1 Introduction

The challenge of green hydrogen obtained through water electrolysis using renewable energy lies in its production cost. The production cost is divided into capital expenditure (Capex) and operating costs (Opex) [1]. The former includes costs for electrode materials, cell components, electrolytes, and other components that make up the electrolysis cell. Among these, the electrode catalysts represent a significant proportion. The latter includes expenses such as electricity costs and replacement costs for components due to degradation of electrodes and electrolytes. To reduce Opex, it is necessary to suppress power consumption by lowering the reaction overpotential and to ensure durability to maintain performance over the long term. Since electrodes and electrode catalysts are the most prone to degradation, it is evident that low-cost, highly active, and durable electrode catalysts are required from both Capex and Opex perspectives.

For the simultaneous achievement of reaction overpotential reduction and durability, platinum (Pt) is used as the electrode catalyst for the hydrogen evolution reaction (HER) [2], and IrO2 is used as the catalyst for the oxygen evolution reaction (OER) [3, 4]. However, these are rare metals with resource constraints and high costs, leading to an increase in Capex. Therefore, many alternative electrode catalysts are under research and development. For example, Ni and its alloys, which possess both conductivity and catalytic activity, are used as electrode catalysts for both poles [5,6,7]. However, Ni is also expensive among transition metals, and there is a need for a transition to more cost-effective and high-performance catalysts.

Among the options being considered are carbon alloys doped with hetero elements in a carbon network surface and metal complex catalysts [8,9,10]. The formation of complexes with carbon and organic compounds reduces the amount of metal used, resulting in low-cost, high-performance electrode catalysts. Carbon alloy electrocatalysts incorporating heteroatoms such as nitrogen and transition metals like iron into carbon frameworks are anticipated as alternatives to platinum group metals (PGMs) for HER catalysis. However, the synthesis of carbon alloys requires high-temperature annealing processes for heteroatoms and metal complexes, involving intricate carburization reactions, posing challenges such as difficulty in structural control.

In recent years, metal phthalocyanine-based catalysts have gained attention for showing diverse catalytic performances based on molecular structure design and the type of metal [11,12,13,14]. Mainly using Co, Ni, and Cu as metals and supporting them on carbon, studies for HER catalysts have been conducted and attempts have also been made to adjust the peripheral structure to modulate interactions.

On the other hand, we developed azaphthalocyanine unimolecular layer (AZUL) catalysts, where metal azaphthalocyanine (AzPc), an analogue of metal Pc, is supported on carbon, as oxygen reduction reaction (ORR) catalysts, and found them to exhibit very high activity and durability [15, 16]. We are also investigating the impact of changing metals and substitution site of nitrogen at different positions on catalytic activity from both experimental and theoretical perspectives [17].

Based on those knowledges of ORR catalysts, this paper discusses the performance of HER catalysts and their application as electrode catalysts for water electrolysis. After evaluating the HER catalytic performance of AZUL catalysts, which use transition metals such as Fe, Co, Ni, and Cu as central metals supported on porous carbon (Fig. 1), by rotating ring-disc electrode (RRDE), we applied the optimal combination of those rare-metal-free HER catalysts by coating gas diffusion electrodes and evaluated the water electrolysis overpotential by using alkaline exchange membrane water electrolysis (AEMWE).

Fig. 1
figure 1

Schematic illustration for preparation of AZUL electrocatalysts

Full size image

2 Experimental

2.1 Materials

Metal azaphthalocyanine molecules (FeAzPc-4N, CoAzPc-4N, NiAzPc-4N, and CuAzPc-4N) are synthesized according to literatures and used stocked samples [15, 17]. 2% Nafion® dispersion was purchased from Sigma-Aldrich, US. Ketjen Black (KB, EC600JD) was purchased from Lion Specialty Chemicals, Co. Ltd., Tokyo. 20% Nafion® dispersion (DE2021, CS type), Dimethyl sulfoxide (DMSO), isopropanol (IPA) and 1.0 M KOH aq. were purchased from Fujifilm Wako Pure Chemical Corporation, Osaka, Japan. Anion exchange membranes (AEMs) including Fumasep® (FAA-3-50, FUMATECH BWT, GmbH, Germany) and Sustainion® (X37-50 Grade RT, Dioxide Materials, Boca Raton, FL, USA) were purchased from Fuel Cell Store.

