Conductive Buffer Layers and Overlayers for the Thermal Stability of Coated Conductors
C. Cantoni, T. Aytug, D. T. Verebelyi, M. Paranthaman, E. D. Specht, D. P. Norton, and
D. K. Christen
one or more insulating oxide buffer layers with a total Abstract—We analyze fundamental issues related to the thermal and electrical stability of a coated conductor during its operation. We address the role of conductive buffer layers in the stability of Ni-based coated conductors, and the effect of a metallic cap layer on the electrical properties of Ni alloy-based superconducting tapes. For the first case we report on the fabrication of a fully conductive RABiTS architecture formed of bilayers of conductive oxides SrRuO3 and LaNiO3 on textured Ni tapes. For the second case we discuss measurements of current-voltage relations on Ag/YBa2Cu3O7-δ and Cu/Ag/ YBa2Cu3O7-δ prototype multilayers on insulating substrates. Limitations on the overall tape structure and properties that are posed by the stability requirement are presented.
Index Terms—Buffer layers, contact resistance, critical currents, metallic oxides, superconducting tapes. I. INTRODUCTION A
t the stage of implementation, coated conductor tapes will have to be electrically and thermally stabilized against transient conditions in which the current may exceed Ic of the HTS coating. Coated conductors are typically formed of a 50 µm thick metallic substrate (Ni or Ni-alloys), Manuscript received September 17, 2000. The submitted manuscript has been authored by a contractor of the U.S. Government under contract DE-AC05-00OR22725. Accordingly, the U.S. Government retains a nonexclusive, royalty-free license to publish or reproduce the published form
of this contribution, or allow others to do so, for U.S. Government purposes.
C. Cantoni is with Solid State Division, Oak Ridge National Laboratory, Oak Ridge, TN 37831-6061 USA (telephone: 865-574-6264, e-mail: cantonic@ornl.gov).
T. Aytug is with Solid State Division, Oak Ridge National Laboratory, Oak Ridge, TN 37831-6061 USA (telephone: 865-574-6271, e-mail: aytug@solid.ssd.ornl.gov). D. T. Verebelyi was with Solid State Division, Oak Ridge National Laboratory, Oak Ridge, TN 37831-6061. He is now with American
Superconductor Corporation (telephone: 508-621-4368, e-mail:
dverebelyi@amsuper.com).
M. Paranthaman is with Chemical and Analytical Sciences Division, Oak Ridge National Laboratory, Oak Ridge, TN 37831-6100 (telephone: 865-574-5045, e-mail: paranthamanm@ornl.gov).
E. D. Specht is with Metals and Ceramics Division, Oak Ridge National
Laboratory, Oak Ridge, TN 37831-6118 (telephone: 865-574-7682, e-mail: spechted@ornl.gov). D. P. Norton was with Solid State Division, Oak Ridge National Laboratory, Oak Ridge, TN 37831-6059. He is now with Department of
Materials Science and Engineering, University of Florida, Gainesville, FL
32611-6400 (telephone: 352-846-0525, e-mail: dnort@mse.ufl.edu).
D. K. Christen is with Solid State Division, Oak Ridge National Laboratory, Oak Ridge, TN 37831-6061 USA (telephone: 865-574-6269, e-mail: christendk@ornl.gov).
