It has been proposed that 80% of electricity generated in the United States has the potential to be made renewable by 2050. [1] With the prospect of completely renewable electricity on the rise, electrochemical water splitting for hydrogen generation poses a promising technology for sustainable energy storage and conversion. [2,3] Implementation of this technology will require improved catalyst design. [4] Platinum has remained a leading electrocatalyst for the multi-electron transfer process of hydrogen evolution reaction (HER), with a high exchange-current density and ideal Tafel slope. [5] However, due to the high cost and low natural abundance of platinum, future catalysts must optimize the use of platinum within the material. [6,7] To achieve this goal researchers have developed various methods to limit the amount of platinum through reducing particle size and the use of platinum oxide rather than platinum metal particles. For instance, Yang et al. used platinum oxide nanoparticles for HER catalysis and found that decreasing particle size from 2 to 0.2 nm resulted in increased catalytic activity, which was attributed to sizedependent proton adsorption properties. [8] More recently, Sarno and colleagues showed that decreasing particle size as well as strong metal oxide-support interactions can help reduce platinum loading and improve hydrogen evolution rates by using a composite material containing platinum metal, platinum(IV) oxide, and iron oxide. [9] Strong metal-support interactions were also reported by Cheng et al. using platinum oxide clusters containing Pt 2+ and Pt 4+ species supported on TiO2. [10] It is widely believed that hydrogen adsorption onto the catalyst surface, known as the Volmer step, is often the rate determining step for HER in both acidic and basic media. [11,12] Thus, to better understand the reaction rates, Cheng and colleagues used density functional theory (DFT) to calculate changes in Gibbs free energy (ΔGH) for hydrogen adsorption. It was found that platinum in a highly oxidized state resulted in a ΔGH close to zero (-0.06 eV) that is ideal for HER. This improved hydrogen binding was attributed to the presence of Pt 4+ sites in Pt-O-Ti bonding. This result suggests that highly oxidized platinum sites can efficiently adsorb hydrogen onto the surface while minimizing the adsorption energy to allow for hydrogen gas formation and release from the catalyst surface. However, the material under study [10] contained a large portion of platinum in a lower oxidation state (Pt 2+ and Pt 0 ) and the HER performance remained subpar as compared to that of commercial Pt/C. In addition, the catalytic contributions and importance of the portion of low-valence platinum were not distinguished. Therefore, further study is required to understand how the platinum's oxidation state can be controlled to tailor the hydrogen evolution performance. With this in mind, controlling platinum oxidation state within the oxide material poses a useful route to effective catalyst design.
The selection of an ideal catalyst support is also critical as this plays a key role in the manipulation of the resulting particle size, stability and other key factors influencing the catalytic performance. [13][14][15] Graphitic carbon nitride (C3N4) represents a promising support for its surface functionality, ease and low cost of production, and inherent ability to control nanoparticle size. Zhu and colleagues have shown how manipulation of C3N4 condensation rate and temperature can be used to control the size of the resulting gold and platinum nanoparticles in a C3N4mesoporous silica (SBA-15) composite material. [16] It was shown that incorporation of the C3N4 resulted in small, homogeneously dispersed nanoparticles and prevented nanoparticle agglomeration, which was attributed to strong support interactions, in comparison to SBA-15 alone. Vinu and coworkers demonstrated the inherent ability of C3N4 to act as a stabilizing agent to produce ultra-small, highly dispersed Au nanoparticles anchored firmly on the functional moieties on the C3N4 surface. This can be attributed to the pyridinic nitrogen on the C3N4 surface that act as a strong Lewis base and π-bond planar layers able to anchor the substrate. [17] For these reasons, C3N4 is an ideal candidate for the fabrication of small metal oxide nanoparticle composites.
