Symposium on Photoelectrochemical and Electrochemical Surface

Electrode Processes", which occurred during the 197th ACS National Meeting, April. 9-14, 1989. In addition to continuing an active series of symposia ...
0 downloads 0 Views 898KB Size
Langmuir 1990,6, 43-50

43

Symposium on Photoelectrochemical and Electrochemical Surface Science: Preface The papers that follow are based on presentations at the symposium entitled "Photoelectrochemical and Electrochemical Surface Science: Microstructural Probes of Electrode Processes", which occurred during the 197th ACS National Meeting, April 9-14, 1989. In addition to continuing an active series of symposia jointly sponsored by the Division of Colloid and Surface Chemistry and the Division of Analytical Chemistry, the symposium had the added distinction of honoring the Kendall Awardee in Surface Science, Arthur T. Hubbard. The symposium consisted of 35 presentations, attesting to the high level of activity in the title subject. A major goal of most areas of chemistry is an appreciation of the relationship between molecular structure and chemical or physical properties. In the case of electrochemistry, the region of interest is an unusual interface at which molecular orientation, electron transfer, high electric fields, and monolayer films can have pronounced effects on chemical behavior. Unfortunately, the electrode/solution interface is difficult to characterize because of the small amount of material present and the sensitivity of the interface to contamination. Thus, the development of microstructural probes of electrode surfaces and nearby solutions is a challenging but crucial task leading to determining structure/activity relationships a t electrodes. The importance of this problem has led to enormous research effort over the past several decades. Progress has accelerated relatively recently with the development of UHV surface characterization techniques, optical spectroscopic probes, and techniques for chemical manipulation a t the monolayer level. The papers presented here are particularly interesting because they demonstrate the capability of microstructural probes for revealing surface and solution structural effects on electrochemical behavior.

R. L. McCreery Ohio State University

SERS Investigation of Interfacial Methanol at Silver Electrodest Raymond L. Sobocinski and Jeanne E. Pemberton* Department of Chemistry, University of Arizona, Tucson, Arizona 85721 Received June 29, 1989 Interfacial solvent structure at Ag electrodes in methanol electrolyte systems has been studied by using surface-enhanced Raman scattering (SERS). The potential dependence of the SERS response suggests that methanol interacts with Ag electrodes through the oxygen end of the molecule. A t all potentials studied, the C-0 bond appears to remain largely parallel to the surface. It is proposed that at potentials positive of the PZC only one oxygen p-orbital is directed toward the Ag surface. Near the PZC, it is postulated that two methanol orientations are detected simultaneously in the presence of Li+. Evidence is presented for the influence of supporting electrolyte cation and anion on the orientation of methanol at these electrodes.

Introduction Considerable research has been focused on understanding interfacial organic solvent structure a t metal electrodes. However, detailed information regarding the ori-

* Author to whom correspondence should be addressed. Presented at the symposium entitled "Photoelectrochemicaland Electrochemical Surface Science: Microstructural Probes of Electrode Processes", sponsored jointly by the Divisions of Analytical Chemistry and Colloid and Surface Chemistry, 197th National meeting of the American Chemical Society, Dallas, April 9-14, 1989.

0143-7463/90/2406-0043$02.50/0

entation of solvent molecules in the inner and outer Helmholtz plane has been somewhat elusive. Until recently, nearly all the work regarding nonaqueous solvent orientation at metal electrodes involved electrochemical methods. The reason for this lies in the fact that Hg is amenable to such studies due to its smooth, clean, and defectfree metal electrode surface. Additionally, spectroscopic methods possessing the requisite sensitivity for surface species have only been available during the last 10 years. Double-layer capacitance measurements have been widely used for the investigation of solvent orientation. 0 1990 American Chemical Society

