Phosphine Oxide Porous Organic Polymers Incorporating Cobalt(II) Ions: Synthesis, Characterization, and Investigation of H2 Production.

Suitably functionalized porous matrices represent versatile platforms to support well-dispersed catalytic centers. In the present study, porous organic polymers (POPs) containing phosphine oxide groups were fabricated to bind transition metals and to be investigated for potential electrocatalytic applications. Cross-linking of mono- and di-phosphine monomers with multiple phenyl substituents was subject to the Friedel-Crafts (F-C) reaction and the oxidation process, which generated phosphine oxide porous polymers with pore capacity up to 0.92 cm3/g and a surface area of about 990 m2/g. The formation of the R3P·BH3 borohydride adduct during synthesis allows to extend the library of phosphine-based monomeric entities when using FeCl3. The porous polymers were loaded with 0.8-4.2 w/w % of cobalt(II) and behaved as hydrogen evolution reaction (HER) catalysts with a Faradaic efficiency of up to 95% (5.81 × 10-5 mol H2 per 11.76 C) and a stable current density during repeated controlled potential experiments (CPE), even though with high overpotentials (0.53-0.68 V to reach a current density of 1 mA·cm-2). These studies open the way to the effectiveness of tailored phosphine oxide POPs produced through an inexpensive and ecofriendly iron-based catalyst and for the insertion of transition metals in a porous architecture, enabling electrochemically driven activation of small molecules.

Introduction

Many classes of porous materials were devised in the past decades, with a great variety of chemical composition, structural order, and functions. These materials encompass purely inorganic zeolites,[1] hybrid metal organic-frameworks (MOFs, crystalline),[2−5] and purely organic materials[6] like covalent organic frameworks (COFs, crystalline).[7,8] The functional properties of these materials are strongly determined by the permanent porosity, large accessible surface, and size and shape of the pores.[3,9−11] More recently, amorphous porous organic polymers (POPs) have been developed, showing a higher stability than MOFs and COFs, and they can be prepared with a number of chemical functionalities within the cavities. Symmetric aromatic synthons, with tetrahedral- or trigonal-planar geometries, condensed through controlled and directional synthetic methodologies, produced porous materials with uniform pore size distribution and high capacity.[12−17] POPs can also be prepared with more conformationally flexible monomers with nonunivocal position of the linkage bond between monomers and linkers.[18,19] Similar to other porous materials, POPs can be prepared by incorporating into the framework Lewis basic sites that can serve as electron-donor systems for metal centers, and thus are suitable for proton transport,[20] or the capture of volatile species.[21]

Additionally, POPs have been investigated in the domain of heterogeneous catalysis,[18,22−25] usually after the incorporation of metal centers.[26−28] In particular, porous frameworks containing phosphine groups coordinated to metal ions were successfully applied in catalysis.[29−32] Although phosphine oxide functional groups can bind metal ions[33] and promote the interaction with heavy elements, providing anchoring sites for metals in the active sites of catalysts, to date, rarely they are incorporated in POPs.[34,35] A high degree of chemical tunability makes POPs potentially adapted to applications in the energy conversion domain, including electrocatalysis for small molecule activation.[36] However, while the electrocatalytic applications of MOFs[37−41] and COFs[42−44] are extensively explored, the study of POP-based electrocatalysts is only in its infancy.[45−48]

In this work, a series of porous organic polymers bearing P=O functional groups was synthesized and loaded with metal ions to obtain hybrid materials containing highly dispersed, yet accessible, catalytic sites, which promoted electrocatalysis for hydrogen evolution.

