Oleylamine (OAm) (>70%), 1-octadecene (ODE, 90%), platinum II) acetylacetonate (Pt (acac)₂, 99%), cobalt (II) acetylacetonate (Co(acac)₂, 97%), borane-tert-butylamine (BTB, 97%), hexane (99%), ethanol (99%), acetone (97%), phosphorus pentoxide (P₂O₅, 95%), potassium peroxodisulfate (K₂S₂O₈, 95%), potassium permanganate (KMnO₄, 99%), hydrogen peroxide (H₂O₂, 30%), sodium nitrate (NaNO₃, 99%), sulfuric acid (H₂SO₄, 98%), dimethylformamide (DMF, >99%) and Ludox^® HS-40 were purchased from Sigma-Aldrich^® and used as received. Natural graphite flakes (average particle size: 325 mesh) and guanidine hydrochloride (GndCI, 98%) were purchased from Alfa-Aesar. Ammonium hydrogen difluoride (NH₄HF₂, 98.5%) was purchased from Fluka. Commercial Ketjenblack carbon was purchased from Aksa Nobel. Deionized water was distilled using a Milli-Q water purification system.

All transmission electron microscope (TEM) images were recorded by transmission electron microscope (TEM, Jem-1400 plus electron microscopy, 120 kV). X-ray diffraction (XRD) patterns were recorded on a PANalytical Empyrean diffractometer with Cu-Kα radiation (40 kV, 15 mA, 1.54051 Å) over a 2θ range from 10° to 90° at room temperature. Elemental analysis measurements were carried out by inductively coupled plasma-mass spectroscopy (ICP-MS, Agilent Technologies 7,700) after each sample was completely dissolved in aqua-regia (HCl/HNO₃: 3/1 vol. ratio).

Monodisperse Co₄₅Pt₅₅ alloy NPs were synthesized by using a surfactant-assisted solvothermal method comprising the decomposition and reduction of metal precursors in a solvent-surfactant mixture at a relatively high temperature. In a synthesis of Co₄₅Pt₅₅ NPs, under an inert atmosphere, 0.2 g of borane-tert-butylamine (BTB), 3 mL of OAm, and 7 mL of ODE were mixed vigorously in a four-necked reactor. Then, a mixture of 0.3 mmol of Pt (acac)₂ and 0.2 mmol of Co(acac)₂ was dissolved in 3 mL of OAm at room temperature and injected in the same reactor at 100°C. The reaction mixture was kept at this temperature for an hour and then cooled down to 40°C. The yielded NPs were separated by centrifugation (8,500 rpm, 12 min) after acetone addition (40 mL). This procedure was repeated with ethanol. The obtained NPs were dispersed in hexane for further use.

Reduced graphene oxide (rGO) was synthesized by using a well-established two-step wet-chemical procedure starting from natural graphite flakes. In the first step, graphite oxide was synthesized by using the modified Hummers method (Hummers and Offeman, 1958). In the second step, graphite oxide was exfoliated to graphene oxide (GO) in DMF (1 mg mL^-1) by using an ultrasonic bath, and then the resultant dispersion was refluxed for 6 h. At the end of the refluxing process, GO was reduced to rGO, and floating black rGO sheets were separated from the solution by filtration (Metin et al., 2014).

For the synthesis of mpg-CN, we used our established hard-templating method published elsewhere (Aksoy and Metin, 2020). The details of the mpg-CN synthesis procedure and its detailed structural characterization can be found in our recent report (Erdogan et al., 2016).

To use the CoPt alloy NPs as catalysts for OER, they were assembled on rGO, mpg-CN, and commercial Ketjenblack (KB) supports by using the liquid self-assembly method that has been utilized by our group for the synthesis of many electrocatalysts (Şener et al., 2015; Martins et al., 2017; Yılmaz et al., 2019). In a general procedure, 40.0 mg of rGO, mpg-CN, or KB was exfoliated in ethanol by sonicating the solution in an ultrasonic bath for 30 min. Next, 20 mg of the NPs were dispersed in 30 mL of hexane, and this dispersion was mixed with the rGO, mpg-CN, or commercial Ketjenblack dispersion in ethanol. The resultant mixture was sonicated for 2 h to ensure the assembly of the NPs onto the carbon-based support materials. In the final step, the mixture was centrifuged at 7,500 rpm for 10 min and the separated catalyst was washed with ethanol. The alloy composition and the metal contents of the catalyst were determined by ICP-MS.

