Electrochemical Studies of Inkjet Printed Semi-Transparent NiCo₂O₄/ITO Supercapacitor Electrodes

Abstract

Transparent supercapacitors find a large number of applications as components of many electronic devices and circuits. Mixed Ni–Co oxides (NCOs) are among the most promising supercapacitor electrode materials exhibiting high pseudo-capacitance and good electronic conductivity, while inkjet printing is a low cost and versatile technique for electrode printing. Surprisingly, although there have been many studies of NCO supercapacitor films on ITO glass substrates, these have not been prepared by the inkjet technique, and their optical properties were not fully characterized. Hereby, we report the fabrication and characterization of thin (295 and 477 nm thick; 0.017 and 0.035 mg cm⁻² NCO loading) semi-transparent NiCo₂O₄/ITO supercapacitor electrodes, showing transparency to visible light (60–30%, from the thinner to the thicker electrode layers tested), typical mass specific capacitance for NCO-based supercapacitor electrodes (1294–829 Fg⁻¹ at 1 mA cm⁻² discharge current density) and high volumetric capacitance (746–608 F cm⁻³ at 1 mA cm⁻²). The NCO nanoparticles were prepared by hydrothermal synthesis followed by thermal treatment and ball milling (ZrO₂ balls, 0.5 mm diameter), resulting in a cubic nickel–cobalt oxide structure and particle size in the 30–150 nm range, whereas the electrode layers were printed from water-propylene glycol solutions using a Dimatix DMP-2850 drop-on-demand (DoD) inkjet printer. Constant current charge–discharge experiments of the supercapacitor electrode (at ca 0.5 mA cm⁻²) for 1000 cycles confirmed stability of performance.

1. Introduction

In the energy storage field, the search for high performance electrochemical devices has drawn significant attention, and supercapacitors, having unique properties such as fast charge–discharge rate, high power density and extended cycle life, are among the most promising of these (see, for example, [1,2,3,4,5,6,7,8,9]). As it is well known, the electrochemical performance of supercapacitors is largely determined by the properties of electrode materials, whether these are of the Electrochemical Double Layer, EDL, type (whereby, low cost carbonaceous materials of high surface area are researched; see, for example, [10]) or of the pseudo-capacitor/battery type (whereby materials of high redox activity, such as conducting polymers and metal oxides, are sought; see, for example, [11]). Mixed Ni–Co oxides (NCOs; usually corresponding to the NiCo₂O₄ formula) are among the most promising supercapacitor electrode materials exhibiting high pseudo-capacitance/redox activity and good electronic conductivity (see, for example, [12,13]). Ni and Co oxides alone already exhibit high theoretical pseudo-capacitance values (2573 and 3560 Fg⁻¹, respectively) due to the relatively fast Ni(II)/Ni(III) and Co(III)/Co(IV) redox transformations, but suffer from relatively low electronic conductivity; however, the mixed Ni–Co oxides, due to their spinel structure, possess adequate conductivity for supercapacitor applications. Additional advantages of the material result from the fact that it is based on inexpensive, earth-abundant elements which are also relatively environmentally benign.

At the same time, thin film supercapacitors find many applications in portable, flexible and wearable electronics (to be used in clothing and fabrics) as well as in transparent devices (to be used for example in touchpad displays and transparent circuits in general), to mention a few of the many applications that small-size power sources find in the ever-expanding world of portable personal devices [14,15,16]. Printed energy storage devices can be built by printing liquid functional materials such as organic and inorganic nanomaterials (see, for example, [17]). Inkjet printing is an ideal method for the deposition of nanomaterials for device fabrication because it is a non-contact, precisely controlled deposition and additive printing process, particularly applicable to thin film preparation. The method is a fast, inexpensive and environmentally friendly technique that also allows microstructuring (see, for example, [18]).

Although there are many reports of thin (active material loading ≤ 1 mg cm⁻²) NiCo₂O₄ supercapacitor electrodes [19,20,21,22,23,24,25,26,27,28,29] (some of which are on ITO substrates [25,26,27,28]) and there are also a few reports of supercapacitor electrodes fabricated on glass substrates by the inkjet printing technique [30,31,32,33], surprisingly, one cannot find any paper reporting semi-transparent supercapacitor NiCo₂O₄ electrodes prepared on ITO glass by the inkjet technique.