Specifically, the porous frameworks have in common the R3P=O structural motif covalently bound within the robust architecture. Various synthetic procedures were applied to prepare phosphine oxide POPs, namely, the Friedel–Crafts (F-C) reaction with FeCl3 or AlCl3 on phenylphosphines or a two-step procedure starting from trichlorophosphine.[30] The P atoms of the resulting porous polymers were oxidized, providing the P=O moiety that was demonstrated to be a good anchoring site for transition metals.[49]

These processes yielded porous frameworks containing isolated phosphine oxide building units (P1, P2, and P3, Figure ) and bidentate units in which two adjacent phosphine oxide units can cooperate to increase the metal-binding ability (P4, P6, P7, and P5). N2 gas-adsorption measurements of the polymers show permanent porosity in both the micro- and mesopore regions, facilitating the diffusion of metal species for the formation of catalytic metal centers. The materials were characterized using thermal methods, multinuclear solid-state nuclear magnetic resonance (NMR) and energy-dispersive X-ray spectroscopy (EDX). In a proof-of-concept study, we screened the hydrogen evolution reaction (HER) activity of the Co-containing materials under neutral pH conditions, which are more environmentally benign with respect to HER in acidic and basic electrolytes,[50−54] even if more challenging to achieve efficient catalysis.[54]

Figure 1

Synthetic pathways (left, a–d) and polymers described in this work (right).

Results and Discussion

Synthesis

Porous organic polymers were obtained by the F-C reaction;[12] in particular, phosphine monomers and external linkers such as benzene with formaldehyde dimethyl acetal (FDA) or 1,3-bis(bromomethyl)benzene were mixed together with a Lewis acid (FeCl3 or AlCl3) to promote polymerization, Figure . In the synthesis, we used monodentate or bidentate P-donors; specifically, the bidentate and conformationally rigid monomers [BPPB (1,2-bis(diphenylphosphino)benzene), and TPPB (1,2,4,5-tetrakis(diphenylphosphaneyl)benzene)] gave rise to P-POPs with the two donor functions properly oriented to provide chelation to the metal center. In the case of bidentate DPPE (1,2-bis(diphenylphosphaneyl)ethane), the complexation of P with BH3 was necessary to prevent the scavenging effect of phosphine groups. The synthesis of compound P3 involved the polymerization of the 4,4′-dibromobiphenyl precursor treated with butyllithium and PCl3.

Characterization of the Frameworks

The materials prepared were characterized by thermal analysis, showing thermal stability above 400 °C (Figures and S21–S27). The thermogravimetric analysis (TGA) profile of P3 was different from that of the other systems, and it showed a sharp multistage decomposition profile with the first weight loss (49%) between 300 and 455 °C and the second weight loss (40%) from 530 to 640 °C. The Fourier transform infrared (FT-IR) spectra of P1-P7 showed similarities with the corresponding phosphine precursors, even though the IR bands were usually larger in the frameworks. All the systems, except P3, exhibited 2900–3000 cm–1 C–H stretching bands, which were associated to the methylene bridge linking the aromatic moieties, Figures S14–S20.

Figure 2

Characterization of P4. (a) TGA in the 25–700 °C temperature range under oxygen flux. (b) EDX spectrum. Chlorine signal from the residual DCE solvent. (c) SEM image.

One-dimensional (1D) 13C, 1H, and 31P and two-dimensional (2D) 1H–13C NMR spectroscopy were used to study the structural organization at the molecular level. 13C spectra of the porous polymers showed peaks between 110 and 150 ppm easily ascribed to the aromatic rings of monomeric units, Figure . Additional signals were present in the aliphatic region for F-C reaction polymers, irrespective of the monomer. The complex pattern was due to multiple alkylation of the aromatic rings.

Figure 3

(a) 13C{1H} CP MAS spectra of P1, P4, P6, P7, and P3 materials performed at a spinning speed of 12.5 kHz and a contact time of 2 ms. (b) Quantitative 13C{1H} MAS spectrum of P4 performed at a spinning speed of 12.5 kHz and a recycle delay of 60 s. (c) 2D 1H-13C PMLG HETCOR NMR spectrum of P4 performed at 12.5 kHz and a contact time of 2 ms. The cross-peaks, highlighted in orange, show the through-space interactions between the aromatic ring and the methylene moiety.