Herein we report a facile, effective approach to properly distinguish contributions from different platinum oxide species toward HER utilizing C3N4 as a unique support material. C3N4 nanosheets are first prepared through condensation of urea by a two-step route. Platinum oxide nanoparticles of about 1.4 nm in diameter are then deposited onto the C3N4 surface by a thermal refluxing method in water with platinum chloride. [18][19][20] Electrochemical measurements show that Pt 4+ species is primarily responsible for the remarkable HER activity, which even surpasses that of commercial Pt/C; and the HER activity increases with increasing Pt 4+ content. Experimentally, C3N4 nanosheets were prepared by thermal treatment of urea, onto which were then deposited platinum oxide nanoparticles by thermal refluxing of C3N4 and PtCl2 or PtCl4 in water at 90 ⁰C. The resulting samples were referred to as Pt2+90C and Pt4+90C (synthetic details in the Experimental Section). Figure 1 and S1 depict representative TEM images of (a) Pt2+90C and (c) Pt4+90C. The overall morphologies are consistent with those of the as-prepared C3N4 (Figure S2), displaying a layered, sheet-like structure. In addition, both Pt2+90C and Pt4+90C samples can be seen to contain a number of dark-contrast nanoparticles deposited on the C3N4 surface (note that no such nanoparticulate objects were observed in the C3N4 sample, Figure S2). From the core histograms in Figure S3, the nanoparticles can be seen to mostly fall within the narrow range of 1.0 to 4.4 nm, with the average particle diameter at 2.0 0.6 nm for Pt2+90C and 2.1 0.7 nm for Pt4+90C. Furthermore, high resolution TEM images reveal clearly-defined lattice fringes of the nanoparticles (insets to Figure 1), with an interplanar spacing of 0.23 nm for Pt2+90C and 0.24 nm for Pt4+90C, which are consistent with several crystalline facets, such as PtO(110), PtO2(011), and Pt(111). [21,22] Yet, further characterizations below suggest that these are actually due to the formation of platinum oxide PtOx (vide infra).
Further structural insights were obtained in X-ray diffraction (XRD) measurements. From Figure S4, one can see that Pt4+90C, Pt2+90C, and C3N4 all exhibit a single, major diffraction peak centered at 2θ = 27°, which can be assigned to the (002) crystalline facets of carbon nitride, as observed previously. [23] In the Pt4+90C and Pt2+90C samples, the diffraction patterns of platinum oxide can not be resolved, in comparison to the standard references of PtO (reference code 00-027-1331) and PtO2 (reference code 01-075-0978), most likely due to the small nanoparticle size, as seen in TEM measurements (Figure 1).
XPS measurements were then carried out to probe the chemical composition and electronic environment of the material. From the survey spectra in Figure 2a, two major peaks can be seen at 288 and 399 eV, due to C 1s and N 1s of C3N4, respectively; and for Pt2+90C and Pt4+90C, two additional species can be identified at ca. 75 eV for Pt 4f, [24] and 532 eV for O 1s, indicating the successful deposition of platinum species onto the C3N4 surface (the small peaks at 199 eV and 270 eV can be assigned to Cl 2p and Cl 2s, respectively, due to residual chloride in the samples); [25] and based on the integrated peak areas, the Pt loading was found to be very comparable at 32 wt% for Pt4+90C and 28 wt% for Pt2+90C. Figure 2b depicts the corresponding high-resolution XPS scans of the Pt 4f electrons. The Pt2+90C sample (red curve) can be seen to exhibit two doublets. The first pair at 76.3 and 73.0 eV corresponds to the 4f5/2 and 4f7/2 electrons of Pt 2+ , whereas the other at 78.0 and 74.7 eV to those of Pt 4+ and a peak splitting of 3.3 eV is consistent with Pt 4f spin-orbit coupling (note that no metallic Pt can be resolved). [9,20,24,26,27] Additionally, based on the integrated peak areas, the atomic ratio of Pt 4+ :Pt 2+ in Pt2+90C is estimated to be 0.45:1. Similar behaviors can be seen with the Pt4+90C sample, where the two doublets appear at 76.2/72.9 eV and 78.0/74.7 eV, respectively. However, the ratio of Pt 4+ :Pt 2+ is significantly higher at 1.66:1, more than 3.5 times that in Pt2+90C. Regardless, this suggests that thermal refluxing of platinum chloride and C3N4 effectively led to the deposition of PtOx nanoparticles onto the C3N4 surface (Figure 1). The high-resolution scans of the C 1s and N 1s electrons are depicted in Figure S5. Deconvolution reveals two distinct peaks at 287.85 and 284.33 eV for the C3N4 sample, which are attributed to the sp 2 hybridized carbon within the C3N4 network (C-N=C) and sp 3 carbon due to defect moieties (C-C), respectively. [23,[28][29][30] The