- Email: [email protected]
Reduction of oxygen to water on cobalt-nitride thin film electrodes prepared by the reactive rf sputtering technique
73
J. Electroanal. Chem., 250 (1988) 73-82 Elsevier Sequoia S.A., Lausanne - Printed
REDU~ON ELECTRODES TECHNIQUE
MASASHI
OF OXYGEN PREPARED
AZUMA,
MINORU
in The Netherlands
TO WATER ON COBALT-NIT~DE BY THE REACTIVE RF SPUTIERING
KASHIHARA,
YOSHIHIRO
NAKATO
TI-IIN FILM
and HIROSHI
TSUBOMURA Laboramy for Cherntcal Comersron of Solar Energy and Department of Chemrstry, Faculty of Engrneering Scrence, Osaka Unruerstty, Toyonaka, Osaka 560 (Japan) (Received
26th June 1987; m revised form 12th Apnl
1988)
ABSTRACT Electrochemical reduction of oxygen was carried out on amorphous cobait-nitride (Co,N) thin film electrodes prepared by the reactive RF sputtering technique. The onset potential of the oxygen reduction current in the Co,N electrode in neutral electrolyte solutions was close to that on a platmum metal electrode, which is known to be the most efficient material for oxygen reduction. Measurements with rotating-disk electrodes (RDBs) and rotating ring-disk electrodes (RRDEs) showed that reduction of oxygen to water (the four-electron-transfer type) proceeded on the Co,N electrodes, unlike that on Co metal electrodes, where only the two-electron-transfer type reduction to hydrogen peroxide took place. Tbe reason for the improved electrocatalytic activity by rutndation is dtscussed briefly.
INTRODUCTION
Electrochemical oxygen reduction is an important reaction in view of energy conversion and storage; for example, in fuel cells. This reaction can be divided into two types: One is the reduction of oxygen to water (four-electron-transfer type), which is reported to occur on a limited number of metal electrodes [l] such as platinum, palladium and silver, and on carbon electrodes coated with tetrasulphonated iron phthalocyanine (Fe-TsPc) [2] and covalently-linked (face-to-face) cobalt po~hyrin dimers (abbreviated as face-to-face Co-Co-4 porphyrins in the literature) 13-71. The other is the reduction to hydrogen peroxide (two-electrontransfer type), which is reported to occur on most other metal electrodes [S] such as gold, cobalt and iron, and on cobalt-phthalocyanine modified electrodes [9]. The four-electron-transfer type is desirable from the viewpoint of energy conversion. For both types of reaction, the overpotential at current densities in practical systems (ca. a few mA cm-*) is 0.3-0.5 V. The reduction of this high overpotential is important from the standpoint of efficient energy conversion. ~22-0728/88/$03.50
0 1988 Elsevier Sequoia
S.A.
74
Many studies have been undertaken in search of low-cost and efficient electrode materials for the oxygen reduction [8,9]. However, no material that is more active than platinum has been reported as yet. Very recently, Vante and Tributsch reported that Mo,.,Ru,.,Se, had fairly high activity for the four-electron-transfer type oxygen reduction [lo]. As mentioned above, transition metals such as Co and Fe are known to show only poor electrocatalytic activity for oxygen reduction of the two-electron-transfer type [8]. On the contrary. complexes of these transition metals with N,-chelates such as porphyrin, phthalocyanine and tetra-aza-anulene are reported to show fairly high catalytic activity [8,9], though these complexes, except face-to-face Co-Co-4 porphyrin and Fe-TsPc, can bring about only the two-electron-transfer type reduction. These results might suggest that Co atoms surrounded by several nitrogen atoms have high electrocatalytic activity for the oxygen reduction. Recently, we have studied the electrochemical properties of transition-metal nitrides such as TIN,, VNx, ZrN, and NbN,, and reported [ll-131 that amorphous ZrN, and NbN, electrodes are electrochemically stable in solutions of a wide pH range and experience fairly efficient oxygen and chlorine evolution, quite unlike the respective metals (Zr and Nb), which are easily passivated. In the present paper we report that amorphous cobalt-nitride (Co,N) thin film electrodes cause the fourelectron-transfer type oxygen reduction, unlike cobalt metal electrodes. EXPERIMENTAL
Cobalt-nitride (Co,N) films were deposited on smooth titanium metal (99.6%) plates as substrates by the reactive RF sputtering technique, using a cobalt plate (Furuuchi Chemicals, 1 mm thick, 99.9%) as a target. Details of our sputtering apparatus are given elsewhere [ll-131. The titanium substrates were etched chemically in 5% HF, rinsed, degreased by immersion in acetone, and further sputter-etched by nitrogen plasma for more than 30 min in the sputtering apparatus. The Co,N films were then sputter-deposited on the Ti plates thus cleaned, in an atmosphere of about 1.0 x lop2 Torr nitrogen (1 Torr = 133.3 Pa) for 2 h at an RF power of 2.5 W cme2. The average thickness of the deposited film was calculated to be 1.72 &-0.18 pm from the weights of the films, with the density assumed to be 9.0 g cmp3. The surface morphology of the deposited films was inspected with an AkashiSeisakusho ALPHA 30 A scanning electron microscope (SEM), and the crystallinity was investigated with a Shimadzu VD-1 X-ray diffractometer (XRD). Depth profiles were obtained with a Shimadzu ESCA 750 X-ray photoelectron spectrometer (XPS), combined with the Ar+-ion sputter-etching technique. The binding energies were corrected by taking the Cls peak of contaminating carbon at 285.00 eV as the standard. The Co,N electrodes for the electrochemical measurements were prepared in the same way as described in refs. 11-13. In addition to electrodes for measurements under static conditions, rotating disk Co,N electrodes (RDEs) and rotating ring
