An in-situ Raman spectroscopic study of the reduction of HNO3 on a rotating silver electrode

An in-situ Raman spectroscopic study of the reduction of HNO3 on a rotating silver electrode

J. Electroanal. Chem , 200 (1986) 231-241 Elsevier Sequoia S.A.. Lausanne - Printed 231 in The Netherlands AN IN-SITU RAMAN SPECTROSCOPIC STUDY HNO...

943KB Sizes 4 Downloads 13 Views

J. Electroanal.

Chem , 200 (1986) 231-241 Elsevier Sequoia S.A.. Lausanne - Printed

231 in The Netherlands

AN IN-SITU RAMAN SPECTROSCOPIC STUDY HNO, ON A ROTATING SILVER ELECTRODE

IORDANES

OF

C.G. THANOS

Lehrstuhl ficr Organrsche Llchtenhergstrasse

(Recetved

OF THE REDUCTION

Chemw III und Btochemle

4. D-8046

Garchmg

(F R.G)

der Techmschen

Uniuersrtht Mimchen,

*

21st June 1985: m revtsed form 29th October

1985)

ABSTRACT

The reactivtty of NO< at a rotating silver electrode is studied by in-sttu Raman spectroscopy. It IS shown that a reversible adsorptton of NO, occurs at - 0 2 V vs. SCE which can however be inhibited by rutrtte in solution. The adsorption IS practtcally rotatton independent. Electrode rotation allows the distmction between vtbrational peaks arising from adsorbed species and species which are diffusionally removed during the anodic dissolution of silver. Spectroscoptc evidence suggesting the formation of -N& bonds at the electrode surface durmg the reduction of NO; is given, even when no hydroxylamme is detected in the electrolyte. A coordmated discussion of all the observed peaks at potentials between 0.4 and -0.8 V vs. SCE and then dependence upon electrode rotation is also presented.

INTRODUCTION

There is a twofold interest for the present study on HNO,. (a) There are a series of by-products and intermediate compounds, which are anticipated to occur during the NO; reduction. Some like N204 and NO, have been postulated from the analysis of electrochemical data on Pt using concentrated acidic solutions at potentials more positive than 0.2 V vs. SCE [1,2]. A spectroscopic search for such species is carried out in the present work. (b) The Ag/HNO, system is in itself of interest in many respects, including the adsorption of NO, on the metal surface. The spectroscopic Raman work on silver is greatly facilitated because of the surface enhanced Raman spectroscopy effect, SERS [3,4]. The conclusions which are drawn from the present study may not however be directly comparable to the ones discussed in the literature in the case of platinum. A major difficulty in the analysis of the spectra is the anomalous observation of depolarisation ratios which may not be used in the usual way to specify the irreducible representations associated with one mode or a particular set of modes. A

* The spectroscoptc U.S.A.

0022-0728/86/$03.50

work was done on a visit at Case Western

Q 1986 Elsevier Sequoia

S.A.

Reserve University,

Cleveland

OH 44106,

232

possible, but not unique explanation of the unusual depolarisation ratios based on proposed models and theories for the SERS effect will however be discussed in a separate communication.

EXPERIMENTAL

The water used was purified by reverse osmosis and subsequently distilled. The chemicals supplied by Fischer Scientific were of analytical grade and used as delivered, except NaNO,, which was recrystallised. The electrochemical cell used for the in-situ Raman investigation is shown in Fig. 1. The working silver electrode is shaped in the form of a conical section of about 3 mm height and 7 mm average diameter. An Arf ion laser, X = 514.5 nm was used, while a He-Ne one (X = 632.8 nm) was available in the experimental setup. The power of the above laser lines at

17

14

SCATTERING

ELECTRIC

FIELD

t a

s

15

e,P

Frg. 1. (1) Glass cell; (2) Teflon cover with holes for suitable tubes used for filling with electrolyte. bridging to reference electrode. and inserting the counter electrode as it ts or m a fritted tube for separation of compartments: (3) quartz windows; (4) cell holder fixed on a goniometer/micrometer assembly mounted with a motor (not shown) having the shaft (12); (5) electrolyte; (6) aluminium seals; (7) sealing liquid, here H,O: (8) sheath and msulation of the working electrode made of Teflon: (9) Ag working electrode; (10) metallic spring; (11) steel wire threaded to the Ag conical section; (12) motor shaft: (13) conducting brush for electrical contact; (14) Nicol prism; (15) laser beam; (16) specularly reflected beam; (17) observed scattering.

