Improved performance of co-sputtered Ni–Ti oxide films for all-solid-state electrochromic devices

Abstract

Thin films of Ni–Ti oxide were co-sputtered by reactive direct current magnetron sputtering and their structures, morphologies, and compositions were investigated by X-ray diffraction, atomic force microscopy and X-ray photo-electron spectroscopy. Their electrochromic (EC) performances were studied using cyclic voltammetry, alternating current impedance and optical transmittance measurements. The proper addition of Ti helps the films achieve excellent EC behavior, including the stable coloration–bleaching cycles, high optical transmittance modulation (∼78%), great coloration efficiency (96 cm² C⁻¹) and low alternating current impedance. Finally, the multiple-layer stacks ITO/NiO_x:Ti/PVB(Li⁺)/WO₃/ITO were laminated based on the optimization of NiO_x:Ti single layers. A large optical contrast (∼60%), a fast response time (3.2 and 4.4 s) and a good durability were obtained for the all-solid-state full EC device.

---

1. Introduction

Electrochromic (EC) materials attract much interest and have a variety of applications such as displays, automotive rearview mirrors, gas sensors, and smart windows.^1,2 Nickel oxide (NiO_x) is a cheap and widely accessible EC material, which exhibits optical transmittance modulation in the process of electrochemical oxidation and reduction.³ NiO_x thin films show anodic electrochromism and are very suitable for device applications in conjunction with tungsten oxide (WO₃) thin films. However, for practical use, one widely debated basic question is the limited durability.^4,5 The poor durability is attributed to the self-discharge phenomenon linked to a partial dissolution of the oxidized phases when NiO cycling in basic electrolyte KOH. In order to increase the stability of the oxidized phases, chemical approaches based on the addition of other elements are considered. Many elements, such as Sn, Mg, Al, Si, W, Co, Ti, are reported to add in nickel-based oxide systems using chemical methods.^6–9 In the process the conditions must be well controlled and extra post treatments are necessary. The increased complexity and cost limit their mass production in industry.

Titanium dioxide (TiO₂) has become one of the most promising semiconductor due to its high chemical stability, excellent functionality, nontoxicity and low cost.¹⁰ Its addition to nickel-based oxide systems shows promising results improving EC efficiency and durability.¹¹ However, little work has been published on the EC performance of the composite Ni–Ti oxide films by employing reactive DC magnetron co-sputtering method.¹² The magnetron sputtering is considered as an efficient way for Ti-doped NiO_x due to controllable power ratios of Ti/Ni and simplified process. In our work, we investigated the effects of Ti/Ni power ratios on the structures, morphologies, compositions and electrochemical properties of Ni–Ti oxide films. This paper reports initial results from a comprehensive study on Ni–Ti oxide films and highlights the advantages of Ti addition to improve EC durability as well as coloration efficiency (CE). The work is an extension of our recent studies on electrochromism in NiO_x.^13,14 In addition, based on the improvement of the EC performances of the single layers, the all-solid-state EC device with the anodic ion-storage NiO_x layer complementary to the cathodic coloring WO₃ layer and the Li⁺ containing polyvinyl butyral (PVB) gel electrolyte is obtained and the transmittance modulation, cycle life, response time and ion diffusion coefficient are discussed. Physicochemical properties of the films were characterized by several measurements such as X-ray diffraction (XRD), atom force microscopy (AFM), UV-VIS spectrophotometer, X-ray photoelectron spectroscopy (XPS), cyclic voltammograms (CVs), alternating current (AC) impedance and chronoamperometry (CA).

2. Experimental

Ni–Ti oxide films were fabricated onto the commercial commonly used indium tin oxide (ITO) coated glass by reactive direct current magnetron sputtering from two metallic targets of Ni and Ti with 6 cm diameter in a versatile deposition system. Depositions of Ni–Ti oxide films took place in a gas mixture with an O₂/(O₂ + Ar) mass flow ratio being as 5/(5 + 200) and a working pressure of 2.5 Pa. The target–substrate separation was 22 cm at all times. In the sputtering process the holder rotated on its symmetrical axis at a constant speed to improve the uniformity of the films. Similarly, the WO₃ layer was deposited on the ITO/glass substrate by applying the magnetron sputtering method at room temperature. More details on deposition parameters can be observed in the values reported in Table 1.

Table 1 Detailed parameters of co-sputter deposited Ni–Ti oxide films and WO₃ films^a

Sample no.	Target	Pressure (Pa)	Ar (sccm) : O2 (sccm)	PowerNi (W)	PowerTi (W)	PowerW (W)	PowerTi/powerNi	Sputtering time (min)
a PowerTi/powerNi denotes the power ratio of Ti/Ni maintained in the co-sputtering process.
1	Ni	2.5	200 : 5	144	—	—	0	60
2	Ni/Ti	2.5	200 : 5	144	29	—	0.2	60
3	Ni/Ti	2.5	200 : 5	144	86	—	0.6	60
4	Ni/Ti	2.5	200 : 5	144	130	—	0.9	60
5	Ni/Ti	2.5	200 : 5	144	187	—	1.3	60
6	Ni/Ti	2.5	200 : 5	144	216	—	1.5	60
7	W	1.8	160 : 100	—	—	132	—	60

---

PVB based Li⁺ containing gel electrolyte was prepared and used as ions transport layer in the full EC device. The solution was obtained by dissolving 12.0 g of PVB in 130 ml ethanol under conditions of magnetic stirring and constant heating of 50 °C. Then, 1 M LiClO₄ propylene carbonate solution was added to the PVB solution. The mixtures were stirred until complete dissolution. The solutions were then poured into Petri dishes covered by PET release film and dried in an oven at 50 °C. After that, the electrolyte films were formed. The all-solid-state EC device with the configuration as glass/ITO/NiO_x:Ti/PVB(Li⁺)/WO₃/ITO/glass can be obtained by laminating the layers of glass/ITO/WO₃ (sample no. 7 from Table 1), PVB(Li⁺) gel electrolyte and glass/ITO/NiO_x:Ti (sample no. 4 from Table 1) together at the temperature of 100 °C in a 0.2 MPa high-pressure autoclave for exhausting extra air.