2.2 Preparation of electrocatalysts

First, 30 mg of Ketjen Black (KB) was dispersed in a DMSO solution of 0.1 mgl/ml metal-AzPc. The dispersion was sonicated with a homogenizer for 5 min and then suction-filtered to collect the samples. The samples were washed three times each with methanol and chloroform and then dried in vacuo.

2.3 RRDE

The HER performance was evaluated by linear sweep voltammetry (LSV) measurements using a potentiostat (2325, BAS, Japan) according to a previous study. For each sample, a catalyst ink was prepared by dispersing the catalyst (0.82 mg) in a solution (1 ml) consisting of Nafion® (6 μl; 527084, Sigma-Aldrich), isopropanol (334 μl), and water (84 μl) with sonication for 5 min. The catalyst ink (20 μl) was then cast onto the glassy carbon (GC) insert (BAS, Japan) of a rotating ring-disk electrode (RRDE; 4 mm diameter, BAS, Japan) and dried. The catalyst loading on the electrode was 300 μg/cm2. A Pt wire and a Hg/HgO electrode were inserted into the electrolyte as the counter and reference electrodes, respectively. A 1.0 M KOH solution bubbled with N2 for 30 min was used as the electrolyte. The potential vs Hg/HgO was converted to the reversible hydrogen electrode (RHE) scale using the following equation:

$${\text{E}}\left( {\text{vs RHE}} \right) \, = {\text{ E}}\left( {{\text{vs Hg}}/{\text{HgO}}} \right) \, + \, 0.0{98 } + \, 0.0{\text{59 V}} \cdot {\text{pH}}$$

To avoid overestimation of HER performance due to deposition of Pt from counter electrode, we evaluated the effect of counter electrode types with using a Pt wire and a glassy carbon (GC). In the three-electrode evaluation using the carbon sheet shown in Fig. 4b, either a Pt wire or a GC electrode was employed as the counter electrodes the reference electrode was Hg/HgO, and a carbon paper coated with FeAzPc-4N @KB as the working electrode was employed. The LSV curves for HER were repeatedly measured. The results shown in Figure Sx(a) demonstrate consistent LSV curves in both cases, proving that LSV curves with using the Pt wire and the GC are identical, that means that Pt dissolution has not occurred, and there is no effect of using Pt wire in this experiment.

Furthermore, when GC was used as the counter electrode, it was observed that the electrolyte turned brown over the course of the reaction (See supporting information, S4). It has been reported in the literature that using carbon as the counter electrode can contaminate the electrolyte and the working electrode due to the oxidation degradation of the carbon [18]. In selecting a counter electrode for alkaline conditions, it is necessary to be more cautious about the oxidation degradation of carbon than about the dissolution of Pt. Therefore, in our research, we used a Pt wire as the counter electrode, which consistently yielded reliable data without contaminating the electrolyte.

Electrochemical surface area (ECSA) values were estimated according to the method shown in the supplementary information, S3.

2.4 LSV measurements with sheet samples

A catalyst ink was prepared by adding catalyst (500 mg) and 20% Nafion® dispersion (DE2021, CS type, Fujifilm Wako Pure Chemical Corporation, Osaka, Japan) into a mixture of IPA (37.9 ml) and water (9.5 ml), followed by ball milling (PULVERISETTE 6, Fritsch GmbH, Idar-Oberstein, Germany). Subsequently, this ink was spray-coated (SimCoat, Sono-Tek Corporation, Milton, NY, USA) onto a carbon paper (Toray Paper TGP-H-120) substrate to fabricate a catalyst-coated sheet. The surface and cross-section morphologies were observed by using scanning electron microscopy (S-5200, Hitachi High-Tech, Co., Ltd., Tokyo, Japan) and electron probe micro analyzer (EPMA, JXA-8530F, JEOL, Tokyo, Japan). The sheet was then precisely sectioned into 5 mm widths and immersed to a depth of 5 mm in a 1.0 M KOH aq. Under these conditions, the carbon sheet served as the cathode, interfaced with a potentiostat, while the remaining conditions were kept consistent with those of RRDE measurements.

