Interfacial Covalent Bonds Regulated Electron-Deficient 2D Black Phosphorus for Electrocatalytic Oxygen Reactions

Developing resource-abundant and sustainable metal-free bifunctional oxygen electrocatalysts is essential for the practical application of zinc-air batteries (ZABs). 2D black phosphorus (BP) with fully exposed atoms and active lone pair electrons can be promising for oxygen electrocatalysts, which, however, suffers from low catalytic activity and poor electrochemical stability. Herein, guided by density functional theory (DFT) calculations, an efficient metal-free electrocatalyst is demonstrated via covalently bonding BP nanosheets with graphitic carbon nitride (denoted BP-CN-c). The polarized PN covalent bonds in BP-CN-c can efficiently regulate the electron transfer from BP to graphitic carbon nitride and significantly promote the OOH* adsorption on phosphorus atoms. Impressively, the oxygen evolution reaction performance of BP-CN-c (overpotential of 350 mV at 10 mA cm-2 , 90% retention after 10 h operation) represents the state-of-the-art among the reported BP-based metal-free catalysts. Additionally, BP-CN-c exhibits a small half-wave overpotential of 390 mV for oxygen reduction reaction, representing the first bifunctional BP-based metal-free oxygen catalyst. Moreover, ZABs are assembled incorporating BP-CN-c cathodes, delivering a substantially higher peak power density (168.3 mW cm-2 ) than the Pt/C+RuO2 -based ZABs (101.3 mW cm-2 ). The acquired insights into interfacial covalent bonds pave the way for the rational design of new and affordable metal-free catalysts.


Introduction
Rechargeable zinc-air batteries (ZABs) have been recognized as one of the most promising energy technologies owing to their unique half-closed configuration, large specific capacity (≈820 mAh g −1 based on Zn metal), high energy density (1086 Wh kg −1 based on Zn metal), and good environmental friendliness. [1,2] As the key component of ZABs, bifunctional electrocatalysts for oxygen evolution reaction (OER) and oxygen reduction reaction (ORR) are of importance to determine the charge/discharge kinetics and energy efficiency of ZABs. [3] Currently, noble metalbased (e.g., Pt, Ru, and Ir) electrocatalysts are the most prevalent category for oxygen electrocatalysis. However, their wide applications seriously suffer from the crustal scarcity and high cost. [1,4] In this regard, metal-free catalysts are highly attractive because only low-cost and resource-abundant elements are utilized. Recently, extensive efforts have been devoted to developing carbon-based metal-free catalysts because Developing resource-abundant and sustainable metal-free bifunctional oxygen electrocatalysts is essential for the practical application of zinc-air batteries (ZABs). 2D black phosphorus (BP) with fully exposed atoms and active lone pair electrons can be promising for oxygen electrocatalysts, which, however, suffers from low catalytic activity and poor electrochemical stability. Herein, guided by density functional theory (DFT) calculations, an efficient metal-free electrocatalyst is demonstrated via covalently bonding BP nanosheets with graphitic carbon nitride (denoted BP-CN-c). The polarized PN covalent bonds in BP-CN-c can efficiently regulate the electron transfer from BP to graphitic carbon nitride and significantly promote the OOH* adsorption on phosphorus atoms. Impressively, the oxygen evolution reaction performance of BP-CN-c (overpotential of 350 mV at 10 mA cm −2 , 90% retention after 10 h operation) represents the state-of-the-art among the reported BP-based metal-free catalysts. Additionally, BP-CN-c exhibits a small half-wave overpotential of 390 mV for oxygen reduction reaction, representing the first bifunctional BP-based metal-free oxygen catalyst. Moreover, ZABs are assembled incorporating BP-CN-c cathodes, delivering a substantially higher peak power density (168.3 mW cm −2 ) than the Pt/C+RuO 2 -based ZABs (101.3 mW cm −2 ). The acquired insights into interfacial covalent bonds pave the way for the rational design of new and affordable metal-free catalysts.
