Electrochromic Performance of NiO_x Thin Film on Flexible PET/ITO Prepared by Nanocrystallite-Dispersion Sol

Nickel oxide thin film has been widely investigated for electrochromic applications.¹ It is commonly used as anodically colored counter electrode with tungsten oxide electrode which is cathodically colored.² The electrochromism of its thin film has generally been demonstrated with high color contrast and colorization efficiency in an alkaline aqueous solution. The electrochromic mechanism of nickel oxide in alkaline aqueous solution is somewhat controversial,³ but what is generally accepted is that NiO is initially converted to colorless Ni(OH)₂ and then Ni(OH)₂ is reversibly oxidized to black NiOOH.⁴ Strictly speaking, the electrochromism in alkaline aqueous solution is a characteristics of nickel hydroxide, not nickel oxide. In addition, the alkaline aqueous electrolyte solutions are not particularly useful for practical smart window devices because of the lack of long-term stability due to the aqueous-solubility of Ni(OH)₂. The electrochromic performance of NiO thin films in nonaqueous electrolyte solution is generally not as good as in alkaline aqueous solution. This is because the cation insertion into the lattice of NiO is involved in the redox mechanism differently from that in the aqueous solution.⁵

Thin films of NiO have been prepared by many methods such as RF-sputtering,⁶ chemical vapor deposition,⁷ electrochemical deposition,⁸ pulsed laser deposition,⁹ spray pyrolysis,¹⁰ chemical bath deposition¹¹ and sol-gel methods.¹² However, most of these methods are not suitable for large scale production except the sol-gel method. Even the sol-gel method typically requires the thermal conversion of the precursor on substrate above 300 °C after sol coating. Therefore, flexible polymer substrates such as polyethylene terephthalate (PET) cannot be used for the thin film preparation on the substrate in most sol-gel methods because such a substrate is not usually stable above 150 °C. In order to implement a practical smart window, it is essential to establish an appropriate mass production method of an electrochromic thin film device with low cost.

Herein we report a very facile method for preparing NiO_x thin film on any kinds of flat substrates as well as flexible polymer substrates that can be easily applied to roll-to-roll mass production. In this method nanocrystalline NiO_x with a crystallite size of 3-4 nm is mechanically dispersed in an alcoholic solvent with a silane compound as binder to prepare a thin film coating sol. Since the synthesized nickel oxide is nonstoichiometric, we denote nickel oxide as NiO_x instead of NiO. This sol differs from the sol of the general sol-gel method in that it does not require the precursor conversion by heat treatment after deposition. This nanocrystallite-dispersed sol method provided a highly reproducible and robust NiO_x thin film on ITO/PET substrates without post high-temperature heat treatment. We investigated the electrochromic performance of this NiO_x thin film on PET/ITO substrate in a nonaqueous electrolyte solution.

NiO nanoparticles are usually synthesized by the thermal conversion of Ni(OH)₂ and NiOOH. Therefore the crystal phase, crystallite size and morphology of the precursor particles, Ni(OH)₂ and NiOOH strongly affect the physical and chemical properties of the final product, NiO. We have prepared nanocrystalline Ni(OH)₂ as precursor of NiO by chemical precipitation method at an elevated temperature using weak base ammonia. Ni(OH)₂ is known to have two pseudopolymorphs of α- and β-phase. The crystal structure and morphology of Ni(OH)₂ synthesized by chemical precipitation method strongly depend upon the temperature, pH and the type of Ni precursor salt.¹⁵ The precipitation method, herein used with ammonia base instead of strong base, could give the delicate phase control between α- and β-phases of Ni(OH)₂ depending both on the temperature and the amount of base.¹⁶ The XRD pattern of the synthesized Ni(OH)₂ in our experimental condition was shown in Fig. 1(a). The pattern indicates most of β-phase structure as assigned in the figure except two low angle peaks.¹⁷ The unassigned peaks at low angles seem to be ascribed to the phase which is neither α nor β phase but contains some structural ingredients of both by an interstratification.¹⁸ The peaks converges to single peak of (001) of β-phase from more widely separated peaks as the precipitation condition changes.¹⁶ As Kamath et al. mentioned,¹⁸ the crystal phase of the synthesized Ni(OH)₂ seems to be in the middle of α→β transformation by interstratification. Fig. 2(a) shows FE-SEM image of the synthesized Ni(OH)₂ powder. The morphology is more like pieces of thin sheet or flower leaf rather than particulate. This nickel hydroxide was annealed at 400 °C in air. It was converted to nickel oxide with a typical cubic crystal structure. Fig. 1(b) shows the XRD pattern of the synthesized nickel oxide with the reflection planes assigned to cubic crystal structure.¹⁹ The peaks correspond to the typical cubic NiO pattern and the unit cell parameter a was estimated as 4.20 Å from Nelson-Riley extrapolation. That was slightly larger than the known value, 4.17 Å. The crystallite size was also estimated to 3.6 nm from (200) peak using Sherrer equation.²⁰ The morphology of the nickel oxide exactly followed that of the precursor hydroxide without any change as shown in Fig. 2(b).

