Constructing a Triangle Ensemble of Pt Clusters for Enhanced Direct-Pathway Electrocatalysis of Formic Acid Oxidation

Abstract

The pursuit of operational advancements in direct formic acid fuel cells (DFAFCs) necessitates the development of high-performance platinum (Pt)-based catalysts for formic acid electrooxidation (FAOR). However, FAOR on Pt-based catalysts follows a dual pathway mechanism, in which the direct pathway is a preferred route due to its efficient dehydrogenation process. Conversely, the indirect pathway results in the generation of adsorbed CO species, a process that deleteriously poisons the active sites of the catalyst, with CO species only being oxidizable at higher potentials, causing a significant compromise in catalyst performance. Herein, we have successfully synthesized Pt-C₃N₄@CNT, where three Pt clusters are precisely dispersed in a triplet form within the C₃N₄ by virtue of the unique structure of C₃N₄. The mass activity for the direct pathway (0.44 V) delivered a current density of 1.91 A ${\mathrm{mg}}_{\mathrm{Pt}}^{-1}$, while the indirect pathway (0.86 V) had no obvious oxidation peak. The selectivity of Pt-C₃N₄@CNT catalysts for the direct pathway of FAOR was improved due to the special structure of C₃N₄, which facilitates the dispersion of Pt tri-atoms in the structure and the electronic interaction with Pt. In this study, we provide a new strategy for the development of highly active and selective catalysts for DFAFCs.

1. Introduction

Formic acid oxidation reaction (FAOR) has been widely investigated with the development of direct formic acid fuel cells (DFAFCs) [1,2,3,4,5,6]. In addition, FAOR has been regarded as an ideal model reaction for exploring the structure–activity relationship of electrocatalysts and understanding the oxidation mechanism of more complex organics. The widely accepted dual-path mechanism of FAOR, as outlined in Scheme 1 [7,8,9,10], presents two distinctive pathways: The direct pathway proceeds via a dehydrogenation process forming a reactive intermediate (COOH*), which is subsequently oxidized to CO₂. In contrast, the indirect pathway generates adsorbed CO species (CO_ads) through a dehydration process, which can be oxidized to CO₂ at higher potentials. Despite the slow initial formation rate, CO_ads can accumulate on the Pt surface and significantly impede the direct pathway [11]. This presents a substantial challenge in the context of DFAFC applications, as the indirect pathway must be substantially suppressed to prevent the detrimental poisoning of active sites by the site-blocking CO_ads species, which in turn impedes the direct pathway of FAOR. To meet the requirements of practical applications, it is desirable to suppress the dehydration of formic acid, thus inhibiting the accumulation of CO_ads. Correspondingly, it is urgent to develop efficient catalysts by realizing the direct pathway of FAOR based on a deeper insight into the elementary steps of the mechanisms.

To date, several strategies have been employed in the design of catalysts aimed at modulating the FAOR reaction pathway on Pt-based catalysts. For instance, Pt alloys were prepared to alleviate the poisoning effect of CO_ads to improve the FAOR performance with a synergistic effect [12,13,14]. Moreover, the surface modification of Pt with organic molecules was proposed to inhibit the indirect pathway of FAOR [15,16,17]. Despite these achievements, an atomic-level understanding of the structure–activity relationship of Pt-based catalyst for FAOR is lacking. Atomic ensemble effect, i.e., modifying an adequate number of atoms in a suitable geometric arrangement at the catalyst surface, plays a crucial role of determining the reaction step, as well as the activity and/or selectivity of the catalyst [15,18]. Although a series of computational works have been contributed to investigate the atomic ensemble effect on the electrocatalytic reaction mechanisms, the experimental evidence based on the well-designing and preparation of the atomic ensemble of catalysts are seldomly reported. Recently, the advancements in confining ultrafine metal clusters provide good opportunities to manipulate the atomic ensemble of catalysts [19,20]. Therefore, it is possible to precisely prepare Pt-based ensembles to enhance the direct electrooxidation of formic acid by virtue of an ensemble effect and illuminate the reaction mechanism based on an explicit structure–activity relationship.

