OF THE RECHARGEABLE MANGANESE ELECTRODE B. SAJDL,K. MICKA* and P.
Acta, Vol. 40, No. 12, pp. 2005-2011. 1995 Copyright 0 1995 Elwier ScienceLtd. Printedin Gnat Britain.All rights ramed 0013~4686/94 $9.50 + 0.M)
Heyrovsky Institute of Physical Chemistry, Academy of Sciences of the Cxech Republic, 182 23 Prague 8, Czech Republic (Received I December 1993; in revisedform 20 February 1994)
Abstract-Manganese dioxide electrode doped with Bi or Pb compounds and containing a conductive component can be reversibly cycled with exploitation of 5040% of the theoretical two-electron capacity; however, after 10 or more cycles there is a negative effect of structural changes consisting in dissolution of Mn oxides, blocking of pores, and impairment of the pore electrolyte conductance. The electrode porosity can be increased by adding a powdered, little soluble compound as pore former, which is gradually leached out, leaving pores. Key words: manganese dioxide, reversible electrode, pore formers.
INTRODUCTION The irreversibility of electrode processes in manganese dioxide electrodes is due to the destruction of the crystal lattice during deep discharge and in volume changes leading to loss of contact between active material particles[l-31. In the initial stage of the reduction, Mn4+ and 02- ions are replaced by Mn”+ and OH- ions with larger radii, the crystal structure being preserved. During this process, which is reversible and limited by the product composition approximately corresponding to MnO,.,, , expansion of the crystal lattice takes place; the overall reaction can be described by the equation[f]
MnO, + mH,O + me- = MnO,_,,,(OH),
where 0 < m < 1 and the product is, in substance, a solid solution of the composition (1 - m)MnO, + mMnOOH. On further discharge, the crystal lattice collapses with the formation of an amorphous phase that is reduced to y-Mn,O, or a-MnOOH and finally to Mn(OH), or the difficultly oxidizable hausmannite, 01= Mn,O,[3, 43. The possibility of realization of a rechargeable manganese dioxide electrode consists, according to the American authors[l, 51, in stabilization of an open birnessite structure by insertion of lead or bismuth ions into the vacancies of the crystal lattice. The development of practical rechargeable alkaline manganese dioxide cells has been perfectioned by Battery Technologies, Inc. (Ontario, Canada). Originally, the cells had a low utilization of the active material (MnO,) not exceeding 35% of the theoretical one-electron capacity (ie low DOD) [2,6}; their capacity was limited intentionally by the zinc anode. An important improvement was the introduction of manganese dioxide doped with titanium, which enabled the utilization of the active material l
Author to whom correspondence should be addressed.
to be increased to 100% of the one-electron capacityC7l. The cited authors also stressed the necessity to eliminate possible loss of contact between the active material particles or between them and the current collector. According to the American authors, manganese dioxide doped with bismuth permits a still higher DOD to be attained. However, they used a high ratio of the active material to graphite added as the conductive component, namely 1:2 or even 1: 10[9-111, which is impractical for cells since the electrode mix is too voluminous. Our present contribution concerns the influence of some factors on the reversibility of the MnO, electrode, namely the method of doping, content of the conductive additives and fillers, electrolyte concentration, etc.
Electrodes were prepared from electrolytic MnO, (Lachema, Brno, Czech Republic) (sedimentation analysis: 49% to 25pm, 87% to 50pm), IBA No. 23 (Kerr-McGee Chem. Corp., U.S.A.) (47% to 30pm, 85% to 60~), Faradiser M WSZ (Chemetals, U.S.A.), and MnO, prepared by oxidation of Mn(OH), precipitated from solutions of Mn(NO,), and KOH, usually in the presence of additives. The conductive components were graphite flakes (particle size in the range lO-ZOO&, graphite CR2 (particle size around 2pm), graphite CR12 (around 12pm), electrographite (around 6Opm) and acetylene black P 1042. Doping agents were Bi,O, (or Bi(NO,), when precipitating Mn(OH),) and PbO, or PbO dissolved in the electrolyte. The following inert substances were used as tillers (pore formers): K,CO,, NazCOj , K,S04, potassium hydrogen tartrate,
