in Wire Development
RABiTS: AN APPROACH TO HIGH CRITICAL CURRENT DENSITY SUPERCONDUCTING TAPES
[Adapted from David P. Norton et al., "Epitaxial
YBa2Cu3O7 on Biaxially Textured Nickel (001): An
Approach to Superconducting Tapes with High
Critical Current Density," Science 274, 75557
Since the discovery of high-temperature superconductivity (HTS) in cuprate materials, significant efforts have focused on developing a high-current superconducting wire technology for applications at 77 K.13 Early in these efforts it was observed that randomly oriented polycrystalline HTS materials have critical current densities (Jcs), < 500 A/cm2. In contrast, oriented YBCO thin films grown epitaxially on single crystal oxide substrates, such as (001) SrTiO3, exhibit Jc values > 1 MA/cm2 at 77 K.4 This huge difference between randomly oriented HTS ceramics and single crystal-like epitaxial films is directly related to the misorientation angles at the grain boundaries in polycrystalline materials. Values for Jc across a grain boundary decrease significantly as the misorientation angle increases, with weak-link behavior observed for grain-boundary misorientation angles greater than ~10°.513 To achieve high Jcs (~105 to 106 A/cm2, 77 K ), the crystallographic orientation of the HTS wire or tape must possess a high degree of both in-plane and out-of-plane grain alignment over the conductor's entire length. Preferably, this would be achieved with YBCO because the limits for dissipation-free current at 77 K in an applied magnetic field are most favorable for this material.14,15
One approach to producing a high-Jc HTS tape is to deposit a thick epitaxial film on a substrate material that has a high degree of in-plane and out-of-plane crystallographic texture and can be produced in long lengths. Epitaxial HTS films on single crystal oxides satisfy the requirements for high Jc, but it is not feasible to produce long lengths of these substrates. Recent efforts have focused on the use of ion-beam-assisted deposition (IBAD) to achieve in-plane alignment of oxide buffer layers on polycrystalline metal substrates for subsequent epitaxial growth of YBCO.1620 Indeed, as shown by Los Alamos National Laboratory and others, a modest degree of in-plane texture for c-axis-oriented YBCO films using IBAD results in a significant increase in Jc, with values ranging from 105 to 106 A/cm2 at 77 K.
We report here an approach for achieving in-plane-aligned, high-critical-current (Ic) YBCO films on long-length substrates without IBAD. A biaxially textured (001) Ni tape, formed by recrystallization of cold-rolled pure Ni,21,22 is used as the initial, in-plane-aligned substrate. A (001)-oriented oxide buffer layer architecture is then epitaxially grown that maintains the sharp crystallographic cube texture of the metal substrate while providing a barrier to chemical interaction with the Ni. Subsequent growth of YBCO on this structure, referred to as a rolling-assisted, biaxially textured substrate (RABiTS), results in c-axis-oriented, in-plane-aligned films with Jcs as high as 700,000 A/cm2 at 77 K. The advantage of this approach over other alternatives, such as IBAD, is in the simplicity of producing the initial in-plane alignment, required for high Jc, by a cold-rolling and annealing process that can easily be scaled to produce arbitrary substrate lengths.
|Fig. 1.1. An XRD -2 scan, along with a schematic representation, of a YBCO/YSZ/CeO2/Ni multilayer structure. The XRD scan shows the out-of-plane (001) orientations of the YBa2Cu3O7 and oxide buffer layers.|
Typically, the epitaxial growth of a (001)-oriented cubic oxide on a (001) Ni surface is inhibited by the formation of (111) NiO at the oxide-metal interface.23 The Ni substrates were annealed at 900°C in a mixture of 4% H2 and 96% Ar gas prior to film growth to reduce any NiO on the substrate surface. To further suppress the formation of NiO and to achieve (001)-oriented epitaxy directly on the (001) Ni surface, H2 gas was introduced into the PLD chamber during the initial stages of CeO2 growth. Hydrogen is effective in reducing NiO, while having little effect on the CeO2 film. This (001)-oriented CeO2 layer provides an oxide template directly on the metal surface for the subsequent epitaxial growth of additional oxide buffer and HTS layers. We have also developed an alternative approach that utilizes Pd as an epitaxial noble metal interface between the oxide buffer layers and Ni substrate to inhibit the formation of NiO.22
After the CeO2 film is deposited, a YSZ layer was grown in situ using PLD. The YSZ layer appears to alleviate cracking of the oxide layers that can occur because of the thermal expansion mismatch between the Ni substrate and the oxides. The CeO2 and YSZ layers are each ~500 nm thick. A thick (>0.5 µm) YBCO film is then deposited at 780°C in an oxygen pressure of 185 mtorr. After deposition, the films are cooled at 10°C per minute, with the oxygen pressure increased to 700 torr at 400°C.
| Fig. 1.2. XRD rocking curves and -scans showing the (a) out-of-plane and (b) in-plane texture of a YBCO/YSZ/CeO2 multilayer structure on a rolled textured Ni substrate.
The in-plane crystallographic alignment of the epitaxial YBCO/YSZ/CeO2/Ni structure, as determined by XRD -scans through the YBCO (226), YSZ (202), CeO2 (202), and Ni (222), is also shown in Fig.1.2. The in-plane FWHM for all of the layers is ~6.8° , indicating excellent epitaxy of the oxide layers with the biaxially textured metal. If the grain-to-grain misorientation angles are uncorrelated with a normal distribution, ~90% of the Ni grains have in-plane misorientation angles of 7° or less.24 The in-plane (100) CeO2 and YSZ principal crystallographic axes are rotated 45° relative to the in-plane (100) Ni axis, whereas the a and b axes of the YBCO are rotated 45° with respect to the YSZ axes, all of which are in agreement with near-coincidence site lattice models.25 From the high-resolution XRD scans, the lattice parameters for the YBCO film are very similar to those reported for fully oxygenated bulk materials, with a = 3.82 Å, b = 3.88 Å, and c = 11.691 Å. The normally cubic CeO2 and YSZ layers show a slight tetragonal distortion, with the in-plane a = b = 5.41 Å, out-of-plane c = 5.422 Å for the CeO2, and a = b = 5.12 Å, c = 5.162 Å for the YSZ layer. The tetragonal distortion of the oxide buffer layers appears related to the larger thermal expansion coefficient of the Ni substrate relative to these oxides, which tends to place the oxides in compression upon cooling after film growth.
| Fig. 1.3. Magnetic vield dependence of Jc, measured at 77 K, is shown for a YBCO/YSZ/CeO2/rolled textured (001) Ni structure. Also shown is Jc (H, 77 K) for YBa2Cu3O7 on (001) SrTiO3, Tl-1223 on polycrystalline YSZ [J. E. Tkaczyk et al., Appl. Phys. Lett. 62, 301 (1993)]; and Bi-2223/Ag tape [S. Kobayashi et al., Physica C 258, 336 (1996)].
| Fig. 1.4. The temperature dependence of Jc, measured at 0, 3, and 8 T, is shown for a YBCO/YSZ/CeO2/rolled Ni structure. For comparison, data are also shown for Bi-2212/Ag wire [R. Wesche, Physica C 246, 186 (1995)]; Bi-2223/Ag tape [P. Haldar et al., Appl. Phys. Lett. 61, 604 (1992)]; Tl-1223/YSZ, NbTi [J. E. Tkaczyk et al., Appl. Phys. Lett. 62, 3031 (1993)]; and Nb3Sn (D. R. Tilley and J. Tilley, p. 235 in Superfluidity and Superconductivity, 3rd ed., IOP Publishing, Bristol, England, 1990).