Herein, we report the construction of triangle clusters of Pt atoms by using graphitic carbon nitride as a structure-guiding template through the coordination between Pt and N atoms. The as-prepared Pt-C₃N₄@CNT showed promising activity for direct pathway FAOR with a mass activity of 1.91 A ${\mathrm{mg}}_{\mathrm{Pt}}^{-1}$ at a peak potential of 0.44 V, which is significantly higher than that obtained on the Pt@CNT (0.11 A ${\mathrm{mg}}_{\mathrm{Pt}}^{-1}$). Besides, the j_d/j_i of Pt-C₃N₄@CNT is 10 times higher than that of Pt@CNT (j_d refers to the peak current density of direct pathway FAOR; j_i refers to the peak current density of indirect pathway FAOR). The single fuel cell device equipped with Pt-C₃N₄@CNT as an anodic catalyst was fabricated, delivering a high Pt mass power density (206.46 mW ${\mathrm{mg}}_{\mathrm{Pt}}^{-1}$), which is 4.8-fold higher than that obtained with Pt@CNT anodic catalyst (43.18 mW ${\mathrm{mg}}_{\mathrm{Pt}}^{-1}$). Moreover, mechanistic investigations show that the high activity and nearly complete direct pathway preference for FAOR obtained with Pt-C₃N₄@CNT can be attributed to the unique ensemble of Pt atoms, where the triangle clusters provide intimately adjacent Pt atoms for a direct dehydrogenation pathway.

2. Results and Discussion

Figure 1a illustrates the preparation procedure of Pt-C₃N₄@CNT catalyst. C₃N₄ was obtained by calcinating urea in a muffle furnace at 550 °C for 4 h. Then, C₃N₄, CNT, and H₂PtCl₆·6H₂O were dispersed in glycol and treated by co-heating at 160 °C for 3 h. Commonly, C₃N₄ has a graphene-like structure, except that it contains an internal triangle-shaped cavity with a side length of ≈0.72 nm. Such a unique structure offers an ideal platform for encapsulating the reactants in the internal triangle-shaped cavity [20]. Therefore, we intend to use C₃N₄ as a structure-guiding template to construct triangle clusters of single-atom Pt. On the one hand, the N sites exhibit a strong affinity for coordinating with Pt salt ions (${\mathrm{PtCl}}_4^{2-}$), thereby aiding the subsequent reduction to produce Pt atoms. On the other hand, the confined space in the cavity of C₃N₄ structure is believed to benefit the generation of isolated atoms. For comparison, a control sample (Pt@CNT) was prepared by following a similar process but without the addition of C₃N₄.

Figure 1b,c show the transmission electron microscopy (TEM) and high-resolution TEM (HRTEM) images, respectively. TEM images confirm the obvious CNTs structure in Pt-C₃N₄@CNT, with typical hollow tubular morphology similar to pristine CNTs. Also, the CNTs structure of Pt@CNT was observed, as shown in Figure S1. For all samples, the morphology and structure of CNTs were well-maintained, indicating that the preparation process did not change the pristine structure of CNTs. HRTEM images of Pt-C₃N₄@CNT, as depicted in Figure 1c, exhibit a typical graphitic structure of CNTs. Interestingly, a shell without distinct lattice fringes can be observed on the outer surface of the CNTs, which corresponds to the C₃N₄ shell. No obvious metallic lattice was observed in the Pt-C₃N₄@CNT, while there was a clear metallic lattice in the Pt@CNT structure identified as metallic Pt nanoparticles (Figure S2). The microstructure of the catalyst was further characterized by aberration-corrected high-angle annular dark-field scanning TEM (AC-HAADF-STEM). In the HAADF-STEM images of the Pt-C₃N₄@CNT sample (Figure 1d), evenly distributed bright spots are observed, indicative of the atomic dispersion of Pt. A particularly fascinating discovery is the significant proportion of atoms found in triangular clusters, in which three metal atoms closely associate with each other at an average distance of 0.17~0.25 nm (Figure 1f). This finding contrasts sharply with the control samples lacking C₃N₄ substrates, which exhibited conventional nanoparticle characteristics, alongside clearly visible lattice fringes of Pt upon further scrutiny (Figure S2). The geometry of this unique tri-metal atom matches nicely with the in-plane void of C₃N₄, which is triangle-shaped with a side length of ≈0.48 nm. This unique structure presents C₃N₄ with the capability to anchor metal atoms effectively, culminating in a tri-metal configuration that retains the distinctive triangular structure [20]. Moreover, energy-dispersive X-ray spectroscopy (EDX) analysis of Pt-C₃N₄@CNT revealed an even distribution of platinum, nitrogen, and carbon signals (Figure 1e), showcasing the uniform dispersion of C₃N₄ within the complex. Additionally, inductively coupled plasma atomic emission spectrometry (ICP-AES) was employed to evaluate the metal loadings in Pt-C₃N₄@CNT and Pt@CNT, which were found to be 1.76 wt% and 2.41 wt%, respectively. These results suggest that C₃N₄ offers an optimal class of templates for crafting such unique tri-metal atom structures. Such triangle clusters of Pt atoms could be highly favorable for FAOR.