Finally, one should mention that combining mixed transition metal hydroxides and metals with transition metal sulfides has been reported to result in excellent supercapacitor materials (see, for example, [34,35]).

The aim of this work has been the fabrication and testing of semi-transparent inkjet-printed NiCo₂O₄ nanoparticulate thin film electrodes on ITO glass substrates for electrochemical supercapacitor applications. To that end, the objectives have been: (a) the hydrothermal synthesis of NiCo₂O₄ nanoparticles and their complete physicochemical characterization (XRD, SEM, AFM), (b) the optimization of the inkjet printing method and of the resulting thin inkjet printed films and their microscopic and optical characterization and (c) the complete electrochemical characterization of the resulting thin films as supercapacitor electrodes (cyclic voltammetry, electrochemical impedance spectroscopy, galvanostatic charge–discharge, stability tests).

2. Results and Discussion

2.1. Structural and Morphological Properties of NCO Nanoparticles

Powder X-ray diffraction (XRD) of NCO nanoparticles as made and after annealing at 350 °C was used to verify the crystal structure and phase of the nickel–cobalt oxide (NiCo₂O₄; NCO) as shown in Figure 1. The XRD pattern before annealing shows the characteristic peaks at 2θ 18.7°, 29.1°, 34.9°, 43.3°, 47.3°, 48.4° and 58.1°, which are attributable to the (202), (114), (022), (224), (602), (026) and (712) reflections of orthorhombic cobalt oxalate hydrate (C₂H₄CoO₆). In addition, the characteristic peaks at 2θ 18.6°, 30.2°, 35.5°, 43.3°, 47.1°, 48.7° and 57.9° are attributable to the (202), (400), (022), (512), (225), (026) and (332) reflections of orthorhombic nickel oxalate hydrate (C₂H₄NiO₆). The XRD pattern after annealing at 350 °C shows characteristic peaks at 2θ 18.91°, 31.15°, 36.70°, 38.41°, 44.62°, 55.44°, 59.09°, 64.98°, 68.31, and 77.55° which correspond to (111), (220), (311), (222), (400), (422), (511), (440) and (531) reflections of the crystalline structure of cubic nickel–cobalt oxide (NiCo₂O₄, NCO) with a cubic crystal system and correspond to the F*3 space group (JCPDS file No. 00-020-0781).

The phase composition of the ITO substrate and inkjet-printed NCO layers was also characterized by XRD, as seen in Figure 2. The XRD pattern of the ITO glass substrate exhibits four peaks which are consistent with the reflections of planes (222), (400), (440) and (622) of indium tin oxide at 30.3°, 35.1°, 50.6° and 59.9°. The main characteristic peak (311 reflection) of cubic nickel–cobalt oxide can be observed at 36.7° for the samples with one and two inkjet-printed layers, and its low intensity can be attributed to the small film thinness.

Figure 3a shows the SEM image of deposited NCO nanoparticles. The powder consists of irregularly shaped particles with a porous morphology. The mean particle size varies significantly with diameters between 200 and 800 nm, as well as the formation of large agglomerates over 1–2 μm. After the ball milling process, the NCO particles have been reduced to a homogenous nanoparticle distribution with diameters less than 200 nm. Specifically, average particle diameter distributions before and after the process are given in Figure 3c,d. Before the milling process, NCO nanoparticle size ranged largely from a few nanometers to micrometers with an average diameter of 450 nm. After the process, nanoparticle diameter narrowed down to 30–150 nm with an average value of 65 nm, demonstrating the effectiveness of the ball milling process.

Figure 4 shows transmission electron micrographs (TEM) of individual NCO nanoparticle aggregates. The material displays large clusters of NCO nanoparticles with diameters of about 100–300 nm, which consist of smaller nanoparticles in the range of 5–9 nm.

EDX elemental analysis of NCO/ITO electrodes (Figure 5) confirms that Ni and Co elements are present in the mixed NCO at a relative ratio of 1:2, as expected for the NiCo₂O₄ compound, as well as the presence of In, Sn, Si and O of the ITO glass substrate.

