OPTICAL AND ELECTROCATALYTIC PROPERTIES OF OXIDE LAYERS

(1)

HAL Id: jpa-00217166

https://hal.archives-ouvertes.fr/jpa-00217166

Submitted on 1 Jan 1977

HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci-entific research documents, whether they are pub-lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.

OPTICAL AND ELECTROCATALYTIC

PROPERTIES OF OXIDE LAYERS

S. Gottesfeld, M. Yaniv, D. Laser, S. Srinivasan

To cite this version:

(2)

JOURNAL DE PHYSIQUE Colloque C5, supplkment au no 11, Tome 38, Novembre 1977, page C5-145

OPTICAL AND ELECTROCATALYTIC PROPERTIES OF OXIDE LAYERS

S. GOTTESFELD, M. YANIV, D. LASER Dept. of Chemistry, University of Tel-Aviv, Israel

and S. SRINIVASAN

Dept. of Applied Science, Brookhaven National Laboratory, U. S. A.

Rburnk.

-

Les propri6tes optiques des oxydes peuvent 6tre modifikes par l'application d'un

traitement Bectrochimique anodique ou cathodique. De tels traitements modifient aussi le compor- tement du systkme m6tal-oxyde Blectrolyte dans les r6actions klectrochimiques qui impliquent un transfert d'Blectrons a travers l'oxyde. Des corrBlations entre ces modifications des propriktks optiques et 6lectrochimiques des oxydes sont dkcrites pour diffkrents types d'electrodes. Des mesures combinkes rkflectomktriques et ellipsom6triques ont kt6 utilisks pour caracteriser les propriBt6s optiques des oxydes.

Abstract.

-

The optical properties of oxides may be altered by the application of electrochemical anodic or cathodic treatments. Such treatments also affect the behaviour of the metal-oxide- electrolyte system in electrochemical reactions which involve electron transfer through the oxide. Correlations between such modifications of optical and electrochemical properties of oxides are described, for noble as well as electropositive electrode materials. Combined Ellipsometric and

Reflectometric measurements were used for the optical characterization of the oxides.

1. Introduction. - The determination of factors which control the electrocatalytic quality of different electrode materials may have a major contribution to efforts aimed at improving the performance of various batteries and electrolyzers. For electrochemical reactions which occur on a bare metal surface, pro-

perties of the metal itself such as the electronic band structure, the work function [I] etc., may determine the catalytic quality of the electrode. However, a large number of electrochemical reactions, most of which are anodic processes, occur at an oxide-electrolyte interface. For this last type of reactions the catalytic properties of the metal/oxide/electrolyte system are to a large extent determined by the characteristics of the oxide layer. The most relevant properties of the oxide are the electronic conductivity, the stability with res- pect to corrosion and, especially, the surface activity

of the oxide, which may be described in terms of the density of surface states in the appropriate energy region and symmetry [2] at the oxide-electrolyte interface or, alternatively, in the more chemical terms of the surface afinity to intermediates adsorbed on it

in the sequence of steps of the electrochemical reac- tion in question. In-situ optical measurements are an

important tool in the evaluation of the properties of such oxide layers under exactly the same conditions as those prevailing during the operation of the interface as an electrocatalyst. They are especially suitable for following changes in the nature of the oxide

introduced by electrochemical treatments, which may bring about significant modifications in the electrocatalytic activity of the interface with a bearing on the possibility to rejuvenate an oxide catalyst the activity

of which has suffered a deterioration with time. This contribution will describe the optical monitor- ing of the growth and properties of oxide layers on two noble metals : Platinum and Iridium, as well as on the electropositive metal Titanium. The electrochemical modification of the oxides' properties, as monitored optically, will be related to their performance in the electrooxidation (or the photoelectrooxidation) of water into molecular oxygen. A combination of ellipsometric and reflectometric measurements was used, and proved to be very helpful for the evaluation of the optical properties of very thin films

-

a quite difficult task when only ellipsometric results are available.

2. Experimental. - The optical measurements of the Pt oxide layers were performed on a Rudolph ellipsometer and those of the Ir oxide

-

on the automatic version of the same ellipsometer (model RR 2

OOO),

both at Brookhaven National Laboratory in the U. S. Measurements of the Ti oxide films were made on the Gaertner model L-119, at the university of Tel-Aviv. The electrochemical-optical cell used was the same in all the cases and has been described in detail elsewhere [3]. Reflectometric measurements

(3)

C5-146 S. GOTTESFELD, S. SRINIVASAN, M. YANIV AND D. LASER

were taken on the same optical set-up with the ellipsometer components set at P = A = 00 or P = A = 900 for measurements of relative changes in the reflec- tance at parallel or perpendicular polarization respecti- vely. A highly stabilized power supply to the tungsten halogen lamp allowed long term measurements of changes in the intensity without the need of a double- beam arrangement. A wavelength of 546 nm, selected by either an interference filter or a monochromator, was used in all the optical measurements. Analytical reagent grade chemicals were dissolved in triply dis- tilled water for the preparation of the appropriate solutions.

