The observation of solvated metal ions in the double layer region at silver electrodes using surface enhanced raman scattering

J. Electroanal. Chem., 146 (1983)367-376

367

Elsevier Sequoia S.A., Lausanne - Printed in The Netherlands

T H E O B S E R V A T I O N O F SOLVATED M E T A L I O N S IN T H E D O U B L E LAYER R E G I O N AT SILVER E L E C T R O D E S U S I N G SURFACE E N H A N C E D RAMAN SCA'ITERING

M. FLEISCHMANNand I.R. HILL Department of Chemzstry, The Umverstty, Southampton S09 5NH (Great Britain)

(Received 2nd August 1982)

ABSTRACT The SERS spectra of water observed when halide ions are adsorbed to roughened silver electrodes is shown to be dependent upon the nature of the supporting electrolyte cation. This cation dependence as well as the potential dependence for any given cation are interpreted as being due to the progressive desolvation of cations with increasing radius and with increasingly negative potentials, the cations being adsorbed as solvent separated ion pairs,

INTRODUCTION The observation of surface enhanced Raman scattering (SERS) from water and halide ions at silver electrodes [1-3] and from water and the pseudo-halide cyanide ion [4] has been reported previously. For the system A g / 1 M KCI the intensity of the OH symmetric stretching mode (p(OH)) was found to follow that of the v(AgCl) mode at - 240 c m - ~ and this was interpreted in terms of the coadsorption of water and the halide ions. For the system A g / 1 M KI two bands were observed in the v(OH) region of the water spectrum. The further investigation of the origin of this behaviour has shown that the spectrum of adsorbed water is more dependent on the nature of the alkali metal cation than the nature of the halide ion. The results of this investigation and an interpretation of the spectra are presented in this paper. EXPERIMENTAL The Coderg T800 spectrometer and associated electrochemical instrumentation and methods have been described elsewhere [1]. Silver electrodes were constructed from 99.99% silver rod supplied by Johnson Matthey Ltd. and were polished using 0.5 # M alumina prior to use. All solutions were prepared using triply distilled water and AnalaR grade reagents. All potentials quoted in this paper are with respect to the saturated calomel electrode (SCE). Electrodes were electrochemically roughened using a 5 s pulse ( - 0 . 1 to +0.3 V for chloride solutions and - 0 . 6 to - 0 . 1 5 V for 0022-0728/83/$03.00

© 1983 Elsevier Sequoia S.A.

368

iodide) and, particularly in the case of iodide, the roughening was carried out in the presence of the laser beam (100 mW/514.5 nm). Partial photolysis of the silver halide layers was found to lead to an overall increase in SERS intensity but no other changes in the spectra [5,6]. This procedure was employed in these experiments owing to the inherent weakness of the bands being studied. SURFACE SPECTRA U S I N G SOLUTIONS C O N T A I N I N G C H L O R I D E IONS

In a previous study [1] the v(AgC1) vibration was observed at 240 c m - l (using 1 M KC1 at - 0 . 2 V) while the OH stretching frequency for coadsorbed water was at 3498 cm-1. Subsequent measurements have shown that both bands shift to slightly lower frequencies as the potential is made more negative, see Table 1 (a shift of the v(AgC1) band has also been reported by Pettinger et al. [7]). A weak shoulder was also found at - 3570 cm-1 and it was suggested that this might be attributed to the antisymmetric stretching mode of the adsorbed water [ 1]. However, the present study (especially of the coadsorption of water with iodide ions described later) has shown that this shoulder should be attributed to a second type of adsorbed water. In contrast to the essentially symmetric shape of the v(OH) band when KC1 solutions are used, this band is found to be highly asymmetric in NaC1 (located at 3553 cm-~ at - 0 . 2 V in 1 M NaCI) with far more intensity being concentrated on the low frequency side (Fig. 1). This asymmetry suggests the presence of a second form of water to lower frequency although it is also possible that the two hydrogens are not equivalent. Figure 1 also shows that at - 0 . 5 V the intensity of the v(AgC1) band has fallen slightly whereas the intensity of the band due to adsorbed water (now at 3540 cm-1) has increased especially on the low frequency side. The earlier

