Robust Evidence for the Breakdown of Standard Gravity at Low Acceleration from Statistically Pure Binaries Free of Hidden Companions

It is found that Gaia DR3 binary stars selected with stringent requirements on astrometric measurements and radial velocities naturally satisfy Newtonian dynamics without hidden close companions when projected separation $s \lesssim 2$ kau, showing that pure binaries can be selected. It is then found that pure binaries selected with the same criteria show a systematic deviation from the Newtonian expectation when $s \gtrsim 2$ kau. When both proper motions and parallaxes are required to have precision better than 0.005 and radial velocities better than 0.2, I obtain 2,463 statistically pure binaries within a `clean' $G$-band absolute magnitude range. From this sample, I obtain an observed to Newtonian predicted kinematic acceleration ratio of $\gamma_g=g_{\rm{obs}}/g_{\rm{pred}}=1.49^{+0.21}_{-0.19}$ for acceleration $\lesssim 10^{-10}$ m s$^{-2}$, in excellent agreement with $1.49\pm 0.07$ for a much larger general sample with the amount of hidden close companions self-calibrated. I also investigate the radial profile of stacked sky-projected relative velocities without a deprojection to the 3D space. The observed profile matches the Newtonian predicted profile for $s \lesssim 2$ kau without any free parameters but shows a clear deviation at a larger separation with a significance of $\approx 5.0\sigma$. The projected velocity boost factor for $s\gtrsim 5$ kau is measured to be $\gamma_{v_p} = 1.20\pm 0.06$ (stat) $\pm 0.05$ (sys) matching $\sqrt{\gamma_g}$. Finally, for a small sample of 40 binaries with exceptionally precise radial velocities (fractional error $<0.005$) the directly measured relative velocities in the 3D space also show a boost at larger separations. These results robustly confirm the recently reported gravitational anomaly at low acceleration for a general sample.