13C Cross-Polarization Magic-Angle-Spinning (CP MAS) NMR spectra of the compounds P1, P4, P6, and P7 synthesized by the Fe-based F-C reaction exhibited a prominent signal at about δ = 37 ppm owing to the methylene bridges connecting the aromatic rings of monomer units. Moreover, minor alkyl and alkoxy (CH3, CH2–O and CH3–O at δ = 16.8–18.7, 56.5, and 73.0 ppm) originated by pendant groups were identified as already observed in porous aromatic polymers by the F-C reaction.[55] A lower content of chlorine-containing pendant groups resonated at 40–44 ppm together with ph-CH2-ph in the broad signal centered at 37 ppm. In 13C CP MAS spectra, the pendant group signals were intensified because of hydrogen-to-carbon magnetization transfer, while in the quantitative 13C MAS NMR spectra, they appeared to be negligible. In the case of Al-based F-C compounds (P2 and P5), we observed CH2-bridging groups and CH3 pendants in the benzyl position (Figure S10). In P3, the aromatic signals substantially dominated, encompassing the carbon–carbon signal of the diphenyl connecting group.[30]

The connectivity of methylene groups, which linked the aromatic rings in the F-C reaction was inferred by 2D 1H-13C NMR spectra, which provided evidence of the close spatial proximity between 1H and 13C nuclei. The 2D 1H-13C MAS spectrum of P4 highlighted the aromatic hydrogens of the main architecture (δH = 6.5 ppm) correlated to the carbons of the bridging methylene groups (δC = 37.3 ppm), Figure . Likewise, the bridging CH2 hydrogens at δH = 4.0 ppm correlated with the aromatic carbons, confirming the reticulation of the monomers by the CH2 linkers. Moreover, the abundance of the CH2 linkers created using the synthetic procedure could be inferred by a quantitative analysis of the 13C MAS spectrum obtained with a long recycle delay of 60 s. The quantitative results of 1 methylene per 9 aromatic carbons were in agreement with the fact that all phenyls of the precursor are reacted and connected through −CH2–benzene–CH2– bridges (Table S2).

According to the 31P SS NMR spectra, the signals of the phosphorus atoms resonated at about 30 ppm, in agreement with the presence of the P=O moiety. In P3, P7, and P6 the minor peaks at δ = −7.2, −10.5, and −14.8 ppm, respectively, suggested the minor presence of reduced aryl phosphorus (Figure S11). This result was in line with previous findings.[30] The 19F SS NMR spectrum of P6 showed the peaks associated to the presence of aromatic fluorine atoms, Figure S13.

The porosity of the frameworks was assessed by N2 adsorption isotherms at 77 K, which exhibited a steep slope in gas uptake at very low relative pressures and a continuous increase at higher pressures, reflecting the simultaneous presence of micro- and meso-pores (Figure ). We could observe that the highest surface area [Langmuir and Brunauer–Emmett–Teller (BET) surface areas of 1125 and 990 m2/g, respectively for P4] was achieved by the rigid monomer structure in which four aromatic rings connected to two phosphorus atoms in the core protrude at different angles, ensuring the expansion of the framework in all directions. An analogous monomer, which contained a flexible ethyl group connecting the two phosphorus atoms, instead of a rigid aromatic ring, allowed higher degrees of conformational freedom and did not generate a framework with a high surface area (P7). Relatively high surface areas of 727 and 640 m2/g were obtained from the monomer triphenylphosphine bearing three phenyl rings. Interestingly, a number of aromatic rings greater than four did not produce any increase in both the surface area and the pore capacity, most likely as a result of the overcrowded reticulation on the same monomer (P6). Hysteresis loops were observed between the adsorption and desorption branches. Such a behavior was indicative of the swelling of the network during sorption owing to capillary condensation in the mesopores, which caused some expansion in the network, as systematically observed in soft polymeric materials.[16,56,57] In the case of P6, the closure at P/P0 of about 0.4 in the hysteresis loop suggested the mesopore shrink to a less extent. Lower surface areas were obtained by the Al-based F-C frameworks (ESI).

