Introduction

Electrochemical water splitting to produce H2 and O2 is one of the key energy conversion processes to store renewable energy in the form of chemical fuel [1]. A polymer electrolyte membrane (PEM) electrolyzer is considered a promising alternative energy storage system to commercial water-alkaline electrolyzers [2, 3]. PEM electrolyzers are more compact and efficient than alkaline electrolyzers and generate hydrogen with greater purity. PEM electrolyzers require the use of conductive rare metal oxide electrocatalysts such as IrO2 and RuO2 as the anode material, because these materials show high catalytic activity for the oxygen evolution reaction (OER) and excellent stability under the harsh acidic conditions for water electrolysis [4–11]. Since these materials are based on platinum-group metals, which are expensive and difficult to procure, extensive studies have been conducted to identify cost-effective, readily available, and robust alternatives, but without success.

In our quest for suitable OER catalysts, we focused on first-row transition metal (TM) titanium phosphides. TM phosphides have many interesting features such as high electrical conductivity at room temperature, higher stability in acidic and basic media as compared to pure metals, and much better cost-effectiveness compared to precious metals [12–14]. In addition, TM phosphides (e.g., TiP, VP, CrP, and TaP) have better corrosion resistance than TMs and transition metal oxides (TMOx) used in alkaline water media, such as Ni metal and Co3O4 spinel, which are good electrocatalysts but unstable in even weak acids [15]. For example, TiP, which is a good metallic conductor, exhibits extremely high corrosion resistance, as it dissolves only in hot aqua regia [16]. Another well-known example of the first-row TM is Fe, i.e., iron promoted with K2O, CaO, SiO2, and Al2O3, used in the Haber-Bosch process.

In the present research, we focused on TMTiP, which is composed of a 3d TM for catalytic activity and TiP for electronic conductivity and high corrosion resistivity as a potential electrocatalyst for electrochemical water splitting. For this type of compounds, synthesis and the crystal structural analysis for CrTiP has been reported [17]. There are some reports about the crystal structure [18] and about the phase diagram Fe-Ti-P system [19].

Experimental

Samples were fabricated by RF magnetron sputtering. Nominal compositions of the TMTiP (TM = Cr–Ni) targets shown in Table 1 were obtained by weighing the precursor powders and thoroughly mixing them for 15 min using an agate mortar and pestle.

Table 1 Composition of sputtering targets
Full size table

The powder mixture was then placed on a Cu saucer and compressed by hand to obtain the sputter target. The raw powders used were TiP (97 %, American Elements), CrP (98 %, American Elements), Mn2P (99 %, Kojundo Chemical Laboratory Co., Ltd), FeP (99 %, American Elements), Co2P (99 %, Kojundo Chemical Laboratory Co., Ltd), Ni2P (99 %, Kojundo Chemical Laboratory Co., Ltd), Ti (99.9 %, Kojundo Chemical Laboratory Co., Ltd), and Fe (99.9 %, Kojundo Chemical Laboratory Co., Ltd). TMTiP films were sputter deposited on glass substrates (EAGLE XG, Corning) and Ti plates for 75 min at 100 W using a sputtering system (Canon Anelva Corp.), where the chamber was evacuated to ca. 5 × 10−4 Pa, and then, Ar was flowed at 10 sccm. Before sputter deposition, the Ti plates were dipped in a mixture of 1 M HF + 2 M H2O2 + 3 g/L caffeine for 3.5 min to remove the surface oxide layer and obtain a clean surface by decomposing contaminants. IrO2 and RuO2 films were also synthesized using IrO2 and RuO2 powders (Wako Pure Chemical Industries, Ltd.) in a similar manner for the comparison of performance of TMTiP films with that of IrO2 and RuO2, because they have been well known as the electrocatalyst with the highest OER activity.

Structural analyses of the resultant films were conducted using X-ray diffraction (XRD; Ultima IV, Rigaku) with Cu Kα radiation. The composition of the sputter-deposited films was determined using Rutherford back-scattering spectroscopy (RBS) with a 2.0 MeV He+ ion beam (AN-2500, Nisshin High Voltage Ltd), or by electron probe microanalysis (EPMA; JXA-8200, Jeol Ltd).

The electrical resistivity of the films was measured using the standard four-probe technique (Laresta-GP MCP-T600, Mitsubishi Chemical Corporation). The film thickness was determined using a contact probe profilometer (Dektak3ST, Veeco Ltd).