44 Langmuir, Vol. 6, No. 1, 1990 At Hg electrodes, the position of the maximum in the entropy versus double-layer charge curve has usually been interpreted in the literature in terms of solvent dipole Orientation. This interpretation is based on a two-state molecular model of solvent dipoles a t a polarizable interface,' which is inadequate for the Hg/solution interface in organic solvent^.^'^ When more solvent states are allowed, the correlation between the position of entropy maximum and dipole orientation is not straightf~rward.~ Consequently, there has been some controversy regarding the orientation of organic solvent molecules at Hg electrodes. In many cases, there is ambiguity in the double-layer capacitance measurements for the evaluation of nonaqueous solvent orientation at metal electrodes. Despite the ambiguity in such measurements, several workers have used double-layer capacitance measurements to study solvent orientation in nonaqueous solvents. Grahame4 observed a positive shift in the potential of zero charge (PZC) of Hg in methanol-water mixtures with greater concentrations of methanol and concluded that the methanol dipole is directed away from the surface. However, Garnish and Parsons4 concluded from the temperature dependence of the double-layer capacitance that the reverse orientation is preferred. More recently, Trasatti has concluded that the predominant interaction of the metal with mononitrile and dinitrile organic adsorbates is through its functional group.5 In particular, the orientation of acetonitrile, propionitrile, butyronitrile, succinonitrile, and glutaronitrile a t Hg electrodes was investigated. While it may be concluded that the functional group interacts with the metal surface, the orientation of bond angles relative to the surface remains unclear from double-layer capacitance measurements. In addition to the lack of fundamental detail on interfacial solvent orientation, relatively little is known about the role of interfacial molecular structure in controlling electrochemical reactivity at solid electrodes. A few electrochemical and spectroscopic investigations have provided some insight regarding interfacial solvent structure and its role in influencing electron-transfer events at solid electrodes. For example, it has recently been demonstrated, with thin-layer electrochemical techniques, that nonaqueous solvents can influence the surface orientation and, therefore, the electrochemical reactivity of 2,2',5,5'-tetrahydroxybiphenyl (THBP) chemisorbed at smooth polycrystalline Pt electrodes.6 Surface-enhanced Raman scattering (SERS) has been particularly useful as an optical probe of interfacial solvent structure. Recent examples include the study of a~etonitrile,~ propylene carbonate,8 and pyridineg at Ag electrodes. Acetonitrile adsorption a t Pt electrodes has also been studied" with infrared reflection-absorption spectroscopy (IRRAS).

(1) Borkowska, Z. J . Electroanal. Chem. 1988,244, 1. (2) Fawcett, W. R.; Borkowska, Z. J. Phys. Chem. 1984,87,4861. (3) Borkowska, Z.; Stafiej, J. J.Electroanal. Chem. 1984, 170, 89. (4) Payne, R. In Aduances in Electrochemistry and Electrochemical Engineering; Delahay, P., Tobias, C. W., Eds.; Interscience Publishers: New York, 1970; Vol. 7, p 10. (5) Trasatti, S. Electrochim. Acta 1987, 32, 843. (6) Song, D.; Soriaga, M. P.; Hubbard, A. T. J. Electrochem. SOC. 1987, 134,874.

(7) Irish, D. E.; Hill, I. R.; Archambault, P.; Atkinson, G. F. J . Solution Chem. 1985, 14, 221. (8) Hill, I. R.; Irish, D. E.; Atkinson, G. F. Langmuir 1986, 2, 752. (9) Pemberton, J. E. Chem. Phys. Lett. 1985, 115, 321. (10) Davidson, T.; Pons, B. S.: Bewick, A.: Schmidt, P. P. J. Electroanal. Chem. 1981, 125, 237.

Sobocinski and Pemberton While the SERS behavior of pyridine and benzene has been studied in methanol and ethanol solvents," the interfacial structure of alcoholic solvents at solid electrodes has not been investigated. This report details the use of SERS to study the effect of electrolyte and electrode potential on interfacial methanol solvent structure at Ag electrodes. Surface selection rules are used to evaluate interfacial solvent orientation. Methanol was chosen, in part, for its waterlike properties and should therefore provide a useful comparison to results obtained in aqueous systems. Additionally, methanol is an attractive and widely investigated fuel for use in organic fuel cells.

