A bifunctional two dimensional TM3(HHTP)2 monolayer and its variations for oxygen electrode reactions

To achieve renewable energy technologies, low-cost electrocatalysts for the oxygen reduction reaction (ORR) and the oxygen evolution reaction (OER) are required to replace Pt and IrO2/RuO2 catalysts. Based on density functional theory, the catalytic activity of TM3(HHTP)2 (2,3,6,7,10,11-hexahydroxytriphenylene) monolayer and its variations (TMX4, where TM 1⁄4 Fe, Co, Ni, X 1⁄4 O, S, Se) for bifunctional ORR/OER have been investigated. The adsorption ability is dominated by the metal center, in the order of Fe > Co > Ni while the ligand shows the minor contribution. Due to the presence of linear relations between the intermediates, the activity of TMX4 for the ORR/OER follows a dual volcano curve as a function of the OH adsorption strength. Considering the overpotential, CoO4 and CoS4 possess superior bifunctional activity, implying their promise as candidates for the oxygen electrode reaction. This systematical work may open new avenues for the development of high-performance non-PGM catalysts for practical applications of ORR and OER.


Introduction
There is growing interest in oxygen electrochemistry as conversions between O 2 and H 2 O play important roles in renewable energy technologies, such as the rechargeable air based battery and devices that require two key electrochemical reactions, oxygen reduction (ORR) and oxygen evolution (OER). 1,2 The electrocatalytic oxygen reduction reaction (ORR) and oxygen evolution reaction (OER) play key roles in such renewable energy devices. 3 The current spectrum of catalysts utilized for these fundamental electrochemical reactions are Pt for the ORR and IrO 2 for applications in the OER. [4][5][6] Their "rare earth" status and associated high cost renders them less than ideal materials for incorporation into commercialization. In addition, the use of two different single function catalysts for the ORR and OER, respectively, makes the air cell signicantly more complex as it requires the combination of three electrodes. In this regard, the development of active and affordable bifunctional electrocatalysts remains a challenging task. 1 As alternatives, great efforts have been devoted to the development of the functional carbon based materials with specic atomic conguration where the heteroatom-doped, such as the nonmetallic as well as the nonprecious transition metal elements, would activate the inert C material and further boost the ORR/OER activity. [7][8][9] Typical example is shown by J. D. Baran et al. that phthalocyanines, porphyrins and their variations with the active sites composed of TMN 4 motif could be acted as the bifunctional catalysts. 10 Furthermore, S. Z. Qiao et al. demonstrate that the TMN 2 embedded in g-C 3 N 4 promotes the oxygen electrode reactions. 2 Based on the mentioned results, the performance is obviously tuned by the selection of TM atom. However, limited investigations have been focused on the inuence of the different TM/ligand combination on the activity.
The metal-organic framework (MOF) provide the abundant active sites due to the structural exibility where the high level of individually coordinated metal and the wide selection of building block. 11,12 Besides the TM coordinated with N ligand, other combinations within MOF have been experimentally synthesized, [13][14][15][16] such as the uniform distribution of TMS 4 and TMO 4 motifs, 14,16 which could offer the prototype for the investigation of the TM/ligand effect. On the other hand, inspired by the attractive bifunctional electrocatalysis exhibited by the TM dichalcogenides, 17-20 the active centers consisted by the TM coordinated to the O/S/Se atoms have raised our attention as the ORR/OER electrode.
To classify the effect of the mentioned combinations, the primary consideration is the theoretical model where the TM 3 (HHTP) 2 monolayer is selected as the prototype in our investigation. 16 In the regard, DFT calculations are used within the electrochemical framework to analyze the ORR/OER reaction. The TM 3 (HHTP) 2 prototype and its variations have been systemically studied to illustrate the critical role of the metal/ligand combination, where the schematic monolayer structures are shown in Fig. 1(a). The corresponding stability of the reaction intermediates is considered, which allows for the evaluation of the free energy and overpotentials. Based on the information, the bifunctional candidates are screened out by a thorough comparison.

