380
Langmuir 1986, 2,380-388
the point that the energy level of the coated Pt as well as that of band edges is energetically shifted by illumination. Being different from cases at Pt-coated Si, photoanodic oxidation of pyrrole on bare smooth electrodes can occur without band-edge shifts. Thus, the onset potential of anodic photocurrents at the bare smooth electrodes is more negative than that at the Pt-coated smooth electrode, as shown in Figure 1. The photogenerated electrons migrate to the semiconductor bulk in the presence of the electric field induced by the space charge layer in the illuminated smooth surface where there is scarcity of recombination centers. Electrons accumulated in the bulk can easily flow to the coated Pt on the back side through intermediate energy levels associated with defect^.'"'^ Appreciable photocatalytic activities are then observed. The Ag deposition on an illuminated damaged surface must occur in a different way. Photogenerated electrons are trapped in surface states associated with defects and then transferred to silver ions in solution. (13) Bard, A. J.; Bocarsly, A. B.; Fan, F.-R. F.; Walton, E. G.; Wrighton, M. S.J. Am. Chem. Soc. 1980, 102, 3671. (14) Gerischer, H.; Mein, F.; Liibke, M.; Meyer, E.; Pettinger, B.; ShBppel, R. Ber. Bumenges. Phys. Chem. 1973, 77,284. (15) Meirhaeghe, R. L. Van.; Cardon, F.; Gomes, W. P. Ber. Bunsenges. Phys. Chem. 1979,83, 236. (16) Tench, D. M.; Gerischer, H. J.Electrochem. Soc. 1977,124, 1612.
The energy diagram for the case of photocatalytic decomposition of formic acid is given itl Figure 4c. The onset potential of anodic photocurrents and a Schottky barrier of 0.4eV between the coated Pt and the Si substrate are taken into consideration to illustrate the diagrams. The most likely potential concerning oxidation of formic acid will be close to the hydrogen electrode potential," which is given in the figure. When hydrogen evolution occurs, the energy level of the coated Pt must become matched with that of the hydrogen owing to a change in work function.18 Thus, the diagram given with solid lines seems to hold for the occurrence of hydrogen evolution.
Acknowledgment. This work was supported by Grant-in-Aid for Energy Research No. 59040049 from the Ministry of Education, Science and Culture. The authors are indebted to Professors S. Kawai and T. Kawai and to M. Fujii for preparation of thin Pt films of known thickness which were useful for determining a calibration curve of absorbance vs. Pt film thickness relations, Registry No. Pt, 7440-06-4;Si, 7440-21-3; Ag, 7440-22-4;H2, 1333-74-0; HC02H, 64-18-6; pyrrole, 109-97-7; polypyrrole, 30604-81-0. (17) Capon, A.; Parsons, R. J. Electroanal. Chem. Interfacial Electrochem. 1973, 44, 1. (18) Heller, A.; Aharon-Shalom, E.; Bonner, W. A.; Miller, B. J. Am. Chem. SOC.1982,104,6942.
SERS and SEM of Roughened Silver Electrode Surfaces Formed by Controlled Oxidation-Reduction in Aqueous Chloride Media David D. Tuschel, Jeanne E. Pemberton,* and Joseph E. Cook Department of Chemistry, University of Arizona, Tucson, Arizona 85721 Received April 15, 1985. In Final Form: October 5, 1985 The SERS intensity of specifically adsorbed C1- ions at Ag electrodes whose surface morphologies vary dramatically has been investigated. Reasonable control of the Ag electrode surface morphology is afforded by the controlled electrochemical rate of oxidation and reduction of the Ag surface prior to SERS experiments. Scanning electron microscopy is used to determine surface morphology on these electrodes of varying roughness. It is found that the SERS intensity increases as the diameter of the surface roughness features decreases and the two-dimensional surface concentration of roughness features increases. These observations are in general agreement with theoretical predictions based on an electromagnetic enhancement model for SERS.
Introduction It is well-known that in order to observe the maximum surface-enhanced Raman scattering (SERS) from adsorbed ions or molecules at Ag, Cu, or Au surfaces in electrochemical systems, an oxidation-reduction cycle (ORC) pretreatment of the electrode surface must first be perf0rmed.l Several methods for accomplishing the ORC have been published. These include potential sweep2and potential step3 techniques. Despite the widespread use of these ORC pretreatments in electrochemical SERS, very little work has been performed in an attempt to systematically study and understand the effect of ORC conditions
* Author
to whom correspondence should be addressed.
0743-7463/86/2402-0380$01.50/0
on the morphology and, therefore, electronic properties of the resulting surface. The results of a systematic study of the effect of oxidation and reduction rate of the ORC on the SERS response for adsorbed C1- at a Ag electrode is reported here. This study was motivated by the desire to better understand the relationship between surface morphology and surface electronic structure in the SERS mechanism. The system chosen for study was C1- ions specifically adsorbed at Ag electrodes which give rise to (1) Jeanmaire, D. L.; VanDuyne, R. P. J . Electroanal. Chem. 1977,84, 1.
( 2 ) Pettinger, B.; Philpott, M. R.; Gordon, J. G., I1 J. Chem. Phys. 1981, 74, 934. (3) Fleischmann, M.; Hill, I. R. J. Electroanul. Chem. 1983, 146, 367.
0 1986 American Chemical Society
Langmuir, Vol. 2, No. 4, 1986 381
SERS and SEM of Roughened Ag Electrode Surfaces the v(Ag-C1) vibrational feature at 235 cm-'. This adsorbate system was chosen on the basis of its simplicity. The atomic nature of the adsorbate probed in these studies precludes complications in interpreting SERS intensities that might arise from adsorbate orientational considerations. The gross changes in surface morphology produced with different ORC rates have been monitored by using scanning electron microscopy (SEM). The results of these investigations indicate a definite relationship between electrochemical SERS intensity and surface morphology. This relationship is in general agreement with electromagnetic enhancement predictions.
