, 3553. (12) Porter, G. Proc. R . SOC.London, Ser. A 1978, 362, 281.
iI I
-"\ ,
0.5
0.3
0.1
W a c e Coverage
Figure 4. (a) After dye is adsorbed on the electrode from a 3 X M solution of rose bengal, 9 is measured as 0 falls with a washing of the surface with 0.1 M KC1 electrolyte without dye. (b) In a similar measurement with a 3 X M solution of rhodamine B, 9 also declines as the surface coverage falls from 0.3 to 0.1.
of rhodamine B absorbs 12.2%of the incident light. A comparable figure can be used for rose bengal. With these considerations it can be seen that CP remains constant down to a surface coverage of 0.001 for rose bengal and 0.002 for rhodamine B. Therefore one may conclude that energy transfer quenching does not play a significant role in determining the efficiency of the current producing reaction and that the rate of the back-reaction of eq 1 and 2 controls the magnitude of a. It is assumed in this discussion that the surface concentration is low enough that the coincidental adsorption of two monomers next to one another is small. This incidence of "contact dimers" is believed to be responsible for an effective concentration quenching of the excitation of dye molecules in solutions and on surfaces even when spectral evidence of a dimer is absent.1'*12 However, a number of studies of the fluorescence lifetimes of sensitizing dye molecules at semiconductor electrode surfaces have shown that at surface coverages below the critical energy transfer level, as is the case here, the fluorescence lifetimes of the dyes increases significantly.'"16 Such results imply that contact dimers of these dyes should not play a significant role at the coverage considered here. The data of Figure 3 show very clearly that is not the same for all dye molecules on the surface of the electrode, an observation which has been reported previ~usly.~ These data seem to indicate that there are two populations of dye molecules, one which sensitizes current production with high efficiency and one which does not sensitize at all. This can be seen through the analysis of the amplified decay in the absorption channel shown in the inset of Figure 3 where the decay of the absorption signal matches the decay of the photocurrent. It is found from a number of analyses that only 13.3% of the absorption signal is attributable to the high efficiency, current-producing dye population. This results in a 9.8% efficiency for this form of the adsorbed rose bengal. In a similar manner, the efficient form of rhodamine B is calculated to have an efficiency of 15.4%. The implication of this result is that the oxidation expression of eq 1 may be split into two possible reactions, one for each dye population: (13) Equation 31, p 51 of ref 10 can be used to calculate the absorption through internal reflection relative to that of a direct transmission measurement. This factor of 4.4 is a function of the angle of incidence of the light and the indices of refraction of the electrode, the solvent, and the adsorbed layer. In this calculation the index of refraction of ZnO at 560 nm was taken to be 2.02, that of the solution to be 1.35, and that about the dye layer to be the mean of the two, 1.74. (14) Liang, Y.; Moy, P. F.; Poole, J. A.; Ponte Goncalves, A. M. J . Phys. Chem. 1983, 88, 2451. ( 1 5 ) Liang, Y . ;Ponte Goncalves, A. M. J . Phys. Chem. 1985.89, 3290. (16) Itoh, K.; Chiyokawa, Y.; Nakao, M.; Honda, K. J . Am. Chem. SOC. 1984, 106, 120.
J . Phys. Chem. 1986, 90, 2159-2163
-
efficient population: D* D+ inefficient population: D*
-
+ e-
D+
current production
+ e-
-
(la) (1b)
D+ + eD recombination (2) In eq l a the excited dye would inject an electron through an efficient pathway into the bulk of the semiconductor to produce photocurrent with high efficiency. In eq l b and 2 the electron would be trapped at the surface from which it returns to the parent oxidized dye, yielding little or not net photocurrent. One is inclined to seek the physical cause for the different @ of these two populations of dye in the nature of the site at which the dye is adsorbed. For example, Bressel and Gerischer” have examined the recombination reaction of (lb) and (2) at SnOa and ZnO and with thermal desorption techniques have characterized these sites as hydroxyl groups on the surface of the semiconductor. In this work an attempt has been made in the experiments of Figure 4 to distinguish these different sites by measuring the relative enthalpies of adsorption of the dye at these two locations on the surface. With the assumption that dye molecules weakly adsorbed to the surface will desorb during a wash procedure before those which (17) Bressel, B.; Gerischer, H. Eer. Bunsenges. Phys. Chem. 1983, 77,963.
2159
are more strongly adsorbed, the data of Figure 4a for rhodamine B indicate that the efficient dye population is less tighly bound to the surface than the inefficient one. This follows from the observation that CP falls rapidly as 8 declines from 0.3 to 0.1, Those dye molecules left on the surface are those most strongly adsorbed, they sensitize photocurrent least efficiently. This also appears to be the case for rose bengal a t higher 8, but to a lesser extent. However, AHadsis higher for rose bengal resulting in a rate of desorption which is much slower than rhodamine B. If the AH difference between efficient and inefficient sites is small, the resultant difference in the desorption rates may not be as significant for rose bengal as for rhodamine B over the 10-min time span of this dynamic mesurement. Such tentative interpretations must await confirmation through more thorough studies of this relation between 0 and AI&&. However, it is apparent that there is a correlation between the arrent-sensitizing ability of a molecule and strength of its adsorptive interaction with the surface.