233

the position of the rotating electrode was 100 and 15 mW respectively. The electric field vector of the incident beam could be rotated with a Nicol prism. Observation of the Raman scattering was made at 90” and the tangential plane to the conical electrode at the point of incidence of the exciting radiation was inclined 20” to the direction of the incident beam, A triple Spex monochromator and an optical multichannel analyser, Tracer Northern 7010, were employed for acquisition of data. For an easier direct comparison of the relative intensities of the Raman spectra shown in different figures, a scaling factor (SF) is used in the ordinates. Hence the length of the indicated bar in the ordinate of one figure corresponds directly to the size of the bar at the ordinate of an other figure showing other Raman spectra. Before dipping the silver electrode into the electrolyte it had been previously polished with emery paper (grade 600) a-alumina 3 pm, and a-alumina 0.3 pm respectively. Potential

dependence of the Raman kpectra on Ag in HNO,

solutions

(a) Prior to the activation of the silver electrode, optical signals from the solution phase dominate the spectra, e.g. Fig. 2a, A, B, recorded while the electrode was rotated at a speed of 1200 rotations per minute (rpm) and keeping the electrode potential at -0.4 V (all potentials are given in reference to the saturated calomel electrode). The measurements were carried out in a dark room. The electrode was stopped from rotating and activated at +0.5 V for 20 s. The transition from one potential to another was made in all cases with a linear sweep at a speed of 200 mV s-i. The potential was then scanned back to more negative values and during electrode rotation at 1200 rpm, the Raman spectra in Figs. 2a, C, D and 2b, E-H were recorded. Essentially the spectra were independent of the speed of rotation, but with a totally stationary electrode, e.g. at -0.6 V, the peaks at 1033, 814 and less so at 1002 cm-’ (& 3 cm-’ ) (which are assigned later on) decreased in intensity, possibly due to photo- or thermal decomposition. At potentials between -0.7 V and -0.8 V, for which the current did not exceed - 6.5 mA, equivalent to approximately 10 mA cm -2, the spectra were similar to that shown in Fig. 2b, H. A subsequent potential increase resulted in the spectra A-D of Fig. 3. It should be remarked that the spectral response to potential changes occurring between 0.0 to -0.7 V within 1 min were nearly constant for 1 h as long as the electrode potential remained unchanged. By further repeating the potential changes which have been described previously potential dependence we see that the strong peak at 1033 cm-’ shows a reproducible of its intensity. If, on the other hand, NaNO, is added to the electrolyte at an electrode potential of 0.0 V, so that the total NaN02 concentration is approximately 0.025 M, the appearance of the 1033 cm-’ peak at lower potentials, e.g. -0.45 V (Fig. 3, E), is prevented. Assuming that the 1033 cm-’ vibration is associated with the adsorption of NO;, for reasons becoming obvious later, we may assume that NO; in solution inhibits the adsorption of NO;. A closely related phenomenon is the following:

234

(bl

r-~

528.0

&I

541.0

WAVELENGTH/ mri

55Q.c

529.0

538.0 WA'IELENZ-TH/

54e.o

558.0

nm

Fig. 2. (a) Raman spectra at the Ag/electrolyte Interphase, h = 514 5 nm. electrolyte, 0.02 M HNO,. The ordinate shows arbitrary Raman intensity: the bar corresponds to one umt to be multlplied by the scaling factor SF. where a higher SF value means a stronger spectrum. Perpendicular (s) and parallel (p) polarisatlon IS defined diagrammatically m Fig. 1. The electrode potential E 1s given m V vs. SCE; observation of the spectrum requires 20 s. (A) s, SF= 1, E = -0.2. (B) p, SF= 1, E = -0 4; (C) p. SF=4. E=-0.2;(D)s.SF=2.E=-0.2.(b)As(a) (E)p.SF=4, E=-0.3;(F)p,SF=4, E=-0.4; (G) p. SF= 4, E = -0.5; (H) p. SF= 4. E = -0 7.

