I am the instrument scientist for the High resolution Infrared Spectrograph for Exoplanet Characterization (HISPEC) project at Caltech. This project is expected to be deployed on the 10 meter Keck II telescope at the W. M. Keck Observatory in Mauna Kea, Hawaii by the end of 2026. I also work on similar instrument, MODHIS, which is in its conceptual design phase for the Thirty Meter Telescope.

Physical Overview

HISPEC has a range of science goals including direct spectroscopy of exoplanets, science of faint dwarfs, and precision radial velocity (PRV) measurements of exoplanets. The last science goal of PRV is the most demanding design-wise and poses milliKelvin temperature stability on the instrument as well as a handful of other requirements.

Due to the extreme stability requirement and limited space on the telescope, the measurement side of HISPEC is decoupled and located in the basement of Keck II in a space originally intended for interferometry. This disjointed instrument design poses the challenge of a long fiber run through the observatory, but gives HISPEC a bit more space to work with for its multiple calibration light sources, its fiber switching mechanisms that moves light around remotely, and the two spectrometers – one for yJ and another for HK.

HISPEC is diffraction-limited instrument meaning it is fed with single-mode fibers that are matched to the telescope’s point spread function (PSF). These fibers are tiny and to have any hope of coupling any light into them useful for doing science, an adaptive optics (AO) system is needed to correct turbulence in Earth’s atmosphere so the telescope PSF looks more like a nice compact Airy disk instead of a giant seeing-limited blob that is moving around left and right!

For HISPEC, we can utilize Keck AO which has a Shack Hartmann wave front sensor to do the adaptive optics correction which also includes a separate camera for doing tip/tilt correction of the PSF. Despite this, and because coupling into single mode fibers is just that difficult, HISPEC has a whole instrument dedicated to getting light into its fibers called the Front End Instrument (FEI). The FEI has lots of cool features to make sure this light gets efficiently coupled into the fibers so the HISPEC spectrographs can do their magic. This includes the ability to back inject light onto the tracking, a fast tip/tilt mirror for keeping the star centered on the fiber, and some beam shaping optics to better match the telescope PSF to the mode field of the fiber.

Differential Limb Coupling

When designing the FEI, we realized that some interesting stuff can happen when you have a single-mode fiber on a big telescope behind an AO system. Stuff that is both good and bad. With a sufficiently good angular resolving power (which comes with a large telescope aperture) things get a little more challenging when measuring radial velocity for single mode fiber-fed spectrographs. Look at the coupling map shown below which shows how much light gets into the fiber as a function of the position from the fiber center, and consider what happens when your rotating star is partially resolved. Do you see how a misalignment might cause errors?

The issue arises when a rotating object is non-centered off of it’s axis of rotation. The side closer to the fiber’s center will couple in better and the final spectrum will be biased toward seeing more of that limb, presenting as a redshift or blueshift of the spectrum. This is problematic when HISPEC will be trying to search for planets by measuring a redshift or blueshift.

What’s the solution to this? Well, there are a few. For one, we could try to better track the true position of our fiber to prevent misalignments. This is hard though due to thermal drifts in the night and we aren’t yet sure how well this can be achieved (though tests are in the works). Otherwise, we could scan the star on the fiber in a way that allows us to average down the bias by purposely sampling either side of the star. For example, we could dither the star around the coupling map by an amount great than our fiber alignment error. This will help equalize the coupling at the cost of a reduction in coupling efficiency. Since only the brightest stars are partially resolved, this can be an acceptable trade. For more information and how you might address the problem without sacrificing photons, see the SPIE proceedings here.