B. SAJDL et al.
BaSO, , KCIO, , and cobalt hydroxycarbonate (xCoC0, . yCo(OH),). Physical doping involves mixing the components; the authors[S] assume that a new phase is only formed during cycling. Chemical doping starts from a solution of a Mn(II) salt and a dopant, which is precipitated by alkali hydroxide and oxidized to birnessite by bubbling oxygen. We used solutions of Mn(NO,), in water and Bi(NO,), in 65% HNO,, whose mixture in the desired ratio was transferred quantitatively under stirring into an excess of 7.5 M KOH solution. The temperature rose by the reaction heat but did not exceed 45°C. The precipitate was oxidized by bubbling oxygen or simply (after filtration) by contact with air. After filtration, the precipitate was filtered off, washed, and dried at a maximum temperature of 60°C. Electrodes prepared from the product were charged to complete the oxidation. The conductive components were added into the suspension prior to precipitation. Electrodes
Initially, the electrode mix was pressed into a mould in the form of tablets. This technique was later replaced by inserting the electrode mix into a special cylindrical cell and applying compression (see below). In either case, the electrodes showed volume changes during cycling and finally solidified. The experimental cell consisted of two parts made of organic glass and provided with a cylindrical cavity of 2 cm diameter holding the electrode system (Fig. 1). The test electrode compartment was connected through a channel with a cavity containing a Hg/HgO tablet reference electrode in the same electrolyte (7.5 M KOH). The potential values given below refer to this electrode. The separator consisted of a disc of non-woven polypropylene fabric and a fritted glass disc. The current collector was a graphite or gold disc and the counter electrode was a perforated stainless steel disc provided with a steel
spring to enable compression in the range lo-20N. A current collector made of nickel or stainless steel proved not to work well since it was gradually covered with an oxide layer showing a high ohmic resistance. (A similar observation was made by Roberge and co-workers; the resistance of the oxide layer between MnO, electrode and nickelplated steel can was measured by Barnard and coworkers[ 133.) The electrode mix in the cell was compressed and the cell cavity was, after assembling, filled with the electrolyte, which communicated with the outer space through a water siphon. The equilibrium potential was attained after several tens of minutes, after which the electrode was capable of cycling. Method of measurement The charge acceptance of the test electrodes was evaluated by galvanostatic cycling using a 16channel galvanostat-potentiostat connected with a SAP1 80 computer (Tesla, Prague). The apparatus permitted the charging and discharging currents to be set individually within the range of k 50mA, the anodic and cathodic potential limits within f 1 V, and allowed a choice in the frequency of data recording. The currents used were in the range l-4OmA, the quantities of MnO, were 150-5OOmg, and the current used for charging was usually the same as that for discharging. The positive limiting potential was set equal to + 5OOmV (vs. Hg/HgO), since at higher potentials the MnO, electrode was oxidized to MnO:- or MnO; ions, as indicated by the green or purple colouring of the electrolyte. The negative potential limit influenced the reversibility of the electrode; it was varied within a range from -500 to - 700 mV. The delivered charge was evaluated from the portion of the discharge curve between + 500 and - 5OOmV, and it was expressed in per cent of the theoretical capacity, C, = 37.0 A min g- ’ (for the two-electron reduction of Mn(IV) to Mn(I1)). Thus, the current for 10 h charging is equal to 62mA per lgofMn0,. The IR spectra were recorded on a NICOLET 205 FTIR spectrometer equipped with a DGTS detector. A small quantity of the electrode material was mixed with KBr (Aldrich, IR grade) in a ratio of 1 : 1000 and homogenized. Thin 200 mg pellets were prepared by the conventional technique.