The crystalline structures of Ni–Ti oxide films were identified by grazing incidence XRD (GIXRD, D/Max 2200 PC, Rigaku, Japan). The surface morphology was characterized by AFM (Nanosurf Easyscan 2). XPS (ESCALAB 250Xi, Thermofisher, UK) was employed to determine the stoichiometry and chemical valence of elements in the films. The optical transmittance of the films was measured by Jasco V-570 UV/VIS/NIR spectrophotometer. CVs, AC impedance and CA measurements were carried out by using conventional three-electrode configuration on a Princeton VersaSTAT 4 electrochemical workstation.

3. Results and discussion

3.1 Effect of the amount of Ti on XRD studies

In Fig. 1, all the diffraction peaks are identified and no peaks of impurity are detected from the films, suggesting that a small amount of Ti addition does not change the cubic structure of NiO. The peaks of NiO can be indexed as (111), (200) and (220) crystal planes. However, when the power ratio of Ti/Ni increases to 1.5, the crystallinity of NiO films is weakened to a great extent. Additionally, no information about a Ni₂O₃ and TiO₂ crystalline phase is found in the XRD results, which indicates that if these compounds exist they are X-ray amorphous.

Fig. 1 XRD patterns of the Ni–Ti oxide films co-sputtered under different Ti/Ni power ratios.

According to Shannon table, we can know that the radius of the ions as listed in Table 2. We find that the radius of Ti⁴⁺ ions is slight larger than Ni³⁺ but smaller than Ni³⁺. Thus, it is not easy to tell the peak shifts towards higher or lower angles after doping. On the one hand, from calculations based on Debye–Scherrer formula and Bragg's law, we observe a slight decrease both in lattice constant and crystallite size. The data are shown in Fig. 3. On the other hand, from detailed analysis of the XRD data in Fig. 2, it is observed the main peak with lattice plane (200) shifts towards higher angles slightly, which is in good agreement with the slight decrease of lattice constant. Replacement of Ni with Ti gives rise to a decrease of the cell parameter. However, as presented in Table 3 in Section 3.3, the percentage of oxygen atom increases from 54% to 61% in the mixed films with the increasing amount of Ti addition. It is reckoned that many Ti⁴⁺ ions takes away the oxygen to form another amorphous phase TiO₂, which can be confirmed in analysis about Fig. 8 in Section 3.3. As a result, there are only a few Ti elements replacing Ni elements to bring about the doping effect.

Table 2 The radius value of different ions

Ion	Ti^4+	Ni^2+	Ni^3+
Radius (Å)	0.605	0.690	0.560

---

Fig. 2 Detailed analysis of the XRD data for the Ni–Ti oxide films co-sputtered under different Ti/Ni power ratios.

Table 3 Composition of Ni–Ti oxide films co-sputtered under different Ti/Ni power ratios

PowerTi/powerNi	0	0.2	0.6	0.9	1.3	1.5
Atomic Ni (%)	46.0	46.0	43.0	42.0	36.0	28.0
Atomic Ti (%)	0	0.4	1.2	2.0	6.0	11.0
Atomic O (%)	54.0	53.6	55.8	56.0	58.0	61.0

---

The detailed analysis of the XRD data indicates that the peak positions are different for films with different Ti/Ni power ratios. Specifically, the most prominent diffraction peaks corresponding to the (111) and (200) lattice planes are shifted towards somewhat higher angels with Ti/Ni power ratios increasing. As noted in our earlier work on NiO_x films,¹⁴ this effect is attributed to tensile and compressive stress for grains originating from doping effect.^15,16

To evaluate the influence of the Ti/Ni power ratios on the microcrystallities of the co-sputtered films, the crystallite size and lattice constant are calculated in Fig. 3. The main peak with lattice plane (200) of NiO is used as a reference. The microcrystalline nature of the NiO_x compounds is clearly observed from the calculated values. Fig. 3 also shows a unit cell of closed-packed (f.c.c.) crystalline NiO with a lattice constant ranging from 0.4167 to 0.4183 nm. From data in Table 2 directly from XPS measurements, it is found atomic Ti percentage increases from 6% to 11% with power ratio increasing from 1.3 to 1.5. So it is easy to understand that lattice shrinkage, if present, is very significant for the last point, for the 1.5 Ni/Ti power ratio. It is observed in XRD figures that Ni–Ti oxide films with the highest power ratio 1.5 become more amorphous related to lattice disorder induced by excess Ti addition. Considering the partial amorphization, we have deleted the cell parameters of the highest power ratio because it is not meaningful to talk about lattice constant here. With error bars added in Fig. 3, there is a slight decrease in cell parameters. It is reckoned that the Ti doping can cause relatively poor crystallities and shrinkage in lattice. H. Moulki et al.¹⁷ have reported on the NiO based thin films by lithium addition and found lithium addition remarkably deteriorates the preferred orientation and creates disorder leading to partial amorphization. Also, their paper gives information that replacement of Ni with Li gives rise to a decrease of the lattice constant correlated to a shrinkage of the unit cell volume. It is the similar case with our work on NiO:Ti films. The replacement of Ni with Ti also results in the decrease of cell parameter. Additionally, NiO:Ti films become more amorphous when the power ratio of Ti/Ni increases to 1.5, as shown in Fig. 1 in manuscript.

