J. Phys. Chem. 1988, 92, 5965-5972 of HS04- should be determined by this rate constant (in contrast to the rate of disappearance of S(IV), which we find to be extremely fast). We attempted to measure the rate of appearance of HSO, by in situ turbidimetry22using Ba(C104)2to (indirectly) support the hypothesis that disulfate is a reaction intermediate. However, these experiments were inconclusive since the rate of growth of BaS04 particles in standard sulfate solutions was of the same order of magnitude as the expected rate of hydrolysis of S2072-. We are thus as yet unable to unequivocally distinguish between these two possible reaction mechanisms (eq 9-10 and 15-17). The selection of a mechanism has no effect on the rate law, which is the same in both cases. The rate constants in eq 11 are replaced by k’l, kLL,and k’2 (as appropriate) to give the rate law for the second mechanism. It is worthwhile noting here that evidence has recently been found for the existence of the disulfate ion as an intermediate in the aqueous oxidation of HS0< by 02.23 (22) Greenberg, A. E.; Trussell, R. R.; Clesceri, L. S., Eds. Standard Methods for the Examination of Water and Wastewater, 16th ed.; American Public Health Association: Washington, 1985. (23) Chang, S. G.; Littlejohn, D.; Hu, K. Y . Science (Washington, D.C.) 1987, 237, 756-58.
5965
The similar rates of oxidation of S(1V) by HS0< and by H202 suggests that the respective reactions may lead to similar transition-state intermediate^.^^ If the intermediate has the structure
where X = SO3- (for HS05-) or H (for H202),then the ability of the substituent, X, to enhance the protonation of 0 and thus to facilitate S-0 bond formation may be compensated by the tendency of X to stabilize the 0-0 bond. Therefore, although peroxymonosulfate is potentially extremely reactive from a thermodynamic point of view, there may be a kinetic barrier for reaction with HS03-. This in turn suggests that under certain conditions in tropospheric water droplets where S(1V) is the dominant reductant, the concentration of HSO, may build up to unexpectedly high levels as predicted by Jacob.2 Registry No. SOz, 7446-09-5; HSO,-,12188-01-1. (24) We thank one of our reviewers for making this suggestion.
Photolysis of p -Nitrobenzoic Acid on Roughened Silver Surfaces Soncheng Sun,+ Ronald L. Birke,* John R. Lombardi,* Department of Chemistry, City College, City University of New York, New York. New York 10031
King P. Leung,t and Azriel Z. Genack* Department of Physics, Queens College, City University of New York, Flushing, New York 1 1 367, and Exxon Corporation, Clinton Township, New Jersey 08801 (Received: January 7, 1988)
Surface-enhanced photochemical reduction of p-nitrobenzoic acid (PNBA) is observed on Ag island films and roughened Ag electrode surfaces. A change in the SERS spectrum on irradiation with laser light is attributed to a photoinduced electrochemical reduction of PNBA. The reduction product could be either p-aminobenzoic acid (PABA) or azodibenzoate or both. It is found that the reduction necessitates both molecular adsorption on a roughened metal surface and the correct laser excitation energy. This reduction is a photoinduced process rather than a thermally induced reaction. The photolysis rate is determined as a function of photon flux,excitation energy, electrode potential, and the nature of solvent. A charge-transfer mechanism is described for the surface-induced photochemical reduction.
Introduction We have observed the photochemical reduction of pnitrobenzoic acid (PNBA) on Ag island films and on roughened Ag electrode surfaces and followed its time dependence over a range of laser frequency and power. The photoreduction of PNBA does not occur in the absence of the surface, molecular absorption of visible light being exceedingly weak either in solution or in the solid state. Thus a new mechanism must account for the observed photolytic reduction which is not simply an enhancement of a process that exists in the absence of the surface. Furthermore, we show that laser heating can be ruled out as the source of the photoinduced process. Surface-enhanced Raman scattering (SERS) and electrochemical methods have been used to investigate the rate, mechanism, and products of the photolysis. The use of SERS had allowed us (i) to achieve submonolayer sensitivity, (ii) to follow reactions in real time, and (iii) to identify possible products and Present address: Department of Chemistry, Virginia Commonwealth University, Academic Center, Richmond, VA 23284. ‘Present address: Avco Research Laboratory, Inc., 2385 Revere Beach Parkway, Everett, MA 02149.
0022-3654/88/2092-5965$01.50/0
intermediates. Investigations of the surface photochemistry on SERS active metal surfaces may in turn lead to a deeper understanding of the influence of a rough metal surface on the optical processes of adsorbed molecules. It has been known that, in addition to Raman scattering, other optical properties of molecules such as absorption] and fluorescence2 are dramatically enhanced and modified when they are adsorbed on or near a rough metal surface. The photochemistry of adsorbed molecules is usually quenched by a smooth metal surface due to a nonradiative energy transfer to the substrate3 and the occurrence of photochemistry is neither expected nor universal. However, a number of photochemical reactions have been observed on some rough metal surfaces. For example, the enhanced photofragmentation of pyridine, pyrazine, and benza l d e h ~ d ephotochemical ,~ degradation of rhodamine 6G (R6G) on silver island films,5 and the demethylation of N-methyl(1) Garoff, S.;Weitz, D. A.; Gramila, T. J.; Hanson, C. D. Opt. Lett. 1981, 6, 245. (2) Glass, A. M.; Liao, P. F.; Bergman, J. G.; Olson, D. H. Opt. Lett. 1980, 5, 368. (3) Gerischer, H. Discuss. Faraday SOC.1974, 58, 219. (4) Goncher, G. M.; Harris, C. B. J . Chem. Phys. 1982, 77, 3767.