SUPERCONDUCTING PROPERTIES OF YBCO DEPOSITS ON RABiTS
We have investigated several systematics regarding the superconductive transport properties of YBCO epitaxial deposits on RABiTS. A series of superconducting films were deposited by pulsed laser deposition of YBCO onto YSZ/CeO2-buffered Ni tapes, along with control samples deposited under similar conditions onto single crystal SrTiO3 substrates. All the samples were short segments of typical length (1215 mm) and width (~3 mm), with YBCO thicknesses ranging from 1 to 3 µm. In all cases reported here, the RABiTS thickness was approximately 126 µm (~125 µm Ni and ~1 µm oxide buffer layer). The focus of the effort was to make a comparative study of the field- and temperature-dependent Jcs among different YBCO/RABiTS samples and with properties of the "benchmark" materials deposited on SrTiO3. In addition, a preliminary test of bend-strain tolerance was conducted, both in compression and tension, and an assessment was made of the prospects for achieving practical current levels through this type of coated-conductor approach.
The samples were prepared for four-terminal transport measurements by sputter deposition of gold or silver current and voltage pads, followed by annealing at 500°C for ½ h, and slow cooling in 1 atm oxygen. The gauge length for the voltage terminals was 4 mm. For some samples, initial measurements of the resistive transition and high temperature Jc were made on the full 3-mm width. To minimize the heating effects at the current contacts and to reduce the overall currents required, thereby extending the measurements to lower temperatures, the samples were subsequently wet-etch patterned to a bridge width of 1 mm. In addition, in some cases pulsed current measurements extended the measurements to the low-field, low-temperature regime. In all cases, the measurements yielded identical results in the regions of overlap, demonstrating that the samples were macroscopically homogeneous and that the measurements were well controlled.
|Fig. 1.5. The magnetic-field-dependent Jc of YBCO/RABiTS compared with other HTS epitaxial thin films that have been deposited on single crystal oxide substrates. Most data are shown for H||c, the applications-limiting orientation. The characteristics with H||ab is typical of the strong intrinsic pinning observed for all HTS materials.|
Figure 1.5 shows the magnetic field dependence of Jc at 77 K. The result for a 1.4-µm-thick YBCO deposit on a RABiTS sample is compared with high-Jc epitaxial layers thinly deposited on single crystal oxide substrates, for several different HTS material classes. In all cases, the films on single crystal substrates have zero-field Jc values in excess of 1 MA/cm2. While all the films show excellent, similar characteristics with the field parallel to the film plane (H||ab planes, the so-called "intrinsic pinning" case), Fig. 1.5 emphasizes the characteristics with H||c, which is the applications-limiting orientation where the flux pinning is limited by the effects of intrinsic material anisotropy. YBCO is the HTS material of choice for conductor applications, and the high-field properties of YBCO/RABiTS may exceed those of prototype YBCO/SrTiO3 films. In fact, the somewhat suppressed Jc(H = 0) for YBCO/RABiTS may arise from the high density of crystalline defects produced naturally from the growth process on RABiTS.
| Fig. 1.6. The Rutherford Backscattering Spectra for films of YBCO/RABiTS and YBCO/SrTiO3. A comparison of the channeling minimum yield Xmin shows that the YBCO/RABiTS has more crystalline structural defects.
|Fig. 1.7. The dependence of Jc on the orientation of the sample in the applied magnetic field of 5 T, comparing YBCO/RABiTS with YBCO/SrTiO3. The enhanced properties near H||c, along with the relatively depressed values near H||ab, imply the existence of strong extended pinning defects in the YBCO/RABiTS sample.|
|Fig. 1.8. The dependence of Jc at 77 K on YBCO thickness for a series of YBCO/RABiTS samples. Open symbols are values in self-field, while closed symbols represent Jc in an applied field H||c = 1 T. The two represent required Jc levels to satisfy different practical operating criteria. Sold curve: overall engineering current density of 104 A/cm2. Dashed curve: tape conductor capacity of 10 A/mm width.|
Figure 1.6 supports this idea in a comparison of the Rutherford Backscattering Spectra (RBS) for a YBCO/SrTiO3 and a YBCO/RABiTS. Analysis of the random spectrum reveals that the YBCO/RABiTS composition is yttrium rich, and slightly barium poor. This result may explain the slightly reduced Tc of 86 to 88 K. Figure 1.6 also shows that the RBS channeling minimum yield Xmin ~ 34% for YBCO/RABiTS compared with Xmin ~ 4% for the YBCO/SrTiO3. Measurements on other YBCO/RABiTS samples indicate that Xmin falls in the range of 30 to 50%. The relatively suppressed channeling characteristics of YBCO/RABiTS imply crystalline defects that may be responsible for the enhanced pinning at high vortex densities. Detailed transmission electron microscopy (TEM) characterizations will be required to help confirm these conclusions.
The enhanced flux pinning at high fields for H||c is accompanied by a slight depression of the Jc for H||ab. This fact is illustrated in Fig. 1.7, which compares the angular dependence of Jc for a YBCO/RABiTS with a YBCO/SrTiO3 as the applied field orientation is swept from near H||c to H||ab. In this example, for an applied field of 5 T at 77 K, a crossover exists in the value of Jc at intermediate orientation. In addition, the Jc peak near H||ab is less sharp and has additional structure for the YBCO/RABiTS sample. These characteristics are reminiscent of film properties before and after the introduction of columnar flux pinning defects by heavy ion irradiation, where Jc enhancements at high fields for H||c are accompanied by decreases in zero field and for H||ab. In this case, one might anticipate that the YBCO/RABiTS materials intrinsically possess some sort of strong-pinning extended defects. Indeed, previous work on YBCO films deposited on vicinal single crystal substrate surfaces has demonstrated enhanced flux pinning and extended "stacking faults" defects propagating upward through the films caused by nucleation and growth at atomic surface steps in the substrates. Again, confirmation of a similar situation must await microstructural analysis.