Figure1.

XRD patterns of the synthesized (a) Ni(OH)₂ and (b) NiO_x powders.

Figure2.

FE-SEM pictures of the synthesized (a) Ni(OH)₂ and (b) NiO_x powders.

The synthesized NiO seemed to be non-stoichiometric as expected from the dark black color of the powder. We have measured XPS data in order to check the oxidation state of Ni in this nickel oxide. Fig. 3 shows the high resolution XPS spectrum of Ni_2p. The spectrum shows 2p_2/3 and 2p_1/2 peaks of Ni at 855 eV and 872 eV, respectively with satellite peaks. These peaks were deconvoluted into two peaks Ni²⁺ (854.1 and 871.5 eV) and Ni³⁺ (856.2 and 873.7 eV) as shown in the colored lines in the figure. The percentage of Ni³⁺ was about 40% in the total Ni species from the relative ratio of the deconvoluted peaks. Since XPS data shows only the surface property, the bulk composition of Ni³⁺ will be lower than this value. These Ni³⁺ species are thought to be due to the nickel cation vacancy and/or interstitial oxygen and mainly located in the grain boundaries of oxygen-rich NiO²¹ since there is no indication of the existence of Ni₂O₃ phase in the pattern of XRD. Therefore, we denoted the synthesized nickel oxide as nonstoichiometric NiO_x.

Figure3.

The high resolution XPS spectrum of Ni_2p in the synthesized nickel oxide.

The NiO_x film on PET/ITO substrate was prepared by bar-coating method using the nanocrystallite-dispersion sol. The synthesized NiO_x was dispersed in ethanol by mechanical ball-milling to prepare a coating sol for its thin film. Fig. 4(a) shows the TEM image of the dispersed NiO_x nanoparticles in ethanol after ball-milling. Although the dispersed individual particles are not clearly shown, the sizes of individual particle are less than 10 nm. One of the important features of the electrochromic thin film is optical transparency, i.e. haze. The haze of the prepared thin film could be controlled at the level of less than 5% with our dispersed sol. The morphology of NiO_x thin film on PET/ITO is shown in Fig. 4(b). The surface morphology of the NiO_x thin film with a thickness of about 400 nm is somehow porous on a scale smaller than a submicrometer but very uniformly smooth on a scale larger than a micrometer.

Figure4.

(a) TEM image of NiO_x nanoparticles in the dispersed sol after being ball-milled in ethanol (bar scale: 20 nm). (b) FE-SEM image of the NiO_x film on PET/ITO (bar scale: 100 nm).

The redox properties of this NiO_x thin film on PET/ITO was first studied by cyclic voltammetry (CV) in a nonaqueous LiClO₄ electrolyte solution. Fig. 5 shows the cyclic voltammograms of NiO_x film on PET/ITO and bare PET/ITO in 0.1 M LiClO₄ PC solution at a scan rate of 25 mV/s for 10 cycles. In this nonaqueous electrolyte solution the redox process of NiO_x can be considered Li⁺ ion intercalationdeintercalation process to be involved according to the following reactions.⁵

Figure5.

Cyclic voltammograms of the NiO_x thin film on PET/ITO and bare PET/ITO in 0.1 M LiClO₄ PC solution for 10 cycles with a scan rate of 25 mV/s.

