| Popis: |
Using Pb on Nb,Sn tunnel junctions we measure the phonon-induced deviations from the BCS density of states. The overall strength of the deviations decreases but the shape does not change appreciably over the composition range from 25 at.% Sn (high Tc, strong-coupling) to 20 at.% Sn (low Tc, weak-coupling) . As a class the A15 materials include most of the known high transition temperature superconductors, and consequently they have been studied from many points of view. Electron tunneling measurements, which are well known to provide a wealth of fundamental information about the microscopic properties of superconductors, have not been generally possible with these superconductors, however, because of the problems of making good junctions on such difficult materials. Of particular interest from tunneling studies is the tunneling density of states and the detailed information it contains in principle about the electron-phonon interaction responsible for superconductivity. In the past, despite much hard work, the only really successful quantitative tunneling study (i.e. good enough to yield a tunneling density of states) on any A15 superconductor has been the work of Shen on NbsSn formed by reacting Sn on a bulk Nb substrate /I/. Recently, however, with the availability of VP--v high quality vapor-deposited A15 thin films, greatly improved tunneling measurements have become possible /2,3,4,5/. These studies have provided useful information about the dependence of the energy gap on composition, important diagnostic information about the quality and nature of the films, and even an unambiguous confirmation of a substantial varrier to flux entry when a magnetic field is applied parallel to the surface of Nb,Sn / 6 / . More ft Work at Stanford initiated under support of the U.S. National Science Foundation. Presently supported by the U.S. Office of Naval Research ? Present Address : Bell Laboratories, Murray Bill, NJ 07974, U.S.A. recently the pair (Josephson) tunneling noted earlier 121 has been studied in detail for the case of Nb Sn and found to have some technological interest 3 171. In this paper, however, we focus on the density of states and the electron-phonon spectral function obtained from tunneling on a series of Nb-Sn films with various compositions. The films of interest were deposited by means of dual electron beam codeposition of the elements using the techniques developed by Hammond /8,9/, and the tunneling barriers were formed either using the thermal oxide of the as-deposited A15 film or using an evaporated layer of Si deoosited on the film before exposure to the atmosphere / 4 / . In all cases reported here the counter-electrodes were Pb. We note that the success of this work is highly dependent on the quality of the thin films that can be obtained with coevaporation. The surface region near the barrier is crucial because of the short sampling depth of the tunneling electrons (QJ C0 = 5 nm) in the A15 superconductors. At high bias voltages this sampling depth is even shorter and the tunneling measurements are sensitive to at most the first few atomic layers. The current-voltage characteristic, I@), of a typical Nb Sn/oxide/Pb tunnel junction is presen3 ted in figure 1. The quasiparticle (Giaever) tunneling curve is excellent and the leakage conduction at voltages below the gap is low, there being a small onset at the lead gap, $b, probably indicative of a finite density of states within the niobium-tin gap. [~ncidentl~, the dc pair (Josephson) tunneling current has been suppressed by application of a small magnetic field.] We obtain a preliminary Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphyscol:19786577 Fig. 1 : Current-voltage characteristic at 1.5 K of a Nb ~n/~xide/Pb tunnel junction 3 estimate of A Nb-Sn by measuring the voltage corresponding to the point of maximum slope in I(V) and subtracting the known energy gap of the counterelectrode, APb (1.5 K) = 1.35meV. For the junction of figure I , A = 3.25 meV yielding a Nb 3Sn 2A/kBTcZ 4.3. More details of the material properties of the films and the procedures for making junctions are given elsewhere /5 / . Figure 2 shows the dynamic resistance, dV/dI, as measured on the junction shown in figurel. Curve (A) shows data taken at 1.5 K, the superconduc tivi ty of the Pb counter-electrode having been suppressed in a magnetic field of 210 mT. Curve (B) is the corresponding data taken at 20 K in the normal state, and displaced slightly in dV/dI to match up at high bias voltages. The phonon-induced structure in (A) is clearly evident. To illustrate this phonon-induced structure in more useful quantitative form, we reduce the data by calculating the normalized tunneling conductance d ~ / d ~ l s / d ~ / d ~ I n (i . e . the tunneling density of states for the superconductor) and subtracting from this the BCS density of states expected for a superconductor with the measured energy gap. The resulting deviation from the BCS density of states, o(w), is shown in figure 3. To illustrate the effect of choosing a smaller energy gap, we also show the data reduced using Am Sn = 2.8 meV. The electron-pionon spectral function, a2(w)F(w), has been calculated from A and o(w) using both the McMillan-Rowell /lo/, and the alternative Fig. 2 : The dynamic resistance, dV/dI, of the junction shown in figure 1, (A) at 1.5 K in a magnetic field to quench the superconductivity of the Pb, and (B) at 20 K method proposed by Galkin et al. /I]/ and implemented at Stanford by D.B. Kimhi. The results obtained using the former procedure are shown in figure 4. There are clearly resolved peaks in a2 (w)F(w) near 9, 18 and 25 meV. Moreover the overall shape is not sensitive to the gap value chosen. From the a2(w)F(w) obtained using the measured gap of 3.15 meV, we calculate A 2 a2(w)F(w)w-l dw = 0.76 and the I Coulomb pseudopotential p P = 0.10. In fact all junctions analyzed to date give p* < 0 indicating that the observed deviation from the BCS density of states is not strong enough to give an a2(w)~(w) consistent with the gap estimated from I(V) within the framework of the Eliashberg gap equation. The origins of this problem are not clear, but we suspect our surfaces and/or barriers are still not ideal. We note, however, that certain useful. integral averages over the electron-phonon spectral function do not depend strongly on the choice of A and the resulting value of u". Such integrals should be physically meaningful and are summarized intable I. Moreover, keeping this same shape of a2 (w)F(w) and fixing u* = + 0.11 we obtain X = 1.6 ?r 0.1. |