75
(Pt)-disk (Co_,N) electrodes (RRDEs) were used together with a Nikko Keisoku model RRDE-1 system and a model DPGS-1 dual potentiogalvanostat. Oxygen reduction currents were studied at room temperature, passing oxygen or nitrogen bubbles into the solution. A saturated calomel electrode (SCE) and a Pt-plate electrode were used as the reference and the counter-electrodes, respectively. Electrolyte solutions were prepared using analytical grade chemicals and water purified with a Japan Millipore Corp. Milli-Q water purification system. RESULTS
Characterization
of sputter-deposited
Co_,N films
The scanning electron micrograph of a Co,N film deposited on a Ti metal substrate showed that the nitride covered the entire Ti surface, though some cracks were present (Fig. 1). No XRD peak was observed in the range of diffraction angles (28) from 20 o to 100 ‘, indicating that the Co,N films prepared in the present work were amorphous, as reported previously by us for nitrides of Ti, V, Zr and Nb [11-131. Figure 2 shows the XPS spectra for Co2p. 01s and Nls, where (a) is for no Arf-ion sputter etching and (b) is after Ar+-ion etching for 30 s. Spectrum (b) did not change upon further etching, indicating that it truly represents the Co,N bulk. The two peaks at 778.4 and 793.6 eV in (b) can be assigned to the Co2p,,, and 2~r,~ peaks, respectively. Spectrum (a) showed another broad peak at about 781 eV, which coincides with the Co2p 3,2 peak of Co0 or Co,O,, lying close to each other [14]. The 01s peak was greatly weakened by the Ar+-ion etching. These results indicate that the surface of the Co,N film is slightly oxidized. The Nls peak is fairly strong in spectrum (a), which probably suggests that the surface oxidized layer may be present as a kind of oxy-nitride. The thickness of the oxidized layer
Fig. 1. Scanning
electron
rmcrograph
of a Co,N
film (1.9 pm thick) deposited
on a Ti metal substrate.
‘5
n s
co2p
I
406
400
392 Brtdmg
energy
538 / eV
I
I
530
Ar’
eichngtin??I
Fig. 2. XPS spectra for Co2p, 01s and Nls peaks for a Co,N film deposlted for no Ar+-ion etching and (b) after Ar+-ion etching for 30 s.
ml,”
on a smooth
Si wafer:
Fig. 3. Relative atomic concentrations for the same Co,N film as shown in Fig. 2 as a function Ar+-ion etching time, together wth the binding energies for the Nls and Co2p,,? peaks.
(a)
of the
was estimated to be 0.5-0.7 nm from spectrum (a) by assuming that the escape depth of the ejected electrons for the Co2p 3/Z peak (having a kinetic energy of about 475 eV) is 0.8-1.0 nm [14]. This oxidized layer was removed by etching in 3 M HCI before the electrochemical measurements as described later. Figure 3 shows the relative atomic concentrations calculated from the integrated peak intensities for the Cq,N film as a function of the Arf-ion etching time, together with the binding energies for the Nls and Co2p,,, peaks. The atomic ratio (Co/N) was constant, about three in the region of Ar+-ion etching times longer than 1 min, suggesting that the deposited film had almost the same chemical composition as that of crystalline cobalt nitride, Co,N [15]. However, as the above-observed atomic ratio is not corrected for the difference in sputter rate between Co and N, the nitride films in the present work will be designated as Co,N (x = 3). It should be noted that the oxygen concentration in the depth profile of Fig. 3 includes a contribution from contaminating oxygen in the XPS apparatus. The
77
true oxygen concentration in the films is estimated to be less than 5% from the weak shoulder at about 781 eV in spectrum (b) of Fig. 2. The binding energy for the Co2p,,, peak for the Cq,N film (778.4 eV) was almost the same as that for cobalt metal (778.45 eV). Electrochemical
measurements
The Co_,N film electrode is chemically unstable in acidic and alkaline solutions; namely, it dissolves in acidic solutions, and is covered with insulating oxide layers in alkaline solutions. In neutral solutions, e.g. in 0.5 M Na,SO, (pH 6.0) the CqN film electrode is stable, although it is electrochemically oxidized and dissolved at potentials more positive than about +0.3 V (vs. SCE). Accordingly, the oxygen reduction currents were studied mostly at potentials less positive than ca. 0.1 V. Before the electrochemical measurements, the Co,N electrodes were etched for several seconds in 3 M HCl in which cobalt oxide and cobalt nitride were dissolved. In Fig. 4, (a) shows the current density (j)-potential (E) curve for a Co,N film electrode in oxygen-saturated 0.5 M Na?SO, (pH 6.0); (b) that for a smooth Pt electrode; (c) that for a Co electrode: and (d) that for a Co,N electrode in nitrogen-bubbled 0.5 M Na,SO,. In all of these electrodes, the hydrogen evolution currents start at potentials of - 1.0 to - 1.3 V. The oxygen reduction current in the Co,N electrode was observed at potentials less positive than +O.l V. The oxygen reduction current in the Co metal electrode was obscured by the electrode oxidation (dissolution) current, which started at potentials more positive than ca. - 0.5 V. The electrode oxidation (dissolution) current for the Co,N electrode started at ca. t-O.3 V, much more positive than the Co metal, as observed for other metal nitrides [13].
I
I
I
!