While the field-dependent Jc levels in YBCO/RABiTS prototypes are adequate to enable new applications in the liquid nitrogen temperature range, the fraction of superconductor in the typical entire structure is presently quite small (~1 µm/126 µm, < 1%). One of the important issues to scaleup is obtaining a sufficient level of overall current density, presumably achieved by the combination of both thinner RABiTS and thicker YBCO deposits. To this end, it is important to determine whether good superconducting properties can be retained in thick HTS deposits. In a preliminary study, we have addressed the dependence of Jc on the YBCO deposit thickness. Figure 1.8 shows these results for applied fields of zero and 1 T applied parallel to the c axis. Most data are given at 77 K, but some results are presented at 64 K (the temperature of pumped liquid nitrogen just above its solidification point). For YBCO thicknesses of ~1 to 3 µm, Fig. 1.8 shows that Jc(1 T) exhibits no systematic thickness dependence with values in the range of ~100 to 150 kA/cm2 at 64 K. For comparison, the two curves represent Jc values required, assuming the present RABiTS dimensions, to satisfy the two applications criteria of overall Je = 1 × 104 A/cm2 (solid line) and Kc = 10 A/mm-width (dashed line) for a tape conductor. Even though the present structures have a small superconductor fraction, the data at 64 K suggest that practical levels are accessible in the liquid nitrogen temperature range.
| Fig. 1.9. The dependence of Jc for two different YBCO/RABiTS samples on the bend strain produced in both compression and tension. Also shown is the room-temperature normal-state resistance, which serves as an indication of a breakdown by cracking.
Conductor applications will require strain
tolerances that permit bending to form coils for
components such as magnets and motors. We
have conducted initial bend-strain tolerance
measurements on two separate samples--one
placed in compression and the other in tension.
Measurements were made sequentially by
bending and straightening the sample around a
specified-diameter mandrel and measuring Jc at
77 K in zero field. The repeated mechanical
cycling associated with this technique resulted
in a noticeable work hardening of the Ni. The
results, along with the effect on the sample
resistance R at room temperature, are shown in
Fig. 1.9. In compression, Jc degrades rapidly
near a strain of 0.5%; the abrupt increase in R at
this same strain level indicates the formation of
cracks. For the present RABiTS dimensions,
this strain level of 0.5% corresponds to a bend
diameter of about 2.5 cm. In tension, the
degradation onset occurs at smaller strain level
but is more gradual. Crack formation is evident
at the level of ~0.2%. If these preliminary
results hold for future configurations having
thinner RABiTS and thicker YBCO, there
appears to be a basis for optimism that several
criteria for real applications can be met. Future
developments must focus on issues of
reproducibility, long-length demonstrations,
rates, and problems presented by the magnetic
substrate for applications requiring rapidly
SOL-GEL SYNTHESIS OF LaAlO3: EPITAXIAL GROWTH OF LaAlO3 THIN FILMS ON SrTiO3 (100)
Sol-gel techniques have been developed as viable methods for the fabrication of superconducting14 and buffer layers57 for severals reasons. Sol-gel represents a low-cost, nonvacuum process and also presents a convenient way of coating long-length conductors and irregularly shaped substrates. Films of LaAlO3 are of particular interest as substrates and buffer layers for high temperature superconductors. Critical current densities exceeding 106 A/cm2 at 77 K have been observed repeatedly from the superconductors YBa2Cu3O7810 and (Tl,Bi)(Sr1.6Ba0.4) Ca2Cu3O911 grown epitaxially on single crystal layers of this material by various deposition techniques. Epitaxial films of the superconductors (Tl,Pb)(Ba,Sr)2Ca2Cu3O912 and (Tl,Pb)Sr2(Ca1xYx)Cu2O713 have also been grown on LaAlO3. The lattice mismatch between LaAlO3 and these superconductors is quite small, and LaAlO3 offers exceptional chemical stability and is possibly an excellent diffusion barrier. Our interests in LaAlO3 were to (1) develop a sol-gel route to grow thin films and (2) demonstrate that these films could be processed to give at least preferred orientation but preferably epitaxy. Epitaxial growth of buffer layers on textured surfaces is considered important in the development of practical HTS conductors with high Jcs.14,15
Films of LaAlO3 have been previously
grown via a sol-gel technique on silicon and
sapphire single crystal substrates.16 The
chemistry employed in that work is different
from the chemistry we used in that, while
starting from alkoxides, a chelating agent
(acetylacetone) was used to obtain solutions for
spin-coating. Fairly complex heat treatments
were also employed, but epitaxial growth was
not observed. We chose instead to use an all-
alkoxide system based on methoxyethoxide
complexes in 2-methoxyethanol. This chemistry
has been used with success in preparing lead-based ferroelectric films17 and epitaxial growth
of lead niobium zirconium titanate (PNZT) on
sapphire.18 We report here our successful
preparation of an all-alkoxide precursor LaAlO3
solution and the growth of epitaxial LaAlO3 on
SrTiO3 single crystals.
|Fig. 1.10. Procedure followed to prepare the LaAIO3 precursor solution, powders, and films.|
The preparation of the LaAlO3 precursor solution was based on the method used in the sol-gel synthesis of PbTiO3 and PbZrO3.17 The weighing of solid materials was done in an argon-filled, inert-atmosphere glove box, and the solution preparation was carried out under argon using a modified commercial solvent still (Ace Glass). Aluminum sec-butoxide (Alfa, 95%), lanthanum isopropoxide (Alfa), and 2-methoxyethanol (Alfa, spectrophotometric grade) were used without additional purification. The preparation procedure, including the formation of the powders and films, is schematically described in a flowchart in Fig. 1.10.
Aluminum sec-butoxide (3.89 g, 15.8 mmol)
was refluxed in 50 mL of 2-methoxyethanol.
Approximately 30 mL of sec-butanol/
2-methoxyethanol were distilled from the
solution, which was repeatedly rediluted with 30
to 50 mL of fresh 2-methoxyethanol followed by
distillation. When the boiling point reached that
of 2-methoxyethanol (124°C), after
approximately 1.5 h, exchange of sec-butoxide
by methoxyethoxide was presumed finished. A
stoichiometric amount of lanthanum
isopropoxide (5.03 g, 15.9 mmol) was then
added to the refluxing aluminum solution.
Approximately 30 mL of isopropanol/
2-methoxyethanol was distilled off, and the
solution was repeatedly rediluted with 30 to
50 mL of fresh 2-methoxyethanol and further
distilled for approximately 1.5 h to again ensure
exchange by the methoxyethoxide ligand. The
final volume of the light yellow solution was
adjusted with 2-methoxyethanol to 32 mL to
make a 0.5 M LaAlO3 precursor solution.
