Surface Raman spectroscopy of the three redox forms of

Lu, Ronald L. Birke, and John R. Lombardi. Langmuir , 1986, 2 (3) ... Characterization of Methyl Viologen in the Channels of Zeolite L. Brian Hennessy...
0 downloads 0 Views 641KB Size
Download PDF
Langmuir 1986,2, 305-309

305

Surface Raman Spectroscopy of the Three Redox Forms of Methylviologen Tianhong Lu, Ronald L. Birke,* and John R. Lombardi* City University of New York, City College, Department of Chemistry, New York, New York 10031 Received April 1 , 1985. In Final Form: November 10, 1985 The surface Raman spectroscopy of the various redox forms of methylviologen was studied on a Ag electrode. The Raman spectrum of the monoradical cation of methylviologen on the Ag electrode at -0.8 V was found to be a superposition of the surface-enhanced Raman and resonance Raman spectra. These two kinds of Raman spectra can be clearly separated and shown to originate from different forms of the monoradical cation of methylviologen. The surface-enhanced Raman and normal Raman spectra of the fully reduced methylviologen are reported for the first time and the assignment of b a n lines is discussed. A demethylation of methylviologen on the Ag electrode due to the photoelectrolysis was also observed.

Introduction

The viologens were originally investigated as indicators in biological redox studies.' At present several applications of viologens have been found; for example, they can be used as herbicides,2 electrochemical display devices,3s4 electron mediators in biological systems5 and in photoelectrolysis,6 and materials to chemically modified electrodes.'+ Recently, surface-enhanced Raman (SER) and resonance Raman (RR) techniques have been used to characterize the viologens such as methylviologen and heptylvi~logen."~~ These investigations were mainly devoted to the Raman spectra of the monoradical viologen cation which shows a RR effect because it displays a wide absorption band in the 450-650-nm region? Lee, Schmidt, Gordon, and MeiseP have reported the RR spectrum of the monoradical cation of methylviologen (MV+) in aqueous solution. When one studies the surface Raman spectrum of MV+, the question arises as to whether it is a SER spectrum, a surface enhanced resonance Raman (SERR) spectrum, or a mixture of a SER and RR process. Some investigators'lJ4 have discussed this question in order to understand the mechanism of the SER effect, but their conclusions are different and not entirely clear. In this paper we present results which show that the Raman spectrum of MV+ on a rough Ag electrode is a superposition of the SER and RR spectra and that we can clearly distinguish between SER and RR bands. In addition, we report for the first time the SER spectrum and normal Raman spectrum of fully reduced methylviologen (MV"). Also the effect on various vibrational modes of redox state is discussed, namely, for the bridging carbon(1) Michaelis, L.; Hill, E. s. J . Gen. Physiol. 1933, 16, 859. (2) Boon, W. R. Chem. Znd. (London) 1965, 782. (3) Jasiski, R. J. J.Electrochem. SOC.1977, 124, 637. (4) Van Dam,H. T.;Ponjee,J. J. J. Electrochem. Soc. 1974,121,1555. (5) Makey, L. N.; Kuwana, T. Bioelectrochem. Bioenerg. 1976,3,596. (6) Fan, F. F.; Reichman, B.; Bard, A. J. J. Am. Chem. SOC.1980,102, 1488. (7) Akahoshi, H.; Toshima, S.; Itaya, K. J. Phys. Chem. 1981,85,818. (8) Landrum, H. L.; Salmon, R. T.; Hawkridge, F. M. J . Am. Chem. SOC.1977,99, 3154. (9) Ciesliski, R. C.; Armstron, N. R. J. Electrochem. SOC.1980, 127, 2605. (10) Regis, A.; Corset, J. J. Chin. Phys. 1981, 78, 687. (11) Ohsawa, M.; Nishijima, K.; Suetaka, W. Surf. Sci. 1981,104, 281. (12) Lee, P. C.; Schmidt, K.; Gordon, S.; Meisel, D. Chem. Phys. Lett. 1981, 80,242.

carbon stretch and the out-of-plane bending motion. Finally, the demethylation of methylviologen on the Ag electrode resulting from photoelectrolysis is discussed.