Finally, EDS mapping of NCO/ITO substrates, presented in Figure 6 below, confirms a uniform and continuous distribution of the NCO material on the ITO substrate.

2.2. Physicochemical Properties of NCO Nanoparticles Ink

The ability of an ink to be jetted is controlled by several properties such as viscosity and surface tension. Ink jet ability is correlated with a dimensionless number Z, the inverse Ohnesorge number, which is defined in Equation (1) as

$$Z={\left(Oh\right)}^{-1}=\frac{\sqrt{\rho \gamma \alpha }}{\eta }$$

(1)

where ρ is the density, γ is the surface tension, η is the viscosity of the suspension and α is the characteristic length scale (nozzle diameter). This dimensionless number relates inertia and surface tension to the viscous force of a liquid. For drop formation in DoD inkjet printing systems, the ideal range is 1 < Z < 10 for reliable droplet formation and jetting. An aqueous-based ink needs to have certain physicochemical properties inherent to water modifications. Despite its great many advantages as a solvent, high surface tension (about 70 mN/m) and low viscosity (about 1 mPa s) make water practically not inkjet-printed. In our case, propylene glycol was used as a viscosity modifier and increased viscosity from 1.00 to 1.71 mPa, thus improving jettability. In addition, Triton X-100 (a non-ionic surfactant) was added accordingly as a surface tension modifier.

Table 1. Physicochemical properties of the aqueous NCO ink and calculation of Z number for different operating temperatures.

T/°C	ρ/(kg/m3)	γ/(10−3 N/m)	η/(10−3 Pa s)	α/(10−6 m)	Ζ
25	1026.4	27.5	2.5	21.5	9.9
30	1023.9		2.1		11.7
35	1021.3		1.8		13.7

shows the various physicochemical properties and calculated Z number for an aqueous NCO nanoparticles ink at different operating temperatures. At room temperature (25 °C), Z is within the desired range with higher temperatures exceeding the Z value of 10. This suggests an optimum operational cartridge temperature of 25 °C for this ink. At higher temperatures, when Z exceeds 10, the ink droplets are prone to splitting and formation of satellite drops which also aligns with the reduction in viscosity.

Figure 7a shows the black color of the milled nanoparticle dispersion and prepared NCO ink. The average particle diameter within the suspension after formulation is around 150 nm, with most of the distribution well under 400 nm, as shown in Figure 7b. This indicates a low agglomeration rate and is highly important for the stability and printability of the ink. To avoid nozzle blockage problems, it is recommended that the maximum particle size does not exceed 1/50 of the nozzle diameter which corresponds to 430 nm. One other critical parameter is the shelf life of the ink. NCO nanoparticle ink shows very good stability through seven storage days (Figure 7d). After seven storage days, minimal precipitation was observed and was reversible with simply stirring by hand.

Thermogravimetric analysis was performed to examine mass loss during the annealing process. Two distinct curves can be seen in Figure 7c, corresponding to the different solvents used for the formulation of the NCO ink. For the temperature range of 25–100 °C, a 75% mass loss is attributed to the evaporation of H₂O. At 100–125 °C, propylene glycol is evaporated/decomposed with a 23.4% mass loss. These losses are in agreement with the solvent ratios used during formulation. NCO loading was estimated at 350 °C using the remaining 4.2% and was found to be 25.7 mg/mL. From the total volume of drops used to prepare one and two NCO layers samples (1L, 2L), the NCO loading is estimated to be 0.0175 and 0.0035 mg cm⁻², respectively.

Figure 8a shows the contact angle behavior of a 2 μL NCO ink drop on an ITO glass substrate at room temperature. The initial angle of 39° indicates good wettability between the ink and the substrate, which is needed during pattern deposition. The angle rapidly decreases during the first 20 s due to the highwater evaporation rate. Using a pendant drop measurement in Figure 8b, the ink’s surface tension was calculated at 27.65 mN/m, a value within the desirable inkjet-printed range.

