ow-Frequency Flux Noise in dc SQUIDS*

tions of the device; in particular, variations in the critical current of small junctions can lead to significant lowfrequency noise [2]. In high-quality junctions, however, these fluctuations may be negligible. For the Stanford experiment, we require well-coupled ck SQUIDs with an energy sensitivity of 1 x lo-*’ J/Hz ai 5 mHz. Early in the development of these devices, we observed both excess low-frequency noise and resonances due to the input coil. (Our investigation of resonance phenomenon will be discussed elsewhere [3] and we focus on the excess low-frequency noise here.) To study the excess noise, we have fabricated devices with two geometries: a well-coupled SQUID with a washer geometry and a self-shielded SQUID with a stripline geometry [4, 51. Results of these experiments indicate that we are observing flux noise originating from within the SQUID. Process improvements have dramatically reduced this noise. We have also performed experiments to reduce the potential for flux motion in the SQUID; this work is still in progress.