Deactivation mechanism of nickel cathodes in alkaline media

Int. J. Hrdrogen Energy, Vol. 19, No. 7, pp. 573 578, 1994

Pergamon

International Association for Hydrogen Energy Elsevier Science ktd Printed in Great Britain 0360-3199/94 $7.00 + 0.00

D E A C T I V A T I O N M E C H A N I S M O F N I C K E L C A T H O D E S IN A L K A L I N E MEDIA D. M. SOARES, M. U. KLE1NKE, I. TORRIANI and O. TESCHKE UNICAMP/1FGW/DFA, 13081-970, Campinas, SP, Brasil (Received fiJr publication 1 September 1993) Abstract Time-dependent hydrogen evolution reaction (HER) on Ni electrodes shows a large increase in electrode over potential with time. This is ascribed to hydride formation at active Ni cathode surfaces. Hydride formation was detected by X-ray diffraction. By taking into account the nickel electronic density of states variation following hydrogen sorption, we are able to satisfactorilyexplain the increase in nickel over potential after a few hours of HER. INTRODUCTION

HzO+e

Nickel cathodes show marked overvoltage increase with time in alkalyne electrolytes [1]. The challenge is to develop methods of electrode activation which give stable performance for prolonged operating periods, for which reason we need a detailed understanding of the factors contributing to electrodes overvoltage. Four possible mechanisms have been postulated by previous investigators as being responsible for the increase in cathodic overpotential rL in time. (i) Increases in q~ caused by a loss of electrocatalytically active material from the cathode [2, 3]. (ii) Increases in qc are caused by a slow reduction of nickel hydroxide to nickel, which may be a poorer evolver of hydrogen. (iii) Increases in r/c are caused by deposition of impurities from the electrolyte into the cathode [4]. (iv) Increases in q, are caused by the absorption of atomic hydrogen into the lattice of the nickel cathode [5]. Time effects are recoverable, and an investigation of the recovery mechanism has yielded evidence that counter-indicates mechanisms (i), (ii) and (iii) and supports the hydrogen absorption mechanism [6, 7]. It has been conclusively shown by volumetric measurement of desorbed hydrogen that nickel does absorb atomic hydrogen during electrolytic evolution of hydrogen gas [8 17]. Even though the resulting hydride is unstabled a few physical properties have been reported. X-ray diffraction studies showed that the face centered cubic structure was retained with a lattice spacing about 5.5°/, larger than that of pure nickel [16]. Further, Szklarska-Smialowska [11] and Smialowski [18] have shown that the/~ nickel hydride phase acts as a diffusion barrier to continued hydrogen diffusion. This supports the assertion of a complicated nickel hydrogen interaction. The hydrogen evolution reaction in alkaline media is assumed to proceed via the well-known steps [19]: Hydrogen sorption with charge-transfer:

M Z HM + O H

.

(l)

Hydrogen desorption with charge-transfer: HM + H 2 0 + e

M 2 H/M + O H

(21

Hydrogen desorption by recombination without charge-transfer: HM + HM 2 H2M + M . The palladium electrode, cathodically polarized at a constant current density i, exhibits after current interruption a characteristic over potential transient decay r/. The overall 1/vs t transient curve is usually found [20 23] to have at least two components. One component of the decay process is the discharge of the capacitance through the charge-transfer reaction. A general treatment of potential relaxation behavior for several of the HER was given by Tilak and Conway [24]. The second component is associated with accumulation of H in the Pd (~-phase PdH) with subsequent removal of the H intermediate. The hydrogen overpotential in alkaline solutions has been measured for nickel cathodes by Lukovzen et al. [25], Legran and Levina [26], and Bockris and Potter [27]. All of them have made measurements in dilute alkaline solutions. Later Lee [28] reported measurements of hydrogen overpotentials on nickel in 6 N and 9 N KOH solutions. In this paper further evidence is given that: a threedimensional hydride layer is formed after the Ni electrode is polarized to a high cathodic over voltages. We claim that this hydride layer is the responsible factory for the time varying effects. This hypothesis was raised after we showed that there is no resistive layer formed on the Ni electrode surfaces after a prolonged HER period [29]. Recent papers on the de-activation process ocurring during hydrogen evolution on nickel support the hydrogen absorption mechanism [6, 7, 30].