TABLE 1 The potential dependence of the SERS bands of water obtained using 1 M chloride solutions and their halfheight bandwidths (Apt/2). Note that K + and Cs + solutions also lead to a very weak band - 3570 c m - i [1] which will be discussed in the following section on iodide Potential/V

-0.1 -0.2 -0.3 -0.4 -0.5 -0.6 -0.7

v(AgC1)/cm - l

p(OH) of adsorbed w a t e r / c m - l Mg 2+

Li +

Na +

K+

Cs +

K+

3516 3516 3516

3522 3522 3520 3515 3510

3557 3553 3549 3545 3540 3527 3495

3499 3498 3488 3479 3468

3501 3497 3492 3485

241 240 237 234 233 232 -

71

72

Avl/2 at --0.1 V / c m - t 180

170

170

369

j ~b

is

15

~5 -3,540

r---3553

¢,, 3800

~

J 000

360o

3800 r--236

/ 310

150

A V/cm-1

Fig. 1. SERS from a silver electrode electrochemicallyroughened in 1 M NaCI. The band pass was 1 cm-] for the 6 cm-i band and 5 crn-] for the others (100 mW/514.5 nm excitation).

observation that in KC1 solutions the intensity of the v(OH) band is proportional to that of the p(AgC1) band [1] must therefore have been fortuitous. A further difference between the spectra for these two alkali metal cations is that the v(OH) band remains intense at more negative potentials for NaC1 than for KC1 solutions. Marked changes in the spectra have also been found for other changes in the cation. For 1 M LiCl a weaker and broader u(OH) band was located at 3522 cm -1 at - 0 . 2 V (AVl/2 = 170 cm - I compared to Av~/2 = 71 cm -~ in 1 M KCI) and this band remained weak at more negative potentials. For 1 M CsC1, a sharp band (Av]/2 --72 cm -1) was found at the same frequency as for KC1 although the peak height was only - 0.07 times the height of the 240 c m - i band compared to 0.2 in the case of KCI. As would be expected, the spectra for 1 M MgC12 resembled those for 1 M LiC1 with a band for v(AgC1) at 240 cm-~ (as for all the other halides) but with a v(OH) band at 3516 cm-1 o f AI, I / 2 = 180 c m - I . The integrated intensity of

370

this band in LiC1 and MgC12 solutions was much lower than, for example, that using NaC1. Finally, in 1 M HCI, the strong band at 240 c m - 1 was accompanied by a new band at 2964 cm-1 whereas only a weak band could be detected at - 3500 cm-1; this weak band may have arisen from low levels of alkali metal ions in the solution. The intensity of the 2964 c m - l band was reduced on adding potassium ions to the electrolyte whereas the intensity of the 3500 cm-~ band increased. It seems likely that the 2964 c m - l band should be assigned to the l'a(H30+) mode of H 3 0 + species [8] and that the coadsorbed water is associated with the alkali metal cations, which are more strongly adsorbed. The spectra of acidic solutions and of mixed electrolytes will be discussed in detail elsewhere [9]. Figure 1 also shows that the spectrum in the very low frequency region ( < 11 c m - l) is dependent on the nature of the cation: the band is shifted from 8 c m - ~ in KC1 to 6 c m - i in NaC1 and can be seen to increase in intensity as the potential is made more negative, indicating a connection with adsorbed water rather than the halide. In ref. 1 we reported that this band shifted with different halides and we now know this to be incorrect. The error probably arose from using NaBr and NaI instead of the potassium salts, coupled with poor background subtraction. We will discuss the behaviour of this band in a further publication [9].

