How to Boost the Activity of the Monolayer Pt Supported on TiC Catalysts for Oxygen Reduction Reaction: A Density Functional Theory Study

Developing the optimized electrocatalysts with high Pt utilization as well as the outstanding performance for the oxygen reduction reaction (ORR) has raised great attention. Herein, the effects of the interlayer ZrC, HfC, or TiN and the multilayer Pt shell on the adsorption ability and the catalytic activity of the TiC@Pt core-shell structures are systemically investigated by density functional theory (DFT) calculations. For the sandwich structures, the presence of TiN significantly enhances the adsorption ability of the Pt shell, leading to the deterioration of the activity whilst the negligible influence of the ZrC and HfC insertion results the comparable performance with respect to TiC@Pt1ML. In addition, increasing the thickness of the Pt shell reduces the oxyphilic capacity and then mitigates the OH poisoning. From the free energy plots, the superior activity of TiC@Pt2ML is identified in comparison with 1ML and 3ML Pt shell. Herein, the improved activity with its high Pt atomic utilization makes the potential TiC@Pt2ML electrocatalyst for the future fuel cells.


Introduction
Proton exchange membrane fuel cells (PEMFC) have attracted widespread attention due to their high efficient and zero carbon emission for the hydrogen economy [1][2][3][4]. To accelerate the sluggish kinetics of the oxygen reduction reaction (ORR) at the cathode, commercial catalysts consist of platinum deposited on a carbon support [5]. However, the high loading of the Pt catalysts results in a major challenge for future commercialization [6]. In this regard, the development of efficient catalysts with reduced Pt content is of great importance.
The core-shell structure with a non-Pt core can significantly improve the Pt utilization, thereby reducing the Pt content and, thus, the cost [7,8]. It as previously revealed that when transition metal (TM) elements acted as the core, such as Pd [9,10], Ru [11], or Ir [12], the ORR activity of the corresponding TM@Pt core-shell was enhanced compared with the commercial Pt/C. However, such a TM core would not be suitable from an economic aspect [13]. On the other hand, titanium carbide (TiC) is a good alternative of Pt due to its similar electronic structure [14], being important in the field of catalysis. As reported, the TiC supported Pt catalysts possess the enhanced performance of the methanol oxidation reaction, hydrogen eVolution reaction, and ORR, implying the positive effect of TiC [15][16][17][18]. It is believed that the TiC core could modify the electronic structure of the corresponding Pt shell to boost the ORR performance. Therefore, the core-shell structure consisting of the TiC core and the Pt shell acting as the ORR cathode could be the solution for the future requirements of the PEMFC cathode material.
TiC suffers from stability degradation due to oxide formation during the electrochemical cycles [19]. To settle the issue, increasing the thickness of the Pt shell is a viable strategy to protect the

The Calculation Details
All calculations are performed within the DFT framework as implemented in DMol 3 code [30,31]. The generalized gradient approximation with the Perdew-Burke-Ernzerhof functional (GGA-PBE) is employed to describe exchange and correlation effects [32]. The DFT semi-core pseudopots (DSPP) core treatment is implemented for relativistic effects, which replace core electrons by a single effective potential and introduce some degree of relativistic correction into the core [33]. The double numerical atomic orbital augmented by a polarization function is chosen as the basis set [30]. Herein, the PBE/DNP combination in Dmol 3 code has been widely employed for the ORR electrocatalysis [7,8,[34][35][36]. Furthermore, these parameters have been used for the TiC@Pt or TiN@Pt system [37]. Our calculation method is consistent with the previous works, indicating the feasibility. A smearing of 0.005 Ha (1 Ha = 27.21 eV) to the orbital occupation is applied to achieve accurate electronic convergence. The spin-unrestricted method is used for all calculations. The minimum energy paths for the ORR were obtained by the LST/QST tools in the DMol 3 code.
The TiC@Pt(001) surfaces are modeled as periodically repeated 2 × 2 supercell. A 15 Å-thick vacuum is added along the direction perpendicular to the surface to avoid the artificial interactions between slab and its images. The corresponding structure of TiC@Pt(001) and its derivatives are schematically illustrated in Figure 1. In all of the structure optimization calculations, the atoms in the bottom two layers are fixed while other are fully relaxed.