Experimental Section The Raman system used for these studies consists of a Spex 1403 double monochromator with 1800 grooves mm-' holographically ruled gratings. The detector is a high-sensitivityGaAs photocathode RCA C31034 photomultiplier tube which is thermoelectridy cooled to ca.-25 OC. The spectrometeris interfaced to a Spex Datamate dedicated microcomputerwhich controls the monochromator and performs data acquisition and analysis. A Coherent Radiation Innova 90-5 Ar+ laser provides excitation at 514.5 nm for all spedra reported here. The laser beam was focused on the Ag electrode surface at 60° with respect to the electrode surface normal. The laser power was 200 mW at the cell. Spectral data were acquired at 0.5-cm-' intervals over a 1.0-5integration time. A spectral band-pass of 5.0 cm-' was used for the entrance and exit slits, and the middle slits were set for 6.0-cm-' spectral band-pass. Spectraldata were collected before and after the ORC. The resultant SERS spectra were obtained by subtraction of the former from the latter. Peak areas were determined digitally by forming a straight base line between the limits of 180 and 270 cm-1 and taking the sum over the enclosed area. The spectroelectrochemical cell is of a previously reported design.' An X-Y-2-0 translator allows control of the electrode position and beam angle of incidence to within fl mm and &lo, respectively. A planar disk polycrystallineAg (Johnson Matthey, 99.999%) working electrode whose geometric area was ca. 0.24 cm2was mechanicallypolished to a mirror f i i h with succeeaively finer grades of alumina (Buehler) down to 0.05 pm and rinsed with doubly distilled water. A Pt wire housed in a compartment separated from the main body of the cell by a medium-porosity glass frit served as the counter electrode. A Ag/AgCl wire immersed in the test solution served as the reference and all potentials are reported vs. this electrode. The ORCs were carried out via symmetric double potential steps that ranged from A25 to *225 mV; all SERS data were collected after carrying out a single ORC. A potential of -200 mV was applied during the acquisition of spectral data. The potential was controlled with an EX0 Instruments Model 551potentiostat. A Princeton Applied Research Model 379 digital coulometer was used to monitor the amount of charge passed during the ORC. Solutions of 1.00 M NaCl were prepared by using doubly distilled water. The second distillation was from basic permanganate solution. Fisher brand certified A.C.S. NaCl was used without further purification. Solutionswere deaeratedby bubbling with nitrogen prior to use. An ISI-DS130scanning electron microscope was used to obtain micrographs of the electrode surfaces. The electrodes were thoroughly rinsed with water and allowed to air dry before being placed in the microscope. No coating of the electrode surface or other form of pretreatment was used for microscopic analysis. The microscope tilt angle was 0'. Average nodule widths with standard deviations were obtained from micrographs by measuring the breadth of 10 nodules and calculating the average and standard deviation. The 10 nodules selected were determined, by visual inspection,to be midway in size between the largest and smallest found on the micrograph. Average interfeature distances with standard deviations were obtained from micrographs by measuring the distance between 10 pairs of adjacent modules and calculating the average and standard deviation. Two-dimensional surface (4)Pemberton, J. E.;Buck, R. P. Appl. Spectrosc. 1981, 35, 571.
concentrationsof nodules were determined by blocking off a unit area on the micrograph and counting the number of nodulea within that area. The electrode surface area determinations were made by underpotentiallydepositing(UPD) Pb on Ag immediatelyfollowing the ORC. The electrode was subjected to a symmetric double potential step ORC in 1.00 M NaC1. The electrode was then transferred to a solution containing M Pb(NO,), and 0.10 M KC1. MCB brand certified A.C.S. Pb(NO3),and KCl were used without further purification. The potential sweep rate for the UPD experiments was 5 mV s-'. The amount of charge passed associated with the underpotentiallydeposited Pb monolayer was determined from the area under the UPD stripping wave. Area measurementswere made by using a Lietz No. 3651-30 planimeter. The electrodearea is obtained from the amount of charge passed under the assumption that one monolayer of Pb on Ag corresponds to 310 pC cm-2 as reported by Dickertmann, Koppitz, and Schultze?
Results and Discussion Surface Morphology Control via Potential Step Methods. A variety of methods for performing an ORC pretreatment of Ag electrode surfaces in aqueous chloride media have been reported in the l i t e r a t ~ r e .Potential ~~~ sweep techniques have been used extensively wherein the potential of the Ag electrode is swept to anodic values such that oxidation of the surface to form AgCl occurs. At a predetermined charge, potential, or time value, the direction of potential scan is reversed and the AgCl surface is reduced to give a roughened surface of reformed Ag. Roughened electrode surfaces can be obtained by using a double potential step method or by controlled current electrodeposition in dilute M) solutions of a Ag salt or complex? This assortment of ORC techniques leads to a variety of electrode surface morphologies. Moreover, the surface morphology obtained by using a particular ORC method can vary if an experimental parameter, such as the amount of charge passed, is altered. Evans and co-workers have demonstrated that the surface concentration of Ag nodules formed during an ORC is critically dependent upon the amount of charge passed.' These workers further noted that the nodule size was fairly independent of the total amount of charge passed during potential sweep ORCs. Schultz, Janik-Czachor, and Van Duyne presented scanning electron microscopy data that showed differences in the surface morphologies of Ag electrodes subjected to potential step as compared to potential sweep ORCS.~ The surface morphology is of critical importance in the SERS experiment, because the sizes and shapes of the roughness features affect the associated surface electronic structure and, hence, the optical properties of the metal s u r f a ~ e . ~AJ ~study is presented in this report that correlates the electrode pretreatment with surface morphology and electrochemical SERS intensity. The electrochemical oxidation of Ag to AgCl and its subsequent reduction to Ag has been studied previously.11-14 In addition, the dependence of AgCl surface (5) Dickertmann, D.; Koppitz, F. D.; Schultze, J. W. Electrochim. Acta 1976, 21, 967. (6) Busby, C. C.; Creighton, J. A. J.Electroanul. Chem. 1982,140,379.
(7)Evans, 3. F.; Albrecht, M. G.; Ullevig, D. M.; Hexter, R. M. J . Electroanal. Chem. 1980, 106,209. (8)Schultz,S. G.; Janik-Czachor, M.; Van Duyne, R. P. Surf.Sci. 1981, 104, 419.
