Direct atomic-level insight into oxygen reduction reaction on size-dependent Pt-based electrocatalysts from density functional theory calculations

Fangren Qian,Lishan Peng,Yujuan Zhuang,Lei Liu,Qingjun Chen,4,5,

1 School of Rare Earths, University of Science and Technology of China, Hefei 230026, China

2 Key Laboratory of Rare Earths, Ganjiang Innovation Academy, Chinese Academy of Sciences, Ganzhou 341000, China

3 Center for Computational Chemistry, College of Chemistry and Chemical Engineering, Wuhan Textile University, Wuhan 430200, China

4 Institute of Process Engineering, Chinese Academy of Sciences, Beijing 100190, China

5 Langfang Technological Service Centre of Green Industry, Langfang 065001, China

Keywords:Density functional theory(DFT)calculations Pt-based electrocatalysts Oxygen reduction reaction

ABSTRACT Developing novel oxygen reduction reaction (ORR) catalysts with high activity is urgent for proton exchange membrane fuel cells.Herein,we investigated a group of size-dependent Pt-based catalysts as promising ORR catalysts by density functional theory calculations,ranging from single-atom,nanocluster to bulk Pt catalysts.The results showed that the ORR overpotential of these Pt-based catalysts increased when its size enlarged to the nanoparticle scale or reduced to the single-atom scale,and the Pt38 cluster had the lowest ORR overpotential(0.46 V)compared with that of Pt111(0.57 V)and single atom Pt(0.7 V).Moreover,we established a volcano curve relationship between the ORR overpotential and binding energy of O* (ΔEO*),confirming the intermediate species anchored on Pt38 cluster with suitable binding energy located at top of volcano curve.The interaction between intermediate species and Pt-based catalysts were also investigated by the charge distribution and projected density of state and which further confirmed the results of volcano curve.

1.Introduction

Proton exchange membrane fuel cells (PEMFCs) are taken for prospective energy conversion technologies due to their great potential for high efficiency and zero emissions [1,2].However,the further commercialization of a PEMFC is strictly hindered by the intrinsically sluggish kinetics of oxygen reduction reaction(ORR) at the cathode,which requires a relatively high loading of Pt electrocatalyst to boost the reaction [3].Indeed,it is desirable to design effective ORR electrocatalysts with lower Pt loading and comparable activity.Pt alloying with transition metals (Fe,Co,etc.)have attracted considerable interest owing to its excellent activity and stability.Nevertheless,the transition metal is easily oxidized and dissolved in the work environment of PEMFCs,which may further cause the structural collapse of catalysts and degrade the proton exchange membrane [3].

Another important strategy is to establish more active sites on electrocatalysts with lower precious metal content,i.e.,decreasing the size of electrocatalysts [4].The active sites of catalysts will be increased by downsizing the size of catalysts to the nanoscale,which is crucial for enhancing the utilization of catalysts and consequently improving the catalytic activity [5-8].The maximum utilization of metal atoms can be obtained when the particle size decreases to about 1 nm(namely,the nanoclusters containing dozens of atoms),by which nearly all catalysts atoms are available for catalysis [9].

Moreover,the electronic structure of catalysts is also dependent on their size especially when the size is below 2 nm.This sizeinduced modulation in electronic structure is resulted from the quantum size effect[10],which has been demonstrated to be vital for endowing the catalysts with superior intrinsic activity[11-14].In the past,many studies have shown that Pt clusters with subnano size possess superior catalytic activity than bulk Pt particles.A significant investigation by Arenzet al.[15] on commercial Pt/C catalysts with an improved experimental scheme demonstrated enhanced catalytic activity when the size of catalysts decreased from 5 nm to 1 nm [4].Also,a tremendous improvement in catalytic activity for ORR was established by Pt60,Pt28,and Pt12obtained by a dendrimer template.As pointed out by Lianget al.[9],metal nanocluster in 1 nm size stabilized on carbon support achieved superior catalytic performance for heterogeneous catalysis.As mentioned above,noble metal-based clusters with sub nano size are considered as potential alternative noble metal catalysts,which has been proved by a considerable amount of experimental works [16].However,the catalytic mechanism of size-dependent Pt-based catalysts for ORR remains controversial,which somehow hinders the large-scale applications of metal nanoclusters in PEMFCs [17].