Fig. 3 Crystallite size and lattice constant of the Ni–Ti oxide thin films determined by Debye–Scherrer formula and Bragg's law.

3.2 Effect of the Ti amount on morphology

Fig. 4 illustrates characteristic topographies for Ni–Ti oxide films with various Ti/Ni power ratios. The pure NiO films surface is rough and compact. After Ti addition, film surface particles become more smaller and uniform, as shown in Fig. 4(d). However, when excess Ti elements are added, film surface becomes very flat. The roughness is very small. So, it is more difficult for ions to interact with films.¹⁸ The reasons and mechanisms leading to such change trends are still unknown. It is probably associated with defects in the host lattice after addition, as discussed by Chin-Yi Chen¹⁹ in the paper about grain size change. The surface roughness of films was measured by using a Veeco Dimension 3100 instrument employed in tapping mode. Both the mean roughness (R_a) and the root mean square roughness (R_q) first increase and then decrease as the Ti/Ni power ratios increase and the peak value appears when the Ti/Ni power ratio equals 0.2, as apparent in Fig. 5. It indicates that a small amount of Ti addition will make the film surface rough and compact. Surface morphologies of NiO_x films can affect the EC properties since electrochemical reactions first occur on the film surface.¹⁸ The outermost layer of the film in direct contact with the electrolyte could influence the EC response. The surface of the films with Ti/Ni power ratio of 0.9 consists of small and uniform nano-particles, showing 0.9 as optimum value of the power ratio. When the power ratio is smaller than 0.9 or exceeds 0.9, the surface demonstrates even bigger particles or much flatter morphology, thus causing a relatively smaller specific surface area and limiting the chances of reaction between the films and the electrolyte.

Fig. 4 The three-dimensional AFM micrographs for the Ni–Ti oxide films co-sputtered under different Ti/Ni power ratios as (a) 0; (b) 0.2; (c) 0.6; (d) 0.9; (e) 1.3 and (f) 1.5.

Fig. 5 The average roughness of NiO_x films as a function of Ti/Ni power ratios deduced from AFM research.

3.3 Effect of the amount of Ti on components and valence states

Fig. 6 illustrates the typical XPS survey spectra of the Ni–Ti oxide films with the Ti/Ni power ratio of 0.9. It can be observed that the surface of the sample is composed of nickel, titanium, oxygen, and carbon contaminants. Table 2 shows the corresponding atomic proportion of Ni/Ti/O in co-sputtered films deposited with various Ti/Ni power ratios. A steady increase in the atomic proportion of Ti is found with the increase of Ti/Ni power ratios.

Fig. 6 Typical XPS survey spectrum of Ni–Ti oxide thin films.

It can be shown in Fig. 7 that there are noticeable chemical peak shifts towards the same direction for Ni 2p with the increase of Ti/Ni power ratios. With the Ti doping, some oxygen nearby Ni elements will be dragged away from the lattice, thus loosen and weaken the chemical bondings between Ni and O formed by the outermost electrons becoming loose and weak. For the Ni elements, the decrease of the outermost electron cloud density will strengthen the hold of inner electrons on their nucleus. As a result, more Ti elements doped, higher binding energy for Ni 2p is detected.

Fig. 7 Displaced peak positions of XPS spectra for Ni–Ti oxide thin films.

To determine the chemical valence state of Ti element in the films, the Ti 2p high resolution XPS spectrum for Ni–Ti oxide films are analyzed in Fig. 8. Shirley instead of linear baseline is substracted and the fitting to 2p_3/2 matches well after adding three peaks. % Lorentzian–Gaussian is taken into consideration and detailed peak parameters are listed in Table 4.^20,21 The peak feature in the Ti 2p XPS spectrum verifies that the element of Ti exists in the form of pure stoichiometric TiO₂. The curve fitting results demonstrate a perfect fit to the double peaks located at binding energies of 457.951 eV and 463.844 eV, respectively, which is in accordance with and core levels of Ti⁴⁺ cations. The energy separation between Ti⁴⁺ 2p_3/2 and Ti⁴⁺ 2p_1/2 peaks is 5.8 eV, reflecting a strong bonding between the Ti and O atoms. The area ratio of Ti 2p_3/2 and Ti 2p_1/2 is 3.704 and their full widths at half-maximum are 1.464 and 2.144 eV, respectively, indicative of the high resolution of the Ti 2p XPS spectrum.^22,23

Fig. 8 XPS Ti 2p high resolution spectra for Ni–Ti oxide films deposited with Ti/Ni power ratio being as 0.9.