Experimental Section Spectroscopic Conditions and Instrumentation. Excita-

tion was provided exclusively by the 514.5-nm line of a Coherent Radiation Innova 90-5 Ar+ laser. All spectra were acquired at a laser power of 130 mW at the sample. The laser was focused to ca. 50 fim at the electrode surface. Scattered radiation was focused onto the entrance slits of a Spex 1403 double monochromator with 1800 grooves/" holographically ruled gratings. Detection was accomplished with a high-sensitivity,GaAs photocathode, RCA C31034A photomultiplier tube which was thermoelectrically cooled to ca. -25 "C. The spectral bandpass was maintained at 8.0 cm-'. Spectra were acquired at 2cm-' increments over a 1-2-s integration period. Electrochemical Conditions and Instrumentation. The working electrode consisted of a polycrystallineAg disk (99.9%, Johnson Matthey) which was mechanically polished to a mirror finish with 0.3-fim alumina particles, rinsed with solvent, and then sonicated for 2 min in methanol to remove any trapped alumina. A Pt wire served as the auxiliary electrode. All potentials are reported versus a Ag/Ag+ reference electrode in methanol containing AgNO,. Electrode potentials were controlled with an IBM Model EC/ 225 voltammetricanalyzer and were measured and reported versus the Ag/Ag+ reference electrode. Linear potential ramps were performed by using a triangle wave format; double potential steps were performed as two single-step experiments with a scan rate of 6000 mV/s. Total charge passed was monitored with a Princeton Applied Research Model 379 digital coulometer. Materials. Methanol was purchased as anhydrous AR grade from Mallinckrodt with water content labeled as 0.006%. This was further purified by reflux of 1.5 L of methanol with 10 g of NaBH, for 3 h. Approximately 6-8 g of Na metal was then added and refluxed for another 2-4 h. This was followed by distillation. The first 200 mL was discarded, and the collected fraction was found to contain 9 X M water by gas chromatography. Alternatively, a 24-h exposure to 3-A molecular sieve beads (Aldrich)followed by distillation was found to be equally effective (10 mM water). LiCl (Fisher Scientific Co.), LiBr (Aldrich,anhydrous,99+ %), and NaBr (Aldrich, anhydrous, 99+%) were dried at 160 "C under a 2 X lo-' Torr vacuum for 24 h. KBr (MCB Reagents, reagent grade) was used as received. All solutions were deaerated with N, for 20 min prior to use. Electrochemical Roughening Procedure. In order to obtain significantly enhanced Raman signals from species present at the electrode-solution interface, it is necessary to activate the surface. This was accomplished through a double potential step oxidation-reduction cycle (ORC). Typically, the electrode is initially poised at ca. -0.60 V versus the Ag/Ag+ reference electrode and stepped to ca. -0.20 V in 0.4 M LiBr and -0.10 V in 0.1 M LiCl for 20-30 s. The potential is then stepped back to -0.60 V. For the systems studied here, the optimum oxidation rate and charge passed for maximum SERS intensities were found to be ca. 2 mA/cm2 and 80 mC/cm2,respectively. Under these conditions, charge recovery is typically only 50-60%,unlike aqueous systems where the charge recovery is typically 95% or more. This lower charge recovery is attributed to the larger solubility of AgCl in methanol. The larger solubility allows the (11) Kim, J. J.; Shin, G. S. Chem. Phys. Lett. 1985, 118, 493.

SERS Investigation of Methanol at Ag I

I

Langmuir, Vol. 6, No. I, 1990 45

I

I

0.60

0.80

0.4 M LiBr f Methanol

i 0.00

0.20

0.40

-E vs

1.00

Ag/Ag+

Figure 1. Cyclic voltammetry of a Ag electrode in 0.4 M LiBr. Sweep rate = 20 mV s-'. AgCl formed during the ORC to diffuse away from the electrode surface, where it can no longer be reduced to Ag metal. A typical cyclic voltammogram of Ag in a 0.4 M LiBr methanol solution is shown in Figure 1. Unlike aqueous systems, a significant cathodic current remains following an oxidation-reduction cycle. This current cannot be attributed to the reduction of dissolved oxygen in these deaerated solutions, as the reduction potential does not correspond to oxygen reduction. Rather, it is attributed to the reduction of some organic species in the methanol solution, possibly formed as a result of the ORC. The morphology of silver surfaces roughened in methanol electrolyte solutions was studied with scanning electron microscopy. The large-scale surface structures resulting from a double potential step ORC in 0.1 M LiCl are approximately spherical and appear to result from nodular growth, similar to the surfaces prepared in aqueous media.12 When the ORC was performed under the optimum conditions described above, the average particle was found to be 74 f 17 nm in diameter.

Results Raman Spectroscopy of Neat Methanol and Methanol Electrolyte Solutions. The spectrum of neat methanol is shown in Figure 2a. The effect of 4 M LiBr electrolyte on this spectrum is shown in Figure 2b. A high concentration was used to mimic the probable electrolyte conditions in the electrochemical double layer. Peak frequencies and band shapes are essentially unchanged in the presence of LiBr, with the exception of the v ( 0 H) band, which becomes narrower and shifted to higher frequencies. The presence of LiCl has a similar effect. These data suggest that the hydroxyl group is involved in extensive hydrogen bonding with the halides. This conclusion is consistent with infrared spectral properties of anion-containing alcohol solution^.'^ These data further indicate that, in spite of the hydrogen bonding, the v(C-0) and u(C-H) vibrational frequencies are relatively unchanged in the presence of dissolved electro(12) Tuschel, D. D.; Pemberton, J. E.; Cook, J. E. Langmuir 1986, 2, 380. (13) Lund, H. Acta Chem. Scand. 1958,12, 298.