In the present work,we gained direct atomic-level insight into oxygen reduction reaction on size-dependent Pt-based electrocatalysts via density functional theory calculation (DFT).A series of models of Pt nanoclusters were established and used to understand the relationship between electronic structures and catalytic activity of size-dependent Pt-based catalysts.As a result of this work,the Pt38cluster with a size of about 1 nm possessed the lowest energy barrier,resulting in an optimal ORR process among all studied clusters [18].Further,we expanded the range of size to bulk and single-atom Pt catalysts as depicted in Fig.1 and the results showed that the Pt38cluster reflected a superior ORR activity compared to Pt bulk particles and single-atom Pt.Subsequently,we established a volcano curve relationship between the ORR overpotential and ΔEO*.

Fig.1.The Pt-based catalysts from single-atom Pt to Nanoparticles.

2.Methods

All DFT calculations[19]were carried out on the Viennaab initiosimulation package(VASP)with spin-polarization.The interactions between ions and electrons were demonstrated by the projector augmented wave potentials[20].The generalized gradient approximation [21,22] with Perdew Burke Ernzerhof functional [23] was used to describe the exchange and correlation effects.In view of the potential van der Waals interactions between ORR intermediates and Pt-based materials,the empirical correction in Grimme’s scheme (i.e.,DFT+D3) was adopted [24].The convergence criteria were set to be 10-5eV in energy and 0.2 eV·nm-1in atomic relaxation for all calculations.An energy cutoff of 400 eV was used for the plane-wave basis set.The Brillouin zone was sampled on the basis of the Monkhorst-Pack scheme with thek-points being 2×2×1.A four layers slab was used to simulate the(1 1 1) surface of Pt in which the bottom two layers were fixed while the top two layers were relaxed.The vacuum layer was set to 1.3 nm to guarantee negligible interactions between the periodically repeated slabs.The differential charge density and density of state (DOS) were used to gain charge transfer and orbital hybridization,respectively.The surface energyEsurf(Pt),cohesive energy(CE(Ptn)),and binding energy(Eb)were carried out to obtain the stability of various surfaces of Pt,Pt clusters,and single-atom Pt according to the following Eqs.(1),(2) and (3) respectively[25-27]:

whereEsurf(Pt)signifies the total energy of the slab;EPtdenote the atomic energy of Pt in their bulk phases andNPtis the number of Pt atoms in the slab,respectively.CE(Ptn)represents the cohesive energy of the Ptncluster andE(Pt)ndenotes the total energy of Ptn cluster,respectively.TheEbsignifies binding energy,Etotalis the total energy of Pt atom adsorb on the graphene andEgraphenerepresents the energy of graphene.

To describe the liquid environment during ORR process,the solvation effect is evaluated using an implicit solvation model implemented in the VASPsol code [28,29].The Gibbs free energy of the adsorbates is calculated by The Gibbs free energy of the adsorbates for each step in the ORR was gained based on the computational hydrogen electrode (CHE) model raised by N?rskovet al[30].as following Eq.(4).where theEDFTis the calculated energy.EZPEand TS respectively represent the zero-point energy correction calculated from vibrational frequencies and entropy contribution at 298.15 K and 0.1 MPa [30].In order to minimize the variations of the absolute Gibbs free energy for adsorbates on Pt-based catalysts,theEZPEis assumed not to change with the surface composition as described in a previous publication,so the catalysts layers were fixed while the intermediate species were relaxed [31].

