The electrochromic process in non-stoichiometric nickel oxide thin film electrodes

Ekcsmhkmlm Act.,, Vol. 37, No . 6, pp . 1033-10311,1992 Panted in Great Britain .

0013-4686/9215.00+0 .O0 0 1992. Petpmon Pane plc .

THE ELECTROCHROMIC PROCESS IN NON-STOICHIOMETRIC NICKEL OXIDE THIN FILM ELECTRODES F. DECKER, S . PABBERINI, R . Plcacct and B. ScaosAn Dipartimento di Chimica, Universitd di Roma'La Sapienza, Rome, Italy (Received 15 April 1991 ; in revised form 9 September 1991)

Abatraet-Non-stoichiometric nickel oxide thin films are very interesting novel electrochromic electrodes for the realization of optical displays and windows of relevant technological importance . However, the exact nature of the electrochromic process in these electrodes is not yet fully clear . In this work we attempt to reach a more precise understanding of this process by examining in detail the electrochemical properties of the intercalation process of lithium ions in the nickel oxide host structure . Investigation based on cyclic voltammetry, potentlometry, impedance analysis, optical transmittance, stress measurements, scanning electron microscopy and secondary ion mass spectroscopy, suggest that the intercalation of lithium induces phase changes in the nickel oxide structure, namely a first phase related to the initial injection of lithium (with which is associated a bleaching process) and a second phase related to an excess of inserted lithium (with which is associated an opposite, darkening process) . Key words : electrochromism, nickel oxide, lithium intercalation.

INTRODUCTION

EXPERIMENTAL

The intercalation process in non-stoichiometric nickel oxide, NiO„ thin-film electrodes and the related electrochromic effect have been studied quite consistently[1] . Following the well-known proton insertion process carried out in aqueous media, many studies have been performed in the past years on the injection reaction of small-size cations in amorphous NiO, in aqueous, as well as in water-free systems[2, 3] . More recently, Scarminio et al have reported a detailed investigation on the electrochromic effect observed during the intercalation-deintercalation process of protons in completely anhydrous NiO, thin films[4] . Similarly, we have demonstrated in our laboratory that under anhydrous conditions, lithium ions may as well undergo an insertion electrochemical reaction in the same NiO, host matrix[5] . The results, even if at a preliminary stage, have clearly evidenced that the reaction is highly reversible and accompanied by a sharp chromic effect . Due to these characteristics, NiO, thin films can be proposed as new, highly reversible electrochromic electrodes for the realization of improved displays and windows of relevant technological interest[5, 61 . However, the exact nature and mechanism of the electrochromic process in NiO, electrodes is not yet fully clear. Consequently, it has appeared to us of importance to complete our previous work[5] by examining in more detail the electrochemical properties of the intercalation process of lithium ions in the NiO, host structure, as well as the chemical-physical characteristics of the lithiated Li,,NiO, compound . In this paper we report and discuss results obtained with various electrochemical and spectroseopical techniques.

Thin-film nickel oxide electrodes were kindly provided by Dr Anne Andersson of the Chalmers University of Technology in Sweden . These electrodes were obtained by dc-reactive sputtering deposition on glass supports, using a nickel target and following the procedure described in the reference[7] . Two types of indium tin oxide (17170)-coated glass supports were used, depending upon the type of measurement required : Coming cover glass plates 150pm thick were selected for stress measurements, while boro-silicate glass plates I mm thick were used for all the other tests . Under the conditions of preparation of the NiO, films, it is not possible to indicate the precise value of x. We can only assume that we are dealing with non-stoichiometric nickel oxide samples . Ultra-pure, HPLC grade chemicals, lithium perchlorate and propylene carbonate were used as received . Since water impurities, even if in traces, are deleterious for the electrochromic process(5], all the chemicals were stored under anhydrous conditions and the chemical manipulation, as well as cell assemblages and tests, were carried out in a controlled-atmosphere (water content less than 10ppm), argon-filled, dry-box . The electrochemical processes were studied in three-electrode, glass cells where the reference electrode (re) was a lithium foil and the electrolyte was a I M solution' of lithium perchlorate in propylene carbonate (hereafter simply abbreviated as LiCIOrPC) . Unless otherwise specified, the counter electrode (ce) was also a lithium foil . The electrochemical measurements, such as cyclic voltammetry and potentiostatic curves, were driven and controlled