Experimental Section Methylviologen chloride (MVC12)was obtained from Sigma Chemical Co. and 4,4'-bipyridyl dihydrate from Aldrich Chemical Co. and both were used as received. Reagent grade KCl was used as the supporting electrolyte. The solutions were prepared with deionized, distilled water and deaerated before use by nitrogen bubbling. Generally, a M MVC12solution with 0.1 M KCl was used. A Tacusel type PRG 5 polarographic instrument and RD3 Pine Instrument Co. potentiostat were used for electrochemical measurements. A saturated calomel reference (SCE) electrode was employed in this study, and all potentialsthat are quoted are with respect to the SCE. The working electrode was Ag or Pt with a Pt counter electrode. The surface area of the working electrode was about 1.4 mm2. In the electrochemical measurements the potential sweep speed was 4 mV/s. The pretreatment of the 99.99% pure polycrystalline Ag working electrode was the following: The working electrode was polished with alumina (particle size: 0.3 pm), then immersed in a solution of H202and NH3 (1:l)for about 10 s, and washed with deionized distilled water. Finally, the electrode was mounted in the cell, and an oxidation-reduction cycle (ORC) applied to the Ag electrode with a double potential step: -0.2 to +0.3 to -0.2 V. The pulse duration was 5 s. The charge passed during oxidation was about 2 mC. In this paper we will refer to a Ag electrode without an ORC as a 'polished Ag electrode" and one with an ORC as a 'rough Ag electrode". Raman spectra were generally taken after an ORC. In order to carry out electrochemical and spectral measurements of surface species without interference from solution species, in some experiments, after an ORC, the cell was washed in situ under potential control with 0.1 M KC1solution;i.e., the bulk solution in the cell was replaced with 0.1 M KCl solution, and then measurements were carried out. The experimental setup for Raman spectroscopy is similar to that described elsewhere.16 The excitation sources were a Spectra Physics Model 164 argon laser at 488 nm with a power of about 10 mW on the cell and a Spectra-Physics Model 375 dye laser at about 585 nm with a power of 20 mW on the cell. The spectra were recorded with a Spex Model 1401 double monochromator with a resolution of about 4 cm-'. Results and Discussion Electrochemistry of Methylviologen on Ag. It is now well-known that methylviologen, Le., 1,l'-dimethyl-

(13) Forster, M.; Girling, R. B.; Heater, R. E. J. Raman Spectrosc. 1982, 12, 36. (14) Melendres, C. A.; Lee,P. C.; Meisel, D. J.Electrochem. Soc. 1983, 130,1523.

0743-7463/86/2402-0305$01.50/0

(15) Birke, R. L.; Lombardi, J. R.; Sanchez, L. A. Adu. Chem. Ser. 1982, No.201.

0 1986 American Chemical Society

306 Langmuir, Vol. 2, No. 3, 1986

Lu et al.

Table I. Raman Bands of Methylviologen and Their Assignment RR lines SEX lines normal Raman of MVt, of MVt, of MVO, lines of M V , cm-' cm-' cm-' cm-l

SER lines

SER lines of M V 2 , cm-' 240 (vs) 280 (m) 410 (w) 560 (w) 653 (w) 674 (w)

280 (w) 360 (w) 550 (w) 618 (w) 678 (w)

678 (w)

790 (8) 840 (8) 1034 (w) 1054 (w) 1188 (5) 1238 (m) 1292 (8) 1480 (w) 1520 (w) 1573 (w) 1638 (s)

816 (w) 1026 (8)

1008 (8) 1190 (m) 1236 (s) 1352 (m) 1506 (m)

1212 (w) 1350 (m)

282 (w) 340 (w) 546 (w) 612 (w) 676 (w) 744 (5) 790 (w)

C

16a 16b 6b d' 11 d 17a 18b 9a 1/12

992 (8)

996 (s) 1208 (w) 1240 (8) 1380 (w) 1512 (s)

W1

1538 (s)

1530 (s) 1612 (s)

1592 (s) 1656 (m)