Table 4 Peak separation parameters for Ti 2p

Peak position (eV)	Area	FWHM (eV)	% GL	Corresponding chemical compounds
457.951	31 518.520	1.464	20	TiO2
459.032	14 490.860	2.047	20	NiTi
463.844	12 423.140	2.144	20	TiO2

---

In Fig. 9 the O 1s spectra are deconvoluted into different components in order to evaluate the proportion of Ni²⁺/Ni³⁺/Ti⁴⁺ in the sputtered films. For the films without Ti addition the O 1s states in the Ni₂O₃ phase and the NiO phase exhibit peaks located at 531.1 eV and 529.3 eV, respectively. However, the corresponding energies for films with the Ti/Ni power ratio of 1.3 are somewhat higher. It is ascribed to the Ti addition, which will result in the decrease of the outermost electron cloud density of the Ni element and the increase of the binding energy between inner electrons and nucleus, which is consistent with the analysis of Fig. 8. Moreover, the proportions of Ni²⁺/Ni³⁺/Ti⁴⁺ in pure NiO_x films and Ni–Ti oxide films are estimated to be 66.41%/33.59%/0% and 60.45%/24.53%/15.02%, respectively. Compared with the O/Ni atomic ratios presented in Table 3, it is found that all the Ni atoms in films are somewhat more than the Ni atoms combined with oxygen. It can be reckoned that incompletely oxidized metallic nickel exists in Ni–Ti oxide films when the Ti/Ni power ratio equals 1.3.

Fig. 9 XPS O 1s high resolution spectra for (a) pure NiO_x films and (b) Ni–Ti oxide films deposited with Ti/Ni power ratio being as 1.3.

3.4 Effect of the amount of Ti on spectroscopic studies

The long-time transmittance stability of the films was checked by cycling up to 6000 s. These transmittance data are from potential sweeps rather than potential steps. Here there are 100 cycles with one minute for per cycle (30 s for coloring and 30 s for bleaching). The sweep potential ranges from −1.5 to +1.5 V and the sweep rate is 100 mV s⁻¹. Due to the limitations of the spectrophotometer in our laboratory, we only exhibit the transmittance evolution in 6000 s. Even though 6000 s is less than 2 hours, it can be observed rapid degradation of the optical contrast takes place for the pure NiO_x films. For Ni–Ti oxide films with the power ratio of 0.6, it does not show any decline. Comparing the pure NiO_x films and Ni–Ti oxide films directly, we can find the Ti addition makes sense without any doubt and it is beneficial to the improvement of durability. Additionally, Matsuoka and Hashimoto^24–26 report on the effect of the incorporation of numerous metallic additives into evaporated electrochromic films and find that Ti is beneficial and able to extend the cycling durability in electrolytes. As demonstrated in Fig. 10, Ti addition turns out to improve the EC performances to a great extent, especially the cycle stability and the transmittance contrast. The cycle life of the films shows a significant dependence on Ti addition. The stability of the films deposited in the absence of Ti is not high and the films degrade after 1000 s when cycled in 1 M KOH electrolyte. After 1600 s almost all of these pure films are dissolved in the liquid electrolyte. A small power ratio of Ti/Ni (0.2) is associated with a slightly improved durability. When the power ratios increase to 0.6 and 0.9, the films exhibit a high optical contrast and a good durability up to 6000 s without any degradation. Films with a larger power ratio of Ti/Ni (1.3 and 1.5) show a decrease in the cycle life due to the excess of Ti addition. For films with proper addition of Ti, the multiple coloration and bleaching cycles occur without any degradation, which makes them practical for use in full EC devices. Specifically speaking, there is an activation period in the initial cycles during which the coulometric cathodic capacity increases for these EC films. Electrochemical quartz microbalance experiments essentially confirm that the coloration/bleaching process in this period arises from the insertion/deinsertion of OH⁻.^27,28

Fig. 10 Evolution of the transmittance at the wavelength 550 nm for Ni–Ti oxide films with diverse Ti/Ni power ratios immersed in 1 M KOH, with an exposed area of 2.5 × 2 cm². Data were taken for the voltage sweep range −1.5 to 1.5 V.

Expanded version of a few of the cycles is shown in Fig. 11. For Fig. 10 and 14 the transmittance and CV data are accumulated from synchronization testing. The expanded version of a few cycles for power ratio as 0.9 is shown as follows. Response time for bleaching and coloring process in the 26^th and 82^nd cycles are presented in Fig. 11 and corresponding potential for synchronous CV measurements is listed in Table 5. It is found that in the 26^th the coloring and bleaching start at potential +1.5 and −0.7 V, respectively. In the 82^nd and 83^rd cycles, the bleaching and coloring start at potential −0.9 and +1.5 V, respectively. The potential does not show significant changes after multiple cycles. In initial cycles ΔT is more stable for power = 0.6 than 0.9, but in later scans there is not much difference. Before the chemical reactions in coloring and bleaching process are kept in balance, there is an activation period as described in our former work.¹⁴ It is more interesting that the calculated CE for power 0.9 is much higher than 0.6, as presented in Fig. 16. The improved CE arises from TiO₂,²⁹ but the exact associations are still unknown.

Fig. 11 Response characteristics of different cycles for Ni–Ti oxide films with power ratio as 0.9.