If a silver electrode activated as described above, but in a 0.02 M H,SO, solution, was subsequently washed in water and transferred into a 0.02 M HNO, solution. the 1033 cm-’ peak was absent in the potential range between 0.2 V and - 0.6 V, though vibrations of the preadsorbed sulphate were observed. The potential dependence of the Raman line 1033 cm-’ was re-established if the Ag electrode was reactivated in the HNO, solution. It is likely that the active sites, where NO, could be adsorbed. were occupied by the sulphate ions, thus inhibiting its adsorption. It should be noted that by using a fritted electrolyticjunction and keeping the counter electrode in a compartment separate from the electrolyte cell it has been verified that the above spectra did not arise from products liberated at the platinum counter electrode. Such a control experiment is necessary because during the recording of the spectra presented here the counter electrode was dipped in the same compartment as the working electrode.

235

I

I

528.0

I

538.0

I

548.0

1

558.0

WAVELENGTH / m

Fkg. 3. (A)-(D) as in Fig. 2a, but potential is changed from negatwe to posItwe dxectlon. (A) p. SF = 4, E=-0.5; (B) p, SF=4, E=-0.2; (C)p, SF=4. E=-0.1; (D) p. SF=4, E=O.O; (E) p. SF=4. E = -0.4 after addition of NaNO, into the electrolyte to yield (NO, ] = 0.025 M: electrode rotation 1200 rpm.

(b) A strong electrode prehistory is observed and the strong SERS signal means that there is a danger of observing vibrations of contaminating substances. For this reason the electrode used for the measurements reported above was originally turned on the lathe and, as described in the experimental part, was subsequently polished. It has been found however, that after a number of experiments in HNO,, NaNO, and NaNO, solutions, an ordinary treatment with emery paper, rw-A120, and ultrasonic cleaning is not sufficient to remove oxygen-nitrogen derivatives, and possibly the poorly soluble Ag,(N,O,), strongly attached to the electrode surface. Their presence is easily demonstrated with ex-situ Raman spectroscopy if the exciting radiation has the electric field vector parallel to the plane defined by the laser beam and the direction of observation (p-polarisation). The dissolution of silver during the activation process of in-situ experiments can clean the surface partially. The main peaks which are most difficult to eliminate by ordinary polishing are shown in Fig. 4, in which the ex-situ Raman spectrum on Ag which has been submitted to the cleaning treatment is presented.

236

528.0

538.0

WAVELENGTH

548.0

558.0

/ nm

Fig. 4. Ex-situ spectrum

of Ag after washing

and polishing,

p-pol.,

SF= 2, electrode

rotation

1200 rpm.

(c) The Raman scattering from the electrode surface depends on the concentration of the electrolyte. In the following experiment the electrode is polished and activated at 0.5 V in 0.002 M HNO,. The spectrum of the rotating electrode with is shown in Fig. 5, A. The 1200 rpm, recorded at -0.4 V with p-polarisation spectrum reveals the bands already shown in Fig. 4 and attributed to preexisting

, 528.0

538.0 WAVELENGTH

I

I

548.0

558.0

/ nm

Fig. 5. Dependence of Raman spectrum at Ag. E = - 0.4 V vs. SCE, on the concentration c of HNO,. (A) p, SF = 1, c = 0.002 M; (B) p, SF = 4, c = 0.014 M; (C) s. SF = 4. c = 0.014 M; (D) p. SF = 4, c = 0.085 M. Electrode rotation 1200 rpm.