2olcathodic n

assignment Ag-C1

1656 (m) 1668 (w)

19b 19a 8b 8a

A

-a d

10

I

-0.2

I

-0.4

I

-0.6

I

I

-0.8

-1.0

3.0Cathodic

2.0-

I

-1.2

Potential ( V ) va. SCE

Figure 1. Differential pulse voltammogram of lo4 M MVClz in 0.1 M KC1 on a polished Ag electrode: wan rate, 4 mV/s; pulse height, IMl=50 mV; (A) cathodic pulse c w e ; (B)Anodic pulse curve. 4,4'-bipyridyl cation, can be generated in three redox forms, as a result of the following equilibria:

+

MV2+ e- = MV+ MV+ + e- = MVO The first reduction step is highly reversible. The further reduction to the fully reduced state is less reversible. Figure 1is the differential pulse voltammogram (DPV) of MVC12 on a polished Ag electrode. We can see that for the first reduction step the peak potential difference is equal to the pulse magnitude, IhEl, and the peak current ratio is unity, i.e., E,C - E pa = W l

Thus, this step is reversible on polished Ag according to the DPV criteria of Birke et a1.16 The second reduction step does not satisfy the above equations and is not reversible. The anodic peak for the second reduction step is quite sharp, suggesting that the fully reduced product is deposited on the electrode. (16) Birke, R. L.; Kim, M. H.;Strassfeld, M. AnaZ. Chem. 1981, 53, 852.

-0.1

-03

-05

-07 Potential ( V )

-09

-I I

Figure 2. Cyclic voltammogram of M MVZt in 0.1 M KC1 solution on the rough Ag electrode: scan rate, 4 mV/s. Generally, there are two reduction peaks in the cyclic voltammogram (CV) or DPV of MV2+;however, after an ORC at the Ag electrode three peaks appear. Figure 2 is a cyclic voltammogram on the rough Ag electrode. We can clearly see that there are three reduction peaks with a new reduction peak located a t about -0.57 V. This new peak was mentioned by Melendres, Lee, and Meisel14 and attributed to the reduction of MV2+with strong adsorption of the product of the cathodic reduction, i.e., MV+ or a related species. The cyclic voltammogram (Figwe 2) shows the prepeaks discussed by Wopschall and Shain17 indicating the reduction of dissolved MV2+to adsorbed MV+. However, after an ORC and thorough washing of the cell with 0.1 M KC1 solution at -0.2 V, the CV curve shows that the normal reduction peaks disappear but the new peaks still exist with a peak potential at ca. -0.57 V. In addition, the SEX spectrum of M V + a t -0.2 V is still observed after washing. Thus M V + is irreversibly adsorbed on the rough Ag electrode. It is apparent that the ORC stablizes the (17) Wopschall,

R.H.;Shain, 1. Anal. Chem.

1967,39, 1514.

Langmuir, Vol. 2, No. 3, 1986 307

Surface Raman Spectroscopy of Methylviologen 1530

I

p w - 4 700

900

I

1100

I

1300

I

I500

I

1700

Roman Shift ~ c m - ' ~

Figure 3. RR spectrum of MVf solution: 488-nm argon laser light, 100 mW.