573

574

D.M. SOARES et al. EXPERIMENTAL

In a strongly alkaline solution nickel electrodes undergo significant deactivation during hydrogen discharge at current densities higher than 100 m A c m 2. For this reason we used in our experiment 300 mA cm -2 and 1 A cm 2 current densities, consequently high cathodic overvoltages were obtained. Tafel and potential lime-dependent curves were obtained for freshly electrodeposited, as-received and annealed nickel electrodes. All electrodes were prepared from the same nickel piece supplied by Niquel Tocantins (electrolytic grade). The electrodes were 1.0 x 0.5 x 3.3 cm 3 pieces and were abraded with emery paper (grid 600). Several electrodes were annealed previous to use, in argon atmosphere for 2 h at ~ 1000°C; then the electrodes had ~0.1 mm of their top surface removed by grinding and were finally abraded with emery paper. Various H g / H g O reference electrode configurations were experimentally examined in recent work from this laboratory [31]; our results show that the H g / H g O electrode may considerably limit the time response for IR-free potential measurements when high current densities ( ~ 1 A cm -2) are used in alkaline solutions. A configuration with a response time of less than 1 Hs was used in this work. The test cell is an improvement upon our earlier version [32]. The electrodes each having an area of 3.3 cm 2 are rectangular nickel plates. The cell module consists of two removable plug-in electrodes. The cell frame is made out of PTFE. The total gap between electrodes is 3.5 cm and the electrode plates are soldered to stainless steel wires. The electrolytic solution (30% K O H , Merck, analytical grade) was pre-electrolyzed for 24 h in a P T F E cell with Pt electrodes and a current density of 1 mA cm-2. This solution was introduced in the test cell with nickel electrode and operated for 4 h at a current density of 1 A cm -2. After this period a new set of electrodes were introduced in the cell and the measurements were performed. The nickel plating solution is prepared with doubledistilled water as follows: 310 g I- NiSO 4 (Merck), 50 g 1- 1 NiC12 (Merck) and 40 g 1 1 boric acid (Quimbras) all analytical grade. A 15 /~m thick nickel layer was deposited on all electrodes in a solution at 60°C, pH = 3.5 and at a current density of 35 MA cm -2. The A s 2 0 3 was supplied by Merck (analytical grade). The apparatus for the time-dependent potential measurements as been previously described [29]. The galwmostat generates current pulses with a rise time of less than 1 ~s and with a maximum current amplitude of 3.5 A. A Hewlett Packard 54201A digitizing oscilloscope, a HP7090A measurement plotting system and an Iwatsu SS-5710 oscilloscope were used for all timedependent measurements. The current interrupter circuit used for D C polarization measurements was also previously described [32]. The electrode surfaces saturated with hydrogen were studied by X-ray diffractometry and intensity vs 20 scans

were obtained at room temperature. The X-ray diffraction patterns were obtained in a Rigaku RU-200 Rotaflex diffraclometer System (40 kV, 60 mA) and Ni filtered Cu(K~) radiation. A rough estimate will give a penetration depth of X-rays of ~ 2 3 Ibm but ~ 5 0 % of the diffracted intensity originates in the first 5 l~m of the sample surface.

RESULTS AND DISCUSSION The open circuit potential following a current pulse was measured. The current pulse duration was enough to charge the electrode capacitance (both double-layer and pseudo-capacitance) but short enough to prevent bubble formation. When the current is interrupted, charges stored at the electrode surface will discharge through the reaction resistance. Cathodic overpotential increase with time and recovery

oJ

lost eJficiency Figure 1 shows the potential decay plots for the HER of a nickel electrode at 30°C. Electrode potential against log t at t = 0, i.e. the instant at which the electrode was removed from a Watt's solution and the HER started is shown in (a). The duration of polarization times for the other curves were: (b) 10 min; (c) 20 rain; (d) 30 rain; (e) 40 rain. These electrode potential decay curves show two 2.10

1.90

o "I-

,7o

1.50

~

1.30

~

ILl 1.10

120 mV/dec

.,,j1~ ,'II

0.90 i0 °

i 0 "l

60 mV/dec I

I

I

I

10-2

iO"3

10-4

10- 5

TIME

10- 6

( s )

Fig. 1. Potential decay plots for the HER in 30% w/w KOH. (a) electrode potential against log t at t = 0, i.e., the instant the electrode was removed from the Watt's solution and the HER was started. For the other curves the duration of polarization times were; (b) 10 rain; (c) 20 min; (d) 30 rain; (e) 40 min. Current pulse amplitude 1 A c m - 2 pulse duration 1 ms at every 100 ms.