All electrochemical measurements were carried out in a 1 M KOH aqueous solution using Ivium V01107 potentiostat/galvanostat in a three-electrode system with graphite rod as the counter electrode, saturated calomel electrode (SCE) as the reference, and CoPt/X (X = rGO, mpg-CN, KB) as the working electrode. All potentials within the paper are given versus the reversible hydrogen electrode (RHE).

Working electrodes were prepared by pipetting 10 μL of catalytic ink onto a glassy carbon tip (0.945 cm²) and drying it at 100°C for 8 h. Catalytic inks were composed of 5 mg of each electrocatalyst, including CoPt/rGO (6.4 wt% Pt), CoPt/mpg-CN (7.2 wt% Pt), CoPt/KB (1.44 wt% Pt), or commercial Co₂₅Pt₇₅ (Tanaka Kikinzoku Kogyo Co. Ltd., 46.7 wt% Pt) on carbon, in 125 μL of 2 wt% polyvinylidene difluoride in N-methyl-2-pyrrolidone solution, which were then ultrasonically mixed for 30 min. The final loadings of Pt on the electrode surface are given in Table 1. Accordingly, considering the Co₄₅Pt₅₅ atomic ratio in the alloy, the carbon-to-catalyst mass ratios were 92.0%, 91.0%, and 98.2% for CoPt/rGO, CoPt/mpg-CN, and CoPt/KB, respectively.

OER studies of the three electrocatalysts were done by LSV in 1 M KOH from the open circuit potential (OCP) to 2 V at 10 mV s^-1 and different temperatures from 25°C to 85°C, adjusting the temperature by water circulation using a Haake F3 bath. Electrochemical impedance spectroscopy (EIS) measurements were conducted at 25°C in the frequency range of 100 kHz to 0.1 Hz, with 5 mV amplitude, at four different potentials. CA study was carried out at the constant potential of 1.8 V for 1 h, while CP was carried out at the constant current density of 10 mA cm^-2 for 10 h.

Nearly monodisperse CoPt alloy NPs were synthesized by using a new wet-chemical protocol comprising the co-reduction of Co(acac)₂ and Pt (acac)₂ precursors with BTB complex in OAm and ODE mixture at 100°C. In the presented recipe, BTB, OAm, and ODE serve as reductant, surfactant, and solvent, respectively. Figure 1 shows the XPS survey and high-resolution spectra for the Pt 4f and Co 2p core levels. The XPS survey spectrum indicates the presence of all expected elements (Pt, Co, O, and C) (Figure 1A). The high-resolution XPS spectrum of the Pt 4f core level (Figure 1B) shows two bands observable at binding energies (BEs) of 71.5 and 74.9 eV, representative of Pt atoms in the metallic state (Alayoglu et al., 2011). In the case of the high-resolution XPS spectrum of the Co 2p core level (Figure 1C), the BEs for Co 2p_1/2 and Co 2p_3/2 at 800.6 and 780.1 eV are characteristic of metallic Co.

The colloidal Co₄₅Pt₅₅ NPs were then assembled on three different carbon-based support materials, namely, rGO, mpg-CN, and commercial KB. The representative TEM images of Co₄₅Pt₅₅ NPs over these 3 support materials are shown in Figure 2. These images confirm that Co₄₅Pt₅₅ NPs were successfully assembled on all support materials, but they show different particle dispersion on each support material. CoPt/rGO and CoPt/KB show almost homogenous particle dispersion (Figures 2A, C) while the NPs do not display a completely homogeneous dispersion over mpg-CN (Figure 2B ). The observation of such a difference in CoPt NPs dispersion over different carbon-based support materials might be due to the difference in their surface area (BET surface areas; S_rGO = 390 m² g^-1 (Metin et al., 2012), mpg-CN = 182 m² g^-1 (Erdogan et al., 2016), and S_KB = 753 m² g^-1 (Inoue et al., 2012)). Although there are regions with a higher concentration of NPs in the TEM image of CoPt/mpg-CN nanocomposite, one can say that, overall, CoPt NPs were not agglomerate and preserved their initial particle size and distribution over the three support materials, which allows studying their catalytic activity for the same reaction.