Computational method
All calculations are performed within the DFT framework as implemented in DMol 3 code. 21,22 The generalized gradient approximation with the Perdew-Burke-Ernzerhof (PBE) functional is employed to describe exchange and correlation effects. 23 The DFT semi-core pseudopots (DSPP) core treat method is implemented for relativistic effects, which replace core electrons by a single effective potential and introduce some degree of relativistic correction into the core. 24 The double numerical atomic orbital augmented by a polarization function (DNP) is chosen as the basis set. 21 A smearing of 0.005 Ha (1 Ha ¼ 27.21 eV) to the orbital occupation is applied to achieve accurate electronic convergence. In order to ensure high-quality results, the real-space global orbital cutoff radius is set as high as 5.2Å. In the geometry structural optimization, the convergence tolerances of energy, maximum force and displacement are 1.0 Â 10 À5 Ha, 0.002 HaÅ À1 and 0.005Å, respectively. The spin-unrestricted method is used for all calculations. A conductor-like screening model (COSMO) was used to simulate a H 2 O solvent environment for all calculations, 25 which is a continuum model where the solute molecule forms a cavity within the dielectric continuum. The DMol 3 /COSMO method has been generalized to periodic boundary cases. The dielectric constant is set as 78.54 for H 2 O. Some previous results have shown that this implicit solvation model is an effective method to describe solvation. 15,26 The 15Å-thick vacuum is added to avoid the articial interactions between the nanosheet and its images.
In the reaction energy landscape, all ORR/OER intermediates are described as proton/electron (H + + e À ) transfers. 4,5 The adsorption energy of the corresponding intermediates are calculated by the following, 10 where the chemical potential of proton/electron (H + + e À ) in solution is equal to the half of the chemical potential of a gas-phase H 2 . 4 The DG for every elemental step can be determined as following: where DE is the electronic energy difference based on DFT calculations, DZPE is the change in zero point energy, T is the temperature (equal to 298.15 K here), DS is the change in the entropy, and DG pH and DG U are the free energy contributions due to variation in pH value (pH is set as 0 in acid medium) and electrode potential U, respectively. In order to decrease the calculation consumption, the approximate correction DZPE À TDS to DE (0.05/0.30/0.35 eV of O*/OH*/OOH*) are used for constructed the DG. As OER are reverse process of the ORR, the corresponding DG of OER intermediates are calculated in the following equation: The thermodynamic CHE model has been applied to interpret the experimental data and design the novel electrocatalysts for metal, oxides as well as carbon-based materials. 10,15,[27][28][29][30][31][32] Besides, the present computational method has been applied to illustrate the ORR mechanism of the TM 3 (HITP) 2 (HITP ¼ hexaiminotriphenylene) monolayer. Therein, the 2e À mechanism is predicted to be prevalent for Ni 3 (HITP) 2 system owing to the insufficient O 2 activation, which is in accordance with the experimental data established by E. M. Miner. 33 Therefore, considered the structural similarity of the TM 3 (HITP) 2 and TM 3 (HHTP) 2 systems, the reliability of our calculation could be conrmed. However, it should be realized that the material stability under the harsh electrochemical environment has been neglected herein. Despite the bifunctional ORR/OER electrocatalytic candidates have been screened out based on our theoretical trend, further performance needs the experimental conrmation.