Figure 4

N2 physisorption isotherms at 77 K (adsorption, ●; desorption, ○) for P4 (a), P7 (b), P1 (c), and P6 (d). Adsorption and desorption branches are denoted by filled and empty symbols, respectively. Insets: differential and cumulative pore size distributions between 0 and 50 Å (dark and light colors, respectively).

Complexation with Transition-Metal Ions

The presence of binding sites within the frameworks was expected to promote the anchoring of transition-metal ions within the cavities. According to 31P NMR, the donor function was represented by P=O, which is suitable for hard transition-metal ions such as lanthanides or the first-row transition metals.[49,58−60] Hence, the frameworks were readily loaded with Co(II) by soaking the frameworks with tetrahydrofuran (THF) solutions of CoCl2·6H2O. The products were extensively washed with THF to remove the excess metal not bound to the frameworks, until a colorless supernatant appeared. The loading of the materials could be easily appreciated by color change from white to green for Co@P3 (Figure ), whereas Co@P1, Co@P4, and Co@P6 were brown powders (Figure S34). Energy-dispersive X-ray spectroscopy (EDX) confirmed the presence of the metal centers in the functionalized polymers, Figure . According to inductively coupled plasma-atomic emission spectroscopy (ICP-AES), the amount of cobalt anchored to the polymers was in the 0.8–4.2% w/w range, (Tables S5 and S6). The N2 adsorption isotherm of P3 at 77 K exhibited both micro- and mesoporosity with notable swellability, as shown by the hysteresis loop, which closed to zero at a low partial pressure (Figure ). In P3, the mesopore fraction allowed easy access to the isolated P=O moieties from the diffusing species, thus favoring specific interactions with metal ions. Consistently, Co@P3 had a considerably reduced mesopore component, a lower surface area, and a shift of the pore width profile to lower values (Figure S30). On the other hand, the presence of the metal ions did not lead to an occlusion of the inner cavities, hence allowing the movement of small molecules toward and from the metal centers.

Figure 5

(a) P3 (white) and Co@P3 (green). (b) N2 adsorption isotherm at 77 K of P3 (orange) and Co@P3 (blue). Inset: cumulative pore size distributions. (c) EDX of P3 (orange) and Co@P3 (blue).

HER Studies

The electrocatalytic activity of the Co-loaded materials for the HER was explored by linear sweep voltammetry (LSV) in aqueous solution at neutral pH (phosphate buffer). In the case of P1, P4, P6, and P3-modified electrodes, a significant increase in current density was observed in the range from ∼ −0.4 to −0.9 V vs RHE range for metallated materials vs metal-free compounds (Figures and S39, half-wave potentials = −0.75 V for P1, −0.74 V for P4, −0.76 V for P6, and −0.62 V for P3). This suggested the catalytic HER activity of the Co@P1,3,4,6 derivatives, induced by the presence of cobalt in the inner cavities of the corresponding porous materials and most likely associated to Co(II) reduction. The overpotentials to reach a current density of 1 mA·cm–2 were 0.68 V for Co@P4, 0.66 V for Co@P1 and Co@P6, and 0.53 V for Co@P3, relatively high compared to (i) other HER electrocatalysts working at neutral pH[50,61] (ii) related MOF[62,63] and COF[43,61,64] materials (working under kinetically more favorable strongly acidic or basic conditions), and (iii) the only reported noncarbonized POP-based material for electrochemical HER (containing intrinsically highly active Pt nanoparticles).[46] The high HER overpotentials of these cobalt-based materials could be tentatively explained either with the low density of catalytic sites, the low wettability, or the low electrical conductivity of the polymer materials.

Figure 6

LSV curves of selected Co@PX vs PX materials (x = 1, 3, 4, and 6).