the electronic structure of carbon materials can be elaborately regulated by heteroatom doping. Various heteroatom-doped carbon materials have shown remarkable ORR activity at a level comparable to Pt/C (e.g., N-doped graphene, [5] B-doped carbon nanotubes, [6] P-doped graphite, [7] and S-doped graphene [8] ). Nevertheless, the employment of carbon-based metal-free catalysts in ZABs has been restricted by their unsatisfying OER performance, including the much higher overpotential than the benchmark catalysts (i.e., Ir/C and RuO 2 ) [9] and the poor catalytic stability due to the carbon corrosion at high anodic potentials. [10] Therefore, developing metal-free bifunctional catalysts based on other resource-abundant elements is critically important to construct cost-efficient and high-performance ZABs.
Recently, 2D black phosphorus (BP) has emerged as a fascinating multifunctional material owing to its appealing features, such as the puckered-honeycomb configuration, large surface-tovolume ratio, tunable bandgap (0.3-2.2 eV), and high charge carrier mobility (≈1000 cm 2 V −1 s −1 ). [11,12] Particularly, the active lone pair electrons of P atoms offer favorable chemisorption sites for the O-containing species (e.g., OH*, O*, and OOH*), empowering BP as a potential catalyst for OER. [13] In an ideal case, all P atoms of single-layer BP can be exposed to the surface and serve as active catalytic sites. [14,15] However, the OER activity of the early reported 2D BP-based metal-free catalysts is still far from satisfactory (overpotential of >550 mV at 10 mA cm −2 ), [14] [16] due to the low adsorption/dissociation kinetics of O-intermediates on BP. Moreover, the active lone pair electrons make BP prone to be oxidized, especially when the OER process provides a strong oxidative environment with a high anodic potential. [17,18] Thus, exploiting effective strategies to regulate the electronic structure of BP will be highly desirable for improving the OER activity and stability. Besides, the development of BP-based metal-free bifunctional catalysts toward OER and ORR remains a blank to be filled.
Herein, we report the first BP-based metal-free bifunctional oxygen electrocatalyst (denoted BP-CN-c) by covalently bonding BP with graphitic carbon nitride (g-C 3 N 4 ). Polarized PN covalent bonds are unveiled to regulate the electron redistribution among the heterointerfaces, inducing electron transfer from P atoms to g-C 3 N 4 . The manipulated electron-deficient feature empowers BP with considerably enhanced OOH* chemisorption capability and chemical stability. Remarkably, the BP-CN-c catalyst presents a high OER activity with a small overpotential of 350 mV at 10 mA cm −2 , which outperforms all the previously reported BP-based metal-free catalysts. Moreover, BP-CN-c presents excellent OER stability with only 10% current loss after 10 h continuous operation. Besides, the BP-CN-c catalyst achieves a superb ORR performance in 0.1 m KOH solution, depicting a halfwave overpotential (390 mV), well comparable to the stateof-the-art Pt/C. As a result, ZABs are constructed to demonstrate the feasibility of BP-CN-c in practical applications. A peak power density of 168.3 mW cm −2 is reached by ZABs based on the BP-CN-c cathodes, which significantly outweighs those based on the commercial Pt/C+RuO 2 catalyst (101.3 mW cm −2 ).

Results and Discussion
Moderate adsorption (neither too strong nor too weak) of adsorbates on catalytic sites is generally considered as the key prerequisite for an outstanding electrocatalyst. To understand the inferior OER performance of pristine BP, we first performed density functional theory (DFT) calculations to elucidate the O-intermediates binding energies (ΔG) on BP during the OER process (Figure 1a). Except for the step from O* to OOH*, all the other steps of the OER process are spontaneous at the potential (U) of 1.23 V. The formation of OOH* with a large energy barrier of 2.81 eV is thereby the potential ratedetermining step (RDS) of the OER process. Moreover, calculations are performed for oxygen-terminated BP (denoted BP-O), as BP-O is recognized to be the stabilized configuration of BP when exposed to air. [19] Similarly, the formation of OOH* is confirmed as the RDS for the OER process on BP-O, showing a large energy barrier of 3.25 eV at U = 1.23 V ( Figure S1, Supporting Information). These results suggest that modulating the local electronic structure of BP to promote the OOH* adsorption is the key to boost its OER performance.