(9)Kerker, M.;Wang, D.-S.; Chew, H. Appl. Opt. 1980, 19, 4159. (10)Schatz, G. C.; Acc. Chem. Res. 1984, 17, 370. (11) Giles, R. D. J. Electroanal. Chem. 1970, 27, 11. (12)Katan, T.;Szpak, S.; Bennion, D. N. J. Electrochem. SOC.1974,
_121. _ _ 757. (13, Katan, T.;Szpak, S.; Bennion, D. N. J.Electrochem. SOC. 1973, 120,883.
382 Langmuir, Vol. 2, No. 4, 1986 I
I
Tuschel et al. I
1.OOM NaCi
11
/ i
T
1
-
1/ 100
50 0 -50 -100 E, mV vs. A g / A g C I w i r e
-150
Figure 1. Cyclic voltammogram of a Ag electrode in aqueous 1.00 M NaCl; scan rate = 20 mV s-l.
Time
Figure 2. Typical chronocoulogram obtained from a symmetric double potential step ORC. Q, = 20 mC cm-2. morphology on anodic current density has been demonstrated.15 It is demonstrated here how constant current density, and therefore constant rate of electrode oxidation and reduction, can be controlled potentiostatically. Furthermore, by maintaining constant current density during the ORC, the resultant Ag surface morphology can be controlled. In order to understand how the rate of the oxidation and reduction can be controlled electrochemically, consider a potential sweep voltammogram of a Ag electrode in 1 M NaCl media as shown in Figure 1. The anodic wave is approximately linear and does not limit or peak within the realm of charge passed for Cl- concentrations used in these SERS experiments. Thus by stepping the potential from the open-circuit value to a particular potential associated with the anodic process, the current, and therefore the oxidation rate, can be controlled by the magnitude of the potential step. After a predetermined amount of charge has been passed (20 mC cm-2),the polarity, but not the potential magnitude, is reversed, and the electrode surface will undergo reduction. If the potential applied during the cathodic portion of the ORC is more negative than -75 mV vs. this reference electrode system, the rate of reduction will also be constant. The amount of charge passed is monitored as a function of time with a coulometer from which chronocoulograms are obtained. An example is shown in Figure 2. These chronocoulogramsare essentially linear with time for both the oxidation and reduction processes. A family of chronocoulograms that have the same profile can be generated (14) Gu, H.;Bennion, D. N. J. Electrochem. SOC.1977, 124, 1364. (15) Harzdorf, C. Anal. Chim. Acta 1982, 136, 61.
I 150
I
1
225
300
WAVENUMBER (cm-')
Figure 3. SERS spectra of v(Ag-Cl). (a) j , = 0.44 mA cm-2,j , = 0.76 mA cm-2; (b) j , = 1.0 mA cm-2,j , = 2.1 mA cm-2;(c) j , = 3.5 mA cm-2,j , = 4.3 mA cm-2; (d) j , = 6.6 mA cm-2, j , = 7.6 mA cm-2; (e) j , = 18 mA cm-2, j , = 28 mA cm-2. by varying the magnitude of the double potential step. Only the slopes of both the oxidation and reduction portion will vary with the potential magnitude. The derivative at any point on a chronocoulogram is equal to the current at that time. Therefore, the current, which is equal to the magnitude of the slope, is constant. Consequently, the rate of the ORC is controlled and maintained at a constant value. The ORC potential steps that were applied range from f25 to f225 mV which in 1 M NaCl generates anodic current densities ranging from 0.44 to 27 mA cm-2, respectively. Typically, 97% of the charge passed during the oxidation portion of the cycle is recovered during reduction. It is well-known that ORCs performed under laser illumination yield SERS signals that are more intense than those obtained after an ORC done in the dark.16J7 However, by performing the ORC in the dark, electrochemical control of the roughening process is maintained without a localized photoeffect. Therefore, only the electrochemical rate controls the resulting electrode surface morphology and surface electronic structure in these studies. Relationship of SERS Intensity and Surface Morphology. Spectra of the v(Ag-Cl) region for specifically adsorbed C1- at Ag were obtained at -200 mV subsequent to ORCs carried out at different rates. Representative spectra are shown in Figure 3. All spectra were obtained following electrode polishing and one double potential step ORC. Spectral data were taken prior to and immediately following the ORC. Subtraction of the former from the latter provides a much improved SERS intensity for the v(Ag41) feature by removing much of the Rayleigh tail. Differences in SERS intensities due to imprecise electrode polishing or positioning were minimized in these experimenta as demonstrated by the fact that spectra taken prior to the ORC were nearly identical from trial to trial. The data in Figure 3 demonstrate that as the oxidation rate increases, the resulting SERS intensity of both the v(Ag-Cl) feature and the background increases. The peak area of this vibrational feature can be determined above the background scattering by assuming a straight line background between the limits of 180 and 270 cm-'. Low-intensity SERS spectra are obtained at low oxidation rates. With increasing oxidation rate, an increase in the intensity of the SERS spectra is observed up to ca. 18 mA cmm2.Beyond this rate, the SERS intensity begins (16) Barz, F.; Gordon, J. G., 11.; Philpott, M. R.; Weaver, M. J. Chem. Phys. Lett. 1982, 91,291. (17) Devine, T. M.; Furtak, T. E.; Macomber, S. H. J. Electroanal. Chem. 1984, 164,299.
SERS and SEM of Roughened Ag Electrode Surfaces
Langmuir, Vol. 2, No. 4, I986 383
I
Ii
Figure 4. Scanning electron micrographs of Ag electrodes subjected to symmetric double potential step ORCs. The calibration bar represents a distance of 1Fm. (a) j , = 1.0 mA j , = 2.1 mA c d , magnification 25900% (h) j , = 3.5 mA em-*, j , = 4.4 mA cm-', magnification26000% (c) j a = 9.8 mA cnP,j , = 9.8 mA cnP, magnification 26200x; (d)j . = 18 mA cm-2,j , = 20 mA cm-', magnification 28100X.