In order to evaluate the durability of the catalysts and to confirm the effect of P1,3,4,6 metallation on the HER performances on a longer timescale, controlled potential electrolysis (CPE) experiments were carried out on the Co@P1,3,4,6 vs the respective P1,3,4,6 materials at −0.68 V vs RHE during 8 h. Significant differences in current densities between the Co-containing and the metal-free versions of the material were observed only for the P1 and P3 materials (Figure S40), which were thus the only ones behaving as stable HER catalysts. The amounts of produced H2 were measured by gas chromatography (Figure S41), which allowed to determine the HER Faradaic efficiency to be 82% for Co@P1 (4.76 × 10–5 mol H2 per 11.25 C) and 95% for Co@P3 (5.81 × 10–5 mol H2 per 11.76 C). Turnover numbers (TONs) and frequencies (TOFs) could be estimated based on the hypothesis that each single cobalt center behaved as an active catalyst: TONCo@P1 = 1.1 × 104, TOFCo@P1 = 1.4 × 103 h–1, TONCo@P3 = 1.2 × 104, and TOFCo@P3 = 1.5 × 103 h–1. These catalytic parameters demonstrated a good intrinsic HER activity of each single active Co center[50,53] and a reasonable catalyst stability, even at noncompetitive overpotentials.

Conclusions

In summary, we have prepared a series of porous materials with incorporated P=O functions aiming at binding metal centers with the purpose to generate hybrid materials suitable for application in the energy conversion domain. Different synthetic strategies were investigated to generate porous polymers. F-C polymerization with iron chloride, which is widely used in industrial applications, led to larger absorption capacity and a high surface area of up to 990 m2/g. The detrimental issue of iron catalyst inactivation due to scavenging operated by the phosphines could be circumvented by the formation of the phosphine·BH3 adduct, which was then used for the successful formation of polymer P7. The presented strategy opens the possibility to employ a wider array of phosphines for the generation of tailored phosphine oxide porous polymers. The presence of a homogeneous distribution of P=O moieties in the cavity allowed the functionalization of all of the polymers investigated with cobalt(II) ions using the impregnation method. In a proof-of-concept investigation, Co@P1 and Co@P3 have shown an electrocatalytic HER activity under environmentally benign neutral pH conditions. To our knowledge, they represent a rare example of HER electrocatalysts based on noncarbonized POPs and including non-noble metals as active species. Although the catalytic performances of these materials are currently limited (high overpotentials), in perspective, their properties (porosity, surface area, density of binding sites, wettability, and electrical conductivity) can be tuned, both by exploring new monomer combinations and by the incorporation of other non-noble transition metals (like iron, molybdenum, and nickel, see Figure S33). These future investigations will have the objective of increasing (i) the density of catalytically active centers, (ii) their accessibility to substrates, and (iii) the activity of each single active center. The goal is also to extend the use of POPs to other electrodriven activation reactions (like ORR, CO2RR, NRR, etc.) toward efficient and sustainable small molecule conversion.

Experimental Section

General Methods

Anhydrous solvents were dried and stored over molecular sieves (3 Å), and all other reagents and solvents were used as received. Anhydrous iron(III) chloride was purchased from Sigma Aldrich and stored in a glovebox under nitrogen. Reactions performed under an inert atmosphere were carried out using Schlenk glassware using nitrogen as the inert gas. Flash column chromatography was performed using silica gel (230–400 mesh). NMR experiments were performed on either a Brüker Avance 400 MHz instrument and JEOL 600 MHz ECZ600R instrument at 298 K. Chemical shifts are quoted in ppm relative to tetramethylsilane, using the solvent residual peak of CDCl3 (δH 7.26, δC 77.00) as a reference standard. TGA was performed with a PerkinElmer TGA 8000 and by heating the polymer (0.5–3 mg) from 30 to 600/800 °C in atmospheric pressure, with a T-ramp of 5 °C min–1 under oxygen flux (30 mL min–1). IR spectra were obtained with a Perkin Elmer spectrum two FT-IR spectrometer (diamond crystal) in the 4000–400 cm–1 interval at room temperature.