231

strongly adsorbed molecules. By increasing the concentration to 0.014 M HNO, spectrum 5, B results. This has to be compared with spectrum 5, C resulting from scattering with s-polarisation. If the electrode is reactivated at 0.5 V and the potential changed by a rectangular potentiostatic step to - 0.4 V the intensity of the background peaks and of that at 1033 cm-’ is increased fairly well in proportion to the increase in the concentration of the electrolyte. On the other hand the peaks at 931, 1004 and 1124 cm-’ substantially decrease in their relative intensity. After a further increase in the HNO, concentration to approximately 0.085 M the depolarised spectrum 5, D is measured at the same potential. The potential dependence of the intensity of the 1033 cm-’ peak was also monitored with the 632.8 nm exciting radiation and the results directly parallel the ones reported using the 514.5 nm Ar+-ion line. (d) Although already known in the chemical literature, the solution phase spectra of HNO,, NaNO,, and NH,OH . HCl are shown in Fig. 6, A-E in order to enable a more direct comparison with the more complicated spectra of e.g. Fig. 2b. For the

528.0

538.0 WAVELENGTH

548.0

558.0

f nm

Fig. 6. Aqueous phase spectra, solutions approximately 0.8 M. Contrary to the spectra observed on Ag, scattenng is here observed for 100 s. HNO,: (A) s. SF = 4; (B) p, SF = 1. NH,OH.HCl: (C) s, SF = 1; (D) p, SF=l. NaNO,: (E) s. SF= 2.

238

case NH,OH . HCl the intensity of the band observed at 1004 cm-t 3.6 times weaker in the spectrum with p-polarised light. Effects arising from adding NaNO,

to the HNO,

was found to be

electrobte

The spectrum observed at 0.1 V during rotation of the electrode with 1200 rpm in 0.002 M HNO, is shown in Fig. 7, A. The electrode was previously activated at 0.5 V and rotated for 10 min at -0.4 V in order to redeposit many of the silver ions produced during electrode activation before spectrum A in Fig. 7 at 0.1 V was recorded. Then NaNO, is added to the solution in order to obtain a concentration of approximately 0.025 M in NaN02. The two peaks at 824 and 1228 cm-’ attributed to NO, are observed in the spectrum of Fig. 7, B, though they are practically absent in the spectrum observed with s-polarisation, Fig. 7, C. The peaks are potential dependent and they almost disappear at potentials < - 0.3 V. e.g. at E = - 0.6 V in Fig. 7, D.

I , 528.0

1

538.0 WAVELENGTH

1

I

I

548.0

558.0

1 nm

Fig. 7 Spectra at Ag m the presence of predominantly cross are strongly rotation dependent. (A) p. SF= E= 0.1, after addmg NaNO,. (C) As (B) but s-pol. electrode not rotated. (F) As (E), but wth electrode after electrode has been washed, rotation 1200 rpm

NO; amon III the electrolyte. Peaks marked wth a 1, E = 0.1. before adding NaNO,. (B) p, SF= 1, (D) p, SF=I. E= -0.6. (E) p. SF=l. E=0.35. rotation 1200 rpm. (G) p. SF = 1. ex-situ spectrum E in V.

739

At potentials higher than 0.25 V silver dissolution begins slowly and on increasing the potential from 0.25 to 0.35 V, along with an increase of the current from 10 to 28 PA, the intensity of the peaks marked with a cross in Fig. 7. E also increases dramatically in the spectrum recorded with a non-rotating electrode. Moreover, the anodic current at 0.35 V is rotation dependent: 28 PA without rotation and 80 PA at 1200 rpm. The intensity of the marked peaks in Fig. 7, E also becomes weaker upon electrode rotation as shown in Fig. 7, F. The ex-situ spectrum of the Ag electrode after the previous treatment and after washing with H,O is shown in Fig. 7, G. It is interesting that 30 min washing in either 0.02 M HNO, or in 8% w/w NH,OH does not cause any significant change in the ex-situ spectrum. AdditIonal spectral characteristm

at higher ti~a~~emmbers

By shifting the grating in the Spex monochromator the Raman spectrum at a different wavenumber range can be studied. The peak dependence upon electrode potential is then studied in an extended vibrational range although longer time intervals between successive potential changes have to be taken into account. After activation of the electrode at 0.5 V for 20 s a sequence of spectra in the wavenumber region 200 to 3800 cm-’ were recorded at four different potentials: 0.1, - 0.3, - 0.8 and 0.4 V with simultaneous silver dissolution in the-last case, Fig. 8a-d. The electrode is rotated at 1200 rpm. The most characteristic features are the weak broad peak at 2760 cm-’ which appears at potentials more negative than - 0.3 V. the vibrational mode around 2300 cm-’ and the triplet about 2860 cm-‘. It is also seen that at - 0.8 V there is possibly an overlap of the water peak at 1650 cm-‘, characteristic for the bending vibration of H,O, with apparently another, sharper peak observed at that potential. Moreover, a further band broadening and shifting to higher wavenumbers is found at 0.3 V. It is interesting that the H-O stretching vibration of 3600 cm-’ gives rise to a band the shape of which is similar for both p- and s-polarisation, Fig. 8d. For comparison the spectrum of bulk water with p-polarisation, showing mainly only one broad peak contrary to the one for s-polarisation with two distinct broad peaks is also displayed in Fig. 8d, G. Liquid phase spectra of electro(t’sed nitrate solutions

