Diffusion-Cooled Tantalum Hot-Electron Bolometer Mixers

NASA Tech Briefs, Oct 2004

Lower T^sub C^s should translate to lower noise and lower required local-oscillator power.

A batch of experimental diffusion-cooled hot-electron bolometers (HEBs), suitable for use as mixers having input frequencies in the terahertz range and output frequencies up to about a gigahertz, exploit the superconducting/normal-conducting transition in a thin strip of tantalum. The design and operation of these HEB mixers are based on mostly the same principles as those of a prior HEB mixer that exploited the superconducting/normal-conducting transition in a thin strip of niobium and that was described in "Diffusion-Cooled hot-Electron Bolometer Mixer" (NPO-19719), NASA Tech Briefs, Vol. 21, No. 1 (January 1997), page 12a.

One reason for now choosing tantalum instead of niobium arises from the fact that the superconducting-transition temperature (T^sub C^) of tantalum lies between 2 and 3 K, while that of niobium lies between 6 and 7 K. Theoretically, the input mixer noise of a superconducting HEB is proportional to T^sub C^ and the power demand on the local oscillator that supplies one of the input signals to the mixer is proportional to T^sub C^^sup 2^. The lower noise and power demand associated with the lower T^sub C^ of tantalum could make tantalum HEBs more attractive, relative to niobium HEBs, in applications in which there are requirements to minimize noise and/or to provide mixers that can function well using the weak signals generated by typical solid-state local oscillators. Of course, to reach the required lower T^sub C^, it is necessary to use more complex cryogenic equipment. Fortunately, such equipment (e.g., helium-3 cryostats) is commercially available.

In order to make a practical tantalum HEB, it is necessary to overcome a challenge posed by the fact that thin films of tantalum tend to contain grains of two different crystalline phases. The presence of the two phases would be unacceptable in a practical device because (1) the additional electron scattering at the grain boundaries would tend to suppress the diffusion-cooling mechanism, and (2) the different T^sub C^s of the two phases would lead to broadening of the transition.

The present HEB mixers contain tantalum microbridges having lengths of 100 to 400 nm, widths of 100 to 200 nm, and thicknesses of 10 nm. The bridges were made from a 10-nm-thick film of tantalum deposited by sputtering onto a 1.5nm-thick seed layer of niobium on a silicon wafer. The niobium seed layer was used to promote the growth of one of the two crystalline phases (the α phase) to ensure the required crystalline purity and thereby keep the superconducting transition (see figure) as sharp as possible.

The results of microwave impedance tests of one of the experimental tantalum HEBs have been interpreted as signifying that the 3-dB roll-off frequency for mixer conversion efficiency can be expected to be about 1 GHz, neglecting the effect of electrothermal feedback. End effects are small enough (as illustrated by the smallness of the "foot" of the resistance-versus-temperature curve in the figure) that it should be possible to use devices as short as 100 nm and possibly even shorter. Inasmuch as the thermal-relaxation time for diffusion cooling is proportional to the square of the device length, the 100-nm-long devices should be capable of "raw" speeds of about 16 GHz. Since only a few gigahertz of bandwidth is needed for most mixer applications, it is expected that tantalum HEBs will be fast enough even in the presence of slowing effects of electrothermal feedback, which is present in most bolometer mixer circuits.

This work was done by Anders Skalare, William McGrafh, Bruce Bumble, and Henry LeDuc of Caltech for NASA's Jet Propulsion Laboratory. For further information, access the Technical Support Package (TSP) free on-line at www.techbriefs.com/tsp under the Physical Sciences category.

NPO-30695

NASA's Jet Propulsion Laboratory, Pasadena, California