Electrochemical measurements were conducted in a standard three-electrode electrochemical glass cell. Electrochemical behaviors in 0.5 M H2SO4 (pH = 0), 1 M Na2SO4 (pH = 6.8), 0.1 M K-Pi (pH = 6.8, 9.3), and 1 M KOH aqueous solution (pH = 14) were studied. Pt foil (10 × 70 × 2 = 1400 mm2) was used as the counter electrode, and a saturated Ag/AgCl electrode was used as the reference electrode. Working electrodes and electrolytes were fixed for all measurements. Prior to measurements, the electrolytes were deoxidized by N2 gas bubbling for at least 30 min.

Cyclic voltammetry (CV) curves were obtained using an electrochemical analysis system (HZ-5000, Hokuto Denko Ltd), where the potential was swept between −1 and 2 V (vs. Ag/AgCl) at a sweep rate of 50 mV/min and 200 mV/s. The CV curves were measured after we confirmed that they were stabilized over repeated potential sweeps. The corrosion potential E corr and corrosion current density i corr were determined as shown in Fig. 3b.

The oxidation current density i ox at 2 V (vs. Ag/AgCl) was also measured by chronoamperometry. Anode polarization curves were measured by sweeping the potential from 0 to 2 V (vs. Ag/AgCl) at a sweep rate of 50 mV/s.

Current density measurements were performed on the cobalt titanium phosphide (CoTiP), IrO2, and RuO2 sputter-deposited films on a Ti plate in 0.5 M H2SO4 aqueous solution (pH = 0) under potentiostatic conditions at 2 V (vs. Ag/AgCl), in order to clarify their short-time stability as an oxygen evolution electrode. The potential was measured as Ag/AgCl standard electrode, we show the experimental results using reversible hydrogen electrode (RHE) standard.

Results and Discussion

The thicknesses of the sputter-deposited films on glass substrates were between 150 and 400 nm. Figure 1 shows the XRD pattern for CoTiP sputter film fabricated on a glass substrate. No diffraction peaks were observed for the samples in the XRD patterns. The XRD patterns for all the TMTiP films sputter deposited on glass substrates were similar to that of CoTiP film, and it is considered all the filmes are amorphous.

Fig. 1
figure 1

XRD data for CoTiP film deposited on glass substrate

Full size image

Table 2 shows the representative compositions of the films determined from RBS and EPMA measurements. The film compositions were in good agreement with the nominal compositions of the targets, except for Ni0.68TiP1.04.

Table 2 Compositions of Ar sputter-deposited films and sputtering targets
Full size table

The resistivities of the films are shown in Fig. 2. All the films fabricated on glass substrates exhibited metallic conductivity, and the resistivity was on the order of 10−6 Ω m.

Fig. 2
figure 2

Electrical resistivities of sputter-deposited films

Full size image

Figure 3a, b illustrates the voltammograms of the CoTiP film measured in 0.5 M H2SO4. No remarkable electrochemical reaction other than the OER was observed. A sweep rate of 50 mV/min was chosen because the system is considered to change through a quasi-static process at this sweep rate. The CV curves of the other TMTiP films, except for MnTiP, were similar to those of CoTiP. Two log │i│ minima were observed (Fig. 3b). In the anodic sweep, the following electrochemical reactions are thought to occur. First, proton reduction up to ca. −0.1 V (vs. RHE) was observed (ideal onset voltage for proton reduction = 0 V vs. RHE), and then, an oxidized layer was formed on the CoTiP film up to ca.1.7 V (vs. RHE). The OER proceeded above ca. 1.9 V (vs. RHE). The minimum in the cathodic sweep profile was considered to correspond to the anodically oxidized surface layer on the CoTiP film. The cathodic sweep profile was similar to the anodic sweep profile, except for a complicated behavior observed between ca. −0.2 and ca. 0.7 V (vs. RHE). In this range, both proton reduction and reduction/recombination of the surface oxide layer are thought to proceed. Therefore, the minima in the anodic and cathodic sweep profile are considered to correspond to the anodically oxidized surface layer on the CoTiP film.