2.3. Fabrication of Inkjet Printed NCO Electrodes

Jetting and droplet formation is presented in Figure 9a, where the ink exhibits an ideal behavior with the formation of a 10 pL cherry-like droplet. While exiting the cartridge nozzle, a small tail is produced that later retracts back to the main body. An SEM image of such a droplet deposited on an ITO glass substrate is shown in Figure 9b. The resulting circular pattern has a diameter of 42.6 um and shows excellent coverage without the appearance of irregularities such as the coffee ring effect. It consists of densely packed/agglomerated NCO nanoparticles (Figure 9c).

Single drop lines at different drop spacing values were deposited as shown in Figure 10a. Low drop spacing values result in the coagulation and distortion of lines due to heavy overlap between droplets. High drop spacing, on the other hand, results in isolated droplets that do not form a continuous pattern. A similar image is shown for thicker lines in Figure 10b. To create homogenous, evenly shaped patterns without gaps, a combination of specific drop spacing with multiple deposition layers can be used. An example of 50 μm thick lines at 40 μm DS and two layers is presented in Figure 10c,d.

Figure 11a,b depict cross-sectional SEM images of 1L and 2L NCO/ITO electrodes, whereby compact NCO overlayers can be clearly seen. The average NCO thickness of 1L electrodes was found to be 295 nm, whereas that of 2L electrodes was 477 nm (indicating the filling in of gaps present in the first printed layer).

UV–VIS spectrometry was used to estimate the transparency of the ITO glass substrate and the 1L-, 2L-coated samples. As seen in Figure 12, over the visible light wavelength range, the transmittance is relatively stable at an average of 83.64% for the ITO glass substrate due to the indium tin oxide thin film. The 1L and 2L semi-transparent NCO thin films have a visible (400–800 nm) transmittance of 40–60% and 35–50%, respectively.

Film morphology and roughness at the nanoscale were studied by AFM (Figure 13). The 1L film appears smoother than the 2L one (average roughness, S_a = 1.35 nm and 37.71 nm, respectively), with the latter showing a higher degree of particle agglomeration.

2.4. Electrochemical Performance/Electrochemical Characterization

Cyclic voltammograms (CVs) have been recorded for the preliminary investigation of capacitive/pseudo-capacitive/battery-type behavior of the electrode material (the latter two requiring the identification of electron transfer/faradaic/redox reactions, surface or diffusion-controlled reactions, respectively) [36], as well as the determination of the useful potential window of the material. Figure 14 shows CVs of 1L and 2L inkjet-printed NCO electrodes in 1 M NaOH, at various scan rates from 5 to 50 mV s⁻¹_, in the potential window of −0.2 to +0.7 V vs. SCE.

A pair of redox peaks is clearly seen at all scan rates, with their separation increasing with potential scan rate, which indicates that the inkjet-printed NiCo₂O₄ electrode is undergoing a quasi-reversible faradaic reaction. Also, both the anodic and the cathodic peak are broad (with the anodic one showing a shoulder in the +0.50–+0.20 V range and the cathodic being split in two in the +0.20–0.00 range) as expected for mixed Ni and Co oxide electrochemistry and indeed recorded in all similar studies [19,20,21,22,23,24,25,26,27,28,29,30,31,32,33]. The redox peaks are attributed to the following redox reactions involving Ni and Co oxides and oxo-hydroxides as well OH⁻ ions from the solution:

NiO + OH⁻ ↔ NiOOH + e⁻

(2)

and

Co₂O₃ + 2OH⁻ ↔ 2CoO₂ + H₂O + 2e⁻

(3)

that occurs in two steps:

Co₂O₃ + H₂O ↔ 2CoOOH

(4)

2CoOOH +2OH⁻ ↔ 2CoO₂ + 2H₂O + 2e⁻

(5)

Plots of the oxidation–anodic peak current density (current per substrate geometric area) with the potential scan rate is given in the insets of Figure 14. A linear correlation is observed for scan rates up to 20 mV s⁻¹ (with a correlation coefficient higher than that of similar plots of peak currents versus the square root of the scan rate), in line with the findings in [21]. This indicates that the very thin NCO layers behave as hybrid supercapacitors, exhibiting pseudo-capacitive behavior (based on the classification given in [36]); it also means that the redox reactions (2) and (3) presented above are controlled by a surface electron transfer step and not by OH⁻ mass transport limitations through the pores of thick battery-type electrodes. (Deviation from linearity, at the highest scan rate of 50 mV s⁻¹ and associated current densities higher than 1 mA cm⁻², may result from the onset of diffusion limitation of OH⁻ through the solution).