The electrolyses were carried out at room temperature in a separated cell with 10 ml anolyte and catholyte capacities respectively. The catholyte was originally of the same composition as the anolyte. Ag and Pt. approximately 5 cm’ each are used as cathode and anode respectively: (a) A 120 h electrolysis of 0.5 M HNO, at -0.36 V did not result in any products in the solution which give rise to vibrational modes with characteristic frequencies other than the ones corresponding to the NO;. Taking into account the sensitivity of the Raman bands of hydroxylamine and of NO, as well as the total

522.0

572.0

542.0

552.0 522.0 572.0 WAYELENGTH / m

542.0

552.0

Fig. 8a. Raman spectra on Ag electrode at four different potenttats. Wavenum~r range appro~mateIy 250-1300 cm-‘. electrode rotation 1200 rpm. (A) p. SF= 2, E = 0.1: (B) p, SF= 2. E = -0.3; (C) s, SF=2,E=-0.3;(D)p,SF=2,E=-0.8;(E\s.SF=2,E=-0.8;(F)p,SF=2.E=0.4. EinV.

2251

I 54.0

‘ 564.0

t 574.0

I , I 584.0 554.0

WAVELENGTH /

Fig. 8b. As Fig. 8a, but m the range appro~mat~ly E=-0.3;(C)p,SF=l, E=-0,3;(D)p,SF=2.

I 564.0

I 574.0

I 584-C

nm

1300-2400 cm-‘. (A) p, SF= 1, E = 0.1: (B) s, SF= 1, E=-O.S:(E)p,SF=4, E-0.4.

241

I 586.0 ,

I

596.0 I

1

616.0 I I 586.0 I

596.0 I

606.0 4

WAVELF,NCTH /

I

616.C

606.0 nm

Fig. SC. As Fig. 8a, but m the range approximately SF=2, E=-0.3; (C)s. SF=2, E=-0.3; (D)p, SF = 4, E = 0.4.

2400-3200 cm-‘. SF=2, E=-0.8;

(A) p, SF= (E)s. SF=2,

2, E = 0.1; (B) p, E=-0.8, (F)p,

-F-

G

‘06.0

616.0

626.0

636.0 606.0 WAVELENGTH 1

616.0

626.0

636.1

m

Fig. 8d. As Fig. 8a, but in the range approximately 3000-3700 cm-‘. (A) p. SF= 2, E = 0.1; (B) p, SF=~,E=-O.~;(C)~,SF=~,E=-O.~;(D)~,SF=~,E=-O.~;(E)S,SF=~, E=-0,8;(F)p, SF = 4, E = 0.4; (G) p, ex-situ recorded spectrum of bulk electrolyte; here, the Raman intensity is of arbitrary scale.

528.0

538.0

548.0

WAVELENGTH /

558.0

nm

Fig. 9 Liquid phase Raman spectra of electrolysed 0.5 M NaNO, + 0.002 M HNO, at - 1.0 V vs. SCE after 6 h electrolysx Recordmg conditions as m Fig. 6: (A) s. SF = 1; (B) p. SF = 1; (C) p. SF = 1. poor to electrolysis.

electrolysis charge, it can be concluded that if any of these compounds is produced electrochemically their current efficiencies would have been less than 2% for NH,OH and 5% for HNO,. The only spectral change from the original solution is that the intensity of the 1048 cm-’ vibration decreases due to migration of the anion into the anolyte for partial compensation of the cathodically discharged Ht. (b) At -0.54 V the predominant cathodic product is hydrogen and during a 5 h electrolysis a strong gas evolution is observed. For the Raman spectra of the solution phase the same remarks as in case (a) above are valid. (c) The liquid phase spectrum of the electrolysis product of a 0.002 M HNO, + 0.4 M NaNO, solution at -1.0 &-0.1 V (the uncertainty in potential is due to ohmic drop effects) reveals only NO; as the final product and the spectrum of the catholyte is reproduced in Fig. 9. The final pH is alkaline, > 13.

