EXPERIMENTAL PROOF THAT NON-JUNCTION DEFECTS CAN SIGNIFICANTLY AFFECT THE PERFORMANCE OF A SINGLE-ELECTRON-TUNNELING TRANSISTOR. Jonathan L. Cobb, Neil M. Zimmerman, Alan F. Clark, Building 220, Room B258, National Institute of Standards and Technology, Gaithersburg, MD, USA. Address NIST bldg 220, Room B258. (301- 975-4270, email: cobb@eeel.nist.gov)

Devices fabricated out of ultrasmall tunnel junctions provide the ability both to manipulate individual electrons and to measure changes in electrical charge with sub-e sensitivity through the phenomenon of single-electron tunneling (SET). The foremost technical challenge precluding the integration of these junctions into useful metrological and digital applications is the unwanted modulation of the junction charge by surrounding defects. Presently, there is a debate about whether or not defects located outside of the junctions themselves can influence the junction charge significantly enough to produce measurable results. We have answered this question by studying the switching noise of a SET transistor. The switching noise we observe consists of multiple levels, has total switching amplitude equal to the total amplitude of the device transconductance, has a 50% duty cycle at all voltages, and has rates which increase near-exponentially with increasing gate voltage. We explain these noise characteristics in terms of a correlated cluster of two-level fluctuators which switch coherently to produce the large total switching amplitude we observe. The fact that the voltage across a tunnel junction in a SET transistor is periodic in gate voltage coupled with the fact that the switching rates do not display any such periodicity demonstrates unambiguously that the defects responsible for the switching noise are not located in the junctions. We also directly observe the wash-out of the transistor performance by the switching noise.