to decrease slightly as the rate approaches the largest value of 27 mA cn-' used in this study. There is no a priori reason to expect that the SERS intensity should be a function of the oxidation rate of the Ag electrode. The relationship between these two parameters comes from the dependence of surface morphology on the ORC rate of the Ag electrode. The importance of roughness feature size and shape to electrochemicalSERS has been debated for some time. However, it is accepted that large-scale roughness plays an important role in the SERS mechanism through electromagneticenhancement of the electric field at the surface?JO In order to address the issue of large-scale surface morphology and its relation to ORC rate, and therefore SERS intensity, scanning eledron microscopy (SEM) was performed on electrodes roughened over the same range of current densities for which SERS experiments have been performed. SEM Characterization of Ag Surface Morphology. A micrograph of an electrode subjected to a symmetric double potential step at an anodic current density of 1.0 mA cm-2 and cathodic current density of 2.1 mA cm-z is shown in Figure 4a. Large, elongated features ca. 600 nm in length predominate. These features are clustered and exhibit three-dimensional growth out into the solution. Smaller features of ca. 200-nm diameter are present although the larger forms dominate the surface. A twodimensional surface concentration of Ag nodules from surfaces oxidized and reduced a t low rates cannot be determined because of the depth of the formations. The pits
in the Ag surface, which appear as dark areas in the micrographs, are formed during the anodic dissolution process.12 The pit edges are sites for deposition during the reduction portion of the ORC yielding the fmal roughened Ag surfaces. A micrograph of a silver surface oxidized at 3.5 mA cm-2 and reduced at 4.4 mA cm-z is shown in Figure 4h. The surface concentration of roughness features is approximately 10 nodules pm-*. This value should be considered only a rough estimate, because a small fraction of the surface features are elongated, although the three-dimensional growth does not appear to be as pronounced as in Figure 4a for the slower oxidation rate. The surface consists of elongated features ca. 340 nm long and less numerow hemispheres of ca. 160-nm diameter. Surface pits are clearly seen, and it appears that the reformed Ag OCCUI'(I along the pits and cracks in the surface. An increase in anodic and cathodic current density to 9.8 mA cm-' yields an increased surface concentration of roughness feature to 33 nodules pm-2 as shown by the micrograph in Figure 4c. At this current density, the surface consists of hemispheres of ca.130-nmdiameter and elongated features that range from 220 to 260 nm in length. These two types of features are present in approximately equal amounts. No three-dimensional growth is observed at this rate, and the reformed Ag is spread uniformly across the surface. The Ag reformation appears more random and does not occur only along pronounced surface pits and cracks.
384 Langmuir, Vol. 2, No. 4, 1986 Table I. Ag Surface Morphological Data from SEM as a Function of Anodic Current Density for a Symmetric Double Potential Step ORC av inter0, Jay nodule width, nm mA feature nodules pmT2 av dist, nm cm-2 largest smallest 1.0 370 110 190 f 20 clustered 3d growth 160 f 10 clustered 3d growth 75 3.5 300 70 180 f 20 clustered 3d growth 520 5.3 150 f 10 30 f 10 21 390 75 6.6 9.8 260 130 f 10 39 f 13 33 75 44 210 60 140 f 10 41 f 6 12 210 51 f 6 54 50 100 f 5 15 180 54 40 100 f 10 62 f 5 18
Figure 4d shows a micrograph of a Ag electrode subjected to an anodic current density of 18 mA cm-2 and a cathodic current density of 20 mA cm-2. The surface concentration of roughness features on this surface has risen to 54 nodules pm-2. Small hemispherea of ca. 100-nm diameter predominate. The elongated features ca. 180 nm long make up less than one-half of the Ag deposits. There exists virtually no three-dimensional growth, and the surface concentration of roughness features and interfeature distance are greater than that in Figure 4c for a slower rate. An increase in both the concentration of roughness features and the interfeature distance is explained by the smaller nodule size. In general, these data suggest that increasing interfeature distance,increasing surface concentration of roughness features, and smaller nodulea coincide with increasing ORC rate. Furthermore, surface pits and cracks appear to provide Ag deposition sites for all ORC rates examined. It should be noted, however, that at greater ORC rates, deposition becomes more random and does not occur only at pronounced surface pits and cracks. Surface morphological information from micrographs of electrodes roughened over the same range of current densities for which SERS experiments have been performed is presented in Table I. Due to the Oo tilt angle for which the SEMs presented here were acquired, the only measurement of the surface roughness features that can be made with any degree of accuracy is nodule diameter. Attempts to measure nodule heights did not prove feasible with the available scanning electron microscopy facilities. Diameter meaeurements preclude calculation of the aspect ratio of the roughness features, the ratio of nodule height to radius. This quantity has been shown to be of enormous importance in theoretical treatmenta of electromagnetic enhancement models.l0 However, qualitative assessment of trends is available from treatment of nodule diameters, since previous calculations have demonstrated a dependence of the electromagnetic enhancement on nodule diameter as well as aspect ratio. The results in Table I confirm that increased ORC rates correspond to an increase in surface concentration of roughness features and interfeature distance and a decrease in nodule diameter. At low rates, large elongated nodules predominate that tend to form clusters with large, barren patches of electrode surface that separate them. In addition, three-dimensional-type growth out into solution occurs. At higher rates, the nodule diameters are significantly smaller, predominantly hemispherical, and distributed more evenly across the surface and demonstrate no three-dimensional growth. The morphological data pertaining to reformed silver in Table I are presented as a function of anodic current density, because it is believed that the rate of oxidation dictates the surface morphology in a symmetric double
Tuschel et al. Table 11. AgCl Surface Morphological Data from SEM as a Function of Anodic Current Density for a Single Potential Step Oxidation av interi., 4 nodule width, nm mA feature nodules cm-2 largest smallest dist, nm pm-2 av 0.61 1200 340 720 f 50 clustered 3d growth 2.1 860 340 560 f 40 clustered 3d growth 4.8 630 170 330 30 clustered 3d growth 9.3 310 170 270 f 10 110 f 25 13 12 400 110 170f 10 130 f 20 15 17 310 110 150f 10 160 f 20 18 23 340 86 130 f 20 170 f 10 19
*
Table 111. Ag Surface Morphological Data from SEM as a Function of Cathodic Current Density at Constant Anodic Current Densityn av inter0, i, nodule width, nm mA feature nodules cm+ largest smallest av dist. nm rrm-2 0.44 540 140 180 f 10 low aspect 2.1 360 140 160 f 20 60 f 10 22 160 f 20 60 f 10 19 3.7 220 120 150 f 10 60 f 10 21 5.2 210 100 10 210 80 170 f 10 60 f 10 22 180 80 150 f 10 55 f 5 23 16 22 210 80 150 10 50 f 7 25
*
ja = 12 f 2 mA cm-2.