thickness of 0.5-1 µm, and an HTS film thicker than 1µm. As a benchmark we can assume that the HTS film (usually YBCO) carries a current density of 1×106 A/cm2 in the superconducting state at liquid nitrogen temperatures (64 - 77 K) and in self magnetic field. In the event of a transient to the dissipative regime, the power generated per unit area in such an electrically-isolated 2 µm thick HTS layer would exceed 104 W/cm2, resulting in destruction of the superconductor. In order to remove the heat generated from the HTS layer, and restore superconductivity, the superconducting film has to be electrically coupled to a conductive metallic layer. This
coupling provides a lower overall resistivity during transient losses of superconductivity, and, consequently, a lower dissipation in the net conductor. In addition, coupling to a good metal greatly increases the thermal conductivity of the entire tape, significantly improving the outflow of heat to the
surrounding refrigerant bath, and, therefore, the cooling rate
of the conductor. The efficiency of cooling depends on many parameters of the application design. However, we consider the simple case of steady-state heat flow removed by contact of the tape with boiling liquid nitrogen. Then, the maximum temperature excursion is determined by limiting the heat flux
to a value below the critical heat flux of LN2 at 77 K (10- 20 W/cm2
) [1]. Electrical coupling of the superconducting film to the metal
layer can be envisioned in two different ways. The textured metal substrate of sufficiently low resistivity, as in the case of pure Ni or Cu, can itself provide the necessary stability for the HTS film if the buffer layer is conducting rather than insulating. For a 2 µm thick YBCO coating that is electrically connected to an underlying 50 µm thick base metal tape of pure nickel, the normal-state heat flux from each unit surface area would be only about 4 W/cm2 per tape surface, well below the critical heat flux of liquid nitrogen. On the other hand, a conductive buffer layer is not an effective route if the substrate of choice is a Ni alloy, which typically would have a resistivity comparable to the normal state resistivity of the superconductor (~100 µΩ-cm @
100 K). In this case, stability can be achieved by applying a metal overlayer such as Cu or Ag on the superconducting film.
Figure 1 shows the calculated dependence on the substrate resistivity of the metal-layer thickness necessary to maintain the transient heat flux below 5 W/cm2 when an HTS film 2 µm thick, carrying a current density of 1 MA/cm2, is driven to the dissipative state. The effect on the overall tape current
18
45
16
40)14
35µm( s1230EJnes10
k205025c8hi154020 (kA/cm2t6103015) pa20c4510102
000246810500020406080100base metal resistivity (µΩ cm)
Fig.1. The approximate calculated thickness required of a highly conductive
cap layer (Cu or Ag), in order to stabilize the transient dissipative heat flux to a value ~5 W/cm2 (2.5 W/cm2 per tape surface), as a function of the 50 µm thick substrate-tape resistivity. The HTS layer is taken to be 2 µm thick, carrying a current density of 1 MA/cm2. The contact resistance
between the HTS and the metal layer is assumed to be negligible. The inset
shows the same plot expanded in the region of low resistivities.
density, JE, is also shown. In this paper we discuss important aspects of both approaches. We report on the development of
a conductive bilayer composed of LaNiO3 and SrRuO3 for Ni-based coated conductors. In addition, we address some stability issues for alloy-based coated conductors by testing transport properties of structures formed by a metal cap layer deposited on YBCO films on single crystal SrTiO3 (STO) substrates.
II. EXPERIMENTAL
Both SrRuO3 (SRO) and LaNiO3 (LNO) are perovskite-type, conductive metallic oxides with pseudocubic lattice parameters of 3.96 Å and 3.86 Å, and room temperature resistivities of about 300 µΩ-cm and 600 µΩ-cm, respectively. We have previously shown that highly textured LNO buffer layers can be successfully deposited onto Ni tapes [2] and we have subsequently optimized the deposition conditions to reproduce in these films the same degree of in-plane and out-of-plane texture of the Ni substrate [3]. The LNO films for this study were deposited by d.c. sputtering from a powder target of 4” diameter. The previously textured and annealed Ni substrates [4] of approximately 3×15 mm in size were mounted on the heater block, and the chamber evacuated to a base pressure of 1 × 10-5 mTorr. After a target presputtering, the temperature was raised to 480 °C and the film grown in a background total pressure P[(96%Ar + 4%H2) + O2] = 10 mTorr, with P[O2]/P[96%Ar + 4%H2] = 0.05-0.1.
The LNO films show several advantages as a first buffer layer for RABiTS tapes. They can be easily deposited by sputtering at a low temperature compared to that used for other oxides like YSZ and CeO2; LNO deposits on Ni are dense and
2
continuous, with no observed microcracks in films to thicknesses of 0.5 µm. However, YBCO films on LNO/Ni
structures show considerably suppressed Tc’s (70-75 K) because of Ni diffusion from the substrate though the LNO film into the YBCO [3,5].