Figure 15a,b display chronopotentiometry results of galvanostatic charge–discharge (GCD) experiments of 1L and 2L inkjet-printed NCO electrodes, at constant charge–discharge current densities in the ± 0.2–2.0 mA cm⁻² range, between the voltage cut-off limits of +0.45 V and 0.00 V (deduced from the CVs to allow for oxide formation and reduction reactions, but excluding the oxygen evolution reaction, OER). The shape of the curves points again to an intermediate picture between that which is typical for electrical double layer (EDL) supercapacitors (of a triangular shape) and battery-type material (showing an extended voltage plateau), i.e., that of a hybrid supercapacitor exhibiting pseudo-capacitive behavior [36,37]. In fact, the shape of the curves implies that one could approximate the discharge part with a straight line between the potential limits (since the area under such a line can graphically be estimated to be within 10% of the experimental curve) and thus use as an approximation for the specific/areal capacitance C estimation, the simple relation that holds for EDL supercapacitors and a potential-independent C [37]:

$$C=\frac{j{t}_D}{{\Delta V}_{corrected}}$$

(6)

where j is the discharge current density, t_D is the discharge time and ΔV_corr is the voltage change corrected for ohmic losses. However, the rigorous equation that should be used for the average areal capacitance, C_ave, of a hybrid supercapacitor is that of the general expression [5,28,37]:

$$C_{ave}=\frac{2j{\int }_0^{t_D}Vdt}{{{(\Delta V}_{corrected})}^2}$$

(7)

The origin of the sudden drop in the potential after current reversal is dual: it results from uncompensated ohmic losses, Ru, (due to contact, ITO and NCO particle electrical resistance as well as due to solution resistance between the Luggin tip of the reference electrode and the working electrode) and ionic resistance through the porous electrode film. (For complete, two-electrode devices, the sum of these two resistances is known as equivalent series resistance, R_ES [37]). To estimate the uncompensated ohmic losses and correct ΔV to ΔV_corr, (and, accordingly, C_ave) we carried out Electrochemical Impedance Spectroscopy (EIS) experiments at potential values in the 0.30–0.45 V vs. SCE potential range.

Figure 16 below presents Nyquist plots of EIS data for 1L and 2L NCO/ITO electrodes in a 1 M NaOH solution in the frequency range of 10⁵–0.1 Hz, obtained with an AC amplitude of 10 mV.

Asymmetric and depressed semi-circles can be seen whose initial, distorted part at the highest frequencies used is typical of ionic resistance through porous electrodes in general [38,39] and particulate supercapacitor electrodes [23,26,27,37,40]. The rest of the semi-circle can be attributed to the surface electron transfer reactions of the NCO, and the smaller diameter of that corresponding to the 2L electrode indicates that it is characterized by a smaller charge transfer areal resistance, R_ct, due to its higher real surface area (being rougher according to the AFM micrographs of Figure 13). The near-linear rise of the curve at lower frequencies could either be assigned to a Warburg impedance (associated to a diffusion process, most likely the transport of OH⁻ ions within the pores of NCO electrodes during the redox reactions) or to a constant phase element CPE with an n value close to 0.5 (associated to a circuit element intermediate between a resistor and a capacitor, in line with the hybrid supercapacitor behavior of the material). Fitting an appropriate electrical circuit to the data ([R_u(R_pCPE_p)([R_ctW]CPE_dl)]; inset of Figure 16) accounts for the above-mentioned processes and is also similar to those appearing in the literature [41,42]. Estimated values of the circuit components are presented in

Table 2. Electrical circuit component values corresponding to the equivalent circuit shown in the inset of Figure 16.