adsorption since without it the new peaks cannot be seen. For the latter conditions the new peak is due to the reduction of MV2+adsorbed on sites which are produced by the ORC process. The reduction product, MV+ or a related species, is adsorbed on the electrode more strongly than MV2+ because the potential of the new peaks are more positive than that of the normal reduction peaks of M V . SER Spectrum of MV2+. After an ORC of the Ag electrode the Raman spectrum of 1.0 X M MVC12 in 0.1 M KC1 was taken on the rough Ag electrode a t -0.2 V. The results are similar to those reported by Regis and Corset.lo For comparison with MV+ and MVO, the bands observed for MV2+are listed in Table I. Typical of most SER spectra of colorless molecules, these bands do not show large changes from the normal Raman spectrum of MV2+in solution except for slight wavenumber shifts in band positions and changes in the relative intensities of some bands. From the electrochemistry we know that MV2+is not reduced at -0.2 V. In addition, after the cell was washed in situ with 0.1 M KC1 solution the spectrum did not change. This experiment illustrates that the SER spectrum is due to the species irreversibly adsorbed on the electrode surface. Nature of Surface Raman Spectra of MV+. When the solution of MVClz is reduced a t -0.8 V, the MV+ radical is obtained and a blue color is produced a t the electrode surface. Figure 3 is the RR spectrum of MV+ obtained a t a polished Pt electrode. The bands observed and their assignments are listed in Table I. The vibrational modes are described by Wilson for the parent benzene ringla and by Kohlrausch for substituted benzenes.le These bands are nearly identical with those seen in the solution RR spectrum of MV+ prepared by chemical reduction with Na2S204.12Normally it is difficult to obtain the RR spectrum of a colored species a t a nonenhancing electrode surface if other RR active species are in the bulk. In the present case, the only RR active species is MV+ which is produced at the electrode surface by the reduction reaction, and the RR spectrum of MV+ in the diffusion layer is easily obtained. Figure 4A is the Raman spectrum of methylviologen on a rough Ag electrode a t -0.8 V, a t which MV2+is reduced to the MV+ radical. After taking this spectrum the cell was carefully washed in situ with 0.1 M KCl solution. During the washing the potential was kept at -0.8 V. Then the spectrum was taken again (Figure 4B). It is seen that the spectrum before washing is different from that after washing with some bands having been lost after washing. The lost bands are exactly the same as those found in the RR spectrum of the MV+ radical (Figure 3). Thus, it is clear that the spectrum of methylviologen on the rough Ag electrode is the superposition of the RR spectrum of MV+ in the solution near the electrode surface and the surface Raman spectrum of MV+ on the electrode surface. (18)Wilson, F. B., Jr. Phys. Rev. 1934,45, 706. (19) Kohlrausch, K. W. F. Phys. 2. 1936, 37, 58.

I 700

90

1100 I300 R A M A N SHIFT ( c m - ' )

1500

I700

Figure 4. Raman spectra of MV' on a rough Ag electrode at -0.8 V (A) before washing the cell with 0.1 M KC1 solution; (B) after

washing.

Melendres et al.'* reported both sets of bands in the same spectrum, indicating the likelihood that they were observing an overlap of bcth spectra. The observed surface Raman spectrum of the irreversibly adsorbed species is quite different from the solution RR spectrum showing strong to medium strength bands a t 790 (s), 1008 (s), 1190 (m), 1236 (s), 1352 (m), 1506 (m), and 1592 (s) cm-' while major bands of the RR spectrum of the solution species appear at 1026 (s), 1350 (m), 1530 (s), and 1656 (m)cm-' (Table I). Only the 1350-cm-l band is common to the two spectra. These large changes in band position make it obvious that the surface Raman spectrum is not a SERR scattering of an adsorbed MV+ radical monomer. The SERR scattering is a RR process additionally enhanced by an electromagnetic effect. In this case one would expect more of a correspondence in the observed band positions of SERR and solution RR spectra. Other possibilities are that the surface spectrum is the SER scattering of a chemisorbed monomer MV+ species exhibiting a combined charge transfer and electromagnetic enhancement or either SER scattering or SERR scattering originating from a chemisorbed dimer MV+%or a polymer MV+ species.a For the reasons indicated below, we believe that the surface spectrum originates from the SER scattering of the monomer. Landrum et al.8 indicated that the polymer film of MV+ radical can be formed by electrochemical means under conditions that, however, are different from the present case. The prominent characteristic of this polymer film is its stability with respect to the application of an electrode potential between +0.50 and -0.95 V vs. AglAgC1. Our experiments show that the MV+ species on the surface of a rough Ag electrode is not stable in the above potential range. For example, after taking the spectrum a t -0.8 V, if the potentia1 is returned to -0.2 V, the SER spectrum of MV2+can be observed again. This procedure can be repeated many times and the same spectra appear (only the intensity is lowered). In addition, the reduction peak a t about -0.57 V in the CV can be still observed. Thus the polymer film of Landrum e t al. is not formed on the surface of the rough Ag electrode under our experimental conditions and the surface Raman spectrum is not due to the polymer of the MV+ radical. On the basis of the CV curves for this system, Melendres et al. reached a similar conclu~ion.~~ Ohsawa et al." reported the RR spectrum of MV+ and its dimer and indicated that only two weak bands at 1513 and 1340 cm-l can be assigned to the dimer. Since these (20) Thorneley, R. N. F. Biochim. Biophys. Acta 1974,333, 487.

