Pareto-Guided Process Optimization of Single-Crystal InP Chemical Mechanical Polishing: Balancing Material Removal Rate and Surface Roughness

Single-crystal indium phosphide (InP) is an important substrate material for high-speed electronic and optoelectronic devices. Its relatively low hardness, brittleness, and sensitivity to polishing chemistry make it challenging to achieve both efficient material removal and low surface roughness during chemical mechanical polishing (CMP). In this study, response surface methodology (RSM) was combined with multi-objective particle swarm optimization and the Technique for Order Preference by Similarity to Ideal Solution (MOPSO–TOPSIS) to investigate the balance between material removal rate (MRR) and surface roughness (Ra). A three-factor Box–Behnken design was used to establish quadratic models incorporating H2O2 concentration, SiO2 abrasive concentration, and polishing speed. The models gave coefficients of determination of 0.9898 for MRR and 0.9894 for Ra. Abrasive concentration had the strongest influence on both responses within the investigated ranges. Higher abrasive concentrations favored material removal, whereas lower roughness was obtained at intermediate concentrations and polishing speeds. A compromise condition of 9.79 wt% H2O2, 1.08 wt% SiO2, and 54.33 rpm was evaluated experimentally. The predicted MRR and Ra were 3871.32 nm/h and 0.6747 nm, respectively, compared with measured values of 4283.79 nm/h and 0.712 nm. The corresponding relative deviations were 10.65% and 5.53%. These results provide a quantitative basis for selecting CMP parameters that balance removal efficiency and surface roughness in single-crystal InP polishing.