The adsorption energies E ads (M) are calculated by the following equations: where E M/slab , E M , and E slab are the energies of the adsorption systems, the ORR intermediates and the catalyst, respectively. Gibbs free energy changes (∆G) of the ORR elemental steps have been calculated according to the computational hydrogen electrode (CHE) model developed by Nørskov et al. where the chemical potential of proton/electron (H+ + e − ) in solution is equal to the half of the chemical potential of a gas-phase H 2 [5]. The ∆G for eVery elemental step can be determined as following: where ∆E is the electronic energy difference based on DFT calculations, ∆ZPE is the change in zero point energy, T is the temperature (equal to 298.15 K here), ∆S is the change in the entropy, and ∆GpH and ∆GU 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 ∆ZPE − T∆S to ∆E (0.05/0.35 eV of O*/OH*) are used for constructed the ∆G [5]. and ∆GU 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 ∆ZPE − T∆S to ∆E (0.05/0.35 eV of O*/OH*) are used for constructed the ∆G [5].

Results and Discussion
In order to characterize the adsorption ability, the high-symmetry adsorption sites are considered, including the top, bridge, and hollow sites [7,38]. The favorable adsorption sites are shown in Figure 1 and the corresponding adsorption energies Eads are listed in Table 1. For TiC@Pt1ML, the favorable adsorption site of O2 is the bridge site with the Eads(O2) of −1.83 eV, indicating the efficiency of the O-O activation, in line with the previous work [39]. Similarly, the O and OH are located at the bridge sites with the Eads of −1.27 and −3.44 eV, respectively. The product H2O is suited at the top site and the Eads(H2O) is −0.80 eV, being stronger than the solvation stabilization energy of

Results and Discussion
In order to characterize the adsorption ability, the high-symmetry adsorption sites are considered, including the top, bridge, and hollow sites [7,38]. The favorable adsorption sites are shown in Figure 1 and the corresponding adsorption energies E ads are listed in Table 1 [40]. Comparing the data of E ads (O 2 ) and E ads (H 2 O), the product H 2 O is readily replaced by the reactant O 2 for the next ORR cycle. For the TiC@ZrC@Pt 1ML or TiC@HfC@Pt 1ML , the difference of the E ads is less than −0.05 eV with respect to the TiC@Pt 1ML , indicating the negligible effects of the ZrC or HfC interlayer on the adsorption behavior. However, the binding between the Pt shell and the adsorbates is significantly enhanced by inserting TiN interlayer. Reserving the stable adsorption sites, the corresponding E ads are −2.39, −1.80, −3.82, and −1.07 eV for O 2 , O, OH and H 2 O, respectively. Herein, the presence of TiN boost the oxyphilic ability compared with TiC@Pt 1ML . Such adsorption variation is reasonable that the interaction between the Pt shell and the substrate of the TiC@TiN@Pt 1ML is via the Pt-N bonds, being different from the Pt-C bonds for TiC@Pt 1ML . On the other hand, increasing the Pt shell thickness would weaken the ligand effect of the TiC core, changing the adsorption capability [11]. As listed in Table 1 Figure 2a. As shown, the scaling relationship is clearly observed, in consistence with the previous results [41]. That is: As is well-known, the adsorption strength is correlated with the d band center of the catalysts according to the d band model where the higher (lower) of the d band center referred to the Fermi energy generally corresponds to the stronger (weaker) adsorption ability [42]. Herein, in order to understand the physical origin of the E ads change, the d partial density of states (PDOS) of the Pt surface is plotted in Figure 2b. As shown in the top panel, the d orbital of the sandwich structures are altered by the different interlayers. Therein, the d bands are almost overlapped for the TiC@ZrC@Pt 1ML and TiC@HfC@Pt 1ML while the obvious upshift is observed for TiC@TiN@Pt 1ML . Quantitatively, the d band centers are calculated and listed in Table 2. The corresponding values are −2.78, −2.80, and −2.40 eV for the mentioned systems, respectively. That is, the enhanced adsorption ability of TiC@TiN@Pt 1ML is attributed by the robustness of the d