2023ApJ...950..104N

Stellar chemical abundances are crucial and fundamental in astrophysics. However, they could suffer from substantial systematic errors according to several investigations but still lack calibrations in bulk. By using Gaia wide binaries, we find the temperature-dependent bias between the two binary components for [Fe/H] and [α/Fe] measurements from the LAMOST low-resolution spectra and Gaia RVS spectra. At T_eff = 4000 K, the LAMOST [Fe/H] is significantly underestimated by approximately 0.4 dex when compared with its typical uncertainty of 0.1 dex. Its [α/Fe] is overestimated by about 0.2 dex. For Gaia, the underestimation of [M/H] and overestimation of [α/Fe] becomes pronounced near 7000 K with smaller magnitudes. We perform an internal calibration by minimizing the differences between the binary components and provide the correction curves. After corrections, the standard deviations of the residuals compared to the PASTEL catalog decrease from about 0.045/0.1 to 0.02/0.043 for LAMOST and Gaia, respectively. The chemical homogeneity of the open cluster M 44 is also improved by a factor of two. We stress that the underestimation of [Fe/H] could lead to an overestimation of binary fractions when selecting binary stars by the excess of luminosity. The method of this work could be applied to other data sets in the future. Our results will benefit statistic studies that use LAMOST and Gaia samples with a wide temperature range.