2.5 AEMWE

AEMWE was performed by using an electrolyzer cell (Complete 5 cm2 AEM Water Elerctrolyzer, Dioxide Materials, Boca Raton, FL, USA). Thin Ni-foam was used as an anode, in which oxygen evolution reaction (OER) occurred. Prepared elecrocatalysts were spray coated onto carbon papers to prepare cathodes. The adsorption of metal-AzPcs onto KB was confirmed by using X-ray photoelectron spectroscopy (XPS, PHI5000 VersaProbe II, Ulvac-Phai, Inc., Chigasaki, Japan). Two types of AEMs, Fumasep® (FAA-3-50, FUMATECH BWT, GmbH, Germany) and Sustainion® (X37-50 Grade RT, Dioxide Materials, Boca Raton, FL, USA), were employed and sandwiched with the anode and the cathode. After setting up the WE cell, I–V polarization curves of WE were measured by using a potentiostat (PARSTAT MC, AMETEK, Berwyn, PA, USA) with circulating 1 M KOH aq. with its flow rate of 0.3 ml/min. iR drop was estimated from galvanostatic electrochemical impedance spectroscopy (GEIS) data ranging from 1 Hz to 100 kHz with 10 mA RMS. From the EIS spectra resistivity of solution (Rsol) was estimated and actual over potential (ηcell) was calculated by following equation;

$$\eta_{cell} = I\,{\text{X}}\,R_{cell}$$

where I was current density and Rcell = Rsol, respectively.

Chronopotentiometric durability test was performed at 200 mA/cm2 for 24 h at same condition.

Faradaic efficiency (FE) values of WE were calculated by following procedures. Detail experimental set up was shown in the supplementary information, S1. A tank is installed in the electrolyte flow path to separate the electrolyte and gas, and it is connected to a soap film type gas flow meter. The flow rate is set at 0.3 ml/min, and the cell temperature is set to 30 °C. A current of 5A (1A/cm2) is applied to the water electrolysis cell, and gas generation is confirmed. Continuously applying 5A, the time required for the soap film to rise by 10 ml is measured using the soap film type gas flow meter (GL Science, Tokyo, Japan). The measurement results are then calculated in ml/s. From the density of hydrogen gas (0.082 g/L @25 °C) and the molecular weight of H2 (2.0158 g/mol), mol/s is calculated. The number of electrons at 5A applied is 5/96485 mol/s. Considering that in the HER (Hydrogen Evolution Reaction), 1 mol of H2 is produced for every 2 mol of electrons, the theoretical hydrogen generation rate is 5/(96485X2) mol/s. The value obtained from the gas flow meter is divided by the theoretical value to determine the Faraday efficiency. The same procedure is applied for oxygen. The density of oxygen gas (0.131 g/L @25 °C) and the molecular weight of O2 (32.0 g/mol) yield a theoretical oxygen generation rate of 5/(96485X4) mol/sec. (See supplementary information, S2).

3 Results and discussion

The formation of the AZUL catalyst layer on the electrode was confirmed by cross-sectional SEM, EPMA and XPS. Figure 2 shows cross-sectional SEM and EPMA elemental mapping images of prepared electrodes. From the topographical and cross-sectional SEM images, uniform distribution of Pt/C and AZUL catalysts, which are comprised of catalyst molecules and KB, are uniformly coated onto carbon papers except for CuAzPc-4N case (Fig. 2(a-1)-2(e-1)). In the CuAzPc-4N case, needle-like aggregates attributed to crystals of CuAzPc-4N are observed. This indicates that CuAzPc-4N has high crystallization properties during adsorption onto KB. Figure 2(a-2)-2(e-2) show elemental mapping of carbon and the signals are clearly observed along to the electrodes. Figure 2(a-3)-2(e-3) show elemental mapping images of metals involved in azaphthalocyanine macrocycles and Pt, respectively. Those results indicate that catalyst molecules and Pt nanoparticles were uniformly distributed on the surface of carbon paper electrodes and the catalyst layer thicknesses were around 100 µm in each case. Figure 2(a-4)-2(e-4) show composite images of elemental mapping and cross-sectional SEM images. These images prove the location of catalysts.