TABLE 2 The potential dependence of the SERS bands of water obtained using 1 M alkali metal iodide solutions. Note the varying potential ranges over which the spectra were obtained Potential/V

~(OH) of adsorbed w a t e r / c m Li +

Na +

l

K+

Rb +

-0.6 - 0.7

3572

-0.8

3538

3571

-0.9

3538

3571

1.0

3538

3558

1.1

3511

3542

1.2

3485

3521

1.3 - 1.4 - 1.5

3470 3447 3443

3483 3454 3439

-

-

-

-

3552 3493 3552 3490 3552 3487 3552 3486

3556 3496 3556 3496 3555 3491 3555 3488

Cs* 3570 3502 3568 3497 - 3565 3497 3496 3493

371 S U R F A C E SPECTRA U S I N G S O L U T I O N S C O N T A I N I N G I O D I D E IONS

Iodide ions are strongly adsorbed at silver electrodes and also give rise to SERS over a wider range of potential than chloride ions. The effects of the nature of the metal cation on the surface spectra are therefore best investigated in iodide solutions. It has already been reported that, when using 1 M KI, two bands can be seen in the OH stretching region [1]. Figure 2 shows the potential dependence of these bands. The frequencies are given in Table 2 and it is clearly seen that the lower frequency band ( - 3490 cm -]) is favoured at more negative potentials. Figure 3

3538 -08V ~

Ll'b3 5 7 ~ lil/lillB~1 3493

-0.9V

3490

No+

3486

~

-1.2v. ~

3ebo

c:~+

3,'o0

A~/cm-I

~;oo

3ioo

Fig. 2. Potential dependence of the SERS of water on a silver electrode roughened in 1 M KI. Fig. 3. Variation of the SERS of water with the metal ion. 1 M MI solutions used with the roughened silver electrode potentiostated at - 0 . 9 V. The intensity scale is expanded 2 x for Li + and Cs + .

372

gives a comparison of the spectra obtained at - 0 . 9 V for a series of alkali metal iodides. T w o different types of spectra are apparent, the broad bands for Li and N a and the narrower double peaks for K, Rb and Cs. The lower frequency band at - 3490 cm-1 is clearly favoured at a given potential as one progresses down the group. The potential dependences are listed in Table 2, and it is interesting to note that Li and Na give rise to spectra at more negative potentials than the heavier metals while K, Rb and Cs each show their characteristic spectra over slightly displaced potential ranges. The spectra given in Fig. 4 show that the two stretching frequencies at - 3550 and - 3 4 9 0 cm -I are associated with two bending modes at - 1620 and - 1590 3571

A.

J

~

-09V

1631

,--'3493

3552-

1596

|

1628-jt

B.

J 3486

355Z'-'~I |

1591

K1 at -1 2V

rvl

C. I

!

3800

IEO0 A~

m -1

Fig. 4. A comparison of the SERS of water, in the OH stretching and bending regions, for molar NaI and KI solutions. The pairing of the bands for the two species of water is apparent. The broad band - 2900 c m - I is always present and is thought to arise from alkyl groups of contaminants [2,10).

373

cm-i. The potential dependences of these bands show that the higher frequency stretching and bending modes belong to the same species; of the two 8(OH) bands the lower frequency mode is generally the more intense as can be seen in Fig. 4 for the case of KI solutions. The clear distinction between the spectra for some of the cations also allows measurements in mixed electrolytes. For example, addition of KI to 1 M NaI to make the final solution 0.25 M in potassium ion leads to a replacement of the spectrum in Fig. 4A by that in Fig. 4B i.e. K ions are preferentially adsorbed at -0.9 V. DISCUSSION It would appear that most of the results obtained in this investigation can be interpreted in terms of the adsorption of ion pairs in the compact double layer. Electrochemical measurements using mercury have shown that halide ions (other than F - ) are specifically adsorbed especially at potentials positive to the point of zero charge (pzc); iodide ions are particularly strongly adsorbed so that their adsorption shifts the PZC markedly to negative potentials. Strong adsorption is also found for silver electrodes [11]. This adsorption of halide ions at the inner Helmholtz plane (IHP) is thought to be accompanied by non-specific adsorption of cations at the outer Helmholtz plane (OHP), the cations retaining their solvation shells. The spectroscopic measurements, however, show that cations may be adsorbed at potentials positive to the PZC at the sites giving rise to the SERS from the adsorbed halide ions. The specific character of this adsorption is shown by the potential dependence of the spectra for a given cation (Fig. 2) and by the variation of the spectra with the nature of the cation at a fixed potential (Fig. 3). In interpreting these changes it is necessary to take into account the likely structure of the solvation shells. Thus Li may have two or three strong solvation spheres, Na two such spheres but for K, Rb and Cs the binding of the secondary solvation sphere will be progressively weakened. The p(OH) band at - 3490 cm- 1 may therefore be assigned to water in the primary solvation sphere and that at - 3550 cm-1 to water in the secondary solvation sphere. At a given potential, the removal of the outer solvation layers will be progressively favoured with increase in the radius of the central cation, Fig. 3, while for a given cation the application of an increasingly negative potential leads to progressive desolvation, Fig. 2. For Na the effects are not sufficiently strong, however, for the band at - 3490 cm-1 to be resolved; Li remains bound via the secondary (or perhaps tertiary) solvation layer throughout the whole accessible potential range. The specific nature of the ion association in the compact double layer would be expected to affect the strength of anion adsorption, which is indeed shown by the experimental data (Table 2). For a given cation the application of increasingly negative potentials (at potentials positive to the PZC) leads to an increase in cation adsorption due to the enhancement of the electrostatic interaction of water with the negatively charged adsorption sites (see below). For example, in the case of KI the SERS of adsorbed iodide ions is intense at - 0 . 6 V and weak at - 1.0 V although