The second reversible redox process which gives the electrochromic color change will depends on the first reduction process of the oxidized Ni³⁺ because the amount of the accompanied Li⁺ intercalation will determine the subsequent reversible process. In the voltammogram as shown in Fig. 5 the current starts with a large cathodic current that corresponds to the reduction of Ni³⁺ and Li⁺ -intercalation and the reversible distinct redox peaks are shown from the first cycle with the background current that corresponds to a typical charging current of a semiconductor electrode. According to the earlier studies of Passerini et al. for NiO_x thin film prepared by sputtering,⁵ the initial activation process was needed to get the second reversible redox process. The activation to give a widening of host structure for Li⁺ could be done by cycling 20 times in cyclic voltammetry or a low-rate galvanostatic polarization. In our case there is no need for that kind of cycling for activation as shown in the cyclic voltammogram. The result can be attributed to the high porosity of the film and the high number of the surface-exposed Ni³⁺ species in the film. Sung et al. also reported the similar result to ours on NiO film that was prepared by the sol-gel method using an aqueous N,N-dimethylaminoethanol solution.^12(f) However, their film characteristics is completely different from ours because the crystallization in the sol-gel method occurred after film formation whereas our film was formed by precrystallized NiO_x.

The electrochromic color change of this film during the cycling in CV was visually shown in Fig. 6. The color change from transparent pale green to deep dark brown is clearly demonstrated at the cathodic and anodic end potentials. The electrochromic performances of this NiO_x film were quantitatively investigated by the repeated step potential chronoamperometry (CA) combined with spectrophotometry. Fig. 7 shows the part of CA response of the NiO_x electrode at the repeated step-potentials of +1.2 V and −0.8 V for each 30 s during 50 cycles. The cathodic maximum current that gives the bleached state is about 2 times higher than the anodic maximum current that gives the colored state and the cathodic response time also looks faster than the anodic one. This means that the reduction process that needs Li⁺-insertion is easier than oxidation process that accompanies Li⁺ extraction. This is apparently an unexpected result because the insertion of ion into crystal structure usually seems to be harder than the extraction. However, this result could be due to the surface redox process where the deep ion insertion is not necessary. Since our NiO_x film was prepared by the pre-crystallized nanoparticles that have a high ratio of Ni³⁺ component on surface and silica binder, the well-dispersed nanoparticles were interconnected by silica and the electroactive surface that were covered initially with Ni³⁺ instead of Ni²⁺ is relatively large as shown in XPS data. Therefore, the redox process might be dominated mostly by the surface components. One more noticeable feature is that the anodic current that remained at the end of the anodic step potential is much higher than that of the corresponding cathodic current. However, the entire current profile was kept well constant without a recognizable degradation for 50 cycles.

Figure6.

The pictures of the NiO_x film electrode during the cyclic voltammetry: (a) the bleached state at −0.8 V vs Ag, (b) the colored state at +1.2 V vs Ag.

Figure7.

The repeated chronoamperometric responses of the NiO_x film electrode at the step-potentials of +1.2 V and −0.8 V vs Ag for each 30 s. (solid line: current, dotted line: applied potential).

We have integrated this CA data to convert into the chronocoulometric (CC) data since the electrochromic effect depends on the amount of charge, not on the current. The data was plotted in Fig. 8. The apparent anodic charge increase (ΔQ_a, from valley to peak in Fig. 8(a)) was always slightly larger than the cathodic charge increase (ΔQ_c, from peak to valley, the actual data is the decrease of charge density in the figure) during the cycles and the ratio of ΔQ_c/ΔQ_a was kept nearly constant. The average charge densities were about 12 and 14 mC/cm² for bleaching at −0.8 V and coloring at +1.2 V vs Ag, respectively. The values were kept relatively constant for 50 cycles. The apparent charge reversibility (ΔQ_c/ΔQ_a) was about 86% that was much higher than the previously reported values (64% for sputtered NiO film in LiClO₄-PC solution^6b and 57.6% for nanorod NiO film in aqueous KOH solution²²).

Apart from the charge density change by the redox cycling, there is a monotonic charge increase in background as being seen easily in Fig. 8(a). This background anodic charge increase could be expected from the CA data as the anodic residual current at the end of the step potential was larger than the cathodic one as mentioned earlier. If the anodic charges are accumulated in the film, the optical transmittance should gradually decrease but no such a tendency was found as the data will be discussed hereafter. Therefore, this background charge is certainly not related to the redox process of NiO_x film. This background anodic process might be related to the catalytic oxidation of the solvent of PC by NiO_x, as PC cannot be oxidized at the ITO electrode by this anodic potential as shown in Fig. 5. To see the change in charge density by the redox process of NiO_x alone, this background charge increase in Fig. 8(a) was estimated by fitting to the anodic maximum values and the charge density change was corrected and shown in Fig. 8(b). After this correction, the anodic and cathodic charge were well balanced, and the charge reversibility was almost 100%. This explains that the CA response remained constant despite the accumulation of anodic charge. As far as we know, this fact has never been discussed in other relevant literature.