Acknowledgment. The author thanks M. A. Ryan and E. C. Fitzgerald for their assistance in the laboratory with these experiments. He is also indebted to Prof. R. Helbig for the gift of the ZnO single crystals and to A. Nozik for his support at SERI while the author was on sabbatical leave. This work was supported by the Office of Basic Energy Sciences of the Department of Energy. Registry No. ZnO, 13 14- 13-2; rhodamine B, 8 1-88-9; rose bengal, 11 121-48-5.
Surface Chemistry of Phosphine on Clean and Oxidized Iron R. I. Hegde and J. M. White* Department of Chemistry, University of Texas, Austin, Texas 78712 (Received: November 12, 1985)
The interaction of PH, adsorbed on clean and oxidized iron at 100 K has been studied by temperature-programmed desorption (TPD) and Auger electron spectroscopy (AES). In TPD after PH3 adsorption on clean iron, both molecular desorption and dissociative processes are evident. On oxidized iron, the dissociation of PH, is inhibited. Coadsorption of D2 and PH, leads to no detectable deuterium incorporation in the desorbing phosphine, but all the isotopic forms of molecular hydrogen desorb in relatively large amounts. Preadsorbed PH3 inhibits D2 adsorption, while postdosed PH, not only displaces D2but also changes its desorption peak shape. Compared to clean Fe, adsorption of D2is strongly inhibited on both oxidized and PH,-covered oxidized iron surfaces at 100 K.
Introduction The surface chemistry of adsorbed organophosphorus molecules has been the subject of recent investigation^.^-'^ Recently we a simple reported the surface chemistry of phosphine (PH3),192,4 phosphorus-containing molecule, and dimethyl methylphosphonate (DMMP),, a relatively large organophosphorus compound, on Rh( 100) using various surface science spectroscopic techniques. (1) Hegde, R. I.; Tobin, J.; White, J. M. J . VUC.Sci. Technol. A 1985, 3, 339. (2) Hegde, R. I.; White, J. M. Surj. Sci. 1985, 157, 17. (3) Hegde, R. I.; Greenlief, C. M.; White, J. M. J . Phys. Chem. 1985,89, 2886. (4) Greenlief, C. M.; Hegde, R. I.; White, J. M. J. Phys. Chem. 1985,89, 5681. (5) Ertl, G.; Kuppers, J.; Nitschke, F.; Weiss, M. Chem. Phys. Lett. 1977, 52, 309. (6) Nitschke, F.; Ertl, G.; Kuppers, J. J . Chem. Phys. 1981, 74, 2911. (7) Kiskinova, M.; Goodman, D. W. Surf. Sci. 1981, 108, 64. (8) Yu, M. L.; Myerson, B. S. J. Vac. Sci. Technol. A 1984, 2, 446. (9) Shanahan, K. L.; Muetterties, E. L. J . Phys. Chem. 1984, 88, 1996. (10) Garfunkel, E. L.; Maji, J. J.; Frost, J. C.; Farias, M. H.; Somorjai, G. A. J. Phys. Chem. 1983, 87, 3629. (11) Friend, C. M.; Muetterties, E. L. J. Am. Chem. SOC.1981, 103, 773. (12) Tsai, M.-C.; Muetterties, E. L. J . Am. Chem. SOC.1982, 103, 2534. (13) Tsai, M.-C.; Friend, C. M.; Muetterties, E. L. J. Am. Chem. SOC. 1982, 104, 2539.
0022-3654/86/2090-2159$01.50/0
As part of our continuing investigation of the surface chemistry of phosphorus-containing molecules on a variety of transition metals, we report here the behavior of PH3 on clean, phosphorus-covered, deuterium-covered, and oxidized Fe surfaces. The experiments include temperature-programmed desorption (TPD), Auger electron spectroscopy (AES), and hydrogen-deuterium isotope exchange. To our knowledge, the surface chemistry of PH3 on an iron surface has not been previously reported. Ertl and co-workers5 have studied a related molecule, PF3, which dissociates on Fe.
Experimental Section The experiments were carried out in a stainless steel ultrahigh-vacuum chamber pumped by a 450 L/s turbomolecular pump which has been described elsewhere.’i3J4 The base pressures were in the low 10-l0-torr range. Briefly, the chamber was equipped with (1) a line-of-sight mass spectrometer which was computer interfaced to multiplex up to nine peaks in TPD, (2) an argon ion gun for sputtering, and (3) a single-pass cylindrical mirror analyzer with a coaxial electron gun for AES. The polycrystalline Fe was a thin rectangular disk, 1 cm X 1 cm X 0.1 cm, metallographically etched and polished. This sample (14) Hegde, R. I.; White, J. M. J . Phys. Chem. 1986, 90, 296
0 1986 American Chemical Society