Crashing through the semiconductor ceiling

The sensitivity of astronomical observations is driven by the size of the telescope's light-gathering area and the sensitivity of its detectors. In the last 50 years, the collecting area of ground-based optical and near-IR telescopes has increased in size by only a factor of four, from the Palomar 200-inch to the 10m Keck telescope. The sensitivity of detectors has increased on a per-pixel basis by at least a factor of 20 as the field transitioned from photographic plates to CCDs. The improvement in quality and size of these optical and near-IR detectors—from the millimeter to the x-ray—has been even more impressive.^{1, 2} However, traditional semiconductor-based detectors (CCD)³ and mercury cadmium telluride, or HgCdTe⁴) are rapidly reaching a plateau in their per-pixel performance, with improvements only coming in the total pixel count of the final mosaic.⁵

To further improve detectors for astrophysics we need to develop other technologies. Potential enhancements for optical and near-IR detectors include reducing or eliminating read noise and dark current (electrons already present), having a wider detection bandwidth, or having inherent spectral resolution.

The most promising solution is the use of superconducting detectors.^6–9 These significantly increase performance by operating below 4 Kelvin and reducing thermal noise, allowing them to measure the energy and arrival time of a single photon with no false counts. Our laboratory is pursuing a relatively new type of superconducting detector called a microwave kinetic inductance detector (MKID).^{6, 10,11} The device comprises a superconducting LC oscillator with a resonant frequency in the microwave region (1–10GHz): see Figure 1. Photons hitting the inductor break Cooper Pairs (pairs of electrons), causing a change in the surface impedance of the superconducting film through the kinetic inductance effect.¹² This change alters the electrical properties of the resonator, allowing a microwave probe signal (consisting of a sine wave at the resonant frequency of the MKID) to readout the low frequency changes (∼1–100kHz) in phase and amplitude caused by the photon. This results in a phase pulse that rises quickly and decays with a time constant equal to the quasiparticle lifetime in the superconductor (typically 10–100μs). The height of this phase pulse gives the energy of the incident photon to several percent, and the arrival time can be determined from the start time of the pulse to several microseconds. There are no false counts—cosmic rays can be rejected since they will trigger many pixels at the same time—and there is no analog to read noise or dark current in a MKID.

Figure 1. A summary of the detection principle of a microwave kinetic inductance detector (MKID). (a) An incident photon breaks Cooper Pairs in a superconducting film that is part of a resonant circuit (b), causing a change in the dissipation (c) and phase (d) of the resonator. (Reprinted with permission).⁶

One of the primary benefits of the technology is that, with their unique resonant frequencies, MKIDs have inherent frequency domain multiplexing—the capacity to transmit several signals over a single channel. We can read out thousands of MKIDs over a single microwave feed line by probing each MKID with a unique sine wave contained in a frequency comb. A 2024 pixel MKID array is shown in Figure 2. The device has the potential to scale to megapixel arrays.

Figure 2. Photograph of a 2024 pixel UV, optical, and IR MKID array in a microwave package. (Reprinted with permission).¹²

We are using MKIDs in the Array Camera for Optical to Near-IR Spectrophotometry (ARCONS)¹³ for the Palomar 200″ and Lick 120″ telescopes. ARCONS has been on these telescopes for 24 observing nights, and has collected data on optical pulsars, compact binaries, high redshift galaxies, planetary transits, and more. As an example, Figure 3 shows a mosaic of the interacting galaxies Arp 147. For observations of rare objects, ARCONS' small field of view (20″×20″) is not a drawback. Rather, it enables the instrument to realize an increase (by a significant order of magnitude) in observing efficiency over filter-based multicolor photometry.

Figure 3. A mosaic of the interacting galaxies Arp 147 made with the ARray Camera for Optical to Near-IR Spectrophotometry (ARCONS) on the Palomar 200″ telescope. The data consists of 36 pointings of 1 min each. The colors were made by breaking the ARCONS data into three wavelength bands and combining. The inset shows a processed image of Arp 147. (Reprinted with permission).¹³

The cryogenic requirements of MKIDs are easily handled with modern cooling techniques even at remote locations, although the cost of these processes will limit their commercial applications. Nonetheless, there are many potential scientific applications for the technology in other fields, such as biological imaging and quantum optics.

In future work, MKIDs will be used in experiments to directly image extrasolar planets. The read noise-free photon- counting MKIDs will allow rapid feedback from the science array to the deformable mirror, allowing us to reduce speckles caused by atmospheric turbulence in a way that is not possible with current detectors. In the long term, MKIDs will be at the heart of superconducting multiobject spectrographs (SuperMOS),¹⁴ which will provide low-resolution spectroscopy (0.32–1.35μm) for billions of objects.

The MKID detectors used in this work were developed under NASA grant NNX11AD55G.