Phase locking up to 2.5 THz in grain-boundary Josephson junctions at 77 K

1l/12, pp. 615419, 1995 Copyright 0 1996 Elsevier Science Ltd Printed in Great Britain. All rights reserved 0964-1807195 $9.50 + 0.00

Applied Superconductivity Vol. 3, No.

0964-1807(95)00085-2

Pergamon

TECHNICAL

NOTE

PHASE LOCKING UP TO 2.5 THz IN GRAIN-BOUNDARY JOSEPHSON JUNCTIONS AT 77 K S. A. ZHGOON,’ G. D. LOBOY,’ YU. V KISLINSKII,z A. R. KUZHAKHMETOV’ and E. A. STEPANTSOV’ ‘M[oscow Power Engineering Institute, 14 Krasnokazarmennaja, Moscow 111250, Russia; ‘Institute of Crystallography, 59 Leninskii Prospect, Moscow 117333, Russia (Received 21 March 1995; in revised form 8 July 1995)

Abstract-Phase locking of two grain-boundary Josephson junction oscillation is directly observed in the frequency range 40 GHz-2.5 THz at 77 K. The junctions are prepared from laser-ablated YBCO films on a bicrystal YSZ substrate. They are connected by a microstrip line for a.c. with an impedance of about 0.3 Ohm. The low impedance is achieved by evaporation of 0.5 pm BaFz followed by 0.3 pm of silver-ground plane. Further cooling up to 70 K results in about 10% upper frequency limit increase.

INTRODUCTION

Oscillation in high critical temperature superconductor Josephson junctions has recently been reported by several research groups [l-4]. Edstam and Olsson obtained phase locking of two Josephson junctions connected by a lumped element circuit [5]. They observed proof of oscillation in the frequency range 0.2-I THz at 4-60 K, and they assumed that the frequency limit of the effect may lie as far as 10 THz with the proper circuitry connecting the junctions. It is certainly of great practical concern to determine the phase-locking frequency limits at 77 K. From this point of view we describe our direct observation of phase locking in a broader frequency range at a higher temperature. TEST

SAMPLE

DESIGN

AND

PREPARATION

Our test sample was initially designed as an YBCO copy of a well-known multijunction oscillator [6]. The major difference was in central frequency. Twenty-five 20-mm wide grain boundary junctions were formed from 0.3~pm thick laser-ablated YBCO film on a OS-mm thick, 10 x 10 mm YSZ bicrystal substrate. These junctions were connected in series for a.c. by a lOOmm wide line and in parallel for d.c. forming an oscillator section as in Ref. 6. The half period of the meander line was equal to 2 mm, and that was expected to correspond to 40-80 GHz synchronous central frequency oscillation depending on the substrate used. The YBCO film was everywhere covered by 0.5 pm silver except the junctions. Two separate junctions were formed in the same meander line at both ends of the oscillator section to serve as spectral detectors of possible oscillation. Only two junctions were connected in the a.c. circuit in this experiment by a continuous microstrip line: lone of the detector junctions and a corresponding closest extreme junction of the oscillator section; other junctions of this section serving only as d.c. stabilizing elements (Fig. 1). In the aim to try new technological approaches and to increase the coupling between neighbouring junctions (with RN = 0.2 Ohm and 1, = 200 pA at 77 K), a 0.5~pm thick BaFz layer was evaporated on the whole meander structure through a shadow mask followed by a 0.3~l.trn thick Ag layer. This resulted in a dramatic decrease of the microstrip line impedance, approaching 0.3 Ohm as estimated with the help of HP AppCAD program. During these subsequent processing steps no significant degradation of junctions T, and 1, was seen, nor during several following cooling cycles. 615

TECHNICAL NOTE

616

Common contact

Gap in silver ground pia?

Bicrystal boundary

Detector

h Oscillator

B btas

bias

Fig. 1. Test sample geometry and connection to the measurement circuit. Note the disruption in the ground plane, disconnecting a part of the oscillator section for a.c.

EXPERIMENTAL

RESULTS

After cooling to 77 K the oscillator and the detector sections were separately biased and the differential resistance of the detector section was registered with different levels of the oscillator section biasing. Deeps, corresponding to the first Shapiro step, are clearly seen on these curves, moving in accordance with the oscillator section bias voltage. As we used two probe measurement circuits and our detector and oscillator sections were d.c. connected by a superconducting path with one common contact pad, there is a shift of the dependence from the origin and a broadening of the slopes. This results in a significant shift of measured voltages corresponding to phase locking especially at high current levels, when we observe the nonlinearity of the contacts (the current in the oscillator section reaches 0.5 A). These features may be taken into account in our model of this sample and extracted to give true voltages on the junctions. We suppose that our detector junction can be modeled as a series connection of a Josephson junction with the contact resistance. Due to formation of YBCO-Ag interlayers the latter may be essentially nonlinear, as follows from our previous experience, while our Josephson junction behavior is assumed to be close to the RSJ model with RN = const. At low bias currents near Z,, where the contact resistance is small, our differential resistance value, R( V= 0), should not deviate far from RN, that may be extracted from the detector Z-V curve as well. Knowing the differential resistance value at each measured bias voltage R(V) we can with some degree of confidence apply a simple relation to extract the true voltage on the junction Vdetector: Vdetector

=

VR(V

= 0)/R(V).