In the evaluation of the apparent properties of the thick oxide on Pt, the electrode potential was lowered to 1.65 V in order to take the optical readings, because at 2.2 V, where the oxide is formed, the current density of the o. e. r. is too high to allow meaningful measurements. On the other hand, ellipsometric measurements (but not reflectometric) were possible on Ir-oxide inside the o. e. r. region as long as the current density did not exceed 10 mA cmF2 [4].

Computer analysis of the optical results was made by using a program for locating the minimum of the function :

C

(X,,,.

-

Xc,,,)2. (Xmeas.

-

measured change in A, Y ,

R,,

or R, ; X

,,,,.

-

calculated changes of the above parameters according to an uniform iso- tropic film growth model).

3. Results and discussion. - 3 . 1 PLATINUM OXIDE LAYERS. -Platinum (or, more precisely, the Pt/Pt oxyde system) is known to be a stable anode material in acid aqueous solutions, but its catalytic activity in the oxygen evolution reactions (0. e. r.) is conside- rably inferior to that of other metal oxides such as Ir-oxide (see next section) or Ru oxide 1.51. A relatively intensive effort has been devoted in the past to the electrochemical as well as optical analysis of the surface oxide on platinum [6]. Most of the conclusions were reached on the basis of experiments which were limited to the so called oxide potential region (up to

--

1.6 V us. the reversible hydrogen electrode (RHE) in the same medium) and to times ranging typically between a few seconds and 10-15 minutes. However, after long term polarization of potentials between 2.10-2.25 V us. RHE in a 0.5 M H,SO, solution (the current densities ranging between 10-100 mA/cm2) a new surface layer can be identified on the platinum electrode, which is very well resolved from the thin superficial oxide layer which forms at lower potentials in the so-called oxide region : A linear cathodic scan following the long term anodic polarization reveals a second large and sharp reduction current peak which appears well inside the Pt-hydrogen region and is well resolved from the regular oxide reduction peak which preceeds it and appears in the same range of potentials as in the absence of the additional reductible layer [7, 81. While the growth of the regular type of Pt-oxide after long periods of polarization is quite

small, the other type of oxide which is formed only well inside the oxygen evolution region may reach a thickness of several hundred layers after a few hours of polarization. This behaviour has already been described by Shibata 171, and his terminology for these films will be used

:$lm

a for the layer formed during anodization at lower potentials, and which is reduced at

--

0.7 V us. the RHE, and film

P

for the thick film formed inside the o. e. r. region.

t

P t , IN H2S04 cl REDUCTION 65; 546.1 n m

C H A N G E O F A, D E G R E E S

FIG. 1.

-

A VS. !P plot for a complete cycle of formation and stepwise reduction of iilms a and /3 on Pt. (20 h growth at 2.25 V

vs. RHE ; cycling for surface smoothing : 20 h at 100 mV s-1

between 0-1.6 V vs, RHE).

Figure 1 gives a A us. Y plot for a complete cycle of formation and reduction of the combined oxide layer on a Pt electrode. The readings during growth were taken at 1.65 V, as explained in the experimental section. Film a was reduced at 0.5 V, and film j? was next reduced at 0.0 V us. RHE. The following points in figure 1 deserve attention : (I) film a behaves as if it were completely independent of film

P.

Its earlier well resolved reduction causes almost the exact rever- sal of the optical effects measured during its formation on the bare metal in the first stages of the an'odization. (11) The reduction of film j? formed after long periods of polarization is accompanied by severe roughening of the Pt surface, which causes a decrease in both A and Y as compared with their values for the bare smooth surface. (Roughening could be confirmed from the increase of the charge passed in a triangular potential scan.) The metal surface could be gradually smooth- ed again by a periodical cycling between 0 and 1.5 V RHE at 100 mV s-I, as can be realized from the last branch of the A us. Y curve in figure 1.