Powder Preparation and Analysis
The gel used to prepare powders for differential thermal analysis/ thermogravimetrical analysis (DTA/TGA) analysis and XRD was made by adding three times the volume of a 1 M H2O in 2-methoxyethanol solution to a volume of the precursor solution. This light yellow, transparent gel was decomposed on a hot plate with a portion of the resulting product being used for the DTA/TGA and the remainder being fired between 500 and 800°C in air for 1 to 12 h to determine the crystallization temperature.
DTA/TGA data were obtained in a Sinku-Riko TFD7000 RH instrument. Samples were
contained in MgO crucibles and heated in air
from room temperature to 900°C and in oxygen
from room temperature to 1250°C at a rate of
10°C/min. The XRD pattern of LaAlO3 powder
was collected using a diffractometer (Bragg-Brentano) equipped with a theta-compensating
incident-beam divergence slit and a graphite
(002) diffracted-beam monochromator. Cu K
radiation was used.
Coating Procedure and Film Characterization
The solution needed for spin-coating was prepared by partially hydrolyzing a volume of the precursor solution using an equal volume of the 1 M H2O in 2-methoxyethanol solution. The SrTiO3 (100) single crystal substrate (Commercial Crystal Laboratories, Inc.) was ultrasonically cleaned in ethanol for 1 h before being coated. Four coatings of the hydrolyzed precursor solution were applied to the substrate using a spin-coater operated at 2000 rpm for 45 s. Between each coating, the substrate was pyrolyzed in O2 in a rapid thermal annealer (RTA, AG Associates Model 610) at 800°C for 2 min. The total thickness of the resulting film was approximately 4000 Å.
The same XRD arrangement used to analyze
the LaAlO3 powders was used to measure the
preferred orientation of LaAlO3 on SrTiO3
(100). Further measurements of the epitaxy of
this film were made using a Rigaku rotating-anode X-ray generator equipped with a graphite
monochromator selecting Cu Ka radiation and
slits defining a 2 × 2 mm incident beam. A four-circle diffractometer was used to collect pole
figures to measure rocking curves of (003)
planes of the (001) textured film, which are used
to determine the out-of-plane alignment, and to
measure scans of the (202) planes, are used to
determine in plane alignment of the film.
Electron micrographs of a LaAlO3 film on a
SrTiO3 single crystal were taken using a Hitachi
S-4100 field emission scanning electron
microscope (SEM). The beam voltage used was
15 kV, and the micrographs were taken at
Results and Discussion
The methoxyethoxide solution produced
both transparent gels and films with excellent
uniformity and coating characteristics. The gels
were converted to ceramic powders to study the
thermal and structural characteristics of the
process as a function of processing
temperatures. Based on results of the study of
powders, thin films were produced on single
crystal substrates to obtain epitaxial films of
LaAlO3.The results are described in detail in the
Effects of Ligand Exchange
The reactivity of the alkoxide ligands with
water is what drives the sol-gel process, but this
process must be controlled to promote the
desired gelling and to avoid precipitation, which
can occur during premature hydrolysis. To
ensure proper gelation, the reactant metal
alkoxides were subjected to ligand-exchange
reactions to convert the sec-butoxide and
isopropoxide ligands to methoxyethoxide
ligands and thus to decrease the moisture
sensitivity of the resulting precursor solution.
Since the bidentate nature of the
methoxyethoxide ligand should "tie up" vacant
coordination sites, the rate of hydrolysis should
be slowed and thus more readily allow the
formation of a gel rather than a precipitate. In
addition, the bidentate nature of the
methoxyethoxide ligand also allows the more
facile formation of mixed-metal alkoxide
complexes. Mixed-metal barium and titanium
complexes with the methoxyethoxide ligand
have been isolated and characterized by X-ray
|Fig. 1.11. DTA/TGA analysis of the pyrolyzed gel between room temperature and 900°C in air; scan rate 10°C/min. Thin line: DTA; thick line: TGA.|
Characterization of LaAlO3 Powders
TGA results (Fig. 1.11) indicated that the pyrolyzed gel lost ~30% of its weight while being fired in standing air between room temperature and 900°C, ensuring complete conversion to LaAlO3. DTA results for the pyrolyzed gel are also shown in Fig. 1.11. The main exothermic peaks at ~315 and ~750°C were observed; however at this time the identity of these intermediate states is not known. The resulting powder was cooled and run again as a baseline correction. A DTA analysis performed in oxygen yielded a very narrow and sharp exothermic peak at 800°C, suggesting a possible melting event that may be necessary to grow a good epitaxial film.
|Fig. 1.12. XRD patterns of the pyrolyzed gel fired between 500 and 800°C. 800°C pattern was indexed based on a simple cubic cell.|
The XRD patterns of the pyrolyzed gel fired between 500 and 800°C in air are given in Fig 1.12. Crystallization begins around 600°C with a well-crystallized product resulting from firing at 800°C. LaAlO3 has a perovskite structure with a 0.2% rhombohedral distortion. Its rhombohedral lattice is described by a hexagonal unit cell with a = 5.374 Å and c = 13.11 Å20 but can also be indexed on the basis of a simple cubic structure with a lattice parameter a = 3.790 Å. The experimental lattice parameter of this well-crystallized, LaAlO3 product, corrected using Si as an internal standard and indexed as cubic, was calculated to be a = 3.796 (1) Å.
| Fig. 1.13. -2 scan of c-axis-oriented LaAlO3 on SrTiO3 (100). S = SrTiO3 and L = LaAlO3. Inset is the rocking curve of the (003) reflection of LaAlO3 (FWHM = 0.87°).
| Fig. 1.14. (a) scan of the (202) LaAlO3 reflection (FWHM = 1.07°) and (b) typical X-ray poly figure from LaAlO3 (202)cubic.
Characterization of Epitaxial LaAlO3 Films
For growing epitaxial films of LaAlO3,
800°C in oxygen was chosen as the growth
temperature because the sharpest exothermic
peak was seen at this temperature in the DTA
analysis performed in oxygen. As shown in
Fig. 1.13, the LaAlO3 film exhibits c-axis
preferred orientation on a SrTiO3 (100) single
crystal substrate. The LaAlO3 peaks are
considered to correspond to (001), (002), and
(003) planes in a cubic structure. The figure in
the inset is the rocking curve scan of the (003)
LaAlO3 reflection and verifies the good c-axis
alignment with the FWHM determined to be
0.87°. Figure 1.14(a) is the scan made from
the (202) plane and indicated that the film had
good in plane texture with a FWHM = 1.07°. A
typical X-ray pole figure of the cubic (202)
reflections [Fig. 1.14(b)] shows a single cube-on-cube epitaxy: SrTiO3(001)||LaAlO3(001)cubic
and SrTiO3||LaAlO3cubic. In hexagonal
LaAlO3 coordinates, this is
SrTiO3||LaAlO3hex. A SEM
micrograph of a LaAlO3 on single crystal
SrTiO3 indicates that the grains are small,
surface roughness is low, and the film appears to
Other Synthesis Efforts
We have investigated other sol-gel routes
for preparing LaAlO3 precursor solutions using
lanthanum nitrate hexahydrate or lanthanum
acetate as the starting material instead of
relatively expensive lanthanum isopropoxide.