Results and discussion
The favorable adsorption properties of the ORR/OER intermediates are the prerequisite for the reaction proceeding. The corresponding adsorption energies are tabulated in Table 1. It should be point out that the values do not signify the absolute strength of the intermediates adsorption. As shown, the adsorption energy decreases monotonically with increasing the d-electron in the valence shell of the metal center, which could be accounted by the d-band model. 34 That is, the adsorption ability is tuned by the variation of the metal center, following the order of Fe > Co > Ni. 2,15,35 In order to reveal the ligand effect, the corresponding Mulliken charges of the TM active center as well as the ORR intermediates are plotted in Fig. 2 where positive and negative represent charge depletion and accumulation, respectively. As shown in Fig. 2(a), the charge is transferred from the TM atom to the C-based skeleton for the TMO 4  and FeSe 4 , the generally tendency of the charge accumulation before adsorption changed toward to the charge depletion with ORR intermediates adsorption is observed. Herein, the ligand effect of the charge distribution possesses the similar behavior on the certain degrees for the TMS 4 and TMSe 4 , as implied by Fig. 2(a). The point is further supported by the partially density of states (PDOS). For CoX 4 as shown in Fig. 3, the appearance of the p-orbital ranged from À3 eV to the Fermi energy is observed for CoS 4 and CoSe 4 , which is missed for the CoO 4 . Besides, the sharp double d-peaks are located at $À1.6 eV and À1.0 eV for the former systems, being different from the d-peaks of the CoO 4 , which conrms the variation of the d-orbital as the ligand charged. The similarity is found for NiX 4 and FeX 4 , where the detail PDOS plots are shown in Fig. S1 and S2 of the ESI. † As discussed, the charge analysis as well as PDOS morphology demonstrates the complex ligand effect on the subtle electronic structure of the TM active center, being dependent on the TM element selection, which would lead to the variation of the adsorption energy. Taken Co combination as an example, the neglect variation is found for E ORR (O) as the ligand changes.
However, the E ORR (OOH) and E ORR (OH) are signicantly weakened as the ligand varied from O/S to Se. The situation is completely different for the Ni/Fe combination. As shown in Fig. 2, due to the charge transfer, the ionic bonding between the O-containing species is formed. 36 The charge accumulations of the Co active center and the O-containing species are observed, implying the presence of electrostatic repulsion, which is different from the situations between Ni/Fe and the ORR intermediates with the electrostatic attraction. 37,38 It is plausibly accounted for the mentioned weakening phenomenon for CoX 4 .
On the other hand, it should been noted that the covalent bonds are dominated by the interaction of the O-p orbital and the TMd orbital, in combination with the minor contribution of the p-p orbital overlap demonstrated in Fig. 3 as well as Fig. S1 and S2. † Herein, both covalent bonds due to the overlap of orbitals and ionic bonds induced by the charge transfer inuence the adsorption. 36 Due to the difference of the electronic structures, the general trend of the ligand effect is difficultly summarized. The data could be plotted as a function of OH adsorption, as shown in Fig. 1(b). From our results, the universal linear relationships between the ORR intermediates are clearly observed, which is in agreement with the previous studies. 10,15,35,39 That is, Compared with the previous data of the porphyrins analogues, 10 it is found that the slopes of the tting lines are comparable. It should be noted that the data deviation of CoSe 4 [E ORR (O) ¼ À2.45 eV, E ORR (OH) ¼ À3.83 eV] from the scaling relation between E ORR (O) and E ORR (OH) is obvious. Such deviation could be observed in the C-based electrocatalysts, 10,15,35 which is caused by the different electron transfer required for ORR intermediates adsorption (formally 2e/1e for O/OH, respectively). 10 Herein, our Mulliken charge analysis shows the consistence with the statement where the adsorbed O possesses the more electrons compared with the adsorbed OH, indicated by the values shown in Fig. 2(b). Furthermore, the intercept of E ORR (OOH) vs. E ORR (OH) generally approaches 3.2 eV, regardless the catalytic materials. 10,31,35,39 Herein, it is well-known that the mentioned scaling relations allow the dependence of the ORR activity on the adsorption strength that too strong means the poisoning of the O-containing species whereas too weak implies the insufficient activation ability, both of which is considered as the origin the ORR overpotential. 