electrons. Conversely, the d band model is unfeasible for the multilayer Pt shell. In the bottom panel of Figure 2b, the d orbital of TiC@Pt 2ML and TiC@Pt 3ML are obviously moved toward the Fermi energy with respect to TiC@Pt 1ML . As the Pt thickness increases from 1ML to 2ML and 3ML, the corresponding d band centers are changed from −2.85 to −1.83 and −2.05 eV, respectively. That is, the d band center follows the order of TiC@Pt 2ML > TiC@Pt 3ML > TiC@Pt 1ML , being contrary against the E ads (OH) tendency. Herein, the higher d band centers is accompanied by the weaker E ads (OH), being deviated from the d band model [43]. In order to explain the abnormal phenomenon, the Mulliken charge is analyzed. As shown in Table 2, the charge transferred from the TiC core to Pt shell is reduced as the thickness increases, indicating Pt shell trends to be electronic neutrality. It implies that the electrostatic repulsion between the multilayer Pt shell and the OH would be lessened. Herein, the charge transformation is unaccountable for the E ads (OH) variation. As previous revealed, the adsorption energy is divided into the interaction energy and the deformation energy where the endothermic latter leads to the energetically loss of the adsorption energy [44]. Therefore, the geometrical factors are considered where the average bond length of the Pt-Pt bonds before and after OH adsorption (D bef and D aft ) are given in Table 2. As shown, no significant change occurs during the OH attachment. However, the Pt-Pt bond underlying the adsorbed OH are elongated with the values of 3.32 and 3.28 Å for TiC@Pt 2ML and TiC@Pt 3ML , compared with the shortened 2.80 Å for the TiC@Pt 1ML , respectively. Plausibly, it is inferred that the deviation from the d band model is attributed from the catalysts deformation [44]. to explain the abnormal phenomenon, the Mulliken charge is analyzed. As shown in Table 2, the charge transferred from the TiC core to Pt shell is reduced as the thickness increases, indicating Pt shell trends to be electronic neutrality. It implies that the electrostatic repulsion between the multilayer Pt shell and the OH would be lessened. Herein, the charge transformation is unaccountable for the Eads(OH) variation. As previous revealed, the adsorption energy is divided into the interaction energy and the deformation energy where the endothermic latter leads to the energetically loss of the adsorption energy [44]. Therefore, the geometrical factors are considered where the average bond length of the Pt-Pt bonds before and after OH adsorption (Dbef and Daft) are given in Table 2. As shown, no significant change occurs during the OH attachment. However, the Pt-Pt bond underlying the adsorbed OH are elongated with the values of 3.32 and 3.28 Å for TiC@Pt2ML and TiC@Pt3ML, compared with the shortened 2.80 Å for the TiC@Pt1ML, respectively. Plausibly, it is inferred that the deviation from the d band model is attributed from the catalysts deformation [44]. Due to the scaling relationship, the optimization prerequisite of the electrocatalysts is located at the trade-off adsorption ability since too strong leads to the poisoning and too weak implies the insufficient capture [45,46]. To evaluate the activity, the simple O2 dissociation are taken into consideration with the elemental steps listing in the following according to the previous report [38]. Due to the small kinetic barrier of proton transfer, which could be ignored at the high potential [47,48], our attentions are focused on the free energies G based on the computational hydrogen model [5]: HO * + (H + + e − ) → H2O + * (8) Figure 3 describes the reaction process at the potential U of 0 V and 1.23 V, respectively. The corresponding free energies change △G are summarized in Table 3 where the positive (negative) △G means the endothermic (exothermic) reaction. For TiC@Pt1ML at the potential of 0 V, the O2 dissociation and the OH formation are exothermic processes with the △G values of −1.23 and −1.23 eV, respectively. Meanwhile, the H2O formation from OH protonation is energetically balanced with the △G of 0 eV. Due to the potential-dependence, at U = 1.23 V, the △G of the OH formation and H2O formation are increased to 0 