Inspired by the heteroatom doping strategy for carbon-based electrocatalysts, we hypothesize that covalently bonding BP surface to other heteroatoms would alter the electron redistribution and thus tailor the adsorption energies of O-intermediates on BP. In this scenario, graphitic carbon nitride (g-C 3 N 4 ) was selected to react with BP sheets because the rich pyridinic-N and graphitic-N sites allow the formation of polarized PN covalent bonds at the BP/g-C 3 N 4 interfaces. Figure 1b illustrates the structural configuration of our developed BP-CN-c catalyst. Few-layer BP nanosheets with lateral sizes of 1-10 µm ( Figure 1c; Figure S2, Supporting Information) and thicknesses of 1.3-9.5 nm (mean thickness of 3.7 ± 1.3 nm) were first prepared via an electrochemical delamination method reported in our previous work. [12] Subsequently, g-C 3 N 4 nanosheets were grown on multiwalled carbon nanotubes (CNTs) to enable a good electrical conductivity (the hybrid is denoted CN). To promote the formation of PN covalent bonds at the BP/g-C 3 N 4 interfaces, BP nanosheets and CN were mixed thoroughly and then subjected to annealing at 300 °C under N 2 atmosphere. The annealing temperature of 300°C was selected to ensure the formation of rich PN covalent bonds ( Figure S3, Supporting Information), while avoiding the thermal decomposition of g-C 3 N 4 and BP ( Figure S4, Supporting Information).
Transmission electron microscopy (TEM) images ( Figure 1d; Figure S5, Supporting Information) verify the uniform coating of ultrathin g-C 3 N 4 nanosheets on the surface of CNTs. Fourier transform infrared (FTIR) spectroscopy ( Figure S6, Supporting Information), X-ray diffraction ( Figure S7, Supporting Information), and SEM images ( Figure S8, Supporting Information) also manifest the successful growth of g-C 3 N 4 nanosheets on the surface of CNTs. Such a 1D g-C 3 N 4 /CNT hybrid can serve as the spacer to prevent the re-stacking when coupled with BP nanosheets, thus enabling the maximum formation of BP/g-C 3 N 4 heterointerfaces. As revealed by TEM images (Figure 1e,f) and energy dispersive X-ray (EDX) spectroscopy elemental mapping ( Figure S9, Supporting Information), CN is homogenously and densely distributed over BP nanosheets, exposing abundant heterointerfaces. Moreover, BP-CN-c exhibits a specific surface area of 119 m 2 g -1 ( Figure S10, Supporting Information).
To highlight the vital role of PN covalent bonds in regulating the electron redistribution at the BP/g-C 3 N 4 interface, physically mixed BP and CN (denoted BP-CN-p) was prepared as a control sample. X-ray photoelectron spectroscopy (XPS) provides direct evidence for the formation of PN bonds in BP-CN-c. In addition to P-P 2p 3/2 peak (130.3 eV), P-P 2p 1/2 peak (131.1 eV) peak, and a broad oxidation peak (133.6 eV) of BP, NP (134.7 eV), and NPO (135.9 eV) peaks are detected in the P 2p XPS of BP-CN-c (Figure 2a). [20] Of note, these two peaks were not observed for BP-CN-p. Additionally, N 1s XPS ( Figure S11, Supporting Information) and FTIR spectra ( Figure S6, Supporting Information) also confirm the formation of PN bonds in BP-CN-c, rather than in BP-CN-p. The strong interfacial interaction in BP-CN-c via PN bonds induces the negative shift of the N-C peak in C 1s XPS spectra by 0.3 eV, which is a sign of electron transfer from BP to g-C 3 N 4 ( Figure S12, Supporting Information). [18,21] Figure 2b compares the Raman spectra of BP, BP-CN-p, and BP-CN-c, which all display three characteristic peaks located at 362 (A 1 g out-of-plane vibrational mode), 438 (B 2g armchair