Burning of high Tc bridgesM. E. Gaevski,a) T. H. Johansen, Yu. Galperin,a) and H. BratsbergDepartment of Physics, University of Oslo, P. O. Box 1048 Blindern, N 0316 Oslo, Norway

A. V. Bobyl, D. V. Shantsev, and S. F. KarmanenkoState Technical University, 195251 St. Petersburg, Russia

~Received 4 February 1997; accepted for publication 24 September 1997!

Burning of superconducting thin film bridges by large transport currents ~up to densities of 23107 A/cm2) is investigated by magneto-optical imaging of flux distribution and low-temperaturescanning electron microscopy providing Tc maps. It is shown that the destruction is preceded bysignificant penetration of magnetic field inside a weak-pinning region. In bridges containingextended defects magneto-optic investigation is sufficient to locate the incipient burning region. Inhigh-quality bridges free from such defects only a combination of the two techniques will allowprediction of the place of fatal destruction. © 1997 American Institute of Physics.@S0003-6951~97!01847-0#

In order to optimize the performance of devices madefrom high Tc superconductors ~HTSs! it is important to havespace-resolved information about the material and its re-sponse properties. Today, several techniques have been de-veloped for this purpose. One of them is low-temperaturescanning electron microscopy ~LTSEM!,1 which monitorsthe transition into the normal state under the local heating byan electronic beam. This method allows one to determine thelocal values of the critical temperature Tc . Another powerfultechnique—magneto-optical ~MO! imaging ~see e.g., Ref. 2and references therein!—enables one to monitor the distribu-tion of magnetic field perpendicular to a plane adjacent to theHTS surface. In this work we combine these two nondestruc-tive techniques to investigate the performance of HTSbridges carrying large transport currents.

When an HTS device carries a transport current, IT ,approaching the critical value, Ic , one expects that localvariations in the material become increasingly important andeventually determines where the device is destroyed. Theusual source of destruction is thermally induced motion ofmagnetic flux which leads to further heat dissipation andlocal increase in the temperature. This, in turn, causes re-duced flux pinning locally, and a process having the charac-ter of an avalanche3 leads to strong overheating or burning.While HTS films carrying small transport currents have beenstudied previously by the MO method,4–7 there exists so farno such study focusing on the behavior when IT;Ic . In thiswork we show that MO images of self-fields from IT corre-lated with Tc maps obtained by the LTSEM technique8 canbe used to predict the location of burning in HTS films.

Superconducting YBa2Cu3O72d ~YBCO! films wereprepared by dc magnetron sputtering of a ceramic stoichio-metric target. The films were grown on ~100! LaAlO3 sub-strates at a rate of 2 nm/min at T5680 K in a gaseous mix-ture of Ar and oxygen. Subsequent heat treatment at T5500 K for 20 min in pure oxygen at a pressure of 1 atmand cooling for 1 h resulted in a high oxygen saturation ofthe YBCO lattice.9 The film thickness was 300 nm. X-ray

analysis confirmed good quality and a ~100! orientation ofthe films. According to scanning electron microscope ~SEM!

studies, the films contained no inclusions or outgrowths >1mm normal to the surface.

Micro-bridges 1003500 (mm!2 were formed by a stan-dard photolithographic procedure. To optimize conditions forMO studies we fabricated samples with all contact platesplaced on the same side of the structure at a sufficient dis-tance from the bridge. The contact pads were covered by Ag,and Au wires were attached by a thermal compression. Twotypical bridges were studied. One of them ~see Fig. 1! re-ferred to as ‘‘good,’’ had a critical current density ;106

A/cm2 at 77 K as determined by current-voltage (I-V) mea-surements. The other sample, referred to as ‘‘bad’’ had criti-cal current ;53104 A/cm2 at 77 K. As shown below, theburning process for the two samples showed distinctly dif-ferent features.

MO images of the ‘‘good’’ sample at 15 K are shown inFigs. 1~b!–1~f!. Here bright areas correspond to large abso-lute values of the perpendicular magnetic field. As an MOindicator we use Bi-doped YIG film with in-plane anisot-ropy, and an optical cryostat system described in Ref. 10. Inthe current-carrying states with IT53 A and 6 A, Figs. 1~b!

and 1~d!, respectively, the field is seen to concentrate nearthe edges. Note that the MO indicator does not discriminatebetween the ‘‘up’’ and ‘‘down’’ directions of the perpen-dicular field when using the present crossed polarizer settingon the microscope, and hence the images are essentiallysymmetric about the center line of the bridge. ~By changingthe polarizer settings the two directions are easily distin-guished.! The increased intensity near the corners is due tobending of the current flowlines as they are governed by thebridge geometry.11 The uniform dark appearance of the innerpart of the bridge demonstrates a very good homogeneousscreening.