4,40 To evaluate the activity of the mentioned systems, the OOH associative mechanisms are taken into consideration with the elemental steps R i listing in the following, 10,35 where asterisks denote active TM sites. Due to the small barrier of proton transfer, which could be ignored at high applied voltages, our attentions are only are focused on the reaction energies. 4,15,35,41 The corresponding free energy value G is analyzed and depicted in Fig. 4. As shown by the following equations, the four-electron ORR pathways in OOH association mechanism can proceed through OOH formation (R 1 ), O formation (R 2 ), OH formation (R 3 ) and H 2 O formation (R 4 ).   From the gures, for NiO 4 monolayer, the whole elemental steps are exothermic at the potential U of 0 V, indicating the thermodynamic favor. However, as the U is raised to 1.23 V, the situation is changed that (R 1 ) and (R 2 ) become endothermic, respectively, implying the ORR reaction would be not proceeded spontaneously. For clearly observation, the DG values are gathered in Table 2. Herein, the rate-determining step (RDS) with the largest DG max value could be acted as a measure of the catalyst activity. 4 The RDS is located at (R 2 ) with DG max of 0.82 eV for NiO 4 monolayer at 1.23 V. Furthermore, other catalytic materials possess good activity at 0 V without the endothermic reaction steps while the unfavorable thermodynamics are observed at 1.23 V, being similar with NiO 4 monolayer. However, it should be noted that two different RDS could be identied. That is, (R 1 ) is for NiS 4  Based on the free energy proles, the highest potential for the feasible thermodynamic ORR steps are obtained and its  As discussed by the previous reports, 10,15,35,39 the ORR activity depends on the adsorption of the intermediates. Due to the linear relationship between the adsorption of ORR intermediates and the OH, the overpotential m ORR as a function of the E ORR (OH) is described in Fig. 6. As the enhancement of the adsorption ability, the m ORR reduces and then increase, demonstrating the classical volcano-shaped activity is found. 2,4,10,35 For NiO 4 , NiS 4 , NiSe 4 and CoSe 4 with weak adsorption strength, the high m ORR originates from the ineffective weakening the O-O coupling. For Fe combination located at the branch of the strong adsorption, the OH poisoning accounts for the increased m ORR . Due to the suitable adsorption ability, the CoO 4 and CoS 4 are situated at the apex of the volcano curve. Our results are in accordance with the previous reports that the bond strength should be compromised for the effective ORR catalysts on the basis of Sabatier principle. 40 Besides the ORR activity, the OER activity is characterized in Fig. 5 where the reversed process of the OOH associative mechanisms is considered. The free energy of OER intermediates are obtained by the eqn (5) and the corresponding data are  Table 3 and its E ORR (OH)-dependence is shown in Fig. 6 where the volcano-curve is roughly observed. Based on the mentioned results, CoX 4 show the bests activity with the exception of CoSe 4 due to the too weak capture ability of the reaction intermediates. Herein, the CoO 4 and CoS 4 are identied as the high efficient electrocatalysts to replace Pt for ORR and IrO 2 /RuO 2 for OER. Besides, FeX 4 provides better activity in comparison with NiX 4 . Generally, the reversible ORR/ OER activity is mainly relied on the selection of the metal center. The different ligand gives slightly tuned. Herein, our stimulation provides the potential candidates for experimental synthesis. However, the structural stability as well as its conductivity are not concerned and out of our scope. Furthermore, due to the structure-dependence property, the model selection leads to distinct possibility for electrocatalysis application. Therefore, due to the limited work, extending our results to the whole MOF systems is not suitable.

Conclusion
Based on density functional theory, ORR/OER activity on a TMX 4 monolayer has been systematically studied. It is found that the combination of the metal center and the ligand affects the ORR/OER bifunctional activity where the classical volcanocurve as a function of E ORR (OH) is roughly observed. Furthermore, based on the overpotential obtained from the free energy proles, the ORR activity follows the order of CoS 4 z CoO 4 > FeS 4 z FeSe 4 > FeO 4 z CoSe 4 > NiO 4 z NiSe 4 z NiS 4 while the OER activity follows the order of CoS 4 z CoO 4 > FeO 4 > CoSe 4 ¼ FeS 4 > FeSe 4 z NiSe 4 > NiS 4 > NiO 4 , suggesting that CoO 4 and CoS 4 exhibit the superior catalytic activity. These results may serve as guidance for rational material design and synthesis.

Conflicts of interest
There are no conicts to declare.