vibrational mode), and 464-466 cm −1 (and A 2 g zig-zag vibrational mode). [22] In comparison to BP and BP-CN-p (464 cm −1 ), BP-CN-c (466 cm −1 ) shows an apparent positive shift in the A 2 g peak, implying that the interfacial interaction in BP-CN-c introduces strong strain into the BP plane. [23] The P L 2,3 -edge X-ray absorption near edge structure spectra (XANES) spectra further provide insights into the local electronic structures of P atoms in different samples (Figure 2c). Peaks ranging from 135.0 to 141.0 eV refer to transitions from 2p electrons of P into the first unoccupied 3s antibonding state. Among, a peak centered at 137.1 eV can be observed for BP, BP-CN-p, and BP-CN-c, corresponding to PO species formed by the slight oxidation of BP. [24] Unlike BP and BP-CN-p, BP-CN-c displays a distinguishable peak at 138.9 eV, which supports the formation of PN covalent bonds. [25] Besides, DFT simulations were performed to investigate the atomic/electronic structure of the PN bonded BP/g-C 3 N 4 hybrid (Figures S13 and S14, Supporting Information). Figure 2d displays the most stable atomic configuration with computed charge-density difference isosurfaces. As expected, the optimized BP/g-C 3 N 4 structure is featured with a slightly deformed BP plane and noticeable electron depletion in P atoms, which agrees well with the experimental analysis. Thereby, we can conclude that the polarized PN covalent bonds can significantly influence the electron redistribution at BP/g-C 3 N 4 interfaces, which produces electron-deficient BP planes. In addition, it was recognized that the lone pair electrons of sp 3 -hybridized P in BP were susceptible to bond with Adv. Mater. 2021, 33,   chemisorbed oxygen atoms and be oxidized. [16] By forming PN covalent bonds, the active lone pair electrons of P in BP-CN-c are substantially occupied, thus prominently promting the stability of BP-CN-c ( Figure S15, Supporting Information).
The modified electronic structure of P atoms in BP-CN-c motivated us to assess the OER catalytic performance of BP-CN-c. It should be noted that, to obtain the optimal catalytic performance, the urea-to-CNTs ratio used for synthesizing CN (Table S1 and Figure S16, Supporting Information) and the BP-to-CN ratio used for synthesizing BP-CN-c ( Figure S17, Supporting Information) were rationally optimized. The significiant role of CNTs in BP-CN-c is also highlighted by the poor catalytic performance of the directly hybrided BP and g-C 3 N 4 ( Figure S18, Supporting Information). Figure 3a presents the linear sweep voltammetry (LSV) polarization curves of CN, the mixture of BP and CNTs (denoted as BPC), BP-CN-p, BP-CN-c, and commercial RuO 2 in 1 m N 2 -saturated KOH solution. Clearly, the as-prepared BP-CN-c catalyst outperforms all other metal-free counterparts and the benchmark RuO 2 catalyst. A current density of 10 mA cm −2 is achieved for the BP-CN-c catalyst with an overpotential of 350 mV, while the overpotentials for CN, BPC, BP-CN-p, and RuO 2 are 650, 520, 430, and 385 mV, respectively (Figure 3a). The smallest Tafel slope (81 mV dec −1 , Figure 3b) and the highest electrochemically active surface area (16.58 mF cm -2 , Figures S19 and S20, Supporting Information) of BP-CN-c also reflect its superior OER kinetics and activity to other counterparts. Excellent reproducibility was demonstrated for the OER performance of BP-CN-c ( Figure S21, Supporting Information). Significantly, the OER activity of BP-CN-c represents the best among the reported BP-based OER catalysts (Table S2, Supporting Information).