Cyclic voltammograms

of solutions containing NO,-,

NO,

In Fig. 10 the cyclic voltammograms on a silver disc electrode (area = 0.22 cm’) are reproduced for two potential windows when the electrode is not rotated, curves A, B, B’. In one of the potential ranges the effect of the electrode rotation at 2000 rpm on the cyclic voltammogram is demonstrated in curve C. In curves B, B’, recorded consecutively with a potential sweep rate of 100 mV s-l, two anodic peaks (1) and (2), and a cathodic “shoulder”, (3), are distinguishable.

243

a

-0.5

-0.3

-0.1 0.0 -0.5 -0.3 -0.1 0.0 -0.5 -0.3 -0.1 0.0 POTENTIAL/ V

YS

SCE

Fig. 10. Cyclic voltammograms on stationary Ag disc electrodes recorded at 100 mV s-‘. Electrolyte: (a) 0.02 M HNO,: (b) 0.02 M HNO, +OS M NaNO,: (c) 0.02 M HNO, +0.5 M NaNO, A and B of (a). (b) and (c) and C of (b) refer to different potential wmdows. B’ m (a) and (b) IS recorded Immediately after B. C of (a) IS recorded during electrode rotation, 2000 rpm. Current scaling factors (in mA/divlsion). (a) A, B. B’ 0.02, C 0.005; (b) A 0.005, B, B’. C 0 05: (c) A 0.0025, B, B’ 0 05.

If the electrolyte contains 0.02 M HNO, and excess of NaNO, is added to give a solution of 0.5 h4 in NO,, peak (1) of Fig. 10a increases drastically, as shown in Fig. lob. A similar increase occurs in the cathodic current due to hydrogen evolution. It has been checked that the increase in current cannot be attributed to the increase of the conductivity of the solution giving rise to less ohmic drop relative to the total potential drop between reference and working electrode. Nevertheless, in the presence of excess of NaNO, the transport of H+ for the reduction is mainly determined by the diffusion coefficient of the hydrogen ion whereas in the absence of NaNO, the effective diffusion coefficient D determines the transport properties of the hydrogen [5]: D = D,+. &o;/&++

&--IO;)

where D,+ and DNOT are the diffusion coefficients of hydrogen ion and nitrate respectively. It is also observed that during cyclic voltammetry a gradual increase of the height of peaks (1) and (3) of Fig. 10b results and a steady state voltammogram cannot be established within 20 cycles. On the contrary if the composition of the electrolyte is 0.02 M HNO, and 0.5 M NaNO, a time independent voltammogram is observed and only one anodic peak is found, Fig. 10~. Obviously, peaks (2) and (3) of Fig. 10a

244

and b are related to the presence of the NO; anion. All three peaks, (l), (2) and (3), disappear upon electrode rotation. Hence the peaks attributed to the NO, can be due to adsorption only if this adsorption is strongly pH dependent, because in this case the H’ ions are depleted close to the electrode surface of a non-rotating electrode. Normally an adsorption peak should be independent of the electrode rotation unless a hindered transport, due to low concentrations or long diffusion paths, limits the entire process. Similar observations are made in electrolytes with the compositions of 0.002 M HNO, + 0.5 M NaNO, or 0.002 M HNO, + 0.5 M NaNO?. At low H+ concentration (0.002 M), cycling between -0.6 V and 0 V vs. SCE results in a pronounced electrode roughening and an increase of the double layer capacitance by more than a factor of 100. The reason is that during the positive sweep oxidation of silver takes place even if E < 0.0V vs. SCE in the alkaline environment which is created in the electrode vicinity by the reduction of H+.