The X-ray diffraction (XRD) patterns of both Pt-C₃N₄@CNT and Pt@CNT show the peaks which correspond to the CNT, and there are also peaks of C₃N₄ in Pt-C₃N₄@CNT, as shown in Figure 2a. However, there is no obvious peak of metallic Pt in the XRD pattern of Pt-C₃N₄@CNT, which indicates that there was no metal or metal compound formed in the sample. In contrast, there is an obvious peak of metallic Pt (PDF#65-2868) in the XRD patterns of Pt@CNT, thereby affirming that no metal or metal compounds materialized in Pt-C₃N₄@CNT. This conclusion aligns with the data procured from earlier HAADF-STEM imaging. Then, Raman spectroscopy and Fourier transform infrared (FTIR) spectroscopy were further performed to investigate the composition of the samples, as shown in Figure 2b,c, respectively. The Raman spectrum of the Pt-C₃N₄@CNT sample (Figure 2b) presents no notable peaks, attributable to the substantial presence of C₃N₄ in the material, thus the results largely mirror those of C₃N₄. In contrast, the Pt@CNT sample exhibits a Raman peak consistent with CNT [21,22]. Additionally, the D band (1360 cm⁻¹) and the G band (1600 cm⁻¹) appear in the Raman spectra, further confirming the presence of C₃N₄ in Pt-C₃N₄@CNT. The incorporation of C₃N₄ into Pt-C₃N₄@CNT is substantiated further through FTIR spectroscopy (Figure 2c). Peaks of the Pt-C₃N₄@CNT samples primarily occur at 811.3 cm⁻¹, between 1200 and 1700 cm⁻¹, and within 2800–3700 cm⁻¹, harmonizing with the characteristic peaks of C₃N₄. The peak at 811.3 cm⁻¹ is linked to the stretching vibration of the triazine ring [23], whereas the peaks between 1200 and 1700 cm⁻¹ correspond to the stretching pattern of aromatic carbon nitride heterocyclic units [24]. Specific peak assignments include 1247 cm⁻¹ (stretching vibration of C–N(–C)–C or C–NH–C), 1312 cm⁻¹ and 1637 cm⁻¹ (stretching vibrations caused by C–N and C=N on the heterocyclic ring, respectively), and 1407 cm⁻¹ (vibration of C=C bond). The broad peak from 3000 to 3600 cm⁻¹ is attributed to vibrations caused by amine or hydroxyl groups [25]. In contrast, the Pt@CNT sample only exhibited a weak peak relating to CNT within the range of 3200–3700 cm⁻¹ in the FTIR spectrum, with no characteristic peaks of C₃N₄ being detected. These findings strongly suggest the successful incorporation of C₃N₄ into Pt-C₃N₄@CNT. To further probe the surface chemical compositions of the resultant Pt-C₃N₄@CNT and Pt@CNT, X-ray photoelectron spectroscopy (XPS) was employed. The Pt 4f spectrum (Figure 2d) of both Pt-C₃N₄@CNT and Pt@CNT reveal a distinct Pt 4f doublet assignable to the 4f_5/2 and 4f_7/2 peaks of Pt (0), corroborating the almost metallic nature of Pt in both configurations [26]. Nevertheless, on examining the precise peak positions of the two samples, a minor positive shift in the binding energy for Pt-C₃N₄@CNT was observed compared to Pt@CNT (4f_5/2: 73.65 eV for Pt-C₃N₄@CNT and 73.35 eV for Pt@CNT; 4f_7/2: 70.35 eV for Pt-C₃N₄@CNT and 70.05 eV for Pt@CNT). This shift suggests a slight decrease in the electron density of Pt in the Pt-C₃N₄@CNT structure, potentially attributable to the presence of C₃N₄ [27]. The inferred electron transfer suggests that the minor reduction in surface electron density on Pt in Pt-C₃N₄@CNT could enhance its CO tolerance. Turning to the N 1s spectra of Pt-C₃N₄@CNT material (Figure 2e), the N 1s peak could be deconvoluted into four peaks of 398.8 eV, 399.8 eV, 401.3 eV, and 404.6 eV, which should be interpreted as C=N–C, tertiary nitrogen group ((C)₃–N), free amino group ((C)₂–NH, C–NH₂), and π excitations, respectively [28]. In contrast, the XPS spectra of Pt@CNT (Figure S5) display no N 1s peak, implying the absence of N elements. In the C 1s spectrum (Figure 2f) of Pt-C₃N₄@CNT, five peaks are discernible. Alongside the peaks at 287.5 eV, 284.8 eV, and 283.3 eV that are attributable to the CNT [29], two additional peaks at 286.8 eV (C–N=C) and 286.1 eV (C–N–C) can be traced back to two distinct carbon species within the basic aromatic CN heterocycles [30]. Contrastingly, only three peaks, all attributed to CNT, are present in the C 1s spectrum of Pt@CNT (Figure S6).