N 1s region of C3N4 was deconvoluted into four components centered at 398.34 eV, 399.81 eV, 400.96 eV, and a weak one at 404.10 eV. These individual components can be attributed to the sp 2 hybridized pyridinic nitrogen (C-N=C), sp 3 hybridized tertiary nitrogen (N-(C)3), quaternary nitrogen (C-N-H), and a π-satellite excitation, respectively. [23,[28][29][30] Upon thermal refluxing with platinum salts, the nitrogen and carbon peaks associated with the C3N4 matrix exhibit an increase in binding energy. The primary carbon peak shifts to 287.97 and 288.04 eV (C-N=C), and the defect carbon to 284.53 and 284.57 eV (C-C) for Pt2+90C and Pt4+90C, respectively. The N 1s electrons also experience a similar increase in binding energy with the three primary peaks centered at 398.47 eV and 398.54 eV (C-N=C), 399.92 eV and 399.88 eV (N-(C)3), 401.10 eV and 401.18 eV (C-N-H), corresponding again to Pt2+90C and Pt4+90C, respectively. This slight increase in binding energy can be attributed to electron donation from the C3N4 support to the Pt centers suggesting a strong interaction between the nanoparticles and C3N4. [31,32] Figure 3. (a) Pt L3 edge XANES data for Pt2+90C and Pt4+90C (Pt foil and commercial α-PtO2 as references). All edge steps are normalized to an edge step height of 1. The energy range selected to normalize the edge is 11700 to 12000 eV. Measurements are performed at the temperature of 10 K. (b) Pt L3 edge r-space data for Pt2+90C (red), Pt4+90C (blue) and commercial α-PtO2 (black). The Fourier transform (FT) window is from 3.0 to 10.8 Å -1 , rounded using a Gaussian function of width, 0.2 Å -1 . The fast oscillating function is the real part R of the FT while the amplitude is the √𝑅 2 + 𝐼 2 where 𝐼 is the imaginary part of the FT. Pt4+90C has been shifted vertically by 0.4 while α-PtO2 has been shifted vertically by 0.8.
Further structural insights were obtained in X-ray absorption spectroscopic (XAS) measurements, where data were collected at the Pt L3 edge to probe the Pt electronic states and investigate the local structures. The data were reduced and analyzed using the RSXAP package. [33] In Figure 3a, the Pt L3 XANES (X-ray absorption near edge spectroscopy) data for Pt2+90C and Pt4+90C are compared with those for a Pt foil and the α-PtO2 reference sample. All samples are electrically conducting and there is no shift of the main edge as typically observed for different valences in insulating materials, suggesting that the Fermi energies are nearly identical among the samples. A strong absorption in this region, referred to as the white line, is observed due to the Pt 2p → 5d transitions. [34,35] Typically, the white line intensity increases with decreasing 5d orbital occupancy (i.e., increasing valence states). [35][36][37] The fact that the white line intensity increases in the order of Pt foil < Pt2+90C < Pt4+90C < PtO2 indicates that the Pt charge state in Pt2+90C and Pt4+90C fell in the intermediate between Pt(0) and Pt(IV), and is higher in Pt4+90C than in Pt2+90C, in good agreement with results from XPS measurements (Figure 2). Consistent behaviors can be seen with the white line energy, which shifts positively in the order of Pt foil (11565.16 eV) < Pt2+90C (11566.00 eV) < Pt4+90C (11566.44 eV) < α-PtO2 (11567.27 eV). EXAFS (extended X-ray absorption fine structure) analysis for Pt2+90C and Pt4+90C was then carried out and compared to that for α-PtO2 in Figure 3b. The same FT range (3.0 to 10.8 Å -1) ) is used for each plot although the k-space data for α-PtO2 extend to much higher k (Figure S6). The k-space plots show increasing disorder from the reference sample to Pt2+90C. Note that the double peak structure near 3 Å in Figure 3b, associated with Pt-Pt second neighbor pairs in α-PtO2, is nearly washed out for the latter sample. α-PtO2 has a hexagonal structure; [38,39] and the environment about Pt consists of six nearest-neighbor O atoms, six second-neighbor Pt atoms and six third-neighbor O atoms. To fit the data, theoretical Pt-O and Pt-Pt functions were calculated for α-PtO2 using FEFF7, [40] plus a weak multiscattering peak, as shown in Figure 4. The data were then fitted to a sum of such standards, allowing the pair distances and pair distribution widths, , to vary. Note that the third neighbor Pt-O2 and the multiscattering peak amplitudes are very small and we only quote the results for the first two neighbors. As shown in Table 1, the Pt-O bond length is somewhat shorter for Pt4+90C than for Pt2+90C. A similar change is also observed for the second neighbors, again, consistent with a higher charge state of Pt in Pt4+90C than in Pt2+90C.