While both slightly absorbing and metallic films may cause a negative change in A and a positive change in Y, as found for the growth of film

8,

the additional reflectometric measurements turned out to be effec- tive in clarifying the meaning of the optical results. (This point will be further elaborated in the next section). The analysis of the optical measurements of film

/?

resulted in the following average properties for 140 min. growth experiments at 2.20 V us. RHE :

(4)

OPTICAL AND ELECTROCATALYTIC PROPERTIES OF OXIDE LAYERS C5-147

In view of the low extinction coefficient in the visible, film

p

is not expected to be a very good electronic conductor. (Indeed, the conductance of film

P has been

recently measured by Shibata [9] and found to be of the order of cm-I). Nevertheless, as figure 2

V, RHE

CURRENT DENSITY. m 4 cm.2

FIG. 2.

-

Tafel Plots for the o. e. r. on Pt.

demonstrates, this oxide even when several hundred angstrems thick, improves the performance of the interface in the o. e. r. by lowering the Tafel slope from 115 to 80 mV per decade of current density. Such an improvement of the o. e. r. characteristics following the appearance of a new higher oxide phase (most probably a partially hydrated form of PtO,) is in accordance with the general views suggested by Tseung and Jassem [lo] : For the o. e. r. to take place at a high rate, OH or 0 intermediates which are reversibly bound to the oxide surface are required. This require- ment of reversible bonding can be fulfilled, however, only as the highest stable oxidation state of the oxide has been reached, because the oxygen atom is bound to the surface at too high an energy in lower oxides. The relatively low electronic conductance is of minor importance here, because of the small thickness of the films concerned. (A current density of 100 mA cm-2 will cause only a 1 mV ohmic drop across a 1000

A

film

p

with a conductance of 1 0-3 0 - I cm-' ). (This

is further supported by the fact that oxide a which is very thin and, apparently, also highly conductive [9], is an inhibitor of the o. e. r. [Ill, most probably because of the too stable bonding of 0 atoms to it). It should be noticed that the effect of film

P

on the performance of the o. e. r. must mean that the surface of film

p

contains the new sites which exhibit the higher catalytic activity (more reversible type of bond- ing to 0 intermediates), and, hence, film

p

must be in contact with the solution, namely it must grow on top of film a and not under it, as suggested by Shibata [9]. 3 . 2 THICK OXIDE LAYERS ON IRIDIUM.

-

Multicycling

of the potential of an Ir electrode in the potential range 0-1.5 V us. RHE (at a rate of

-- 100

mv s-l), results in the current-potential curve which appears in figure 3. The same figure exhibits also the changes recorded in

FIG. 3. - Current and A vs. potential curves during a triangular

scan at 10 mV/s on an Ir electrode covered with a thick oxide layer (0.5 M HzS04 electrolyte).

A (on the automatic ellipsometer) during a single

triangular potential sweep, following the overnight multicycling treatment. The magnitude of the optical effect indicates a marked change in the properties of the layer as a function of applied potential, while the positive 6A/6V cannot be explained by the growth of

an oxide layer on the bare Ir substrate during the anodic half-cycle. The total charge measured in a single half-cycle between 0.25 V and the onset of 0, evolution was

-- 15

000 yCb cm-2 (geo.). Rand and Woods have argued [12] that the multicycling treatment produced an oxide layer which was irreducible at OV us. RHE under the multicycling conditions, while

the large reversible and growing oxide peak currents were due to a change of the stoichiometry in this layer. In order to examine optically the nature of the surface layer left on the Ir electrode as a result of multicycling, ellipsometric and reflectometric readings were taken a t 0.25 V and at 1.40 V at eight points in time during a cycling experiment which lasted 15 hours. The changes in each of the parameters were related

FIG. 4.

-

A YS. 'P plots for the hydroxide layer detected on Ir

(5)

C5-148 S. GOTTESFELD, D. LASER, M. YANIV AND S. SRINIVASAN

to the reference readings obtained at 0.25 V for the freshly polished Ir surface

(n^,,,

measured for the polished surface was 3.19-4.95 i). The fit between experimental 6 A and 6 Y readings at 0.25 V and the best calculated curve obtained under the assumption of uniform iso- tropic film growth due to multicycling, is demonstrated in figure 4. The same figure also shows that experimental ellipsometric readings obtained for the thin- ner films produced in the earlier stages of growth, could be also reasonably well fitted to a solution of a metallic Jilm. This, apparently, caused the earlier misinterpretation of the optical results obtained for an Ir electrode subjected to potential multicycling [13]. Since the theoretical 6 A us. 6Y curve for a metallic

Jilm converges upon the readings for a two-phase (ambient-absorbing film) system as the thickness of the film exceeds 600-700

A

(Fig. 4), ellipsometric measurements may be sufficient to differentiate between slightly absorbing films and metallic films provided the films can be grown to larger thicknesses. Figure 5 reveals the important advantage of combin-