To test the electrochemical behaviors of the as-prepared samples, pristine samples were subjected to a cyclic voltammetry (CV) test in Ar-saturated 0.5 M H₂SO₄ solution. In order to investigate the effect of the Pt content on FAOR, samples with a different Pt content were evaluated through linear sweep voltammetry (LSV) tests (Figure S7). The Pt content was modulated at 1%, 2%, 3%, and 4%, with the respective samples being denoted as Pt-C₃N₄@CNT-1, Pt-C₃N₄@CNT-2, Pt-C₃N₄@CNT-3, and Pt-C₃N₄@CNT-4. As depicted in Figure S12, Pt-C₃N₄@CNT-2 exhibited the highest activity for FAOR via the direct pathway, thus denominated as Pt-C₃N₄@CNT. The Pt-C₃N₄@CNT and Pt@CNT samples showed different CV curves (Figures S11 and S12). Firstly, an absence of obvious underpotential deposition of hydrogen (H_upd) peaks was observed for Pt-C₃N₄@CNT, suggesting a dearth of Pt nanoparticles in this configuration. Secondly, the typical Pt–OH reduction peak, generally observed at approximately 0.65 V, was missing in the case of Pt-C₃N₄@CNT [31]. It is well-documented that H_upd would be formed on Pt of step, defect, and terrace sites. The lack of a H_upd peak on Pt-C₃N₄@CNT offers robust evidence corroborating the absence of Pt nanoparticles [31].