0 1988 American Chemical Society
5966
The Journal of Physical Chemistry, Vol. 92, No. 21, 1988
pyridinium6 and methyl~iologen~ on Ag electrodes have been reported. Furthermore, several calculations have been given which predict that the photochemical reactions can be enhanced near a rough metal surface by an electromagnetic effect.'~~ The surface enhancement of a photochemical reaction might be the result of enhancement of the light intensity near the surface produced by the excitation of localized electron plasma resonances. On the other hand, deactivation processes are greatly enhanced by the coupling to the metal. Thus the surface-induced enhancement of any optical process depends on the balance between enhanced pumping and decay from various deexcitation channels. Therefore, only those molecules which have a greater pumping rate than decay rates would show photochemistry. In this paper we suggest as another possibility a charge-transfer mechanism in which the excitation can take place by a charge transfer between the metal and states of the charge-transfer complex. Then the photochemistry may be assisted by lowering of the charge-transfer threshold. Both a charge transfer and an electromagnetic effect can operate simultaneously. For a physisorbed molecule the charge-transfer effect can be thought of as enhanced photoemission from metallic microstructures. Localized plasmons associated with multipoles higher then the dipolar mode may create strong electric fields just under the surface of metal microstructures,10which might enhance photoelectron emission by several orders of magnitude over that of a planar metal surface.1° Although PNBA has been widely used as a model molecule l-~~ in SERS studies on metal surfaces or tunnel j u n c t i ~ n s , ~less attention has been given to its photolysis. There have been reports in several studies of two different kinds of SERS spectra from PNBA adsorbed on a Ag surface.l4-I7 In addition to a spectrum similar to the normal Raman spectrum of PNBA, a second distinct spectrum is observed in which a band near 1460 cm-' and a doublet near 1150 cm-' occur, which is very different from the normal Raman spectrum. A satisfactory explanation for the diverse spectra observed in these studies has not been discussed. A recent report, which appeared during the course of our investigation, by Roth, Venkatachalam, and Boerio,'* gives a detailed study of the possible chemical change of PNBA on a Ag island film. These authors conclude that PNBA undergoes a thermally induced chemical reduction and that the reduction is probably azodibenzoate. In the present investigation the SERS spectra on both Ag island films and Ag electrode surfaces are studied. The results described here have led us to the conclusion that the chemical change observed on SERS activated surfaces is a photoinduced rather than thermally induced process. Experimental Section The silver island film is produced by slow thermal evaporation" onto a quartz substrate to a mass thickness of about 50 A. The substrate is mechanically polished and chemically cleaned. An electron micrograph has shown that the islands are on average (5) Garoff, S.; Weitz, D. A,; Alverez, M. S. Chem. Phys. Lett. 1982, 93, 283. (6) Bunding, K. A,; Durst, R. A.; Bell, M . I. J . Electroanal. Chem. 1983, 150, 437. (7) Lu, T. H.; Birke, R. L.; Lombardi, J. R. Langmuir 1986, 2, 305. (8) Nitzan, A,; Brus, L. E. J . Chem. Phys. 1981, 75, 2205. (9) Nitzan, A,; Brus, L. E. J . Chem. Phys. 1981, 74, 5321. (10) Kerker, M. J. Colloid Interface Sci. 1985, 105, 257. (11) Liao, P. F.; Stern, M. 9. Opt. Lett. 1982, 7 , 483. (12) Tsang, J . C.; Avouris, Ph.; Kirtley, J. R. Chem. Phys. Lett. 1983, 94, 172. (13) Tsang, J. C.; Kirtley, J. R.; Theis, T. N.; Jha, S . S. Phys. Reu. B 1982, B25, 5070. (14) Murray, C. A.; Allara, D. L.; Rhinewine, M. Phys. Rev. Lett. 1981, 46, 57. (15) Dornhaus, R.; Benner, R. E.; Chang, R. K.; Chabay, 1. Surf. Sci. 1980, 101, 367. (16) Tsang, J. C.; Avouris, Ph.; Kirtley, J. R. J . Chem. Phys. 1983, 79, 493. (17) Heritage, J . P.; Allara, D. L. Chem. Phys. Lett. 1980, 7 4 , 507. (18) Roth, P. G.; Venkatachalam, R. S.; Boerio, F. J. J . Chem. Phys. 1986, 85, 1150. (19) McCarthy, S. L. J . Vac. Sci. Technol. 1976, 13, 135.
Sun et al.
4 2 I
400
I
600
I
800
I
I
1000 1200
I
1400
I
I
1600 I800
Wavenumber / cm-' Figure 1. SERS spectra of PNBA on a Ag island film a t different times of irradiation with 5 m W of 476.5-nm laser light.