308 Langmuir, Vol. 2, No. 3, 1986 Table 11. Overtone and Combination Bands of RR Spectrum of MV+ frequencies of possible overtone or frequencies of exptl bands, cm-' combination bands, cm-' 2052 (1026 + 1026) 2050 2376 (1026 + 1350) 2376 2556 (1026 + 1530) 2532 2562 (1212 + 1350) 2556 2682 (1026 + 1656) 2682 2700 (1350 + 1350) 2868 (1212 + 1656) 2870 2880 (1350 + 1530) 3014 3006 (1350 + 1656) 3060 3060 (1530 + 1530)

dimer bands are not observed in the surface Raman spectrum mentioned above, it is not due to the RR effect of the dimer of the MV+ radical. Additional experiments support the conclusion that the surface Raman spectrum is due to the typical SER effect. For example, if a polished Ag electrode or Pt electrode is used, only the RR spectrum can be observed, and the surface Raman spectrum only develops after an ORC pretreatment on Ag which is typical SER scattering behavior. Further evidence that the surface Raman spectrum is due to the SER effect is obtained by observing the overtone or combination bands. We observe bands in the 2040-3064-cm-' region on the rough Ag electrode a t -0.8 V before washing the cell with 0.1 M KC1 solution, but most of the bands disappear after washing. The frequencies of those bands which disappear correspond to overtone or combination bands of the RR spectrum (Table 11). They are around 5% of the intensity of the fundamentals. The question arises as to whether or not overtones are observed in a SER spectrum. The only specific report of the observation of overtones in a SER spectrum is that by Pettinger.21 However, as far as we can tell their intensities are less than 1% of the fundamentals. Since in the theory of RR spectroscopy relative intensities depend only on relative Franck-Condon factors, there is no particular restriction on the intensities of overtones. In principle an overtone could be more intense than a fundamental. However, in NR spectroscopy, as well as the charge-transfer theories of SER spectroscopy based on Herzberg-Teller the relative intensities come from terms involving the normal IR matrix elements. In the harmonic oscillator approximation, one predicts zero intensity for overtones and combinations. However, if anharmonic corrections are included one expects some intensity in overtones and combination bands for NR and SER spectroscopy. The larger the anharmonic correction, the greater would be the overtone intensity. Note for MV+ the overtones are almost exactly twice the fundamental frequencies (Table 11) which indicates only slight anharmonicity. Thus, we would expect almost no intensity of overtone or combination band in the SER spectrum, and these bands can be considered as the overtone or combination bands of RR bands. This is consistent with the theoretical prediction given by Adrianz3and more recently by Lombardi et ai.zz The SER scattering may, in fact, be due to the MV+ radical monomer or a dimer which is in equilibrium with the monomer. However, when the surace Raman spectra were taken for different concentrations of MVCl, in solu(21) Pettinger, B. Chem. Phys. Lett. 1981, 78, 404. (22) Lombardi, J. R.; Birke, R. L.; Lu, T. H.; Xu, J. J. Chem. Phys., in press. (23) Adrian, F. J. J. Chem. Phys. 1982, 77, 5302.