I
I
-1.0
I
I,
,,,,,,,,,,
-0.5 / V
0
E vs SCE
Fig. 4. Current density (J)-potential (E) curves for Co,N (a). Pt (b) and Co (c) electrodes m O,-saturated 0.5 M Na2S0, (pH 6.0) and that for Co,N (d) m deoxygenated 0.5 M Na,SO,. The scan rate was 20 mV/s.
II
1
I
-1.0
I/II
““‘NJ
-0.5 E vs SCE / V
0
I
0
I
I
I
1
0.02 0.04 f-l” I (rpm)“R
Rg. 5. Current (I)-potential (E) curves for a CqN RDE at various rotation frequencies O,-saturated 0.5 M Na,SO, (solid lines), compared with that in deoxygenated 0.5 M Na,SO, line). The scan rate was 20 mV/s.
Fig. 6. Koutecky-Levich plots for the 0, reduction Na,SO,: (a) for Co,N, E = -0.75 V; (b) for Co,N, glassy carbon, E = - 0.80 V.
(f) in (dotted
current (I) at a potential E in oxygenated 0.5 M E = -0.70 V; (c) for Pt, E = -0.20 V; and (d) for
However, the j-E curves for the Co,N electrodes changed gradually during several repeated cyclic scans, especially when the potential was extended to a region more positive than - 0.4 V. The original j-E curve was restored by chemical etching in 3 M HCl. Figure 5 shows the disk current (I) vs. E curves measured by sweeping from negative to positive potentials for the oxygen reduction on a rotating disk (Co,N) electrode at various rotation frequencies (f). The curves show a step at around plots [16]) for the - 0.8 V. Figure 6 shows I-’ vs. f ‘I2 curves (Koutecky-Levich Co,N electrode (a and b) compared with that for a Pt (c) and a glassy carbon (d) electrode. It has been reported [17] that the disk current (I) can be expressed as l/Z = l/i, + l/i,
(1)
i, = Bf I/*
(2)
B = nFAc, K
(3)
where i, and i, are the diffusion- and the reaction-controlled currents, respectively, n is the number of electrons transferred in the reaction, F is the Faraday constant; A is the electrode area, c, is the concentration of the reactant (O,), and K is a constant determined by the kinematic viscosity of the solution and the diffusion coefficient for the reactant.
19 TABLE
1
B values for the KouteckL-Levich reaction calculated from them Electrode
E/V
Pt
-0.15 -0.70 -0.80
Co.rN GC
a This value was assumed
plots (Fig. 6) and the electron
(vs. SCE)
lo5 B/A
to -0.40 to -0.80 to -0.90
7.21 7.47-8.15 4.39
for the calculation
(rpm)-“”
number
(n) for the oxygen
reduction
n 3.30 3.40-3.70 2.0 a
of n for Pt and Co,N
The plots in Fig. 6 are straight lines for all the electrodes. The slope (B-l) for the Co,,N electrode is constant, (7.5-8.2) X 10e2 mA (rpm)-‘I2 in the potential region between -0.5 and -0.8 V. The product of c, and K in B was determined experimentally to be 2.9 X lo-’ mol cm2 s 1 from the observed slope (BP’) of the Koutecky-Levich plot (rpmY2 for the glassy carbon (GC) electrode, by assuming that the two-electron-transfer type reduction proceeds on the GC electrode. i.e. n = 2 [l]. The electron transfer numbers for the Cq,N and Pt electrodes, calculated using the above-obtained c,K value, are given in Table 1, together with the B values. The n value for Pt is 3.30, indicating that both the four- and the two-electron-transfer type reduction occur on the Pt electrode, in agreement with the results reported in ref. 1. The value of n for the Co_,N electrode ranges from 3.40 to 3.70, indicating that the four-electron reduction reaction is more dominant than it is for Pt. Figure 7 shows the j-E curve for a Cq,N electrode in deoxygenated 0.5 M Na,SO, solution containing 8.8 mM H,O, (solid line) compared with that in the solution without H,O, (dotted line). It is seen that reduction of H,O, occurs on the Co_,N electrode in the potential region less positive than about 0.1 V. This result strongly supports the above conclusion that the four-electron reduction reaction proceeds on the Co,N electrode. Figure 8 shows the results obtained using a rotating ring (Pt)-disk (Co,N) electrode (RRDE) in oxygen-saturated 0.5 M Na,SO,; the disk current (I,) and the ring current (I,) at a constant ring potential E, = 1.0 V (vs. SCE) against the disk potential (E,) are given. An E, value of 1.0 V was chosen so as to detect H,O, when it was formed by the oxygen reduction on the Co,N-disk electrode. A small ring current (Zg = 14 PA) was observed at this E, even when no disk current flowed. The steeply descending disk current (Zb) in the potential region less positive than - 1.2 V is due to hydrogen evolution on the Co_,N-disk electrode, and the steeply rising ring current (In) in the same potential region is due to re-oxidation of the produced hydrogen on the Pt-ring electrode. The disk current in the potential region between - 0.3 and - 1.2 V should be due to the reduction of oxygen to either H,O or H20z on the Co_,N-disk electrode as mentioned before (cf. Fig. 4a). Accordingly, the ring current in this potential region, as measured from the above-mentioned Zg value, should correspond to the re-oxidation of the produced Hz02. The current efficiency q(H202) for HIOz production, i.e. the ratio of the
80
-10 Ftg. 7. Current density (I)-potenttal wtth (solid line) and wtthout (dotted
-0.5 E,vs. SCE / V
(E) curves for the Co,N electrode in deoxygenated line) 8.8 mM HzO,. The scan rate was 20 mV/s.
0
0.5 M Na,SO,
Fig. 8. Disk current (I,)-disk potential (En) and ring current (I,)-disk potential (En) curves for the Co,N electrode in O,-saturated 0.5 M NasSO,. The rotation frequency was 3600 rpm, the nng potential E, was 1.0 V, and the scan rate was 20 mV/s.
