1：Experimental Design and Method Selection:

The study employed an artificial-intelligence-assisted Raman analysis method to quickly probe the quality of centimeter-large graphene samples in a non-destructive manner. Chemical vapor deposition (CVD) of graphene was used to obtain two types of samples: layer-plus-islands and layer-by-layer graphene films. An unsupervised learning algorithm was integrated with automated Raman spectroscopy to cluster Raman spectra collected from these samples.

2：Sample Selection and Data Sources:

Graphene samples were grown on bare and treated Cu substrates using a low-pressure chemical vapor deposition (LPCVD) system. The samples were then transferred onto Si/SiO2 substrates for characterization.

3：List of Experimental Equipment and Materials:

A 532 nm excitation laser with a 100 × objective lens (WITec Alpha 300 micro-Raman imaging system) was used for Raman spectra acquisition. The laser power was kept below 80 μW. Cu foil from Alfa Aesar (#46365) was used as the growth substrate.

4：Experimental Procedures and Operational Workflow:

Graphene samples were grown on Cu substrates, transferred to Si/SiO2 substrates, and then analyzed using Raman spectroscopy. The k-means clustering algorithm was applied to classify the Raman spectra.

5：Data Analysis Methods:

The k-means algorithm from the Scikit-learn package was used to cluster the Raman spectra based on three parameters: ωG, ω2D, and Γ2D.