To get insight into the crystal structure of Co₄₅Pt₅₅ alloy NPs, powder XRD analyses were performed on CoPt/rGO, CoPt/mpg-CN, and CoPt/KB nanocomposites. Although all the supported catalysts exhibit mostly amorphous structure, the broad peaks observed at around 2θ = 40° are attributed to the (111) plane of the face-centered cubic crystal phase of CoPt NPs (Figure 3) (Tzitzios et al., 2005; Wang et al., 2018). The broadening of (111) peak is due to the very small particle size of CoPt alloy NPs. Moreover, it should be noted that the (111) peak of CoPt is slightly red-shifted, indicating the solid-solution formation between Pt and Co metals. Additionally, the peak aroused at 2θ = 27.4° for the CoPt/mpg-CN nanocomposite is readily assigned to the (002) plane of mpg-CN and the peaks observed at 2θ = 23° in the XRD patterns of CoPt/rGO and CoPt/KB nanocomposites are attributed to the graphitic layered structure of rGO and commercial carbon.

CV data (not shown) revealed notably higher capacitance in the case of CoPt/rGO than in the case of CoPt/mpg-CN and CoPt/KB. This reflects a higher number of active sites that can participate in the adsorption processes in the O adsorption region and, thus, it is expected to lead to higher OER activity (Eftekhari, 2017). Additionally, the electrocatalytic reactivity of the active sites will depend on their accessibility (as OER is not ideally a surface reaction and it proceeds within a thin layer of ca. 10 nm), as well as on their oxidation state (Milikić et al., 2018).

OER activity of CoPt/rGO, CoPt/mpg-CN, and CoPt/KB electrocatalysts was compared by recording polarization curves in 1 M KOH (Figure 4A ). The activity was evaluated based on three parameters: onset potential, E_onset, overpotential necessary to reach a current density of 10 mA cm^-2, η₁₀, and Tafel slope, b. The OER onset potential, defined as the potential at which a current density of 1 mA cm^-2 is reached (Inoue et al., 2012), was observed to be ca. 10 and 40 mV lower at CoPt/rGO (1.61 V) compared to those of CoPt/mpg-CN (1.62 V) and CoPt/KB (1.64 V), respectively, and ca. 20 mV more positive than that of commercial Pt/C (1.59 V), respectively (Table 2). CoPt/rGO (0.484 V) showed 28 and 308 mV lower overpotential to reach a current density of 10 mA cm^-2 compared to CoPt/mpg-CN (0.512 V) and CoPt/KB (0.792 V), respectively. Moreover, this value is ca. 100 mV lower than the overpotential value of commercial CoPt/C (0.584 V) and Pt/C (0.576 V) (Mladenović et al., 2021; 2022). Lower overpotential and higher current densities observed in LSVs of the herein-prepared CoPt/rGO point out its superior activity compared to the commercial CoPt/C. This superior performance of CoPt/rGO under OER polarization conditions could be related to the 2D layered structure of rGO that facilitates mass transfer. Moreover, the high surface area of rGO enhances the dispersion of CoPt alloy NPs and facilitates electron transport between the NPs and graphene (Murillo Leo et al., 2017; Meng et al., 2019). As mentioned above, adsorption/desorption is an essential step in the catalysis of water electrolysis, i.e., catalysis of both hydrogen evolution reaction and herein studied OER. rGO was demonstrated to have a key role originating in its unique reactivity related to the O surface groups present in the graphene structure. Namely, it has been shown that rGO acts as H adatoms acceptor facilitating their recombination to form H₂. H atoms generated by H₂O discharge on transition metals such as Ni, spill onto the rGO where they are re-combined. At the same time, the electrocatalyst’s surface is continuously cleaned, i.e., free active sites are continuously formed enabling HER/OER to proceed (Chanda et al., 2015).

