850 zyxwvutsrqponml IEEE TRANSACTIONS zyxwvu ON MAGNETICS, VOL. MAG-21, NO. zyx 2, MARCH 1985 CRITICAL CURRENT BEHAVIOR AND OXIDE BARRIER PROPERTIES OF Ta SURFACE LAYERS ON Nb S.T. Ruggiero and G.B. Arnold University of Notre Dame Notre Dame, IN 46556 and E. Track and D.E. Prober Yale University zyxwv P.0.BOX 2157 New Haven, CT 06520 Abstract We have investigatedthecriticalcurrent, IcR, and oxide barrier shape in Nb/Ta/Ox/C .E. tunnel junc- tions. Here, layers of Ta in the thickness range O<D~~<1000# were deposited, in situ, on 20008 thick Nb underlayers. Junctions were. completed with Pb, PbBi, and Ag counter-electrodes. W e find that as D is in- creased,there i s a more rapid decrease of I Kacompared with the effective energy gap, in accord with an ex- tended version of the Gallagher theory1 including strong-coupling and electron-scattering effects. In addition,-we have investigated the average barrier height, $, and width, s, of the oxide barriers which form on the Ta overlayers. It is observed empirically that T - zyxwvut 6/(s-so) where so - 108 and 5 is measured in eV. This relationshipisalso found to hold for barrier formation on a wide variety of pure and com- posite metallic systems. These results are discussed in conjunction with the Fromhol d-Mott-Cabrera theory2 for self-1 imiting oxide growth on metal surfaces. Introduction In this paper we discuss the critical current and oxide barrier properties of tunnel junctions with base electrodes comprising T surface layers on Nb. As pre- viouslf-$iscussed, Ta33$ and a number of other metals with desirable.oxidation properties have been deposited on Nb to form high quality tunnel junctions. However, along with the improvement in I-V charac- t e r i s t i c s come an associated decrease in the energy gap and IcR product and a strong change in oxide barrier properties. A study of the systematics of these effects in Ta overlayer systems on Nb is discussed. Experimental The junctions were prepared by depositing thin Ta overlayers, in situ, on 20008 Nb f i ms by ion-beam deposition as previously discussed.l In order to pre- pare a series of samples each with a different Ta thickness and an identical Nb/Ta interface, a sample bolder was employed which sequentially shuttered up to 5 samples after the Ta deposition had been initiated. Afterdeposition, and - 30 min. had elapsedtoallow internal cryogenically cooled surfaces to warm, the chamber was either vented with room a i r , or first brought to - 1-2 Torr with pure oxygen for 3-4 minu- tes. The samples were then transferred to a thermal evaporation chamber (within - 12 minutes) wherein 20008 of Ge was deposited on samples through a wire mask to insulate base elecrode edges and define a 75um line. To c r e a t e 75um x 75um area junctions a second Ge deposi- tion was made orthogonal to thefirst. Otherwise (and more t y p i c a l l y ) 75pm x 350um junctions were made by de- positing counterelectrodes through a mechanically slot- ted mask. Ag, Pb, and 29 wt. % Bi PbBi films were used to complete the junctions. zyx 0 I zyxwvut 2 3 4 VOLTAGE (MILLIVOLTS 5 Fig. 1. Current-voltage curves for a tunnel junction with a base electrode comprising 758 Ta on 20008 Nb. Energy Gap and IcR Product Surfacelayers on Nb, while providing sharp, low leakage I-V characteristics, can also have a strong effect on junction characteristics in the-vicinity of the sum gap. Since over1 ayer materials charac- teristically have Tc s less than that of Nb, any overlayer metal remaining afteroxidation will form a proximity system wit the underlying Nb. As discussed by Arnold and Wolf ,I6 because of the lower air poten- tial, A ,in the surface layer compared wit! that of Nb, As,%nd consequent Andreev scattering at the N/S boundary, a bound quasi-particle state will be created in the surface layer. This effect is clearly evident in Fig. 1 for a junction with 758 Ta on Nb, the bound state beingmanifestas a clear peak or "knee" a t an energy Eo + ApbBi. The position of this peak, which decreases in energy with increasingoverlayerthick- ness, DN, sets an upper limit on the effective energy gap of the system. In addition, as discussed by 0018-9464/85/0300-0850$01.0001985 IEEE