Long length superconducting cables have many potential industrial applications, notably power cables and magnet windings for magnetic resonance imaging, high field research magnets, motors and transformers. Bi-based superconducting materials are beginning to be commercialised through demonstrator applications in the transmission industry. However, owing to their poor current carrying performance in the presence of magnetic fields, it is unlikely that the Bi-based superconductors will achieve the property requirements for magnetic winding applications.

Rare earth barium cuprate superconductors of the formula: REBa2Cu307 x (or REBCO), on the other hand, do have adequate intrinsic properties for such applications.

What is currently limiting the performance of these materials is the difficulty in the processing of them because it is necessary to achieve very highly aligned crystallites to allow the superconducting current to be transferred across the crystallites. Another issue is achieving highly aligned metallic substrates and highly aligned buffer layers onto which the superconductor can be grown. There are currently two main methods to achieve the required texture. The first is a method called IBAD (ion beam assisted deposition) which gives a highly textured buffer layer on a polycrystalline, randomly oriented Ni-based substrate, and the second is called RABITs (rolling assisted biaxial texturing of substrates) which produces a highly textured Ni-based substrate by mechanical working.

Typically, the requirements for balancing various different properties of the materials used in the fabrication of such super conductors lead to a multilayer structure, of the kind illustrated in Figure 1. This comprises a metallic/ceramic substrate tape 2, having a protective layer 4 on its reverse surface, whilst the front surface carries a buffer

layer 6 formed by one of the above methods, having a"ReBCO"layer 8 formed on the buffer layer 6. On top of this are a further"shunt"layer 10, and one or more further insulator and stress relief layers 12.

For the IBAD route, the superconducting layer needs to be deposited on the IBAE, buffer. For the RABiTs route, the buffer layer and superconducting layer need to be deposited on the RABiTs substrate. There are a variety of ceramic processing techniques available to deposit these layers. There are advantages and disadvantages to each. The present invention provides improved processing routes and new buffer layer compositions for RaBiTs substrates, including processing methods and compositions having advantages over existing methods for fabrication of both the buffer layer and superconducting layer on a RABiTs substrate.

Accordingly, a first aspect of the present invention provides a method of forming a buffer layer in an elongate superconductor, comprising first forming a nickel oxide seed layer, on a nickel substrate, by surface oxidation epitaxy (SOE), and then growing a further layer of nickel oxide, on the surface of the SOE seed layer, by liquid phase epitaxy.

The use of liquid phase epitaxy to grow the NiO on the surface of the SOE seed layer leads to a much improved NiO buffer layer, which is smoother, more highly textured and has improved grain connectivity.

A further aspect of the invention provides a method of fabricating a superconducting layer or a buffer layer in an elongate superconductor structure comprising (a) feeding a rod of superconductor or buffer layer material plus flux into the focus of a heat source to produce a molten zone at a predetermined higher temperature; and (b) moving an elongate textured substrate in a horizontal plane beneath the molten zone at a suitable rate, the substrate being maintained at a suitable lower temperature to promote epitaxial growth; whereby the liquid from the molten zone falls onto the moving substrate and epitaxial growth is thereby promoted on the substrate.

Preferably, the substrate is a RABiTs Ni-based substrate. The NiO layer has around a 9% lattice mismatch with REBCO and so it is just about possible to grow an epitaxial REBCO film on the surface of the NiO. However, NiO alone is not an ideal buffer layer since there is still some poisoning of the REBCO by Ni in the NiO. On the other hand, NiO so readily forms on the surface of the Ni substrate that it is preferable to fabricate a high quality NiO layer first, and then develop subsequent buffers, rather than trying to develop other buffers on the Ni and then find that a poor quality NiO layer forms during the fabrication of these oxides: the poor quality oxide may lead to spalling or increased surface roughness in those other buffers. The drawback of the SOE approach on its own is that the NiO layer surface roughness is quite high and the grain connectivity is not ideal.