Role of the conductive components
Fig. 1. Schematic cross-section of the measuring cell: (1) test electrode; (2) graphite disc; (3) auxiliary electrode; (4) referenceelectrode; (5) stainless steel disc; (6) polypropylene separator; (7) glass frit; (8) steel spring; (9) gasket; (10) electrolyte inlet (closed by a water siphon during measurement).
The purpose of the first experiment was to find a suitable method of electrode preparation. Electrodes prepared without a conductive component showed a utilization of the active material from l-7% C, corresponding apparently to a thin layer of MnO, in contact with the current collector. Addition of lo-18% electrographite increased the utilization to 36-38% C, in the first cycle at a molar ratio of Pb : Mn from 0.02 to 0.1. However, the utilization dropped from one half to one third after 5-10 cycles. Similar results were obtained with electrodes doped with B&O,.
Study of rechargeable MnO, electrode Replacement of half of the electrographite (content 11%) by acetylene black resulted in an increase of the utilization to 4554% C, during the first several cycles with electrodes doped with Bi,O, . Electrodes containing 6-20% of the conductive component including l-6% of acetylene black gave 41-58% C, after the first discharge, although the utilization dropped to 15-34% C, after 3-10 cycles. Several experiments were made with a cathode mix composition analogous to that in commercial alkaline primary cells, namely 2% of acetylene black, 13% of graphite CR2, and MnO, doped with B&O,. The utilization of the active material dropped from the initial 39% to 32% C, after 10 cycles. On increasing the content of the conductive component to 30%, electrode the utilization of the doped (Bi : Mn = 0.158) increased to 47% C, in the 15th cycle. With a still higher content (eg 85% of graphite CR2 and 5% of acetylene black) the high utilization was maintained approximately constant around 60%. However, a better composition was close to 42% of graphite CR12, 43% graphite flakes, 5% of acetylene black, and 10% of doped MnO,, giving a stable value of 76% C,. It is worth noting that such a high excess of the conductive component was also used by other authors, who reported optimistic results[9-11-J. Influence of the electrolyte and other factors
The utilization of the active material was the same in 7.5 M KOH as in 9 M KOH; however, in 0.75 M KOH it decreased appreciably (not exceeding 30% C,), as observed already by Kordesch and coworkers. The addition of zincate ions (final concentration 0.1 M) had no effect, but higher concentration (0.5 M) caused breakdown of the electrode after l-2 cycles (Fig. 2). A similar effect was observed by Dzieciuch and co-workers[lO], who attributed it to the formation of the inactive hetaerolite, Mn,O, . ZnO. Electrodes showing (in the absence of zincate ions)
a decreased discharge capacity could be partly regenerated by washing, drying, grinding, and placing into the test cell again. This, together with the favourable effect of the increasing content of the conductive component (to 30% and more), suggests that the loss of the discharge capacity on cycling is not related to the microstructure of MnO, but rather with the porous structure of the electrode. Also, the partial dissolution of Mn oxides may be a factor, as evidenced by the formation of black-brown deposits in the fritted glass disc, on the separator, and on other cell components. These deposits originate partly from the decomposition of manganate ions formed during overcharging, but mainly from the decomposition of [Mn(OH)$ions formed during cycling, as shown in a recent detailed study. The loss of the active component caused by dissolution had a non-negligible influence on the drop in electrode capacity. For example, the quantity of Mn oxides deposited in the separator during 30-50 cycles was found to represent 13-39% of the original MnO, content in the electrode. A suitable cell construction, such as that proposed by Dzieciuch and co-workers[lO], could help to decrease the volume of electrolyte and the loss of active material. Scanning electron microphotographs showed that the originally porous and granular electrode surface (a virgin MnO, layer shown in Fig. 3a) changed after cycling to a compact, non-porous one, with cracks at some places (Fig. 3b, c). This development was accompanied by an increase in the electrode resistance (eg from 0.15 to 0.3 R after six cycles) as indicated by impedance spectroscopy records made at different stages of cycling. The resistance increase was also apparent from the magnitude of the potential jump when the current was reversed. Interestingly enough, the electrode mix adhered very firmly to the graphite or gold current collector after cycling, suggesting that the contact resistance is not a problem here. When the positive limiting potential exceeded 5OOmV, the galvanostatic curve showed a plateau due to oxygen evolution. A green colouring of the electrolyte, caused by MnO:- ions, was observed in the potential region 500-6OOmV; and after a longer time (tens of hours) even MnO; ions were formed, causing a purple colouring. Discussion of the potential delays