Unlike LNO, SRO behaves as a good Ni diffusion barrier and
is chemically compatible with YBCO as demonstrated by the
growth of YBCO films with Jc(0 T, 77 K)=4 MA/cm2 on SRO buffered LaAlO3 single crystals [5]. Since we were not able to grow epitaxial SRO films directly on Ni, SRO was deposited
after the first layer of LNO had been sputtered on the Ni substrate. Highly (200) oriented SRO films were deposited by
both sputtering and PLD on LNO/Ni samples that had usually been exposed to air after the LNO deposition. For both methods the deposition temperature ranged between 600 and
650 ºC. In the case of PLD the deposition was carried out in a
background oxygen pressure of 1-5 mTorr and with a laser
pulse energy of 4 J/cm2. In the sputtering system SRO was grown using pure Ar at a pressure of 10 mTorr. YBCO films subsequently deposited on these samples showed
Jc(0 T, 77 K) values around 1.2 MA/cm2
and therefore comparable to those of epitaxial YBCO films on standard, insulating CeO2/YSZ/CeO2 buffered Ni substrate. The YBCO
films were deposited by PLD at a substrate temperature of 780 ºC and an oxygen background pressure of 100 mTorr.
III. DISCUSSION Figure 2 shows a comparison between four-probe resistivity measurements for the Ni substrate, the net SRO/LNO/Ni parallel and the completed YBCO/SRO/LNO/Ni structure. m)c µΩ (ytvistisiRe4080120160200240280
Temperature (K)
Fig.2. The net resistivity measured using four-terminal potentiometric measurements of: the bare, textured nickel substrate; the nickel tape coated with the conductive bilayer buffer SrRuO3/LaNiO3; and the complete structure of YBCO/SRO/LNO/Ni. The distance between current probes is 10 mm, and the distance between voltage probes is 4 mm.
The thickness values are 140 nm, 150 nm, 325 nm, and 50 µm for YBCO, SRO, LNO and the Ni substrate, respectively. Because of the electrical connection between the YBCO and the Ni substrate through the buffer layers, the resistivity of the YBCO/SRO/LNO/Ni structure is nearly a factor 30 smaller,
3
and the resistance a factor 104 smaller than that of the isolated
-212YBCO film at 100 K. However, were there complete 1.4 10
77 Kelectrical connection between the YBCO film and the Ni
-21.2 10substrate the resistivity shown in Fig.2 for the total structure 10
HTSwould be indistinguishable from that of the bare Ni substrate.
NetCross-sectional SEM studies of the YBCO/SRO/LNO/Ni 1 10 -28architecture have shown that the additional resistivity is
possibly due to a discontinuous NiO layer that forms under 8 10 -3Capthe oxidizing conditions of YBCO deposition [3]. Since NiO 6has a lower free energy of formation than either LNO or SRO, 6 10 -3its formation is expected, at a rate that is dependent upon the
4kinetics of oxygen diffusion to the LNO/Ni interface. The Data4 10 -3w (W/cm 2) V (Volts)formation of this insulating layer could be explained by the
diffusion of Ni from the metal substrate into the LNO layer,
thereby resulting in deterioration of the stability of LNO and probably formation of a polycrystalline mixture of La-O and
Ni-O. The deterioration of the interface between LNO and Ni
and the formation of a NiO layer (although thin and discontinuous) is not desirable for the development of a robust conductive architecture. Ongoing research is focusing on solving this problem by changing the buffer layer architecture or by passivating the Ni surface with a different metal layer. In the case of coated conductors that use a high resistivity metal substrate, stabilization is carried out by depositing a metal layer on top of the superconducting film. Such a metallic film does not have to grow epitaxially on YBCO, and could be deposited at room temperature. However, as in the case of the conductive buffer layers, the interface between metal and YBCO must have a low areal resistivity in order to provide an effective electrical connection. It is known that the YBCO surface reacts with air to form Ba(OH)2 and BaCO3. Humidity can give rise to other types of surface reconstruction that involve absorption of H2O molecules inside the YBCO crystal lattice. For this reason, ex situ sputter-deposited Ag on the YBCO surface shows surface resistivities that are too high [6,7]. A post-anneal in O2 at 500 °C reduces the contact resistivity to values in the
10-5 Ω-cm2 range, which may be suitable (note, such a procedure is not possible for Cu on YBCO). However, if the contact surface resistivity of the Ag/YBCO interface is higher than 10-8-10-9 Ω-cm2, then the I-V characteristics of our short-segment measurements will show a resistive behavior below Jc, with a slope determined by the fraction of current that partitions into the metallic cap layer.
