Standardized dataset of CO-TPD spectra on transition-metal single-crystal surfaces

This study constructs a standardized CO-TPD dataset covering 717 experimental records published between 1976 and 2023. It involves single-crystal surfaces of 13 transition metals, including Ag, Au, Co, Cu, Fe, Ir, Ni, Os, Pd, Pt, Re, Rh, Ru, and W, encompassing both low-Miller-index facets (such as (111), (110), and (100)) and stepped surfaces (e.g., (355), (533)), with the data focused on the scale of surface science experiments. The dataset was built by systematically retrieving literature, collecting original spectra, and digitizing the images using WebPlotDigitizer software to extract raw coordinate points. Subsequently, baseline correction and normalization based on maximum peak height were applied to eliminate variations in signal intensity across different studies while fully preserving spectral features such as peak shape. During data generation, 14 parameters were recorded, including metal type, crystal orientation, initial temperature (in K), pre-adsorption conditions (in molecules/cm², L, mbar·s, or ML), heating rate (in K/s), peak temperature (in K), peak type, temperature measurement method, and baseline quality. All data are stored in structured CSV and ZIP files. The core index file, TPD_list.csv (84.3 KB), contains 717 entries, each row corresponding to one TPD measurement, with column labels representing the aforementioned parameters. TPDdata.zip (11.4 MB) contains the original spectral images and the unscaled extracted data points, while TPDdata_scaled.zip (645 KB) provides normalized standard data points, facilitating cross-study comparison. Some entries in the dataset have incomplete information, such as heating rates or adsorption doses annotated as "~5 K/s" or "<12 L," which are explicitly marked in the corresponding fields. Through rigorous quality control—including the inclusion of only clean single-crystal surface data, cross-verification of peak temperatures with literature values, interpolation, and removal of anomalous fluctuations—the dataset comprises 478 high-reliability entries and 239 entries with lower reliability. The latter exhibit some uncertainty in parameter annotation and baseline quality but remain valuable for reference. All files in this dataset are in common formats (CSV and ZIP) and can be accessed and processed without specialized software.