Low temperature detectors for the near-IR, optical, and UV provide greatly improved performance over conventional semiconductor detectors, but it has proven difficult to make large arrays of these detectors.  To address this problem, our group at UCSB, Caltech, and JPL has developed the MKID, which is particularly well suited to large arrays because of a simple and powerful readout scheme.

    The operational principle of the MKID is shown in Figure 1.  Photons with energy hν are absorbed in a superconducting film, breaking Cooper Pairs and producing a number of unpaired electron excitations, called quasiparticles, equal to Nqp = ηhν/Δ, where Δ is the gap parameter of the superconductor, and η is an efficiency factor (about 0.6 for our devices).  The principle is similar to electron-hole generation by high-energy photons in a semiconducting detector, with the difference that Δ is only tenths of millielectron volts for a superconductor as opposed to electron volts for a semiconductor.  This very low gap energy means that tens of thousands of quasiparticle excitations are created for a single photon absorbed.  The fundamental energy resolution of the detector, known as the Fano limit, is limited by statistical fluctuations in this number of quasiparticles created.  The quasiparticles produced by an absorption event are sensed through their effect on the kinetic inductance and microwave surface resistance of the superconducting film.  To sensitively measure these quantities, the film is placed in a high frequency planar resonant circuit (Figure 1(b)).  The figure shows a lumped element resonant circuit that forms a single MKID pixel.  This is circuit layout can be realized in a devices like the one shown in Figure 2, which is about 150 μm on a side and made of a single superconducting TiN film.

    The amplitude and phase of a microwave excitation signal sent through the resonator are shown in Figures 1(c) and 1(d).  The increase in the kinetic inductance and surface resistance of the film following a photon absorption event pushes the resonance to lower frequency and changes its amplitude.  If the detector (resonator) is excited with a constant on-resonance microwave signal, the energy of the absorbed photon can be determined by measuring the degree of phase and amplitude shift.  The phase and amplitude change quickly (several μs) during the initial absorption event, followed by a slow (~100 μs) return to the unexcited state as the quasiparticles recombine into Cooper Pairs.  This produces the characteristic pulse shape shown in Figure 3 for each arriving photon.

    As can be seen from Figure 1 (c), the transmission through the resonator is very high for signals detuned from the resonance.  Since the quality factor (the resonant frequency divided by the line width) of these resonators is very high, in the range of 10,000-3,000,000, the degree of detuning is very small before the transmission is perfect.  This leads to a simple readout multiplexing scheme.  Many resonators with slightly different resonance frequencies may be coupled to the same feedline, as shown on the right top panel of Figure 1. The array is excited with a microwave comb function containing a sine wave at the resonant frequency of each of the resonators in the array. The resonators pick out one particular excitation signal while passing all of the rest unaltered.  The signals for the entire array can be amplified with a single cryogenic amplifier without saturation.  We have fabricated arrays with greater than 250 resonators and have demonstrated multiplexed readout of a subset of the array with a mean resonator to resonator frequency jitter of 0.8 MHz using CPW MKIDs, as shown in the bottom right panel of Figure 1.  It is possible to read out an array of thousands of detectors with this technique using only a single transmission line.

    Our previous optical/UV MKIDs used a strip detector geometry (also known as a distributed readout), where a rectangular strip of a high superconducting gap absorber (usually tantalum) is terminated on each end by a low gap (usually aluminum) MKID.  This allows 1-d multiplexing since a photon hitting the absorber can be localized along the length of the strip by observing the ratio of the detected energy in the MKIDs at either end of the absorber strip.  This approach allows ~10 pixels for each resonator, but has two significant disadvantages.  First, because of quasiparticle trapping the ultimate energy resolution is limited by the larger absorber gap instead of the lower MKID gap.  For an operating temperature of 100 mK this is the difference between a maximum theoretical energy resolution R=E/∆E at 400 nm of 50 for a strip detector and 150 for a single pixel detector.  Second, the maximum count rate per pixel is reduced by strip multiplexing, which limits how far into the near infrared the detector can be used.  For instance, a MKID at the Palomar 200” telescope with strip detectors would be limited by sky counts to a bandwidth of 350-750 nm, while a single pixel device could cover 350-1350 nm.

    Due to significant developments and reduction in cost of the digital readout of MKIDs, the recent discovery that it is possible to make high quality resonators out of TiN, and significant technical problems trapping quasiparticles in thin aluminum films, it appears than strip detectors are no longer the favored technical path.

    An OLE MKID pixel is a single microwave resonator consisting of a lumped element inductive meander and an interdigitated capacitor, as shown in Figure 1(b) and 2.  The virtue of the single pixel per resonator OLE MKID design is that the photon absorber is also the inductor, so a separate absorber and quasiparticle trapping are not needed to get efficient absorption of photons.  This vastly simplifies the design and fabrication process.

    The inductive meander of an OLE MKID is patterned with nearly the smallest feature size practical in order to increase the active area of the sensor.  Our prototype OLE MKID uses 0.5 μm wide slots and 2 μm wires.  This arrangement reduces detector quantum efficiency by 25% at wavelengths shorter than 0.5 μm, but provides the full quantum efficiency of the TiN film for wavelengths longer than 0.5 μm where the inductor looks like a uniform superconducting sheet to the incoming photons.  Slot widths of 0.25 μm should be straightforward with modern lithographic techniques, removing this 25% loss except for UV wavelengths.

The capacitor is a vital component of the OLE MKID.  Since the voltage peaks in the capacitor, two-level systems (TLSs) located on the surfaces or bulk dielectric near the capacitor with frequencies close to the MKID’s resonant frequency can interact, causing an excess phase noise in the resonator.  TLS noise will likely be the dominant noise source in OLE MKID, and careful efforts to reduce its contribution will be a substantial part of the research in the lab.

Figure 2.  An optical lumped element (OLE) MKID array made of a 40 nm thick TiN film.

Figure 1.  (Left) The basic operation of an MKID. (a) Photons with energy hν are absorbed in a superconducting film, producing a number of excitations, called quasiparticles.  (b) To sensitively measure these quasiparticles, the film is placed in a high frequency planar resonant circuit.  The amplitude (c) and phase (d) of a microwave excitation signal sent through the resonator.  The increase in the kinetic inductance and surface resistance of the film following a photon absorption event pushes the resonance to lower frequency and changes its amplitude.  If the detector (resonator) is excited with a constant on-resonance microwave signal, the energy of the absorbed photon can be determined by measuring the degree of phase and amplitude shift.  (Right) An example of frequency domain multiplexed (FDM) MKIDs.

Figure 3.  The measured response of an OLE MKID to illumination by a single 254 nm photon.  The inset shows a histogram of ~5000 photon events.  A Gaussian fit to this histogram reveals an energy resolution R=16.

Figure 4.  The measured absorption of a 40 nm thick TiN film on sapphire.  This is roughly the detector quantum efficiency we can expect from the OLE MKID.