200 A in diameter and cover about 3C-40% of the surface.20 The PNBA molecules were deposited on the substrate by a dipping technique2' in which the substrate is slowly inserted into and M solution of PNBA in ethanol. The withdrawn from scattered light was collected in a back-scattering geometry. The incident laser beam was focused onto the island film with an f/1.2 lens with 5 cm focal length, giving a spot approximately 25 wm in diameter. The same lens was used to collect the light. The laser radiation was provided by either a Spectra Physics Ar' laser or a Coherent Kr' laser, with the substrate either stationary or spinning. The SERS spectra on Ag island films were recorded by using a Spex Triplemate (Spex 1877) spectrometer with a PAR optical multichannel analyzer, OMA. The experimental setup for the SERS study of PNBA on silver electrodes has been described elsewhere.22 Both a scanning monochromator (Spex 1401) and an OMA based spectrometer (Spex 1877 and Tracor Northern OMA system) were used in studies of molecules adsorbed on electrode surfaces.22 All electrode potentials are quoted versus the saturated calomel electrode, SCE. The p-nitrobenzoic acid (PNBA) and p-aminobenzoic acid (PABA) compounds were reagent grade and used without further purification. In the electrochemical study, the solution pH was adjusted by adding NaOH to give a pH 11 in order to increase the solubility of PNBA. At this pH the predominant form in solution is p-nitrobenzoate (pK, = 3.47) ion (hereafter we also abbreviate these anions as PNBA). It has been found that monocarboxylic acids chemisorb on Ag or metal oxide surfaces as symmetric, bidentate carboxylate ions; that is, the carboxylic acid group loses its hydrogen to form a carboxylate anion. Both oxygen atoms on the carboxylate are then symmetrically bonded to the surface. This is indicated by the absence of C=O stretching band in the 1700-cm-' region of the SERS spectrumI2 and from studies using tunneling spectro~copy~~ and second-harmonic generation24as well as FTIR-ATR studies.25 (20) Garoff, S.; Stephens, R. B.; Hanson, C. D.; Sorenson, G. K. Opt. Commun. 1982, 41, 257. (21) Yang, C . C.; Josefowicz, J. Y.; Alexandrov, L. Thin Solid Films 1980, 74. 117. 1
~~
(22) Sun, S. C.; Bernard, I.; Birke, R. L.; Lombardi, J. R. J . Electroanal. Chem. 1985, 196, 359. (23) Hall, J. T.; Hansma, P. K. Surf. Sci. 1978, 6, 61. (24) Heinz, T. F.; Tom, H. W. K.; Shen, Y. R. Phys. Reu. 1983, 28, 1883.
Photolysis
p-Nitrobenzoic Acid on Silver Surfaces
The Journal of Physical Chemistry, Vol. 92, No. 21, 1988 5961
100 50 V
2
20
i=
10
E
I1
h
l
L
h
1 1000
1200
1400
1600
W a V en umber/ c m-1
Figure 3. SERS spectra of PNBA on a Ag electrode surface at different times of irradiation with 20 mW of 606-nm laser light. The electrode 500
was roughened by an oxidation reduction cyclic (ORC) pretreatment with a pulse of -0.2 to +0.5 to -0.2 V vs SCE, with 2-s duration. Both ORC treatment and SERS measurement are in 50 mM PNBA and 0.1 M Na2S04electrolyte and at pH 11.
1000 1500 Wavenumbevcm-'
Figure 2. SERS spectrum of PNBA on a Ag island film under cyclohexane. Excitation at 514.5 nm with 40-mW power. (a) The first 1-s exposure was recorded. The labeled peaks correspond to the SERS spectrum of unphotolyzed PNBA. The other main peaks correspond to cyclohexane. (b) After 30 min of exposure with 40 mW of laser light. (c) The background spectrum of cyclohexane and the SERS spectrum of unphotolyzed PNBA has been subtracted out. Labeled peaks correspond to the photolysis products.
Sodium sulfate was used as the electrolyte. We found that halides quench the SERS spectrum of PNBA due to the competition of halide anions with the anions of PNBA on the Ag surface. The SERS intensity observed for PNBA was much larger when sulfate or nitrate was used as the electrolyte.
Results and Discussion The Observation ofthe Photolysis of PNBA. Figure 1 shows the time dependence of the SERS spectrum of PNBA on a stationary Ag island film which is irradiated by laser radiation. The initial SERS spectrum is the same as the spectrum obtained when the substrate is spinning and is very similar to the normal Raman of PNBA in solution phase. The main bands at around 866, 1115, 1355, 1395, and 1602 cm-I correspond to 6(NO2), ~ ~ ( ~~ 0 ~3 ,( yS(co,),and 7 8 ring modes, respectively (where 6 represents the deformation and y represents the stretching mode). The absence of the C=O stretching mode around 1700 cm-' indicates the loss of a hydrogen atom when the carboxylic acid group is adsorbed on the Ag island surface with bonding to the surface through both oxygens. As the sample is irradiated with laser light, scattering from a set of new modes at around 825, 1131, 1145, 1175, 1283, 1390, 1453 and 1595 cm-' increases in intensity. This observation is very similar to that of Roth, Venkatachalam, and Boerio recently reported.I8 We have also observed that a structured background around 1350 and 1550 cm-I grows as the intensities of the new modes increase. This background is very similar to the broad bands which were identified as surface carbon due to surfaceenhanced ph~tofragmentation.~ When the sample is immersed in cyclohexane, Figure 2, the same change in spectrum is observed as when the sample is exposed to air. However, under cyclohexane there is no remarkable background growth as the intensities of the new modes increase. The presence of cyclohexane seems to protect the sample from photofragmentation; the cyclohexane may inhibit an air oxidation reaction which is part of the photofragmentation process or it may facilitate the dissolution of photofragments. We have found that the rate of increase in intensity of the new Raman modes is faster for irradiation at 488.0 nm than for 514.5 ( 2 5 ) Kimura, F.; Umemura, J.; Tamenara, T. Langmuir 1986, 2, 96.