However, such attempts using these reactants
were not particularly successful so far, resulting
in either premature hydrolysis (nitrate route) or
failure to form a mixed-metal complex (acetate
route). We plan to focus now on identifying the
proper conditions to grow high-Jc
superconductors on these epitaxial LaAlO3
A LaAlO3 precursor solution has been
prepared by an all alkoxide sol-gel route. A gel
was formed upon complete hydrolysis of the
solution and yielded crystalline powders when
pyrolyzed in air beginning around 600°C. The
cubic lattice parameter of well-crystallized
LaAlO3 pyrolyzed at 800°C, a = 0.3796 ±
0.0001 nm, agrees quite well with the literature
value. Highly c-axis-oriented films of LaAlO3
[FWHM = 0.87° for the (003) plane] with good
in plane texture [FWHM = 1.07° for the (202)
plane] have been grown on SrTiO3 (100) single
crystals using a partially hydrolyzed solution
and firing in O2 at 800°C.
GROWTH OF BIAXIALLY TEXTURED BUFFER LAYERS ON ROLLED NI SUBSTRATES BY ELECTRON BEAM EVAPORATION
For conductors developed for high
temperature and high-field applications,
YBa2Cu3O7x (referred to as YBCO) deposits are
very promising. Also, Dimos et al.1,2 have
demonstrated from their YBCO bicrystal studies
that high Jc can only be obtained on oriented
YBCO films with a high degree of texture both
normal to and within the basal plane. Two
approaches have been used to deposit biaxially
textured YBCO films. The first approach was to
grow biaxially textured YSZ buffer layers on
polycrystalline Ni-based alloys such as Haynes
230 and Hastelloy C276 through an IBAD
process.36 A high Jc of over 1 × 106 A/cm2 at
75 K and zero field was obtained on 1-µm-thick
YBCO films on Ni-based alloys with textured
YSZ buffer layers grown by IBAD.7 The second
approach was developed at Oak Ridge National
Laboratory (ORNL) using the RABiTS8
concept. Our approach was to use
rolling-induced texture to obtain biaxially
textured face-centered cubic (fcc) metal strips
and to deposit epitaxially on the strips both
buffer layers and superconductors to form a
conductor. Ni was chosen as the substrate
because it readily develops the cube texture and
is more oxidation resistant than Cu. Recently,
we demonstrated that a Jc of 0.73 × 106 A/cm2 at
77 K and zero field can be obtained for films
with a layer sequence of YBCO/YSZ/CeO2/Ni.9
The crystallographic orientations for all the
layers were (100). In this architecture, all oxide
layers were grown by PLD. A clear need exists
to develop a capability to deposit buffer and
superconductor layers by other techniques.
These layers must be grown as continuous
epitaxial films on Ni. The purpose of the
chemical buffer layers is to retard oxidation of
Ni, to reduce the lattice mismatch between Ni
and YBCO, and to prevent diffusion of Ni into
YBCO. Previous studies proved that CeO2,1013
Pd,13,14 and YSZ15 films could be grown
epitaxially by vapor-deposition techniques on
single crystal substrates. In this paper, we
describe our successful development of the
growth of buffer layers on rolled Ni substrates
using an electron beam evaporation technique.
In related work, these buffer layer architectures
also have been developed using rf and dc
magnetron sputtering; those results are reported
separately.16 In this report, the deposition
conditions for growing epitaxial Pd, CeO2 and
YSZ films on textured Ni substrates by electron
beam evaporation are reported for the first time.
Experimental Results and Discussion
The cube (100) texture in Ni (99.99 %) was
produced by cold-rolling to over 90%
deformation followed by recrystallization at
temperatures ranging from 400 to 1000°C.8 The
thickness of the textured-Ni substrate used was
125 µm. The deposits were produced without
any substrate polishing in an AIRCO Temescal
CV-14 system with three electron guns. It was
operated by a Temescal FDC-8000 Film
Deposition Controller. The films were analyzed
by detailed XRD studies. A Philips Model
XRG3100 diffractometer with Cu Ka radiation
was used to record powder diffraction patterns.
For texture analysis, a Rigaku rotating-anode
X-ray generator was used, with a graphite
monochromator selecting Cu Ka radiation, and
slits defining a 2 × 2 mm incident beam. A four-circle diffractometer collected pole figures to
measure rocking curves ( scan) of (002) planes
of the (001) textured film, which measure
out-of-plane alignment. The diffractometer was
also used to measure scans of the (202)
planes. These scans indicate the in-plane
alignment of the film. SEM micrographs were
taken using a Hitachi S-4100 Field Emission
Scanning Electron Microscope. The beam
voltage used was 15 kV. The experimental
details for the growth of two-buffer-layer
architectures on rolled Ni substrates are
described in the following sections.
Growth of Pd on Rolled Ni Substrates by Electron Beam Evaporation
The as-rolled Ni substrates were cleaned
ultrasonically with both acetone and methanol
and were mounted on a substrate holder with a
heater assembly in the electron-beam system.
After the vacuum had reached 1 × 106 torr at
room temperature, the substrates were annealed
in situ at 400°C for 4 h. The temperature of the
substrate was measured using a thermocouple.
Then the Pd layer was grown on the textured Ni
at temperatures ranging from 100 to 500°C. The
typical deposition rate for Pd was between 0.5
and 1.0 nm/s at a pressure of 106 torr, and the
final thickness varied from 200 nm to 1 µm. The
thickness of the film was measured by a quartz
crystal monitor during the deposition. The -2
scan for a 400-nm-thick Pd film deposited on Ni
at 500°C showed the presence of a (001)-oriented film. From the and scans for 400-nm-thick Pd films deposited on Ni at 500°C, the
FWHM for Ni (002) and Pd (002) obtained were
7.3 and 4.1°, and that of Ni (202) and Pd (202)
were 8.8 and 7.4°, respectively. The XRD
results show that Pd can be deposited epitaxially
Growth of CeO2 on Pd-buffered Ni Substrates by Electron Beam Evaporation
| Fig. 1.15. The room-temperature powder XRD for a 100-nm-thick CeO2 layer deposited on Pd-buffered Ni at 400°C.