potential step ORC. Two types of experiments were performed to test this hypothesis. First, single potential step oxidations were carried out over the same range of current densities for which SERS experiments had been performed and scanning electron micrographs were obtained. Morphological information on these electrochemically fabricated AgCl surfaces is presented in Table 11. The dimensions of the AgCl features are larger than those of the corresponding reformed Ag features as indicated by the data in Tables I and 11. However, the trend in surface feature size with rate is the same for surfaces subjected to a full ORC and those subjected to oxidation only. Large AgCl nodules, clusters, and three-dimensional growth are observed at low oxidation rates for the oxidized Ag surfaces. As the rate is increased, smaller diameter AgCl nodules, larger interfeature distances, and larger two-dimensional surface concentration of AgCl features are observed. In addition, the AgCl nodules are more evenly distributed at high oxidation rates. These data suggest that the oxidation rate plays the major role in determining the surface morphology of a silver electrode subjected to a symmetric double potential step ORC. A second type of experiment to verify the critical role played by oxidation rate was also performed. The effect of reduction rate on the surface morphology was determined by carrying out asymmetric double potential step ORCs in which the oxidation rate was kept constant and the reduction rate varied. Morphological data taken from micrographs of these surfaces are shown in Table III. The electrodes in these experiments were oxidized at 12 mA cm-2, and the rate at which they were reduced was varied from 0.44 to 22 mA cm-2. This anodic current density is approximately midway between the upper and lower current limits which yield the extremes of surface morphology. Therefore, effects due to reduction rate should be revealed by these experiments. As the data in Table I11 suggest, the surfaces reduced at the slower rates have roughness features with diameters slightly larger than those reduced at all of the other rates. However, the surface morphology is largely independent of the reduction
SERS and SEM of Roughened Ag Electrode Surfaces
Langmuir, Vol. 2, No. 4, 1986 385
rate. The independence of surface morphology over nearly the entire cathodic current density range studied further supports the contention that the oxidation rate is primarily responsible for determining the surface morphology of a Ag electrode subjected to a double potential step ORC. The SEM data presented above clearly establish the dependence of surface morphology on oxidation rate. Specifically, the average nodule diameter, interfeature distance, and two-dimensional surface nodule concentration change as a function of oxidation rate. An understanding of the relationship between surface morphology and oxidation rate comes from consideration of the chemical mechanisms associated with the ORC process. Katan and co-workers have extensively studied the electrochemical oxidation of Ag to AgCl and have proposed a mechanism that consists of the following steps? Ag e Ag+
+ e-
Ag+ + (n+l)Cl- a AgC1n+l-"
(dissolution)
(1)
(complex formation) (2)
AgC1,+l-n(a) e AgCl,,,-"(b)
(solution transport) (3)
+ nC1-
(deposition & growth) (4) It has been proposed that a t lower oxidation rates, eq 4 is the rate-determining step. Under these conditions, deposition takes place at energetically more favorable sites of continued AgCl growth. Consequently, the roughness features are large and develop outward into solution. The large elongated features seen in Figure 4a support this hypothesis. At faster oxidation rates, these researchers conclude that eq 3 is the rate-determining step. The imposition of a "high rate" potential in the experiments reported here would decrease the time allowed for the solution transport step such that the number of AgCl nodules would increase. This reasoning would lead to the further conclusion that the nodules would be smaller and that the distance between them would increase. The micrograph in Figure 4d supports these results. The changes in surface morphology reflected by the data in Table I1 are consistent with eq 4 as the rate-determining step as low rates and eq 3 as the rate-determining step at high rates. Furthermore, if either eq 1 or 2 were the rate-determining step, the amount of charge recovery of the reduction step might not be expected to approach loo%, because these solution AgCln+l-" species would have time to diffuse away from the surface into the bulk solution. This is clearly not the case since the charge recovered on the reduction portion of the ORC is equal to or greater than 97% of the charge passed during the oxidation portion. Furthermore, the high C1- ion concentration drives step 2, thereby making it unlikely that it would be the rate-determining step. Jaenicke and co-worker@ and Briggs and Thirskl9 have proposed that the reduction of AgCl deposited on a Ag electrode surface takes place at the Ag/AgCl interface and at the AgCl grain boundaries. The independence of surface morphology with respect to reduction rate, as indicated by the data in Table 111, agrees with this reduction model. That is, no effects of solution transport are observed, except at extremely low reduction rates. Therefore, it is the AgC1n+l-n s AgCl(s)
(18)Jaenicke, W.; Tischer, R. P.; Gerischer,H. Z . Elektrochem. 1956, 59, 448. (19) Briggs, G. W. D.; Thirsk, H. R. Trans. Faraday SOC.1952, 48, 1171.
T
17cm2
I1 -200 -3CO
-400 -500 -600 -700
E,mV vs A g i A g C l
Figure 5. Cyclic voltammogram of Pb UPD at a roughened polycrystalline Ag electrode in aqueous M Pb2+and 0.100 M KCl; scan rate = 5 mV s-l.
