Facile low-temperature synthesis of nickel oxide by an internal combustion reaction for applications in electrochromic devices

Electrochromic nickel oxide ion storage anodes compatible to tungsten oxide as cathode layer are prepared by a combustion reaction with urea as fuel and nickel nitrate as oxidizer at temperatures not higher than 230 C to stay well within the temperature range in which thermally sensitive intercalated tungsten oxide layers are still stable. The precursors for nickel oxide are well available materials and the preparation of the layers can be performed at low energy input by spin-coating onto the substrate and moderate heating on a hot-plate. Ethanol and acetone where used as solvents and reveal large differences in the resulting film morphologies, electric and electrochromic characteristics. Cyclic voltammetry measurements in contact to lithium perchlorate (LiClO4) in propylene carbonate show remarkable colouration efficiencies at 550 nm of 47 cm C for NiO prepared from ethanol solution and 92 cm C for NiO prepared from acetone solution, which are, to the best of our knowledge, among the highest reported values for Li-intercalation into NiO prepared at moderate substrate temperature reported so far.


Introduction
Energy saving is one of the relevant strategies to ensure sustainable modern societies [1]. Housing as one important sector offers great opportunities [2,3]. Combinations of low-energy houses with green power generation units are already able to produce more power than they use (energy-plus houses) [4]. One of the key factors of this concept is the proper management of incoming and emitted heat [5,6]. Aside from appropriate thermal insulation of walls, roofs and windows, the control of incoming solar radiation is important [7]. While in winter, incoming heat is mainly beneficial, in summer the benefits of room lighting have to be balanced against overheating of rooms which in many cases might then require energy-consuming airconditioning by refrigeration [4]. Smart windows represent an efficient way to control the incoming heat by solar radiation [8]. Electrochromic devices, in which the transmission of light can be tuned electrically can form a relevant part of smart windows. They can easily be integrated into central building control systems to optimize the energy management. Commercial systems are available that can control the transmission of visible light between 9 and 51% with a solar heat gain coefficient between 8 and 36% [9]. Despite the fact that electrochromic windows are still quite expensive, they can be found in commercial and public buildings as well as in the form of rear-view mirrors in automobiles, sunglasses or windows in the Boeing 787 Dreamliner [10][11][12][13][14][15]. Typical electrochromic devices consist of a tungsten oxide (WO 3 ) colouring cathode, and an anodic counterpart of vanadium titanium oxide (almost not colouring) or nickel oxide (colouring) with Li ? as intercalating ions and an ion exchange layer in between, typically an organic liquid or polymer gel electrolyte [16][17][18][19]. Nickel oxide (NiO) has a great potential to improve the overall characteristics of the devices due the brown colouration upon oxidation and excalation of Li ? , which perfectly fits to the blue colouration of reduced WO 3 upon intercalation of Li ? to get a gradual and deeper darkening in a stack with minor changes in colour [16,20,21]. NiO is a well-known material to serve as counterpart for tungsten oxide and, because of the large number of possible manufacturing processes and precursor materials for its preparation, NiO offers great potential to meet the necessary technical requirements. As an inorganic material, e.g. used in combination with a transparent inorganic solid electrolyte such as lithium phosphorous oxynitride (LiPON) in an all-solid-state device, it preserves the high mechanical and thermal stability which can be of particular advantage when used with large-area electrochromic windows [22,23] When prepared to an equal thickness, e.g., it can store an identical amount of Li ? as WO 3 [16,[24][25][26][27][28]. To improve the Li ? -capacity and/or the diffusion coefficient of Li ? in NiO mixed oxides with rare earth metals can be used at, however, reduced electrochromic effect [29,30]. On the way to an all-solid-state device, a solid Li ? -conductor should replace the liquid or polymeric component [31]. To prepare such stacks it is mandatory that the preparation methods of the individual layers have to be compatible with already deposited materials. Therefore, the preparation for NiO has to fit, e.g., to the wellestablished WO 3 and a transparent solid electrolyte like LiPON. Tungsten oxide loses its excellent electrochromic characteristics at about 250°C due to a phase transition [32]. LiPON can be prepared at moderate temperatures via sputtering deposition but shows high instability against water [33][34][35][36].