The standard potential of the reaction Bi,Op + 3H,O + 6e- = 2Bi + 60H-, is equal to -0.46OV. The cathodic potential delay corresponding to deposition of bismuth was observed with a Bi,O;-doped electrode at - 600 mV (C, , Fig. 4), and in this case the electrode performed well. Whereas an electrode prepared from B&O, without MnO, behaves reversibly, the anodic delay, A,-at about -500mV on the curve in Fig. 4-is shorter than the cathodic one, Ci,. Nevertheless, the potentials of the two delays are m good agreement with those of the voltammetric peaks for a Bi,O, electrode in the same medium. The anodic portion of the galvanostatic curve shows three regions, related to the charge acceptance
Fig. 2. Influence of zincate ions on utilization of the active material (MnO, doped with Bi,O, , 15% of graphite flakes, 15% of graphite CR12 and 5% of acetylene black; 10mA). (A) 7.5M KOH without zincate, (0) with 0.1 M ZnO:-, (0) with 0.5 M ZnOs-.
of the electrode : the delay A, around - 250 mV, the delay A, around + lOOmV, and the hump in the region 300-500 mV, corresponding most probably to the oxidation of Mn(II1) to Mn(IV) (Fig. 4). The potentials of the delays A, and A, correspond approximately to those of the voltammetric peaks for the MnO,/Bi,O, electrode in the same medium, namely -0.2 and O.OV[l, 5, lo]. However, the delay A, disappears after about 10 chargdischarge cycles, and with electrodes containing an excess of graphite it does not appear at all. A somewhat analogous phenomenon was observed with nickel hydroxide electrodes giving a second discharge step; this depends on ohmic phenomena in the active mass and tends to disappear with increasing content of the conductive component[l$ 161. Potential delays close to A, and A, were also observed by Mondoloni and co-workers during deep cycling of a non-doped y-MnO, electrode; they attributed the more positive one to oxidation of Mn(I1) to y-Mn,O, or (isostrtG&al) hausmannite. potential of the system The standard Mn,O,/Mn(OH), is equal to about + 0.1 V, that of the system MnO,/Mn,O, to about + 0.2 V[ 18, 193. (The potential of the Hg/HgO electrode in 8 M KOH is practically the same as that of the standard hydrogen electrode.) From this point of view, the interpretation of the delay A2 would be difficult. Concerning the cathodic delay C, at -0.35 V (Fig. 4), which also corresponds to the cathodic peak on the voltammetric curves at -0.35 to -0.4V[l, 5, lo], it is close to that observed by Mondoloni and co-workers[ 171 at -0.42 V with a y-MnO, electrode and attributed by them to the reduction of Mn,O, to Mn(OH), .
of cycling conditions
Repeated deep reduction to a potential limit of - 700 mV had an unfavourable effect : the utilization of the active material dropped from 40 to 14% C, after 10 cycles, as compared with a similar electrode reduced only to -5OOmV, where the utilization dropped from 39 to 32% C, after 10 cycles. It seems that the best conditions are such that the cut-off potential for the first discharge is -700mV and for the subsequent ones - 5OOmV (Fig. 5). During this cycling regime, the utilization increased from 48 to 63% C, after 10 cycles and attained the original value after 25 cycles. On the other hand, when the cut-off potential of -5tKlmV was constantly maintained, the utilization dropped during 10 cycles from 48 to 44% C,. On increasing the discharge current from 5 to lOmA, the utilization dropped only by a few per cent. Role of the method of doping Somewhat different results were obtained with chemically doped MnO,. The utilization of the active material (with 35% of conductive component, Bi:Mn = 0.1) was close to 90% at the beginning of cycling, but dropped to about 40% C, after 10 cycles. At a lower Bi content (Bi:Mn = 0.05) with 42% of the conductive component the utilization was up to 67% at the beginning, but only 23% after 10 cycles, showing a more rapid decrease with cycling than with the physically doped MnO, electrodes. For comparison, some experiments were made with lead dioxide as a dopant. The reduction of PbO, (pure or in mixture with MnO,) was attained
Fig. 3. (a) SEM microphotograph of a virgin MnO, layer (Faradiser M WSZ); (b,c) cycled MnO, elec. .