In practice, at least one further oxide buffer layer is required on the surface of the NiO to allow the growth of Ni free REBCO, and according to a further feature of the invention, a rare earth transition metal oxide buffer layer such as Nd2Cu04 may be grown by liquid phase epitaxy on the surface of NiO.

Another aspect of the invention therefore provides a superconductor structure comprising a metallic substrate, a buffer layer structure including (a) a metal oxide seed layer formed on the substrate by surface oxidation epitaxy (SOE) and (b) a further layer of metal oxide such as a rare earth transition metal oxide grown on the surface of the seed layer ; and a superconducting layer formed on the buffer layer or layers. Preferably the further layer of metal oxide is grown by liquid phase epitaxy (LPE).

A further aspect of the invention provides a superconductor structure comprising a metallic substrate of a first metal, and also including a thin metallic buffer layer of a second metal formed by physical vapour deposition (PVD) and a superconducting layer formed on the metallic buffer layer by liquid phase epitaxy (LPE).

The present invention also provides methods which enable Nd2-xCexCu04 (x=0- 0.15) thick films to be grown directly on LaAIO3 substrates and surface oxidised Ni tapes by fast liquid-phase processing methods.

It provides films with a smooth surface and a very good biaxial texture, typically with FWHM=0. 8° and 5° on LaAlO3 substrates and surface oxidised Ni tapes, respectively. Films of thickness of 5-15 pLm have been grown at rates in excess of 2 tm/min. Nd2-xCexCu04 has a good lattice and thermal-expansion match to REBCO, minimum reaction with the high-temperature CuO: BaO solutions, and is non-poisoning to superconductivity. It is an ideal buffer for liquid phase expitaxy (LPE) processing of REBa2Cu307 8 (RE=rare earth, REBCO) thick films.

Since the discovery of high Te superconductors in 1986, there has been an enormous effort worldwide to establish a saleable method for the fabrication of coated conductors on metallic tapes. Although short length samples with Jc over 1 MA/cm3 have been prepared by a number of vapour deposition methods, e. g. pulsed laser deposition (PLD),'they have so far failed to produce long-length conductors at reasonable economic cost. An alternative route to vapour deposition methods is LPE, which has been widely used for the growth of semiconductor and garnet thin films during the 1970s and 1980s.2 LPE does not require a high vacuum system, so the system is much cheaper to build. The advantages of LPE compared to vapour methods are the very fast growth rate, typically 1-10 p. m/min, and the capability of growing thick films without degrading the structural perfection. LPE growth of REBCO thick films on single crystal substrates has so far produced very encouraging results. In-plane texture of the films is generally excellent, with a FWHM less than 3° from XRD pole figures. 3 Typical growth rates of 1 jum/min are achievable and films of thickness up to 10 pm with 100% density can be obtained in minutes. Tu's achieved are above 90 K. In a self-field at 77 K, transport Jc's over 1 MA/cm2 have been achieved and this value does not degrade severely with increasing film thickness up to-10 plu. 4

However, LPE growth of REBCO thick films on metallic substrates is much more difficult. There are special problems associated with LPE. The high growth temperature largely increases the risk of reaction between the substrate and the film. The solubility of RE's in the BaO: CuO solution is very low, therefore nucleation and growth is more difficult than the traditional LPE of semiconductors and garnets. Compared to PLD and other vapor deposition methods, LPE has a much lower available supersaturation and hence a much smaller driving force for growth. As a consequence, a close lattice matched substrate is required. At present, most of the buffer layers have been developed for the use in vapor deposition methods and are therefore not suitable for LPE growth. There is an urgent need to find a suitable metallic substrate and its buffer layers, which can meet the special conditions of LPE. Nd2CuO4, a tetragonal crystal with a=b~0. 394 nm, is a good potential buffer for LPE. Since its composition falls under the RE-Ba-Cu-O system, Nd2Cu04 neither poisons superconductivity nor reacts strongly with the barium cuprate solution at high temperature. It has a fairly close lattice match and thermal expansion match to REBCO. In this letter, we show that it is possible to grow Nd2CuO4 thick films on metallic substrates by a saleable method.

