1666 IEEE TRANSACTIONS ON MAGNETICS, VOL. MAG-23, NO. 2, MARCH 1987 SUBPICOSECOND OPTOELECTRONIC STUDY OF SUPERCONDUCTING TRANSMISSION LINES* C. C. Chi, W. J. Gallagher, I. N. Duling 111, D. Grischkowsky, N. J. Halas**, M. B. Ketchen, and A. W. Kleinsasser IBM Thomas J. Watson Research Center P. 0. Box 218 Yorktown Heights, New York 10598 Abstract We have studied the propagation of subpicosecond elec- trical pulses on coplanar superconducting Nb transmission lines. Pulses with 0.6 ps full width at half maximum were generated by photoconductively shorting a 10 pm region between two charged 1 to 5 pm lines separated by a 2 to 10 pm gap. The propagating pulses were sampled by the delayed shorting of a fast phototconductive switch between a sampl- ing probe and one of the transmission lines at variable dis- tances away from the generation point. Silicon-on-sapphire wafers served as the transmission line substrate, with the 0.5 pm thick Si layer heavily damaged by an oxygen implant t o provide the subpicosecond carrier life time for the excitation andprobeswitches.Measurementsandanalyses of pulses propagated up to 8 mm distance at temperatures from 2 K to 10 K showed a threshold for strong attenuation and dispersion at a frequency reflecting the onset of pair breaking in the superconducting transmission lines. The results at least qualitatively confirm the superconducting microstrip trans- mission line calculations of Kautz based on Mattis and Bardeen's formulae for the complex conductivity of super- conductors. I. Introduction It is well known that superconducting transmission lines are superior to their normal metal counterparts because of their nearly lossless and dispersion-free pulse propagation for pulses with frequency components less than the supercon- ducting gap frequency.' Theoretical calculations of pulse propagation with frequency components exceeding gap fre- quency have been carried out by Kautz2 using the Mattis- Bardeen formulae3 for the complex conductivity of superconductors. Kautz predicted that the strong dispersion and attenuation due to pair breaking would cause progressive pulse degradation such as the reduction of pulse height and the occurence of trailing edge oscillations. For typical superconducting transmission lines made of Nb or Pb alloys, subpicosecond pulses and sampling capabilities are required to explore these pair breaking effects. Only recently, were such capabilities demonstrated using AI coplanar trans- mission lines with subpicosecond photoconductive switches for pulse generation and dete~tion.~ We have applied these techniques t o study pulse propagation on coplanar supercon- ducting Nb transmission lines.5 A different optical sampling method has been used to study the propagation of fast-rise- *Partially supported by the U. S. Office of Naval Research. **Bryn Mawr College. Manuscript received September 30, 1986. time steps on superconducting Pb-alloy transmission lines by Dykaar et a1.6 In this paper, we will briefly describe the ex- perimental technique, and present some experimental results and comparison to the calculation of Kautz. 11. Experimental Details The transmission line geometries we studied consisted of two or three equal width parallel Nb lines separated by gaps that were twice the metal line width. High quality Nb films were e-gun evaporated on commercial silicon-on-sapphire (SOS) wafers. Liftoff techniques were used to pattern transmission Iine structures in thin Nb films (70 to 80 nm) and plasma etching was used for thicker Nb films (150 and 300 nm). After the Nb deposition and patterning, the sam- ples were implanted with 0' at 100 and 200 keV. The im- plant dosage ( lOI5 cm-*) and energy were chosen to damage the Si epilayer heavily and uniformly so thatthephoto- excited carrier lifetimes were shortened into subpicosecond region. The implant also slightly degraded the T, of the Nb film from 9.4 K to 8.9 K in the worst case. For low temperature measurements, the transmission line samplesweremountedona cryostat insert with low fre- quencyleadsconnected t o the transmissionlinesand the sampling probes. An optical dewar positioned on the same optical bench with the laser source was used to cool the sam- ple with forced flow of cold He vapour. Temperature regu- lation was achievedby a feedback controlledresistive heater embedded in the copper sample mount. There are several ways to generate and sample electrical pulses on the transmission line^.^.^ Our prefered method4J is illustrated in Fig. 1. The transmission line is charged with a dc voltage bias (1 to 22 V). Focusing a beam of laser pulses on a spot between the two charged lines causes momentary localdischarges to generate electrical pulses, We call this "sliding contact" pulse generation since the generation point can be arbitrarily movedalong the transmission line. The pulse sampling was usually achieved by connecting one of the two lines to an electrical probe via a fast photoconductive switch, i. e. the sampling "gap", driven by a second time- delayed beam of the same laser pulse train. The laser source used for this experiment is a compensated, colliding pulse, passive mode-locked dye laser producing 80 fs pulses at a 100 MHz repetition rate with a typical average power of 10 mW. The details of electrical measurement setup are the same as described in Ref. 4. It is worth to note that this optoelectronic sampling technique is quite suitable for fast electrical measurements inside a dewar since only slow signals (0 to 2 kHz) have to come out. 0018-9464/87/0300-1666$01.0001987 IEEE