If our assumption about the behavior of RN( V) is wrong, and it grows with applied voltage (that is not very probable) this will result in an even higher corresponding upper frequency limit, that we state here. The processed data in Fig. 2 demonstrate that the deep position on the differential resistance dependence corresponds sufficiently well to the oscillator voltage when the influence of all contact resistances is excluded. The example of unprocessed data at high current levels is shown in Fig. 3.

TECHNICAL NOTE

617

80

60

0

100

300

200

400

V detector FV Fig. 2. Example of the detector differential resistance (dR/dQ dependencies on the calculated detector voltage (V detector) with different levels of calculated oscillator section biasing (Vc) at low current levels. The curves are shifted upwards from the origin for clarity, data are processed to exclude contact resistances.

As the differential resistance R(v) for large bias voltages reaches almost 6-lOR( V= 0), the calculation precision with the above algorithm could be quite low. But as the oscillator section contacts in this :sample have quite a large area, the deviation of the oscillator 1-V curve from linear behavior is smaller (about 1.5 times); this gives us confidence in the overall voltage comparison results presente’d in the inset in Fig. 3. The R(V) dependence for the largest observed calculated phase-locking voltage (5.2 mV) is not present, as we did not dare to repeat measurements near a possible breakdown region. All this allows us to suppose that as in Ref. 5 we deal with phase locking of Josephson oscillations in the junctions. The calculated junction lowest voltage, when we clearly observe the drop, is about 40-80 uV, which corresponds to about 20-40 GHz. The calculated junction highest voltage, where the deep abruptly disappears in this test sample is about 5.2 mV, which corresponds to 2.5 THz. Lowering the temperature to about 70 K by pumping liquid nitrogen resulted in about a 10% increase of upper frequency limit. This fact strongly supports the hypothesis that we are reaching the limit, implied by the gap structure of the material in this sample, rather than by the absorption in our microstrip lines. Unlike in Ref. 5 we do not clearly observe the presence of second and higher harmonics. Neither do we notice any strong dependence of the phase-locking strength on the oscillation frequency. We can suppose that, unlike in Ref. 7, the loss level is high in our very low impedance and relatively long microstrip line, so that its behavior is nearly aperiodic.

ADDITIONAL

EXPERIMENTAL

FEATURE

The two-probe measurement circuit with one common contact pad for the oscillator and the detector sections allowed us to observe one more interesting feature of the phase locking behavior. The synchronous oscillation region (in terms of the deep width) in Fig. 2 looks unbelievably broad. It is much broader, that could be estimated from calculations following [8], including junctions loading. Measured detector junction voltages include the variations of measured oscillator voltage and vice versa. We estimate that the resistance of our common contact pad acts in this circuit as a self biasing network. We suppose, that with proper understanding of the

TECHNICAL NOTE

0

6

4

2

Measured detector voltage, mV

V

oadllator pV

Measured detector voltage, mV Fig. 3 .Measured (unprocessed) differential resistance dependencies on the measured detector voltage at (a) intermediate and (b) high current levels. The inset shows the comparison of the calculated voltages corresponding to the beginning and the end of the differential resistance deep vs the calculated oscillator section voltage. The straight line corresponding to V, = V,,,, is a guide for the eye.

operation mechanism such biasing networks could be intentionally applied to broaden the phase locking region in multijunction systems with high level of parameter scatter. CONCLUSION Using a very low impedance metal microstrip line with BaFa dielectric to couple low impedance Josephson junctions has allowed us to demonstrate that YBCO Josephson junctions can phase lock up to at least 2.5 THz at 77 K.

TECHNICAL NOTE

619

The upper frequency limit in our test sample is probably set by intrinsic reasons, such as the superconducting gap behavior in our test sample that follows from its increase with temperature decreasing, but extrinsic reasons, such as wave attenuation in our BaFz dielectric and in YBCOAg on YSZ coating cannot be neglected either. Acknowledgements-This work ir supported in part by the Russian state HTSC program. The author8 are grateful to A. A. Ivanov for YBCO thin film llaser ablation and to all colleagues for help and encouragement.

REFERENCES 1. J. S. Martens et aL, IEEE lk~ns. Appl. Supenzond. 3, 3095 (1993). 2. J. S. Martens, A. Rance, K. Char, L. Lee, S. Withley and M. Hietelo, Appl. Phys. L&t. 63, 168 1 (1993). 3. J. Edstam, I! A. Nilsson, E. A. Stepant8ov and H. K. O&on, Appf. Phys. Leti. 62, 896 (1993). 4. R. Kleiner and I? Muller, Phy8. Rev. B 49, 1327 (1994). 5. J. F&am and H. K. Olsson, Appl. Phys. Lett. 64, 2587 (1994). 6. K. L. Wan, A. K. Jain and J. E. LllkellS, Appl. Phys. Lerr. 54, 1805 (1989). 7. J. F&am and H. K. Olsson, Appl. Phys. L&t. 64, 2733 (1994). 8. K. K. Liharev, Dynamics ofJosephson Junctions and Cimits. Gordon and Breach, New York (1986).