	1L	2L
Ru/Ω cm2	41	11
Qp/Ω−1 sn1 cm−2	0.0018	0.0003
n1	0.485	0.720
Rp/Ω cm2	118	63
Qdl/Ω−1 sn2 cm−2	0.010	0.012
n2	1.1	0.93
Rct/Ω cm2	27.50	4.00
W/Ω S1/2 cm2	0.00080	0.00085
Χ2	0.007	0.005

, and an interpretation follows.

The difference in uncompensated resistance values, Ru, of the two samples is too large to be attributed to solution resistance variations due to a change in the reference electrode Luggin capillary—working electrode distance (especially since this was fixed)—and it points to better interparticle electrical contact in the case of the more compact thicker (2L) film. This is also in line with the 2L electrode having a lower pore resistance, R_p, pointing to a surface pore network, not extending to the ITO substrate. Converting the Q_dl admittance values to electrical double layer capacitance values, C_dl, based on the expression of the Brug equation proposed by Jovic et al. [43], one obtains almost identical values of 0.012 and 0.010 F cm⁻² for 1L and 2L electrodes, respectively, that can be explained by the 1L electrode being more porous and only the outer layers of the more compact 2L electrode exhibiting adequate porosity to render them accessible to the electrolyte; hence, they are electroactive (this is also indicated by similar W diffusion resistance values). Finally, the much higher charge transfer resistance of R_ct for the 1L electrode, when compared to that of the 2L one (unexpected for the same material and similar electroactive areas—similar double layer capacitance values) can only be rationalized if the higher porosity of the 1L film allows for the NCO redox surface reactions to occur not only at the NCO/electrolyte interface but also at locations of the ITO/NCO/electrolyte triple phase interphase, with ITO having an apparently inhibitory effect.

The values of 41 and 11 Ω cm² that were estimated for R_u in experiments with 1L and 2L electrodes, respectively, have been used to correct the potential drop during the discharge part of the GCD curves:

ΔV_corrected = ΔV − |j| × R_u

(8)

Using Equations (7) and (8), we can calculate the values of C_ave from the GCD data of Figure 15. Figure 17 shows the specific average capacitance (per mass of NCO) as a function of the charge–discharge current density. High values were obtained and remained nearly constant up to current densities higher than 1 mA cm⁻², indicating good process reversibility. (Also note that an area-normalized current density of 1 mA cm⁻² is significantly high for very thin layer supercapacitor electrodes as it translates to 58 and 29 A g⁻¹ (for 1L and 2L electrodes) and lies in (or exceeds) the range studied for most similar NCO electrodes [19,20,21,22,23,24,25,26,27,28,29,30,31,32,33]).

Finally, medium-term stability GCD tests were carried out for both 1L and 2L inkjet-printed NCO/ITO electrodes by carrying out 1000 GCD cycles and monitoring the capacitance over time. The results for the areal average capacitance shown in Figure 18 below indicate that no capacitance deterioration was observed during the tests. (In fact, there seems to be an activation of the 1L NCO/ITO electrode during the first 300 cycles.)

3. Discussion

The proof-of-concept for preparing low cost and high-performance semi-transparent metal oxide supercapacitor electrodes by inkjet-printing has been presented in the case of NiCo₂O₄ layers on ITO glass substrates. Very low loading (0.017 and 0.035 mg cm⁻²), thin (295 and 477 nm) and uniform NiCo₂O₄ films were deposited on ITO from dispersions of hydrothermally produced nanoparticles of the mixed oxide, by an appropriately optimized inkjet printing technique (using a water-propylene glycol-Triton X-100 ink formulation). The films showed a visible light transparency in the 30–60% range, making them ideal candidates for transparent electronics applications.

Preliminary electrochemical testing of the electrodes by cyclic voltammetry (CV), galvanostatic charge–discharge (GCD) and electrochemical impedance spectroscopy (EIS) confirmed that these very thin, redox-active materials (due to Ni and Co centers) behave as hybrid-type supercapacitors, as indicated by the appearance of quasi-reversible redox peaks in CVs, distorted GCD curves bearing narrow plateaus and a combination of semi-circles and straight lines at lower frequencies in Nyquist plots of EIS experiments.