1033



1034

F. Dscxaa et al.

by an AMEL potentiostat-galvanostat (model 551), an AMEL square wave generator (model 565), an AMEL ramp generator (model 567) and by YEW X-T and X-Y recorders . The impedance measurements were carried out with a Solartron frequency analyser (model 1255) over the frequency range of 1 .0 mHz to 65 kHz in combination with a Solartron electrochemical interface (model 1286) . The entire equipment was dynamically controlled by a personal computer (HP Vectra) used for both the optimization of the measurements and the data collection . A fitting analysis of the experimental impedance data was performed using the NLLS fitting program written by Boukamp[8] . The procedure for the optical and stress measurements was described in detail in previous works[5, 9] . In synthesis, the transmittance of the electrodes was followed in situ using a He-Ne laser (A = 6328 A) as the light source and an UDT light detector, all the procedure being carried out on an optical bench assembled and maintained in the dry-box . The transmittance values are reported using as the reference (100% transmittance) the system formed by the bare glass cell plus electrolyte . A similar procedure, ie a laser light source and a position detector (UTD, LSC-30D) to analyse the direction of the reflected laser beam, was used for the stress measurements . Again, the entire optical bench was maintained and operated in the dry-box . Finally, the elemental and morphological analyses on thin-film Li,NiO, samples were performed by an Atomika Secondary Ion Mass Spectroscopy (SIMS) equipment using oxygen as primary ions source and by a Cambridge 360 Scanning Electron Microscope (SEM), respectively .

upon cycling. In fact, while in the few following cycles only some modification in the cathodic peak and a small anodic peak occurs (Fig . Ib), after repeated cycles a consistent rise in the anodic stripping peak is clearly observed (Fig. Ic). Furthermore, when the cathodic limit is lowered to I V, two cathodic peaks are well evidenced (Fig. ld) and, upon continuing the cycling test, the main cathodic and anodic peaks get closer, this final behavior remaining reproducible and constant over a large number of cycles (Fig . le) . This sequential voltammetric response is interpreted assuming that the cathodic current flowing in the initial cycles is associated with an irreversible process of the type (1), which in turn may be described as the uptake of lithium ions, this producing a permanent change in the nickel oxide film structure . Such a change is favorable for promoting further and progressively more reversible insertion of lithium ions [process (2)], probably by widening the structural slabs within which the lithium ions can diffuse . Unfortunately, it is almost impossible, except with the use of highly sophisticated equipment, to O 0

Cycle

3

-20 0

,root

cycle

6

-20 0

RESULTS AND DISCUSSION j

It has been shown in previous works[5, 10] that, under rigorously anhydrous conditions, thin-film, non-stoichiometric nickel oxide NiO, electrodes, are able to intercalate lithium ions, according to the following, oversimplified electrochemical reaction NiO,+yLi'+ye - -*Li,,NiO„

(1)

-20

Cy

cle 8

0 0 U

g

-20

and that the injection of the initial ions "activates" the host structures for a subsequent fast and reversible insertion and deinsertion process of the type

_ (*Mm ~ Cycle 16

mlo vg

Liy NiO, Lily _„NiO,+zLi*+ze - ,

(2)

bleaching

which in turn is accompanied by a fast and reversible electrochromic effect . The initial "activation" process can be accomplished using different electrochemical means . One possibility is to drive and follow the process by cyclic voltammetry and a typical response in terms of developments of subsequential cycles is illustrated in Fig. I a-e. Following an initial cycling potential sweep between .11 V and 3 .7 V (vs . Li), a cathodic current flows below 1 .5 V while no comparable anodic current appears in the reverse anodic scan (Fig . la) . However, the shape and the characteristics of the current-voltage curves changes consistently

-2

f

Subsequent cycles

c cle 1 .0

20

2 .0

3 .0

Voltage, V vs. Ll Fig. 1 . Subsequential cyclic voltammetry of NiO, electrodes in the LiC1O,PC solution . Curve a : 3rd cycle, voltage limits 1 .1-3 .7 V . Curve b : 6th cycle, voltage limits 1 .1-3.7 V . Curve c : 8th cycle, voltage limits 1 .1-3 .5 V . Curve e: 20th and subsequent cycles, voltage limits 1 .0-2 .5 V. Scan rate: 10mVs' .