lyte. A summary of solution frequencies in neat methanol, 4 M LiCl/methanol, and 4 M LiBr/methanol is given in Table I. SERS Behavior of Methanol at Ag Electrodes. Following a double potential step ORC in 0.4 M LiBr, SERS spectra of interfacial methanol can readily be observed. These data are tabulated in Table I. A survey spectrum of methanol between 600 and 4000 cm-' acquired a t a Ag electrode poised a t -0.60 V is shown in Figure 3. Comparison of this spectrum with those shown in Figure 2 indicates that the most prominent features in the SERS spectrum are the vs,(C-H) and v,,(C-H) vibrations. This is in contrast to the behavior observed in neat methanol, in which both of these vibrations and the v(C-0) vibration are of considerable intensity. In going from neat solution to an adsorbed state, the intensity ratio of the v(C-0) band to the v,,(C-H) band decreases from 0.5 to ca. 0.1. This decrease may be attributed to a preferred orientation of the surface-confined methanol such that the v(C-0) vibration cannot effectively couple with the incident electric field. The ability of preferred bond orientations to couple with an electric field a t the surface is described by surface selection rules for SERS.14v'5 One interpretation of these rules is that vibrational motion perpendicular to the surface will couple more effectively with surface electric fields than vibrational motion parallel to the surface. In addition to a change in relative intensity, the u(C0) frequency is shifted to lower energies when the molecule is adsorbed. The v(C-0) vibration is observed a t 1036-1040 cm-' in the bulk and 1028 cm-' when adsorbed at Ag at positive electrode potentials. This observation suggests that the oxygen atom in methanol interacts with the Ag surface. Two bands at 1390 and 3522 cm-' are also observed in the SERS spectrum that cannot be attributed to methanol. The band a t 1390 cm-' may be due to the presence of an impurity species formed during the ORC. Although the exact nature of the species is not known, its presence can be minimized by proper choice of the reduction potential during the ORC. The band at 3522 cm-l is well-established in the literature and is ascribed to the v(0-H) of surface water with its oxygen end toward the metal surface. This band will be discussed in greater detail below. The potential dependence of the spectra in the C-H stretching region is shown in Figure 4. The band a t 2946 cm-' is due to v,,(C-H), while the band a t lower energies is due to vs,(C-H). Qualitatively, as more negative potentials are applied, the v,,(C-H) intensity loss is greater than the v,,(C-H) intensity loss. The I(C/I(C-H)s, ratio change suggests a potential-deF k g n t surface orientation. The relationship between this ratio and the electrode potential is shown more quantitatively in Figure 5 . This plot suggests a preferred orientation a t the most negative (-1.20 V) and positive (-0.60 V) potentials. Between these potentials, the I(C-H),,/ I(C-H)sm ratio is sensitive to changes in electrode potential. Consequently, one may conclude that the orientation is similarly sensitive to potential in t h i s region. Important potential-dependent behavior is also exhibited by the v(C-0) band. As shown by the spectra in Figure 6, this band is shifted to lower energies for adsorbed methanol a t positive potentials relative to its value in (14) Creighton, J. A. In Spectroscopy ofsurfaces;Clark, R. J., Hester, R. E. Eds.; Wiley: New York, 1988; Chapter 2. (15) Moskovits, M. J . Chem. Phys. 1982,77, 4408. (16) Chen, T.T.; Smith, K. E.; Owen, J. F.; Chang, R. K. Chem.

Phys. Lett. 1984, 108, 32.

Sobocinski and Pemberton

46 Langmuir, Vol. 6, No. I , 1990

a

4 M LiBr

b

Methanol

/

MeOH

1038

1036

r,

E

5

1455

2 1800

900

900

1800

2836

2840

'

2947 I

2944 h I

3380 J

'

--L-/ 2600

4

Wavenumber

(cm')

2600

4000

wavenumber

4000

(cm

Figure 2. (a) Raman spectra of neat methanol. (b) Raman spectra of 4 M LiBr/methanol solution. Table I. ComDarison of Bulk and Surface Raman Spectral Frequencies of Methanol freauencv, cm-'

SERSb

solution"

MeOH 1036 1112 1454 2836 2944 3337

LiCl/MeOH 1040 1106 1453 2842 2950 3361

LiBr/MeOH 1038 1108 1455 2840 2947 3380

a Electrolyte concentrations are 4 M for bulk solution spectra. frequencies at E = -0.7 V.