These supercapacitor electrodes, during GCD experiments at 1 mA cm⁻², exhibited areal capacitance values in the 22–29 mF cm⁻² range and very high mass specific capacitance values that can reach 1294 F g⁻¹ for the thinner samples tested.

Table 3. Preparation and performance parameters of thin (loading ≤1 mg cm⁻²) NCO supercapacitor electrodes found in the literature and studied in this work. (Parameters were either taken as quoted in the literature or calculated from data therein).

Electrode Substrate [Ref.]	Method of Preparation	Loading/ mg cm−2	Film Thickness/ nm	jD/ mA cm−2	Careal/ mF cm−2	Cmass/ F g−1	Cvol/ F cm−3
Graphite [19]	electro-deposition	0.43		0.26	301	700
CNT/ Graphite [20]	thermal decomposition	0.48		1	273	569
Ni [21]	hydrothermal, annealing	1		1	480–610	480–610
carbon cloth [22]	hydrothermal, annealing	0.83		1	581	700
Ni foam [23]	hydrothermal, annealing	1		1	758	758
ITO [25]	co-precipitation, annealing	0.3	2000–3000	2	105–147	350–490	525–735
Ni/ZnO/ ITO [26]	co-electro- deposition	0.1–0.2		1	120	600–1200
Ni/ITO [27]	co-electro- deposition	0.1–0.2		1	35–79	394–787
ITO [28]	hydrothermal, spin-coating, annealing	0.02	43	1	17	850	3000
ITO [this work]	hydrothermal, annealing, inkjet-printing (1 layer)	0.017	390	1	22	1294	746
ITO [this work]	hydrothermal, annealing, inkjet-printing (2 layers)	0.035	1000	1	29	829	608

below presents a comparison of supercapacitor NCO electrode characteristics and performances for the electrodes of this work and similar systems that appear in the literature. (Note that for very thin film supercapacitor electrodes, the areal capacitance is much smaller than state-of-the-art thicker supercapacitor electrodes—see, for example, values as high as ca 3500 F cm⁻² that are reported for hierarchically structured composite materials in [44]). One can see that the semi-transparent 1L NCO/ITO inkjet-printed electrode of this work exhibits the highest mass-specific capacitance among all the thin NCO electrodes tabulated and good volumetric capacitance too (comparable to that of one [25] of the two other values [25,28] that could be estimated from results found in the literature).

4. Materials and Methods

4.1. Materials

All the reagents were used as received and without further purification. Nickel(II) acetate tetrahydrate ((CH₃COO)₂Ni·4H₂O), cobalt(II) acetate tetrahydrate (CH₃COO)₂Co·4H₂O) and oxalic acid (HO₂CCO₂H₂·H₂O) were purchased from Merck (Darmstadt, Germany)).

4.2. NCO Nanoparticles

In a typical synthesis procedure, 5 mmol nickel acetate, 10 mmol of cobalt acetate and 10 mmol of oxalic acid were dissolved in 30 mL of distilled H₂O and magnetically stirred for 30 min until a white, homogenous suspension was obtained. This was then transferred to a hydrothermal process bottle for 12 h at 90 °C. The resulting dispersion was centrifuged at 4000 rpm for 20 min. The white precipitate was washed twice with ethanol at 4000 rpm for 10 min and later dried at 80 °C for 4 h in a vacuum oven. The produced white powder was annealed at 350 °C for 2 h with a heating rate of 5 °C/min to obtain the final black nickel–cobalt oxide (NCO) particles.

4.3. Ink Preparation

For the formulation of the ink, the reagents used were distilled water, propylene glycol (C₃H₈O₂) and Triton X-100 (C₁₄H₂₂O(C₂H₄O)_n) from Merck. (Darmstadt, Germany) To reduce the particle size of the NCO powder, a ball milling technique was used where 3 g of NCO powder was dispersed in 25 mL of distilled H₂O and placed in a ZrO₂ vessel along with 100 g ZrO₂ balls (0.5 mm diameter). The NCO dispersion was then centrifuged at 700 rpm for 30 min in a process that was repeated four times. A black dispersion with high NCO nanoparticle loading was thus obtained and subsequently mixed with propylene glycol in a 7:3 volume ratio (NCO/H₂O:propylene glycol). Triton X-100 surfactant (0.5 mg/mL) was added in the suspension to improve the dispersion of the NCO nanoparticles. Then, the aqueous dispersion was filtered through a 0.45 μm syringe filter with a nylon membrane to remove large agglomerate particles which can cause blockage-related problems to the inkjet printer nozzles.

