We use more than a decade of radial velocity measurements for \alpha~ { } { Cen } ~ { } A , B , and Proxima Centauri from HARPS , CHIRON , and UVES to identify the M \sin i~ { } and orbital periods of planets that could have been detected if they existed .
At each point in a mass-period grid , we sample a simulated , Keplerian signal with the precision and cadence of existing data and assess the probability that the signal could have been produced by noise alone .
Existing data places detection thresholds in the classically defined habitable zones at about M \sin i~ { } of 53 M _ { \oplus } ~ { } for \alpha~ { } { Cen } ~ { } A , 8.4 M _ { \oplus } ~ { } for \alpha~ { } { Cen } ~ { } B , and 0.47 M _ { \oplus } ~ { } for Proxima Centauri .
Additionally , we examine the impact of systematic errors , or “ red noise ” in the data .
A comparison of white- and red-noise simulations highlights quasi-periodic variability in the radial velocities that may be caused by systematic errors , photospheric velocity signals , or planetary signals .
For example , the red-noise simulations show a peak above white-noise simulations at the period of Proxima Centauri b .
We also carry out a spectroscopic analysis of the chemical composition of the \alpha { Centauri } stars .
The stars have super-solar metallicity with ratios of C/O and Mg/Si that are similar to the Sun , suggesting that any small planets in the \alpha~ { } { Cen } ~ { } system may be compositionally similar to our terrestrial planets .
Although the small projected separation of \alpha~ { } { Cen } ~ { } A and B currently hampers extreme-precision radial velocity measurements , the angular separation is now increasing .
By 2019 , \alpha~ { } { Cen } ~ { } A and B will be ideal targets for renewed Doppler planet surveys .