Electrochromic process in NiO, thin film electrodes

0

200

400

600

Charge, mC cm'pm'

Fig . 2. Variation of voltage (vs. .Li) and of optical transmittance (top) of NiO, electrodes in the L1C1O,PC solution in dependence of the charge injected by a low-rate galvanostatic polarization . Current density : 1 .1 pA cm- ' . confirm by structural studies the proposed "activation" process. On the other hand, this type of process is common in insertion electrochemistry and it has been observed and experienced in various intercalation compounds[] 1] . Therefore, if the assumption is also correct in the case under study, the activation process should be achieved by any technique which promotes the controlled injection of lithium ions in the NiO, host structure . Effectively, pristine nickel oxide electrodes can be activated by cathodic polarization carried out either potentiostatically (fixed potential) or galvanostatically (fixed current) . In the former case, the electrolysis must be run at the voltage suitable for the injection process, ie 1 .1 V vs. Li (compare Fig. la-c), while in the latter case, very low current rates (typically few pA cm -3 ) should be used to cope with the unavoidably slow initial process of expansion of the host structure . Figure 2 illustrates the variation of voltage and of optical transmittance of the nickel oxide electrode following a galvanostatic (1 .1 pA cm') polarization . We notice that upon passage of current is upon injection of lithium ions, the potential steeply decays to reach the value typical of the injection process (namely around I V vs . Li, see Fig. 1), then rises and finally stabilizes . If the current is temporarily interrupted, the potential of the Li,NiO, electrode sharply increases, to gradually return to the plateau value after restoration of the current flow . This behavior is again typical of intercalation electrodes and it is indicative of a slow diffusion of the inserted ions from the interface into the bulk of the host structure . The voltage trend is accompanied by a significant variation of the transmittance, detected in situ during the evolution of the polarization . The transmittance steeply grows, to reach a maximum (approaching 100%) when the injected charge is about 150mC cm -2 pm - ' . Further insertion of lithium produces a slow but consistent decay in transmittance . The change in optical transmittance upon lithium insertion, suggests that the modifications of lithiated

1035

nickel oxide electrodes are more complicated than a simple "activation" process consisting of widening the host structure . The fact that, following the bleaching process associated with the initial charging step, the electron becomes colored in a progressive fashion, suggests the presence of different phases, at least two, in the Li,NiO, structure . On the basis of the experimental results, it is reasonable to assume, a first phase (phase 1) related to the initial insertion of lithium and to which is associated the bleaching process . This first phase remains stable until the number of inserted lithium ions is lower than that of the available interstitial sites in the host structure . When the forced injection exceeds this number, the structure rearranges into a second phase (phase 2) to accommodate the excess of lithium, this being associated with an opposite, darkening process . In this two-phase transition, the potential remains fixed and becomes characteristic of the process of the insertion of lithium in the second formed phase . The experimental results (see Fig. 2) indicate that the value of the potential representative of the two phases is around 1 .2 V vs . Li, while the critical charge for the phase rearrangement is around 150 mC cm -2 pm - ' . There are further experimental results which confirm this interpretation . Figure 3 illustrates the parallel variation of the stress angle and the optical transmittance of a Li,NiO, sample, induced by a potentiostatic (1 .1 V vs . Li) polarization . We notice that the inversion in transmittance at the critical value (l50 mC em'put') is accompanied by a correspondent change in the slope of the stress angle trend . A change in stress angle is representative of a modification in the mode of the structural expansion induced by the intercalating ion and thus the results of Fig. 3 confirm that beyond the critical charge a phase rearrangement takes place in the host structure . Figure 4 shows a SEM analysis of the surface of a heavily lithiated (l057mCcm - 'pm `) Li,NiO,

Phase I

Phase 2

NIO, film 0.2 pm thick

CPS

200

400

put' Fig. 3. Variation of the optical transmittance (top) and of the stress angle (bottom) of Nio, electrodes in the UCIO,-PC solution in dependence of the chargee injected by a potentiostatic (1 .1 . V us . Li) polarization . Charge . mC cm'



F.

1036

DECKER

et al.