Table 5 Detailed parameters for films with power ratio 0.9 in bleaching/coloring process

Cycle number	26	26	27	82	83	83	83
Bleaching/coloring time (s)	1530	1552	1580	4914	4938	4949	4974
Transmittance (%)	91.7	32.3	92.4	24.8	93.8	93.1	24.8
Electrochromic process	Coloring		Bleaching	Bleaching		Coloring
Synchronous potential (V)	+1.5	−0.7	+0.5	−0.9	+0.4	+1.5	−1.0

---

The optical transmittance spectra in Fig. 12 show that the colored and bleached states of the Ni–Ti oxide films range from 300 to 760 nm. Higher voltages of −2 V and +2 V are applied to expand the optical contrast op to 78% so that the difference between the optical gaps of bleached and colored states becomes more obvious. The optical transmittance of the bleached and colored states at the wavelength of 550 nm are 97% and 19%, respectively, so the optical transmittance modulation and the optical density are 78% and 0.71, respectively. Additionally, the optical banding gaps (E_g) of the bleached and colored states are calculated on the basis of the optical transmittance spectra by the following equation:^30,31

[hν ln(1/T)]ⁿ = A(hν − E_g)

where hν is the photo energy; T is the transmittance; A is the constant related to the transition probability of the material; n refers to either 1/2 or 2 for indirect transition and direct transition, respectively. Accordingly, the optical band gap of the films at bleached/colored states can be obtained by extrapolating the linear portion of [hν ln(1/T)]ⁿ – hν curve to zero, which is presented in Fig. 13. The E_g of the original films is 3.52 eV, whereas in the EC process the E_g are found to be 3.53 and 3.28 eV for the bleached and colored states, respectively. This means that blue shift is shown by the bleached films and the red shift by the colored films in the energy band gap. This is associated with the micro mechanisms in the EC process. For colored films, this would be mainly due to a large number of charges are injected into the microstructures leading to the decrease of the E_g.³² Therefore, it is reckoned that if the transmittance of the bleached film decreases, its energy band gap will also tend to decrease. Furthermore, no linear relation is found for n = 1/2, indicating that the Ni–Ti oxide films are semiconductors with direct transition at this energy level.

Fig. 12 Transmittance of the bleached and colored states for the Ni–Ti oxide films deposited with Ti/Ni power ratio as 0.9 corresponding to the 101^th cycle.

Fig. 13 Diagram of geometry solving for optical band gap of the Ni–Ti oxide films at bleached and colored states based on Tauc formula.

3.5 Effect of the Ti amount on electrochemical properties

CV recording was performed to observe the electrochemical fingerprints of the EC films in the coloration/bleaching process. The EC behavior of the Ni–Ti oxide films is explored using CV measurements for intercalation and deintercalation of the OH⁻ ions. The dependence of the CV response on the Ti/Ni power ratios is depicted in Fig. 14. Films prepared without Ti addition cycled in 1 M KOH electrolyte display weak adhesion on ITO glass substrates. They present very small electrochemical activities after 50 cycles as shown in Fig. 14(a), because serious corrosions occur within the films. However, the addition of Ti to the oxide film improves the adhesion of the film to the substrate and reinforces its abrasion resistance.¹¹ Therefore, the CV results verify that Ni–Ti oxide films with the Ti/Ni power ratio of 0.9 has more stable and persistent EC effects than the films with the power ration of 0.6. Films with the power ratio of 0.2 after 100 cycles show abnormal electrochemical behavior. It is mainly because the ITO is involved in the electrochemical reaction.³³ The CVs results combined with the evolution of the transmittance presented in Fig. 10 shows Ni–Ti oxide films with the Ti/Ni power ratio of 0.9 have better cycle stability and enlarged optical contrast. The improved cycle life of the films with Ti addition is ascribed to the high chemical stability in the EC process.

Fig. 14 CVs of Ni–Ti oxide films deposited with Ti/Ni power ratio being as (a) 0, (b) 0.2, (c) 0.6, (d) 0.9, (e) 1.3 and (f) 1.5, for applied voltage of −1.5 and 1.5 V (versus Ag/AgCl), after 30/50/80/100 cycles, with an exposed area of 2.5 × 2 cm².

For Fig. 10 and 14, the transmittance and CV data are accumulated from synchronization testing. The expanded version of a few cycles for power ratio as 0.9 is shown as follows. Response time for bleaching and coloring process in the 26^th and 82^nd cycles are presented in Fig. 11 and corresponding potential for synchronous CV measurements is listed in Table 3. It is found that in the 26^th the coloring and bleaching start at potential +1.5 and −0.7 V, respectively. In the 82^nd and 83^rd cycles, the bleaching and coloring start at potential −0.9 and +1.5 V, respectively. The potential does not show significant changes after multiple cycles.

The evolution of the inserted and extracted charge density Q during multiple voltammetric cycles is shown in Fig. 15. It is apparent that in general the Ni–Ti oxide film with the Ti/Ni power ratio being as 0.6 displays a decline in its charge capacity with cycle number accumulated. It is also interesting to note that there is an opposite tendency for the Ni–Ti oxide film with the Ti/Ni power ratio being as 0.9. The charge density increases with cycle number to some extent, which is linked with more active electrochemical behaviors within the films. Furthermore, the inserted and extracted charge densities are different for both samples in the process of voltammetric cycles. The disequilibrium phenomenon, associated with the irreversibility of the film, is possibly a result of irreversible ions incorporation.^34,35

Fig. 15 Inserted and extracted charge density Q during voltammetric cycling for pure NiO_x films and Ni–Ti oxide films. Symbols denoting data were joined by straight lines.