DEACTIVATION

>v"'

-

,

[

MECHANISM

r

T

o

C

,
E E I :2 ::: t.

0

] 0()

200

CATHODES

IN A L K A L I N E

575

MEDIA

the slope of all decay curves tend to 60 mV s ~ [33, 34]. The initial slope immediately after current interruption for a fresh electrode is around 120 mV dec -~ and q~ decreases linearly with time. This slope increases for a period of about 2 h. Hydride formation and decomposition rates have a strong dependence on temperature; in order to check this fact the following experiment was performed. Electrode potential vs time curves for electrodes polarized at 300 mA cm -2 were measured at 19°C and at 85°C as shown in Fig. 2. The curve obtained for 85°C shows a very small increase in electrode potential with time, while the electrode curve obtained at 19°C shows an accentuated increase with time, in accordance with the temperature dependence of the hydride formation and decomposition rates.

•

,~;~ t 9 .~ ~;~,:~,,

16

OF NICKEL

:~00

'i'illl~' ( n ~ i t l . )

Detection of nickel hydride presence hy X-ray diff)'action in hard nickel electrodes

Fig. 2. IR-free electrode potential vs time for a polarization of 300 m A c m -2 at 19°C a n d 85°C, respectively.

different features: (a) there is an increase in the steady electrode over potential reached after each applied current pulse, for at least 2 h; (b) there is a change in the slope without a change in the pulse current amplitude. For long times ( ~ 10-~ s) after current interruption

The two nickel peaks at 20 angles equal to approximately 44.5 ° and 51.85 ° are due to Bragg reflections from (111) and (200) planes [35]. Figure 3(a) shows the X-ray diffraction pattern recorded from a nickel sample that was polarized with a 300 mA cm 2 current for approximately 2 h as shown by the electrode potential vs time curve in Fig. 3(b). The sample was removed from the solution when the electrode potential reached a region of less accentuated potential increase. The X-ray pattern shows peaks at 20 angles equal to approximately 42 °, 44 °, 49 ° and 51.9°; these are assigned to both nickel and nickel hydride [35]. The pattern was obtained 20 rain after electrode removal from the electrolyte solution. Figure 4 shows a diffractogram obtained for the same electrode

I0--

~8

=,t

t

)I.-

eL

[

z ILl

~6

4 0 KV

~

:

60 mA

"

LO >

--I

b.J 4 n,"

Z

~

l (200) NiH

(I I I ) N i H

38

I

I

40

42

~

~

I

,

44

I

,

46 SCAN

I

~

48 ANGLE

p ~)

1 O0

:20{) "]'iI'~l("

I

,

50 (28)

it

I 52

,

I 54

,

I

;3()0

(l:l).iIl.)

,

56

DEGREES

Fig. 3. (a) X-ray diffraction pattern recorded from a nickel electrode polarized with a 300 mA cm 2 current for 2 h and observed 20 min after removal and electrolyte solution. (b) Electrode potential vs time curves for two identical nickel electrodes, one of them removed from the electrolyte after HER for 2 h for observation of its X-ray diffraction pattern shown in (a).

576

D.M. SOARES et al.

8000

1

o

~/I

,\~,a()

6 ~

° ° O ° e ° O O o o o o ~ , i ~ (el

6 0 0 0

i'

' add(~d 1 ~ / 1 3,s,():l

• ,:

z

•

~4000

c

z M

i

I

i

I (2/O)NO H ~

I

'

:

'

. . . .

'

1 | ann~'aled

;' _

0

i

i

.

?,0

j

(;0

i

90

1:20

Time, (mm.)

2000

I 40

,

I 42

,

I 44

46

48

50

52

SCAN ANGLE(2e) DEGREES

Fig. 4. X-ray diffraction pattern obtained from the same electrode as shown in Fig. 3, 40 min after the first spectrum was obtained.

Fig. 6. Electrode potential vs time curves (a) for an annealed nickel electrode as shown in Fig. 5; (b) with 1 g I ~ As203 added to the solution after 1 h of electrode polarization: (c) with 1 g l I As203.

