Table 1 Configurations included in the complete training dataset, encompassing both the primary training set and additional structures incorporated during iterative learning

From: Neural network-driven molecular insights into alkaline wet etching of GaN: toward atomistic precision in nanostructure fabrication

Category

 

Structure type

Number of structures

Number of atoms

Bulks

GaN

747

68,884

 

KOH

1222

197,964

Reaction products

Ga2O3+NH3

414

53,820

 

Ga(OH)3+NH3

414

45,540

 

Ga(OH)3+NH3+mH2O

414

46,368

 

[Ga(OH)4]+K++NH3

414

47,196

 

[Ga(OH)4]+K++NH3+mH2O

207

27,738

 

[Ga(OH)4]+K++H2O

414

55,890

 

NH3+K+OH+H2O

414

53,406

GaN/solution interface

m-plane

Top

668

156,312

  

Sub

668

156,312

 

Iterative learning

5308

1,554,126

Total

11,304

2,463,556