Rapid evaluation method for anisotropic growth of WS2 monolayers by combining machine learning algorithms and kinetic Monte Carlo simulation data

When Machine Learning Meets 2D Materials: A Review

“…[ 94 ] Since the materials that are to be prepared can differ, the experiment conditions for CVD can vary as well, and hence the ML databases in this case are often small databases. [ 93 ] CVD can be applied to the preparation of high‐quality 2D materials, [ 186 ] and ML‐enabled CVD preparation of 2D materials includes WTe 2 , [ 93 ] MoS 2 , [ 94 , 96 ] WS 2 [ 97 , 98 ] and h‐BN. [ 187 ] For instance, the trained model proposed by Xu et al can optimize the CVD synthesis parameters (reaction temperature, rising time, deposition time, airflow rate) to enable controllable growth of multilayer 1D WTe 2 .…”

“…[ 94 ] Xia et al developed a DL‐based framework for the analysis of data from kinetic Monte Carlo simulations, and employed KNN, SVM, and RF classifiers to predict the anisotropic growth of WS 2 monolayers. [ 97 ]…”

“…[94] Xia et al developed a DL-based framework for the analysis of data from kinetic Monte Carlo simulations, and employed KNN, SVM, and RF classifiers to predict the anisotropic growth of WS 2 monolayers. [97] As pointed out in Zhang et al's work, the major challenge facing controllable CVD synthesis of 2D materials involves complex correlations between variables in CVD growth, which means that the proper control of these variables is a must to regulate the inherent thermal and dynamic properties of crystal growth. [95] To address this problem, they started the Materials Genome Initiative (MGI) to establish morphological diagrams of 2D crystals based on crystal growth experiments, modeling, and databases, as shown in Figure 10b.…”

“…[ 78 , 79 , 80 , 81 , 82 , 83 , 84 , 85 , 90 , 91 , 92 ] In terms of preparing 2D materials, ML methods have been applied to deposition and exfoliation to enable easier and more controllable preparation of 2D materials such as WTe 2 , [ 93 ] MoS 2 , [ 94 , 95 , 96 ] and WS 2 . [ 97 , 98 ] When it comes to characterizing 2D materials, ML has been combined with characterization techniques, like Raman spectroscopy, transmission electron microscopy, optical microscopy, and imaging, to acquire accurate information such as the thickness [ 99 , 100 , 101 , 102 , 103 , 104 , 105 , 106 , 107 , 108 , 109 , 110 , 111 , 112 , 113 ] and defects [ 114 , 115 , 116 , 117 ] of materials. The utilization of ML in 2D materials research, along with the algorithmic processing of experimental data, has the potential to support data analysis, leading to more conclusive results on the reliability and reproducibility of given datasets.…”