CE of the device is expressed as the optical density change per insertion charge unit. The change of the CE at 550 nm is consequently different as illustrated in Fig. 16(a). Data for different values of efficiency indicate that films with the Ti/Ni power ratio of 0.9 give a much higher magnitude of the CE than that with the power ratio of 0.6. The high value of CE is estimated around 90 cm² C⁻¹, which is much better than the reported values.^36,37 Hence there is a correlation between CE and the amount of Ti addition. Specifically, for films with the power ratio of 0.9 the CE show a slight decrease with cycle numbers while there is an increase trend for films with the power ratio of 0.6. The bleaching efficiency is larger than the related coloring efficiency and the calculated corresponding data are shown in Fig. 16(b). As demonstrated in Fig. 11, in coloring process the transmittance decreases gradually but does not reach a flat level, while in bleaching process the transmittance increases sharply and then becomes stable. It indicates the coloring is not a symmetry process to bleaching and it is more difficult for extraction of ions. If the equilibrium exists between insertion and extraction, the films can exhibit even better optical contrast.

Fig. 16 Evolution of (a) coloration and (b) bleaching efficiency at the wavelength 550 nm for Ni–Ti oxide films immersed in 1 M KOH. Symbols denoting data were joined by straight lines.

Fig. 17 shows the impedance spectra of EC glass with NiO_x electrode and Ni–Ti oxide electrode. The corresponding equivalent circuit for impedance analysis is displayed in the inset of Fig. 17. In the fitting circuit R_s, R_ct, C_dl, W and C_p is solution resistance, charge transfer resistance, double layer capacitance, Warburg impedance and pseudo capacitance, respectively. The Bode plot curves show that pure NiO_x films from EC glasses have impedance values between 10^1.4 and 10^1.6 ohms in high-frequency ranging from 10¹ to 10⁵ Hz. However, for co-sputtered Ni–Ti oxide films, there is a significant decrease in measured impedance values with the addition of the Ti elements. They have impedance values between 10^1.2 and 10^1.4 ohms in the same high-frequency range, which is lower than that of pure NiO_x films. It is reckoned that the Ni–Ti oxide film exhibits promising electrode properties due to its lower AC impedance value than films without Ti addition. Additionally, because the working and counter electrodes consisting of ITO, Pt, Ag/AgCl and NiO_x semiconductor films have much lower impedance than that of the electrolyte composing of 1 M KOH, the electrodes and electrolyte characterizations are represented by the impedance in high- and low-frequency.³⁸ In Fig. 18, the Nyquist plots, where bigger radius of the fitting circle refers to bigger impedance values, also show similar impedance decreasing tendency with Ti addition.^39,40 The analysis is therefore fully compatible with the results displayed in the Bode plot.

Fig. 17 Bode plot of AC impedance curves of pure NiO_x electrode and Ni–Ti oxide electrode from EC glass in 1 M KOH. Inset shows the equivalent circuit used to fit the experimental impedance spectrum.

Fig. 18 AC impedance curves of pure NiO_x electrode and Ni–Ti oxide electrode from EC glass in 1 M KOH: (a) Nyquist plot and (b) Nyquist plot in a high-frequency range.

Fig. 19 shows that the AC impedance of the colored films is lower than that of the bleached films. From bleached to colored state, a transfer process related to ions and charges moving between the electrolyte and electrode interface will occur in the mixed ionic and electronic conducting electrode/electrolyte system. More ions are gathered into the Ni–Ti oxide film electrode during the coloring process, reducing the charge difference between the electrolyte and the electrode. Hence, the difficulty of the charge transfer between the electrolyte and electrode interface is decreased and the AC impedance of the film becomes lower for colored films. The Nyquist plot in Fig. 20 also shows that the films have lower impedance in a colored state than in a bleached state.

Fig. 19 Bode plot of alternating current impedance curves of Ni–Ti oxide electrode from EC glass with the Ti/Ni power ratio being as 0.9 under bleached and colored states.

Fig. 20 Alternating current impedance curves of Ni–Ti oxide electrode from EC glass with the Ti/Ni power ratio being as 0.9: (a) Nyquist plot and (b) Nyquist plot in a high-frequency range.

3.6 Studies of Ni–Ti oxide films based devices

In light of the improvements on the durability, optical contrast and EC efficiency deduced from CVs and long-time transmittance testing, the Ni–Ti oxide film is employed to form EC device with WO₃ film as the complementary EC layer and PVB(Li⁺) as the gel electrolyte layer. For the all-solid-state EC device, the co-sputtered Ni–Ti oxide films serve as anodically coloring material, while as cathodic material WO₃ films deposited via sputtering technique are used. To visualize the configuration of the device, the cross-section is demonstrated in Fig. 21. In order to evaluate the transmittance variation versus time, the durability of the device is tested by performing life-time cycles measurements as shown in Fig. 22(a) and corresponding individual cycles are exhibited in Fig. 22(b). The optical modulation at a mid-luminous wavelength of 550 nm reaches 75%, with a good durability up to 6000 s without any loss of transmittance contrast. In early cycles, the modulation is enhanced with time. The activation period appears because the films can evolve during EC cyclings.¹⁴

Fig. 21 A cross-sectional view of the glass/ITO/NiO_x:Ti/PVB(Li⁺)/WO₃/ITO EC device.

Fig. 22 (a) Life cycle transmittance and (b) the partial enlargement of the glass/ITO/NiO_x:Ti/PVB(Li⁺)/WO₃/ITO EC device.

