A classic and extremely productive technique for detecting planets around other stars is to measure the tiny reflex motion their gravity imparts to the parent stars. The signature of that motion is the presence of small radial velocity shifts in the stellar spectra. However, to detect the most interesting rocky (or earthlike) planets (one of the top three scientific priorities laid out by the 2010 Decadal Survey "New Worlds, New Horizons") will require an improvement in measurement precision by a full factor of ten over the current state of the art, which is roughly 1 meter per second. Extraordinary control of systematic error sources will therefore be required to detect a planet comparable to our Earth. The "New Worlds, New Horizons" report indeed recommended improvement of radial velocity techniques as the highest priority for ground-based exoplanet research.
Dr. D. Fischer and coworkers at Yale University are investigating a number of aspects of the problem that involve determining the optimal coupling of light from the telescope into the spectrograph, the instrument that actually measures the radial velocity of interest. To control variations in illumination of the optical fiber that first accepts light from the telescope and then channels it into the spectrograph, octagonal fibers will be investigated for their superior "scrambling" of input modes relative to a slit or round fiber. Scrambling in this way makes the output intensity less sensitive to details of the input illumination. A thorough study of this and other effects, such as how the image is sliced into different contributions for spectral analysis, will be performed, and when the comprehensive study is completed, the conclusions are expected to lead to substantial reductions in the systematic sources of error for radial velocity planet finding.
In situ tests of the recommendations emerging from this work will be carried out using the highly stable cross-dispersed echelle spectrometer CHIRON on the Cerro Tololo 1.5 meter telescope in Chile, as well as on the HIRES spectrometer at Keck Observatory on Mauna Kea, Hawaii. However, the results will have implications for all spectrographs targeted for extremely high velocity precision.
Funding for this work is being provided by NSF's Division of Astronomical Sciences through its Advanced Technologies and Instrumentation program.
Hundreds of exoplanets (planets orbiting nearby stars) have been detected by measuring the radial velocity of the host star, a method called the "Doppler technique." The precision of radial velocity measurements is not sufficient to detect analogs of the Earth (i.e., planets that have a mass between 0.7 and a few times the mass of the Earth, and orbiting at distances from their host stars where liquid water might pool on the surface of the planet). Before this NSF award, we obtained Doppler measurements by focusing light collected from the telescope onto the entrance slit of the spectrometer (the instrument that disperses light into component wavelengths that can be used to measure stellar velocities). This was the approach used by most astronomers. However, that approach results in variable illumination of the optics because of focus changes of the telescope, guiding errors or scintillation in the atmosphere. As a result, two consecutive observations will give slightly different results - not because of photon statistics but because the starlight is shining on the optics differently. More importantly, these changes are modeled by our computer code as tiny variations in the velocity of the star and they prevent us from detecting low mass planets. We wanted to find a way to stabilize the illumination with the goal of improving our measurement precision. To reach this goal, we began investigating the use of fiber optics. Fiber optics have the effect of scrambling the light so that the spatial illumination at the front of the fiber is not remembered. As a result, the light emerging from the fibers is (more or less) independent of the angle and the intensity distribution that goes into the fiber. We hypothesized that this stabilization would improve our precision and enable the detection of lower mass planets. We tested different geometries of optical fibers (circular, square, rectangular and octagonal) and we also quantified the impact of scrambling the light at two different parts of the beam. The first part of the beam is the in-focus image at the entrance to the spectrometer, also called the "near field." The near field is always in focus and this is what is detected by our cameras. The second part of the beam is the diverging or collimated beam the pupil plane (the "far field"). This part of the beam illuminates the optics inside of the spectrometer. We learned that multimode fibers with octagonal cross-sections provide the best scrambling of the near field image. We also learned that we could further stabilize the illumination of the optics by scrambling the far field. We built a system that used an octagonal fiber to collect light from the telescope (and scrambled the near field) and then sliced the far field image and stacked the two half moons before injecting into a rectangular fiber. The result of this work showed that using a circular fiber to scramble only the near field light improved the illumination stability by a factor of 10. However, we did not see any improvement in our Doppler precision. We redesigned our system to use an octagonal fiber and a rectangular fiber (scrambling both the near field and the far field), giving us a factor of 50 improvement in illumination stability. That is, from exposure to exposure, we saw that the variability of the recorded image decreased by a factor of 50. With the double-scrambler, we also saw improvement in the radial velocity precision of every star we observed. In most cases, the measurement errors dropped by a factor of two; in one case, the improvement was 400%. This was a critical benchmark. We improved our radial velocity precision, but we also learned that our Doppler technique will require a new wavelength calibration to gain significant precision towards the search for Earth analogs around nearby stars. With the knowledge gained here, we are now beginning that work.