H,O,-producing equation
disk current
to the total disk current,
can be calculated
from the
(4) where N is the current collection efficiency determined by the geometrical structure of the RRDE. For our Co,N-Pt RRDE, N was determined to be 0.175 by in 10 mM Fe(CN)zsolution. The current efficiency measuring I, and I, n(H,O,) thus calculated was 4-5s at E, = - 0.7 and - 0.5 V and 17.4% at - 1.1 V.
DISCUSSION
We have found for the first time that the onset potential for the oxygen reduction current in Co,N amorphous film electrodes is rather close to that for Pt electrodes, known to be the most active material for oxygen reduction [8]. In addition, on this electrode the four-electron-transfer type reaction was more prevalent than on Pt electrodes. Unfortunately, the Co,N film electrodes prepared in the present work were not sufficiently stable near the onset potential of the oxygen reduction current. In spite of this instability, the present result is interesting enough from the viewpoint that it indicates the influence of nitridation on the reaction mechanism since only the two-electron-transfer type reduction occurs on Co metal electrodes.
81
It is reported [9] that carbon electrodes modified with Co-N, chelate compounds such as Co-phthalocyanines and Co-porphyrins have fairly high electrocatalytic these compounds cause only the activity for the oxygen reduction. However, two-electron-transfer type reduction to hydrogen peroxide [9], with a head-on type adsorbed oxygen molecule on cobalt as the intermediate [18,19]. On the other hand, two Co-porphyrin molecules held parallel to each other by side alkyl chains (abbreviated as face-to-face Co-Co-4 porphyrins in the literature) have been reported to cause four-electron-transfer type reduction [3-71 with the oxygen molecule held in the form of a Co-O-O-Co bridge as the reaction intermediate [4-71. It has also been reported that the four-electron-transfer type reduction in such compounds proceeds only when the Co-Co distance is in~~~ange 0.4-0.6 nm [4-61, and that the Co-Co-4
porphyrin
forming
the cis Co’
‘Co
bridge
configuration
has a
reaction rate higher than that forming the tram configuration [6]. In the Co_,N film electrodes of the present work, the Co atoms are surrounded by several nitrogen atoms, and therefore it can be expected that the electronic structure of the Co atoms will be modified by the surrounding N atoms so as to have a high electrocatalytic activity like the Co-N, chelates. Also, both the head-on type (Co * . - O=O) and the are possible stereochemically in bridge-type (Co . - - 0=0 . . . Co) oxygen adsorption the Co_,N electrodes. These facts probably constitute the reason why the four-electron-transfer type reduction occurs mainly on the CqN electrodes. One may point out the possibility of an important role of the amorphous structure of the present Co,N electrodes in addition to the effect of nitridation. In this respect, it would be interesting to prepare crystalline cobalt nitride films and to investigate their electrocatalytic activity. We have attempted to prepare crystalline Co-nitride films, but so far we have had no success with the reactive RF sputttering technique. The present result has shown, however, the interesting finding that the catalytic activity of cobalt metal is improved by nitridation.
REFERENCES
1 J.P. Hoare in A.J. Bard (Ed.), Encyclopedia of Electrochemistry of the Elements, Vol. 2, Marcel Dekker, New York, 1974. p. 305. 2 J. Zagal. P. Bindra and E. Yeager, J. Electrochem. Sot., 127 (1980) 1506. 3 J.P. Collman, M. Marrocco, P. Demsevich, C. Koval and F.C. Anson, J. Electroanal. Chem., 101 (1979) 117. 4 J.P. Collman, P. Denisevich, Y. Konai, M. Marrocco, C. Kovai and F.C. Anson. J. Am. Chem. Sot., 102 (1980) 6027. 5 R.R. Durand, Jr., C.S. Bencosme and J.P. Collman. J. Am. Chem. Sot., 105 (1983) 2710. 6 H.Y. Liu. M.I. Weaver, C.-B. Wang and C.K. Chang. J. Electroanal. Chem.. 145 (1983) 439. 7 C.K. Chang, H.Y. Liu and I. Abdalmuhdi, J. Am. Chem. Sot., 106 (1984) 2725. 8 E. Yeager, Electrochim. Acta, 29 (1984) 1527. 9 A. Van der Putten, A. Elzmg, W. Visscher and E. Barendrecht, J. Electroanal. Chem., 205 (1986) 233. 10 N.A. Vante and H. Tnbutsch, Nature (London), 323 (1986) 431. 11 M. Azuma, Y. Nakato and H. Tsubomura, Mater. Res. Bull., 22 (1987) 527. 12 M. Azuma, Y. Nakato and H. Tsubomura, J. Eiectroanal. Chem., 220 (1987) 369.
82 13 M. Azuma, Y. Nakato and H. Tsubomura, J. Electroanal. Chem., in press. 14 G.E. Mullenberg (Ed.), Handbook of X-ray Photoelectron Spectroscopy, Perkin-Elmer, Phystcal Electronics Division, 1979. 15 G.W. Bnndley (Ed.), X-ray Powder Data File, 6th set, American Society for Testing Materials, 1957, Card No. 6-0691. 16 J. Zagal, R.K. Sen and E. Yeager, J. Electroanal. Chem., 83 (1977) 207. 17 J. Newman, J. Phys. Chem., 70 (1966) 1327. 18 B.Z. Nikolic, R.R. Adzic and E.B. Yeager, J. Electroanal. Chem., 103 (1979) 281. 19 B. Simic-Glavaski, S. Zecevic and E. Yeager, J. Phys. Chem., 87 (1983) 4555.