Fig. 23 displays CVs of the full glass/ITO/NiO_x:Ti/PVB(Li⁺)/WO₃/ITO EC device for 100 cycles. Broad oxidation and reduction features appear and it is evident that some evolutions take place during the initial cyclings, but the properties remain stable after a few cycles. With ±2 V voltage applied, the cathodic peak occurs at +1.25 V and keeps constant in the cycles. The value of the anodic peak is +0.75 V and it keeps relatively constant with a minor shift of 0.2 V. The extra small peak does not originate from NiO or WO₃ redox reactions. It is due probably to the ITO reduction process.^41,42 However, this subject is controversial as discussed previously by R. Schiller in details.⁴³

Fig. 23 Multiple curves of CVs of the glass/ITO/NiO_x/PVB(Li⁺)/WO₃/ITO EC device with an active area of 2.5 × 2 cm², for applied voltage of ±2 V.

Fig. 24 displays the switching response characteristics of the full EC device. The switching responses, including the bleaching time (t_b) and coloration time (t_c), are very important aspects in the performances of EC devices. The voltage is stepped from its rest potential of −1.5 to 0 V for 30 s and then reverses to +1.5 V for the next 30 s during the CA cycles. Here it may be necessary to mention that t_b or t_c, is defined as the time required to reach 90% of its full modulation. The extracted bleaching time and coloration time are 3.2 and 4.4 s, respectively. The short switching response time stands for excellent performance of the EC device. To be more specific, the slope of the cathodic current density is higher than that of the anodic current density, signifying that the bleaching process is faster than the coloration process. It is reckoned that the insertion of Li⁺ ions in the bleaching process is easier than the exaction of Li⁺ ions in the coloration process. The fast insertion process is related to the high electronic conductivity of the oxide film in the charged state.⁴⁴

Fig. 24 Time response of the current density of the full EC device.

There is little change for the current density and slope value in the multiple CA curves, as represented in Fig. 25, indicating the stability of the full EC device. The relationship between current density and time is expressed by the following Cottrell equation. The diffusion coefficient within the device is evaluated from the slope of the curve as depicted in inset of Fig. 25.⁴⁵

i = nFACD^1/2π^−1/2t^−1/2

where i, F, A, D, C, t denote current, Faraday's constant, area of the electrode, diffusion coefficient, concentration of analyte and time, respectively. The D is evaluated to be 1.05 × 10⁻¹⁰ cm² s⁻¹, which is in accordance with a fast switching process.

Fig. 25 Time response of the current density of the full EC device taken for 12 steps. Inset shows the plot of current vs. t^−1/2.

4. Conclusion

In order to improve the cycle stability and EC efficiency of sputtered NiO_x thin films, the effect of titanium addition is investigated. Co-sputtered Ni–Ti oxide films are prepared in optimized conditions for all-solid-state EC devices, namely power ratio of Ti/Ni = 0.9 and room temperature. A decrease in crystallite size, lattice constant and surface particle size is observed with the increase of Ti content. When cycled in KOH liquid electrolyte, Ni–Ti oxide films prepared with the power ratios of Ti/Ni being as 0.9 exhibit far better EC properties including improved optical contrast, better cycle stability and higher efficiency than pure NiO_x films. With the conditions optimized for Ni–Ti oxide single layers, the glass/ITO/NiO_x:Ti/PVB(Li⁺)/WO₃/ITO/glass laminated all solid state devices are successfully fabricated. The device exhibits a high optical modulation up to 58% at 550 nm and a good durability up to 6000 s without any decline of optical properties.

Acknowledgements

This work was financially supported by the National Program on Key Research Project (2016YFB0303901) and the Beijing Natural Science Foundation (2161001) and the Fundamental Research Funds for the Central Universities (Grant No. YWF-16-JCTD-B-03).