Lu et al. I

c

700

1538

900

1100

1300

1500

1700

Roman Shift ( c m - ' )

Figure 5. Normal Raman spectrum of MVo on a polished Ag electrode at -1.2 V.

tion, it was noted that the relative intensities of the bands in the surface Raman spectra did not change with respect to each other as the MVC1, solution concentration was varied from 5 X to M. One would expect some change with concentration if a dimer species was the source of SER scattering. The question arises as to why the surface Raman spectrum is very different from the RR spectrum. EfrimaU indicated that the interaction between adsorbed molecule and the metal can lead to shift in molecular states, i.e., a nonchromophric molecule can be "pushed" into resonance and exhibit SER scattering, and a chromophoric molecule can, on the other hand, be "pushed" out of resonance. The MV+ radical may be good example of the later situation. Therefore, the surface Raman spectrum is very different from the RR spectrum. In fact, the SERS spectrum may be similar to the solution NR of MV+; however, the strong absorption of the colored MV+ throughout the visible region prevents the observation of an off-resonance spectrum. The surface Raman band assignments of MV+ are listed in Table I. The following observations should be mentioned: (a) We observed the characteristic shift of the 1292-cm-' band of MV2+to 1350 cm-I for MV+. This band is attributed to the C4-C4' stretch. As shown by Dewar and TrinajstiF the lowest unoccupied molecular orbital in biphenyl, @7, contains considerable electron density in this bridging carbon-carbon bond, thus accounting for the sizeable upward shift in frequency. These resulb have also been observed by Takahashi and Maedaz6 in biphenyl. (b) The assignment of other bands is of some dispute. Lee et al.I2 give their 818-cm-' band the same assignment as the 840-cm-' band in MVz+(d mode, principally C-C-C bend.z7 On the other hand Regis and Corsetloassign their 1027-cm-' band to the same vibrational mode as the 840cm-l band in MVz+. This latter assignment is unlikely, for such a large shift (about 180 cm-l) in a C-C-C bending motion could not be so strongly affected by addition of an electron to the 6, orbital where most of the additional electron density is added perpendicular to the molecular plane. On this basis we tend to agree with the assignment of Lee et a1.12 (c) Regis and Corsetlodid not observe some bands, such as the strong SER bands at 790,1008,1506, and 1592 cm-'. By means of the assignment of bands from MVClz solution by Regis and Corsetlo and the assignment of SER bands of pyridine by Fleischmann and the 790-cm-' SER band is assigned to vll, the 1008-cm-' SER band is given the same assignment as the 1026-cm-' RR band, (vlTa), and the 1506-cm-' SER band is assigned to Vlgb, and the 1592 cm-' SER band to Vgb. (24) Efrima, S. J.Phys. Chem. 1985, 89, 2843. (25) Dewar, M. J. S.; Trinajstic, N. Collect. Czech. Chem. Commun. 1970,35, 3136. (26) Takahashi, C.; Maeda, S. Chem. Phys. Lett. 1974, 24, 584. (27) Zerbi, G.; Sandroni, S. Spectrochim. Acta, Part A 1968, !HA, 511. (28) Fleischmann, M.; Hill, I. R. J.Electroanal. Chem. 1983,146,353.

Langmuir, Vol. 2, No. 3, 1986 309

Surface Raman Spectroscopy of Methyluiologen

IT

1592

A 1236

n

I

B

700

900

A

992

1604 1292

1100

1300

I500

1700

Roman Shift (cm-ll

Figure 6. Raman spectra of MV" on a rough Ag electrode a t -1.2 V (A) washing the cell with 0.1 M KC1 solution a t -0.8 V, then adjusting the potential to -1.2 V; (B) without washing.

I

J

700

900

I100

1300

1500

1700

Raman Shift ( c m - ' )

Figure 7. SEX spedra of MV+ on the rough Ag electrode by using the different excitation sources: (A) 488-nm argon laser light (20 mW);