and 1.23 eV, respectively. Thus, the rate-determining step (RDS) of TiC@Pt1ML is located at the H2O formation. Based on the data in Table 3, the similar situation is found for the sandwich structures. The RDS are reserved at the final step of H2O formation with the △G of 0.01, 0.05 and 0.37 eV at U = 0 V or 1.24, 1.28, and 1.60 eV at U = 1.23 V for inserting ZrC, HfC, and TiN interlayer, respectively. Herein, no activity improvement is achieved in comparison with Due to the scaling relationship, the optimization prerequisite of the electrocatalysts is located at the trade-off adsorption ability since too strong leads to the poisoning and too weak implies the insufficient capture [45,46]. To eValuate the activity, the simple O 2 dissociation are taken into consideration with the elemental steps listing in the following according to the previous report [38]. Due to the small kinetic barrier of proton transfer, which could be ignored at the high potential [47,48], our attentions are focused on the free energies G based on the computational hydrogen model [5]: Figure 3 describes the reaction process at the potential U of 0 V and 1.23 V, respectively. The corresponding free energies change ∆G are summarized in Table 3 where the positive (negative) ∆G means the endothermic (exothermic) reaction. For TiC@Pt 1ML at the potential of 0 V, the O 2 dissociation and the OH formation are exothermic processes with the ∆G values of −1.23 and −1.23 eV, respectively. Meanwhile, the H 2 O formation from OH protonation is energetically balanced with the ∆G of 0 eV. Due to the potential-dependence, at U = 1.23 V, the ∆G of the OH formation and H 2 O formation are increased to 0 and 1.23 eV, respectively. Thus, the rate-determining step (RDS) of TiC@Pt 1ML is located at the H 2 O formation. Based on the data in Table 3, the similar situation is found for the sandwich structures. The RDS are reserved at the final step of H 2 O formation with the ∆G of 0.01, 0.05 and 0.37 eV at U = 0 V or 1.24, 1.28, and 1.60 eV at U = 1.23 V for inserting ZrC, HfC, and TiN interlayer, respectively. Herein, no activity improvement is achieved in comparison with TiC@Pt 1ML . On the other hand, being different from TiC@Pt 1ML , the elemental steps of the TiC@Pt 2ML at U = 0 V are energetically downward with the ∆G of −1.18, −0.98, and −0.30 eV for the O 2 dissociation, the OH formation and the H 2 O formation, respectively. At U = 1.23 V, the OH formation and H 2 O formation are changed to be endothermic and the corresponding ∆G are 0.25 and 0.93 eV, indicating the RDS is reserved as the H 2 O formation. Analogously, for 3ML Pt shell, the RDS of H 2 O formation with the ∆G of −0.14 eV at U = 0 V and 1.09 eV at U = 1.23 V are observed. Herein, from the thermodynamic aspect, the promotion effect on the ORR activity is observed for the multilayer Pt shell.  TiC@Pt1ML. On the other hand, being different from TiC@Pt1ML, the elemental steps of the TiC@Pt2ML at U = 0 V are energetically downward with the △G of −1.18, −0.98, and −0.30 eV for the O2 dissociation, the OH formation and the H2O formation, respectively. At U = 1.23 V, the OH formation and H2O formation are changed to be endothermic and the corresponding △G are 0.25 and 0.93 eV, indicating the RDS is reserved as the H2O formation. Analogously, for 3ML Pt shell, the RDS of H2O formation with the △G of −0.14 eV at U = 0 V and 1.09 eV at U = 1.23 V are observed. Herein, from the thermodynamic aspect, the promotion effect on the ORR activity is observed for the multilayer Pt shell.    In addition to the thermodynamic analysis, the kinetic barriers of the O 2 dissociation mechanism on TiC@Pt 2ML are further considered. The reaction pathway of TiC@Pt 2ML is plotted in Figure 4. The corresponding reaction barriers E a and reaction energy E r are tabulated in Table 4. Herein, TiC@Pt 1ML and TiC@TiN@Pt 1ML are selected as references. For O 2 splitting into the O atoms, the E a of TiC@Pt 2ML is 1.28 eV, being slightly higher than 1.05 eV of TiC@Pt 1ML and 0.76 eV of TiC@TiN@Pt 1ML . The weaker E ads (O 2 ) correlates to the higher E a of O 2 dissociation, indicating the degradation of O 2 activation, in consistence with the previous reports [39,49]. Noting