In the best case, the I-V characteristic of a metallic layer in parallel with a superconductive layer shows zero voltage up to a current equal to the Ic of the superconductor, a small non-linear transition region for current just above Ic, and, finally for I > Ic, a linear behavior with a differential resistivity given by that of the metal layer.
To conduct controlled tests of the stabilization provided by a metal cap layer for the case of YBCO insulated from the substrate, we measured I-V curves of Ag/YBCO and Cu/Ag/YBCO multilayers grown on insulating single crystal STO. The latter might simulate the ultimate stabilizing structure for coated conductors. After growth, the YBCO films were cooled in a background oxygen pressure of
2 10 -32000102030I (A)
Fig. 3. Comparison between the I-V characteristic of a 130 nm isolated YBCO film (diamonds) and the expected I-V dependence for the same YBCO film when electrically connected to a 1.5 µm thick Ag film (net solid
curve). The dotted line indicates the power per unit area dissipated in the
Ag/YBCO bilayer assuming a negligible interface contact resistivity. The open triangles are the experimental data for the Ag/YBCO bilayer. The cap solid line is the Ag film I-V characteristic.
400 Torr at a rate of 10 °C/min. At room temperature the chamber was evacuated to a background pressure in the range
of 10-6
Torr and a 0.1 µm thick Ag film deposited in situ by PLD at a temperature below 100 ºC, and using a laser pulse
energy of 4 J/cm2
and a repetition rate of 10 to 40 Hz. The samples were then annealed ex situ in a furnace at 500 ºC for 30 min. Finally, a Ag or Cu film of a few microns in thickness was deposited to complete the stabilizing layer. Cu films were deposited at room temperature by d. c. sputtering using an Ar background of 20 mTorr, with a deposition rate of 30 Å/s. Figure 3 is a comparison between the I-V curve taken in LN2
for an isolated YBCO film 130 nm thick, and for the same YBCO film after electrically coupled to a 1.5 µm thick Ag cap layer. The substrate dimensions were 3×13 mm, and the
YBCO films showed critical current densities in the range of
3 MA/cm2
at 77 K in self-field. In the plot, the solid curves labeled “cap”, “HTS”, and “net” model the expected partitioning of the total current among the various layers,
assuming a negligible interfacial resistance. The data agree well with the expected I-V curve for the net, parallel combination of the non-linear superconductor and the ohmic cap layer. The power dissipated in the bilayer is greatly reduced with respect to the isolated YBCO film: 1.7 W/cm2 at
I = 1.7×Ic = 20 A for the bilayer versus ~ 1700 W/cm2
at the
same applied current for isolated YBCO. We also notice that the voltage measured in the bilayer below Ic lies in the noise level, indicating a sufficiently low contact resistivity of the Ag/YBCO interface.
Figure 4 shows the same plot generated for a sample formed of a 28 µm Cu film deposited on Ag/YBCO/STO. The Ag thickness is 0.1 µm and the YBCO thickness is 130 nm. In
this case the stabilizing effect of the metal layer is much more
1 10 -21
9 10 -3 77 K0.98 10 -3HTS0.8
7 10 -30.7
)2 m)6 10 -30.6cs/tl5 10 -30.5
Wo( (Vw V4 10 -30.4 xul3 10 -3Net0.3
F t2 10 -3Cap0.2aeH1 10 -3Data0.1
000102030
I (A)Fig.4. Comparison between the I-V characteristic of a 130 nm isolated YBCO film (diamonds) and the expected I-V dependence for the same YBCO film when electrically connected to a 28 µm thick Cu film (net solid
curve). The dotted line indicates the power per unit area dissipated in the entire structure assuming a negligible interface contact resistivity. The open triangles are the experimental I-V data for the Cu/Ag/YBCO trilayer discussed in the text. The cap solid line is the Cu film I-V curve. Stabilizing caplayerYBCO +buffersMetal tapeFig.5. A plot of the calculated cap layer thickness and resulting overall current density, JE, as a function of the HTS layer thickness, for an envisioned coated conductor with a 50 µm thick alloy substrate tape. The curves are generated for the case where Jc=1 MA/cm2, and the generated heat flux is limited to 5 W/cm2.
evident. In fact, the linear differential behavior in the net I-V curve is much more pronounced for currents just above Ic, and the power dissipated at 2×Ic = 21 A is only 0.24 W/cm2.