~
nm at the same photon flux of laser irradiation, indicating a frequency dependence of the change in spectra. In order to determine whether this change of the spectrum is caused by a thermally induced effect or by a photoexcitation, we heated the Ag island film sample, on which the PNBA molecule had been deposited, in an oven. No change in the SERS spectrum of PNBA was observed except for a decrease of the overall SERS intensity. This decrease in the overall spectrum is probably due to either a change of the Ag morphology or desorption of the adsorbate. After heating to 110 "C for 30 min, the whole spectrum disappeared. Roth et al.'* similarly found no observable change in the SERS spectrum of PNBA on the Ag island film even on heating to 160 OC. The SERS spectra of PNBA on a roughened Ag electrode surface in aqueous solution has also been studied. The change in SERS spectra at the electrode/electrolyte interface, as shown in Figure 3, is similar to that on the Ag island film. Initially, the SERS spectrum is almost the same as the normal Raman of PNBA in aqueous solution. The same modes as those on the Ag island film grow with time on irradiation with laser light. Similar ~ ) ~ to0 the observation on island films covered with cyclohexane, on Ag electrode, which is under an aqueous solution, the backthe ground remains flat. The other isomers of PNBA, o-nitrobenzoic acid and mnitrobenzoic acid, as well as similar molecules such as benzoic acid, nitrobenzene, and p-, 0-,and m-aminobenzoic acids were studied under the same conditions as those for PNBA. In all these systems, except for a growth of the background, no spectral change was observed. The solid PNBA and p-nitrobenzoate salt powder were tested under the same excitation conditions as those in the study of PNBA on the Ag island film. It was found that the spectra from these solids do not change even after irradiation for an hour. In addition, there is no observable change in the normal Raman spectrum of PNBA in solution phase under irradiation even with a laser power of 600 mW. These experiments demonstrate that the change in spectrum only occurs for the molecules which are adsorbed on the roughened metal surface. On the other hand, at a Ag electrode in the absence of blue or green laser light, there is no change of the SERS spectrum (recorded with red excitation at low power) even several hours after an oxidation-reduction cycle (ORC) pretreatment. The transformation only starts once a laser of high enough photon energy irradiates the electrode surface and ceases when the laser is shut off. From the above observations it is seen that both the rough metal surface and laser excitation are necessary for the observation of a spectrum which varies with time. Therefore, the change of the spectrum is neither a simple
5968
1000
1200
1400
1600
Wovenumber/ cm-’
Figure 4. SERS spectra on a Ag electrode surface at -0.2 V, using 5 mW of 606-nrn laser excitation. (a) 0.05 M PNBA. The first 1-s exposure was recorded. (b) 0.05 M PABA. (c) 0.05 M PNBA after 100 s of exposure under 20 rnW of laser light.
photochemical process nor a simple electrochemical process, nor is it a thermally induced chemical process. We conclude it is a surface-induced photolysis on Ag island films and a photoinduced electrochemical process on the roughened Ag electrode. Investigation of the SERS of the Photolysis Products. A possible photolysis product is p-aminobenzoate formed by the complete reduction of the nitro group of PNBA. Figure 4a-c shows the SERS spectrum of p-aminobenzoate, PABA, compared with the SERS spectra of PNBA before and after photolysis. Figure 4a is the SERS spectrum of PNBA recorded within 1 s of irradiation by 606-nm laser light at a low power (5 mW) to suppress the photochemical reaction. Thus the spectrum corresponds to an unchanged spectrum of PNBA. It is, indeed, very similar to the normal Raman (NR) spectrum of PNBA in aqueous solution. Figure 4c is the spectrum of PNBA after 100 s with the irradiation at 20 mW of 606-nm laser light, and Figure 4b is the SERS spectrum of the PABA under the same conditions as Figure 4c. Comparing the spectrum in Figure 4b with Figure 4a,c, it can be seen that the spectrum 4c is nearly identical with the superposition of spectra 4a and 4b; that is, PABA has the same SERS spectrum as the photolysis product of PNBA. It is possible, therefore, that these spectra characterize the same surface adsorbed molecular species. The SERS spectrum of PABA is significantly different from the solution Raman spectrum of PABA. The solution spectrum is dominated by two bands: one at 1383 cm-’ most likely corresponding to a COO- stretching vibration and one at 1607 cm-’, a benzene ring stretching mode. However, the SERS spectrum is dominated by four bands: one at 1600 cm-’, the ring stretching, one at around 1454 cm-I probably corresponding to the ring ~ closely , spaced bands at around in-plane deformation Y ~ and~ two 11 30 and 1145 cm-l. In most SERS spectra, the band positions are only slightly shifted (usually less than 20 cm-’) from the normal Raman, since the molecules are only weakly bonded to the surface, and vibrational energy levels are slightly perturbed by the interaction between molecules and metal surface. The remarkable difference between the SERS spectrum and normal Raman spectrum of PABA indicates that there must be some other effect which causes the change in the spectrum. A two-phase model, which accounts reasonably well for the observed SERS spectrum of PABA, has been proposed by Moskovits and DiLella26based on the concentration dependence and temperature dependence of the SERS spectrum of PABA on Ag sols. They suggested that there are two different phases for adsorbed PABA molecules. One is called a “gaslike” phase, which (26) Suh,J. S.; DiLella, D. P.; Moskovits, 1540.
Sun et al.