J. Electroanal. Chem., 250 (1988) 73-82 Elsevier Sequoia S.A., Lausanne - Printed
REDU~ON ELECTRODES TECHNIQUE
MASASHI
OF OXYGEN PREPARED
AZUMA,
MINORU
in The Netherlands
TO WATER ON COBALT-NIT~DE BY THE REACTIVE RF SPUTIERING
KASHIHARA,
YOSHIHIRO
NAKATO
TI-IIN FILM
and HIROSHI
TSUBOMURA Laboramy for Cherntcal Comersron of Solar Energy and Department of Chemrstry, Faculty of Engrneering Scrence, Osaka Unruerstty, Toyonaka, Osaka 560 (Japan) (Received
26th June 1987; m revised form 12th Apnl
1988)
ABSTRACT Electrochemical reduction of oxygen was carried out on amorphous cobait-nitride (Co,N) thin film electrodes prepared by the reactive RF sputtering technique. The onset potential of the oxygen reduction current in the Co,N electrode in neutral electrolyte solutions was close to that on a platmum metal electrode, which is known to be the most efficient material for oxygen reduction. Measurements with rotating-disk electrodes (RDBs) and rotating ring-disk electrodes (RRDEs) showed that reduction of oxygen to water (the four-electron-transfer type) proceeded on the Co,N electrodes, unlike that on Co metal electrodes, where only the two-electron-transfer type reduction to hydrogen peroxide took place. Tbe reason for the improved electrocatalytic activity by rutndation is dtscussed briefly.
INTRODUCTION
Electrochemical oxygen reduction is an important reaction in view of energy conversion and storage; for example, in fuel cells. This reaction can be divided into two types: One is the reduction of oxygen to water (four-electron-transfer type), which is reported to occur on a limited number of metal electrodes [l] such as platinum, palladium and silver, and on carbon electrodes coated with tetrasulphonated iron phthalocyanine (Fe-TsPc) [2] and covalently-linked (face-to-face) cobalt po~hyrin dimers (abbreviated as face-to-face Co-Co-4 porphyrins in the literature) 13-71. The other is the reduction to hydrogen peroxide (two-electrontransfer type), which is reported to occur on most other metal electrodes [S] such as gold, cobalt and iron, and on cobalt-phthalocyanine modified electrodes [9]. The four-electron-transfer type is desirable from the viewpoint of energy conversion. For both types of reaction, the overpotential at current densities in practical systems (ca. a few mA cm-*) is 0.3-0.5 V. The reduction of this high overpotential is important from the standpoint of efficient energy conversion. ~22-0728/88/$03.50
0 1988 Elsevier Sequoia
S.A.
74
Many studies have been undertaken in search of low-cost and efficient electrode materials for the oxygen reduction [8,9]. However, no material that is more active than platinum has been reported as yet. Very recently, Vante and Tributsch reported that Mo,.,Ru,.,Se, had fairly high activity for the four-electron-transfer type oxygen reduction [lo]. As mentioned above, transition metals such as Co and Fe are known to show only poor electrocatalytic activity for oxygen reduction of the two-electron-transfer type [8]. On the contrary. complexes of these transition metals with N,-chelates such as porphyrin, phthalocyanine and tetra-aza-anulene are reported to show fairly high catalytic activity [8,9], though these complexes, except face-to-face Co-Co-4 porphyrin and Fe-TsPc, can bring about only the two-electron-transfer type reduction. These results might suggest that Co atoms surrounded by several nitrogen atoms have high electrocatalytic activity for the oxygen reduction. Recently, we have studied the electrochemical properties of transition-metal nitrides such as TIN,, VNx, ZrN, and NbN,, and reported [ll-131 that amorphous ZrN, and NbN, electrodes are electrochemically stable in solutions of a wide pH range and experience fairly efficient oxygen and chlorine evolution, quite unlike the respective metals (Zr and Nb), which are easily passivated. In the present paper we report that amorphous cobalt-nitride (Co,N) thin film electrodes cause the fourelectron-transfer type oxygen reduction, unlike cobalt metal electrodes. EXPERIMENTAL
Cobalt-nitride (Co,N) films were deposited on smooth titanium metal (99.6%) plates as substrates by the reactive RF sputtering technique, using a cobalt plate (Furuuchi Chemicals, 1 mm thick, 99.9%) as a target. Details of our sputtering apparatus are given elsewhere [ll-131. The titanium substrates were etched chemically in 5% HF, rinsed, degreased by immersion in acetone, and further sputter-etched by nitrogen plasma for more than 30 min in the sputtering apparatus. The Co,N films were then sputter-deposited on the Ti plates thus cleaned, in an atmosphere of about 1.0 x lop2 Torr nitrogen (1 Torr = 133.3 Pa) for 2 h at an RF power of 2.5 W cme2. The average thickness of the deposited film was calculated to be 1.72 &-0.18 pm from the weights of the films, with the density assumed to be 9.0 g cmp3. The surface morphology of the deposited films was inspected with an AkashiSeisakusho ALPHA 30 A scanning electron microscope (SEM), and the crystallinity was investigated with a Shimadzu VD-1 X-ray diffractometer (XRD). Depth profiles were obtained with a Shimadzu ESCA 750 X-ray photoelectron spectrometer (XPS), combined with the Ar+-ion sputter-etching technique. The binding energies were corrected by taking the Cls peak of contaminating carbon at 285.00 eV as the standard. The Co,N electrodes for the electrochemical measurements were prepared in the same way as described in refs. 11-13. In addition to electrodes for measurements under static conditions, rotating disk Co,N electrodes (RDEs) and rotating ring
