Title: Systematic bias due to eccentricity in parameter estimation for merging binary neutron stars

Abstract: We study the impact of eccentricity on gravitational-wave parameter estimation for binary neutron star systems. For signals with small eccentricity injected into the advanced LIGO sensitivity, we perform Bayesian parameter estimation using the circular waveform model and show how the recovered parameters can be biased from their true values, focusing on the intrinsic parameters the chirp mass ($M_c$), the symmetric mass ratio ($\eta$), and the tidal deformability ($\tilde{\lambda}$). By comparing the results between the Bayesian and the analytic Fisher-Cutler-Vallisneri (FCV) methods, we obtain the valid criteria for the FCV approach. Employing the FCV method and using the realistic population of binary neutron star sources distributed in the $m_1$-$m_2$-$e_0$ space, where $e_0$ indicates the eccentricity at 10Hz, we calculate the measurement errors ($\sigma_{\theta}$) and the systematic biases ($\Delta \theta/\sigma_{\theta}$) and obtain their generalized distributions in the range of $0 \leq e_0 \leq 0.025$. We find that for all of the three parameters, the biases increase with increasing $e_0$, and this increase is faster for larger $e_0$. The bias is mainly dependent on the value of $e_0$ and weakly dependent on the component masses, and thus the distribution shows a narrow band in the $e_0$-$\Delta \theta/\sigma_{\theta}$ plane. We present various posterior examples to illustrate our new findings, such as the bimodality of posteriors. In particular, we give a specific injection-recovery example to demonstrate the importance of including eccentricity in parameter estimation to avoid incorrect predictions of the neutron star equation of state.