Nickel oxide is well available, of rather low toxicity and can be prepared by a variety of methods like electrochemical deposition [37][38][39], chemical bath deposition [40][41][42][43], flame/aerosol-assisted chemicalvapour-deposition [44][45][46] or sputtering [47][48][49]. Low processing temperatures are attractive to protect underlying layers in a given stack and to widen the range of applicable substrates. To reach sufficient NiO layer quality, however, a high substrate temperature of 300°C or above is often needed during preparation or subsequent annealing as seen in a comparison across different methods of preparing NiO films (see Table 1 and Table S1). Among low temperature techniques like sputtering, electrochemical deposition, deposition of pre-formed nanoparticles [50] or pulsed laser deposition which can even be performed with the substrate at room temperature [37], acceptable coloration efficiency could only be reached by electrodeposition or by spin-coating of pre-formed nanoparticles [50]. Sol-gel processes offer great possibilities to produce thin layers at moderate temperature, too, but with increasing film thickness the preparation becomes less reliable [51,52]. All methods offer different advantages and disadvantages and must be selected depending on the needs and conditions in a given device structure. Therefore, a selection from a large variety of methods is important. By use of solution combustion synthesis (SCS), moderate process temperatures for NiO thin films with several nanometer thickness have also been reported [53][54][55]. Such layers, however, do not show sufficient lithium capacity to serve as counterpart to tungsten oxide in a smart window. To provide a beneficial electrochromic effect and to store an adequate amount of charge, a thickness of several hundred nanometers is mandatory.
In this work, we present an economic way to prepare NiO films with such increased thickness prepared below 250°C. Some approaches to such films do exist in the literature. Zhang and Li reported a method in which they used nickel nitrate and urea as oxidizer and fuel to prepare thin films of approximately 65 nm at a temperature around 225°C [54]. In Table 1 Coloration efficiency CE at the given wavelength k EC active electrode layer Method of film preparation CE cm 2 C -1 (k /nm) Aside from the selection of fuel and oxidizer, the ratio of the two serves to optimize the combustion synthesis. By varying this ratio the reaction can be directed to different products, as shown by Bai et al., where the combustion of nickel nitrate with acetyl acetone or glycine leads to metallic nickel or nickel oxide or a mixture of both [55]. According to González-Cortés and Imbert [60], the overall reaction in a combustion synthesis with nickel nitrate and n mole urea, with a fuel to oxidizer ratio corresponding to the ratio of total valences of 3/5 n, can be written as: In the present method, however, organic solvents were used instead of water, the precursor concentration was higher and thicker films were achieved than in typical SCSs [21,61]. In addition, the combustion reaction is based on precipitated precursors in the form of dried films. Because of these substantial differences to SCS, we just use the term combustion synthesis (CS) for this method.
In the following, we present a facile method to prepare thin films of nickel oxide of several hundred nanometres thickness which can be used as anodic electrochromic layers. The components are readily available, no vacuum process is necessary and routine lab equipment suffices. The method yields sufficient film thickness at low preparation temperatures, which provides full compatibility to other components in electrochromic cells. We, therefore, believe that this new preparation method can contribute to develop efficient electrochromic windows that are well affordable, and that may lead to widespread commercialization and significant contributions to energy-saving in modern architecture.

Experimental
For all experiments we approached the solubility limit of nickel nitrate. Therefore, 8.00 mg or 7.12 mg Ni(NO 3 ) 2 Á 6 H 2 O (98%, Alfa Aesar) were completely dissolved in 5.1 ml ethanol (C 99.9%, Roth) or 8.0 ml acetone (C 99.5%, Roth), respectively, and were allowed to rest overnight, before 0.5508 g or 0.4902 mg urea (C 99.5%, Sigma Aldrich) were added, respectively, to establish a molar ratio of 1/3 between the nickel and urea and were allowed to rest overnight, again. Finally, the solutions were filtered by a 0.47 lm Teflon filter and were observed to be stable for months. As substrate, fluorine doped tin oxide glass (FTO) (Zhuhai Kaivo, \ 15 X sq -1 ) was used, which had been cut and cleaned by RBSÒ 35 solution containing sodium hydroxide, tetrapotassium diphosphate, sodium carbonate, sodium hypochlorite and surfactants (Roth), acetone (Roth, C 99.5%) and isopropanol (Roth, C 99.8%) in an ultrasonic bath for 15 min each. Directly before the spin-coating process the FTO substrates were UVozone-treated for 10 to 15 min. The NiO films were prepared by spin-coating with 4000 rpm for the ethanol (ethNiO) or 3500 rpm for the acetone (ace-NiO) based solution for 30 s. The optimum rotation speed to reach a homogenous coverage and a uniform film thickness for the different solutions had been determined in separate experiments before. The samples were then dried at 100°C for at least 20 h in an oven and finally annealed at 230°C for 90 min on a heat plate. Figure 1 shows a schematic representation of the procedure.
