Improving MKID Performance at the UVOIR Wavelengths

Microwave kinetic inductance detectors (MKIDs) operating in the UV, optical, and near-IR wavelengths are single photon counting and energy resolving sensors. However, their spectral resolving power around 1000 nm has remained stubbornly around 8 for the past several years—about an order of magnitude away from what is needed for accurate spectral characterizations of exoplanet atmospheres. Additionally, the quantum efficiency at these wavelengths remains around 30%. Increasing this efficiency is crucial for low light applications like exoplanet detection. This proposal targeted improving these two areas for the next generation of MKID designs.

This project succeeded in identifying phonon losses as the main source of resolving power degradation in our detectors. We used quantum limited parametric amplifiers to significantly reduce the readout noise to show that phonon losses were restricting the resolving power to <10. Introducing an indium phonon-blocking layer in between the device and the substrate then allowed us to double the resolving power to about 20 at 1000 nm. Additionally, we developed a new analysis technique which allows us to extend these results over a wider range of energies, doubling the dynamic range of our detectors. These results may be improved upon by exploring the range of possible phonon blocking layer materials as well as focusing on further reductions to the detector noise.

We also developed an anti-reflection coating for our MKIDs consisting of a two-layer film of Ta2O5 and SiO2 deposited onto the photo-sensitive portion of the sensor. We incorporated this coating on a full 20,000 pixel array and verified that it decreases neither the yield nor the performance of the pixels. In these tests, we saw a modest 25% improvement in the quantum efficiency. The smaller than expected improvement was likely because of poorly tuned coating thicknesses. Further testing of the coatings without the MKIDs showed that we can improve the quantum efficiency from 30% to roughly 70%.