40 min after the first was obtained; the hydride X-ray lines decreased substantially; consequently, the layer formed at the nickel electrode surface is unstable at room

'

f

>o ;:I2

:20

I )l

E ~5

temperature in agreement with the fact that nickel hydride is unstable at ambient temperature. The hydride content decreases with time after hydrogen charging and approaches zero at extended periods indicating that all of the hydrogen can be desorbed. Similar results were obtained using electrodeposited Ni surface. X-ray diffraction patterns were recorded from a nickel specimen with a 1 Itm thick layer electrodeposited in a Wart's solution at 25°C. Sets of lines for both nickel and nickel hydride were registered. Nickel samples annealed for 2 h under argon atmosphere were polarized and tested for X-ray hydride lines. Twelve samples were tested and no X-ray hydride lines were found. The annealed samples show substantially lower over voltages when compared to nonannealed ones as shown in Fig. 5.

al!nc,,t led

Cathodic poison as an efficient hydrMe formation agent

¢,

P I

;:2

0

b0

/ 00 Tim.

150

200

~:~0

(mm.)

Fig. 5. Electrode potential vs time curves for (a) an annealed nickel electrode at 1000°C in argon atmosphere for 2 h and (b) nonannealed.

Finally, 1 g l ~ o f A s 2 0 3 was added to a 30% w / w K O H solution and overvoltages were registered as a function of time. The effect of the arsenic is shown in Fig. 6; there is a substantial overvoltage increase in a matter of minutes. The X-ray lines showing the presence of the hydride layer at the surface are shown in Fig. 7. The effects of arsenic on the electrode metallic surface are the immediate increase of the electrode overvoltage, and the increase of the hydrogen loading of the electrode [10], in agreement with the data shown in Fig. 6. The overvoltage increase takes place in a few minutes but its final value is not as high as the one shown in Fig. 5.

DEACTIVATION MECHANISM OF NICKEL CATHODES IN ALKALINE MEDIA

eJ

d

o

Z w hZ uJ >

4

J UJ n-

(III)NiH

2

,

40

I

577

,

i

(200)NiH

I

42

44

46 48 50 52 54 56 SCAN ANGLE ( 2 e ) DEGREES

58

Fig. 7. X-ray diffraction pattern recorded form an electrode with a potential vs time curve as shown in (b). The solution was 30% KOH with 1 g 1-1 As:Os.

Nickel densi o, o [ electronic states

If our assumption that the hydride presence is the basic condition for the initial increase ( ~ 2 h) in overvoltage is proved correct then it remains to be explained why the electrode activity decreases substantially for an electrode surface formed by mixed metal Ni and Ni H. Figure 8 shows a schematic diagram of the density of states function of the d and sp electrons for Ni and N i - H calculated by G u p t a [-36]. Since for Ni H the Fermi level cuts only the metal s p band, the geometry of the Fermi surface is expected to be similar to that of the egg reactive metals, Ag or Cu. It would follow, therefore, that the catalytic activity in nickel is determined by the percentage d-character of the metal since the Fermi level is at the d-band. The hydrogen absorption changes the d-character of the metal to a s p character for the intermetallic

0") W

'd'

Ni-H. The density of states (DOS) is substantially decreased at the Fermi energy and its electrocatalytic character is lifted to that of an s p band as in metal Cu or Ag.

CONCLUSIONS It is shown by X-ray diffraction that increasing overvoltages for high current densities are associated with the formation of hydride at nickel cathode surfaces. The catalytic activity in nickel is determined by percentage d-character of the metal since the Fermi level is at the d-band; whereas nickel hydride has a substantially lower density of electronic state value at the Fermi level since the DOS of nickel was modified by the hydrogen presence to an s- p band. Acknowledgements The authors thank Bernardo Lacks for helpful discussions and Luiz Orivaldo Bonugli for technical assistance. This work was supported in part by CNPq and FAPESP.

band

or) LL © >I-" z Lld 0

~

' s p ' band ;:.i,: :; ):i:'I

f

i

EF

EF

(M)

(MH)

ENERGY

Fig. 8. Nickel and nickel hydride electronic density of states from [36].