General Procedures for the Polymerization

The products P1,[29]P2,[31] and P3(30) were prepared according to published procedures. P4, P7, and P6 were prepared by reacting the organophospine monomer, FDA, benzene, anhydrous FeCl3, and anhydrous 1,2-dichloroethane as the solvent under the inert gas atmosphere (N2). The mixture was stirred at 80 °C for the required time. P5 was prepared by reacting the phosphine monomer, 1,4-bis(bromomethyl)benzene, anhydrous AlCl3, and anhydrous 1,2-dichloroethane as solvents under the inert gas atmosphere (N2). The mixture was stirred at 80 °C for the required time. The details of the synthesis and purification steps are reported in the Supporting Information (Figures S1–S9).

Solid-State NMR

Solid-state NMR experiments observing 13C, 1H, 31P, and 19F nuclei were performed using a Bruker Avance 300 instrument equipped with high-power amplifiers (1 kW) and a 4 mm double-resonance MAS probe. 13C{1H} ramped-amplitude cross-polarization (CP) experiments were carried out at a spinning speed of 12.5 kHz using a 5 s recycle delay and 0.05–2 ms contact times. The 1H 90° pulse length was 2.5 μs. As an external chemical shift reference, crystalline polyethylene was set at 32.8 ppm. For a quantitative analysis, single-pulse excitation (SPE) MAS NMR spectra were performed using a recycle delay of 60 s. 31P{1H} ramped-amplitude CP experiments were performed at a spinning speed of 12.5 kHz using a recycle delay of 5 s and a contact time of 8.5 ms. 1H MAS NMR spectra were performed at a spinning speed of 12.5 kHz using a recycle delay of 20 s. The 1H chemical shift was referenced to adamantane. The 19F MAS NMR spectrum of P6 was performed at a spinning speed of 12.5 kHz using a recycle delay of 20 s. The 90° pulse for 19F was 2.5 μs. The 19F chemical shift was referenced to sodium fluoride. Phase-modulated Lee–Goldburg (PMLG) heteronuclear 1H–13C correlation (HETCOR) experiments coupled with fast magic-angle spinning (MAS) allowed the recording of the 2D spectra with a high resolution in both 1H and 13C dimensions. The line widths of hydrogen resonances are on the order of 1–2 ppm, as obtained by homonuclear decoupling during t1. The 2D 1H–13C PMLG HETCOR spectra were run with an LG period of 18.9 μs. The efficient transfer of magnetization to the carbon nuclei was performed by applying the RAMP-CP sequence. Quadrature detection in t1 was achieved by the time proportional phase increment method (TPPI). The carbon signals were acquired during t2 under 1H decoupling by applying the two-pulse phase modulation scheme (TPPM). The 2D 1H–13C PMLG HETCOR NMR spectra of P4 were conducted at 298 K under MAS conditions at 12.5 kHz with a contact time of 2 ms.

Gas-Adsorption Measurements

N2 adsorption isotherms at 77 K were collected on a sorption analyzer (Micromeritics ASAP 2020). The samples were treated overnight at 100 °C under high vacuum before adsorption experiments (p < 5 μbar). Surface areas were calculated from the N2 adsorption isotherm at 77 K using the data in the pressure range P/P0 from 0.015 to 0.1, according to the BET and Langmuir models. The total pore volume was calculated from the N2 adsorption isotherms at 77 K using the nonlocal density functional theory (NLDFT) method with the carbon slit pore model up to P/P0 0.98.

Functionalization with Co(II)

To 40–50 mg of porous polymers P1–6, a 0.35 M CoCl2·6 H2O solution (3 mL; 1.05 mmol), and 3 mL of dry THF were added under an inert atmosphere, forming a blue reaction mixture. The system was stirred at r.t. for 48 h and then was washed with THF and centrifuged until the supernatant became colorless. After this time, the solvent was discarded, and the solid was dried under vacuum for 4 days. The functionalization of P3Li with other transition metals [Ni(II), Mo(III), and Fe(II)] is reported in the Supporting Information (Figure S33).

