Towards quaternary alloy Au–Pd catalysts for direct synthesis of hydrogen peroxide

https://doi.org/10.1016/j.mtener.2020.100399Get rights and content

Highlights

  • Extensive DFT Study of Active & Selective Catalyst Surface Site Structure for H2O2 Synthesis alloy-based Au–Pd materials.

  • Effect of quaternary doping via oxophilic and oxophobic transition metal pairs.

  • Candidate catalyst materials identified which improve activity, selectivity, and cost.

  • Recommendations for extension of results to guide further ongoing work in catalyst screening.

Abstract

The direct synthesis of hydrogen peroxide (H2O2) in situ to replace legacy large-scale commercial anthraquinone process is a critical industrial technology required to advance applications in sustainable green chemistry and reduce energy consumption associated with transporting reagents and oxidants. Current state-of-the-art Au–Pd transition metal alloy catalysts show promise to selectively synthesize H2O2 however activity is not optimal and material costs and sustainability concerns hinder widespread use. In this manuscript, using target values from previously derived Oxygen Reduction Reaction (ORR) Volcano Plots, we analyze and filter potential (Au44Pd44)M6N6 {M = metal 1, N = metal 2} quaternary alloys by their associated descriptor values, the adsorption energy of mono-atomic oxygen and hydrogen. We report possible surface structures and compositions which have adsorption sites that simultaneously optimize the adsorption energy of both descriptors and explain possibilities for using these results to leverage in future and ongoing work for truly optimal catalyst design for transition-metal alloys for direct synthesis of hydrogen peroxide. These results and recommendations should ultimately help guide developments to increase the performance (activity and selectivity) of direct synthesis catalysts for hydrogen peroxide synthesis while simultaneously lowering the costs of materials in these catalysts and making them more sustainable.

Introduction

Hydrogen peroxide (H2O2) is one of the most common household chemicals and it also has many uses in industrial processes. Such uses include as a pharmaceutical disinfectant, an effective oxidant and a growing use in the pulp and paper industry, and can be an {undesired} product of the 2-electron transfer in the Oxygen Reduction Reaction (ORR) relevant to Proton Exchange Membrane Fuel Cells (PEMFC) [[1], [2], [3], [4], [5]]. The main benefit of using H2O2 in such roles is that it can be a “greener” alternative to many of the other oxidants in use today. Accordingly, it is being looked at as a replacement for many of the common household cleaners and ingredients for laundry detergent [3,5]. Hydrogen peroxide is also used in municipal wastewater treatment for the removal of contaminants such as hydrogen sulfide, cyanide, nitrite and many more [5].

Hydrogen peroxide is primarily produced through the anthraquinone auto-oxidation process [6,7]. This process is an indirect synthesis that uses organic solvents to achieve the formation of H2O2 which avoids the use of H2/O2 mixtures. The avoidance of using such a volatile mixture is one of the primary advantages that the anthraquinone process provides. However, due to high capital and operating costs, this process is only economical when performed on a large scale which prevents many companies that need the chemical for their processes from producing it on a small scale in situ. This scenario also creates a need for H2O2 to be transported which introduces even more cost and safety concerns since H2O2 can decompose explosively [5,6]. Beyond the cost and safety concerns previously mentioned, degradation of the ethylanthraquinone, one of the reactants along the pathway to form hydrogen peroxide, is a major concern for the efficiency of this process. Ethylanthraquinone degradation is also correlated with the deactivation of the palladium catalysts used in the process which also decreases the efficiency and increases the cost of the process [8,9].

One of the methods currently being investigated to minimize the issues that are associated with the indirect synthesis of hydrogen peroxide is using direct synthesis instead. If H2 and O2 can be safely extracted from common sources such as air and water, then given the right catalyst(s), H2O2 can be produced on a smaller scale in situ which will also help eliminate transportation concerns, further contributing to its value as a “greener” oxidant. Therefore, in hopes of reducing the cost and eliminating some of the safety concerns of hydrogen peroxide production, researchers are looking at determining a catalyst for the direct synthesis of hydrogen peroxide (DHPSP). Currently, AuPd catalysts have been shown to be the current state-of-the-art catalysts in regards to activity and selectivity for this reaction. In this context, the use of selectivity to mean the ratio of desired to undesired products and, correspondingly, activity meaning the multiplicative factor increase in the rate of the reaction as compared to the non-catalyzed reaction. However, because Au and Pd are both rare metals, the primary concerns are the cost and sustainability of such catalysts. Another concern is that while Au–Pd based alloy catalysts currently show state-of-the-art performance for the DHPSP, the selectivity and activity can be improved according to the predictive Volcano Plot(s) previously generated for the ORR on single crystal transition metal catalysts [1].