Accordingly a further aspect of the present invention provides a method of growing layers of Nd2Cu04 or Nd2Cu04 doped with Ce, directly on an LaAIO3 substrate or a Ni or Ni alloy substrate, comprising applying a precursor composition of approximately 25% Nd203 : 75% CuO in an organic binder onto the substrate by screen printing, heating to a suitable temperature to reach a single liquid state and then cooling at a controlled rate so as to promote the nucleation and growth of Nd2Cu04.

In a preferred method according to this latter aspect of the invention, the composition is heated to 1150°C and held at that temperature for about 15 minutes before cooling to room temperature at 3000C/hr.

According to another aspect of the present invention, a hydraulically pressed and sintered source rod composed of approximately 40% Nd203 : 60% CuO is heated to

produce a molten zone at one end so that the liquid material drops onto a LaAIO3 substrate or a textured Ni or NiO substrate located beneath it whilst the substrate is maintained at a suitable temperature to induce epitaxial growth.

This latter method in particular allows the formation of a highly in-plane aligned NiO layer on textured Ni or Ni alloy tapes, so that the Nd2Cu04/NiO/Ni architecture can be fabricated using a one-step process, without separate pre-oxidation step. In addition, when doped with Ce4+, Nd2xCex Cu04/NiO/Ni is itself a superconductor at about 25°K, and consequently provides a coated conductor with potential magnetic applications.

The substrates used in the experiments were (001) LaAIO3 single crystals (Epi- polish, size: 10x5x0. 5 mm3) and (002) -textured Ni foils (FWHM=8-10°, size: lOx5xO. 1 mm3). Before being used in the film growth, the (002) textured Ni foils were oxidised in flowing oxygen at 1240-1250 °C for two hours. The NiO layers were formed pseudomorphically on the (002) Ni foils with a biaxial texture around 10° and a thickness of 20 Mm 5~7 Some embodiments of the invention will now be described, by way of example, with reference to the accompanying drawings in which: Figure 1 is a cross-section through a known type of elongate superconductor; Figure 2 is a diagrammatic perspective view of the structure of a first type of superconductor in accordance with the present invention; Figure 3 is a diagrammatic perspective view of the construction of a second type of superconductor; Figure 4 is a schematic view of a horizontal LPE process in operation; Figure 5 is a typical XRD @-29 scan of Nd2Cu04 thick films grown from liquid phase; Figure 6 is a XRD pole Figure measurement of an Nd2Cu4 film grown on an (001) LaAI03 ;

Figure 7a is a photomicrograph of a pure Nd2Cu04 film grown on LaAIO3 ; Figure 7b is a photomicrograph of a pure Ndi. 85CI0 15CuO4 film grown on LaAI03 ; Figure 8a is an XRD pole Figure of an Nd2Cu04 grown on an NiO/Ni substrate; Figure 8b is a photomicrograph of a Nd2CuO4 film grown on surface oxidised Ni; and Figure 8c is an SEM image of a growth island boundary in the film of Figure 8b.

Referring to Figure 2 of the drawings, the process begins with a substrate 12 of Ni alloy, corresponding to the layer 2 in Figure 1, which is formed by RABiTS. This has a lattice constant of 3.52 to 3.6 Å, and a NiO seed layer 14 is then formed on the surface by surface oxidation epitaxy. This layer has a lattice constant of 4.172 Å.

In order to improve the quality of the NiO layer, liquid phase epitaxy may be utilised and leads to a much improved NiO buffer. It is smoother, more highly textured and has improved grain. connectivity. The LPE NiO is grown on a textured piece of Ni foil which has had a short surface oxidation treatment to give a thin SOE NiO layer which is textured but whose surface is not evenly covered and is rough. The growth of the LPE NiO layer occurs by dipping and spinning the foil in a crucible of a LiB02-NiO flux at around 900-1000°C (the composition of the flux is in the primary crystallisation phase field of NiO-for example, a composition 85% {Li20-B203} +15% NiO).

