1-13. High-throughput calculation-aided exploration of performance descriptor of single atom-doped MoS2 nanosheet for OER

Ge Wang, Guangtong Hai, Hongyi Gao, Wenjun Dong

Beijing Advanced Innovation Center for Materials Genome Engineering, University of Science and Technology Beijing, Beijing, 100083

Abstract: Oxygen evolution reaction (OER) is the fundamental technology of fuel cell, metal-air battery and other next generation clean and efficient energy system. Because of the multi-step electron transfer process, OER usually requires a high overpotential to drive, resulting in a series of problems, such as low energy conversion efficiency and serious side effects. Therefore, OER usually requires the introduction of catalysts to effectively reduce the overpotential, and the development of novel catalysts with high activity and low cost is a core subject in this field.

        Based on the high-throughput concurrent thermodynamic calculation, the OER theoretical overpotential of catalytic materials could be predicted, and then rapidly screen out new high-performance OER catalysts, so as to effectively shorten the research cycle and reduce the development cost

        Due to many advantages of single atom load type catalyst, we take the materials genome thoughts as the guideline, in terms of high-throughput calculation as the main research method, and further by means of data correlation analysis technology, systematically investigated the structure-activity relationship of disulfide nanosheets doped with isolated single atom (M-UMONs) in OER catalytic system (Figure 1). Through high-throughput modeling and structural optimization, and then theoretically predict the structure of disulfide nanosheets doped with different single metal atoms as well as the synthetic feasibility. On this basis, we further studied the free energy change trend of different M-UMONs and then obtain the theoretical overpotential accordingly. Taking the theoretical overpotential as the "efficiency index" and different structural parameters (such as Fermi energy level, d orbital center, eg orbital electron number, etc.) as the "structure index", data correlation analysis technology is applied to dig into the correlation relationship between the two indexes. Through a series of analysis and screen of potential descriptor, we found that the difference between surface of metal-oxygen bonds (M-O) and metal-sulfur chemical bonds (M-S) orders is linearly correlated with the overpotential. Hence we selected the difference between surface M-O and M-S bond orders as a descriptor OER activity. At the same time, compared with the timeliness of high-throughput computing, descriptor-based filtering can reduce at least one or two orders of magnitude of computation while still ensuring the accuracy and reliability of the predicted results, thus greatly reducing the computational cost. To experimentally verify the predictive power of the descriptor, we synthesized more than ten kinds of different M-UMONs materials and performed a series of characterizations and OER activity measurements, which strongly confirmed the reliable predictive power of the descriptor.

      This research idea changes the material screening mode from "high-throughput computational screening" to "description-based screening", realizing a breakthrough in the research mode of new materials field and effectively accelerating the material screening process. At the same time, the introduction of descriptor not only profoundly reflects the nature of the structure-activity relationship of materials, but also provides a convenient means for material performance prediction and screening, which is of great significance for the research and development of new high-performance catalytic materials.

Figure 1. (a) The correlation between the theoretical overpotentials and M-S bond orders.
(b) The correlation between the theoretical overpotentials and the difference between M-S and M-O bond orders.

图 1. (a): M-UMONs 催化 OER 的理论过电位与 M-S 键级的关系。(b): M-UMONs 催化 OER 的理论过电位和 M-S 与 M-O 键级差的关系

基于高通量计算挖掘单原子负载型 OER 催化材料的活性描述符 

王戈，海广通，高鸿毅，董文钧

北京材料基因工程高精尖创新中心，北京科技大学，北京 100083 

摘要:析氧反应(OER)是燃料电池、金属-空气电池等下一代清洁高效能源系统的核心技术。 由于涉及多步电子转移过程，OER 通常需要较高的过电位来驱动，导致能源转化效率低、副反 应严重等一系列问题。因此，OER 通常需要引入催化剂来有效地降低过电位，开发高活性、低 成本的新型催化剂是目前 OER 研究领域的核心任务。

       以高通量并发式的热力学计算为依据来衡量催化材料的 OER 过电位，从而快速筛选出新型 高性能 OER 催化材料，以实现有效地缩短研发周期，降低研发成本，是当前 OER 领域研究的 一种新模式，也是材料基因工程理念的重要体现。

      鉴于单原子负载型催化剂的诸多优点，我们以材料基因组思想为指导，以高通量计算为主 要研究手段，并结合数据关联性分析技术，对单原子负载型的二硫化物纳米片(M-UMONs)在 OER 催化体系中的构效关系进行了系统地研究(图 1):通过高通量建模与结构优化，从理论上 预测了不同金属原子负载的二硫化物纳米片的结构以及合成制备的可行性，在此基础上，针对 OER 反应的多电子转移过程，进一步研究了不同的 M-UMONs 在催化 OER 过程中的自由能变 化趋势，据此得到催化 OER 的理论过电位。以理论过电位为“效能指标”，以不同的结构参数(例 如费米能级、d 轨道中心，eg 轨道电子数等)为“结构指标”，采用数据关联性分析技术深入挖掘 两个指标之间的关联关系。通过对一系列潜在的描述因子进行分析筛选，发现表面金属-氧化学 键(M-O)的键级和表面金属-硫化学键(M-S)的键级差与 OER 过电位之间存在显著的线性关系，因此选择表面金属-氧化学键(M-O)的键级和表面金属-硫化学键(M-S)的键级差作为OER 活性的描述符。同时，相对于高通量计算的时效性，基于描述符的筛选可以至少降低 1~2 个数量级的计算量而依然能够保证预测结果的准确性与可靠性，实现了计算成本的大幅度降低。 为了从实验上验证描述符的预测能力，我们合成了十余种不同的 M-UMONs 材料，并对其进行 一系列表征和 OER 活性测试，有力地证实了该描述符可靠的预测能力。

      该研究思路将材料的筛选模式从“基于高通量计算筛选”变革为“基于描述符筛选”，实现了 新材料领域研究模式的一次突破，有效地加速了材料的筛选进程。同时，描述符的提出，不仅 深刻反映了材料构效关系的本质，更提供了一种材料性能预测与筛选的便捷手段，对新型高性 能催化材料的研发具备重大意义。

关键词:高通量计算;密度泛函理论;描述符;析氧反应