DISCUSSION

OF THE RAMAN

SPECTRA

The following factors complicate the peak assignment: (a) A number of molecular species giving rise to vibrations with similar frequencies are present. (b) The frequencies of molecules interacting with the electrode are different from those of the same modes in the bulk of the electrolyte. Even in cases where the symmetry of the molecules is lowered upon adsorption and degenerate peaks split, there are different types of bonding of the free molecule which may result in the same symmetry group; e.g. NO; with D,, symmetry may interact with one or two 0 atoms at the surface and the symmetry may be C,,. for both cases. (c) The depolarisation ratio cannot be used as a criterion for mode identification with SERS. Nevertheless, the Raman spectrum can yield valuable information not obtainable from pure electrochemical techniques. Before activation of the Ag electrode, Fig. la. A, we cannot identify any compounds at the surface of the electrode when the p-polarised scattering is observed. s-Polarised scattering shows the 1048 cm-’ A,, vibration of the solution phase. After activation a strong potential-dependent peak is possibilities for the assignment of this observed at 1033 cm-‘. Three reasonable peak are the following: (1) weakly adsorbed NO, result in a frequency shift of 15 cm-’ of the A,, mode relative to the one found for the ion in the solution phase, cf. Figs. la, lb and 2 with Fig. 5; (2) the symmetric (N02),Y, mode of bidentate NO,-Ag bonds; (3) N-O stretch of unidentately bonded NO, unit to Ag. In the last two cases, however comparable complexes exhibit frequencies for the respective modes which are usually less than 1025 cm-’ and simultaneously the asymmetric stretch in the range 1450-1580 cm-’ is observed (Table 6.14 of ref. 6

245

and references cited therein). Since the latter frequencies have not been observed in the potential range - 0.2 to - 0.4 V despite the strong 1033 cm-’ peak, explanation (1) is preferred. There is a strong potential dependence of the intensity of the 1033 cm-’ peak but hardly any frequency shift between -0.2 and -0.8 V vs. SCE. The intensity of the bending 6(NO,) vibration at 720 cm-’ is such that it is detectable only at high sensitivity. It is then found that the ratio of the intensities of the modes at 720 and 1033 cm-’ for scattering with p-polarised light from the Ag surface resembles the intensity ratio of the corresponding peaks when p-polarisation is applied to study nitrate in the solution phase. A similar anomaly is found in the vibration at 822 cm-’ is best seen with case of NO, ; the totally symmetric p-polarised exciting radiation (Fig. 6, spectra A, B). These findings are understandable in terms of the amplification of the perpendicular electrical field component at silver and the excitation of surface plasmons. The doublet between 790 and 822 cm-’ can be in part due to the Raman-forbidden r(NO;) vibration; upon ion adsorption the symmetry is lowered from D,, to C,,, or C,, and the respective selection rule relaxes. Otherwise the two peaks can be attributed to NO,. In this case the 794 cm-’ peak must be due to adsorbed NO; while that at 822 cm ’ due to NO; in solution but very close to the electrode surface. In such a case however it would be hard to explain why the intensities of the 794 cm-i and 1033 cm-’ peaks reveal the same potential and concentration dependence and even more why the 794 cm-’ peak should be observed at -0.5 to - 0.7 V, a potential range where NO; shows no adsorption, Fig. 6, B, D. It is interesting that NO, inhibits the adsorption of NO, to such an extent that the 1033 cm-’ peak disappears. In this way it is possible to associate certain peaks with particular molecules or ions. For instance the 1002 cm-’ peak disappears upon inhibition of the NO, adsorption when excess NO; is present. The same peak is also the single peak which is observed in the solution phase spectrum of NH,OH . HCl. At about 1000 cm-’ typical N-OH vibrations are found and so the one at 1002 cm-’ can be attributed to such a stretching vibration. Figure 7c shows that at -0.3 V an increased signal of the broad band system between 2800 and 2925 cm-‘, most likely due to hydrogen stretching appears. Unless all such vibrations are attributed to O-H stretch we may assume that the formation of N-H bonds at -0.3 V takes place so that, through charge transfer to NO; species, (-N (g,) are created. This hints at the synthesis of NH,OH, although in solution no hydroxylamine was detected. This can possibly be explained through the following decomposition reaction: NH,OH

+ 2 NO,

+ OH-+

3 NO,

+ 2 H,O

At -0.8 V the intensity of the broad signal due to the probable H-N stretching vibrations at 2800 to 2925 cm-’ increases and two more well defined peaks at 2785 and 2853 cm-’ are clearly observed. The first one is relatively weak and may be a combination band from the intensive vibration observed in the range 1250 to 1450 cm-‘.

