The design and testing of integrated circuits for submillimeter-wave spectroscopy

Abstract

Optoelectronic techniques have extended the bandwidth of electronic spectroscopic systems to the submillimeter wavelengths. In a significant class of these systems the submillimeter-wave source, detector and device of interest are monolithically integrated. Such systems are attractive because of their reliability and small size and cost, because an integrated circuit is the highest-bandwidth environment for testing microelectronic devices, and because of their potential application to on-chip chemical and biological sensing. This thesis focuses on three separate topics in the field of submillimeter-wave spectroscopy with integrated circuits. The first topic is the decrease in bandwidth of photoconductive submillimeter wave emitters with increasing voltage bias, which limits the output power of these devices at frequencies near 1 THz. We performed measurements of a photoconductor made of low-temperature grown GaAs embedded in a coplanar waveguide with both static and dynamic illumination. We investigated the bandwidth decrease and an increase in de photocurrent that occurs at the same bias voltages. We attribute both phenomena to a reduction of the electron capture cross section of donor states due to electron heating and Coulomb-barrier lowering. The second topic is a novel circuit for ultrafast measurements with coplanar waveguide transmission lines. The circuit contains photoconductive switches that allow tunable generation and reception of a coplanar waveguide's two propagating modes. The circuit has fewer discontinuities than other circuits with similar capabilities and does not require air bridges. We show how the photoconductive switch can be biased to compensate for pump laser beam misalignment. The third topic is the first demonstration of an integrated circuit's use for submillimeter- wave frequency-domain spectroscopy. Such an application is attractive because of its inherently good frequency resolution, which is necessary for chemical and biological detection. The amplitude and phase of the measured spectrum of a circuit without a device under test agree with a model that takes into account circuit resonance, photoconductive-switch dynamics, and resistive loss. We discuss why photoconductive frequency-domain spectroscopy has an inherently lower output signal than similar time-domain spectroscopy, and how this drawback can be compensated for.