In addition, the NP bond in BP-CN-c greatly improve the resistance of BP against structure degradation during the longterm OER operation. In a chronoamperometric test at 1.58 V for 10 h, BP-CN-c exhibits high current retention of ≈90%, which substantially outperforms BP-CN-p (63%) and RuO 2 (53%) (Figure 3c). Moreover, BP-CN-c still retains high current retention of 86.3% after the 20 h chronoamperometric test ( Figure S22, Supporting Information). We further examined the stability through a standard accelerated durability test with a scan cycling potential window between 1.3 and 1.7 V versus reversible hydrogen electrode (RHE). After 1000 cycles, the OER overpotential of BP-CN-c at 10 mA cm −2 exhibited a negligible increase of less than 1 mV, which is much better than that of BP-CN-p with an increase of 40 mV ( Figure S23, Supporting Information).
In situ Raman spectro-electrochemistry of BP-CN-c during the OER process elucidates that P atoms are active catalytic sites for the OER (Figure 3d; Figure S24, Supporting Information). When the imposed potential increases from the open circuit potential (OCP, 0.875 V vs RHE) to 1.0 V versus RHE, the slight positive shift of the A g 2 peak can be explained by the compressive strain in BP planes induced by the adsorption of OH − on P atoms. [26] A further positive shift can be observed for the A g 2 peak at a potential exceeding the thermodynamic OER potential (1.4 V vs RHE), which can be assigned to the conversion of the adsorbed O-intermediates. [23] DFT calculations further provide insights into how PN bonds promote the OER performance of the BP-CN-c catalyst. Figure 3e displays the calculated density of states (DOS) for the PN bonded BP/g-C 3 N 4 . Impressively, the Fermi level (EF) of the BP/g-C 3 N 4 heterostructure lies inside the bands, indicating the metallic feature of BP/g-C 3 N 4 . The BP/g-C 3 N 4 interface produces interface states that are  composed of N 2p and P 3p states. This is due to the covalent bonding between N and P in the BP/g-C 3 N 4 heterostructure. The appearance of interface states in the bandgap results in the metallic feature, which can significantly improve the catalytic performance of the BP/g-C 3 N 4 heterostructure compared with semiconducting BP and g-C 3 N 4 ( Figure S25, Supporting Information). The free energy diagrams of the OER process on BP and BP/g-C 3 N 4 hybrid at U = 0 and 1.23 V are summarized in Figure 3f. As discussed earlier, the formation of OOH* is regarded as the RDS for BP. In comparison with BP (4.04 eV at U = 0 V and 2.81 eV at U = 1.23 V), the BP/g-C 3 N 4 hybrid presents a much lower energy barrier for the OOH* formation (2.85 eV at U = 0 V and 1.62 eV at U = 1.23 V), significantly promoting the conversion from O* to OOH*. Similar calculations were also performed for BP-O and the BP-O/g-C 3 N 4 hybrid, which offer the same conclusion for the role of PN bonds ( Figures S26 and S27, Supporting Information).
Furthermore, the BP-CN-c catalyst shows a comparable ORR catalytic activity to the commercial Pt/C in an O 2 -saturated 0.1 m KOH solution. The cyclic voltammetry curve ( Figure S28, Supporting Information) of BP-CN-c shows a distinct cathodic peak centered at ≈0.80 V in the O 2 -saturated 0.1 m KOH electrolyte, indicating the pronounced electrocatalytic activity for ORR. LSV curves at 1600 rpm indicate that the BP-CN-c catalyst achieves a half-wave potential of 0.84 V versus RHE and a diffusion-limiting current density of 5.34 mA cm −2 (Figure 4a). This catalytic performance is well comparable to that of Pt/C (0.85 V vs RHE, 5.60 mA cm −2 ), and considerably better than those of BPC (0.68 V vs RHE, 1.92 mA cm −2 ), CN (0.81 V vs RHE, 5.15 mA cm −2 ), and BP-CN-p (0.80 V vs RHE, 4.55 mA cm −2 ).