|Fig. 1.16. The and scans for a 100-nm-thick CeO2 layer deposited on Pd-buffered Ni at 400°C.|
Initially, biaxially textured Ni substrates
were cleaned and Pd films were grown on them
as discussed in the previous section. CeO2 films
were then deposited on the Pd-buffered Ni
substrates. After the vacuum in the chamber had
reached 1 × 106 torr at room temperature, a
mixture of 4% H2 and 96% Ar was introduced
until the pressure inside the chamber reached
~104 torr. The gas flow was controlled by a dc-powered piezoelectric valve. The Pd-buffered Ni
substrates were then annealed at ~600°C for
30 min at ~104 torr. After the substrates were
annealed, the chamber was maintained at a
pressure of 2 × 105 torr with a mixture of 4%
H2 and 96% Ar. The textured CeO2 layers were
grown on the Pd-buffered Ni at temperatures
ranging from 300 to 750°C. The deposition rate
for CeO2 was 0.1 nm/s. The final thickness
varied from 50 to 200 nm. Cerium metal was
used as the source. The crucibles used were
usually graphite. A -2 scan for a 100-nm-thick CeO2 film deposited on Pd-buffered Ni at
400°C is shown in Fig. 1.15. The strong CeO2
(200) lines show the presence of a good
out-of-plane texture. The oxygen impurity
present in the chamber was apparently enough
to oxidize the film to form stoichiometric CeO2.
Figure 1.16 shows the and scans for the
same film. The FWHM for Ni (002), Pd (002),
and CeO2 (002) are 6.6, 4.6, and 5.9°,
respectively. The rocking curves for Pd and
CeO2 are smooth because these are fine-grained
films. By contrast, the Ni substrate is
coarse-grained, so its rocking curves consist of
many sharp peaks corresponding to individual
grains. The FWHM from scans for Ni (202), Pd (202), and CeO2 (202) were 8.3,
7.8, and 10.0°, respectively. The XRD results demonstrate
that CeO2 and Pd can be deposited epitaxially on Ni.
Growth of CeO2 on Rolled Ni Substrates by Electron Beam Evaporation
The electron beam evaporation technique
was also used to deposit CeO2 films directly on
Ni. Biaxially textured Ni substrates were
mounted on a substrate holder that contained a
heater assembly. After the vacuum in the
chamber had reached 1 × 106 torr at room
temperature, a mixture of 4% H2 and 96% Ar
was introduced until the pressure inside the
chamber reached ~1 torr. The Ni substrates were
annealed at ~700°C for 60 min at ~1 torr.
During CeO2 deposition, the chamber was
maintained at a pressure of 2 × 105 torr with a
mixture of 4% H2 and 96% Ar. The textured
CeO2 layers were deposited on the Ni substrates
at temperatures ranging from 300 to 750°C. The
deposition rate for CeO2 was 0.1 nm/s, and the
final thickness varied from 5 to 150 nm. The
XRD results from the -2 scan as well as from the and scans for 100-nm-thick CeO2 films
deposited on Ni at 600°C revealed good
epitaxial texturing. The CeO2 (002) from the
-2 scan showed the presence of a good
out-of-plane texture. The use of a mixture of 4%
H2 and 96% Ar gas presumably prevents the
formation of NiO during the CeO2 growth.
These characterizations show that CeO2 can be
deposited epitaxially on Ni.
| Fig. 1.17. The room-temperature powder XRD for a 100-nm-thick YSZ layer deposited on 10-nm-thick CeO2-buffered Ni at 600°C.
| Fig. 1.18. The and scans for a 100-nm-thick YSZ deposited on 10-nm-thick CeO2-buffered Ni at 600°C.
| Fig. 1.19. The CeO2 (111) (top) and YSZ (202) (bottom) pole figures for a 100-nm-thick YSZ layer deposted on 10-nm-thick CeO2-buffered Ni at 600°C.
|Fig. 1.20. The SEM micrographs for (a) 50-nm-thick and (b) 100-nm-thick CeO2 films deposited directly on rolled Ni substrates. The micrographs were taken at 50-kX magnification.|
Growth of YSZ on CeO2-Buffered Ni Substrates by Electron Beam Evaporation
The electron beam evaporation technique
was used to deposit YSZ on CeO2-buffered Ni
substrates. Biaxially textured CeO2-buffered Ni
substrates were cleaned with methanol and were
mounted on a heated substrate holder in the
electron-beam system. After the vacuum in the
chamber had reached 1 × 106 torr at room
temperature, a gas mixture of 4% H2 and 96%
Ar was introduced until the pressure inside the
chamber reached ~1 torr. The CeO2-buffered Ni
substrates were annealed at ~700°C for 60 min
at that pressure. The chamber was then
maintained at a pressure of 2 × 105 torr with a
mixture of 4% H2 and 96% Ar. The textured
YSZ layers were grown on the CeO2-buffered
Ni substrates at temperatures ranging from 650
to 750°C. The YSZ deposition rate was
0.1 nm/s, and the final thickness varied from 50
to 150 nm. Yttria (10%) stabilized zirconia was
used as the source. The -2 scan is shown in Fig. 1.17 and the and scans are shown in
Fig. 1.18. These results were obtained on
100-nm-thick YSZ films grown at 600°C. The
strong YSZ (200) and CeO2 (200) peaks shown
in Fig. 1.17 indicate the presence of a good
out-of-plane texture. The FWHM for Ni (002),
CeO2 (002), and YSZ (002) were 7.4, 6.6, and
6.8°, and that of Ni (202), CeO2 (202), and YSZ
(202) are 9.5, 8.8, and 8.5°, respectively. As
shown in Fig. 1.19, the CeO2 (111) and YSZ
(202) pole figures demonstrate that the buffer
layers are epitaxial with a single orientation.
The XRD results show that YSZ can be grown
epitaxially on CeO2-buffered Ni substrates. To
date, our efforts to grow YSZ directly on rolled
Ni substrates have produced either randomly
oriented YSZ or (111)-oriented films.
Thickness Dependence and Crack Formation in CeO2 Layers
In our studies of CeO2 layers of various
thicknesses deposited on rolled Ni substrates,
we found that the as-grown 100-nm-thick CeO2
layers were cracked whereas 50-nm-thick CeO2
layers were crack-free. The SEM micrographs
that demonstrate these features are shown in Fig. 1.20. After the growth of YSZ on a
50-nm-thick CeO2 layer, the CeO2 layers were
also cracked but the YSZ layers were crack-free.
The presence of YSZ layers on top of CeO2
layers seems to alleviate the cracks that are
formed underneath. The CeO2 layer thickness
was found to be critical. Both CeO2 and YSZ
layers were crack-free for a CeO2 underlayer
thickness of <10 nm. Our SEM studies showed
that both CeO2 and YSZ layers were smooth and
continuous, demonstrating that a very thin layer
of CeO2 is needed to grow textured YSZ layers.
