Antenna Coupled Planar Integrated Array

Bolometer systems would be simplified considerably if all components could be made photolithographically on a single substrate. We are developing a design for a single-substrate array with several frequency bands that incorporates a planar antenna for coupling to the telescope, transmission-line filters for separating and defining frequency bands, and lithographed bolometers. Antenna Coupling: Bolometer arrays are usually coupled to telescopes either by horns or by placing the absorbers directly in the focal plane. Another possible method is to use planar antennas. Planar antennas control the solid angle that is visible to the bolometer as with a horn, but are much easier to fabricate in large arrays. Our current design uses a double-slot dipole [filipovic97]. This antenna has many good features for use in bolometer arrays. It has a symmetric beam pattern and an impedance suitable for matching to microstrip transmission line. A dielectric lens in direct contact with the antenna is typically used, and the combination antenna/lens can be matched to the f/# of a telescope by adjusting the parameters of the lens. A dual polarization version of the double-slot dipole has been developed by Chattopadhyay and Zmuidzinas [chattopadhyay98]. Transmission-line Filters and RF Multiplexers: A planar antenna couples radiation from the telescope to a transmission line, for example, microstrip. Transmission-line elements can be used to separate the RF energy into multiple well defined frequency bands. There is a considerable literature on such devices from the electrical engineering community. We are investigating superconducting Nb microstrip resonant stub filters. The stubs are ¼ lambda in length and can be modeled as ``LC'' resonators. The in-band transmission of such filters can approach unity due to the low microwave loss of the superconductor. Lithographed VSBs: We are developing two types of VSB, a suspended device where the thermal isolation is achieved by using long thin legs of silicon nitride and a ``hot-electron'' device where the isolation occurs due to the weak electron-phonon interaction in a small-volume metal sensor. Both types can be coupled to a planar transmission line such as microstrip. The suspended device can be coupled via transmission lines on the legs of the bolometer terminated in a lithographed resistor adjacent to the sensor. The hot-electron device can act as a resistive load at frequencies above those corresponding to twice the gap energy of the superconductor. An appropriate impedance transformer can be used to match to the typically low normal resistance of the sensor. We have built and tested suitable suspended devices with thermal-fluctuation limited white noise and 1/f knees of several hundred mHz [gildemeister00b]. The hot-electron VSBs would be more simple to fabricate, and we are currently exploring devices using an Al/Ti proximity-effect sandwich superconductor. SQUID readout: Readout SQUIDs or multiplexed readout SQUIDs could eventually be integrated onto the same substrate as the antennas, RF multiplexer, and bolometers. The low RF loss of niobium microstrip, roughly 1% per wavelength of line, allows the RF energy to be taken outside the field-of-view of the telescope where there is room for the RF multiplexer, readout SQUIDs, and readout multiplexers.

Absorber Coupled Filled VSB Array

We have developed a prototype absorber-coupled VSB, shown below, that is suitable for building a nearly filled array on a single substrate. This type of array is placed at the focal plane of a telescope and directly absorbs the incident radiation as with the GSFC pop-up [benford98] or the CEA [agnese99] designs, but our design may be considerably easier to fabricate. The absorber consists of a square micromesh of silicon nitride covered with metal of the appropriate resistivity. There is a TES at the center of the absorber made from a proximity effect sandwich of Al and Ti. The Tc of the sandwich is 304 mK. The silicon nitride supports give strong thermal isolation while occupying very little area. We have measured an NEP of 2.3 x 10^(-17) and a time-constant of 24 ms at a temperature of 304 mK. We have also built a 32 x 32 array of micromeshes, as shown in the second figure below, with the same design as the device in the first figure. All of the micromeshes were intact although several had defects due to dust contamination. Full details are given in reference [gildemeister00a].

Prototype single array element designed for filled focal-plane array. The absorber is a micromesh 1.5 mm on a side built from 1um thick silicon nitride. The TES at the center is made from an Ti/Al/Ti proximity-effect sandwich. The absorber is also made from this trilayer. The leads are made from the trilayer with an additional layer of Al on top. The bolometer is attached at 4 points (see arrows) to tensioned silicon-nitride members running up to down in the photograph. This device has a G = 2.7 x 10^(-11) W/K and an NEP of 2.3 x 10^(-17) W/rtHz at 304 mK. 32 x 32 array of micromeshes with the same geometry as that of the bolometer above. Each square micromesh is 1.5 mm on a side, and the entire array is 4.8 cm on a side. All 1024 suspended micromeshes are intact although several have defects due to dust contamination.