246

Regarding the rest of the peaks the following speculations can be made: The 1126 and the 1056 as well the 1405 and the 932 cm-’ peaks are due to stable surface adducts in a wide potential range and are observed even when the most concentrated ion is the NO, ion. Peak 1126 cm -I is one of several peaks appearing during the anodic dissolution of Ag in the presence of NO; and can be apparently attributed to a nitrite complex, Fig. 6, E, F. The potential at which this process takes place matches the electrode potential for the following reaction: AgNO,

+ e- = Ag + NO;

E = 0.59 V vs. NHE [7]

It is interesting that at -0.8 V vs. SCE one distinct peak at 1615, and possibly another at 1671 cm-‘, are observed which overlap the bending vibration of H,O. as would The vibration at 1615 cm-’ lies in the range of typical O=N- vibrations have been expected for a monodentate adduct: Ag-O-N=0 peak may be explained in the same way, since a high In fact the 1671 cm-’ frequency doublet with a separation of approximately 59 cm-’ is found in organic nitrito bonding. We have not been able to detect any NO with characteristic frequency around 1800 cm-’ or even any N,O,. About NO, ( Y, = 1358, v2 = 757, v3 = 1666 cm-’ [8]) which has been proposed as an accumulating intermediate during the reduction of HNO, on platinum [2] at high electrode potential, no conclusive remarks can be made: a weak peak appears in fact at 752 cm-’ when E = -0.8V whereas it is controversial whether the peaks observed at 1370 and 1671 cm-’ may be correlated with the other two vibrations listed above (Figs. 8a, 8b and 2b). It has not been possible to detect any NH, or NH,OH spectroscopically. However, if we draw an analogy from the NH, spectra, where the symmetric and antisymmetric stretching vibrations are separated by many hundreds of wavenumbers, we can tentatively assign the 2273 and 2951 cm-’ peaks to stretching vibrations of -N,,E groups. The 1440 cm-’ peak which appears at potentials < -0.4 V is most likely to be due to (ONO) stretching. The way in which this group is bound to the electrode surface cannot be predicted. For this reason it is necessary to have further clarifications of the vibrational modes by using isotopically substituted NaNO,, NaNO, and H,O with “N, “0 and ‘H and observing the shifts of the peaks shown here.

ACKNOWLEDGEMENTS

I would like to thank Case Center for Electrochemical Sciences for the stipend awarded to the author and especially Professor Ernest Yeager for corrections and constructive discussions. Drs. A. Nazri, B. Simic-Glavaski, A. Wilkinson and S. Zecevic are greatly thanked for experimental suggestions. Professor H.P. Fritz has read and suggested modifications to the original manuscript.

247

REFERENCES 1 2 3 4 5 6 7

K.J. Vetter. 2. Phys. Chem., 194 (1950) 199. K.J. Vetter, 2. Phys. Chem., 194 (1950) 284. M. Fleischmann, P.J. Hendra and A.J. McQuiIlan, Chem. Phys. Lett., 26 (1974) 163. D.J. Jeanmaire and R.P. Van Duyne, J. Electroanal. Chem., 66 (1975) 235. V.G. Levich, Physicochemical Hydrodynamics, Prentice Hall, Englewood Chffs. NJ, 1962, pp. 281-286. S.D. Ross, Inorganic Infrared and Raman Spectra, McGraw-Hill, London, 1972. R.C. Weast (Ed.), The Handbook of Chemistry and Physics, 56th ed., CRC Press. Cleveland. OH, 1971, p. D-141. 8 G.R. Bird. J.C. Baird, A.W. Jacke, J.A. Hodgeson, R.F. Curl Jr., A.C. Kunke, J.W. Bransford, J. Rastrup-Andersen and J. Rosenthal, J. Chem. Phys., 40 (1964) 3378.