Efforts are being made to demonstrate the
growth of high-Jc YBCO films on these buffer
We developed two buffer layer architectures on textured-Ni substrates using the electron beam evaporation techniques. The two-buffer layer sequences are CeO2/Pd/Ni and YSZ/CeO2/Ni. The cube (100) texture in Ni was produced by cold-rolling followed by recrystallization at temperatures ranging from 400 to 1000°C. The CeO2 layer was deposited epitaxially on both Pd-buffered and textured Ni substrates. The YSZ layer was deposited epitaxially on CeO2-buffered Ni substrates. For thicker CeO2 films on Ni substrates, crack formation was observed. Growth of a 3- to 10-nm-thick CeO2 layer prevents crack formation and also assists the epitaxial growth of YSZ films.
Oxford Instruments, Inc., and ORNL have a CRADA that involves development of Ag/Bi-2212 conductors for high-field applications. Both dip-coated and powder-in-tube (PIT) conductors are being studied. The ORNL role in the project has involved evaluating batch-to-batch variations in the 2212 powder supplied by a commercial vendor and investigating difficulties encountered during process development.
Processing the Bi-2212 conductors involves partially melting the powders and slow cooling. Reproducible melting behavior is therefore a basic process requirement. Table 1.1 gives melting data for seven lots of commercial powders and nitrate pyrolysis material produced at ORNL. The variables include Ag content, carbon content, cation concentrations, and intra-lot uniformity. The samples were measured in two conditions, as received from the vendor and after a 750/835°C O2 decarburization heat treatment. Small (0.10.2 g) samples suitable for DTA/TGA were decarburized in flowing oxygen.
The endotherm measurements show that the two Ag-free commercial lots of Bi-2212 are nearly identical and the melting behavior was not altered by the decarburization heat treatment. Results for the Ag-containing powders were more variable, but data on decarburized commercial lots show that Ag lowers the melting temperature by about 15°C. This value is close to the 18 effect obtained by adding Ag2O to stoichiometric nitrate pyrolysis powder.
With a single exception (lot 5), decarburization produced an increase in melting temperature. It also reduced the inter-lot-variability which is a requirement for the conductor fabrication process. The most serious powder defect identified in this work is the 11°C intra-lot variation observed for lot 3. This lot was inhomogeneous both before and after decarburization. Intra-lot uniformity was considerably better for lots 1 and 4. Cation content also has an effect. The data for five different lots of decarburized Ag-containing powders had a range of 15°C, but part of this variation is caused by differences in the two values for lot 3. This decarburization procedure produces material with a carbon content of 500 to 700 ppm.
The results on melting behavior helped in identifying the source of a problem encountered with Bi-2212/Ag PIT conductors. Before processing, these tapes, which contained decarburized powder from lot 5, were heat treated in vacuum for 2 d at 650°C. This treatment decomposed some 2201 into Bi6Ca4O13 but did not prevent blistering during subsequent processing. Heating samples to increasing temperatures showed that the blisters began to form at ~800°C. Powder samples were removed from the Ag sheath and DTA/TGA data were obtained in O2 at temperatures up to 900°C. These results showed that the onset of melting fell between two values shown in Table 1.1. The samples also experienced a weight loss of about 0.4% between 700 and 810°. Because the small sample decarburized for the melting study did not lose weight in this temperature range, inadequate decarburization and CO2 release was suggested. This was confirmed by making measurements in another DTA/TGA that has an attached mass spectrometer. The CO2 content of the gas phase reached a maximum at ~800°C.
Dip-coated Bi-2212/silver conductors are also being developed for high-field applications. The conductors contain two layers of superconductor and three layers of silver. The Bi-2212 layers are directly exposed at an edge and this is required because the dip-coating process leaves residual organics in the superconductor layers. The residual organics are removed by oxidation at ~350°C. Experience has shown that removing the organics from coils causes property degradation that is not observed with single conductors. Some coils, heated to temperatures as low as 800°C, showed evidence of extensive melting.
The most obvious difference between a coil and a single conductor is that oxygen can reach the Bi-2212 in the single conductor both through the exposed edge and by diffusion through the outer silver layers. In a tightly wound coil the silver layer route is largely eliminated. To investigate the behavior in a low-oxygen environment, a Bi-2212/organic mixture was exposed to flowing CO2 in the DTA/TGA apparatus. A thermal event was noted at ~400°C, and the runs were terminated at 840°C. Post-test examination showed that the samples had melted, and XRD revealed that the Bi-2212 had completely decomposed. The decomposition products included elemental Bi and Cu, indicating that the decomposition involved reduction.
This experiment suggested that with a limited supply of oxygen, oxidation of the organics produces some CO, which in turn reduces Bi and Cu to the elemental state. The melting could then be attributed to eutectic formation; the Bi-Ag and Bi-Cu eutectics form at 262°C and 270°C, respectively.
Two methods for avoiding this problem
were investigated. Slowly heating samples to
400C in vacuum removed the organics, and
after a second 400°C cycle in air, XRD showed
that the product was phase pure Bi-2212.
Oxidation at 250°C also removed all of the
organics and a lower-temperature oxidation
would avoid eutectic formation.
LOCAL TEXTURE, CURRENT FLOW, AND SUPERCONDUCTIVE
TRANSPORT PROPERTIES OF Tl-1223 DEPOSITS ON PRACTICAL SUBSTRATES
In FY 1996, we extended quantitative investigations of the crystal grain orientations and electrical transport properties of HTS TlBa2Ca2Cu3O8+x (Tl-1223) deposits on polycrystalline substrates. The results confirm that in these systems current flow comprises percolative networks of strongly coupled material. In the following we describe how superconductive transport properties on different samples, on the same samples at different widths, and on samples with artificially induced strong flux pinning defects confirm the nature of current flow.
It has been demonstrated that Tl-1223 thick films can be synthesized by nitrate-spray pyrolysis of Ba2Ca2Cu3Ox precursors on polycrystalline YSZ or silver, followed by Tl-vapor-phase reaction in a two-zone furnace. These deposits can yield Tl-1223 with Jcs that exhibit strong behavior in magnetic fields, similar to those of high-Jc epitaxial thin films but with an overall reduced magnitude. The thick films have c-axis texture with (001) X-ray rocking curve FWHM of less than 2° but show no long-range in-plane texture using large-area XRD. However, microstructural grain-orientation mapping using electron beam backscattering patterns (EBSP) and X-ray microdiffraction has revealed the presence of local in-plane grain alignment extending over grain colonies of dimensions from 100 µm to 1 mm. These observations have suggested that current transport occurs through a network of percolative paths across adjacent colony boundaries. These paths select the low-angle grain boundaries of the overlapping grain orientation distributions of neighboring colonies.
This type of conduction mechanism should lead to transport properties that are characterized by parallel paths of weak-linked and well-coupled material. In substantial magnetic fields the weak-linked component is quenched, and the properties take on field dependencies that are characteristic of strong material occupying an overall reduced volume fraction. The following experimental results support this view and provide motivation for further development of Tl-1223 thick deposits for conductor applications.