Fig. 2
figure 2

Scanning electron micrographs (a-1-e-1), cross-sectional scanning electron (SE) images (a-2-e-2), elemental mapping images of respective metals measured by EPMA (a-3-e-3), and merged images of SE and elemental mapping (a-4-e-4), respectively

Full size image

Figure 3 shows the XPS spectra of electrode with each electrocatalyst. In addition to peaks from carbon and binder-derived fluorine in the GDE in the wide scan spectra, peaks corresponding to metal elements derived from the AZUL catalyst were observed in the narrow scan spectra, confirming the formation of an AZUL catalyst layer consisting of carbon supported on the electrode. Normally, metal complex shows peaks attributed to 2p1/2 and 2p3/2 of each metal atom. Those peaks found in 710–730 eV (Fe), 780 eV and 795 eV (Co), 855 eV and 874 eV (Ni), 933 eV and 954 eV (Cu), respectively. Those spectra strongly support formation of catalyst layers on the surface of GDE.

Fig. 3
figure 3

Wide scan XPS spectra of sheet electrode after coating with AZUL catalysts of FeAzPc-4N (a), CoAzPc-4N (b), NiAzPc-4N (c), and CuAzPc-4N (d), respectively. The inset spectrum of each plot is a narrow scan XPS spectra of respective metal elements

Full size image

Figure 4a shows the LSV polarization curves of each electrocatalyst measured by RRDE. The reference experiment using Pt/C demonstrates high HER performance, with current rising nearly from around 0 V vs RHE. On the other hand, it became apparent that the catalytic performance varies depending on the central metal in the AZUL catalysts. Table 1 shows the reaction overpotentials at 10 mA/cm2 and 0.5 A/cm2, which obtained from three-electrode evaluation results obtained from sheet samples shown in Fig. 4b, for each electrocatalyst. Among the AZUL catalysts, those supported with CoAzPc-4N exhibited the highest performance, with ηj = 10 mA/cm2 = 378 mV, followed by FeAzPc-4N (ηj = 10 mA/cm2 = 408 mV), NiAzPc-4N (ηj = 10 mA/cm2 = 508 mV), and CuAzPc-4N (ηj = 10 mA/cm2 = 547 mV). The catalyst activity of metal atoms for HER has been evaluated experimental and theoretically and its important parameter is adsorption energy between metal atom and hydrogen. From that previous research, volcano plot for HER has been reported [19, 20]. The volcano plot for the HER of single atom catalysts reports that Fe, Co, and Ni exhibit almost equivalent adsorption energies and catalytic activities, whereas only Cu shows a slightly smaller energy. This result is consistent even in azaphthalocyanine systems with the same molecular structure, where only CuAzPc-4N exhibited a significant overpotential due to low catalytic activity itself and crystalline aggregate structures.

Fig. 4
figure 4

LSV curves of HER on electrocatalysts measured in 1.0 M KOHaq by RRDE (a) at its rotating speed of 1600 rpm and sheet samples (b), respectively. c shows Tafel slope obtained from respective LSV curves

Full size image
Table 1 Over potentials measured in various electrochemical methods and FE of H2
Full size table