It is worth noting that the requirement of stabilization of a coated conductor (whether it is achieved by a metal cap layer or electrical connection to a low resistivity substrate) and the requirement of maximum overall current density JE are not independent. The simple model of conduction through parallel layers of different resistivity gives the dependence of JE on the HTS layer thickness ds with the condition that, when the superconductor is in the dissipative state (Js › Jc), the power generated in the entire structure is smaller than a
4
criterion value. The calculated JE shows a maximum JEmax as function of ds. The expression for JEmax and the corresponding superconducting film thickness dsmax are:
JEmax
= Jc {(ρs - ρc)/ρs + 2[Jc2 ρc2 ρm,c-1(dm /w)] ½}-1;
dsmax = [(w dm)/(Jc 2 ρm,c)]½,
where ρm,c = ρm ρc/(ρm - ρc), ρm is the resistivity of the metal
substrate, ρc is the resistivity of the metal cap layer, ρs is the
resistivity of the superconductor, dm is the thickness of the
metal substrate, Jc is the critical current density of the
superconducting film, and w is the areal power dissipated in
the net conductor, respectively. Figure 5 shows a plot of JE and the thickness of the stabilizing cap layer versus the YBCO thickness, assuming an upper limit of 5 W/cm2 for the
dissipated power and a Jc of 1 MA/cm2 for the
superconducting layer on an alloy tape. In this case JE has a
maximum value of 34.1 kA/cm2 that corresponds to a YBCO
film thickness of 3.5 µm and a metal cap layer thickness of 46.3 µm. Further increase in the YBCO thickness beyond 3.5 µm, while providing more overall current, actually decreases JE due to the rapid increase in cap layer thickness required to maintain stability. In conclusion, further development of coated conductor tapes
cannot neglect these fundamental considerations. The issues involve the thickness of a metallic stabilizing layer, its interface resistivity with the superconductor, and economic considerations, as required by stability limitations on the power generated in the dissipative regime.
REFERENCES
[1] E. G. Brentari, and R. V. Smith, “Nucleate and film pool boiling
design correlations for O2, N2, H2, and He,” Adv. Cryog. Eng. vol. 10, pp. 325-341, 1965.
[2] Qing He, D. K. Christen, R. Feenstra, D. P. Norton, M. Paranthaman,
E. D. Specht, D. F. Lee, A. Goyal, and D. M. Kroeger, “Growth of biaxially-oriented conductive LaNiO3 buffer layers on textured Ni tapes for high-Tc coated conductors,” Physica C, vol. 314. p. 105, 1999. [3] T. Aytug, Synthesis and characterization of compatible buffer layers
for development of high temperature superconducting coated conductors (Ph.D. graduation thesis), University of Kansas, July 2000, pp. 62-112.
[4] A. Goyal, R. Feenstra, F. A. List, M. Paranthaman, D. F. Lee, D. M.
Kroeger, D. B. Beach, J. S. Morrell, T. G. Chirayil, D. T. Verebelyi, X. Cui, E. D. Specht, D. K. Christen, and P. M. Martin, \"Using RABiTS to fabricate high-temperature superconducting wire,\" J. of Metals, pp. 19-23 (July,1999).
[5] T. Aytug, J. Z. Wu, C. Cantoni, D. T. Verebelyi, E. D. Specht, M.
Paranthaman, D. P. Norton, D. K. Christen, R. E. Ericson, and C. L. Thomas, “Growth and superconducting properties of YBa2Cu3O7-δ films on conductive SrRuO3 and LaNiO3 multilayers for coated conductor applications,” Appl. Phys. Lett., vol. 76, pp. 760-762, December 1999.
[6] S. E. Russek, S. C. Sanders, A. Roshko, and J. W. Ekin, “Surface
degradation of superconducting YBa2Cu3O7-δ thin films,” Appl. Phys. Lett., vol. 64, pp. 3649-3651, April 1994.
[7] J. W. Ekin, S. E. Russek, C. C. Clickner, and B. Jeanneret, “In situ
noble metal YBa2Cu3O7-δ thin-film contacts,” Appl. Phys. Lett., vol. 62, pp. 369-371, November 1992.
因篇幅问题不能全部显示，请点此查看更多更全内容