Xu et al. reported that Pt as an alloying metal to AuPd catalysts provides better activity while retaining great selectivity [10]. They also predicted that W, Pb, Mo, and Ru would also be good candidates to form ternary catalysts with Au and Pd. Mahata and Pathak looked at binary and ternary catalyst materials composed of Co, Au, and Pt [11]. They concluded based on energy barriers for the desired and undesired reactions of the direct H2O2 synthesis pathway (DHPSP) that Au10Co9Pt60 was highly selective towards H2O2 production and therefore the best of the alloys tested. Nugraha et al. studied Pd–Au and Pd–Hg catalysts in comparison to pure Pd [12]. Their results concluded that Pd–Hg and Pd–Au showed similar properties and are both better than pure Pd. Other works have examined the structure of the Au–Pd catalyst(s) and determined possible methods of improving the selectivity and activity of the catalyst. These works studied varying lattice strain, changing the arrangement of the atoms, and different layering in the sub-surface structure [[13], [14], [15]].

The new results in this manuscript are based on density functional theory (DFT) calculations. The focus was on two different aspects of interest to help understand how to improve the Au–Pd catalyst system. First, eight different AuPd surfaces with different arrangements of the Pd and Au atoms were studied. This was done with the purpose of determining whether the arrangement of the atoms significantly changes the O and H binding energies which are descriptors for the activity and selectivity of AuPd catalyst [1]. (Descriptors are key variables or parameters upon which all other properties of the system can be correlated to and predicted from.) In addition to these tests, the AuPd surfaces were then alloyed with minority loadings of other transition metals (Pt, Rh, Ir, Ag, Cu, and Ni) to determine if these alloying metals could simultaneously increase both the activity and selectivity of the catalysts and further decrease the rare metal content of the catalysts. More details on how this alloying was done are provided in the Methods section of this manuscript.

Among the 576 different systems (combinations of surface and site structure, and composition) reported in this manuscript, 80 fell within the desired target’ range’ for the descriptor value(s) for the ideal catalytic activity and selectivity. Over a third of these surface/site structure variations were from pure Au50–Pd50 systems which indicates (and agrees with known experiments) that seemingly random ordered/structured Au-Pd catalyst with varying surface structure and local adsorption site compositions are innately well suited as among the best catalysts for the direct synthesis of H2O2. These Au50–Pd50 catalysts however can still be improved in activity, selectivity, as predicted by their position on ORR Volcano Plots, and potentially, at significantly cost reduction by inclusion of less rare metals. The remaining ~two-thirds of the best identified systems varied throughout the range of combinations studied in this work. However, systems with Pt and Ag and systems with Rh and Ag had the most frequent occurrence indicating these materials might have robust average(d) activity and selectivity performance for the DHPSP. While the introduction of Pt has shown promise in increasing catalytic activity for the DHPSP in other work, it does not solve the cost issue associated with using rare metals [10]. The same argument can be said qualitatively for Rh unfortunately. However, to the contrary, the introduction of Ag would help with cost reduction and future research (including increasing the loading as a replacement atom for Au) may be warrant to further study the effect of Ag as a dopant or alloying agent in these catalysts. The use of Ir, Ni, and Cu also showed promising results, but these were not as numerous{frequent} as the results seen for Pt, Ag, and Rh indicating the exact surface structure and site composition of alloys of AuPd including these elements may be influential on creating a disperse range of non-ideal descriptor value adsorption sites for O∗ and H∗ in the catalytic process of the DHPSP; this may indicate challenges for synthesis of specific optimal catalysts using these dopant elements.