The FAOR performances of Pt-C₃N₄@CNT and Pt@CNT were explored in 0.5 M H₂SO₄ and 0.5 M HCOOH solutions. Figure 3a illustrates the superior performance of Pt-C₃N₄@CNT over Pt@CNT via anodic polarization, as evidenced by its elevated current density, particularly at lower potentials. It is noted that the anodic polarization profile of FAOR on Pt-based catalysts typically displays two peaks. The weaker peak at lower potential correlates to the oxidation of HCOOH to CO₂ via the direct pathway, while the stronger peak at a higher potential aligns with the oxidation of CO via the indirect pathway [32]. Furthermore, the anodic polarization curves of FAOR on Pt-C₃N₄@CNT and Pt@CNT were contrasted. The oxidation peak for CO_ads on Pt@CNT appears at ~0.86 V; a peak is nearly absent on Pt-C₃N₄@CNT. This suggests that FAOR primarily proceeds via the direct pathway on Pt-C₃N₄@CNT, underscoring its capacity as a highly active catalyst for FAOR with a strong predilection for the direct pathway. In addition, Cuesta et al. have also reported that the indirect pathway of FAOR to CO necessitates at least three linearly contiguous Pt atoms [33]. Consistent with this, experimental studies have shown that single-Pt-atom catalysts display no electrocatalytic FAOR current, implying that single-Pt-atom catalysts fall short for FAOR [6]. Maurer et al. carried out a comprehensive examination of single-atom platinum supported on CeO₂ [34], shedding light on the dynamic shifts induced during oxidation reactions and their implications for the activity of such catalytic systems. Their findings suggested that the initiation of CO oxidation is associated with the migration of platinum single sites from four-fold hollow sites to form diminutive clusters comprising several platinum atoms [34]. These insights raise the prospect that single-atom catalysts might not necessarily provide the optimal structure for catalysis. In the context of Pt-C₃N₄@CNT catalyst, which features triangular clusters of Pt, the near-complete selectivity of the direct pathway for FAOR can be attributed to the presence of triads of Pt atoms within the structure. In order to demonstrate the results more visually, the mass activity and the specific activity of Pt-C₃N₄@CNT and Pt@CNT at 0.44 V were compared in the bar chart (Figure 3a). Compared with Pt@CNT, Pt-C₃N₄@CNT showed the highest mass activity of 1.91 A ${\mathrm{mg}}_{\mathrm{Pt}}^{-1}$ and specific activity of 0.79 mA cm⁻² at a potential of 0.44 V, signifying 17.4-fold and 2.6-fold enhancements relative to Pt@CNT (0.11 A ${\mathrm{mg}}_{\mathrm{Pt}}^{-1}$ and 0.31 mA cm⁻²), respectively. In addition, the current density of Pt-C₃N₄@CNT normalized by ECSA is 0.086 mA cm⁻² mg, which is significantly higher than that of Pt@CNT (0.016 mA cm⁻² mg). These results provide a compelling testament to the superior activity of Pt-C₃N₄@CNT as a catalyst for FAOR. Importantly, Pt-C₃N₄@CNT catalyst, structured with triangular clusters of Pt atoms, exhibited outstanding selectivity for the direct FAOR pathway, aligning with expectations. CVs of CNT (Figure S13), C₃N₄ (Figure S14), and C₃N₄@CNT (Figure S15) revealed an absence of FAOR current, thereby discounting the potential activity of CNT in FAOR. Furthermore, as shown in Figure S10, the mass activity of Pt@CNT is higher than that of Pt@CNT mixed with C₃N₄. The reason for this may be that the active site of Pt in Pt@CNT material is covered up after the addition of C₃N₄. Against this backdrop, we posit that the enhanced electrocatalytic performance of our catalysts can be chiefly attributed to the unique atomic dispersion structure of Pt-C₃N₄@CNT, which stabilizes the three-atom structure of the Pt cluster, forestalling aggregation into nanoparticles. This structural configuration amplifies catalytic activity and bolsters the utilization of precious metals [20]. Given the triangular clusters of Pt atoms within the Pt-C₃N₄@CNT catalyst, the near-total bias towards the direct pathway can be credited to the presence of two proximate Pt atoms in the Pt-C₃N₄@CNT structure.