75
(Pt)-disk (Co_,N) electrodes (RRDEs) were used together with a Nikko Keisoku model RRDE-1 system and a model DPGS-1 dual potentiogalvanostat. Oxygen reduction currents were studied at room temperature, passing oxygen or nitrogen bubbles into the solution. A saturated calomel electrode (SCE) and a Pt-plate electrode were used as the reference and the counter-electrodes, respectively. Electrolyte solutions were prepared using analytical grade chemicals and water purified with a Japan Millipore Corp. Milli-Q water purification system. RESULTS
Characterization
of sputter-deposited
Co_,N films
The scanning electron micrograph of a Co,N film deposited on a Ti metal substrate showed that the nitride covered the entire Ti surface, though some cracks were present (Fig. 1). No XRD peak was observed in the range of diffraction angles (28) from 20 o to 100 ‘, indicating that the Co,N films prepared in the present work were amorphous, as reported previously by us for nitrides of Ti, V, Zr and Nb [11-131. Figure 2 shows the XPS spectra for Co2p. 01s and Nls, where (a) is for no Arf-ion sputter etching and (b) is after Ar+-ion etching for 30 s. Spectrum (b) did not change upon further etching, indicating that it truly represents the Co,N bulk. The two peaks at 778.4 and 793.6 eV in (b) can be assigned to the Co2p,,, and 2~r,~ peaks, respectively. Spectrum (a) showed another broad peak at about 781 eV, which coincides with the Co2p 3,2 peak of Co0 or Co,O,, lying close to each other [14]. The 01s peak was greatly weakened by the Ar+-ion etching. These results indicate that the surface of the Co,N film is slightly oxidized. The Nls peak is fairly strong in spectrum (a), which probably suggests that the surface oxidized layer may be present as a kind of oxy-nitride. The thickness of the oxidized layer
Fig. 1. Scanning
electron
rmcrograph
of a Co,N
film (1.9 pm thick) deposited
on a Ti metal substrate.
‘5
n s
co2p
I
406
400
392 Brtdmg
energy
538 / eV
I
I
530
Ar’
eichngtin??I
Fig. 2. XPS spectra for Co2p, 01s and Nls peaks for a Co,N film deposlted for no Ar+-ion etching and (b) after Ar+-ion etching for 30 s.
ml,”
on a smooth
Si wafer:
Fig. 3. Relative atomic concentrations for the same Co,N film as shown in Fig. 2 as a function Ar+-ion etching time, together wth the binding energies for the Nls and Co2p,,? peaks.
(a)
of the
was estimated to be 0.5-0.7 nm from spectrum (a) by assuming that the escape depth of the ejected electrons for the Co2p 3/Z peak (having a kinetic energy of about 475 eV) is 0.8-1.0 nm [14]. This oxidized layer was removed by etching in 3 M HCI before the electrochemical measurements as described later. Figure 3 shows the relative atomic concentrations calculated from the integrated peak intensities for the Cq,N film as a function of the Arf-ion etching time, together with the binding energies for the Nls and Co2p,,, peaks. The atomic ratio (Co/N) was constant, about three in the region of Ar+-ion etching times longer than 1 min, suggesting that the deposited film had almost the same chemical composition as that of crystalline cobalt nitride, Co,N [15]. However, as the above-observed atomic ratio is not corrected for the difference in sputter rate between Co and N, the nitride films in the present work will be designated as Co,N (x = 3). It should be noted that the oxygen concentration in the depth profile of Fig. 3 includes a contribution from contaminating oxygen in the XPS apparatus. The
77
true oxygen concentration in the films is estimated to be less than 5% from the weak shoulder at about 781 eV in spectrum (b) of Fig. 2. The binding energy for the Co2p,,, peak for the Cq,N film (778.4 eV) was almost the same as that for cobalt metal (778.45 eV). Electrochemical
measurements
The Co_,N film electrode is chemically unstable in acidic and alkaline solutions; namely, it dissolves in acidic solutions, and is covered with insulating oxide layers in alkaline solutions. In neutral solutions, e.g. in 0.5 M Na,SO, (pH 6.0) the CqN film electrode is stable, although it is electrochemically oxidized and dissolved at potentials more positive than about +0.3 V (vs. SCE). Accordingly, the oxygen reduction currents were studied mostly at potentials less positive than ca. 0.1 V. Before the electrochemical measurements, the Co,N electrodes were etched for several seconds in 3 M HCl in which cobalt oxide and cobalt nitride were dissolved. In Fig. 4, (a) shows the current density (j)-potential (E) curve for a Co,N film electrode in oxygen-saturated 0.5 M Na?SO, (pH 6.0); (b) that for a smooth Pt electrode; (c) that for a Co electrode: and (d) that for a Co,N electrode in nitrogen-bubbled 0.5 M Na,SO,. In all of these electrodes, the hydrogen evolution currents start at potentials of - 1.0 to - 1.3 V. The oxygen reduction current in the Co,N electrode was observed at potentials less positive than +O.l V. The oxygen reduction current in the Co metal electrode was obscured by the electrode oxidation (dissolution) current, which started at potentials more positive than ca. - 0.5 V. The electrode oxidation (dissolution) current for the Co,N electrode started at ca. t-O.3 V, much more positive than the Co metal, as observed for other metal nitrides [13].
I
I
I
!
I
I
-1.0
I
I,
,,,,,,,,,,
-0.5 / V
0
E vs SCE
Fig. 4. Current density (J)-potential (E) curves for Co,N (a). Pt (b) and Co (c) electrodes m O,-saturated 0.5 M Na2S0, (pH 6.0) and that for Co,N (d) m deoxygenated 0.5 M Na,SO,. The scan rate was 20 mV/s.