The Journal of Physical Chemistry, Vol. 92, No. 21, 1988
M. J . Phys. Chem. 1983, 87,
is present in low concentration or at high temperature. Another is called a “solidlike” phase, which appears in high concentration or at low temperature. The SERS spectrum of the surface “solidlike” phase resembles that of crystalline PABA in many aspects, while the surface “gaslike” phase is much more similar to PABA molecules in solution. For example, the prominent band at 1452 cm-’ in the SERS spectrum of solidlike phase has its counterpart (1432 cm-’) in crystalline PABA, but no such band is seen in the solution spectrum. The solution spectrum, on the other hand, is dominated by the two bands at 1383 and 1607 cm-’, close to their counterparts (at 1372 and 1600 cm-’) in the SERS spectrum of the gaslike phase. Furthermore the COO- vibration (at 1383 cm-’) is a prominent band in both the solution spectrum and gaslike SERS spectrum, but quite weak in both the spectra of crystalline and solidlike phase. The SERS spectrum of PABA observed in our experimental conditions is very similar to the SERS spectrum of adsorbed PABA in the solidlike phase as reported by Moskovits and DilellaaZ6Therefore, it is possible that the main photolysis product of PNBA is PABA. It has been proposed that the solidlike phase26 is formed by “condensed” molecules which are linked together by hydrogen bonds between -NH2 and -COO- groups on neighboring molecules. Such linkages are in fact known to exist in the ordinary solid form of PABA.27 In the Case of the photolysis of PNBA, this kind of linkage can exist not only between PABA molecules but also between PABA and PNBA, that is, between the -NH2 group in PABA and the -COO- group in PNBA. This may be the reason why the intensities of modes associated with -NO2 (at 866, 11 15, and 1355 cm-’) do not decrease to the same extent as the product modes increase during the course of the photolysis of PNBA as seen in Figure 3. As the PNBA molecules are reduced, the PNBA molecules in the solution phase can diffuse to the surface and link with the PABA. This may compensate for the loss of the intensity of PNBA due to the reduction. When we replaced the PNBA solution with an electrolyte solution after ORC pretreatment (a flow cell and replacement procedures have been described elsewhere22),we found that upon laser irradiation the intensity of the PNBA modes decreases much faster with respect to the product modes than when PNBA is present in solution. Another possible product of the photolysis of surface adsorbed PNBA is an azo compound formed by a condensation reaction. Roth et a1.18 concluded that an azodibenzoate is the main reduction product based on the similarity of the SERS spectrum of photolyzed PNBA on a Ag island film with the normal Raman spectrum of azodibenzoate. Futher evidence that is consistent with this conclusion will be presented below. Thus some uncertainty exists as to the identity of the main photolysis product. Determination of the Photolysis Rate. The photolysis rate is determined based on the change of the SERS intensity, assuming that the SERS intensity is proportional to the concentration of the species adsorbed on the surface. We use the intensity of the 1450-cm-’ band (background subtracted) to represent the change of the surface concentration of the photolysis product, since this band has less interference from the background and adjacent bands than do other bands. Because each change of laser power required repolishing the electrode, the intensity of PNBA was normalized by the initial intensity of the 1350-cm-’ band of the PNBA SERS spectrum to eliminate uncertainty due to the change in optical alignment and sample preparation. A plot of the intensity of product vs time at various photon flux is shown in Figure 5. As the power of the irradiation increases, the photolysis rate increases. The plot of intensity vs time was fit to a polynomial, and the initial photolysis rate taken as the coefficient of the linear term which is just the first derivative at t = 0. The initial rate is plotted as a function of laser power as shown in Figure 6, and a linear relation is observed. Since the photolysis rate is found to be directly proportional to excitation power, a one-photon absorption process is indicated. This is distinct from the surface-enhanced photofragmentation of pyridine (27) DiLella,
D. P.; Moskovits, M. J . Phys. Chem.
1981, 85, 2042.
Photolysis of p-Nitrobenzoic Acid on Silver Surfaces
The Journal of Physical Chemistry, Vol. 92, No. 21, 1988 5969
4.0
3.0
2.0
1.o
0.0
I '
0
20
'
40
'
I
60
"
80
'
'
'
I
100 120
'
I
140
'
J
160
Time/ seconds
- 0 1 -03 - 0 5
Figure 5. Time dependence of the intensities of the 1450-cm-' peak of
the reduction product at different powers of laser light. 606-nm laser excitation was used.
-09 -1 1 - I 3 -1 5
-07
Potential
/ V vs SCE
Figure 8. Cyclic voltammogram of 0.001 M PNBA (in 0.1 M Na2S04, pH 11) on a Ag electrode at 100 mV/s scan rate. The arrow shows the direction of the scan.
1
2 800
i:::~,;/l,l,l,,,l -0.2
3 00
0
IO
20
30
40
50
60
-0.4
70 Potential
Power ( m W I
Figure 6. The linear relation between initial rate of photolysis and laser
power.
-0.6
/
-0.8
-1
0
V vs. SCE
Figure 9. Cyclic voltammogram of PNBA (1) on smooth Ag electrode and (2) on roughened Ag electrode at 100 mV/s scan rate. The same ORC pretreatment as in SERS measurements was used for roughening
the electrode surface.
al C
5
40
-
30
0
u
a
20 10
00
00
05
10
15
20
25
lime / minutes Figure 7. The time dependence of the intensities of photolysis product at different electrode potential; a constant laser power of 20 mW at 606
nm was used.