4.4. Fabrication of Inkjet Printed NCO Supercapacitor Electrodes

All patterns and devices were inkjet printed by a Dimatix DMP-2850 drop-on-demand (DoD) inkjet printer (Fujifilm Dimatix (Santa Clara, CA, USA)) equipped with a 10 pL piezoelectric cartridge (DMC-11610) and 3 mL of ink capacity. NCO nanoparticle ink was deposited on ITO glass substrates (10 Ohm/square). Before use, the ITO glass substrates were cleaned with a mixture of water and ethanol. The printed pattern dimensions were 5 × 1 cm with a 1- or 2-layer thickness, at a drop spacing of 40 μm and a cartridge temperature of 25–35 °C. Printing was followed by annealing in air at 350 °C for 2 h with a heat rate of 5 °C/min. (A schematic representation of NCO particles, ink and printed film preparation is given in Scheme 1 below).

4.5. Characterization

X-ray diffraction (XRD) was conducted on a Rigaku D/MAX-2000H (Tokyo, Japan) rotating anode diffractometer (1.54 Å CuKa radiation) equipped with a secondary pyrolytic graphite monochomator; X-Pert Highscore Plus was used to analyze and identify the diffraction patterns/determine the composition and structure of the NCO nanoparticles. Surface morphologies of the synthesized products were examined by Scanning Electron Microscopy (SEM) that was conducted on a JEOL 7000 electronic microscope JEOL Ltd., Tokyo, Japan), operated at 20 kV. Effective particle diameter was measured using DLS measurements on a Malvern Zetasizer Nano ZS90 (Malvern, UK). Thermogravimetric analysis was performed in a TGA SDT Q600 V8.3 Build 101 equipment (TA Instruments, New Castle, DE, USA) in an Argon gas environment (100.0 mL/min, 5 °C/min heating ramp). The goniometer used for contact angle and sessile drop measurements was an OCA35 by Dataphysics Instruments (Filderstadt, Germany). Viscometry was conducted using a DMA 4100 M density meter from Anton Paar (Albans, UK), with a Lovis 2000 ME micro-viscometer extension and a 0.59 mm stainless steel ball inside the capillary. The surface morphology and the thickness of the 1- and 2-layer thin films electrodes was examined via Atomic Force Microscopy, AFM, by means of a Solver Pro AFM (ΝΤ-ΜDT, Moscow, Russia)).

4.6. Electrochemical Measurements

The electrochemical characterization of inkjet-printed NiCo₂O₄ thin film electrodes was carried out by Cyclic Voltammetry (CV), Galvanostatic Charge–Discharge curves (GCD) and Electrochemical Impedance Spectroscopy (EIS) using a PGSTAT302N (Autolab, Metrohm, Herisau, Switzerland) electrochemical workstation. All the experiments were performed using fresh aqueous 1 M NaOH electrolyte. CV experiments were carried out in the 2–50 mV s⁻¹ potential sweep rate range; EIS measurements were performed in the frequency range of 10⁵–0.1 Hz with an AC amplitude of 10 mV, while GCD experiments were carried out by chronopotentiometry at constant charge/discharge current densities (in the 0.2–2 mA cm⁻² range and with voltage cut-off values as determined by the potential window of interest, based on CV experiments). Stability tests were carried out at ca 0.5 mA cm⁻² for 10,000 charge–discharge cycles. A three-electrode system consisting of platinum as a counter electrode, a SCE as a reference and NiCo₂O₄/ITO samples of different loadings (1 and 2 layers; 1L and 2L; 0.017 and 0.035 mg cm⁻²) was the working electrode. The area of deposited film in contact with the electrolyte was 0.5 cm². A Luggin capillary, whose tip has been at a distance of 1 mm from the surface of the working electrode, was used to minimize the IR drop errors in the electrolyte solution.