To assess the electrocatalytic performance of these materials toward HER, linear sweep voltammetry was performed under a nitrogen atmosphere in 0.5 M H2SO4 at the potential scan rate of 10 mV s -1 . Figure 5a displays the polarization curves of Pt2+90C, Pt4+90C, and commercial Pt/C, in comparison to that of C3N4.
One can see that C3N4 exhibited essentially zero HER activity, whereas apparent catalytic activity appears with the other three Pt-containing samples. Remarkably, at equivalent platinum mass loadings, Pt4+90C stood out as the best among the series, even outperforming commercial Pt/C, with an ultralow overpotential (η10) of only -7.7 mV to reach the current density of 10 mA cm -2 , in comparison to -26.1 mV for Pt/C and -214.6 mV for Pt2+90C, suggesting that Pt 4+ species played a critical role in the HER activity. In fact, to the best of our knowledge, the Pt4+90C sample outperforms most platinum oxide-based HER catalysts in acidic media that have been reported in recent literature (Table S1). To further understand the reaction mechanisms of these materials, Tafel plots were derived from the polarization curves, as shown in Figure 5b, from which the Tafel slopes was estimated to be 21.0 mV dec -1 for Pt4+90C, 26.3 mV dec -1 for Pt/C and 55.4 mV dec -1 for Pt2+90C [41] . This suggests that kinetically Pt4+90C behaved similarly to Pt/C where the Tafel reaction is the rate determining step, while for Pt2+90C, HER is likely limited by the slower Heyrovsky reaction.
Figure S7 depicts the Nyquist plots (squares) and the corresponding fits (solid lines) using a typical Randle's equivalent circuit modified with a Warburg diffusion term (W2). The chargetransfer resistances (RCT) was estimated to be 2230 Ω for the asprepared C3N4, and diminished markedly to 26.8 Ω for Pt2+90C, and only 1.6 Ω for Pt4+90C. That is, the charge-transfer kinetics of HER was dramatically improved with the deposition of platinum onto C3N4, and the Pt4+90C sample showed the lowest charge transfer resistance among the series of samples. To further examine the influence of the Pt oxidation state on the HER performance, several additional electrochemical measurements were performed. In the first test, the Pt2+90C sample was found to show a marked improvement of the HER performance after 50 cycles at high potentials between +0.9 and +1.2 V vs RHE (where Pt 2+ was electrochemically oxidized to Pt 4+ ) at the potential rate of 10 mV s -1 . From Figure 5c, one can see that η10 diminished drastically by 120 mV to -93.8 mV (blue solid and dotted curves). XPS measurements (Figure S8) show that the primary peak appears at 73.9 eV, ca. 0.9 eV higher than that of the as-prepared sample, suggesting the formation of Pt 4+ species that led to the improved HER performance. [9,20,24,26,27] Additional tests were carried out with the Pt4+90C sample. When the catalyst was subject to potential cycling between 0 and +0.02 V vs RHE (where Pt 4+ was electrochemically reduced to lower valence states) for 1000 cycles at 10 mV s -1 , the η10 in the subsequent HER measurement was found to deteriorate slightly, from -7.7 mV to -21.7 mV (red solid and dotted curves). However, after 50 potential cycles between +0.9 and +1.2 V to regenerate the Pt 4+ species, the HER performance was almost fully recovered, with η10 = -9.8 mV (green solid curve). By sharp contrast, electrochemical treatment (reduction or oxidation, black solid and orange dotted curves) of C3N4 alone did not lead to any change of the electrochemical response, suggesting that it is the Pt species that is responsible for the HER activity and Pt 4+ is far more active than Pt 2+ . In fact, the HER activity increases markedly with increasing Pt 4+ loading in the Pt4+90C sample (Figure S9).