Figure8.

The repeated chronocoulometric responses of the NiO_x film electrode at the step-potentials of +1.2 V and −0.8 V vs Ag for each 30 s: (a) as is measured, (b) as corrected for the background charge.

The optical transmittance change of NiO_x film was measured simultaneously with the CA measurement. Fig. 9 shows the change in the transmission spectra of the NiO_x film electrode during the first two cycles in the CA measurement in Fig. 7. The transmittance at the bleached state was not so high in the low wavelength range compared to the spectrum of NiO film in an aqueous KOH solution.²³ NiO film can be converted completely to colorless Ni(OH)₂ in aqueous KOH solution, but the reduction of NiO_x in nonaqueous solution is limited by Li⁺-intercalation and Ni³⁺ in the lattice is difficult to be completely reduced. Therefore, some color remains as shown in Fig. 6(a).

Figure9.

The transmission spectrum changes of the NiO_x film electode according to the repeated CA measurements for the first two cycles.

The time dependence of the transmittance change at 550 nm was also plotted from the transmission spectrum change and shown in Fig. 10. The changes in transmittance during the bleaching and coloring cycles were quite symmetric as shown in Fig. 10. This matches well to the charge density change shown in Fig. 8(b). The transmittance difference (ΔT₅₅₀) between the colored and bleached states was about 44.5% at this potential difference (−0.8/+1.2 V vs Ag). If we define the response times of coloring and bleaching as a time that need to reach 70% of ΔT_c and ΔT_b, respectively, both the response times evaluated from Fig. 10 are about 5.5 s without an appreciable difference within a variation of 0.5 s. The colorization efficiency (CE) which is defined as the change in optical density per charge density change (ΔOD/ΔQ) is one of the important parameters to evaluate the electrochromic performance. The CE value at 550 nm was calculated to be 40.6 cm²/C with the step potentials of +1.2/−0.8 V vs Ag for 30 s. This value is comparable to the CE values obtained for various NiO films in aqueous KOH solution.²⁴ Due to the wide variety of measurement conditions, especially such as wavelength and potential difference, it is difficult to directly compare this value with other reported values.

Figure10.

The transmittance changes of the NiO_x film electrode at 550 nm and for whole visible light according to the CA measurement.

The integrated luminous transmittance (T_lum), referred to as the transmittance of visible light in the industrial standard, is defined and calculated according to the following equation:

T l u m = ∑ λ D λ V λ T λ ∑ λ D λ V λ for  380 ≤ λ ≤ 780  nm

where T_λ represents the measured spectral transmittance, D_λ and V_λ are the spectral energy distribution and the standard luminous efficiency for the photopic vision of human eye.²⁵ The change of T_lum of NiO_x film on PET/ITO for the CA cycling was also plotted in Fig. 10, together with T₅₅₀. The value of T_lum is very similar to T₅₅₀. The visible light modulation, ΔT_lum was 44.1% and the CE for visible light, not specific wavelength, was 38.2 cm²/C. The visible light modulation value of the electrochromic film is often presented in the literature, but the CE for “visible light” has not been presented in any other literature as far as we know. For the industrial standard of electrochromic film, this colorization efficiency for visible light might be necessary.

A new and facile method to fabricate electrochromic NiO_x thin films on any kinds of substrates was presented. This method utilized the coating sol prepared by dispersing nanocrystalline NiO_x into an alcoholic solvent with a silane binder. A robust NiO_x thin film on flexible PET/ITO substrate with relatively low haze was prepared by bar-coating of nanocrystalline NiO_x dispersion sol without high-temperature thermal treatment. The electrochromic performance of this NiO_x film on PET/ITO electrode in non-aqueous LiClO₄ electrolyte solution was comparable to that of other NiO_x films on ITO/glass in a KOH aqueous solution. The NiO_x film of a 400 nm thickness exhibits visible light modulation of ca. 40%, colorization efficiency of 40.6 cm²/C at 550 nm and a response time of 5 s at step potentials of −0.8/+1.2 V vs Ag without the activation cycle for Li⁺ ion intercalation-deintercalation. If we neglect the side reaction related to solvent, the almost 100% charge reversibility was obtained with this NiO_x thin film. Further investigations on the applied potential dependence and long-term stabilty of this NiO_x film are undergoing.