oxidation process that controls the surface morphology developed during the reduction process. Consequently, the SERS intensity depends on the rate at which the Ag electrode is oxidized. Hence, the surface morphology is quite uniform and can be predicted and controlled according to previously established electrochemical rate theory. Surface Area Considerations. The relationship between electrochemical rate and surface morphology has been established. Therefore, the relationship between the ORC conditions and the resulting SERS intensity for specifically adsorbed C1- can be addressed. One issue that deserves comment is the question of whether or not the different ORC conditions produce surfaces of different surface areas. An increase in surface area would allow a higher population of C1- species to be adsorbed at the Ag surface. This, in turn, would allow a greater number of Raman scatterers to be sampled during the SERS experiment yielding a larger SERS intensity. The total anodic charge passed during all of the ORC pretreatments presented here was kept constant. Therefore, the surface area of the Ag surfaces should remain constant. Verification of this assumption comes from the underpotential deposition (UPD) of P b monolayers onto the various Ag electrodes produced for the SERS experiments. Underpotential deposition of foreign metal monolayers is the formation of the monolayer on the substrate electrode at potentials positive of the reversible Nernst potential for the metal couple at which bulk deposition of the metal onto the monolayer occurs. A significant potential difference usually separates the underpotential and bulk deposition processes. This feature is shown by the cyclic voltammogram for the UPD of P b onto a roughened polycrystalline Ag electrode from lob3M Pb2+/0.1M KC1 in Figure 5. The surface area of the Ag electrode is calculated from the amount of charge required to strip the P b monolayer from the Ag surface. In these calculations, a value of 310 pC cm-2 for the formation of a Pb monolayer was used as determined previously by Dickertmann, Koppitz, and S ~ h u l t z e . ~ The results of UPD experiments in which the surface area of Ag electrodes subjected to double potential step ORCs giving a series of rates for which SERS experiments were performed are shown in Table IV. These data confirm the expectation that the surface area of the Ag electrodes remains reasonably constant and independent of the rate at which the ORC is performed. Small fluc-
386 Langmuir, Vol. 2, No. 4, 1986
Tuschel et al.
r
Table IV. Ag Electrode Surface Area by UPD of Pb a s a Function of ORC Rate JW
mA cm-* 1.0 3.6 9.9 18
25
i, mA cm-' 1.6 3.8 9.8 17 22
lead UPD coverage, pC 150 4 160 2 180 4 180 4 140 f 0
** *
*
true electrode area, cm2 0.50 0.53 0.60 0.60 0.47
roughness facto? 2.1 2.2 2.5 2.5
e
'
>
e e
u)
z
ZI F
1.9 e
a Roughness factor calculated as ratio of true electrode area determined from Pb UPD to geometrical electrode area of 0.24 cm2.
tuations in the electrode area are observed. However, these fluctuations are not large enough to account for the approximate 2 orders of magnitude difference in the SERS intensity measured for these electrodes as shown in Figure 3. These results clearly indicate that increased surface area is not the cause of the increased SERS intensity observed with an increase in ORC rate. Similar conclusions regarding the role of increased surface area after an ORC in electrochemical SERS have been reported previously. Hupp, Larkin, and Weaver have reported differential capacitance measurements of surface roughness of differently anodized Ag surfaces.20 These researchers found that Ag electrodes roughened by an ORC in which 10-40 mC cm-2 of anodic charge was passed exhibited roughness factors between 1.5 and 2.0. Implications for SERS Mechanisms. A plot of SERS intensity for the v(Ag-Cl) band for specifically adsorbed C1- on electrodes subjected to different controlled ORC rates in 1 M NaCl as a function of average Ag nodule diameter measured from SEMs of these electrode surfaces is presented in Figure 6. These data demonstrate that the SERS intensity of the v(Ag-Cl) band at -200 mV increases as the Ag particle diameter decreases until a maximum in intensity is reached at a particle diameter of ca. 90-100 nm. These data are qualitatively consistent with electromagnetic models for SERS. Theories have been proposed for SERS which suggest that large electromagnetic fields associated with the oscillations of conduction band electrons of small metal particles or surface roughness features are, at least partially, responsible for the surface enhancement phenomenon.21$22Furthermore, interaction between small metal particles due to the Coulombic fields of their electric dipole moments yields an even greater e n h a n ~ e m e n t . ~Calculations ~ of this effect predict an optimum distance between the small metal particles or roughness features of a particular size at which the maximum enhancement is obtained.23 In the limit of a twodimensional concentration of roughness features on a flat dielectric medium, a relatively smooth surface is produced, and the enhancement decreases accordingly. It is important to note from Table I that an increase in the surface concentration of roughness features for the surfaces studied here is accompanied by an increase in the interfeature distance and a corresponding increase in the SERS intensity. These data further suggest that electromagnetic effects may be responsible for the observed trends. The coincidence between experimental data from electrochemically fabricated Ag surfaces and electromagnetic theory has not been previously established in the literature. (20) Hupp, J. T.;Larkin, D.; Weaver, M. J. Surf. Sci. 1983,125,429. (21) Gersten, J. I. J. Chen. Phys. 1980, 72, 5779. (22) Moekovita, M. J. Chem. Phys. 1978,69,4159. (23) Aravind, P. K.; Nitzan, A.; Metiu, H. Surf. Sci. 1981, 110, 189. (24) Barber, P. W.; Chang, R. K.; Massoudi, H. Phys. Reu. Lett. 1983, 50, 997.
8
e
e
e
II I l i l l l l l l I . I I e e
0
40
80
120
160
200
240
PARTICLE DIAMETER ( n m 1
Figure 6. Normalized SERS intensity of v(Ag-Cl) as a function of average particle diameter in nanometers determined from SEM.
However, the data presented in this report suggest that the ORC functions to fabricate Ag surfaces with defined morphologies which affect SERS intensities, at least in part, through electromagnetic effects. From a purely theoretical basis, it is difficult to predict what combination of nodule size, nodule surface concentration, and interfeature distance will produce the maximum enhancement. The magnitude of enhancement calculated from electromagnetic theory depends sensitively on the specific shape and size of the roughness feature used as the model. Calculations have been made by using sphere^,^^^^^^ prolate and randomly roughened s u r f a ~ e as a ~models. ~ ~ Electrochemical control of the size and shape of the surface roughness will never be achieved at the level necessary to validate or invalidate these models. It is useful, however, to compare the trends in SERS intensity (and therefore, enhancement factor) observed with surfaces fabricated by controlled electrochemical ORCs with the trends predicted by theoretical calculations. Comparisons between the experimental results and the calculations can only be made on the basis of particle diameters. This parameter is the only physical descriptor that could accurately be quantified in the studies presented here. Previous calculations on noninteracting spheroids have demonstrated that the field enhancement decreases rapidly as the size of the spheroid i n c r e a s e ~ . ~ J ~As. ~ ~ - ~ ~ discussed above, the data presented in this report are generally consistent with these calculations. A discrepancy does exist, however, between the optimum particle diameter predicted from theory and the particle diameter at which the maximum SERS is observed for v(Ag-Cl). Calculated values of particle diameter at which the optimum enhancement factor should be realized typically range from 10 to 50 nm depending on the shape of the Kerker, M.; Wang,D.-S.; Chew, H. Appl. Opt. 1980, 19, 2256. Kerker, M.; Wang, D.43.; Chew, H. Appl. Opt. 1980, 19, 3373. Kerker, M. Acc. Chem. Res. 1984,17, 271. Zeman, E. J.; Schatz, G. C. In Dynamics on Surfaces; Pullman, B.; Jortner, J.; Nitzan, A.; Gerber, B.; Eds.; D. Reidel Publishing: Dordrecht, Holland, 19&4; p 413. (29) Barber, P. W.; Chang, R. K.; Massoudi, H. Phys. Reu. B 1983,27, (26) (26) (27) (28)
7251. (30) Laor, U.; Schatz, G. C. J. Chem. Phys. 1982, 76, 2888. (31) Laor, U.; Schatz, G. C. Chem. Phys. Lett. 1981, 82, 566. (32) Ayra, K.; Zeyher, R.; Maradudin, A. A. Solid State Commun. 1982, 42, 461.