neat methanol. As the potential is made more negative, however, the frequency shifts toward the bulk value of 1038 cm-'. The v(C-0) frequency reaches a value of 1034 cm-' a t the most negative potential studied, -1.4 V. This increase in frequency suggests an increase in C-0 bond strength a t more negative potentials. The v(C-0) SERS intensity remains weak and is relatively constant with change in electrode potential, suggesting that the C-0 bond orientation does not change significantly with electrode potential. Therefore, the C-O bond orientation must remain largely parallel to the surface as the electrode potential is altered. SERS spectra in the v(0-H) region are shown in Figure 7 as a function of potential. Trace amounts of water are detected a t the surface as evidenced by a relatively narrow band a t ca. 3500 cm-l. This band has been observed previously and is consistent with the 4 0 - H ) vibration of surface water not extensively hydrogen bonded and oriented with its oxygen directed toward the surface.16 As the potential is made more negative between -0.6 and -0.9 V, this band is relatively constant in intensity, although the loss of Br- a t more negative potentials allows more extensive hydrogen bonding, as evidenced by the asymmetry on the low energy side of this band. In the presence of Li+, there are two additional bands in the v ( 0 - H ) region, at 3256 and 3354 cm-l, which cannot be ascribed to surface water. These bands are tentatively assigned to methanol species in two distinct chemical environments. The observation of two interfacial methanol species in the electrochemical environment has not been made previously. As inore negative potentials

LiCl/MeOH 1026-1034

LiBr/MeOH

assignment

1026-1034

s(C-O),, u(C-O),, 8(C-H),, u(C-H),, dC-H),, 40-H)

n.0.'

n.0.

1456 2836 2940 3256, 3340d

1456 2842 2946 3256, 3354d

0.1 M LiCl and 0.4 M LiBr in SERS spectra.

Not observed.

SERS

are applied, the intensity of these two bands increases until a potential of -0.9 V is reached. At potentials more negative than -0.9 V, both methanol and water 40-H) bands decrease rapidly in intensity with potential. This decrease may be attributed to a sudden loss in SERS activity or a drastic change in surface orientation. If a sudden loss in SERS activity beyond -0.9 V were occurring, all surface bands should undergo a similar loss in intensity. This is not observed for the v(C-0) or the v(C-H) band. Therefore, the intensity loss beyond -0.9 V must be associated with large changes in surface orientation. This is corroborated by the data in Figure 5 , which show that the I(C-H)aBp/ I(C-H)symratio is extremely sensitive to electrode potential near -0.9 V. The large spectroscopic changes observed in this potential region are consistent with unique behavior" often associated with potentials near the potential of zero charge (PZC). Therefore, it is assumed that the PZC is located in the potential region near -0.9 V, similar to its value in aqueous media." The effect of electrolyte anion (Br-, C1-) and cation (Li+, Na+, K+) on the v(0-H) vibration was also studied. Parts a and b of Figure 7 illustrate the influence of the anion on the v(0-H) spectral region. In 0.4 M LiBr, the 0-H intensities from interfacial methanol reach a maximum at -0.9 V, while in 0.1 M LiC1, the intensities are maximized at -0.7 V. The difference in potentials of (17) Anson, F.C. Acc. Chem. Res. 1975,8, 400. (18) Larkin, D.; Guyer, K. L.; Hupp, J. T.; Weaver, M. J. J.Electroanal. Chem. 1982,138, 401.

Langmuir, Vol. 6, No. 1, 1990 47

SERS Investigation of Methanol at Ag 2946

',

0.00 , , , > , , ,, , , , , 0.40 0.60

-E I 600

4000

Wavenumber (cm-')

0.80

1.00

(V

VS.

Figure 5. Z(C-H)m,,,JZ(C-H)sm trode potential.

Figure 3. SERS spectrum of 0.4 M LiBr/methanol solution at Ag.

0 4 M LiBr

/

1.20

1.40

Ag/Ag+)

ratio as a function of elec-

Methanol

+I

1

5 -e

.

-s,.rJ

960

1030

1100

Wavenumber (cm-1)

Figure 6. SERS spectra in u(C-0) region as a function of electrode potential.

1 2600

-1.2 2885

v 3170

Wavenumber (cm-1)

Figure 4. SERS spectra in u(C-H) region as a function of electrode potential. maximum intensity may be the result of more extensive Br- adsorption at negative potentials.18 Another difference between these anions is that consistently more intense SERS of interfacial methanol v(0-H) bands are observed after an ORC in Br- relative to C1-. This is apparently due to the nature of the Br- interaction with methanol and the Ag in the interface rather than a concentration effect. In the presence of 0.4 M LiC1, methanol SERS signals are no larger than those in the presence of 0.1 M LiC1. The v(Ag-Cl) and v(Ag-Br) vibrations are easily observed in methanol. The potential dependence of these bands is shown in Figure 8. The C1- signal is lost a t about -1.0 V, while the Br- signal is not lost until potentials of ca. -1.2 V, consistent with observations in aqueous media.lg The v(Ag-Cl) vibration is observed a t 230-222 cm-l depending on electrode potential and is relatively unshifted from aqueous media. The v(Ag-Br) vibration in methanol is observed a t 156-132 cm-' depending on potential. Additionally, a band a t 176 cm-' is observed in this fre(19) Coria-Garcia, J. C.; Pemberton, J. E.; Sobocinski, R. L. J.Electroanal. Chem. 1987,219,291.