Fig. 3
figure 3

Cyclic voltammetry curve of CoTiP in 0.5 M H2SO4 (sweep rate 50 mV/min). The black broken line and red dotted line correspond to the cathodic and anodic sweep profiles, respectively. a Linear vertical scale. b The logarithm of the absolute value of current density is plotted. E corr and i corr were determined as shown in the inset of (b), i.e., similar to Tafel extrapolation

Full size image

E corr and i corr for the TMTiP films, which are indices of chemical stability, measured in 0.5 M H2SO4 aqueous solution, were determined from the anodic sweep profile (inset of Fig. 3b). The measured values are plotted in Fig. 4a, b. The E corr values for TMTiP (TM = Cr–Cr0.33Mn0.67) were as low as −0.25 V (vs. RHE), while that for the MnTiP film could not be measured because the film dissolved in H2SO4. E corr gradually increased as the TM was changed from Cr to Co0.33Ni0.67, reached a maximum of 0.089 V (vs. RHE) for Co0.33Ni0.67TiP, and then decreased to −0.08 V (vs. RHE) for NiTiP (Fig. 4a). i corr gradually decreased as the TM was changed from Ni to Fe0.33Co0.67, with minima on the order of ca. 10−7 A/cm2 in the vicinity of Co, and then increased for Co0.33Ni0.0.67TiP and NiTiP (Fig. 4b).

Fig. 4
figure 4

Corrosion potential E corr and corrosion current density i corr of sputter-deposited TMTiP film in 0.5 M H2SO4, determined as indicated in the inset of Fig. 3b. a E corr. b i corr

Full size image

Both sets of results imply that the chemical stability decreases as we move from Cr to Fe, reaches a minimum in the vicinity of the composition with TM = Co, and then increases from Co to Ni. Thus, a material that has both high corrosion resistance and metallic conductivity was obtained in the vicinity of the composition CoTiP.

For stability and OER activity measurement, the oxidation current density i ox (at 2.199 V vs. RHE) in 0.5 M H2SO4 aqueous solution (pH = 0) was measured with respect to systematic changes in the TMTiP composition (TM = Cr to Ni). The electrochemical behavior was elucidated at the scan rate of 50 mV/min. There was no difference in the current density at 2.199 V vs. RHE between the scan rates of 50 mV/min and 200 mV/s. Moreover, no difference or hysteresis was observed between the anodic and cathodic profiles in the OER region. Therefore, the data corresponding to the sweep rate of 200 mV/s are shown in Fig. 5 as an index of the OER activity. The maximum i ox was obtained for the composition CoTiP. Although O2 detection was not performed, almost all the i ox was thought to be consumed for O2 generation via the decomposition of water, for the following reasons. First, a large number of gas bubbles were observed to form on the CoTiP working electrode under a high i ox. Second, the thickness of the film was only ca. 400 nm; hence, if i ox had been due to the dissolution of the CoTiP film, it would have immediately decreased in value and no oxidation current would have been detected thereafter. Finally, even after 0.5 h of i ox, the surface of the CoTiP film exhibited no apparent change. It should be noted that i ox, which is an index of the OER catalytic activity of TMTiP films, gave the “volcano” as the TM changed from Cr to Ni, i.e., with a change in the 3d electron number of the TM. Similar observations have been reported previously; that is, d electron filling in the 3d transition metal oxide is essential for electrocatalytic activity in the OER or oxygen reduction reaction (ORR) [20–23]. In those studies, material design was carried out based on electronic structure calculations, and materials with optimized electrocatalytic activity were obtained. Similarly, in our case, we expect to obtain OER catalysts with compositions other than CoTiP or new materials with high OER activity by computational design based on the data obtained in detailed experimental analyses, such as TMTiP surface structure observation and surface electronic structure determination [24–27].

Fig. 5
figure 5

Oxidation current density, i ox, for TMTiP sputter-deposited films at 2.199 V vs. RHE in 0.5 M H2SO4 (sweep rate 200 mV/s)

Full size image

Figure 6a, b shows the polarization curves for the Pt foil and sputter-deposited CoTiP, IrO2, and RuO2 films on Ti plates in 0.5 M H2SO4 (pH = 0) and 1 M KOH aqueous solution (pH = 14). The small peak around 1.4 V (vs. RHE) in the polarization curve of CoTiP (Fig. 6b) may be attributed to the formation of the surface oxide layer. From Fig. 6, it is clear that the OER activity in 0.5 M H2SO4 follows the trend IrO2 ≈ RuO2 > Pt > CoTiP, while that in 1 M KOH shows the trend IrO2 ≈ RuO2 ≥ CoTiP > Pt. The overpotential η at 1 mA cm−2 is plotted in Fig. 7 as a function of pH. We cannot simply discuss the difference in OER activity between acid and alkaline media because data corrections, for example, elimination of the effect of resistance of the electrolyte and the electrode itself, are needed to discuss the intrinsic features of the OER activity. In addition, careful analysis of the i–v behavior in the Tafel region, where the current density is very low (≤10−4 A cm−2), is essential to elucidate the OER mechanism. Hence, from Figs. 6 and 7, we can only discern that the OER activity of CoTiP is comparable to that of Pt but far inferior to the activity of the state-of-the-art IrO2 and RuO2.