Furthermore, Tafel slopes were determined from overpotential η vs log current density j plots, Figure 4A (Table 2). The evaluated Tafel slopes of CoPt/KB, CoPt/rGO, CoPt/mpg-CN, and CoPt/C were 61, 101, 160, and 208, mV dec⁻¹, respectively, suggesting higher OER rate on CoPt/KB and CoPt/rGO than on CoPt/mpg-CN and commercial CoPt/C. Besides, the OER Tafel slope value for CoPt/rGO was observed to be lower than values reported in the literature, for instance, for OER at commercial Pt/C (198 mV dec⁻¹) and PtNi at different transition metal oxide supports (values ranging from 140 to 224 mV dec⁻¹) in 0.1 M KOH (Mladenović et al., 2021; Mladenović et al., 2022). Furthermore, PtM (M = Ni, Fe, Cu) on graphene nanoplatelets presented notably higher OER Tafel slope values (280–478 mV dec⁻¹). Several Ir_xPd_y/C and Ir/C electrocatalysts were tested for OER in 0.1 M NaOH (Nguyen and Shim, 2018), where Ir₂₃Pd₇₇/C and Ir/C presented the same value of Tafel slope of 164 mV dec⁻¹ that is again higher than Tafel slope value determined for herein tested CoPt/rGO. Ir-Pd nanowires (Zhang et al., 2018) showed a somewhat higher Tafel slope (118 mV dec⁻¹) than herein tested CoPt/rGO.

The value of the Tafel slope can further be used to determine the OER mechanism (Eqs. (3)–(7), i.e., the rate-determining step (RDS). Thus, Tafel slopes of ca. 120, 40, and 28 mV dec⁻¹ are observed when first (Eq. (3) and (second (Eq. (5)), and third (Eq. (7)) electron transfer is the RDS, respectively (Bandal et al., 2017). In addition, a Tafel slope of 60 mV dec⁻¹ can also be observed when the first electron transfer step is followed by the rate-determining chemical transformation (Doyle et al., 2013; Bandal et al., 2017). Shinagawa et al. (2015) thoroughly examined the OER mechanism in alkaline media, where the single-site OER mechanism is assumed. They concluded that the Tafel slope is coverage-dependent. A Tafel slope of 120 mV dec⁻¹ was obtained when the adsorbed species formed before the RDS were predominant. On the other hand, a lower Tafel slope was obtained when adsorbed species formed in the first stage of the reaction were predominant. The Tafel slope value evaluated for OER at herein prepared CoPt/rGO electrocatalyst (101 mV dec⁻¹) suggests the first electron transfer as the RDS. For comparison, for Pt/C, a Tafel slope of 60 mV dec⁻¹ increasing to 120 mV dec⁻¹ with increasing potential was reported for OER in 1.0 M KOH (Shinagawa et al., 2015). M + O H − → M O H + e − (3) M O H + O H − → M O + H 2 O l + e − (4) 2 M O → 2 M + O 2 g (5) M O + O H − → M O O H + e − (6) M O O H + O H − → M + O 2 g + H 2 O l + e − (7)

Figure 4B reveals specific current densities of 1402, 1132, 999, and 104 A g^-1 _Pt for CoPt/rGO, CoPt/KB, CoPt/mpg-CN, and CoPt/C, respectively, at the potential of 2 V. CoPt/rGO again showed higher activity for OER compared to the commercial CoPt/C electrocatalyst, followed by CoPt/KB, where their specific current densities were ≥10 times higher than the specific current density of commercial CoPt/C. It is worth mentioning that commercial CoPt/C contains 46.7 wt% Pt, whereas CoPt/KB, CoPt/rGO, and CoPt/mpg-CN catalysts contain only 1.4, 6.4, and 7.2 wt% Pt, respectively. Specific current density value at 1.48 V of CoPt/rGO (3.2 A g^-1 _Pt), was found to be comparable/higher than values reported for rutile IrO₂ and RuO₂ NPs, and of the same order of magnitude as that of commercial Ir/C (40 wt% Ir, Premetek, ca. 9 A g^-1 _Ir in 0.1 M KOH solution) (Lee et al., 2012).