Figure 4c shows the Tafel Slopes of each catalyst. The reaction steps of HER are divided into following three steps [21];

$$\begin{array}{*{20}c} {{\text{H}}_{{3}} {\text{O}}^{ + } + {\text{ e}}^{ - } \to {\text{H}}_{{{\text{ads}}}} + {\text{ H}}_{{2}} {\text{O}}} & {\left( {\text{Volmer step}} \right)} \\ \end{array}$$
$$\begin{array}{*{20}c} {{\text{H}}_{{{\text{ads}}}} + {\text{ H}}_{{3}} {\text{O}}^{ + } + {\text{ e}}^{ - } \to {\text{H}}_{{2}} + {\text{ H}}_{{2}} {\text{O}}} & {\left( {\text{Heyrovsky step}} \right)} \\ \end{array}$$
$$\begin{array}{*{20}c} {{\text{H}}_{{{\text{ads}}}} + {\text{ H}}_{{{\text{ads}}}} \to {\text{H}}_{{2}} } & {\left( {\text{Tafel step}} \right)} \\ \end{array}$$

Pt/C shows low Tafel slope (55.6 mV/dec), which attributed to the Heyrovsky or Tafel step. On the other hand, AZUL catalysts had higher Tafel slope values as shown in Fig. 4c and Table 1, that indicates the Volmer step is the rate limiting step of HER with using AZUL catalysts. Especially, the Tafel slope value of CuAzPc-4N is much higher than the other three molecular catalysts, this result strongly agree with the low adsorption energy of hydrogen atom onto single atom Cu [16].

While examples evaluating HER catalytic performance by mixing phthalocyanine-based molecules with carbon are relatively rare, a CoPc introducing alkylpyrrolidone groups to the peripheral part mixed with oxidized graphene [22], or (n-OctSe)8-CuPc [23], both requiring multistep synthesis, exhibits a similar alkaline condition with ηj = 10 mA/cm2 = c. a. 370 mV. Therefore, the AZUL catalyst, especially using FeAzPc-4N and CoAzPc-4N, which can be synthesized in one step, has very high activity as a molecular catalyst without mixing PGM catalysts [24,25,26,27]. From the values of ECSA, the surface areas of metal AzPc-4N catalysts had higher surface areas (See supporting information, S3). This also contributes to high activity of AZUL electrocatalysts.

There are only small numbers of papers have been reported about AEMWE with using metal-Pcs as electrocatalysts for HER. We developed an AEMWE cell as shown in Fig. 5 and evaluated the performance of cathode. Figure 6a and b show the I-V polarization curves of AEMWE using cathodes with various catalyst electrodes using two types of AEMs, Fumasep® (Fig. 6a) and Sustainion® (Fig. 6b), with circulating 1.0 M KOHaq. at room temperature. The dashed lines show original I–V polarization curves, and solid lines show iR-corrected I–V polarization curves with using respective electrocatalysts for hydrogen electrodes. The iR-corrected voltages were calculated based on EIS data. When using Fumasep®, overall reaction overpotentials are high, with overpotentials at 1.0 A/cm2 being 1.94 V for Pt/C, followed by 2.11 V for CoAzPc-4N, 2.21 V for FeAzPc-4N, 2.62 V for NiAzPc-4N, and 2.22 V for CuAzPc-4N, respectively (Fig. 6c). The most high-performance case except for Pt/C was WE with using CoAzPc-4N. In contrast, when using Sustainion®, reaction overpotentials at 1.0 A/cm2 were 1.87 V for Pt/C, followed by 2.19 V for CoAzPc-4N, 2.02 V for FeAzPc-4N, 2.18 V for NiAzPc-4N, and 2.47 V for CuAzPc-4N, respectively (Fig. 6d). Most of the overpotentials of WE with using Sustainion® were lower than those of using Fumasep® due to the resistivity of Sustainion® (18–25 m Ω cm2) [28] was much lower than that of Fumasep® (1.2–2.0 Ω cm2) [29], it is reasonable to assume that the overvoltage of WE using Fumasep® is greater in that of sustaininon® [30]. However, it is noteworthy that the overvoltage ordinates varied greatly depending on the combination of AEM membrane and catalyst, for example, CoAzPc-4N case was not so much different in both cases, and the relationship between using these two AEMs were inverted in the CuAzPc-4N case.