References

1. C. G. Granqvist, Handbook of Inorganic Electrochromic Materials, Elsevier, Amsterdam, 1995.
2. C. G. Granqvist, E. Avendano and A. Azens, Thin Solid Films, 2003, 442, 210.
3. G. Bodurov, P. Stefchev and T. Ivanova, Mater. Lett., 2014, 117, 270–272.
4. Y. Ushio, A. Ishikawa and T. Niwa, Thin Solid Films, 1996, 280, 233.
5. A. Fukazawa and C. Minami, EP1394598, 2004.
6. N. Penin, A. Rougier, L. Laffont, P. Poizot and J. M. Tarascon, Sol. Energy Mater. Sol. Cells, 2006, 90, 422–433.
7. Y. Makimura, A. Rougier and J. M. Tarascon, Appl. Surf. Sci., 2006, 252, 4593–4598.
8. A. Al-Kahlout, A. Pawlicka and M. Aegerter, Sol. Energy Mater. Sol. Cells, 2006, 90, 3583–3601.
9. K. Zhang, X. Q. Zhang, C. X. Zhang, S. J. Zhang, X. C. Wang, D. L. Sun and M. A. Aegerter, Sol. Energy Mater. Sol. Cells, 2013, 114, 192–198.
10. A. Fujishima and K. Honda, Nature, 1972, 238, 5358.
11. M. Martini, G. E. S. Brito, M. C. A. Fantini, A. F. Craievich and A. Gorenstein, Electrochim. Acta, 2001, 46, 2275–2279.
12. C. T. Wang, W. K. Zhang, H. Huang and X. Y. Tao, J. Inorg. Organomet. Polym., 2011, 21, 852–857.
13. D. M. Dong, W. W. Wang, G. B. Dong, Y. L. Zhou, Z. H. Wu, M. Wang, F. M. Liu and X. G. Diao, Appl. Surf. Sci., 2015, 357, 799–805.
14. D. M. Dong, W. W. Wang, G. B. Dong, Y. L. Zhou, Z. H. Wu, M. Wang, F. M. Liu and X. G. Diao, Appl. Surf. Sci., 2016, 383, 49–56.
15. R. C. Tenent, D. T. Gillaspie, A. Miedaner, P. A. Parilla, C. J. Curtis and A. C. Dillon, J. Electrochem. Soc., 2010, 157, 318–322.
16. T. Kubo, Y. Nishikitani, Y. Sawai, H. Iwanaga, Y. Sato and Y. Shigesato, J. Electrochem. Soc., 2009, 156, 629.
17. H. Moulki, D. H. Park, B. K. Min, H. Kwon, S. J. Hwang and A. Rougier, Electrochim. Acta, 2012, 74, 46–52.
18. S. R. Jiang, P. X. Yan and B. X. Feng, Mater. Chem. Phys., 2002, 77, 384–389.
19. D. Xu, K. Li, Y. Zhou, Y. Gao, D. Yan and S. Xu, J. Eur. Ceram. Soc., 2017, 37, 419–425.
20. A. P. Dementjev, O. P. Ivanova, L. A. Vasilyev, A. V. Naumkin, D. M. Nemirovsky and D. Y. Shalaev, J. Vac. Sci. Technol., A, 1994, 12, 423–427.
21. C. M. Chan, S. Trigwell and T. Duerig, Surf. Interface Anal., 1990, 15, 349–354.
22. Y. Wang, H. J. Sun, S. J. Tan, H. Feng, Z. W. Cheng, J. Zhao, A. D. Zhao, B. Wang, Y. Luo, J. L. Yang and J. G. Hou, Nat. Commun., 2013, 4, 375–381.
23. H. Zhou and Y. Zhang, J. Phys. Chem. C, 2014, 118, 5626–5636.
24. H. Matsuoka, S. Hashimoto and H. Kageshika, J. Surf. Finish. Soc. Jpn., 1991, 42, 104–110.
25. S. Hashimoto and H. Matsuoka, J. Electrochem. Soc., 1991, 138, 2403–2408.
26. S. Hashimoto and H. Matsuoka, Surf. Interface Anal., 1992, 19, 464–468.
27. A. Al-Kahlout, S. Heusing and M. A. Aegerter, Electrochromism of NiO–TiO₂ sol gel layers, J. Sol-Gel Sci. Technol., 2006, 39, 195–206.
28. A. Al-Kahlout and M. A. Aegerter, Sol. Energy Mater. Sol. Cells, 2007, 91, 213–223.
29. A. Kongkanand, K. Tvrdy, K. Takechi, M. Kuno and P. V. Kamat, J. Am. Chem. Soc., 2008, 130, 4007–4015.
30. S. M. Gupta and M. Tripathi, Cent. Eur. J. Chem., 2011, 10, 279–294.
31. A. G. Al-Sehemi, A. S. Al-Shihri, A. Kalam, G. Du and T. Ahmad, J. Mol. Struct., 2014, 1058, 56–61.
32. L. Kumari, W. Z. Li, C. H. Vannoy, R. M. Leblanc and D. Z. Wang, Cryst. Res. Technol., 2009, 44, 495.
33. M. Lao, P. Li, P. Wang, X. Zheng, W. Wu and M. Shui, Electrochim. Acta, 2015, 176, 694–704.
34. M. S. Burdis and J. R. Siddle, Thin Solid Films, 1994, 237, 320–325.
35. M. A. Arvizu, C. A. Triana, B. I. Stefanov, C. G. Granqvist and G. A. Niklasson, Sol. Energy Mater. Sol. Cells, 2014, 125, 184–189.
36. G. Cai, J. Tu, D. Zhou, L. Li, J. Zhang, X. Wang and C. Gu, J. Phys. Chem. C, 2014, 118, 6690–6696.
37. R. T. Wen, G. A. Niklasson and C. G. Granqvist, Thin Solid Films, 2014, 565, 128–135.
38. C. C. Chien, J. Nanomater., 2013, 1, 749–756.
39. S. Mirra, A. K. Shukla and S. Sampath, J. Power Sources, 2001, 101, 213–218.
40. Y. W. Li, J. W. Wang and Y. D. He, Electron. Compon. Mater., 2008, 27, 26–28.
41. P. M. S. Monk and C. M. Man, J. Mater. Sci.: Mater. Electron., 1999, 10, 101.
42. M. J. Neto, R. Leones, F. Sentanin, J. M. S. S. Esperanca, R. M. Medeiros, A. Pawlicka and M. M. Silva, J. Electroanal. Chem., 2014, 63, 714–715.
43. R. Schiller, G. Battistig and J. Rabani, Radiat. Phys. Chem., 2005, 72, 217.
44. O. M. Hussain, A. S. Swapnasmitha, J. John and R. Pinto, Appl. Phys. A: Mater. Sci. Process., 2005, 81, 1291–1297.
45. P. M. S. Monk, Fundamental of Electroanalytical Chemistry, John Wiley and Sons Ltd, England, 2001.

---