FIG. 5. - (l/R,) 6Rl vs. A plots for the hydroxide layer detected on Ir at 0.25 V, as a result of potential cycling between

0-1.5 V. Solid- calculated for uniform £dm growth with

2

= 1.41-0.01 i. Broken

-

calculated for uniform film growth

A

with nf = 1.90-4.0 i [13]

.

ing ellipsometric and reflectometric measurements : The solution of a metaIlicJilm is shown to be rejected at very early stages in film growth in view of the additional reflectometric results, whereas at such early stages in film growth it is actually impossible to dis- tinguish between metallic and slightly absorbing films on the basis of ellipsometric readings alone (Fig. 4). Figure 6 shows the results of the analysis of the optical results obtained a t 1.40 V during the growth experiment, again under the assumption of uniform iso- tropic film growth. The optical properties of the films monitored at 0.25 V and at 1.40 V mean that a hydrated hydroxide overlayer is produced by multicycling, and that it is converted to a light absorbing layer by the application of an anodic potential across it. (The layers at 1.40 V were found to be 10-15

%

FIG. 6. - A vs. Y and A vs. (1/R) 6R plots for the hydroxide

on Iridium following its oxidation at 1.40 V. The curves are A

calculated for an uniform film with nf = 1.41-0.08 i.

thicker compared to those detected at 0.25 V after the same number of cycles). Slightly higher extinction coefficients of the oxide layer were found inside the o. e. r. region at 1.50-1.55 V, where the current density of the o. e. r. was 5-10 mA cm-' and ellipsometric measurements still meaningful [4].

Since the thick oxide layer on Ir could be easily removed by electrooxidation at potentials higher than 1.60 V [4], this allowed a comparison of the Ir electrode catalytic activity in the presence and the absence of the phase oxide, as given in figure 7. Again, a thick

TAFEL PLOTS FOR THE 0 E R ON II ( 0 5 M H,SO.. ROOM TEMPI It! AFTER ANODIZINF AT ZOV HE

121 IN THE PRESENCE OF TOO^ THICK SURFACE OXIDE

L. "A ~ m ' l g e o )

FIG. 7. - Tafel plots for the o. e. r. on Iridium.

oxide layer of the appropriate oxidation state is shown to exhibit an improved performance in the o. e. r. : The increased light absorption due to anodic applied potential is most probably due to a process such as :

(6)

OPTICAL AND ELECTROCATALYTIC PROPERTIES OF OXIDE LAYERS C5-149

empty appropriate states located on the surface metal ion of the higher valence [4]. The fact that a consi- derable deterioration in the performance of the thick film electrolyte interface in the o. e. r. at 1.50 V was

not accompanied by strong apparent changes of the thick oxides' properties [4] means, again, that the nature and long term fate of the surface states formed by anodization of the Ir(OH), layer may be different (and more significant) than that of the bulk states. 3 . 3 THE ELECTROCHEMICAL MODIFICATION OF OPTI- CAL AND ELECTROCATALYTIC PROPERTIES OF TiO, FILMS.

- Thin oxide films grown on fresh Ti surfaces which have been mechanically abraded in-situ, can be optically monitored according to a procedure similar to that suggested by Ambrose and Kruger [14]. Such films can be then electroreduced at potentials inside the H, evolution region and are then found to have a much higher absorption coefficient. (Thick TiO, photoanode materials are routinely prereduced by gaseous Hz at high temperature, to produce the n- type semiconductor which exhibits an improved performance in the photooxidation process 1151). Optical results for the growth and electroreduction of TiO, films are presented in table I. The associated

Optical constants for the oxide on Ti at variouspotentials (at 5 461

A)

Applied Voltage,

V. i,pAcml' nf ref df(iP)

- - - - -

+

4.0

+

60 2.43 0.0 138

-

0.9 - 20 2.58 0.0 100

-

1.2 - 200 2.60 0.14 86

4.0

+

60 2.40 0.0 135

photocatalytic behaviour as demonstrated for the process of the photooxidation of water to 0, at the Ti/TiO,/electrolyte interface is presented in figure 8. It is quite understandable why the increase of qil,,

which testifies most probably to an increased electronic conductivity as well as an increase in the number of new states inside the forbidden band gap, should be associated with an improved photoanodic performance. It is less obvious reanodization which seems to restore the apparent optical properties of the film back to the as-grown state, should further improve it. This seems to be, again, a good case for demonstrating the higher relative importance of surface states as compared with bulk states both formed by electrochemical treatments. This is especially amplified in this

Refer

FIG. 8. - Photocurrent vs. potential curves for water oxidation at Ti oxide layers irradiated with UV light. (a) TiOz film as

grown by anodic treatment of Ti electrode ; (b) After reduction

at

-

1.3 V for 10 minutes ; (c) After reoxidation (0.5 M Na2S04 ;

pH = 1.7).

case because, apparently, bulk states act as traps for the non-equilibrium electrons produced in the photo- process, and, hence, cause a lowering of the faradaic photocurrent especially at the less anodic applied potentials where trapping is most significant. This interpretation is supported by the improvement due to reanodization being mostly emphasized in the potential region corresponding to the ascending branch of the photocurrent wave [16].

