We present a strategy, by taking a prototypical model system for photocatalysis (viz. N-doped (TiO₂) _n clusters), to accurately design photocatalyst with enhanced reactivity at a given environmental conditions (i.e. temperature (T), pressure () and doping ( _e )). Since free energy potential energy surface consists of many competing isomers even in a small energy window, computational design of specific metastable photocatalysts with enhanced activity is extremely challenging. This requires fixing various parameters as follows; (i) favorable formation energy, (ii) low fundamental gap, (iii) low excitation energy, (iv) high vertical electron affinity (VEA) and (v) low vertical ionization potential (VIP) to drive the hydrogen evolution reaction (HER) and oxygen evolution reaction (OER) for water splitting. Here, we show that by integrating several first principles based methodologies that consideration of the global minimum structure alone can severely underestimate the activity. First, we have used a suite of genetic algorithms (namely, searching clusters with conventional minimum total energy ((GA)); searching clusters with specific property i.e. high VEA ((GA)), and low VIP ((GA))) to model the N-doped (TiO₂) _n ( n = 4–10, 15, 20) (meta)stable clusters. Next, we have identified its free energy using ab initio atomistic thermodynamics to confirm that the metastable structures are not too far from the free energy global minima. We find that N-substitution ((N)) prefers to reside at highly coordinated oxygen site, whereas N-interstitial ((NO)) and split-interstitial ((N) favor the dangling oxygen site. (NO) and (N doped states are thermodynamically stable at finite temperature (T) and pressure (). Interestingly, each types of defect (viz. substitution, interstitials) reduce the fundamental gap and excitation energy substantially. However, (N doped clusters are found to be less probable in the Pourbaix phase diagram, whereas (N) and (NO) doped metastable clusters show significant occupance probability near the phase boundaries. The latter ensures much higher electrocatalytic activity for water splitting than the most stable configurations.

我们提出一种策略，通过采用光催化的原型模型系统（即N掺杂（TiO ₂）_n团簇），在给定的环境条件（例如温度（T），压力（）和掺杂 _Ë ））。由于自由能势能表面即使在很小的能量窗口内也由许多竞争性异构体组成，因此具有增强活性的特定亚稳光催化剂的计算设计极具挑战性。这需要固定各种参数，如下所示；（i）有利的形成能；（ii）低的基极能隙；（iii）低的激发能；（iv）高的垂直电子亲和力（VEA）和（v）低的垂直电离势（VIP）以驱动氢释放反应（HER ）和氧气分解反应（OER）进行水分解。在这里，我们表明，通过整合基于几个基本原理的方法，仅考虑全局最小结构可能会严重低估活动。首先，我们使用了一套遗传算法（即以常规最小总能量（（GA））; 搜索具有特定属性的簇，即高VEA（（GA））和低VIP（（GA））），以模拟N掺杂（TiO ₂）_n（n = 4-10、15、20）（亚）稳定簇。接下来，我们使用从头算式原子热力学确定了其自由能，以确认亚稳结构与自由能全局极小值相距不远。我们发现N取代（（N））倾向于驻留在高度协调的氧位点，而N间隙（（NO））和分裂间隙（（N ）则有利于悬挂的氧位。（NO）和（N掺杂态在有限温度（T）和压力（T）下热力学稳定。 ）。有趣的是，每种类型的缺陷（即缺陷，填隙）都会大大减小基本间隙和激发能。但是，（在Pourbaix相图中发现N掺杂的团簇可能性较小，而（N）和（NO）掺杂的亚稳簇在相界附近显示出显着的占据概率。后者确保了对水分解的电催化活性远高于N）。最稳定的配置。