II
1
I
-1.0
I/II
““‘NJ
-0.5 E vs SCE / V
0
I
0
I
I
I
1
0.02 0.04 f-l” I (rpm)“R
Rg. 5. Current (I)-potential (E) curves for a CqN RDE at various rotation frequencies O,-saturated 0.5 M Na,SO, (solid lines), compared with that in deoxygenated 0.5 M Na,SO, line). The scan rate was 20 mV/s.
Fig. 6. Koutecky-Levich plots for the 0, reduction Na,SO,: (a) for Co,N, E = -0.75 V; (b) for Co,N, glassy carbon, E = - 0.80 V.
(f) in (dotted
current (I) at a potential E in oxygenated 0.5 M E = -0.70 V; (c) for Pt, E = -0.20 V; and (d) for
However, the j-E curves for the Co,N electrodes changed gradually during several repeated cyclic scans, especially when the potential was extended to a region more positive than - 0.4 V. The original j-E curve was restored by chemical etching in 3 M HCl. Figure 5 shows the disk current (I) vs. E curves measured by sweeping from negative to positive potentials for the oxygen reduction on a rotating disk (Co,N) electrode at various rotation frequencies (f). The curves show a step at around plots [16]) for the - 0.8 V. Figure 6 shows I-’ vs. f ‘I2 curves (Koutecky-Levich Co,N electrode (a and b) compared with that for a Pt (c) and a glassy carbon (d) electrode. It has been reported [17] that the disk current (I) can be expressed as l/Z = l/i, + l/i,
(1)
i, = Bf I/*
(2)
B = nFAc, K
(3)
where i, and i, are the diffusion- and the reaction-controlled currents, respectively, n is the number of electrons transferred in the reaction, F is the Faraday constant; A is the electrode area, c, is the concentration of the reactant (O,), and K is a constant determined by the kinematic viscosity of the solution and the diffusion coefficient for the reactant.
19 TABLE
1
B values for the KouteckL-Levich reaction calculated from them Electrode
E/V
Pt
-0.15 -0.70 -0.80
Co.rN GC
a This value was assumed
plots (Fig. 6) and the electron
(vs. SCE)
lo5 B/A
to -0.40 to -0.80 to -0.90
7.21 7.47-8.15 4.39
for the calculation
(rpm)-“”
number
(n) for the oxygen
reduction
n 3.30 3.40-3.70 2.0 a
of n for Pt and Co,N
The plots in Fig. 6 are straight lines for all the electrodes. The slope (B-l) for the Co,,N electrode is constant, (7.5-8.2) X 10e2 mA (rpm)-‘I2 in the potential region between -0.5 and -0.8 V. The product of c, and K in B was determined experimentally to be 2.9 X lo-’ mol cm2 s 1 from the observed slope (BP’) of the Koutecky-Levich plot (rpmY2 for the glassy carbon (GC) electrode, by assuming that the two-electron-transfer type reduction proceeds on the GC electrode. i.e. n = 2 [l]. The electron transfer numbers for the Cq,N and Pt electrodes, calculated using the above-obtained c,K value, are given in Table 1, together with the B values. The n value for Pt is 3.30, indicating that both the four- and the two-electron-transfer type reduction occur on the Pt electrode, in agreement with the results reported in ref. 1. The value of n for the Co_,N electrode ranges from 3.40 to 3.70, indicating that the four-electron reduction reaction is more dominant than it is for Pt. Figure 7 shows the j-E curve for a Cq,N electrode in deoxygenated 0.5 M Na,SO, solution containing 8.8 mM H,O, (solid line) compared with that in the solution without H,O, (dotted line). It is seen that reduction of H,O, occurs on the Co_,N electrode in the potential region less positive than about 0.1 V. This result strongly supports the above conclusion that the four-electron reduction reaction proceeds on the Co,N electrode. Figure 8 shows the results obtained using a rotating ring (Pt)-disk (Co,N) electrode (RRDE) in oxygen-saturated 0.5 M Na,SO,; the disk current (I,) and the ring current (I,) at a constant ring potential E, = 1.0 V (vs. SCE) against the disk potential (E,) are given. An E, value of 1.0 V was chosen so as to detect H,O, when it was formed by the oxygen reduction on the Co,N-disk electrode. A small ring current (Zg = 14 PA) was observed at this E, even when no disk current flowed. The steeply descending disk current (Zb) in the potential region less positive than - 1.2 V is due to hydrogen evolution on the Co_,N-disk electrode, and the steeply rising ring current (In) in the same potential region is due to re-oxidation of the produced hydrogen on the Pt-ring electrode. The disk current in the potential region between - 0.3 and - 1.2 V should be due to the reduction of oxygen to either H,O or H20z on the Co_,N-disk electrode as mentioned before (cf. Fig. 4a). Accordingly, the ring current in this potential region, as measured from the above-mentioned Zg value, should correspond to the re-oxidation of the produced Hz02. The current efficiency q(H202) for HIOz production, i.e. the ratio of the
80
-10 Ftg. 7. Current density (I)-potenttal wtth (solid line) and wtthout (dotted
-0.5 E,vs. SCE / V
(E) curves for the Co,N electrode in deoxygenated line) 8.8 mM HzO,. The scan rate was 20 mV/s.
0
0.5 M Na,SO,
Fig. 8. Disk current (I,)-disk potential (En) and ring current (I,)-disk potential (En) curves for the Co,N electrode in O,-saturated 0.5 M NasSO,. The rotation frequency was 3600 rpm, the nng potential E, was 1.0 V, and the scan rate was 20 mV/s.