In addition, the following phenomena are worth mentioning. Generally, an ORC a t the Ag electrode was applied by a double potential step: -0.2 to +0.3 to -0.2 V. When the potential was adjusted to -0.8 V and the Raman spectrum of MV+ was taken, we found that the relative intensity of the SER and RR bands in the spectrum of MV+ changed with the duration time, t+2, for which the electrode is held a t -0.2 V after ORC. The shorter t4,2is, the weaker the surface Raman bands are. There are two factors involved in this phenomena. First, the development of the intensity of SER bands of MV+ after ORC is a slow process. The intensity quickly reaches about 50% of the maximum in about 10 s after ORC, and then it slowly rises to a maximum in approximately 10 min. Second, after the electrode potential is adjusted to -0.8 V, the intensities of SER bands do not increase any more. This means that the surface Raman-active adsorbed MV+ are mainly formed by reduction of adsorbed MV2+. Raman Spectrum of MVO. In Figure 5 we show the Raman spectrum of MVClz on a polished Ag electrode a t -1.2 V. We believe this spectrum to be the normal Raman spectrum of an MVO polymer film formed on the polished Ag electrode, because under these conditions a barely visible transparent layer of material is observed to be coated on the surface of the Ag electrode. This observation is consistent with the electrochemical results (Figures 1, 2). Figure 6B is the Raman spectrum of MVO on a rough Ag electrode a t -1.2 V. Similar to the situation of MV+, this spectrum consists of two kinds of bands: SER bands of MVO adsorbed on the electrode surface and NR bands of the MVO polymer. If we wash the cell with 0.1 M KC1 solution a t -0.8 V, only the adsorbed MV+ species is left on the surface of the rough Ag electrode. On adjusting the potential to -1.2 V, the SER spectrum of MVO (Figure 7A) is obtained. There are no NR bands of the MVO polymer in this spectrum, and the SER bands of MVO are not very stable. Their intensities decrease with time and they almost disappear in about l h. The SER and normal Raman bands of MVO observed are listed in Table I. In the SER spectrum of Mvo no band is seen at 1292 or 1350 cm-' but instead a band is observed at 1380 cm-'.Most likely the bridge carbon-carbon stretch at 1352 cm-' for MV+ has been shifted to still higher frequency when a second electron is added to the c$17 orbital. The second electron will not have as large an effect as the fist (1292 cm-l for M V + shifts to 1352 cm-' for MV+) due to electron-electron repulsions which will tend not to allow

A

(B)585-cm dye laser light

(20 mW).

a simple doubling of the electron density buildup in this bond. There is a band at 996 cm-' in the SER spectrum of MVO that is analogous to the band 1008 cm-' for MV+ and the 1034-cm-' band for MV2+. The decrease in frequency on adding an electron is consistent since this vibration does not involve a stretching motion of the bridge carbons. Instead it arises from the out-of-plane bending motion of the ring carbon atoms. The lowest unoccupied molecular orbital c$17 has a number of nodes which cross several carbon-carbon bonds. Thus addition of an electron to this orbital would result in a slight weakening of those bonds and consequently a slight decrease in frequency. A strong band at 744 cm-l in the SER spectrum of MVO is also seen. This corresponds to the 741-cm-' band in biphenyl. However, the assignment is not certain. In addition, we are not certain of the assignment of the 1668-cm-' band. Photoelectrolysis of Methylviologen. On comparison of the SER spectrum of MV+ (Figure 7A) for which a 488-nm argon laser light (20 m W) is used as an excitation source with the spectrum (Figure 7B) for which a yellow or red dye laser light (20 mW) or low-power (less than 10 mW) 488-cm excitation is used, some differences are observed between the two spectra. For example: (a) There is no 1292-cm-' band in Figure 7B, instead the 1350-cm-' band appears as mentioned above. A strong 1292 cm-' band is observed in Figure 7A. (b) A 768-cm-' band is observed in Figure 7A. (c) The 1592-cm-' band in Figure 7B shifts to 1604 cm-' in Figure 7A. We find that the 768, 1292, and 1604-cm-' bands are characteristic bands for the SER spectrum of 4,4'-bipyridyl. These phenomena seem to indicate that a demethylation reaction has taken place on the surface when the 488-nm laser light at higher power is used and that there is an apparent threshhold for the process. These observations of surface photochemistry are of sufficient interest to warrant further study.

Acknowledgment. We are indebted to Dr. Xinhua Li for helpful discussions concerning this work. W e are also

indebted to the National Science Foundation (CHE7911159 and RT-8305241),the PSC-BHE Research Award Program (RF-664197) of the City University of New York, and MBRS program DRR/NIIT Grant 08168 for financial assistance. Registry NO.Ag, 7440-22-4;M V , 25128-26-1; MV', 25239-558;

MV2+, 4685-14-7.