quency region which is independent of potential and attributed to a Ag-cluster vibration at the activated Ag surface on the basis of previous work of Furtak." The cation dependence of the surface methanol ~ ( 0 H) was also investigated. In addition to Li+, the effects of Na+ and K+ were studied. The v(O-H) of surface methanol was only observed in the presence of Li+, in spite of strong v(C-0) and u(C-H) SERS signals in Na+ and K+ solutions. Figure 7c illustrates the absence of both surface methanol v(0-H) bands in 0.4 M NaBr despite the strong signal from adsorbed water. Similar spectra are observed in the presence of 0.4 M KBr.

Discussion Interfacial Methanol Orientation as a Function of Applied Potential. The potential dependence of the v(C-0), v(C-H), and v(0-H) bands provides insight into the orientation of methanol species interacting directly with the Ag electrode. A model for the potential-dependent methanol orientation a t Ag electrodes is shown in Figure 9. At positive potentials, the v(C-0) band is observed at ca. 1026 cm-l. This v(C-0) surface vibration is shifted down by 10 cm-' relative to solution, suggesting that methanol interacts with the Ag surface through the oxygen a t these potentials. At these potentials where the methanol is strongly bound, this orientation forces (20) Roy, D.; Furtak, T. E. Phys. Reu. B 1986, 34, 5111.

Sobocinski and Pemberton

48 Langmuir, Vol. 6, No. I , 1990

a

1 Methanol

0.4 M LiEr

i

3100

b

a 3520

3425 Wavenumber (cm-1)

0 1 M LiCl

1 Methanol

3526

3750

120

6 V(0-H)

100

b

0.1 M LiCl

Wavenumber (cm ' I

1 Methanol

250

y(Ag-CI1 226

3100

3425 Wavenumber (cm

3750 1)

C 0.4 M N a B r

3100

/

Methanol

3425 Wavenumber (cm-'1

~(0-H)

3750

Figure 7. SERS spectra of u(0-H) region in different electrolyte solutions: (a) 0.4 M LiBr, (b) 0.1 M LiCl, and (c) 0.4 M NaBr.

the methyl group to assume a specific orientation relative to the surface normal as evidenced by the change in intensities compared relative v,,,(C-H) and u,,,(C-H) to bulk methanol. The C-0 bond axis is most likely parallel to the surface at these potentials, constraining the methyl group near the surface. Additionally, the uYym(C-H) vibration must be in a plane largely perpendicular to the surface. In this orientation, the vaBym(C-H)

150

300 Wavenumber (cm ' )

Figure 8. (a) SERS spectra in v(Ag-Br) region as a function of electrode potential. (b) SERS spectra in v(Ag-Cl) region as a function of electrode potential.

vibration apparently couples with the electric field more vibration, as indicated by effectively than the u,,(C-H) the relative SERS intensities in Figure 4. At positive electrode potentials, the 40-H)bands from methanol are weak or nonexistent, depending on the electrolyte. This behavior is in contrast to that of the v ( 0 H) from surface water, which is of significantly greater

Langmuir, Vol. 6, No. 1, 1990 49

SERS Investigation of Methanol at Ag Methanol

+E

-

PZC

on

Ag

-

-E

Figure 9. Proposed model of the potential-dependentinterfacial methanol orientation and chemical interactions at a Ag electrode in LiX, where X = C1-, Br-.