Fig . 4 . Scanning Electron Microscope (SEM) picture of thee surface of a heavily fithiated Li,NiO, sample. sample . The occurrence of the two phases is clearly indicated by the coexistence of the dark spots (phase 2, Li-rich) on the clear substrate (phase 1, Li-poor) . Further information on the characteristics of lithiated nickel oxide samples is provided by SIMS analysis . Figure 5a illustrates the response obtained with a pristine, lithium-free sample : the count-rate is here independent from the sputter time until the boundary NiO,/ITO is reached, as evidenced by the increment of the count-rate for tin . In the analysis of 2 lithiated samples (Fig . 5b, 125 mC cm_ pm - ' and Fig . 5c, 1057 mC cm -2 µm - ' ), we notice that, as expected, the more is the charge injected the higher is the count-rate for lithium . In addition, the most heavily intercalated sample (Fig. 5c) shows a surface accumulation of lithium (as evidenced by the change in slope of the Li curve at the lower (0 to 10 min) sputter time), as well as an apparent Li injection in the ITO layer (as suggested by the tail in the same curve at higher sputter time) . The latter is the typical 10°

instrumental response for a non-homogeneous system, this again confirming that lithium-rich samples indeed present a two-phase structure . All the discussed results support the two-phase model here proposed for describing the Li insertion process in nickel oxide electrodes . However, the final confirmation of this model would require the exact identification of the two phases and work, mainly based on spectroscopic techniques, is indeed planned in our laboratory to achieve this goal . It is important to point out here that the process involving phase I [ie process (2) up to about 150 mC cm -2 pm - ', corresponding to about 15 mC cm - t under the constructive characteristics of our samples], is fast,, reversible and accompanied by very small changes in transmittance . This fact, which has been clearly established by the above discussion, is finally confirmed by Fig . 6 which reports the cyclic voltammetry and the in situ transmittance of a Li,NiO, sample cycled within the cited charge limits .

NiO,

LN

0

0

- 0 Sn02 ;)02 03

m

--o

g td 10

I

2 0

10

20

30

0

10

20

30

,Ni 40

0

10

20

30

40

Time, min

Fig .

5. Secondary Ion Mass Spectroscopy (SIMS) analysis of a pristine NiO, electrode (curve a) and of lithiated (125 mC M -2 AM -t' curve b, and 1057 mC cm - rpm ', curve c) Li,.NiO, electrodes .



Electrochromic process in NiO, thin film electrodes This is a very interesting property from the practical point of view, since it shows that lithiated nickel oxide films may be used as an almost optically passive electrode and thus as a very convenient counterelectrode for the realization of efficient windows based on tungsten oxide, WO j , as main electrochroniic material. In fact, the amount of lithium necessary to induce in W0 3 a change in transmittance of AT = 50%, is around 14.7 mC cm'' (considering an electrochromic efficiency for WO, of 0 .024 cm' mC -1 ) . Assuming as a counter electrode a lithiated nickel oxide having a thickness of 1200 A, the amount of charge to be cycled to assure the WO, coloration results to be 123 mC cm - ',um - ', is a value lower than that which critically affects the phase change and far below than that which induces the appearance of black spots on the Li Y NiO, electrode. In such a way, this latter electrode can safely operate within the stability range of its optically favorable phase 1, and use the remaining excess of charge to account for the unavoidable losses of efficiency experienced by any practical devices, including electrochromic windows . The promising behavior of lithiated nickel oxide as optically passive counterelectrode in WO,-based smart windows, has been ascertained in our laboratory[5,12] . The experimental results have clearly shown that these new type of windows do indeed show very good performance in terms of optical contrast and cyclability . However, these characteristics, even though they are of great importance, are not sufficient to ensure a versatile application of the device, whose universal success is not only related to good and repetitive optical responses but also to fast switching times. This latter property is directly connected to the kinetics of the electrochromic process which, being based on insertion reactions, are likely to be controlled by the diffusion of lithium ions to and from the nickel oxide host structure . ~, 100 U m

Ew

80

m

20

1037

1000

a

Ni

500

0 0

200

0

400

800

z-"Q Fig. 7. Impedance response of a pristine, Li-free NiO, electrode (curve a) and of a lightly intercalated Li,NiO, electrode (curve b). LiCll),PC electrolyte. Frequency range: 0.001 Hz-65 kHz .

Therefore, kinetic studies appear to be of particular importance in the characterization of Li,Ni0 . thinfilm electrocbromic . These studies have been performed using impedance spectroscopy, which has been proven to be a very convenient technique for investigating the electrode kinetics in solid-state reactions[13] . Figure 7a shows the impedance response of a pristine, lithium-free NiO, electrode . As expected, the sample behaves as a poor conductor and the response approaches that of a pure capacitance . More significant is in comparison the behavior of a lightly intercalated (phase 1) sample (Fig. 7b) . Here the high to low frequency (f) response develops with a relaxation arc characteristic of the charge transfer process, followed by a 45° Warburg line, representative of a mass transport process and, finally, a very low frequency spike which is indicative of a limiting capacity behavior . This overall response approaches that expected on the basis of a model developed for thin-layer intercalation electrodes[14] . This fact, beside further confirming the insertion nature of the electrochromic process of nickel oxide electrodes, gives us confidence in using the model equations for evaluating the kinetic parameters of this process[14] . In particular, the analysis of the response in the low frequency, limiting capacity regions, allows us to obtain qualitative information on the value of the lithium diffusion coefficient, D, in the nickel oxide host matrix, by the use of the equation 12 D