In summary, PtOx nanoparticles were deposited onto C3N4 surfaces by thermal refluxing of C3N4 nanosheets and platinum chloride in water. The nanocomposites displayed similar platinum loading and particle size but with stark differences in the platinum valence states. It was shown that the platinum oxidation state greatly influenced the hydrogen evolution performance under electrochemical conditions, and platinum oxide nanoparticles in primarily a 4 + charge state exhibited an HER performance even superior to that of commercial platinum on carbon in acid media. This catalyst showed excellent recoverability after recycling the material under oxidizing conditions. Results from the present study suggest that future design of platinum oxide catalysts should attempt to optimize the valence states of platinum by minimizing the formation of less oxidized, and less active species.
Urea (Certified ACS, Fisher Chemicals), platinum(II) chloride (PtCl2, 73% Pt, ACROS Organics), platinum(IV) chloride (PtCl4, 99%, ACROS Organics), platinum on carbon (Pt/C, nominally 20 wt.%, Alfa Aesar), platinum(IV) oxide (-PtO2, Matheson Coleman & Bell), carbon black (Vulcan XC 72R), and Nafion 117 (Sigma-Aldrich) were used as received. All solvents were obtained through typical commercial sources and used as received. Water was supplied with a Barnstead Nanopure water system (18.3 MΩ cm).
Graphitic carbon nitride (C3N4) was prepared by adopting a method reported previously. [23,28] Briefly, urea (15 g) was placed in a crucible, covered, heated in air to 300 C at a rate of 2 C min -1 and held for 3 h. The resulting white solid was ground to a fine powder, heated again in air to 520 C at a rate of 20 C min -1 and held for 4 h. The resulting light-yellow solid was collected and washed with Nanopure water, filtered, and dried in a vacuum furnace for 24 h at 60 C, affording C3N4.
Platinum oxide deposition was performed by following a method described previously. [8,20] Briefly, 50 mg of C3N4 prepared above was dispersed in Nanopure water (50 mL) under ultrasonication for 1 h. The resulting dispersion was then placed onto a hot plate under stirring. PtCl2 or PtCl4 (0.27 mmol) was slowly added to the stirring solution and allowed to mix at 90 C for 48 h. In the case of PtCl2, the salt was first dissolved in HCl and neutralized with anhydrous sodium carbonate to achieve a neutral pH before addition. The products were collected by centrifugation at 4500 rpm for 10 min, washed with Nanopure water and acetone, and dried in a vacuum oven overnight, which were named as Pt2+90C and Pt4+90C. It should be noted that the supernatant exhibited an orange color, indicative of an excess of platinum chloride in the solution, and became clear after washing, signifying effective removal of excess metal salts.
TEM measurements were carried out on a JOEL JEM 2100F microscope. XRD patterns were acquired with a Bruker D8 Advance diffractometer with Cu K radiation (λ = 0.15418 nm). XPS measurements were performed with a Phi 5400/XPS instrument equipped with an Al Kα source operated at 350 W and 10 -9 Torr. XAS measurements were carried out at 10 K on beamline 4-1 at the Stanford Synchrotron Radiation Lightsource using an Oxford liquid helium cryostat.
Electrochemical measurements were performed with a CHI710 workstation, and electrochemical impedance measurements were carried out with a Gamry Reference 600 instrument. A glassy carbon electrode (5.60 mm in diameter, 0.246 cm 2 ) was used as the working electrode, while a Ag/AgCl (1.0 M KCl) electrode and graphite rod were used as the reference and counter electrodes, respectively. The Ag/AgCl electrode was calibrated versus a reversible hydrogen electrode (RHE) and all potentials in the present study were referenced to this RHE. In a typical experiment, 2 mg of a dry sample prepared above was mixed with 3 mg of carbon black and sonicated for 20 min in 1 mL of isopropanol followed by an addition of Nafion (40 µL) and sonicated for an additional 10 min. 30 µL of the prepared ink was dropped onto the surface of the glassy carbon electrode and dried at room temperature, corresponding to a catalyst mass loading of 0.244 mg cm -2 . Pt 2+