374

adsorbed water reaches maximum intensity near - 1.0 V. The loss of SERS intensity near - 1 . 1 V in the case of Cs ÷ compared with - 1.3 V for K ÷ (Table 2) must reflect the fact that Cs forms a stronger ion pair which is more weakly adsorbed. Clearly, in the light of these data differential capacitance studies must be carried out on these systems in order to determine whether SERS is sampling the overall surface excess of specifically adsorbed species. Further information on the nature of the bonding of the water in the solvation shells of the cations can be obtained by comparing the band shapes and frequency of the SERS bands with those in the Raman spectra of aqueous solutions. Concentrated solutions of chloride, bromide and iodide perturb the Raman spectra of water with iodide having a pronounced effect. In the case of 7.9 M LiI the OH stretching mode is shifted from 3425 to 3466 c m - l, is 1.7 times as intense and half the width compared with bulk water whilst the bending mode, little shifted, is 8.4 times more intense and half the width [12]. This behaviour is qualitatively explained in terms of the structure breaking properties of iodide with respect to the hydrogen bonded water. The OH bonds of the water molecules weakly interacting with iodide are less polarised than those in the strongly hydrogen bonded bulk water and hence should lead to intrinsically higher Raman intensities as well as higher frequencies [12]. In the present electrochemical environment we appear to have an extreme case of this intensity behaviour. In Fig. 4C we see that the O H bending mode for the K ÷ solution is extremely intense: in comparison with bulk water this band is about 12 times more intense relative to the stretching mode, whilst for Na ÷ the ratio is only about 3 times. Clearly the lower frequency species of water is the more strongly bound one and is quasi crystalline. Unfortunately, the intensity difference between the bending modes in K ÷ and Na ÷ cannot be explained in terms of bond polarisation because the OH bonds for the strongly bound water of the K ÷ I - solvent separated ion pairs should be more highly polarised. An explanation of this particular behaviour is lacking at this time. The sharp v(OH) bands in the SERS spectrum can therefore be most easily interpreted in terms of the electrostatic interaction of the solvation shells with the adsorbed anions; increased electrostatic interaction with increasingly negative potentials leads to a decrease in v(OH) (Table 2). Further, this increased interaction results in displacement of water from the cation's secondary solvation sphere, specifically that part between the two ions. It is probable that in the SERS experiment we are only sampling those water molecules between the ion pairs. It is also possible that we are only sampling a single hydration sphere for any given ion pair, namely that one nearer to the electrode surface; but because of local anisotropies in the electric field at different lattice sites the degree of interaction of these ion pairs varies so we are observing the average interaction. Certain apparently specific effects such as the higher intensity of the band at - 3550 c m - ~ in potassium iodide solutions compared to potassium chloride solutions could also be interpreted in terms of electrostatic interactions, here a reduced interaction in the case of the larger anion resulting in more of the secondary solvation shell of the cation being present. The frequencies of the two species of water support our interpretation; compared