Langmuir, Vol. 2, No. 4 , 1986 387
SERS and SEM of Roughened -Ag Electrode Surfaces roughness feature assumed in the calculations. These values are significantly smaller than the particle diameter of 90-100 nm at which the maximum SERS intensity for the v(Ag-C1) band is observed. This discrepancy most likely arises from the fact that the relatively crude surfaces afforded with the controlled electrochemical ORC procedures presented here do not closely approximate the model systems used for calculations. Correlations between data presented here and calculations on the interaction between surface roughness features are, unfortunately, not as straightforward. Metiu has suggested that surface nodules in close proximity form a grating, thereby yielding a smaller enhancement than a single, segregated surface feature.33 howe ever, an optimal distance between surface roughness features is known to produce an increase in the surface enhancement due to interaction between the enhanced electric fields of each feature. Therefore, the effect of interaction between nodules would be to maximize the enhancement for a “low” concentration of surface roughness features. The enhancement would decrease if the roughness features were brought too close together to form, in Metiu’s terminology, a grating. The term low concentration implies a greater internodule distance. The data in Table I suggest that interaction between the surface roughness features may be important in electrochemical SERS. The surfaces that give rise to the largest SERS intensity are those possessing the largest internodule distance. The sizes and shapes of the roughness features are also important considerations in understanding the response of these surfaces as discussed above. A critical feature of the data presented here is that the enhancement associated with roughness feature size and internodule distance cannot be differentiated by these experiments, because both parameters change with different ORC rates. It is anticipated that excitation studies will provide further insight into the electromagnetic effects occurring in the SERS from the electrochemically controlled surface electrode morphologies under investigation. These studies are currently under way in this laboratory and will be reported in a future publication. One final feature of these experiments must be emphasized. The changes in SERS intensity have been associated with large-scaleroughness features only. However, the possible effects of “atomic”- or small-scale roughness features cannot be discounted, because the scanning electron microscope used for these experiments is unable to resolve features smaller than 40 nm. I t is conceivable that changes in surface roughness at this level also contribute to the results presented in this report. A series of experiments were designed to better assess the contributions of atomic-scale roughness to the trends presented above. These experiments consisted of first preparing a series of electrodes according to the controlled ORC procedures discussed previously. SERS spectra were acquired on these electrodes in the v(Ag-Cl) frequency region. The electrodes so prepared were then subjected to a negative potential excursion to -800 mV and back to -200 mV at 5 mV s-l. This treatment has been reported to destroy the majority of atomic-scale roughness features due to the desorption of stabilizing adsorbates at potentials negative of the potential of zero ~ h a r g e . ~The ~ - ~SERS ~ (33) Aravind, P. K.; Metiu, H. Surf. Sci. 1983, 124, 506. (34) Wetzel, H.;Gerischer, H.; Pettinger, B. Chem. Phys. Lett. 1981,
78,392. (35) Wetzel, H.; Gerischer, H.; Pettinger, B. Chem. Phys. Lett. 1981, 80., 159. ~~
(36)Owen, J. F.; Chen, T. T.;Chang, R. K.; Laube, B. L. Surf. Sci. 1983, 131, 195.
e
BEFORE 0 AFTER
e*
e
0
e 0
se
0 0
e
e
0
40
80 120 160 ZOO PARTICLE DIAMETER ( n m )
240
Figure 7. Normalized SERS intensity of u(Ag-Cl) as a function of average particle diameter in nanometers determined from SEM both before ( 0 )and after (0) the destruction of a majority of the
atomic-scale roughness by a negative potential excursion.
intensities of the v(Ag-C1) band were then redetermined after the destruction of the majority of the atomic-scale roughness features. These intensities are plotted in Figure 7 as a function of large-scale roughness feature diameter along with the intensities measured before the destruction of atomic-scale roughness. The absolute intensities of the v(Ag-Cl)feature decrease 5 0 4 0 % with the destruction of atomic-scale roughness. For large-scale Ag diameters larger than ca. 140 nm, the overall enhancement is below the minimum detectable level. Scanning electron microscopy of the Ag electrode surfaces after controlled ORC treatments and after the destruction of the atomic-scale roughness demonstrates that the large-scale morphology of the surface is not significantly altered from that presented in Table I by the negative potential excursion. Therefore, the changes in S E W intensity shown in Figure 7 must be due to changes in surface roughness smaller than ca. 40 nm as proposed previ~usly.~~-~~ The data in Figure 7 indicate that the trend in SERS intensity with large-scale Ag feature diameter remains unchanged after the destruction of atomic-scale roughness. Furthermore, the large-scale Ag feature diameter at which the maximum SERS intensity is observed also remains unchanged. These results are quite significant, because they suggest that electromagnetic effects make important contributions to the surface enhancement phenomenon in electrochemical systems.