intensity. The weakness of the v(0-H) bands of methanol at these potentials may be explained by a parallel 0-H bond orientation, as pictured in Figure 9. These spectral observations are consistent with a molecule which interacts with Ag through a single oxygen lone pair of electrons rather than both lone pairs. Only in this configuration can the v(0-H) and v(C-0) vibrational motions remain parallel to the surface as required by surface selection rules. A t more negative potentials, the u(C-0) increases in frequency, suggesting an increase in C-0 bond strength and, consequently, a weaker interaction with the surface. The C-0 bond remains reasonably parallel to the surface as evidenced by the constant, weak, intensity as the applied potential is made more negative. There is no indication that the C-0 bond becomes perpendicular to the surface as the PZC is reached. Negative of the PZC, the v(C-0) intensity is constant with changes in potential. Therefore, it is concluded that the C-0 bond axis remains parallel even a t the most negative potentials. However, the v(C-0) frequency continues to increase, up to a value of 1034 cm-' a t the most negative potentials. This behavior implies that the oxygen-Ag interaction continues to decrease a t potentials negative of the PZC, consistent with what is expected on the basis of Coulombic considerations. In 0.4 M LiBr, the v(0-H) bands from methanol grow in intensity as the PZC is approached, while the 40-H) from water is relatively constant in intensity. These data suggest that near the PZC the 0-H bond points toward the surface with the acidic hydrogen near the electrode. Therefore, as the applied potential approaches the PZC, the 0-H functionality must rotate about the C-0 bond axis. The driving force for this rotation is apparently the presence of Li+ in the outer Helmholtz plane, since the methanol 0-H bands are not observed in the presence of Na+ or K+. The surface methanol species can then solvate the Li+ cation as it is drawn toward the negatively charged surface. As lone pairs of electrons solvate this cation, the 0-H bond naturally rotates about the C-0 bond axis. Two additional factors may contribute to this rotation. First, the acidic hydrogen is not repelled by the negative surface. Second, the anion appears to stablize the hydrogen at the surface, perhaps through hydrogen bonding. In LiC1, interfacial methanol u(0-H) bands are observed at 3256 and 3340 cm-' and in LiBr a t 3256 and 3354 cm-l. The higher 0-H frequency of the second band in the presence of Br- is consistent with the shift to higher frequencies in solution relative to C1- as shown in Table I and discussed in ref 13. These frequency shifts are indicative of hydrogen bonding with the anion. Thus, the two observed bands may be assigned to different interfacial methanol species, one which has its acidic hydrogen directed toward the surface and hydrogen bonded with the surface halide and another with the acidic hydrogen

directed away from the surface. The data suggest that these species exist simultaneously and, in fact, are intimately related. These species may be, for example, hydrogen bonded to each other. Evidence for this surface dimer is the concomitant growth of these bands as the potential approaches the PZC as shown in Figure 7 . For this arrangement on the surface, the lower energy u(0-H) a t 3256 cm-l is assigned to the methanol species with its oxygen atom pointing toward the electrode and the acidic hydrogen pointing away from the surface. The 0-H vibrational frequency of this species would be predicted to be relatively independent of anion type. On the other hand, the higher energy band is dependent on anion. In LiC1, the higher energy band is observed at 3340 cm-l, while in LiBr it is observed as high as 3354 cm-l. The anion dependence of this band suggests that only the species with acidic hydrogens near the surface are involved in hydrogen bonding with surface anions. It is interesting to note that single, linear hydrogen bonding (LHB) between two methanol molecules in bulk solution predominates at low temperatures (-77 OC).'l The frequency of the u(0-H) in LHB species is found to be centered at 3220 cm-l, close to the 3256 cm-l observed a t Ag electrodes. Therefore, the surface species associated with the 3256cm-l band is expected to be the most strongly hydrogen bonded of the two interfacial species observed. The frequency differences between the two observed u(0-H) bands are ca. 80 and 100 cm-' in C1- and Br-, respectively. Chang and co-workers'' have previously reported that surface water exhibits two vl bands depending on supporting electrolyte cation. These results were interpreted in terms of the influence of the cation on the orientation of surface water. Cations of low hydration energy (Cs', Rb+, K+) allow the water dipole to be oriented with its oxygen end toward the metal surface. In the presence of these cations, water gives a narrow, symmetric band a t ca. 3510 cm-l. Cations of high hydration energy (Li+, Na+, Ba2+,Ca2+,Sr2+,Mg2+)orient the water dipole such that the oxygen end is solvating the cation. In the presence of these high hydration energy cations, the v1 water band is observed at ca. 3550 cm-l. The frequency difference for water in these two orientations is ca. 40 cm-l. The frequency difference observed in methanol (80-100 cm-' depending on electrolyte) is of similar magnitude, which supports the hypothesis that one methanol species is oriented with its acidic hydrogen toward the surface, while a second species is simultaneously oriented with its acidic hydrogen away from the surface. Hydrogen bonding of methanol species with the surface anion is further supported by a sudden loss in v ( 0 H) intensity just negative of the PZC. At these potentials, all of the 40-H) bands lose more than 50% of their intensity. This sudden loss in intensity is not observed with the u(C-0) or v(C-H) bands; therefore, the loss of v(0-H) intensity is not attributed to a loss in SERS activity. Rather, the loss of v(0-H) intensity may be attributed to a drastic change in 0-H bond orientation. Just beyond the PZC, the halide is driven from the electrode surface, removing the opportunity for hydrogen bonding of the acidic hydrogen oriented toward the surface. Under these conditions, no drivingforce exists for maintenance of the 0-H bond perpendicular to the surface. As a result, the intensities for this band decrease. In total, these data suggest that both the cation and anion can play an important role in controlling methanol surface orientation. SERS spectra of the v(C-H) frequency region indicate (21) Giguere, P. A.; Pigeon-Gosselin,M. J. Solution Chem. 1988, 17,1007.