that the O 2 dissociation on TiC@Pt 1ML would be significantly boosted by lowing O 2 coverage where the E a reduces from 0.86 eV to 0.36 eV as the O 2 coverage changes from 1/4 ML to 1/9 ML [39]. Therefore, it is reasonably believed that the mentioned phenomenon is occurred on TiC@Pt 2ML , implying that the barrier of O 2 splitting would be overcome at the room temperature [50]. Furthermore, the similar situation is observed for the OH formation where the unfeasibility of TiC@Pt 2ML is identified compared with TiC@Pt 1ML and TiC@TiN@Pt 1ML . However, the E a of the H 2 O formation is 0.91, 1.04, and 1.41 eV for TiC@Pt 2ML , TiC@Pt 1ML , and TiC@TiN@Pt 1ML , respectively. The lower value implies that the OH hydrogenation is speeded by the presence of TiC@Pt 2ML . Herein, the kinetic benefit of TiC@Pt 2ML is confirmed that the low oxyphilic character avails the OH poisoning, in line with the thermodynamic data [51]. In addition to the thermodynamic analysis, the kinetic barriers of the O2 dissociation mechanism on TiC@Pt2ML are further considered. The reaction pathway of TiC@Pt2ML is plotted in Figure 4. The corresponding reaction barriers Ea and reaction energy Er are tabulated in Table 4. Herein, TiC@Pt1ML and TiC@TiN@Pt1ML are selected as references. For O2 splitting into the O atoms, the Ea of TiC@Pt2ML is 1.28 eV, being slightly higher than 1.05 eV of TiC@Pt1ML and 0.76 eV of TiC@TiN@Pt1ML. The weaker Eads(O2) correlates to the higher Ea of O2 dissociation, indicating the degradation of O2 activation, in consistence with the previous reports [39,49]. Noting that the O2 dissociation on TiC@Pt1ML would be significantly boosted by lowing O2 coverage where the Ea reduces from 0.86 eV to 0.36 eV as the O2 coverage changes from 1/4 ML to 1/9 ML [39]. Therefore, it is reasonably believed that the mentioned phenomenon is occurred on TiC@Pt2ML, implying that the barrier of O2 splitting would be overcome at the room temperature [50]. Furthermore, the similar situation is observed for the OH formation where the unfeasibility of TiC@Pt2ML is identified compared with TiC@Pt1ML and TiC@TiN@Pt1ML. However, the Ea of the H2O formation is 0.91, 1.04, and 1.41 eV for TiC@Pt2ML, TiC@Pt1ML, and TiC@TiN@Pt1ML, respectively. The lower value implies that the OH hydrogenation is speeded by the presence of TiC@Pt2ML. Herein, the kinetic benefit of TiC@Pt2ML is confirmed that the low oxyphilic character avails the OH poisoning, in line with the thermodynamic data [51].   Noting that he experimental verification should be urgently needed to confirm the DFT prediction. Herein, we believe our results realizable due to the following reasons: firstly, CHE model has been successfully applied to interpret the experimental data and design the novel electrocatalysts for metal, oxides as well as carbon-based materials [41,[52][53][54][55][56][57][58]; secondly, the development of the synthesis technology leads to the feasibility of the TiC@Pt materials with different core composition as well as shell thickness [17,[59][60][61]. Therefore, it is reasonably believed that the DFT candidate of TiC@Pt 2ML materials could be experimentally achieved.

Conclusions
In this study, DFT calculation is used to investigate the effect of the interlayer and shell thickness on ORR activity. Compared with the TiC@Pt 1ML , the comparable adsorption ability was found for TiC@ZrC@Pt 1ML and TiC@HfC@Pt 1ML whilst the presence of TiN causes a sharp enhancement of the adsorption energy. From the PDOS analysis, the upshifted d band of TiC@TiN@Pt 1ML supports the variation of the adsorption behavior. On the other hand, the multiplayer Pt shell generally weakens the oxyphilic affinity with the order of TiC@Pt 1ML > TiC@Pt 3ML > TiC@Pt 2ML . The deviation from the famous d band model is plausibly attributed from the structural deformation. Furthermore, the RDS of the considered systems are identified as the H 2 O formation. The decrease of the adsorption capacity alleviates the OH poisoning and boosts the ORR activity. Herein, the enhanced activity of the TiC@Pt 2ML is confirmed compared with TiC@Pt 1ML . The promising ORR performance of the multilayer Pt supported on TiC supplies the theoretical guide for the synthesis.