REFERENCES 1. R. L. Leroy, M. B. I. Jangua, R. Renand and U. Leuenberger, J. Electrochem. Soc. 126, 1674 (197% 2. K. Christiansen and T. Grundt, in Industrial Water Electrolysis, (S. Srinivasan, F. J. Salzano and A. R. Landgrebe, Ed), p. 29. The Electrochemical Society Softbound Proceeding Series, Princeton, NJ (1978). 3. A. Nidola, P. M. Spaziante and L. Griuffre, ibid., p. 102 (1978). 4. J. N. Murray, J. B. Laskin and W. C. Kincaide, ibid., p. 39 (19781.

578

D . M . SOARES et al.

5. P.J. Moran, in Proceedings of the ERDA Contractors Review Meeting in Chemical Energy Storage and Hydrogen Energy Systems, p. 39. National Technical Information Service, Springfield, VA (1976). 6. H. E. G. Rommal and P. J. Moran, J. electroehem. Soc. 132, 325 (1985). 7. J. Y. Hnot and L. Brossand, 20, 281 (1990). 8. E. O. Wollan, J. W. Cable and W. C. Koehler, J. Phys. Chem. Solids 24, 1141 (1963). 9. T. V. Lipets, Z. L. Vert and 1. P. Tverdovski, Elektrokhimiya 5, 71 (1969). 10. B. Barankowski and M. Smialowski, J. Phys. Chem. Solids 12, 206 (1956). 11. Z. Szklarska-Smialowska, Bull. Acad. Pol. Sci., Set'. Sci. Chim. 8, 305 (1960). 12. Z. Szklarska-Smialowska and M. Smialowski, J. dectroehem. Soc. 110, 444 (1963). 13. A. R. Troiano, Proeeedings of the ConJerenee on Hydrogen in Metals, ASM Materials/Metalworking Tech. Series, No. 2, p. 3. Champion, PA (1973). 14. G.C. Smith, ibid., p. 485. 15. R. M. Latanision and H. Opperhauser Jr, ibid., p. 539. 16. T. Boniszewski and C. G. Smith, J. Phys. Chem. Solids 21, 115 (1961). 17. T. B. Flanagan and F. A. Lewis, Trans. Faraday Soc. 55, 1400 (1959). 18. Z. Szklarska-Smialowska and M. Smialowski, Bull. Pol. Sci., Ser. Sci. Chim. 6, 187 (1958). 19. M. Enyo in Comprenhensive Treatise ~fElectrochemistry (B. E. Conway, J. O' M Bockris, E. Yeager, S. U. M, Khan and R.E. White, Ed), Vol. 7, p. 241. Plenum Press, New York 0983).

20. A. Frumkin and N. Aladjalova, Acta Physicochim., URSS 19, 1 (1944). 21. R. Clam Roth and C. A. Knorr, Z. Electrochem. 57, 399 (1953). 22. M. V. Stackelberg and H. Bischofl; Z. Electrochem. 59, 407 (1955). 23. J. C. Barton and F. A. Lewis, Z. Phys. Chem. N. F. 33, 99 (1962). 24. B. V. Tilak and B. E. Conway, Electrochim. Acta 21, 745

( 1976). 25. P. Lukovzen, S. Lavina and A. N. Frumkin, Acta Physicochim. 11, 21 (1939) 26. A. Leg ran and S. Lavina, ibid., 12, 243 (1940). 27. J. O" M Bockris and E. C. Potter, J. electrochem. Soc. 99, 169 (1952). 28. T. S. Lee, ibid., 118, 1278 (1971). 29. W. Botter Jr and O. Teschke, J. electrochem. Soc. 138, 1028 (1991). 30. B. E. Conway, H. Angerstein-Kozlowska, M. A. Satter and B. V. Tilak, J. Electochem. Soc. 130, 1825 (1983). 31. W. Botter Jr., D. M. Soares and O. Teschke, J. electroanal. Chem. 267, 276 11989). 32. O. Teschke, D. M. Soares and C. A. P. Evora, J. appl. Electrochem. 13, 371 (1983). 33. M. Elam and B. E. Conway, J. electrochem. Soc. 135, 1678 (1988). 34. J. G. N. Thomas, Trans. Faraday Soc. 57 1603 (1961}. 35. T. Boniszewski and G. C. Smith, J. Phys. Chem. Solids 21, 115 (1961). 36. M. Gupta, in The Electronic Structure of Complex Systems, (P. Phariseau and W. M. Temmerman, Ed), p. 243. NATO ASI Series B, Physics, Vol. 113. Plenum Press, New York.