In the remanent state after switching off the current, Fig.1~c!, one can clearly observe two pairs of bright bands nearthe edges separated by the dark shielded area. The inner twobands show the trapped flux, which has opposite direction inthe two bands. The outer two bands show re-magnetized re-gions, i.e., the return field of the trapped flux partly penetrat-

a!Also with A. F. Ioffe Physico-Technical Institute, 194021 St. Petersburg,Russia.

3147Appl. Phys. Lett. 71 (21), 24 November 1997 0003-6951/97/71(21)/3147/3/$10.00 © 1997 American Institute of Physics

ing into the film. The flux bands therefore represent an os-cillating flux profile across the bridge where the fieldchanges sign three times, as predicted by theory.12 A moredetailed comparison is presented in a forthcoming paper.

Increasing the current above 6 A caused burning, i.e.,Jc523107 A/cm2 at 15 K, leaving the bridge in the rema-nent state shown in Fig. 1~e!. More than 50% of the bridgeappears dark because the trapped flux escaped this regionduring the heating. In order to localize where fatal destruc-tion took place an external field was applied to the sample.Figure 1~f! shows such an MO image with an applied field of5 mT. The destroyed region is the bright channel crossing thebridge.

From the details of the pre-burning MO images thebridge had two regions near the edge, marked by arrows,showing enhanced flux penetration in the current carryingstate. Other irregularities can also be seen in the flux front,but these are of less magnitude. Of the two major defects theone associated with region 1 is the more pronounced, andshould therefore be the most probable candidate to initiatethe burning. Surprisingly, the bridge ‘‘chooses’’ instead re-gion 2, as evident from Fig. 1~f!. Note also that none of thedefects at any pre-burning stage extends across a substantialpart of the bridge even when JT.Jc , as in Fig. 1~d!. It istherefore clear that MO imaging alone is not capable of pre-

dicting the burning location in a bridge which is free fromextended defects. This situation is quite different from thecase studied in Ref. 5, and also for the ‘‘bad’’ sample dis-cussed below, where pronounced extended defects domi-nated the behavior.

Using the LTSEM technique, described in detailelsewhere,8 we made prior to the MO investigation, a map ofthe local Tc for each bridge. Such a Tc map of the ‘‘good’’bridge is shown in Fig. 1~right! where the dark regions cor-respond to reduced values of Tc . Here one clearly recog-nizes regions 1 and 2 from the MO images as also beingareas with relatively low Tc . If one assumes that regionswith low Tc also have reduced Jc it is evident that region 2 isthe more probable candidate as the area of low Tc here per-colates across the entire bridge. Indeed, the channel of de-struction seen in the MO image strongly correlates with thepercolating path of low Tc .

One can also notice that a low Tc path exists near theupper boundary of the bridge. The reason why we believethat this region is not fatal is as follows. First, the MO im-ages show no distinct features in this part of the bridge,indicating that the region has an enhanced concentration ofpinning centers which prevents reduction in Jc in spite oflow Tc . Second, the central part of the bridge is likely to bein the worst situation with respect to thermal stability. Thus

FIG. 1. ~a! Sketch of the bridge geometry. ~b!–~f! MO images of the ‘‘good’’ bridge @scale bar in ~b! is 100 mm long#. ~b!,~d! The self-field distributions forcurrents IT53 and 6 A, respectively. ~c!,~e! The remanent state after switching off the 3 and 6 A currents, respectively. ~f! Destroyed bridge in an externalfield of 5 mT. In this frame the the bright areas near the contact pads are due to the magnetic prehistory of the sample, and thus irrelevant for the illustrationof the burned region, which appears as a bright channel near the center. ~Right! Tc map of the ‘‘good’’ sample: ~black! 90 K, ~white! 92 K.

3148 Appl. Phys. Lett., Vol. 71, No. 21, 24 November 1997 Gaevski et al.

one expects that the device will burn at some ‘‘weak region’’close to the central part.

The MO images of the ‘‘bad’’ bridge are shown in Figs.2~a!–2~e!. In 2 ~a! one sees an image of the remanent stateafter applying a magnetic field of 150 mT which caused fullpenetration. In this case, the pair of dark lines separate theuni-directed trapped flux in the center from the reverselydirected return flux. Adding now a transport current, the fielddistribution becomes asymmetric, as shown in 2~b!–2~d!.According to Ref. 12 an additional line with Bz50, i.e., withdark appearance, should exist within an interval of IT values.Figure 2~b! gives clear evidence for this prediction of thetheory, not previously verified for a state with such a com-bined magnetic prehistory.