Conclusions A study has been performed in which the SERS intensity of the v(Ag-Cl) vibrational feature at Ag electrodes in 1M NaCl is found to be dependent upon the large-scale morphology of the electrode surface. The large-scale morphology is generated via a double potential step oxidation-reduction cycle for which at least 97% of the charge passed during the oxidation step is recovered during reduction. Scanning electron micrographs of the electrode surfaces reveal different surface morphologies that are dependent upon the ORC rate. At low oxidation rates, the roughness features are large, elongated, and clustered and grow out into bulk solution. Very weak SERS intensities are ob-
Langmuir 1986,2, 388-392
388
tained at these low rates. As the oxidation rate is increased, the features become smaller, hemispherical, and well separated from each other. Also, the two dimensional surface concentration of roughness features and the interfeature distance increases. It is this latter surface structure that yields the most intense electrochemical SERS response. A relationship between the ORC rate and surface morphology has been shown to exist. Furthermore, it appears that the oxidation process determines the surface morphology, surface electronic structure, and SERS intensity subsequent to reduction. A correlation between SERS intensity and large-scale surface morphology is interpreted in terms of significant contributions from electromagnetic effects in electrochemical SERS. The corre-
lation between SERS intensity and large-scale surface morphology is found to be independent of the presence of atomic scale roughness. Through electrochemical control of the roughening process, the surface morphology and surface electronic structure necessary for SERS can be regulated. The results presented here confirm the results of previous investigations of the morphology of SERSactive surfaces.
Acknowledgment. We gratefully acknowledge financial support of this research by the National Science Foundation (CHE-8309454). We are also grateful to one of the reviewers of this paper who suggested the negative potential scan experiments. Registry No. C1-, 16887-00-6;Ag, 7440-22-4; Pb, 7439-92-1.
Potential-Dependent Water Orientation: An in Situ Spectroscopic Study -
M. A. Habib* and J. O'M. Bockris Department of Chemistry, Texas A&M University, College Station, Texas 77843 Received June 17, 1985. In Final Form: November 25, 1985 This paper concerns the possible orientation of water molecules at the surface of platinum electrodes in aqueous NaF solutions. The work was carried out by using in situ FTIR spectroscopy in SNIFTIRS (substractively normalized interfacial Fourier transform infrared spectroscopy) mode. The intensity of the IR spectral band around 3300 cm-', characteristic of 0-H vibration, was found to increase from about 0.0 V on the NHS when the potential was shifted anodically. Corresponding bands were observed for the 0-D stretch in a 50-50 H20-D20 mixture. However, the intensity of these were 4-8 times less than those for the 0-H vibration. This difference is explained in terms of their differences in hydrogen bonding. The dependence of the intensity of OH and OD vibration upon potential fib a model for water orientation in which water molecules lie flat (i.e., both oxygen and hydrogen atom of a water molecule lie on the surface) near the pzc and gradually orient with their oxygen end toward the surface with the increase of electrode potential in the anodic direction. The results would also probably fit other models in which the water dipoles change their direction on change of electrode charge.
Introduction A consideration of the effect of water structure in the double layer on electrode kinetics was first made by Bockris and Potter.' A model of water in various orientations at the surface was published by MacDonald and Barlow.2a A two-state water model of eat simplicity was introduced by Mott and Watts-Tobin,gbut it was written in terms of a condenser in parallel model and without water-water interactions. The condenser series model, with water-water interactions, was first published by Bockris, Devanathan, and Muller: Thefi7 showed that the adsorption behavior of certain solutes in aqueous solutions can be predicted correctly, largely by considering the energy of adsorbed solvent dipoles, which undergo reorientation as the electric field in the double layer changes direction and strength. At high field strengths, organic molecules are expelled from the double layer by the competing solvent dipoles, which interact more favorably with the field. Since these founding contributions to water molecule theory, improvements of the two-state water model have been given by various authors,916 taking into account dimers and clusters of water dipoles and orientation of *Present address: Electrochemistry Department, General Motors Research Laboratories, Warren, MI 48090-9055.
0743-7463/86/2402-0388$01.50/0
dipoles parallel to the electrode surface. Recent models also include consideration of electron-density distribution at the metal ~urface'~J'and a Monte Carlo simulation of several array models of dipoles at a metal s ~ r f a c e . ' ~ J ~ In spite of these lively modelistic developments, no direct evidence has been obtained in respect to the orientation of water molecules on the surface, (cf. Neff et al.18), although Bewick et al.lgconcluded from an infrared (1) Bockrie, J. OM.; Potter, E. C. J. Chem. Phys. 1952,20,614. (2) MacDonald, J. R. J. Chem. Phys. 1954,22, 1857. (3) Barlow, C. A., Jr. In Physical Chemistry: An Advance Treatise; Eyring, H., Ed.; Academic Preae; New York, 1970. (4) Mott, N. F.; Watts-Tobin, R. J. Electrochim. Acta 1961, 4 , 79. (5) Watta-Tobin, R. J. Philos. Mag. 1961, 6, 133. (6) Bockria, J. O'M.; Devanathan, M. A. V.; Miiller, K. Proc. R . SOC. London, Ser. A 1963,274, 55. (7) Bockris, J. O'M.; Gileadi, E.; Miiller, K. Electrochim. Acta 1967, 12, 1301. (8) Bockria, J. OM.; Habib, M. A. J.Electroanal. Chem. 1975,65473. (9) Bockris, J. OM.; Habib, M. A. Electrochim. Acta 1977, 22, 41. (10) Dameskin, B. B.; Frumkin, A. N. Electrochim. Acta 1974,19,173. (11) Dameskin, B. B. J . Electroanal. Chem. 1975,59, 229. (12) Parsons, R. J. Electroanal. Chem. 1975, 59, 229. (13) Parsons, R.; Reeves, R. M. J.Electroanal. Chem. 1981,123, 141. (14) Fawcett, W. R. J. Phys. Chem. 1978,82, 1385. (15) Fawcett, W. R. Isr. J. Chem. 1979, 18,3. (16) Schmickler, W. J. Electroanal. Chem. 1983, 157, 1. (17) Schmickler, W.; Henderson, D. J.Electroanal. Chem. 1984,176, 383. (18) Neff, H.; Lange, P.; Roe, D. K.; Sass, J. K. J . Electroanal. Chem. 1983, 150,513.
0 1986 American Chemical Society
