Signatures of unresolved binaries in stellar spectra: implications for spectral fitting

The observable spectrum of an unresolved binary star system is a superposition of two single-star spectra. Even without a detectable velocity offset between the two stellar components, the combined spectrum of a binary system is in general different from that of either component, and fitting it with single-star models may yield inaccurate stellar parameters and abundances. We perform simple experiments with synthetic spectra to investigate the effect of unresolved main-sequence binaries on spectral fitting, modeling spectra similar to those collected by the APOGEE, GALAH, and LAMOST surveys. We find that fitting unresolved binaries with single-star models introduces systematic biases in the derived stellar parameters and abundances that are modest but certainly not negligible, with typical systematic errors of $300,rm K$ in $T_{rm eff}$, 0.1 dex in $log g$, and 0.1 dex in $[rm Fe/H]$ for APOGEE-like spectra of solar-type stars. These biases are smaller for spectra at optical wavelengths than in the near-infrared. We show that biases can be corrected by fitting spectra with a binary model, which adds only two labels to the fit and includes single-star models as a special case. Our model provides a promising new method to constrain the Galactic binary population, including systems with single-epoch spectra and no detectable velocity offset between the two stars.