375 with bulk water (3425, 1640 cm -1) the 3552 cm - I b a n d is indicative of weak hydrogen b o n d i n g with the iodide and the bending m o d e at - 1628 c m - 1 of an interaction with the metal ion [13]. O n the other h a n d the - 3490 c m - 1 b a n d is indicative of a stronger interaction with the iodide and the - 1596 c m - 1 b a n d of a stronger interaction with the metal ion. This strongly b o u n d water would be expected to give rise to narrow R a m a n bands, which is seen to be the case. In contrast the bands observed for Li ÷ and N a ÷ are broader and only one b a n d is resolved in the O H stretching region. In both these cases we are sampling filled secondary hydration spheres, which are m u c h less rigidly held than those in the primary sphere. However, the secondary sphere of Li ÷ is more strongly b o u n d than that of N a + , leading to a lower O H stretching frequency. While our observations can be accounted for in terms of a model of ion pairing it would be helpful to have confirmatory data. F o r instance, v ( M - O H 2) vibrations have been reported for the alkali metal cations [14] (180, 179 and 158 cm -1 for N a ÷ , K ÷ and Cs ÷ respectively) but none were detected in our experiments, possibly because the bands are too weak or out of range of the surface enhanced mechanism. W e should, however, point out that bands already reported at 110 and 150 c m - 1 in the SERS of adsorbed chloride are still present on the dried silver surface [1] and are cation independent. With respect to the orientation of the bridging water molecules in these solvated ion pairs, the symmetric looking v(OH) bands for K ÷ , Rb + and Cs ÷ do indicate that both the hydrogens of water are equivalent, whereas the b a n d shape for N a + appears to indicate inequivalence for that system. Finally, we have previously reported SERS from water coadsorbed with cyanide [4] but that work was carried out in N a ÷ media and, in the light of the present results, K ÷ would be expected to yield different spectra. ACKNOWLEDGEMENT The authors thank J. R o b i n s o n for several helpful discussions. The support of the U.S. Office of Naval Research for this project is gratefully acknowledged. REFERENCES 1 2 3 4 5 6 7 8 9 10

M. Fleischmann, P.J. Hendra, I.R. Hill and M.E. Pemble, J. Electroanal, Chem., 117 (1981) 243. B. Pettinger, M.R. Philpott and J.G. Gordon II, J. Chem. Phys., 74 (1981) 934. B. Pettinger, M.R. Philpott and J.G. Gordon II, Surf. Sci., 105 (1981) 469. M. Fleischmann, I.R. Hill and M.E. Pemble, J. Electroanal. Chem., 136 (1982) 361. M. Fleischmann and I.R. Hill, in R.K. Chang and T.E. Furtak (Eds.), Surface Enhanced Raman Scattering, Plenum Press, New York, 1982, p. 275. M. Fleischmann and I.R. Hill, J. Electroanal. Chem., 146 (1983) 353. B. Pettinger, M.R. Philpott and J.G. Gordon II, J. Phys. Chem., 85 (1981) 2746. P.A. Gigu~re and S. Turrell, Can. J. Chem., 54 (1976) 3477. M. Fleischmann and I.R. Hill, Chem. Phys. Lett., to be published. M.W. Howard, R.P. Cooney and A.J. McQuillan, J. Raman Spectrosc., 9 (1980) 273.

376 11 12 13 14

G. Valette, A. Hamelin and R. Parsons, Z. Phys. Chem. N.F., 113 (1978) 71. J.W. Schultz and D.F. Hornig, J. Phys. Chem., 65 (1961) 2131. D.E. Irish, in S. Petrucci (Ed.), Ionic Interactions, Vol. 2, Academic Press, New York, 1971, Ch. 9. M. Moskovits, in E.D. Schmid (Ed.), Proc. 5th Int. Conf. Raman Spectrosc., Schultz Vedag, Freiburg, Germany, 1978, p. 768.