Water electrolyzers typically operate at temperatures ranging from 65°C to 90°C. Polarization curves of CoPt/rGO in 1 M KOH illustrate the increase of current densities from 77.2 to 214 mA cm^-2 with increasing temperature from 25°C to 85°C (Figure 5A ). CoPt/mpg-CN electrocatalyst showed some instability with increasing temperatures above 35°C due to vigorous gas bubble evolution. Tafel plots of CoPt/rGO were constructed for all temperatures in the potential range from 1.65 to 1.75 V, allowing the determination of the Tafel slope and exchange current density values (Figure 5B). The exchange current density was observed to increase more than three orders of magnitude, i.e., from 0.13 μA cm^-2 at 25°C to 170 μA cm^-2 at 85°C (Table 3), evidencing a strong effect of temperature on OER kinetics (Miles, 2006).

The resistance during OER was assessed by EIS, with the Nyquist plots of three studied electrocatalysts being presented in Figure 6A, B. The electrocatalysts showed the same trend, where resistance values decreased with the increasing applied potential; Figure 6A illustrates the case of CoPt/rGO. The obtained impedance spectra could be best fitted with an equivalent electric circuit having five elements (inset of Figures 6A,B): solution resistance (R₁), charge transfer resistance at the working electrode (R₂), mass-transfer resistance of the adsorbed intermediates (pseudo-resistance, R₃) and capacitance C₁ and C₂ related to the double-layer capacitance and the pseudo-capacitance of the working electrode, respectively. Evaluated EIS parameters for OER at CoPt/rGO at four potentials are presented in Table 4 along with the EIS parameters for OER at CoPt/mpg-CN and CoPt/KB electrocatalysts at a potential of 1.75 V. CoPt/rGO, CoPt/mpg-CN, and CoPt/KB showed similar charge transfer resistance values of 1.9, 2.1, and 2.9 Ω, respectively, at 1.75 V (Figure 6C). These low values suggest boosted charge transfer and effective utilization of active material during the electrochemical reaction. The uncompensated solution resistance, R₁, was slightly lower in the case of CoPt/rGO (4.42 Ω) compared to CoPt/mpg-CN (5.8 Ω) and CoPt/KB (13 Ω); these small variations in solution resistance value originate in small differences in the cell geometry during measurements.

The chronoamperometric response of CoPt/rGO under OER polarization conditions (constant potential of 1.8 V) is shown in Figure 7A. It could be observed that CoPt/rGO showed OER specific current density of ca. 32 A g^-1 at 200th s. However (specific) current density at CoPt/rGO was observed to continuously decrease with time. Thereafter, long-term chronopotentiometric measurements were carried out to assess the electrocatalysts’ stability (Figure 7B). CoPt/KB showed instability after less than 1 h. The overpotential to reach a current density of 10 mA cm^-2 after 10 h was observed to be 0.527 and 1.190 V for CoPt/mpg-CN and CoPt/rGO, respectively. Comparison with values recorded at the beginning of the experiment, reveals the stability of overpotential in the case of CoPt/mpg-CN, but a significant increase in the case of CoPt/rGO. This instability of CoPt/rGO is rather surprising as rGO was reported to stabilize anchored metal NPs through strong electrostatic interactions and thus lead to electrocatalysts’ high long-term stability/durability (Murillo Leo et al., 2017; Meng et al., 2019). Instability might partially come from electrocatalyst’s film deterioration due to intense bubble evolution rather than just NPs detachment from the support. Still, this holds a promise that the stability of CoPt/rGO could be improved and that, along with its high performance for OER catalysis, it could enable its use in water electrolyzers, for instance.