Optimal Electric Field Estimation for Exoplanet Imaging Observatories in Space

Just over two decades ago the first planet outside our solar system was found, and thousands more have been discovered since. Nearly all these exoplanets were indirectly detected by sensing changes in their host stars’ light. However, exoplanets must be directly imaged to determine their atmospheric compositions and the orbital parameters unavailable from only indirect detections. The main challenge of direct imaging is to observe stellar companions much fainter than the star and at small angular separations. Coronagraphy is one method of suppressing stellar diffraction to provide high star-to-planet contrast, but coronagraphs are extremely sensitive to quasi-static aberrations in the optical system. Active correction of the stellar wavefront is performed with deformable mirrors to recover high-contrast regions in the image. Estimation and control of the stellar electric field is performed iteratively in the camera’s focal plane to avoid non-common path aberrations arising from a separate pupil sensor. Estimation can thus be quite time consuming because it requires several high-contrast intensity images per correction iteration. Research in coronagraphy and focal plane wavefront correction (FPWC) is of high priority for NASA’s proposed Wide Field Infrared Survey Telescope’s (WFIRST’s) Coronagraph Instrument (CGI), which is in Phase A of development. The research undertaken during this fellowship award advanced both coronagraphy and focal plane wavefront estimation (FPWE) for the WFIRST CGI and other proposed exoplanet imaging missions. The major focus of this fellowship work was on efficient FPWC for coronagraphy. Observation time is a precious commodity for a space telescope, so there is a strong incentive to reduce the total exposure time required for FPWE. Much of my work emphasized faster, more robust estimation via Kalman filtering, which optimally combines prior data with new measurements. In coronagraphic lab experiments at Princeton’s High Contrast Imaging Laboratory (HCIL) and the Jet Propulsion Laboratory’s (JPL’s) High Contrast Imaging Testbed (HCIT), I demonstrated that Kalman filtering enables faster FPWC than the commonly used batch process estimation method. This result means either that a higher contrast is reached in an allotted amount of correction time, or that less time is necessary to reach a given contrast. In FPWC simulations of the WFIRST CGI, I showed that Kalman filtering enables much shorter exposure times per estimation image than the batch process estimator, which can allow an order of magnitude less total exposure time for FPWC. The primary contribution of this thesis is a paradigm shift in the use of estimation images. Time for FPWC has generally been considered to be lost overhead, but I demonstrated that estimation images for FPWC can be used for the detection and characterization of exoplanets and disks. These science targets are incoherent with their host stars, so I implemented an iterated extended Kalman filter (IEKF) for simultaneous estimation of the stellar electric field and the incoherent signal. In Princeton’s HCIL, I demonstrated in four different cases that the IEKF could disambiguate an injected incoherent planet signal from the starlight during FPWC. I then performed a statistical analysis of this method with Monte Carlo simulations of FPWC for the WFIRST CGI. I showed that with photon shot noise and some model error that the IEKF could still detect exoplanets up to ten times fainter than the residual starlight. This method of exoplanet detection offers a different approach from conventional, state-of-the-art image post-processing techniques. Exoplanet detection during FPWC does not require slewing or rolling of the telescope as the other methods do, so the stellar PSF is less affected by thermally-induced wavefront changes. My work was the first demonstration of integrated wavefront correction and exoplanet detection and will need much follow up to characterize it’s possible performance on future space observatories. The secondary focus of my work addressed coronagraphs. Neil Zimmerman, during his time at Princeton University, and I created a new type of coronagraph combining a shaped pupil apodizer and a Lyot-type coronagraph. We used linear optimization techniques in the development of this shaped pupil Lyot coronagraph (SPLC). We then designed SPLCs to meet the science requirements of the WFIRST CGI, and the SPLC is baselined as one of the two coronagraphs for that instrument. I worked closely with JPL’s Microdevices Laboratory (MDL) to tolerance and manufacture designs for use in JPL’s HCIT. I also collaborated with the HCIT team to perform wavefront correction experiments with the SPLC.