Fig. 6
figure 6

Polarization curves for CoTiP, IrO2, and RuO2 films deposited on Ti plates and for Pt foil (sweep rate 50 mV/s). a 0.5 M H2SO4, b 1 M KOH

Full size image
Fig. 7
figure 7

Relationship between oxygen overpotential η and pH. The overpotential η corresponds to the current density, which is 1 mA cm−2. Red and blue asterisks indicate the reference data corresponding to the results measured for 1 M Na2SO4

Full size image

In order to examine the very short-time stability of the CoTiP electrode, continuous electrochemical water oxidation was carried out under potentiostatic conditions at 2.199 V vs. RHE, i.e., the same corrosive environment. The values of the normalized current density i(t)/i(0), where i(t) is the electrical current density at time t (s) and i(0) is that at time 0 (s), are shown in Fig. 8. The actual values of i(0) and i(2000) are also plotted in Fig. 9. In these measurements, a potential of 2.199 V vs. RHE was simultaneously applied to the electrodes maintained at the near open-circuit potential. Therefore, the effects of the capacitive current consumed for the formation of the surface oxide layer and the electric double layer, which are related to i(0), must be taken into consideration. As indicated in Fig. 9, the i(0) values for the IrO2 and RuO2 electrodes were so high that the capacitive current could be neglected. For the CoTiP electrode, the i(0) value was low and hence, the effect of the capacitive current should be taken into account. However, the i(0) value was comparable to i(2000). At t = 2000 s, the capacitive current is considered to be nearly equal to 0. The experimental result that i(0) ≈ i(2000) for CoTiP indicated that the effect of the capacitive current is small enough to be neglected. Hence, it was concluded that the effect of the capacitive current can be neglected for all the samples. We also show the i cv values measured for the anode polarization curves at 2.199 V vs. RHE. The i cv value is almost equal to i(0), implying that the effect of the capacitive current is small. Consequently, as can be seen in Fig. 8, the very short-time stability of the CoTiP film was higher than that of the RuO2 film and comparable to that of the IrO2 film sputter-deposited on the Ti plate.

Fig. 8
figure 8

Current–time curves for electrochemical water oxidation in 0.5 M H2SO4 (pH = 0) under potentiostatic conditions at 2 V vs. SSE using CoTiP, IrO2, and RuO2 films sputter deposited on a Ti plate

Full size image
Fig. 9
figure 9

Actual i(0) and i(2000) data in 0.5 M H2SO4 (pH = 0) at 2.199 V for CoTiP, IrO2, and RuO2 films sputter deposited on Ti plate. i cv, the current density for anode polarization measurements, is plotted for the purpose of reference

Full size image

From the abovementioned observations, it is confirmed that CoTiP is a potential candidate for oxygen evolution electrode in water oxidation, in both acidic and alkaline media.

Conclusion

TMTiP films were fabricated on glass substrates and Ti plates by RF Ar sputtering, and the electrochemical properties of the electrodes were systematically investigated when the TM was changed from Cr to Ni. XRD measurements showed that all the electrodes were amorphous. RBS and EPMA analyses indicated that the film compositions agreed well with the sputter target compositions. The TMTiP films exhibited metallic resistivity on the order of 10−6 Ω m. Films with a nominal composition of TMTiP (TM = Fe–Ni), especially in the region of the CoTiP composition, showed excellent electrochemical resistance to acids, while maintaining high electrical conductivity. A volcano relationship was clearly observed for the i ox (electrical current density at 2.199 V vs. RHE) values of the TMTiP films, i.e., the OER activity. The electrocatalytic activity of the electrodes for the OER reached a maximum at the CoTiP composition, with excellent corrosion resistance over a wide pH range (0–14). The OER activity of CoTiP was comparable to that of Pt but far inferior to that of state-of-the-art IrO2 and RuO2.