Thermogravimetric analysis (TGA) and differential thermal analysis (DTA) were performed on a Setsys Evolution 16/18 system under argon atmosphere. Samples for the TGA/DTA analysis were prepared as the films on FTO, but directly in the ceramic crucible. The resulting masses of the intermediate products in the crucibles, which were initially completely filled with solution, after drying for 24 h, were 23.3 mg and Figure 1 Schematic representation of the facile combustion synthesis procedure of nickel oxide optimized for electrochromic applications. The colours of the EC devices represent the real optical impression measured by a spectrometer and calculated to RGB values by use of the D65 standard illumination (see S1).
15.6 mg for ethNiO and aceNiO, respectively. Samples were heated from room temperature to 230°C at 24 K h -1 and kept at this temperature for 2 h. X-ray diffraction (XRD) was performed on a Siemens D5000 using Cu Ka for samples prepared as described for the TGA/DTA but in a ceramic vessel and from 10 ml of solution. Up to 20 h at 230°C annealing were needed for a complete conversion from the green precursor to the black final product for the XRD analysis. High resolution scanning electron microscopy (SEM) was performed on a Zeiss MER-LIN with an acceleration voltage of 5 kV, an emission current of 120 pA and charge compensation with ionized N 2 . Cyclic voltammetry experiments were performed by use of an IviumStat (Ivium Technologies) with a scan rate of 50 mV s -1 and vertex potentials of -1.2 V and 1.5 V vs Ag/AgCl, if not otherwise noted. Spectroelectrochemical characteristics were determined with the cell placed in a lighttight box by simultaneous use of an Evaluation Line (tec5) optical spectrometer with a spectral range of 310-1100 nm (specified range 360-900 nm). Reference spectra were obtained by use of the empty box. A three electrode setup was used with a platinum wire as counter electrode and a leak-free Ag/AgCl (242 mV vs SCE, Harvard Apparatus) as reference electrode. 1 mol L -1 lithium perchlorate (C 95%, Sigma-Aldrich) in propylene carbonate (anhydrous, Sigma-Aldrich) was used as an electrolyte.
The colouration efficiency CE was calculated by use of the wavelength dependent transmission T for the intercalated and the de-intercalated case (denoted by indices i and d) and the observed charge density DQ: Additionally, we use a calculated visual representation of the colour of films. The calculations were performed according to CIE (International Commission on Illumination) standards by use of an algorithm from the NVIDIA cooperation (more details in the Supplemental Information). In short, the transmission spectra of the sample, D65 illumination and the sensitivity of a typical human eye are used to calculate XYZ colour coordinates, which are converted to the corresponding RGB values. Superior to photographs, such calculated colours are independent of any camera and its position, any contrast and brightness settings, background and ambient light, etc. Furthermore, they can directly be generated from spectra collected during spectroelectrochemical measurements which would not have been possible by use of a camera.

Results and discussion
Films of NiO were prepared (Fig. 1) from mixed solutions of Ni (NO 3 ) 2 and urea with ethanol (eth-NiO) or acetone (aceNiO) as solvent. Other solvents were also tested but with ethanol and acetone the best film quality was obtained. Precursor films also formed from 1-propanol, but films pealed off from the substrate during the combustion reaction (CR). With dimethyl sulfoxide the whole material was spun off the substrate in the coating step. When using highly concentrated water-based solutions (essential to reach high films thickness), too much gaseous products were generated during the CR, leading to large bumps in the film. For ethanol and acetone, the spinning rate, drying time and temperature were varied in a wider range and the optimum conditions were chosen as indicated above. 230°C was determined as the minimum temperature for ignition of the CR by observing a dried film on a hot plate which was slowly heated from room temperature under detailed monitoring of the surface temperature. With n = 1/3 according to the ratio of reactants and according to (1), after rearrangement and multiplication by three, the reaction equation of the present combustion reaction reads: Using the present ratio n = 1/3 ambient oxygen is not needed for the combustion. The end of the conversion to NiO during the heating process can be observed by a change from the complete transparent state to a yellowish/brownish colour, which occurs after approximately 1.5 h. A delayed ignition of the CR is thereby indicated which can be caused by a chain breaking mechanism [62] as directly proven by the TGA analysis of a powder sample (Fig. 2).