It thus seems that, in general, an increase of K

at a wavelength in the visible does signify a change in the stoichiometry which is accompanied by an increased number of both bulk and surface states, as well as conduction electrons. However, while the surface states, which are vital for catalysis, may have a cha- racteristic fate as a result of a given long term working scheme, the larger number of bulk states makes their contribution to the apparent oxide optical properties dominant. Evaluation of the absorption as a function of wavelength may reveal the separate contri- butions of bulk and surface states which are simultaneously introduced by the same electrochemical treatments.

Acknowledgments.

-

(1) Part of this work was conducted under the auspices of the U. S. Energy Research and Development Administration, during the stay of S. Gottesfeld as a Visiting Scientist a t Brookhaven National Laboratory.

(2) Partial assistance of the U. S.-Israel Binational Science Foundation (the part on Ti oxide) is gratefully acknowledged.

[I] See, for example, TRASATTI, S., J. Electroanal. Chem. 39 [2] MORIN, F. J. and WOLFRAM, T., Phys. Rev. Lett. 30 (1973)

(7)

C5-150 S. GOT'TESFELD, S. SRINIVASAN, M. YANIV AND D. LASER [3] GOTTESFELD, S. and REICHMAN, B., Surf. Sci. 44 (1974) 377.

[4] GOITESFELD, S. and SRNVASAN, S., J. Electroanal. Chem.,

in press.

[5] HOARE, J. P., The Electrochemistry of Oxygen (Interscience Pub].) 1968.

[6] See, for example, CONWAY, B. E. and GOTTESFELD, S.,

J. Chem. SOC. Faraday Trans. 69 (1973) 1090. [7] SHIBATA, S. and S u m , M. P., Electrochim. Acta 17 (1972)

2215.

[8] GOTTESFELD, S. and SRINNASAN, S., Extended Abstracts of

the Electrochemical Society Meeting (May 1976) p. 907. 191 SHIBATA, S., Electrochim. Acta 22 (1977) 175.

1101 TSEUNG, A. C. C. and JASSEM, S., Electrochim. Acta 22

(1977) 31.

[l 11 DAMJANOVIC A. et al., J. Electrochem. Soc. 121 (1974) 11 86. [12] RAND, D. A. J. and WOODS, R., J. Electroaizal. Chem. 55 (1974) 375 ; see also BURKE, L. D. and BUCKLEY, D. N.,

Extended Abstracts of the Electrochemical Society Meeting (Oct. 1974) p. 581.

[I31 OTTEN, J. M. and VISSCHER, W., J. Electroanal. Chem. 55 (1974) 1 ; 55 (1974) 13.

[14] AMBROSE, J. R. and KRUGER, J., Corrosion 28 (1972) 30. [15] BODDY, P. J., J. Electrochem. Soc. 115 (1968) 199. [16] MOLLERS, F., TOLLE, H. J. and MEMMING, R., J. Electrochem.

SOC. 121 (1974) 1160.

DISCUSSION

K. NAEGELE. - The actual state of art of ellipso- metry and reflection spectroscopy does not allow

- to my opinion the determination of optical constants better than to the first digit after the decimal point. So I would say that you have demonstrated pretty well that the optical constant of the Ir-oxide film at 0.25 V and 1.2 V at this special wavelength are identical. The second remark I would like to make is the fact that one cannot deduce for only two optical constants at one wavelength a different state oxide. This has to be demonstrated by a spectral depen- dance of n, k, -+ hv.

S . GOTTESFELD. - I think the statement on the inability to detect differences in k between 0.01 and 0.08 is very wrong ! If you take, for example, the fact that

differences in light absorption of such two films, both about 2 000

A

thick, may be as high as 80

%

or more (depending on angle of incidence) you can realize how easy it is to distinguish between such two cases. (It is also important to note, however, that roughening did not occur simultaneously, and, therefore could not have caused the apparent high K values, the original value of n metal could be reobtained after oxide disso- lution (4).)