A further buffer layer 16 comprising a transition metal oxide such as Nd2Cu04 is then formed on top of the nickel oxide layer, by liquid phase epitaxy. This layer has a lattice constant of 3. 92A, as against 3. 86A for REBCO, and is consequently a good lattice match. It also has the advantage that it contains only the significant elements present in REBCO, namely RE and CU.

This is important for the LPE process since elements different to those in the flux used to grow the REBCO layer 18 will dissolve rapidly in that flux during growth of the REBCO layer. It is also a potential buffer layer for deposition using other routes. It

should be noted that other RE elements could be substituted for Nd in Nd2Cu04 to be compatible with the RE in REBCO.

Nd2Cu04 melts incongruently but has a high Nd solubility in the self flux of CuO ana so is easily grown by LPE. For example it can be grown on LaAI03 substrate at 1100°C. The approximate growth rate obtained is 5pm/mm, providing films which are highly textured.

An alternative construction is illustrated in Figure 3, in which a thin metallic buffer layer 20 is provided instead of an oxide layer. Once again, the process starts with a RABiTs Ni alloy substrate 12, whose lattice constant is in the range 3.5 to 3. 6A.

Ideally, it would be preferable to grow REBCO straight onto Ni without buffer layers. In general this cannot be done due to poisoning effects, so that a buffer layer is required. Oxide buffers are generally used since they are well lattice matched to the REBCO layer and they are generally inert. However, it is beneficial to use a thin metallic buffer layer instead of oxide layer since it is quicker and easier to grow. Pd buffer layers have been tried for growth on RABiTS substrates using physical vapour deposition (PVD) but Pd alone is not, in practice suitable for PVD.

In Figure 3 the buffer layer is Pt or a Pt alloy grown by a PVD technique. Pt is quite closely lattice matched to REBCO and with a small Cu or Ni concentration, is even better lattice matched. It can be grown cold by sputtering in a cube orientation and need only be very thin. Also, the surface roughness is not critical for subsequent growth of a high quality REBCO layer (22) by LPE. As shown in the Figure, the Pt alloy buffer layer 20 has a lattice constant approaching 3. 92A which, again, is well matched to the REBCO (22) lattice constant of 3. 855A.

It will of course be appreciated that the buffer layer of Figure 3. i. e. Pt grown by a PVD technique could be used in addition to the buffer layer of Figure 2 in certain circumstances.

Conventions) methods of liquid phase epitaxy suffer from the drawback that they are difficult to implement on a continuous basis, and Figure 4 illustrates a novel horizontal LPE process which provides fast growth of a REBCO buffer layer. This employs a traditional zone refining furnace with an infra red heating source, as illustrated in the Figure.

A hydraulically pressed and sintered source rod (24) of RE2Cu04 + flux (CuO) is fed into the focus of an IR/laser beam (26) to produce a molten zone at ~1200°C.

When the increasing size of the molten zone (28) exceeds the limit of the surface tension, the liquid (30) falls down onto the moving substrate (32) below, which is maintained at a suitable temperature (-900C) for the epitaxial growth induced from the highly textured substrate. (Nio/Ni etc).

Advantage of this novel LPE process are that it provides a faster growth rate, and allows lower growth temperature at the substrate, which greatly reduces the risk of reaction between the substrate and the growing film. It is also practical for long lengths.

Nd2CuO4 thick films can-also be grown from the liquid phase using a number of other techniques. One method that has been used is top dipping LPE growth. For example a high temperature solution of 25% Nd03 : 75% CuO is prepared at 1180 °C and then quickly cooled to about 1100 °C, at which temperature the substrates are dipped into the liquid and rotated at 60 rpm. After growing in liquid for 2 to10 minutes, the substrates are pulled clear and brought to room temperature in about 30 minutes. In this way it is possible to grow films between 5 and 15 um thick, corresponding to an averaged growth rate of 2 pm/min.

