Abstract: Kinetic Inductance detectors have held a promise for the last decade to enable very large arrays, in excess of 10.000 pixels, with background limited sensitivity for ground- and Space Based sub-mm observatories. First we present the development of the detector chips of the A-MKID instrument: These chips contain up to 5400 detector pixel divided over up to 5 readout lines for the 350 GHz and 850 GHz atmospheric windows. The individual detectors are lens antenna coupled KIDs made of NbTiN and Aluminium that reach photon noise limited sensitivity at sky loading levels in excess of a few fW per pixel using either phase readout or amplitude readout. The ability to use phase readout is crucial as it reduces the requirements on the readout electronics of the instrument. Cross coupling between the KID resonators was mitigated by a combination of numerical simulations and a suitable position encoding of the readout resonance frequencies of the individual pixels. Beam pattern measurements are performed to demonstrate the absence of any cross talk due to resonator- resonator cross coupling. Second we present experiments on individual lens-antenna coupled detectors at 1.5 THz that are made out of aluminium. With these devices we have observed, as a function of the irradiated power at 1.5 THz, the crossover from photon noise limited performance to detector-limited performance at loading powers less than 0.1 fW. In the latter limit the device is limited by intrinsic fluctuations in the Cooper pair and quasiparticle number, i.e. Generation-Recombination noise. This results in a sensitivity corresponding to a NEP = 3.8·10 -19 W/√(Hz).

Abstract: NbN hot electron bolometer (HEB) mixers are at this moment the best heterodyne detectors for frequencies above 1 THz. However, the fabrication procedure of these devices is such that the quality of the interface between the NbN superconducting film and the contact structure is not under good control. This results in a contact resistance between the NbN bolometer and the contact pad. We compare identical bolometers, with different NbN – contact pad interfaces, coupled with a spiral antenna. We find that cleaning the NbN interface and adding a thin additional superconductor prior to the gold contact deposition improves the noise temperature and the bandwidth of the HEB mixers with more than a factor of 2. We obtain a DSB noise temperature of 950 K at 2.5 THz and a Gain bandwidth of 5-6 GHz. For use in real receiver systems we design small volume (0.15x1 micron) HEB mixers with a twin slot antenna. We find that these mixers combine good sensitivity (900 K at 1.6 THz) with low LO power requirement, which is 160 – 240 nW at the Si lens of the mixer. This value is larger than expected from the isothermal technique and the known losses in the lens by a factor of 3-3.5.

Abstract: NbN phonon cooled HEB’s are one of the most promising bolometer mixer technologies for (near) future (space) applications. Their performance is usually quantified by mea- suring the receiver noise temperature at a given IF frequency, usually around 1 – 2 GHz. However, for any real applications it is vital that one fully knows all the relevant properties of the mixer, including LO power, stability, direct detection, gain bandwidth and noise bandwidth, not only the noise temperature at low IF frequencies. To this aim we have measured all these parameters at the optimal operating point of one single, small volume quasioptical NbN HEB mixer. We find a minimum noise temperature of 900 K at 1.46 THz. We observe a direct detection effect indicated by a change in bias current when changing from a 300 K hot load to a 77 K cold load. Due to this effect we overestimate the noise temperature by about 22% using a 300 K hot load and a 77 K cold load. The LO power needed to reach the optimal operating point is 80 nW at the receiver lens front, 59 nW inside the NbN bridge. However, using the isothermal technique we find a power absorbed in the NbN bridge of 25 nW, a difference of about a factor 2. We obtain a gain bandwidth of 2.3 GHz and a noise bandwidth of 4 GHz. The system Allan time is about 1 sec. in a 50 MHz spectral bandwidth and a deviation from white noise integration (governed by the radiometer equation) occurs at 0.2 sec., which implies a maximum integration time of a few seconds in a 1 MHz bandwidth spectrometer.