50 Langmuir, Vol. 6, No. 1 , 1990

that the methyl group orientation is also sensitive to changes in electrode potential near the PZC. This observation is clearly shown in the plot of I(C-H), /I(CHIsym as a function of electrode potential in G u r e 5. The absolute intensity of each band decreases with a more negative applied potential; however, Figure 5 indicates that the uasP(C-H) intensity decreases more rapidly than the u,,,(C-H) intensity. This observation is consistent with rotation of the methyl group about the C-0 bond axis as the PZC is reached. As a result of this rotation, the uasym(C-H)vibration maintains a significant component perpendicular to the surface; however, the C-H bond orientations may no longer be optimum for coupling of this vibration with the surface electric field. For example, one of the C-H bonds may be approaching a more parallel orientation and, therefore, would be less capable of coupling with the surface electric field. More negative of the PZC, the C-H bond further rotates about the C-0 bond axis until two of the three methyl hydrogen atoms are in close proximity to the metal surface, as shown in Figure 9. In this orientation, the vas ,(C-H) vibration could conceivably be parallel to the surface. One would predict that the result of this orientation would be a loss in intensity of the vas ,(C-H) band at potentials negative of the PZC. As d o w n by the data in Figures 4 and 5, this is, in fact, observed. Comparison of Solvent Structure at Methanol/ Ag and Water/Ag Interfaces. Cations play an important role in controlling the interfacial structure in both of these solvents. In solution, cations are more strongly solvated in methanol than in water.22 This solvation is related to the more basic nature of the oxygen atom on methanol. In contrast to solution behavior, the SERS spectra suggest that water may solvate cations more strongly than methanol at the Ag/ electrolyte interface. The predominant water orientation in the presence of Li+ is with the oxygen end solvating the cation in the outer Helmholtz plane. On the other hand, methanol does not have a preferred orientation in the presence of Li+. Rather, two different orientations are allowed to exist simultaneously in the presence of Li+. This obser(22) Gordon, J . E.The Organic Chemistry of Electrolyte Solutions; Wiley: N e w York, 1975;p 229.

Sobocinski and Pemberton vation may also be the result of the poorer ability of methanol to solvate anions at the surface. In solution, it is well-established that methanol does not solvate anions as strongly as water," due to the smaller dielectric constant and weaker hydrogen bonding of methanol. Consequently, the observation of two methanol orientations may be due to weaker hydrogen bonding with surface C1- as compared to water. These conclusions emphasize the importance of the combination of cation-solvent and anion-solvent interactions in determining interfacial solvent orientation. They further imply that both the dielectric constant (a factor in hydrogen bonding) and basicity (a factor in cation solvation) are important in dictating the orientation of protic solvents such as methanol a t metal surfaces.

Conclusions SERS has been used to study the electrolyte and the potential-dependent orientation of methanol a t Ag electrodes. One of the most notable observations is the presence of two u ( 0 - H ) vibrations from interfacial methanol which can exist simultaneously in the presence of Li+. The two bands are ascribed to different interfacial methanol species, one which has its acidic hydrogen directed toward the surface and another with the acidic hydrogen directed away from the surface. It is postulated that a methanol dimer exists at the metal surface, giving rise to 0-H vibrational bands of different frequency. The v(0-H) of higher energy is anion-dependent, suggesting that the species may be involved in hydrogen bonding with the surface anion. Two methanol species are only observed in the presence of Li+. More weakly solvated cations such as Na+ and K+ cannot compete effectively with the electrode for the methanol electron lone pairs on the oxygen atom. Finally, the importance of both cation-solvent and anion-solvent interactions implies that both the dielectric constant and basicity play a role in determining the surface orientation of methanol. Acknowledgment. We gratefully acknowledge the support of this work by the National Science Foundation (CHE-8614955). Registry No. Ag, 7440-22-4; MeOH, 67-56-1; LEI, 7447-418; LiBr, 7550-35-8; NaBr, 7647-15-6; KBr, 7758-02-3.