In another method according to the present, invention, a precursor with composition around 25% Nd203 : 75% CuO was screen-printed on the substrates using an organic binder (Johnson Matthey 63/2 binder) and then heated up to 1150 °C to reach a single liquid state. This temperature was kept for about 15 minutes and then cooled

down to room temperature at 300 °C/hr. Because of the large surface area to volume ratio and the faster evaporation of CuO than Nd203, the liquid became supersaturated rapidly, leading to the nucleation and growth of Nd2CuO4 on the substrates. In order to obtain a smooth surface, it was essential to make sure that the film growth had finished and the extra CuO liquid had been evaporated completely before the fast cooling to room temperature started. This was done by proper control of the precursor composition and the dwell temperature and time. There was a large difference in the evaporation rates between the CuO and Nd203 (or CeO2) in air or low p02. This was often reported as a problem in the bulk single crystal growth of Nd2-xCexCu04 and higher oxygen pressure (p02=1-2 atm) was used in order to prevent the evaporation of CuO from the liquid. g Therefore, there was a fairly large temperature and composition window for processing by evaporation.

In a further method according to the invention, a growth system similar to the traditional zone-refining furnace was used. A hydraulically pressed and sintered source rod of 40% Nd203 : 60% CuO was fed into the focus of an infrared beam to produce a molten zone at one end. When the increasing size of the molten zone exceeded the limit of the surface tension, the liquid dropped down onto the substrate below, which was maintained at a suitable temperature (-900 °C) for the epitaxial growth induced from the highly textured substrate. Because of the extremely good wetting property of the high- temperature cuprate solution, the liquid drops quickly spread out on the substrates.

The top-dipping method LPE has the advantage of simplicity and easy control.

The main problem for this method is the low available supersaturation and hence low driving force for dense nucleation in large lattice mismatch substrates such as NiO/Ni.

Therefore, it was hard to achieve 100% coverage. The problem may actually limit its actual application as a saleable process. This led to the development of the methods of the present invention. Films grown by the screen-printing method generally have a better coverage as well as an excellent texture, although it requires a precise control and a

relatively high growth temperature. The"molten rod"method has the important combined benefit of high supersaturation and largely reduced growth temperature at substrates, although a more sophisticated growth system was used. Both of these methods are actually suitable as a saleable process for rapid growth of Nd2-xCexCuO4 buffers.

Among these latter three methods, the screen-printing method has so far been studied in most detail and the results presented below (Figs. 5-8) were measured from the samples made by this method. A typical XRD 6-29 scan of Nd2Cu04 thick films grown from liquid phase is shown in Fig. 5, where the main recorded reflections were (00/) (I=2, 4, 6,..., 12), indicating the film was dominantly c-oriented. The XRD pole figure measurement revealed that the Nd2CuO4 films grown on (001) LaAlO3 substrates had a nearly single crystal quality with the FWHM=0. 8°, as shown in Fig. 6. There was a 18° rotation of the film orientation from the a-axis of substrates, indicating the epitaxial relation was Nd2Cu04 {001} <100>//LaA) 03 {001} <130>. The <100>//<100> epitaxy did not occurred because the <130> directions of LaAlO3 (a=0.378nm) had a better lattice match. Three unit cells of Nd2Cu04 (a=0.394nm) could grow along the <130> axis of LaAlO3 with a relative lattice mismatch of about 1%, i. e.

1/3#<130>#LAO = 1/3 #(3aLAO = 0.398nm By doping 15% smaller ion of Ce4+ into Nd2Cu04, the epitaxial relation was switched to <100>//<100>, as shown in Fig. 7. In the pure Nd2CuO4 film in Fig. 7 (a), the boundaries of squared growth islands were found tilted an angle of about 18° to the substrate edge (a-axis), while in the Nd1. g5Ceo. i5Cu04 films in Fig. 7 (b), the growth boundaries were parallel to the substrate edges. Thus the Ce doping gives a better oriented Nd2CuO4 as well as having its own superconducting characteristics.