In addition, the good ORR reproducibility of the BP-CN-c catalyst is supported by the LSV curves collected from four RDEs with BP-CN-c prepared from the same batch and BP-CN-c prepared from different batches ( Figure S29, Supporting Information). Meanwhile, the electron transfer number of ORR on BP-CN-c was measured by rotating ring-disk electrode measurements ( Figure S30a,b, Supporting Information). A closeto-4-electron (3.87-3.92 electrons) pathway could be verified at 0.8-0.1 V versus RHE, demonstrating that O 2 was almost completely reduced to H 2 O (Figure 4b). RDE polarization curves at a variety of rotating speeds were also collected for BP-CN-c, which identify the first-order reaction kinetics of BP-CN-c concerning the concentration of dissolved oxygen ( Figure S30c, Supporting Information). Moreover, BP-CN-c exhibited a remarkable ORR durability with a negligible current loss (3%) at 0.7 V versus RHE for 10 h ORR operation (Figure 4c). This result stands in contrast to the poor durability of BP-CN-p and Pt/C catalysts with large current decay of 23% and 55%, respectively. The superior ORR activity of the BP-CN-c catalyst can be attributed to the accelerated electron transport of the PN bonded BP/g-C 3 N 4 hybrid and the reduced energy barriers of ORR steps occurring on carbon atoms of g-C 3 N 4 ( Figure S31, Supporting Information).
Rechargeable ZABs were subsequently assembled by employing BP-CN-c as the catalyst for cathodes (Figure 5a). For the sake of comparison, reference ZABs were assembled by using Pt/C+RuO 2 catalyst (Pt/Ru 1:1 mass ratio) for cathodes. The open-circuit voltage of BP-CN-c-based ZABs achieved 1.47 V (inset of Figure 5b; Figure S33, Supporting Information). Due to the excellent bifunctional activity of BP-CN-c, BP-CN-c-based ZABs exhibited apparently narrower charge/discharge voltage gaps than Pt/C+RuO 2 -based ZABs in a broad current density range (1-180 mA cm −2 ) (Figure 5b). Much higher discharging plateaus were also detected for BP-CN-c-based ZABs than Pt/ C+RuO 2 -based ZABs at the same current densities ( Figure S34 (Table S3, Supporting Information). Besides, the specific capacity and energy density of BP-CN-c-based ZABs reached 793.9 mAh g −1 and 952.7 Wh kg −1 (based on the mass of Zn metal) at a discharge current density of 5 mA cm −2 , which also outclasses those of Pt/C+RuO 2based ZABs (751.3 mAh g −1 and 826.4 Wh kg −1 ) ( Figure S37, Supporting Information).
The cycling stability of the fabricated ZABs was further examined at a current density of 10 mA cm −2 with 20 min per cycle (10 min for discharge, 10 min for charge). In the first cycle, the discharge and charge voltages of BP-CN-c-based ZABs were 1.10 and 1.95 V, respectively (Figure 5d). After 900 cycles (operation for 300 h), the devices still featured a stable discharge voltage of 1.09 V and a charge voltage of 1.98 V. In sharp contrast, battery failure was observed for Pt/C+RuO 2based ZABs after 700 cycles. Finally, two BP-CN-c-based ZABs were connected in series, by which a green light emitting diode (LED) with an operating voltage of 2.5 V could be successfully lighted up (Figure 5e). This result demonstrates the feasibility of BP-CN-c-based ZABs for practical use.   [27][28][29][30][31][32][33] The dotted lines refer to ΔE at the same values.

Conclusion
We have developed the first 2D BP-based metal-free bifunctional oxygen catalyst. Experimental studies accompanied with DFT calculations corroborated the vital role of interfacial PN bonds in regulating the electron redistribution between BP/g-C 3 N 4 interfaces and thus the chemisorption properties of O-intermediates. The as-developed BP-CN-c catalyst achieved the best OER catalytic performance among BP-based metal-free catalysts and a comparable ORR catalytic activity to the commercial Pt/C catalyst. Additionally, the application of BP-CN-c for ZAB cathodes was also demonstrated, which considerably surpasses the benchmark Pt/C+RuO 2 catalyst. This work emphasizes that interfacial covalent bond engineering can be a powerful strategy for devising bifunctional metal-free hybrid catalysts required by diverse energy storage and conversion applications.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.