15% of graphite flakes, 15% of graphite CR12 and 5% of acetylene black) after 23 cycles.
Study of rechargeable MnO, electrode
Fig. 3. (Continued) at -700mV. A potential delay at about + 150mV, similar to A3 in Fig. 4 but smaller and rapidly disappearing, was attained during repeated cycling to a cut-off potential of -750 mV; under these conditions the electrode gave 45543% C, during 23 cycles. On the whole, the results were not as favourable as in the case of electrodes doped with Bi,O, . InJSence ofjillers
Since the electrode utilization is favoured by high concentration of KOH, which is likely to form soluble hydroxo complexes with Mn(III), we studied the effect of tartrate ions, which also form soluble
and stable complexes with Mn(II1). However, addition of potassium tartrate only had a favourable effect if it was added to the electrode mix in powdered form (10%). Thus, the utilization of the active material attained 60% after 22 cycles (Fig. 6). Addition of tartrate into the electrolyte had rather a negative effect. It appears that the particles of potassium tartrate after dissolution contribute to the electrode porosity, enhancing the ion conductivity in the pores and thus the electrode performance. Addition of 10% powdered K&O, had no effect since it was obviously leached out rapidly before the porous structure of the electrode could be stabilized.
B. SAJDL et al.
t / hours
curve for a MnO, electrode doped with B&O, (composition as in Fig. 2) in 7.5 M KOH (7th cycle; 20mA). Fig. 4. Gahanostatic
However, addition of 10% K,SO, had a very favourable influence, leading to a utilization of 81% C, after 10 cycles and 50% after 50 cycles. A similar, although weaker, influence was observed with KClO, , which is sparingly soluble. Barium sulphate, which is insoluble, did not influence the utilization of the electrode material. It is notable that addition of 5% of powdered cobalt hydroxycarbonate, forming particles of a porous structure, also helped to preserve the electrode capacity (the effect being similar to that of KC10.J. These observations show again that the porous structure of the electrodes plays a decisive role in their performance.
Fig. 6. Influence of potassium tartrate on utilization of the active material. (0) Without tartrate, (0) 10% of solid potassium tartrate in the electrode mix, (A) 7.5M KOH with 0.1 M C,H,O, K. Same electrode composition as in Fig. 2.
tally inert. However, with an electrode containing 35% of graphite CR2 and polarized to + SOOmV for about 8 h, the formation of carbonyl groups was observed by ir spectroscopy using a KBr tablet. Besides absorption bands of water in KBr (stretching vibration, about 3500 cm- ‘, and bending vibration, 164Ocm- ‘) a new absorption band at 174Ocm-’ appears in the spectrum (Fig. 7). Its position, together with the composition of the studied electrode, indicates the presence of C-0 bonds. This was not observed with electrodes in the reduced state. A further piece of evidence for increased content of CO groups in samples that had been
Evidence for corrosion of graphite
The conductive component, such as graphite or carbon black, is usually considered as electrochemi-
Fig. 5. Enhancement of utilization of the active material by reduction of the dopant (Biro,) during the first discharge. (0) Without reduction of the dopant, (0) with reduction at -700mV. Electrode mix: 55% MnO,, 39% of graphite CR2 and 6% of acetylene black. Current lOmA.