Although the texture of surface oxidised Ni foils was only about 10°, the Nd2Cu04 films grown on such NiO/Ni substrates showed an improved biaxial texture over substrates, as shown in the XRD pole figure in Fig. 8 (a). The typical FWHM of Nd2Cu04

was about 5°. Such an improvement in film texture has often been observed in the films grown by LPE, 10 which is another advantage of liquid assisted growth. Under the right growth conditions, very smooth surfaces of Nd2Cu04 films were achieved despite the rough surface of NiO/Ni substrates. The surface roughness of Nd2Cu04 films was measured to be 5-1 Onm by an optical interferometer, compared to 200nm for the NiO/Ni substrates. Such a surface is of high quality for subsequent REBCO growth. Shown in Fig. 8 (b) is a typical optical micrograph of the surface of Nd2CuO4 films grown on surface oxidised Ni. A couple of lines on the surface in Fig. 8 (b) are not macroscopic cracks, instead they are faceted boundaries of growth islands, as revealed in the SEM image in Fig. 8 (c), which shows the contrast due to the different heights of the edges of the boundaries.

Similar results were obtained for the Ce doped films on the surface oxidised Ni.

We note that Nd2-xCexCu04 is an electron-doped superconductor with a Tc of above 24 K'1 Hence, the single layer of Nd2-xCexCu04 itself may have potential low temperature magnetic applications without any additional REBCO layer.

In conclusion, Nd2-xCexCu04 thick films have been grown on LaAIO3 substrates and surface oxidised, textured Ni tapes using saleable, liquid phase processing methods. The films had excellent biaxiai texture and a good surface smoothness suitable for subsequent REBCO growth by LPE and other processing routes.

REFERENCES: 1 J. O. Willis, P. N. Arendt, S. R. Foltyn, Q. X. Jia, J. R. Groves, R. F. DePaula, P. C.

Dowden, E. J. Peterson, T. G. Holesinger, J. Y. Coulter, M. Ma, M. P. Maley and D. E.

Peterson; Physica C 335,73 (2000)..

2 R. L. Moon; Liquid Phase Epitaxy, In : B. R. Pamplin, editor, Crystal Growth, 2nd ed., Oxford, Pergamon Press, 1980, pp. 421-61.

3 X. Qi and J. L. MacManus-Driscoll ; J. Crystal Growth 213,312 (2000).

4 S. Miura, K. Hashimoto, F. Wang, Y. Enomoto and T. Morishita; Physica C 278, 201 (1997).

5 Z. Lockman, X. Qi, A. Berenov, R. Nast, W. Goldacker and J. L. MacManus-Driscoll, Physica C 351,34 (2001).

6 K. Matsumoto, K. Seokbeom, G. W. Jian, l. Hirabayashi, T. Watanabe, N. Uno, and M. Ikeda ; IEEE Transactions on Applied-Superconductivity 9,1539 (1999).

7 K. Matsumoto, Y. Niiori, I. Hirabayashi ; N. Koshizuka, T. Watanabe, Y. Tanaka, and M. Ikeda ; Advances in Superconductivity X; Proceedings of the 10th International Symposium, Springer-Verlag Tokyo, Tokyo, Japan; vol. 2, pp. 611-14,1998 8 A. N. Maljuk, A. A. Jokhov, I. G. Naumenko, I. K. Bdikin, S. A. Zver'kov and G. A.

Emel'chenko, Physica C 329, 51 (2000).

9 K. Gamayunov, f. Tanaka, and H. Kojima; Physica C, 228,58 (1994).

10 Y. Ishida, T. Kimura, K. Kakimoto, Y. Yamada, Z. Nakagawa, Y. Shiohara and A. B.

Sawaoka; Physica C, 292,264 (1997).

11 M. Lang, R. Kursch, A. Grauel, C. Geibel, F. Steglich, H. Rietschel, T. Wolf, Y.

Hidaka, K. Kumagai, Y. Maeno, and T. Fujita; Phys. Rev. Lett. 69,482 (1992).