The work reported in this manuscript is not intended to be an end-all investigation complete predictive design of the best Au–Pd based DHPSP catalyst. Instead, this work is designed to pre-filter materials and guide future studies which leverage the results presented in this work to move closer to optimal DHPSP catalyst design by identifying overall trends and similarities in structure and composition of quaternary alloy catalyst sites which express optimal adsorption energies of the previously identified activity and selectivity descriptors for the DHPSP. The results presented in this work prove the first step in helping move such progress forward by proving that at least some of the quaternary (AuPd)88 (MN)12 catalysts have at least a fraction of their surface sites which have descriptor values for activity and selectivity that fall very close to the predicted Volcano Plot peak. The novel results presented in this work add to what is known in the literature by: 1) showing that quaternary Au–Pd rich alloys can have active sites which move closer to the “ideal” values for both the descriptors of adsorption energy of O∗ and H∗, 2) that for some of these quaternary alloys such sites can occur with a variety of surface-site structure and atomic configurations {at a given composition} 3) the effect of alloying at this loading is highly non-linear and can break linear scaling/interpolation rules, 4), that some of the promising quaternary systems can include guest/dopant elements (including Ag) which are lower cost/rarity than Pd or Au, and 5) that accordingly, in whole it should be possible to make more selective and active Au–Pd alloy catalyts at lower cost for direct hydrogen peroxide synthesis than the current state-of- the-art. Further investigation is warranted however, especially for Ni and Cu systems, because similar to Ag they would help decrease the cost of the catalyst even more. Ongoing and future work related to this project is underway and will focus on studying the to-date best identified composition(s), surface structure, and adsorption sites (by way of descriptor values) as a function of varying composition (decrease AuPd loading) and varying catalyst surface order/disorder (random alloy, intermetallic, core-shell, etc); ongoing work will also continue to leverage results from this manuscript to identify AI algorithms to help guide optimal alloy/structure relationships with respect to optimizing descriptor values at minimal precious metal loadings.

Section snippets

DFT calculations

Plane-wave DFT calculations were performed to study the energetics of both the mono-atomic oxygen and mono-atomic hydrogen adsorbate atoms on the catalyst surface(s). These adsorption energies serve as explicit descriptors of activity and selectivity, on the Au–Pd based surfaces described in this work [1]. Plane-wave DFT calculations accurately reproduce both the quantitative and qualitative trends in the activity and selectivity for single crystal surfaces of transition metal catalysts for the

Background

As previously mentioned, Au–Pd catalysts are currently the best for the direct synthesis of H2O2. It was determined from the calculations for this work that structure is a very important factor in the determination of the effectiveness of a catalyst with respect to both activity and selectivity. Our calculations from MedeA-VASP output’ raw’ energies in units of kJ/mol. These must be converted to units of eV and in the same reference basis as used in the work where the prior Volcano Plot was

Conclusion

In this work, we studied the effect of atom arrangement and substitution of select transition metals on the activity and selectivity of Au–Pd based catalyst for the direct synthesis of H2O2. Using Δ Eads (determined using DFT calculations) of monoatomic O and H as descriptors for activity and selectivity along with previously generated data several “good” candidates were determined that could potentially decrease the rare metal content in Au–Pd catalysts. We compared our results both

Data availability

The raw/processed data required to reproduce these findings is available by request to the corresponding authors. They cannot be shared in full to public at this time as they constitute part of an ongoing study.

Author contribution

CT Waldt: Software, Investigation, Formal Analysis, Validation, Writing – Original Draft , Visualization

S Ananthaneni: Investigation, Formal Analysis, Validation, Writing – Original Draft

RB Rankin: Conceptualization, Methodology, Validation, Supervision, Writing (Editing), Project Administraiton, Data Curation , Funding Acquisition, Resources

Declaration of Competing Interests

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.

Acknowledgments

R.B.R. gratefully acknowledges financial and research support from the Department of Chemical and Biological Engineering at Villanova University through his startup package fund. C.W. and S.A. acknowledge financial support from the Department of Chemical and Biological Engineering and the College of Engineering at Villanova University as well. All the authors acknowledge assistance from U.N.I.T. team at Villanova University for maintenance of computational resources necessary in this work.

Rees B. Rankin Assistant Professor. (Ph.D. Chemical Engineering, Carnegie Mellon University 2006)

References (31)

  • R. Kosydar et al.

    Degradation reactions in anthraquinone process of hydrogen peroxide synthesis

    Catal. Lett.

    (2010)
  • H. Xu et al.

    Design of high performance Pd-based alloy nanocatalysts for direct synthesis of H2O2

    ACS Catal.

    (2017)
  • A. Mahata et al.

    Bimetallic core-based cuboctahedral core–shell nanoclusters for the formation of hydrogen peroxide (2e reduction) over water (4e reduction): role of core metals

    Nanoscale

    (2017)
  • M. Nugraha et al.

    Descriptor study by density functional theory analysis for the direct synthesis of hydrogen peroxide using palladium–gold and palladium–mercury alloy catalysts

    Mol. Syst. Des. Eng.

    (2018)
  • T.S. Choksi et al.

    Predicting adsorption properties of cat- alytic descriptors on bimetallic nanoalloys with site-specific precision

    J. Phys. Chem. Lett.

    (2019)
  • Rees B. Rankin Assistant Professor. (Ph.D. Chemical Engineering, Carnegie Mellon University 2006)

    View full text