To highlight the superior performance of the Pt-C₃N₄@CNT catalyst in formic acid electrooxidation, its catalytic activity was benchmarked against other Pt-based FOR catalysts, as documented in Table S1; the Pt-C₃N₄@CNT catalyst displayed both exceptional and comparable specific and mass activities when compared to other Pt-based catalysts. To further elucidate the resistance to poisoning exhibited by the two samples, a CO stripping experiment was undertaken. The CO stripping curves (Figure 3b,c) reveal the oxidation of adsorbed CO on Pt@CNT at 0.64 V, manifested as a prominent peak, which corroborates previous reports [35,36]. Intriguingly, Pt-C3N4@CNT displays only a minor peak at 0.61 V ascribed to the oxidation of adsorbed CO. The down-shift of the CO oxidation peak suggests an increased electron density on Pt, responsible for an enhancement of the back-donation of the platinum d orbital in the 2π* antibonding orbital of CO, leading to a decrease in the Pt−CO bond strength [37]. Consequently, the triangular Pt clusters embedded within Pt-C3N4@CNT can curb the dehydration of formic acid and CO poisoning and promote the direct dehydrogenation pathway of formic acid [38,39]. To further appraise the resistance to poisoning of the two samples, CO gas was purposefully introduced into the electrolyte in the midst of chronoamperometry (CA) tests (Figure 3d) [40]. The relative reduction in mass activity of Pt-C₃N₄@CNT proved to be substantially less than that of Pt@CNT during CO introduction, thereby confirming the superior anti-poisoning ability of the Pt-C₃N₄@CNT catalyst. Furthermore, the morphology of Pt-C₃N₄@CNT has been preserved after stability tests verified by TEM and HRTEM images in Figures S3 and S4, demonstrating the good durability of Pt-C₃N₄@CNT.

In addition, the electrochemical oxidation reaction kinetics of Pt-C₃N₄@CNT and Pt@CNT catalysts were studied using Tafel plots (Figure 3e). Notably, the Tafel plot for Pt-C₃N₄@CNT shows a smaller slope (261 mV dec⁻¹) than that of Pt@CNT (314 mV dec⁻¹) at lower potentials, suggesting that at these potentials, Pt-C₃N₄@CNT displays a predominant dehydrogenation pathway and a negligible dehydration pathway [41]. This smaller Tafel slope suggests that a lower overpotential is required to achieve the same current density, implying that Pt-C₃N₄@CNT can attain a higher catalytic current at the same overpotential and thus exhibits faster charge-transfer kinetics for formic acid oxidation compared to Pt@CNT [42]. CO stripping experiments were also employed to estimate the electrochemically specific areas (ECSAs) of the two samples. The ECSAs of Pt-C₃N₄@CNT and Pt@CNT were found to be 240 m² ${\mathrm{g}}_{\mathrm{Pt}}^{-1}$ and 35 m² ${\mathrm{g}}_{\mathrm{Pt}}^{-1}$, respectively. This indicates that the Pt-C₃N₄@CNT exposes a larger number of electrochemically active sites than Pt@CNT, further solidifying the superior electrocatalytic activity and predominant dehydrogenation pathway of Pt-C₃N₄@CNT in formic acid oxidation reaction. To further probe the intrinsic catalytic activity of the samples, the turnover frequency (TOF) of the catalysts during the FAOR was calculated. As depicted in Figure S16, Pt-C3N4@CNT exhibited a remarkably higher TOF of 1.88, compared to Pt@CNT (0.45).

To validate the exceptional performance of Pt-C₃N₄@CNT under practical operating conditions, single-cell tests were executed to holistically assess the electrochemical efficacy of the catalysts [43]. Employing Pt-C₃N₄@CNT and Pt@CNT as anodic catalysts and 60% Pt/C as cathodic catalysts for DFAFCs, Pt-C₃N₄@CNT exhibits a superior mass power density of 206.46 mW ${\mathrm{mg}}_{\mathrm{Pt}}^{-1}$ (Figure 3f), marking a 4.8-fold increase over Pt@CNT at 80 °C (43.18 mW ${\mathrm{mg}}_{\mathrm{Pt}}^{-1}$). These observations indicate the promising potential of Pt-C₃N₄@CNT catalysts in DFAFCs.