which was observed to be a nonlinear photochemical process under the irradiation of UV light.4 The linear laser power dependence of the photolysis rate shows that laser heating is not the source of the photochemical reaction because a thermally induced process would be expected to follow Boltzman kinetics. In fact the temperature rise at the electrode surface due to laser heating is quite modest (less than 5 "C). The photolysis rate was also studied at different electrode potentials as shown in Figure 7. The power of the dye laser (606 nm) was kept at 20 mW. The intensities of the 1450-cm-' band were plotted as a function of time. At potentials between -0.1 and -0.3 V a small increase in photolysis rate was observed as the potential becomes more negative. Then a dramatic increase of the photolysis rate was observed when the potential reaches -0.4 V. This change is most likely attributable to the initiation
of a photoelectrochemical reduction. The relationship of the spectrum to the stage of electrochemical reduction is elucidated by a study of the cyclic voltammetry of PNBA as shown in Figure 8. There are two main steps for the reduction of PNBA in the potential range between -0.1 and -1.5 V. The first step at around -0.55 V is expected to be a four-electron reduction process to form a hydroxylamine compound, and then the hydroxylamine is further reduced at around -1.1 V to form an amino compound. There is no corresponding oxidation peak for these two main steps in the anodic scan, indicating both reduction steps are irreversible. However, there is another reversible oxidation-reduction process at about -0.3 V which occurs during a second scan. This process has been reportedZsto correspond to the reduction and oxidation between the nitroso group and hydroxylamine. This reduction peak does not occur during the first scan since originally there is no nitroso, the nitroso being formed as the oxidation product of hydroxylamine. The potential dependence of the photolysis rate indicates that the change of the spectrum is related to an electrochemical reduction. From the study of the cyclic voltammetry, it can be seen that the electrochemical reduction starts at around -0.5 V. However, under the irradiation of laser light a reduction is observed with SERS on a rough Ag surface at a much more positive potential (up to the positive limit around +0.1 V where the Ag electrode is oxidized). The decrease of the charge-transfer threshold is not simply caused by the roughness of the surface. The cyclic voltammetry on a smooth Ag surface and on a roughened surface is compared in Figure 9. Only about a 100-mV shift of the reduction potential in a positive direction is observed. This shift might be due to the catalytic properties of small Ag clusters at the Ag surface which conceivably could lower the (28) McIntire, G. L.; Chiappardi, D. M.; Casselberry, R. L.; Blount, H. N. J . Chem. Phys. 1982,86, 2632.
5970
Sun et al.
The Journal of Physical Chemistry, Vol. 92, No. 21, 1988
SCHEME I: Electrochemical Reduction of Arylnitro Compounds (Ar = Benzoate Ion) e-+ H f ArN02
)
H20
>-
e-+
ArN02H.
H
I
I
1
I
+ ArN(OH)?
I
+ ArNO (nitroso)
400 alkali electrode surface
600
500
700
Wavelength / nm (hydroxylamine)
H 0 + AIN=NAI
r
'
2 ~ + + 2e-
Ai- N=N Ax (azo)
H O
(amine
Figure 10. Frequency dependence of the initial rate of photolysis of PNBA (solid line with bars), compared to the absorption spectrum of a Ag island film (dash-dot line) and to the excitation profile of SERS of PNBA on the Ag island film (dashed line). The latter two lines are taken from ref 30. 35,
20 3 5
,
l
t
~
l
~
l
'
I
'
1
+ products
overvoltage required for reduction.29 The chemical and electrochemical reduction of aromatic nitro compounds has long been s t ~ d i e d . ~ ' -The ~ ~ electrochemical reduction process can be represented by Scheme I.34 Generally speaking, the reduction proceeds in two ways, depending on whether the solution is alkaline or acidic. If the reducing agent is strong enough or the electrode potential is negative enough, the ultimate product in both cases is the primary amine. In alkaline medium the nitroso compound may act as an electrophilic center and the hydroxylamine compound can react with it with the formation of azoxy compounds. The hydroxylamine, as it is the unprotonated form, may act as a nucleophile, and a higher pH may make the electrophilic center more reactive. Thus if the nitro compound is reduced in alkaline solution, the main products are those formed by the condensation of ArNO and ArNHOH. In contrast, the amino compound will be predominant in acidic medium. In our work the solution pH was adjusted to pH 11. This pH would seem to be more favorable for the formation of the azo compound. In highly basic solution (0.1 M KOH) the formation of trans-azobenzene from the reduction of nitrobenzene can be determined by cyclic voltammetry and SERS.3s-37 There is an additional peak around -0.5 V, which has been assigned to the trans-azobenzene. The SERS spectrum of trans-azobenzene can be detected only in a narrow potential range between -0.4 and -0.55 (vs. SCE), since at more positive potential azobenzene has not been formed and at more negative potential the azobenzene will be further reduced. This study indicates that azobenzene can be formed in basic medium and observed by SERS. However, (29) Asmus, K. D.; Wiggerand, A,; Henglein, A. Ber. Bunsen-Ges. Phys. Chem. 1966, 70, 862. (30) Weitz, D. A.; Garoff, S.; Gersten, J. I.; Nizan, A. J . Chem. Phys. 1983, 78, 5325. (3 1) Brand, K. Die Elektrochemische Reduktion Organischer Nitrokorper; Stuttgart, 1908. (32) Titova, 1. A,; Levinson, I . M.; Mairanovskii, V. G.; Ershler, A. B. Elektrokhimiya 1973, 9, 424. (33) Kariv, E.; Terni, H. A,; Gileadi, E. Electrochim. Acta 1973, 18, 433. (34) Tomilov, A. T.; Mairanovskii, S.G.; Fioshin, M. Ya.; Smirnov, V . A. Elecfrochemistry of Organic Compounds; Halstead: New York, 1972; p 248. (35) Shindo, H.; Nishihara, C. Surf. Sei. 1985, 158, 393. (36) Nishihara, C.; Shindo, H. J . Electroanal. Chem. 1986, 202, 231. (37) Nishihara, C.; Shindo, H.; Hiraishi, J. J . Electroanal. Chem. 1985, 191. 425.