3.Results and Discussion

3.1.Structures and stability

Firstly,we investigated the geometrical structures of the singleatom Pt supported on perfect or defective grapheme [9,10],isolated Ptnclusters,and bulk Pt nanoparticles.In order to improve the stability of single-atom on substrates,we designed seven different initial structures (Fig.2(a)).After DFT optimizations,it is found that the initial top (1) and hollow site (3) evolved into the bridge site ultimately.Hence,five various single atom coordinated structures were obtained,which are demonstrated in Fig.S1(Supplementary Material).The binding energy of all single-atom Pt structures are depicted in Fig.2(b) and these results indicated the single-atom Pt with the N4Pt coordination is most stable due to the strong electronegativity of four-nitrogen can increase the interaction between Pt atom and coordinate anion,which can modulate the adsorption of intermediates such as O*,OH*,OOH*and thereby affects its catalytic properties.Given Pt clusters with nano-scale exhibit highly effective metal atom utilization in the electrocatalytic process,the cluster models with various sizes were established as displayed in Fig.2(d).All the cluster structures were optimized by DFT and the results showed that all the Ptnclusters were in 3D geometries forn＞4,and their corresponding structural parameters were well consistent with previously reported works[31-35].To obtain the stability of various clusters,we also calculated the cohesive energy,which represents the interaction between all Pt atoms in a cluster,as depicted in computation methods and models.The larger the value ofCE,the more stable the corresponding structure [36].According to Fig.2(e),the Pt38cluster with a size of about 1 nm is the most stable among all the studied cluster structures and may be easier to be obtained in experiments.To gain deeper insight into the oxygen reduction reaction on size-dependent Pt-based electrocatalysts,we also extended the size to bulk and investigated the stability of pure Pt nanoparticles with various surfaces,the optimized surface of Pt,and corresponding surface energy as demonstrated in Fig.S2 and Fig.2(c),respectively.According to Fig.2(c),the(1 1 1)surface possessed the lowest surface energy,indicating the (1 1 1) lattice plane was most stable and will dominant exposure in the synthesis process.Thereby we will focus on (1 1 1) surface in the following works [25].

Fig.2.(a) The various coordinated structures for single-atom Pt and (b) the Eb,(c) Esurf, and (e) CE were calculated by DFT,(d) The optimized structure of Ptn cluster with various sizes.Color legend: C brown,Pt lilac,N modena.

3.2.ORR mechanism

Subsequently,we revealed the ORR activity on various sizedependent Pt-based materials by calculating the Gibbs free energy of each step in the ORR process.Based on previous works[37-39],the ORR process was supposed to occur on designed materials according to four processes as illustrated in Fig.3(a): (1)*+O2(g)+H++e-→OOH*,(2) OOH*+H++e-→O*+H2O,(3) O*+H++e-→OH*,and (4) OH*+H++e-→H2O.The optimized adsorption structures containing OOH*,O*,and OH*intermediates on various size-dependent Pt-based catalysts as illustrated in Fig.S3-S5.As described above,the ORR process begins with the hydrogenation of O2transfer to*OOH anchored on the Pt sites(Fig.S6).The Gibbs free energy diagram for ORR on sizedependent Pt-based materials systems were summarized in Fig.S7,indicating the single atom PtC4and Pt38cluster possessed low energy barrier in these studied single atom and clusters,respectively.In the following,we focus on investigating the ORR process among Pt111,Pt38,and single atom PtC4and insight into the ORR process on size-dependent Pt-based electrocatalysts.First,we calculated the Gibbs free energy diagram for ORR of Pt111,Pt38,and single atom PtC4at 0 V.It is indicating all the reaction stages in the entire ORR were exothermic at 0 V as illustrated in Fig.S8.In addition,some processes of ORR on single atom PtC4,Pt38cluster,and Pt111catalysts changed to endothermic process at the equilibrium potential of 1.23 V (Fig.3(b)) in which the situation of Pt111and single-atom Pt were similar.For single atom PtC4,the formation of the*OOH species was uplifted by 0.70 eV.Whereafter,the OOH*is broken and yield (O*+H2O).Interestingly,we discovered the formation of(O*+H2O)from the OOH*species was exothermic with the free energy diagram by-0.69 eV.Once the first H2O molecule was obtained,the remaining O*intermediate was anchored on a single atom PtC4,which will be integrated with H to generate OH*species.Obviously,the stage of O*→OH*is a non-spontaneous process because of the positive ΔG(0.10 eV) on the single-atom PtC4.Finally,the OH*intermediate was incorporated with H to generate another H2O and this final step was still exothermic by-0.11 eV on a single atom PtC4.Therefore,the rate determined step(RDS)for single-atom PtC4was the formation process of*OOH due to the energy barrier (0.70 eV) being the highest among all the steps.In the case of Pt111,the change of Gibbs free energy for the formation of*OOH was positive by 0.57 eV,which was lower than single-atom PtC4.Subsequently,the hydrogenation of*OOH and remaining O*were anchored on Pt111,of which the ΔGis negative(-0.49 eV).Then the formation process of*OH on Pt111was still spontaneous and the value of ΔGwas-0.13 eV.Finally,the hydrogenation of*OH and the second H2O were obtained.This process was non-spontaneous and the corresponding ΔGwas 0.06 eV.The results indicated the RDS for Pt111in the ORR process was also the first step (0.57 eV),and the energy barrier is lower than single-atom Pt.On the other hand,the ORR process of Pt38was different from single atom PtC4and Pt111,and the formation process of*OOH was spontaneous due to the ΔGbeing negative (-0.07 eV).The next process for hydrogenation of*OOH was also spontaneous and the ΔGwas -0.77 eV.Then the ΔGfor formation of*OH on Pt38is 0.24 eV.Finally,the hydrogenation process of*OH was non-spontaneous and the ΔGwas 0.46 eV.This result showed the RDS of the Pt38cluster was the final step,i.e.,the hydrogenation process of*OH and the energy barrier for the Pt38cluster (0.46 eV) was lower than both single-atom PtC4(0.70 eV) and Pt111(0.57 eV).This result further confirmed that the Pt38cluster with the size of about 1 nm possessed the highest ORR activity among all the studied size-dependent Pt-based catalysts.