DFT calculations were carried out to further elucidate the oxidation mechanism of formic acid on Pt-C₃N₄@CNT and Pt@CNT (Figure 4a,b). Figure 4c–f display the optimized structures of the intermediates in the two pathways. The Gibbs free energy profiles of formic acid oxidation (Figure 4a) unveil that the indirect oxidation pathway on Pt-C₃N₄@CNT spontaneously generates CO* + H₂O(l) from HCOOH*. However, the subsequent steps to form CO* + OH* from CO* + H₂O(l) and CO₂(g) from CO* + OH* are not thermodynamically favorable, exhibiting endothermic reaction energies of 0.61 eV and 1.24 eV, respectively. Hence, the production of CO₂(g) from CO* + OH* represents the rate-determining step. In the direct oxidation pathway, the formation of COOH* from HCOOH* requires only a nominal reaction energy of 0.11 eV. The subsequent conversion of COOH* to CO₂(g) is indeed endothermically, necessitating 1.04 eV of reaction energy. Despite being the rate-determining step, it still demands less energy than its equivalent in the indirect oxidation pathway. Consequently, from a thermodynamic standpoint, the direct oxidation route in Pt-C₃N₄@CNT offers a clear advantage over its indirect counterpart.

Based on the experimental observations and the calculated thermodynamic and kinetic data, we propose that the electrochemical oxidation of HCOOH on the Pt-C₃N₄@CNT surface predominantly follows the direct pathway, with COOH* as the preferred reactive intermediate, via:

$$\mathrm{H}\mathrm{C}\mathrm{O}\mathrm{O}{\mathrm{H}}^{\ast }\to \mathrm{C}\mathrm{O}\mathrm{O}{\mathrm{H}}^{\ast }+{\mathrm{H}}^{+}+{\mathrm{e}}^{-}$$

(1)

$$\mathrm{C}\mathrm{O}\mathrm{O}{\mathrm{H}}^{\ast }\to \mathrm{C}{\mathrm{O}}_2^{\ast }+{\mathrm{H}}^{+}+{\mathrm{e}}^{-}$$

(2)

$$\mathrm{C}{\mathrm{O}}_2^{\ast }\to \mathrm{C}{\mathrm{O}}_2\left(\mathrm{g}\right)$$

(3)

with the overall reaction:

$$\mathrm{H}\mathrm{C}\mathrm{O}\mathrm{O}\mathrm{H}\left(\mathrm{g}\right)+\ast \to \mathrm{C}{\mathrm{O}}_2\left(\mathrm{g}\right)+\ast +2{\mathrm{H}}^{+}+2{\mathrm{e}}^{-}$$

(4)

Figure 4b highlights the Gibbs free energy curves for the direct pathways of the Pt-C₃N₄@CNT and Pt@CNT catalysts. Remarkably, the energy required for the adsorption of HCOOH(l) on Pt-C₃N₄@CNT is a mere 0.58 eV, significantly less than the 2.16 eV necessitated by Pt@CNT. This underlines that the formic acid oxidation on Pt-C₃N₄@CNT is thermodynamically more favorable.

3. Conclusions

In summary, we have successfully synthesized Pt-C₃N₄@CNT, equipped with triangular clusters of Pt atoms, serving as a high-performance anode electrocatalyst for DFAFCs. The as-prepared Pt-C₃N₄@CNT exhibits a high mass activity of 1.91 A ${\mathrm{mg}}_{\mathrm{Pt}}^{-1}$ and an enhanced specific activity of 0.79 mA cm⁻² at a potential of 0.44 V, which are 17.4- and 2.6-times higher than those of Pt@CNT (0.11 A ${\mathrm{mg}}_{\mathrm{Pt}}^{-1}$ and 0.31 mA cm⁻²), respectively. Importantly, we observed significantly weakened CO adsorption on the surface of Pt-C₃N₄@CNT, leading to its facile CO oxidation. Additional investigations employing single-cell tests with Pt-C₃N₄@CNT as the anode catalyst resulted in an enhanced mass power density of 206.46 mW ${\mathrm{mg}}_{\mathrm{Pt}}^{-1}$. DFT calculations further elucidated the mechanism of formic acid oxidation on Pt-C₃N₄@CNT, indicating a thermodynamic preference for a direct pathway. This nearly complete preference of a direct pathway for FAOR can be ascribed to the presence of adjacent Pt atoms in the Pt-C₃N4@CNT. This study affirms the advantageous role of triangular Pt atom clusters for promoting the direct pathway for FAOR and underscores the unique contribution of the C₃N₄ structure in assembling these advantageous Pt clusters. These findings have substantial implications for the development of advanced DFAFCs, setting a new precedent for future explorations in the field.