The main observation here, however, is that in thissample an extended defect is visible in the MO images. Theflux pattern in Fig. 2~c! is severely distorted near a ‘‘weakregion’’ where the field penetrates more than half of thebridge width. Note however that, at large IT close to thecritical value as in 2~d!, the field distribution becomes moresymmetric. We believe that in this regime the transport cur-rent drives the cross section containing the ‘‘weak region’’into a resistive state with resistivity of the same order ofmagnitude as in the nondefected part. Finally, at I5Ic53 A,the bridge burns out. From 2~e! showing how the destroyed

bridge shields an external field it is clear that the burningtook place in the ‘‘weak region.’’ The micro-photograph ofFig. 2~f! shows that the damaged part of the film consists ofa narrow straight line, not directly across the bridge, butinstead forming a certain angle. The ‘‘bad’’ bridge was evi-dently destroyed along a linear defect present in the film.

We conclude therefore that MO imaging can be used tolocate extended defects which become fatal for HTS bridgescarrying large transport currents. In films showing no ex-tended defects, like the ‘‘good’’ sample in our study, thesituation is more complex. From MO imaging one can findseveral candidates for the location of burning, and henceadditional diagnostics is required. The method of LTSEMmapping of Tc provides another set of candidates, which ingeneral do not completely overlap with the ones from MOimages. We have shown that bands of low Tc extendingacross the bridge, and also overlapping with edge defectsseen by MO imaging, are the actual place of fatal destructionwhen IT>Ic . The burning is usually seen to take place nearthe central part of the bridge. This indicates that in order tooptimize the performance of superconducting bridge devicesit is important to design them in a convex shape in order toincrease the heat transfer. In spite of the considerable fieldamplification near the sharp corners, see e.g., Fig. 1~d!, wefound that burning never takes place here. Thus, a designwith rounded corners will probably not improve the strengthof the bridge with respect to burning.

The authors thank The Norwegian Research Council andthe Russian State Program on HTSC ~Grant No. 94048! forfinancial support.

1 R. P. Huebener, in Advances in Electronics and Electron Physics, editedby P. W. Hawkes ~Academic, New York, 1988!. Vol. 70, p. 1.

2 M. R. Koblischka and R. J. Wijngaarden, Supercond. 8, 199 ~1995!.3 A. V. Gurevich and R. G. Mints, Rev. Mod. Phys. 59, 941 ~1987!.4 V. K. Vlasko-Vlasov, M. V. Indenbom, V. I. Nikitenko, A. A. Polyanskii,R. L. Prozorov, I. V. Grakhov, L. A. Delimova, I. A. Liniichuk, A. V.Antonov, and M. Y. Gusev, J. Supercond. 5, 1582 ~1992!.

5 M. V. Indenbom, A. Forkl, H. Kronmuller, and H.-U. Habermeier, J.Supercond. 6, 173 ~1993!.

6 T. H. Schuster, M. R. Koblischka, B. Ludescher, W. Gerhauser, and H.Kronmuller, Phys. Status Solidi A 130/2, 429 ~1992!.

7 M. D. Johnston, J. Everett, M. Dhalle, A. D. Caplin, J. C. Moore, S. Fox,C. R. M. Grovenor, G. Grasso, B. Hensel, and R. Flukiger, in High-Temperature Superconductors: Synthesis, Processing, and Large-ScaleApplications, edited by U. Balachandran, P. J. McGinn, and J. S. Abell~The Minerals, Metals & Materials Society, Warrendale, PA, 1996!, p.213.

8 A. V. Bobyl, M. E. Gaevski, S. G. Konnikov, D. V. Shantsev, V. A.Solov’ev, and R. A. Suris, Scanning Microsc. 10, 679 ~1996!.

9 S. F. Karmanenko, V. Y. Davydov, M. V. Belousov, R. A. Chakalov, G.O. Dzjuba, R. N. Il’in, A. B. Kozyrev, Y. V. Likholetov, K. F. Njakshev,I. T. Serenkov, and O. G. Vendic, Supercond. 6, 23 ~1993!.

10 T. H. Johansen, M. Baziljevich, H. Bratsberg, Y. Galperin, P. E. Lindelof,Y. Shen, and P. Vase, Phys. Rev. B 54, 16 264 ~1996!; L. A. Dorosinskii,M. V. Indenbom, V. I. Nikitenko, Yu. A. Ossip’yan, A. A. Polyanskii, andV. K. Vlasko-Vlasov, Physica C 203, 149 ~1992!.

11 M. Baziljevich, T. H. Johansen, H. Bratsberg, Y. Shen, and P. Vase, Appl.Phys. Lett. 69, 3590 ~1996!.

12 E. Zeldov, J. R. Clem, M. McElfresh, and M. Darwin, Phys. Rev. B 49,9802 ~1994!.

FIG. 2. ~a!–~e! MO images of the ‘‘bad’’ sample. ~a! Remanent state ob-tained by switching off a field of 150 mT. ~b!–~d! Increase of transportcurrent up to 3 A. ~e! The destroyed bridge in an external field of 5 mT. Thescale bar is 100 mm long. ~f! SEM image of the area marked by whiterectangle on ~e!.

3149Appl. Phys. Lett., Vol. 71, No. 21, 24 November 1997 Gaevski et al.