Fig.3.The mechanism of ORR for size-dependent Pt-based catalytic materials.(a)the proposed electrocatalytic mechanism for ORR,(b)Gibbs free energy diagram for ORR on single-atom Pt (PtC4),Pt38 cluster,and Pt111 at 1.23 V,(c) The liner relationship between ΔG*O,ΔG*OH,and ΔG*OOH,(d) The volcano curves between ΔEO* and overpotential.

To deeper understand the origin of the excellent ORR activity of the size-dependent Pt-based catalysts,we examined the relationship between ΔG*O,ΔG*OH,and ΔG*OOH.As illustrated in Fig.3(c),ΔG*OHand ΔG*Oindicated a linear scaling relation (ΔG*O=1.3-ΔG*OH-0.009 eV)with anR2of 0.91 for all the size-dependent catalytic materials.The relation between ΔG*OOHand ΔG*OHwas similar to ΔG*OHand ΔG*O,and we obtained ΔG*OOH=0.9ΔG*OH-0.49 eV with anR2of 0.97.Therefore,it is suitable to describe the ORR activity only by O*(or*OH,*OOH).As shown in Fig.3(d),a volcano curve relationship between the ORR overpotential and ΔEO*was found for size-dependent Pt-based catalytic materials.For the ORR,the optimal peak position was about at ΔEO*=-4.2 eV,where ΔEO*on the Pt38cluster was closest to the top.If ΔEO*on the Pt38cluster increases or decreases,the energy barrier will be enhanced and resulted in the ORR activity decay.These results showed the Pt38cluster possessed optimal ORR activity among all the studied Pt-based catalysts,resulting from the Pt38cluster can adsorb the O*intermediate with desirable binding energy.This volcano-shaped curve was similar to the Sabatier principle in catalysis,that is,the reasonable catalysts should adsorb the intermediates neither too strong nor too weak,which benefits the desorption of adsorbates and conducting to the next reaction.

3.3.Electronic properties

In order to further explain the variation tendency of Gibbs free energy in the ORR process on size-dependent Pt-based catalysts and reveal the mechanism of the ORR process,we calculated the difference in charge density and binding energy between intermediate species and catalysts as depicted in Fig.4.The electronic structure at the atomic level will regulate the binding energy and further influence Gibbs free energy,resulting in an energy barrier increase or decrease which is related directly to ORR activity.As described above,the RDS of single-atom Pt and Pt111was the first step for the formation of*OOH while the RDS of the Pt38cluster was the final step.

Fig.4.The difference charge density and corresponding binding energy for each step of ORR:(a)Pt111,(b)Pt38 cluster,and(c)single atom PtC4.Color legend:C brown,Pt lilac.(1 ?=0.1 nm).