1
00 00
05
10
15
20
25
30
Tlme/minutes
Figure 11. Time dependence of the intensities of photolysis product at (a) aqueous solution and (b) a mixture of 25% of isopropyl alcohol and 75% water solution; a 30-mW Kr+ laser (647.1 nm) was used. the formation of the azobenzene is not a surface-induced photochemical process, it is merely an electrochemical reduction step, since the azobenzene is only observed when it is electrochemically reduced. The photolysis rate was further investigated as a function of excitation frequency at a constant electrode potential (-0.2 V) and at a constant laser power (30 mW). The initial rate of photolysis is obtained from the plot of the intensity of photolysis product modes vs time, and it is plotted as a function of excitation frequency in Figure 10. The error bars indicate the range for repeated measurements. The frequency dependence of SERS intensity (excitation profile) of PNBA on a Ag island film and the absorption spectrum of the Ag island film are also plotted on the same graph for comparison in an arbitrary scale on the vertical axis. Both the SERS and photolysis rate exhibit a pronounced resonance, with the photolysis rate peaking around 490 nm at a frequency very close to that of the excitation profile of Raman scattering. The photolysis rate is found to be dependent upon the solvent. Figure 11 shows the time dependence of the intensity of photolysis product. It can be clearly seen that the photolysis rate in an aqueous solution (Figure 1l a ) is much faster than in a mixture of isopropyl alcohol and water (Figure 1 lb). The process of the photochemical reduction of aromatic nitro compounds in solution phase with irradiation by UV light can be written as in Scheme IL3* In this scheme R H represents the solvent which contains abstractable hydrogen atoms. In the photochemical reduction the ArN02 molecule first adsorbs a photon. The energy of the photon must be high enough to cause an electronic transition; then the excited state is followed by a hydrogen abstraction from the solvent. This hydrogen abstraction (38) Barltrop, J. A,; Bunce, N. J. J . Chem. SOC.C 1968, 1467. (39) Lombardi, J. R.; Birke, R. L.; Sanchez, L. A,; Bernard, I.; Sun, S. C. Chem. Phys. Lett. 1983,86, 3166.
Photolysis of p-Nitrobenzoic Acid on Silver Surfaces
The Journal of Physical Chemistry, Vol. 92, No. 21, 1988 5971
SCHEME 11: Photoinduced Reduction of Arylnitro Compounds
(nj
\L
Ar N (OH) 2
Ar NOH*
1
AI "OH
"2
-
charge-transfer mechanism of SERS. For a Ag electrode-molecule system, the general theory of such a scattering mechanism has been recently given.40 Both of these chargetransfer models would show a similar lowering of the chargetransfer barrier with potential; however, the plasmon model would not show a resonant charge transfer. Only if the absorption of this system is over a broader frequency range than the plasmon resonance which is the width of the electromagnetic (EM) enhancement would the width of the plasmon resonance itself correspond to the frequency width over which photochemistry occurs. The features that are distinct in the conceptualization of models i and ii are that the molecule-metal coupling is not explicitly considered in (i) whereas enhancement of the laser EM field by the surface is not explicitly considered in (ii). From Figure 10 it can be seen that the photochemical excitation profile is similar to the SERS excitation profile which could be consistent with either of the above models. In our case an electromagnetic enhancement effect by itself is probably not the source of photochemical reaction of PNBA adsorbed on the Ag surface, since the frequencies we used are in the visible range, but the absorption bands of PNBA are in the UV range. The energy needed to cause an electronic transition in the molecule is much larger than the excitation energy of the laser. Thus, in a pure electromagnetic surface-enhanced photolysis process a multiphoton absorption process would be required. An energy accumulation is needed as reported in the photofragmentation processes4 with a rate that is nonlinear in laser power. However, our observations indicate a one-photon process for the photochemical reduction of PNBA. Of course, the limiting step may be the absorption of the first photon, and we cannot rule out some EM contribution to the photolytic process. Based on the observation of the potential dependence and frequency dependence, a mechanism involving charge transfer seems important in the enhancement of the photochemical reduction on the Ag surface. Similar to our view of the chargetransfer theory for SERS,39,40we can consider the metal and chemisorbed molecules as a single system as in the previously discussed model (ii). First of all, the interaction between the adsorbed molecules and the metal surface may lower the charge-transfer threshold. A charge-transfer acceptor level may exist in the molecule and there is a Fermi level in the metal which can be changed by a change of the electrode potential. A negative shift of the electrode potential will raise the Fermi level, and vice versa. If the energy difference between the Fermi level EF and the low-lying excited state of the charge-transfer complex ECT matches the energy of the excitation radiation, a resonant charge transfer from metal to the excited state of the complex will take place. The electron-transfer step is viewed as a direct optically induced charge transfer from the Fermi level to a low-lying LUMO of the adsorbed surface complex. Such a direct laserinduced electron transfer has recently been proposed for the photoreduction of metal ammine complexes on the basis of photocurrent measurement^.^' Following the electron-transfer step the excited state of the complex may undergo two types of deexcitation processes. One involves the return of the electron back to the surface followed by a radiative process. Another possible process could be chemical one. The excited state may be attacked by a proton in solution to form a radical ArN0,H' followed by subsequent chemical or electrochemical reactions. In (40) Lombardi, J. R.; Birke, R. L.; Lu, T. H.; Xu,J. J . Chem. Phys. 1986, 84, 4174. (41) Corrigan, D. S.;Weaver, M . J. J . Electroannl, Chem. 1987, 228, 265.