Wavenumbers /cm-’ Fig. 7. Ir spectrum of an electrode sample (65% of MnO, doped with B&O, and 35% of graphite CR2) after prolonged polarization to + 500mV showing corrosion attack of the admixed graphite (absorption band at 174Ocm-I). The absorption band at 164Ocn-’ corresponds to traces of water.
Study of rechargeable MnO, electrode
overcharged for a longer time was obtained by the ESCA method, mainly in the surface layer of the tablet electrodes.
CONCLUSIONS Manganese dioxide electrodes doped with Bi or Pb oxides undergo structural changes during many discharge-charge cycles leading to gradual decrease of the discharge capacity from 50-90% of the theoretical two-electron capacity to less than 50%. This is due to dissolution of Mn oxides in the KOH electrolyte, clogging of the pores and the resulting increase of the pore electrolyte resistance. A high content of the conductive component (above 25%) and/or addition of a pore-former, help to preserve the porous structure and thus the high utilization of the active material. Acknowledgements-The authors are indebted to Dr Z. Bastl for analyses by the ESCA method, to Dr V. Masafik for SEM microphotographs, and to Dr Z. ZBbransk$ for sedimentation analyses.
REFERENCES 1. Y. F. Fao, N. Gupta and H. S. Wroblowa, J. electroanal. Chem. 223,107 (1987). 2. K. Kordesch, J. Gsellmann, M. Peri, K. Tomantschger and R. Chemelli, Electrochim. Acta 26, 1495 (1981). 3. D. Boden, C. J. Venuto, D. Wisler and R. B. Wylie, J. Electrochem. Sot. 114,415 (1967).
4. A. Kozawa and J. F. Yeager, J. Electrochem. Sot. 112, 959 (1965). 5. H. S. Wroblowa and N. Gupta, J. electroad. Chem. 238,93 (1987). 6. J. Gsellmann, W. Harer, K. Holzleithner and K. Kordesch, in Proc. Symp. MnO, Electrode Theory and Practice for Electrochem. Applications (Edited by B. Schumm Jr, R. L. Middaugh, M. P. Grotheer and J. C. Hunter) Vol. 85-4, p. 567. The Electrochem. Sot., Pennington, N. J. (1985). 7. K. Kordesch and J. Daniel&ad, Progr. Batteries and Battery Mat. 11,70 (1992). 8. Y. F. Yao, U.S. patent 4,520,005 (1985). 9. M. A. Dzieciuch, H. S. Wroblowa and J. T. Kummer, U.S. natent 4.451.543 (19841. 10. M. A. Dzie&ch, N.‘Gup;a and H. S. Wroblowa, J. Electrochem. Sot. 135,2415 (1988). 11. L. Bai, D. Y. Qu, B. E. Conway, Y. H. Zhou, G. Chowdhury and W. A. Adams, J. Electrochem. Sot. 140,884 (1993). 12. P. R. Roberge, M. Farahani and K. Tomantschger, J. Power Sources 41, 321 (1993). 13. R. Barnard, L. M. Baugh and C. F. Randell, J. Appl. Electrochem. 17,165 (1987). 14. D. Y. Qu, B. E. Conway, L. Bai, Y. H. Zhou and W. A. Adams, J. Appl. Electrochem. 23,693 (1993). 15. B. Klipst6, J. Mrha, K. Micka, J. Jindra and V. MareEek, J. Power Sources 4,349 (1979). 16. B. KllpF;t&, K. Micka, J. Mrha and J. Vondrik, J. Power Sources 8, 351 (1982). 17. Ch. Mondoloni, M. Laborde, J. Rioux, E. Andoni and C. Levy-Clement, J. Electrochem. Sot. 139,954 (1992). 18. Spravochnik po elektrokhimii (Edited by A. M. Sukhotin) Khimiya, Leningrad (1981). 19. W. M. Latimer, Oxidation Potentials, Prentice-Hall, Englewood Cliffs, N. J. (1961). 20. G. Socrates, Infrared Characteristic Group Frequencies, J. Wiley, Chichester (1980).