In the beginning,the*OOH species was anchored on the top site for the Pt111(Fig.4(a1)) and single atom PtC4(Fig.4(c1)) while on the bridge site for the Pt38cluster.Besides,the charge transfer between*OOH and catalyst was enhanced by the Pt38cluster as depicted in Fig.4(b1),leading to the easier formation of*OOH species on Pt38compared to Pt111and single atom PtC4due to the binding energy being stronger than the two others.Subsequently,the hydrogenation of*OOH to H2O and remaining O*(Fig.4(a2)-(c2)) was anchored on catalysts which further determined the formation of*OH.Finally,the hydrogenation of*OH accompanied by the second H2O molecule was obtained,and the*OH was anchored tightly on the Pt38cluster ascribing to obvious charge transfer(Fig.4(b3)),resulting in the step for hydrogenation of*OH turned into the RDS of Pt38cluster for ORR process.While the RDS of Pt111and single atom PtC4was the first step in the ORR process due to the difficulty of the formation of*OOH.In addition,the binding energy between O*and Pt38cluster (Fig.4(b2)) is more suitable compared to Pt111(Fig.4(a2)) and single atom PtC4(Fig.4(c2)),which benefited accelerating the ORR process and reducing the energy barrier (Fig.3(d)).To gain a deep insight into the electronic structure and charge transfer in size-dependent Ptbased materials,we calculated the projected density of state(PDOS) and charge density contour plot of Pt-based catalysts(Fig.S9).The orbital hybridization between Pt and C was observed from the partial density of states(PDOS)(Fig.S9(a))which also can be verified by the difference charge density (Fig.S9(d)),indicating the obvious electronic metal-support interaction and benefiting the electron transfer from Pt to graphene supporting [40].At the Fermi level (EF),d orbitals contributed to the PDOS and finally exhibited the metallicity of Pt111(Fig.S9(c)) and Pt38cluster(Fig.S9(b)),which was beneficial to improving electrocatalysis[41].

To unravel the origin of the ORR in size-dependent Pt-based electrocatalysts,we further investigated the PDOS between the catalysts and intermediate species (*OOH,O*,and*OH) due to the ORR activity depends heavily on the electronic structure(Fig.5) [42].First,the*OOH species was anchored on the Pt111(Fig.5(a1)) and single-atom PtC4(Fig.5(c1)),and the orbital hybridization between*OOH and Pt111(or single-atom PtC4) was relatively weaker compared to Pt38(Fig.5(b1)),indicating the*OOH intermediate may be easier adsorb on Pt38cluster which will accelerate the formation of*OOH on Pt38,further confirming the ΔG1of Pt111and single atom PtC4is larger than Pt38cluster.As illustrated in Fig.5(b2),the orbital hybridization between*O and Pt38cluster was stronger than single atom PtC4(Fig.5(c2)),resulting in the hydrogenation of O*to*OH for Pt38being more difficult than another.Finally,the orbital hybridization between*OH and Pt38cluster (Fig.5(b3)) was more obvious than Pt111(Fig.5(a3))and single atom PtC4(Fig.5(c3)),resulting in the further hydrogenation of*OH to H2O from Pt38being more difficult than the others.All the above results in the RDS of Pt111and single atom PtC4located at the formation of*OOH while the Pt38cluster change to the hydrogenation of*OH to H2O.

Fig.5.The PDOS of Pt-based catalysts for each step of the ORR process: (a) Pt111,(b) Pt38 cluster,(c) single atom PtC4.

4.Conclusions

In summary,we self-consistently investigated the activity origin of a group of size-dependent Pt-based catalysts materials using DFT calculations.After optimizing all the structures and screening more stable single-atom coordination,cluster configuration,and crystal plane,we further revealed its in-depth origin of induced activity at the atomic level.According to the electronic structure,we found a parameter,namely ΔEO*,as a descriptor for ORR activity due to the linear relationship between O*with*OH and*OOH,and a volcano curve between the ORR overpotential and ΔEO*was obtained.Finally,the results indicated Pt38cluster with a size of about 1 nm possessed the highest ORR activity among all the size-based catalysts materials according to Gibbs free energy by DFT calculation,and we described the high ORR activity of the Pt38cluster by the suitable ΔEO*,benefiting the adsorption and desorption for ORR intermediates and endow low energy barrier to trigger ORR process.This work lights up a novel method for studying the origin of the ORR mechanism of size-dependent Pt-based catalysts and their activity in other catalytic reactions and further induces the development of novel catalysts with high atom utilization.

Data Availability

Data will be made available on request.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

The work was supported by the National Natural Science Foundation of China (92061125,21978294),Beijing Natural Science Foundation (Z200012),Jiangxi Natural Science Foundation(20212ACB213009),DNL Cooperation Fund,CAS (DNL201921),Self-deployed Projects of Ganjiang Innovation Academy,Chinese Academy of Sciences(E055B003),and Hebei Natural Science Foundation (B2020103043).

Supplementary Material

Supplementary data to this article can be found online at https://doi.org/10.1016/j.cjche.2023.02.019.