We constructed a nine-dimensional (9D) potential energy surface (PES) for the dissociative chemisorption of H₂O on a rigid Ni(100) surface using the neural network method based on roughly 110 000 energies obtained from extensive density functional theory (DFT) calculations. The resulting PES is accurate and smooth, based on the small fitting errors and the good agreement between the fitted PES and the direct DFT calculations. Time dependent wave packet calculations also showed that the PES is very well converged with respect to the fitting procedure. The dissociation probabilities of H₂O initially in the ground rovibrational state from 9D quantum dynamics calculations are quite different from the site-specific results from the seven-dimensional (7D) calculations, indicating the importance of full-dimensional quantum dynamics to quantitatively characterize this gas-surface reaction. It is found that the validity of the site-averaging approximation with exact potential holds well, where the site-averaging dissociation probability over 15 fixed impact sites obtained from 7D quantum dynamics calculations can accurately approximate the 9D dissociation probability for H₂O in the ground rovibrational state.

中文翻译：

---

刚性Ni（100）上的水离解：全尺寸势能表面的量子动力学研究

我们基于从广泛密度泛函理论（DFT）获得的大约11万能量，使用神经网络方法构造了九维（9D）势能表面（PES），用于H ₂ O在刚性Ni（100）表面上的解离化学吸附）计算。基于较小的拟合误差，以及拟合的PES与直接DFT计算之间的良好一致性，生成的PES是准确且平滑的。与时间有关的波包计算也表明，就拟合程序而言，PES很好地收敛了。H ₂的解离概率最初从9D量子动力学计算得出的基振状态下的O与七维（7D）计算得出的特定于现场的结果有很大不同，这表明了全维量子动力学对于定量表征此气体表面反应的重要性。发现具有精确电势的位点平均近似的有效性保持良好，其中通过7D量子动力学计算获得的15个固定碰撞位点的位点平均解离概率可以准确地近似地面中H ₂ O的9D解离概率。振动状态。