In Fig. 2a the mass loss in percentage (line connected circles), the heat flow in lV (lines) for ethNiO (black) and aceNiO (red) and the applied temperature (green line) are plotted against the time. From 0 min (room temperature) up to 525 min (230°C) a constant heating rate of 24 K h -1 was applied. In Fig. 2b the same data for ethNiO and aceNiO are shown with respect to the applied temperature. In the beginning of the analysis at approximately 60°C a little but ongoing loss in mass can be detected, which comes along with a sudden but short decrease in the heat flow for both ethNiO and aceNiO. This is surprising since the materials had been dried for at least 20 h at 373 K (100°C) before the TGA/DTA analysis. Melting of Ni(NO 3 ) 2 Á 6 H 2 O is known to occur around 56°C [63], indicating that some water had readsorbed before the TGA/DTA analysis leading to the re-formation of Ni(NO 3 ) 2 Á 6 H 2 O and gets released at around 60°C again.
The heat flow then decreases for both precursors up to ca. 140°C-the temperature where urea starts to decompose [63]-from where on it stays around zero. In parallel, the rate of mass loss decreases. Chain breaking is thereby shown and obviously has reached a level, where as much heat is produced by the CR as consumed by further drying. While for ethNiO the heat flow crosses the zero line several times between 140 and 175°C the heat flow stays slightly positive for aceNiO throughout this regime. Beyond 175°C the heat flows get negative showing the exothermic character of the reactions still, however, without a significant increase in the rate of mass loss. Only at an applied temperature around 210°C, the rates of mass loss increase and around 215°C for ethNiO and 225°C for aceNiO the ignition is reached indicated by a rapid mass loss and significant changes in heat flow. Following an intermediate rise to the positive region presumably related to phase transitions into the gaseous products of Eq. (3) the heat flow shows a strongly exothermic reaction characteristic for the reaction enthalpy of (1). Consistently, this temperature had also turned out as the optimum reaction temperature of film formation (Fig. 1). The highly exothermic character of (1) is consistent with the reported temperatures up to 973 K (700°C) for the nickel nitrate urea combustion when performed under adiabatic conditions [60].
The XRD analysis in Fig. 3 clearly shows the complete conversion to NiO for both precursors. The ethNiO and the aceNiO diffraction pattern in black and red, respectively, perfectly fit to the cubic fcc structure measured for powder of NiO (ICSD 98-064-6098) depicted in green. Quite comparable relative peak intensities obtained for the films and the powder show the formation of crystalline films with random crystal orientation. Rather broad peaks are observed for the films and by applying the Scherrer equation to the peaks at approximately 37°we calculate an average crystal size of 22.9 nm for ethNiO and 23.4 nm for aceNiO, indicating crystals of quite similar size in the films prepared from both solvents.
The resulting film morphologies characterized by SEM are shown in Fig. 4. The cross-sections and the top views of the films show the high homogeneity of both NiO layers. They have a thickness of approximately 350 nm for the ethNiO and 450 nm for the aceNiO. This trend was also confirmed when the spin-coating was done with the same rotation speed despite higher precursor concentration in ethanol than acetone based solution. A closer look to (a) and (c) reveals, that the aceNiO layer seems to be more porous. Especially close to the FTO layer larger cavities can be found, which would explain the higher thickness for a given amount of material.