Single SQUID Readout Multiplexer

We have developed a SQUID readout multiplexer that is suitable for use with VSBs. It reduces the number of wires by a ratio similar to the NIST design [chervenak99], but only requires a single SQUID per row of detectors. Each of the VSBs is ac biased with a unique frequency. The output currents of the bolometers are added in a superconducting summing loop as shown below. A single SQUID measures the current in the summing loop, and all the individual signals are lock-in detected after the room temperature SQUID electronics. The current in the summing loop is nulled by feedback to eliminate crosstalk. We have built and tested an eight-channel prototype fabricated with superconducting niobium. Resistors were used as mock bolometers and the sin-wave biases were amplitude modulated to simulate signals. Test data are shown below. The simulated signals were recovered with < 1% crosstalk. The output noise of the system was consistent with the Johnson noise of the resistors. We are currently exploring the limit on the number of multiplexed bolometers. Full details are given in reference [yoon00].

Schematic diagram of the single SQUID multiplexer. The bolometers are ac biased each with a separate frequency. The signals are added together in the summing loop, and separated at room temperature by lock-in demodulation. Each bolometer is coupled to the summing loop by a transformer. The current in the loop is nulled to prevent crosstalk. Photograph of a multiplexer chip. Each large square is a spiral wound niobium transformer. There are nine transformers, eight for the bolometers and one for the feedback circuit. The outer size of the transformers is 1 mm. Connections for the SQUID input coil are visible at the top. Test multiplexer data using resistors as mock bolometers. Eight channels are multiplexed, and for seven of these the biases are sin-wave amplitude modulated to simulate a signal. The two traces show demodulated data obtained by locking to the same bias frequency as a channel that is amplitude modulated at 8.6 Hz at the top and to the single unmodulated bias frequency at the bottom. Crosstalk is less than 1%.

Model for High Frequency Noise

Suspended VSBs sometimes exhibit high-frequency noise in excess of that expected from simple thermal fluctuations as shown in the figure below. We have developed a model that accurately predicts this excess noise in our devices. In this model, the excess noise arises from a combination of the distributed nature of the bolometer's thermal circuit, i.e. different parts of the bolometer can have different temperatures, and the action of electrothermal feedback. The model shows that this noise can be moved to frequencies higher than the usable optical bandwidth by choosing bolometer materials with reduced heat capacity. A significant reduction in heat capacity compared to our current bolometers should be possible, for example, by using niobium for the electrical leads since the niobium would be far below its critical temperature. The model was tested by fabricating and testing a set of three bolometers with varying heat capacities. It was found that the model predicted the noise of the three bolometers well as shown below for one of the bolometers. A full description of this work can be found in reference [gildemeister00b].

Measured noise and model prediction for suspended VSB. The measured noise is close to the expected white thermal noise for frequencies between 200 mHz and 20 Hz. There is a high-frequency peak above 20 Hz that arises from a combination of the distributed nature of the bolometer's thermal circuit and the action of electrothermal feedback. The model predicts that the noise can be moved above the useful bandwidth of the bolometer by reducing the heat capacity of the bolometer materials.

References

[debernardis00] DeBernardis, P., et al., 2000, Nature, 404, 995 [hanany00] Hanany, S., et al., 2000, submitted to ApJ, astro-ph/0005123 [gildemeister00a] Gildemeister, J., Lee, A.T., Richards, P.L, submitted to Applied Physics Letters [yoon00] Yoon, J., Clarke, J., Gildemeister, J.M., Lee, A.T., Myers, M., Richards, P.L., Skidmore, J., submitted to Applied Physics Letters [gildemeister00b] Gildemeister, J., Lee, A.T., Richards, P.L, submitted to Applied Optics [filipovic97] Filipovic, et al., IEEE Transactions on Antennas and Propa gation, V 45, No 5, 1997 [chattopadhyay98] Chattopadhyay, G., Zmuidzinas, J., IEEE Transactions on Antenn as and Propagation, V 46, No 5, 1998 [benford98] Benford, D.J., Serabyn, E., Phillips, T.G., Moseley, S.H., Proceedings of the Advanced Technology MMW, Radio, and Terahertz Telescopes, 1998 Kona, Hawaii, SPIE Vol. 3357. [agnese99] Agnese, P., Buzzi, C., Rey, P., Rodriguesz, L., Tissot, J.-L., Proceeding of the Infrared Technology and Applications XXV, 1999 Orlando, Florida, SPIE Vol. 3698 [chervenak99] Chervenak, J.A., Irwin, K.D., Grossman, E.N., Martinis, J.M., Reintsema, C, Huber, M.E., Bull. of the Am. Phys. Soc. 43, 231 (1999)