H,O,-producing equation
disk current
to the total disk current,
can be calculated
from the
(4) where N is the current collection efficiency determined by the geometrical structure of the RRDE. For our Co,N-Pt RRDE, N was determined to be 0.175 by in 10 mM Fe(CN)zsolution. The current efficiency measuring I, and I, n(H,O,) thus calculated was 4-5s at E, = - 0.7 and - 0.5 V and 17.4% at - 1.1 V.
DISCUSSION
We have found for the first time that the onset potential for the oxygen reduction current in Co,N amorphous film electrodes is rather close to that for Pt electrodes, known to be the most active material for oxygen reduction [8]. In addition, on this electrode the four-electron-transfer type reaction was more prevalent than on Pt electrodes. Unfortunately, the Co,N film electrodes prepared in the present work were not sufficiently stable near the onset potential of the oxygen reduction current. In spite of this instability, the present result is interesting enough from the viewpoint that it indicates the influence of nitridation on the reaction mechanism since only the two-electron-transfer type reduction occurs on Co metal electrodes.
81
It is reported [9] that carbon electrodes modified with Co-N, chelate compounds such as Co-phthalocyanines and Co-porphyrins have fairly high electrocatalytic these compounds cause only the activity for the oxygen reduction. However, two-electron-transfer type reduction to hydrogen peroxide [9], with a head-on type adsorbed oxygen molecule on cobalt as the intermediate [18,19]. On the other hand, two Co-porphyrin molecules held parallel to each other by side alkyl chains (abbreviated as face-to-face Co-Co-4 porphyrins in the literature) have been reported to cause four-electron-transfer type reduction [3-71 with the oxygen molecule held in the form of a Co-O-O-Co bridge as the reaction intermediate [4-71. It has also been reported that the four-electron-transfer type reduction in such compounds proceeds only when the Co-Co distance is in~~~ange 0.4-0.6 nm [4-61, and that the Co-Co-4
porphyrin
forming
the cis Co’
‘Co
bridge
configuration
has a
reaction rate higher than that forming the tram configuration [6]. In the Co_,N film electrodes of the present work, the Co atoms are surrounded by several nitrogen atoms, and therefore it can be expected that the electronic structure of the Co atoms will be modified by the surrounding N atoms so as to have a high electrocatalytic activity like the Co-N, chelates. Also, both the head-on type (Co * . - O=O) and the are possible stereochemically in bridge-type (Co . - - 0=0 . . . Co) oxygen adsorption the Co_,N electrodes. These facts probably constitute the reason why the four-electron-transfer type reduction occurs mainly on the CqN electrodes. One may point out the possibility of an important role of the amorphous structure of the present Co,N electrodes in addition to the effect of nitridation. In this respect, it would be interesting to prepare crystalline cobalt nitride films and to investigate their electrocatalytic activity. We have attempted to prepare crystalline Co-nitride films, but so far we have had no success with the reactive RF sputttering technique. The present result has shown, however, the interesting finding that the catalytic activity of cobalt metal is improved by nitridation.
REFERENCES
1 J.P. Hoare in A.J. Bard (Ed.), Encyclopedia of Electrochemistry of the Elements, Vol. 2, Marcel Dekker, New York, 1974. p. 305. 2 J. Zagal. P. Bindra and E. Yeager, J. Electrochem. Sot., 127 (1980) 1506. 3 J.P. Collman, M. Marrocco, P. Demsevich, C. Koval and F.C. Anson, J. Electroanal. Chem., 101 (1979) 117. 4 J.P. Collman, P. Denisevich, Y. Konai, M. Marrocco, C. Kovai and F.C. Anson. J. Am. Chem. Sot., 102 (1980) 6027. 5 R.R. Durand, Jr., C.S. Bencosme and J.P. Collman. J. Am. Chem. Sot., 105 (1983) 2710. 6 H.Y. Liu. M.I. Weaver, C.-B. Wang and C.K. Chang. J. Electroanal. Chem.. 145 (1983) 439. 7 C.K. Chang, H.Y. Liu and I. Abdalmuhdi, J. Am. Chem. Sot., 106 (1984) 2725. 8 E. Yeager, Electrochim. Acta, 29 (1984) 1527. 9 A. Van der Putten, A. Elzmg, W. Visscher and E. Barendrecht, J. Electroanal. Chem., 205 (1986) 233. 10 N.A. Vante and H. Tnbutsch, Nature (London), 323 (1986) 431. 11 M. Azuma, Y. Nakato and H. Tsubomura, Mater. Res. Bull., 22 (1987) 527. 12 M. Azuma, Y. Nakato and H. Tsubomura, J. Eiectroanal. Chem., 220 (1987) 369.
82 13 M. Azuma, Y. Nakato and H. Tsubomura, J. Electroanal. Chem., in press. 14 G.E. Mullenberg (Ed.), Handbook of X-ray Photoelectron Spectroscopy, Perkin-Elmer, Phystcal Electronics Division, 1979. 15 G.W. Bnndley (Ed.), X-ray Powder Data File, 6th set, American Society for Testing Materials, 1957, Card No. 6-0691. 16 J. Zagal, R.K. Sen and E. Yeager, J. Electroanal. Chem., 83 (1977) 207. 17 J. Newman, J. Phys. Chem., 70 (1966) 1327. 18 B.Z. Nikolic, R.R. Adzic and E.B. Yeager, J. Electroanal. Chem., 103 (1979) 281. 19 B. Simic-Glavaski, S. Zecevic and E. Yeager, J. Phys. Chem., 87 (1983) 4555.