Corresponding inhomogeneities can also be seen on top of the surface in (d). Small holes between 1 and 10 nm can be identified. The ethNiO layer, on the other hand, shows more grains but a widely closed structure aside from a few little cracks distributed over the layer. The cyclic voltammograms for the first 10 cycles in Fig. 5 look qualitatively similar for ethNiO and ace-NiO but show an almost doubled current density for the layer obtained from the ethanol-based precursor solution in contrast to the lower film thickness, indicating more efficient charging. For both samples, the initial cycle is consistently measured at slightly different current density indicating a slight conditioning toward Li ? -intercalation. For subsequent cycles, however, a steady state is reached within about four cycles. Even the positions of the cathodic peaks at 0.0 V (I red) and 0.5 V (II red) and the anodic peak at 0.3 V (I ox) are quite fixed from the beginning. Just the second anodic peak (II ox) with 0.9 V for the aceNiO is a little bit shifted to a more positive value from 1.0 V for the ethNiO. The cycling behaviour of the spectroelectrochemical response of the ethNiO and aceNiO layers, also shown in Fig. 5, are in accordance to the observations of Granqvist et al. [48]. The transmittance of the initial state is found close to the average of the coloured and bleached states. Although the ethNiO is thinner than the ace-NiO layer the overall transmittance of ethNiO is lower. Granqvist et al. showed that the transparency of NiO relates to a non-stoichiometric content of oxygen. They observed decreasing transparency for NiO x layers with increasing x. It is thereby indicated that ethNiO is characterized by higher x in NiO x than aceNiO.
A closer look to the peak current densities of I and II plotted in Fig. 6 reveals more differences. Compared to ethNiO, aceNiO shows much less variation in the peak current densities during cycling. Especially the second oxidation/reduction process (II) at aceNiO quickly establishes a steady state. Generally, a quite parallel increase of absolute current densities is observed for all peaks. The observed changes seem to refer to an identical underlying process. In the context of the results of Granqvist et al. [48] and  Rougier et al. [64] this could indicate irreversible surface reactions of NiO with the anions and cations of the electrolyte upon cycling. In Fig. 7, the charge density for each cycle is plotted for the two kinds of films ethNiO and aceNiO. The initial cycle now appears significantly shifted from all subsequent cycles for both materials and a charge density unlike zero can be observed at the end of all cycles, in particular for the first cycles and for cycles of ethNiO whereby an ongoing modification might be indicated.
Clear electrochromic switching was observed for ethNiO and aceNiO. The ethNiO film changes its transparency at 550 nm by DT = 25% from 30.4 to 55.3% and the aceNiO sample by DT = 34% from 41.5 to 75.7% despite smaller charge exchanged by ace-NiO compared with ethNiO (Fig. 5). A colouration efficiency of 47.1 cm 2 C -1 at 550 nm is calculated for ethNiO and of 92.3 cm 2 C -1 for aceNiO upon cycling between 1.5 V and -1.2 V which is one of the highest values for NiO in LiClO 4 /PC reported (see Table 1). The scan rate of 50 mV s -1 results in a corresponding switching time of 54 s, considerably shorter than 10 min which are typical for use in smart windows [65,66]. To explore the maximum accessible switching range, potential steps of -2 V and ? 2 V were applied for 90 s (Fig. 8) each. ethNiO shows a change from 23.4 to 56.8% and the aceNiO sample a change from 26.5 to 81.1% in transmission at 550 nm. This corresponds to changes of DT = 33% and DT = 55% for the two materials, respectively, and can be considered technically attractive. Whereas, the derived parameters differ significantly from the cyclic voltammogram to the double potential step measurement, this is not the case for the colour impression, as can be seen by comparing the top and bottom part of Fig. 8. The colours surrounding the data points are calculated from the collected spectra (as described in the supplementary information) and represent the visual colour impression very well. The cycle in Fig. 8a shows a CE of 92 cm 2 C -1 for the sweep from -1.2 V to 1.5 V but the last potential step in Fig. 8b provides a CE of only 20 cm 2 C -1 , without significant differences in the optical impression. The switching time of 2.3 s to reach 90% of maximum transparency and of 16.9 s to reach 90% of the maximum absorbance at 550 nm are in a typical range for NiO [50,51,[56][57][58]. However, both materials suffer from chemical instability even within the first 5 cycles under the present conditions. This problem was stated also earlier, albeit at higher cycle numbers but less intense changes in transmission [48].