i

0

3C, RL,

(3)

c

e C) U -20

-40 1 .0

1 .5

2 .0

2 .5

Voltage. V vs. L I

voltammetry (bottom) and in situ optical transmittance (top) of a Li NiO, electrode maintained within the limits of phase 1 . LiCIO,PC solution. Scan rate: 10 mV s - ' . Fig . 6. Cyclic

where l is the thickness of the sample, C L and RL , are the limiting capacity and the limiting resistance, respectively . One has to point out that these latter values are obtained by the experimental curve of Fig . 7 using a suitable fitting program[8] which accounts only to a certain degree for the deviations from the ideality . Consequently, the obtained diffusion coefficient values are affected by some uncertainties and, without confirmation by other, independent techniques, they must necessarily be regarded as indicative. On the other hand, the impedance method is generally accepted as one of the most appropriate



1038

F . DEcran et al.

and convenient for providing rapid, qualitative information on the diffusion of ions in solid host structures(151 . On the basis of this consideration and in line with this approach, we may here propose for the diffusion coefficient of lithium ions in nickel oxide the tentative value of D = 1 .0 x 10 - " cm' s'' at room temperature. This value is lower than those typically associated with fast diffusion in intercalation processes (ie D > 10 -9 cmr s - ') but still much higher than those representative of unacceptably slow processes (ie D < 10 -1 d cinz s - '). Also in conclusion, the impedance result may be regarded as a further indication of the promising properties of nickel oxide as a novel, high-performance electrochromic electrode . Work is in progress in our laboratory to confirm the diffusion data, using alternative electrochemical and physical techniques as well as to improve the response of the nickel oxide electrode, possibly by properly modifying its surface morphology . Acknowledgements-We would like to thank Dr Anne Andersson of the Department of Physics of the Chalmers University of Technology in Sweden for having kindly provided the NiO, samples, and V. Vittori of ENIRICERCHE of Monterotondo, Rome for the SIMS measurements. Finally, the financial support of the Consiglio Nazionale delle Ricerche, Progetto Finalizzato Materiali Speciali per Tecnologie Avanzate, is acknowledged .

REFERENCES 1 . J. S . E M . Svensson and G . C . Granqvist, App! . Phys . Lett. 49, 1566 (1986). 2 . W . Estrada, A . M . Andersson and G . C. Grangvist, J. appl. Phys . 64, 3678 (1988) . 3 . F . Hahn, D . Floner, B . Beolen and C . Iamy, Ekctrochim . Acta 32, 1631 (1987). 4. J . Scarminio, W . Estrada, A . Andersson, A. Gorenstein and F. Decker, J. electrochem . Soc., submitted . 5. S . Passerine, B . Scrosati and A . Gorenstein, J. electrochem. Soc . 137, 3297 (1990) . 6 . B . Scrosati, Chum . Oggi 7, 41 (1989) . 7 . T. Eriksson and G . C. Granqvist, J . app! . Phys . 60, 2081 (1986) . 8 . B. A . Boukamp, Solid St. Ionics 20, 31 (1986) . 9 . S. N . Sahu, J. Scarminio and F . Decker, J. electrochem. Soc. 137, 1150 (1990). 10 . R . Pileggi, B . Scrosati and S . Passerine, Proc. Symp. on Solid State Ionics, MRS Conference, Boston, MA (Nov . 1990). 11 . S . M . Whittingham, Prog. Solid St. Chem, 12, 41 (1970) . 12 . S . Passerini, B . Scrosati, A. Gorenstein, A . M . Andersson and G . C . Grangvist, J. electrochem . Soc . 136, 3394 (1989) . 13 . J . Ross Macdonald, Impedance Spectroscopy . John Wiley & Sons, New York (1987) . 14 . C . Ho, 1 . D . Raistrick and R . A . Huggins, J. electrochem . Soc . 127, 343 (1980) . IS . S . Panero, P. Prosperi, B. Scrosati and D . D . Perlmutter, J. electrochem . Soc . 136, 3729 (1989) .