5972
J . Phys. Chem. 1988, 92, 5972-5977
the case of a very weakly interacting metal-molecule system the interfacial barrier must give rise to a lifetime for the chargetransfer state that is long enough for a fast chemical step to take place. Evidence for charge transfer is the similarity of the frequency dependence of the photolysis and the SERS intensity profile shown in Figure 10. At a constant potential the photolysis rate reaches a maximum at nearly the same excitation frequency as the SERS intensity. From the potential dependence of the photolysis rate we cannot observe a resonance profile as seen in the SERS spectra since at around -0.5 V PNBA is electrochemically r e d u d . Thus, the present experiments cannot distinguish between models (i) and (ii). However, the increase in photolysis rate that occurs before the electrochemical reduction (-0.1 to -0.3 V) is evidence of a charge-transfer mechanism. The change of the potential in the negative direction will raise the Fermi level and result in the decrease of the charge-transfer threshold or to the barrier to tunneling. Thus a higher photolysis rate will be observed at more negative potentials. As a matter of fact, there are many other damping processes that will also be enhanced by the surface. The reason why onitrobenzoic acid or m-nitrobenzoic acid does not show photochemical reduction under the same conditions as in the case of PNBA might be attributed to the competition of the chemical step with other damping processes. Since adsorbed PNBA has its nitro group pointing out, it may be that PNBA is in a more favorable position to react with the solvent than o-nitrobenzoic acid or m-nitrobenzoic acid. On the other hand, reactivity differences between isomers might also be due to differences in
charge-transfer interactions. These differences would be expected to be especially pronounced for interactions through an extended 7r system as in the present surface complex. In conclusion, the change of the SERS spectra of PNBA on Ag island films and Ag electrode surfaces is attributed to a surface-induced photochemical reduction. Heating the sample does not cause the conversion of the spectra nor is the reaction rate vs laser power consistent with a reaction that is thermally induced. The potential dependence and frequency dependence of the photolysis rate also indicate that the change of the spectra is not thermally induced. It is found that both the rough metal surface and excitation with laser light is necessary for the surface-induced photochemical reaction. The reduction takes place at much more positive electrode potentials than a simple electrochemical reduction and with a much lower excitation energy (visible excitation) than a simple photochemical reduction which requires UV excitation. It is possible that the reaction product is either PABA or an azo compound or both. Although the molecular details of the mechanism require further study, it is clear that a charge-transfer mechanism may explain the potential and frequency dependence of the photolysis rate. Important clues to the mechanism may come if related photochemistry can be found and studied for other adsorbed molecules.
Acknowledgment. R.L.B. and J.R.L. are indebted to the PSC-BHE Research award program of the City University of New York (RF666367 and RF667261) and the National Science Foundation (CHE-87 11638) for financial assistance. Registry No. PNBA, 62-23-7; Ag, 7440-22-4.
Kinetics of the Reactions of C,H,S with NO2, NO, and 0, at 296 K Graham Black,* Leonard E. Jusinski, Chemical Physics Laboratory, SRI International, Menlo Park, California 94025
and Roger Patrick LSI Logic, Santa Clara, California 95050 (Received: February 1 , 1988; In Final Form: April 28, 1988)
The laser-induced fluorescence technique has been used to study the reactions of the ethylthio radical (C2H5S)with NO, NO,, and O2 at 296 K. The reaction of CzH5Swith NO involves a third body and has been shown to be in the transition region between the low- and high-pressure limits. Using an expression developed by Troe to fit the results, we have obtained X values of the broadening factor F = 0.49foqE, k , = (5.2 f 0.5) X lo-" cm3 molecule-' s-l, ko(He,Ar) = (75.);: cm6 molecule-2s-I, and ko(SF,) = (2.22,:) X cm6 molecule-z s-'. The reaction with NO2 has a rate coefficient of (9.2 & 0.9) X lo-" cm3 molecule-' s-I. No reaction could be found with 01,for which an upper limit on the rate coefficient of 2 X cm3 molecule-' s-' was estimated.
Introduction Alkylthio radicals (RS) are important intermediates in the reaction of reduced sulfur compounds (RSH, R S R , RSSR) with O H radicals in the atmosphere. They are also important intermediates in the photolysis and hydrogen atom reactions of these compounds. Although these compounds are minor constituents, they may play a part in the atmospheric sulfur cycle and contribute to the acid precipitation problem. It is, therefore, important to understand the atmospheric chemistry of these radicals. Until recently, the only reported absolute rate coefficients involved the HS radical.'-7 Recently, laser-induced fluorescence (1) Black, G. J . Chem. Phys. 1984, 80, 1103. (2) Black, G.; Patrick, R.; Jusinski, L. E.; Slanger, T. G. J . Chem. Phys. 1984, 80, 4065. (3) Bulatov, V. P.; Kozliner, M. J.; Sarkisov, 0. M. Khim. Fiz. 1984, 3, 1300.
0022-3654/88/2092-5972$01.50/0
(LIF) spectra of CH3S were reported.8 This has led to the first measurements9J0 of the rate coefficients for the reactions of CH3S with NO, NO,, 02,and 0,. Very recently, LIF spectra of C2H5S were reported." Using the L I F technique, we report the first (4) Friedl, R. R.; Brune, W.
H.; Anderson, J. G. J. Phys. Chem. 1985,89,
5505.
( 5 ) Schonle, G.; Rahman, M. M.; Schindler, R. N. Ber. Bunsen-Ges. Phys. Chem. 1987, 91, 66. ( 6 ) Stachnik, R. A.; Molina, M. J . J . Phys. Chem. 1987, 91, 4603. (7) Wang, N. S.; Lovejoy, E. R.; Howard, C. J. J . Phys. Chem. 1987,91, 5743. ( 8 ) Susuki, M.; Inoue, G.; Akimoto, H. J . Chem. Phys. 1984,81, 5405. (9) Balla, R. J.; Nelson, H. H.; McDonald, J. R. Chem. Phys. 1986, 109, 101.
(IO) Black, G.; Jusinski, L. E. J . Chem. SOC.,Faraday Trans. 2 1986,82, 2143. ( 1 1 ) Black, G.; Jusinski, L. E. Chem. Phys. Lett. 1987, 136, 241
0 1988 American Chemical Society