Characterization of the Metal Content

Scanning electron microscopy (SEM) experiments were performed using an ESEM instrument Quanta 250 FEG (FEI, Hillsboro, OR) equipped with an energy-dispersive spectrometer for X-ray microanalysis (Bruker Nano GmbH, Berlin, Germany). The energy-dispersive X-ray spectrometer is equipped with a QUANTAX XFlash 6 | 30 detector with energy resolution ≤126 eV full width at half maximum (FWHM) at Mnkα. The spectra were collected and analyzed using ESPRIT 1.9 software (Bruker Nano GmbH). ICP-AES analyses were performed with an ULTIMA 2 instrument JOBIN YVON in the radial configuration, with a JY 2501 monochromator calibrated against carbon lines. The optical path was continuously purged with nitrogen (2 L/min). The samples of the functionalized frameworks were dissolved in 2 mL of a mixture of HNO3 65% and H2O2 30% and then heated by microwave irradiation (Milestone, MLS-1200 MEGA, equipped with TFM inner vessels). Calibration was performed with standard solutions, 10% of HNO3 on six different metal concentration levels, ranging from 0.5 to 100 mg/L. No significant spectral interferences were detected. Data were acquired by considering the following emission lines: Fe 238.204 nm, Co 228.616 nm, and Mo 202.030 nm. Data acquisition and processing were performed using the ICP JY v 5.4.2 software (Jobin Yvon).

Electrochemical Measurements

Electrochemical measurements were performed using an electrochemical workstation (Metrohm-Autolab potentiostat/galvanostat, PGSTAT100N) with a standard three-electrode setup, with Ag/AgCl (in 3.5 M KCl solution) as the reference electrode, a platinum plate as the counter electrode, and a glassy carbon electrode (GCE, 3 mm in diameter) coated with as-prepared catalysts as the working electrode. All the measurements were carried out in 0.1 M phosphate buffer (pH 6.93) and conducted in an argon-saturated solution at ambient temperature. In a typical experiment, 5 mg of the target material and 5 mg of carbon black powder (Vulcan XC 72R) were dispersed in 950 μL of isopropanol and 50 μL of Nafion solution (5 wt. %). The mixture was vigorously sonicated for about 1 h to form a “homogeneous” ink suspension. The obtained ink (5 μL) was drop-casted onto a GCE (3 mm diameter, mass loading of ∼90 μg cm–2), previously polished with diamond paste, sonicated in water for 10 min, washed with acetone, and oven-dried. All the measurements were referred to the reversible hydrogen electrode (RHE) using the following equation:

Each newly prepared electrode was first stabilized by cyclic voltammetry (CV) between 0 and −1.18 V vs RHE at a scan rate of 50 mV s–1 until the CV curves remain roughly stable (10 cycles). After this step, LSV experiments were carried out at a scan rate of 5 mV s–1 in the same potential window. The linear portions of the Tafel plots (i.e., overpotential vs log(|j|) plot), as derived from iR-corrected LSV curves, were analyzed using the fitting Tafel equation:where j is the current density (mA.cm–2), η is the overpotential vs RHE, b is the Tafel slope, and A is the intercept of the linear regression. For H2 quantification, a custom-made four-neck cell was used and equipped with rubber septa, allowing for the introduction of three electrodes as well as the gas inlet and outlet tubing. The counter electrode (Pt) was separated from the working electrode compartment with a glass frit.

The free volume of the closed cell after fitting the septa and electrodes was determined (38.0 mL), and the electrolyte (15.0 mL, 0.1 M phosphate buffer) was introduced. The electrolyte was purged with N2 (10 mL min–1) for 30 min before conditioning the working electrode (3 mm GC, coated with the desired material) as mentioned above. The electrolyte was further purged with N2 for 5 min before running 8 h long CPE at −0.68 V vs RHE. The quantification of produced H2 was performed using a Perkin Elmer Clarus 580 gas chromatograph. CPE was run under constant N2 purging (5 mL min–1), and automated injections were programmed to sample the composition of the exhausting gas mixture every 2 min. The instant production of H2 could therefore be monitored over time, and the total quantity of H2 produced during the CPE was determined upon integration over 8 h of the experiments.