Fig. 5
figure 5

Schematic illustration of WE cell

Full size image
Fig. 6
figure 6

I-V polarization curves of AEMWE using cathodes with various catalyst electrodes using two types of AEMs, Fumasep® (a) and Sustainion® (b), with circulating 1.0 M KOH aq. at room temperature, and comparison of overpotentials at 1.0 A/cm2 with using Fumasep® (c) and Sustainion® (d), respectively. The dashed lines show original I-V polarization curves, and solid lines show iR-corrected I-V polarization curves with using respective electrocatalysts for hydrogen electrodes in (a) and (b)

Full size image

Generally, the HER activity of electrocatalysts is measured using RRDE. However, the results of AEMWE did not show straight forward correlation with the electrocatalyst activity measured by RRDE. When using Fumasep®, CoAzPc-4N exhibited the lowest overpotential, followed by FeAzPc-4N. On the other hand, when using Sustainion®, the overpotential of CoAzPc-4N remained almost unchanged, but a significant reduction in overpotential was observed with FeAzPc-4N and NiAzPc-4N. On the other hand, over potential of AEMWE with using CuAzPc-4N exhibited increasing overpotentials with using Sustainion®. This suggests that the catalyst activity varies depending on the type of AEM. Fumasep® is an AEM comprised of quaternary ammonium functionalised polysulfone [31, 32], and Sustainion® is an AEM that utilizes imidazolium in its side chain [28, 33]. Generally, pyridine and imidazolium groups are known to coordinate with the central metal of metal phthalocyanine compounds, particularly iron and nickel, and it has been reported that these ligands coordinated to metals enhance catalytic activity [34, 35]. This result is the first example showing significant interaction between AEMs and molecular catalysts such as metal-AzPc. It is expected that further detailed analysis will reveal the optimal combination of metal complex molecular catalysts and AEMs in AEMWE.

With the exception of the combination of Pt/C-Fumasep®, the Faraday efficiency (FE) based on hydrogen generation rate exceeded 90%, confirming the high efficiency of water electrolysis. Particularly, when using Sustainion®, an FE of nearly 95% was achieved. Therefore, from the perspectives of overpotential and FE, the combination of FeAzPc-4N and Sustainion® as an alternative catalyst to Pt/C demonstrates the most efficient hydrogen production.

Durability evaluation at high current density regime is a crucial parameter for practical application. Thus, we demonstrated voltage stability for over 24 h even at a relatively high current density of 200 mA/cm2 (Fig. 7). We investigated parameters from past literatures for catalysts composed of non-precious metal nanoparticles, non-metal catalysts, and metal phthalocyanines embedded on carbon similar in structure to our catalyst (see supporting information, S5). While many papers report reaction overpotentials, only a few have evaluated long-term durability with using carbon-based catalysts. Additionally, while some catalysts show very high activity with reaction overpotentials of only a few tens of mV, most measurements of durability at low current densities of around several tens of mA/cm2 lasted only a few hours. This indicates that the HER catalyst we developed possesses higher durability compared to previously reported non-precious metal HER catalysts.

Fig. 7
figure 7

Durability test of AEMWE with using FeAzPc-4N and Fumasep® at 200 mA/cm2 for 24 h

Full size image

4 Conclusion

We elucidated the HER catalytic activity of AZUL electrocatalysts comprising of different metal-AzPc and compared the overpotentials in AEMWE. Different central metal AZUL catalysts were fabricated, and their HER activities, similar to the case of single atom catalysts, showed that Cu with lower hydrogen adsorption energy had the lowest performance, while the others exhibited similar activities. While the overpotentials are higher compared to Pt/C, the system utilizing the highest-performing FeAzPc-4N with Sustainion® showed a low overpotential of 2.02 V at 1.0 A/cm2, even among the previous alternative catalysts. It is noteworthy that the AZUL catalysts are not using precious metals and even comparing with other transition metal electrocatalysts, metal atom usage is much lower because only the single transition metal atom in the azapthalocyanine macrocycle works as a catalytic site. It is expected that further investigation into the structure, loading amount of catalyst molecules, and interaction with AEMs will lead to realizing highly efficient AEMWE. Moreover, the interaction between metal complex catalysts and AEMs discovered in this study provides deep insights for the design of future AEMWE systems.