In order to discuss the efficiency of a material as an electrochromic layer, the observed change in transmission of a layer in contact to a given electrolyte should be compared to the charge exchanged. From Eq. (2) it directly follows that a high coloration efficiency is achieved for large changes in transmission at small charge uptake. The values of Q/V in Table 1 can be used to discuss the level of charging of the films at which this change in transmission was observed. Among the low-temperature, solutionbased films, ethNiO provided most complete charging of the film (158 C cm -3 ) at the given CE, slightly higher than aceNiO (63 C cm -3 ) or NiO prepared from nanoparticles (89 C cm -3 ) [50]. As another relevant criterion, significantly high charge uptake in a given film area Q/A is needed in order to qualify as an anode for, e.g., WO 3 as electrochromic cathode, for which 4 mC cm -2 to 50 mC cm -2 are desirable [67][68][69][70]. NiO-layers characterised in OH --based electrolytes often show a higher charge density (see Table S1) but this electrolyte is incompatible to electrochromic switching of WO 3 and, therefore, the present films are compared to those in Table 1. Q/ A in Table 1 represents the charge uptake for which CE was determined and is of interest to discuss the methods regarding a good combination of high CE and high Q/A. NiO layers from the spray pyrolysis technique offer the highest Q/A in this comparison at, however, rather high temperatures during processing and moderate CE. Low-temperature NiOlayers with a high CE like the nanoparticle NiO show rather small Q/A. In comparison, the present ethNiO and aceNiO may offer a good compromise of a decent CE at moderate Q/A. An optimum has to be established because CE of NiO across methods of film preparation typically drops with increased Q/A or Q/V. This was also seen in chronoamperometric experiments (Fig. 8) with ethNiO-and aceNiO-layers, for which Q max /A values of 114 mC cm -2 and 46 mC cm -2 were achieved at, however, decreased CE around 4 cm 2 C -1 and 10 cm 2 C -1 , respectively. The Extracted peak current densities from the reduction and oxidation peaks for each cycle for ethNiO (black squares, left axes) and aceNiO (red circles, right axes) with identical step sizes but different current ranges for the axes of a given peak.
corresponding maximum achieved volume charge densities (Q max /V) of 3260 C cm -3 and 1020 C cm -3 compare favourably to values of 180 C cm -3 and 690 C cm -3 calculated from published values (Table 1). Optimum working conditions considering both parameters CE and Q/A, therefore, have to be established for each given type of device and application. The values of CE, Q/A and Q/V in Table 1 (and Table S1) can serve as benchmark values for choosing an appropriate preparation method for an envisaged device.

Conclusions
We have demonstrated that the preparation of NiO layers by a combustion reaction of urea and nickel nitrate as fuel and oxidizer leads to very attractive film formation even at low processing temperatures. It is strongly influenced by the solvent used in the precursor solution. Such impact of the solvent indicates that the mechanism for the combustion reaction of nickel nitrate and urea as depicted in Eq. (1) is not complete for the presently presented reaction in dried layers but may show an additional input from solvent or film texture. Ethanol as solvent leads to more compact grainy films with a thickness of approximately 350 nm and a low general transparency. On the contrary, NiO layers prepared from acetonebased precursor solutions are porous and show a high transparency despite a thickness of approximately 450 nm. Both films have the necessary thickness to serve as a counter electrode for typical electrochromic tungsten oxide films and can be prepared below 250°C. The transmission could be  In both graphs the black points represents the data points, which are surrounded by the corresponding colours calculated from the collected spectra and calculated by the method described in the supplemental.
reversibly changed by about 25 and 34% at 550 nm for the ethNiO and aceNiO samples, respectively. This leads to a colouration efficiency of 47.1 cm 2 C -1 for ethNiO and a remarkable high efficiency of 92.3 cm 2 C -1 for aceNiO during cycling, which belong to the highest CE values reported in the literature for NiO in contact to LiClO 4 /PC so far. These changes in colouration can be considered attractive for an addition of a counter anode to the 55 cm 2 C -1 typically reached when switching WO 3 cathodically. In addition, the blue characteristics of coloured WO 3 devices might be spectrally broadened by the use of these NiO layers as counter electrodes to WO 3 . The method presented here, therefore, shows the usefulness of solvents in a combustion reaction for NiO film preparation and their potential application in electrochromic devices. It may, therefore, help to reduce the production costs of electrochromic windows and glasses, broaden their commercial application and contribute to energy-saving solutions in architecture.

Supplementary information
Extended information about the calculation of the colors for the graphical representation of the samples from the recorded spectra can be found in the supplementary information as well as a table with a comparison of the presently achieved characteristics (in analogy to Table 1) with reported data on different NiO films in contact to OHbased electrolytes. by the machine-shop of the Physics Department (Univ. Giessen). We also thank Omeir Khalid (Inst. of Physical Chemistry, Univ. Giessen) for the TGA and DTA measurements.

Compliance with ethical standards
Conflict of interest There are no conflict to declare.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licen ses/by/4.0/.