In this work, precursors were deposited on polished polycrystalline YSZ substrates by aerosol spray pyrolysis of nitrate solutions with cation content Ba2Ca2Cu3Ag0.37. Typical thicknesses are 3 µm. Tl-1223 is formed by reaction of the precursor films with oxygen/Tl2O vapor in a two-zone furnace, with the sample zone maintained at 860°C and the Tl2O3 source zone at ~740°C. Reaction times are typically ~1 h, although the reaction occurs very rapidly (within 20 min). Electrical transport measurements were made on short samples (approximately 15 mm long and 8 mm wide). Electrical contact to the Tl-1223 surface was provided by sputtered gold pads. For studies of sample dimension effects, the samples were photolithographically patterned to 3-mm-long bridges of widths down to ~80 µm. Measurements of Jc were conducted in fields to 8 T, and Jc was defined at a criterion of 1 µV/cm.
Figure 1.21 illustrates the dependence of the transport Jc on magnetic fields oriented parallel to the c axis at 77 K. Data are shown for different Tl-1223/YSZ samples: (1) for two fully epitaxial thin films grown on single crystal LaAlO3 substrates and (2) for one sample that has been patterned to progressively narrower widths (although the all widths remained large compared with the colony dimensions). The logarithmic scale illustrates the common dependence of Jc on magnetic field; the data in field are similar aside from a multiplicative factor, even among different samples and despite large differences in absolute values. This observation is consistent with current flow occurring in a well-coupled component that occupies a reduced geometrical fraction of the sample. By regarding the epitaxial films with Jc(H = 0) > 1 MA/cm2 as fully connected benchmarks, one may infer that only 5 to 10% of the Tl-1223/YSZ material is active in a magnetic field, and for that component the Jc(H) characteristics are nearly identical among all samples.
|Fig. 1.21. Field dependence of Jc at 77 K for different TI-1223 samples and for a sample of different widths. The epitaxial films, deposited on single crystal LaAlO3 are ~1 µm thick and represents a fully connected benchmarks. The inset shows the same data, with Jc scaled by appropriate multiplicative factors.|
Further confirmation of the strongly coupled, percolative-path model of current flow is provided by the effects of artificially induced flux pinning defects. In previous work, we found that Tl-1223/YSZ samples irradiated parallel to the c axis with energetic heavy ions showed substantial increases in Jc and the irreversibility fields. In this case, it is reasonable that the effects arise from the columnar-defect-induced flux pinning in the well-coupled component and that this component is responsible for virtually all the loss-free current flow in substantial magnetic fields (we know of no mechanism whereby damage will raise the tunneling currents through weak-linked grain boundaries).
An as additional test of this current-path
model, we investigated the dependence of the
apparent Jc on lateral sample dimensions. This
effect arises because well-coupled current-path
options over long lengths are statistically
restricted if sample widths become comparable
to or smaller than the colony size. A quantitative
analysis of this phenomenon has been conducted
through both experiment and numerical
modeling. In the former case, the transport
Jc(H)s of two Tl-1223/YSZ samples were
measured as the samples were
patterned to progressively narrower bridge widths, ranging from 8 mm to 75 µm. The numerical modeling was based on a limiting-current-path analysis applied to the actual colony orientation distribution of one of the samples, measured by X-ray microdiffraction using a 100-µm-diam beam. Current transfer between adjacent colonies was calculated from the observed dependence of Jc on grain misorientation as determined in the epitaxial bicrystal film experiments of Nabatame et al.1 The model describes the effect as essentially a two-dimensional phenomenon, where c-axis currents are assumed to be negligible, in contrast to the view of the "brick-wall" model. Figure 1.22 illustrates the ratio of observed Jc to intra-grain Jc for different sample widths, compared with the numerical predictions at zero applied field. The measurements show qualitative agreement but fall somewhat below the numerical predictions. This effect might be explained by the relatively large size of the X-ray beam (comparable to some grain colony dimensions), which may overestimate the intra-colony Jc values within small colonies having a relatively large distribution of grain orientations.
|Fig. 1.22. The dependence of the apparent Jc on the width of two TI-1223/YSZ samples. The solid curve is a numerical calculation based on a limiting path model of current flow across basal plane tilt boundaries. The calculation assumes a fully connected Jc= 1 MA/cm2, consistent with the observed values of epitaxial films.|
|Fig. 1.23. Dependence of electric field on current density for two sample width regimes. (a) w = 4 mm > grain colony size; (b) w = 0.2 mm colony size.|
In addition to the decline of Jc at narrow widths, its overall field dependence and the accompanying current-voltage characteristics become qualitatively different. The latter is demonstrated in Fig. 1.23, where the field dependence at 77 K of electric field E on current density J is compared for examples in the two width regimes: w = 4 mm > colony size, and w = 0.2 mm < colony size. For the case w = 4 mm, the E(J) relations are qualitatively similar to those of the epitaxial film (negative curvature and power law at low fields, evolving toward the resistive thermally activated flux flow regime at high fields). For the case w = 0.2 mm at low and moderate fields the curves exhibit a constant linear differential resistance as dissipation sets in at currents just above the (low) Jc values. At high fields, the behavior again follows that of intra-grain dissipation. Both these observations are consistent with a picture of parallel conductive channels, one being a well-coupled component and the other being a grain-boundary-limited, weak-link channel. The differences in the overall character of Figs. 1.23(a) and (b) are simply determined by the relative fraction of each component. In Fig. 1.23(b), the dashed line represents a dominant weak component having a volume resistivity of ~8 × 109 cm.
We have shown that the transport properties
of Tl-1223 thick deposits are consistent with the
observed grain orientations of the as-formed
material on smooth, polycrystalline YSZ
surfaces. (Recently, these observations have
been extended to Tl-1223 deposits on
polycrystalline silver.) The growth
characteristics, which produce c-axis
perpendicular deposits, with large colonies of
grains having low-angle in-plane grain
boundaries, are necessary for well-coupled
current paths. Studies of the dependence of
Jc(H) and of the voltage-current characteristics
on lateral sample dimensions uphold a picture of
percolative, well-coupled current paths through
a fraction of the overall sample volume.
Although existing Jc values are not high enough
to provide practical conductor current levels,
large improvements could occur through
improved processing for enhanced colony
formation, or the epitaxial growth of Tl-1223 on
biaxially oriented tapes. These results provide
motivation for the further development of Tl-1223 thick deposits for conductor applications.
GROWTH OF TlBa2Ca2Cu3O9Y SUPERCONDUCTING FILMS WITH LOCAL BIAXIAL ALIGNMENT EXTENDING UP TO 5 MM ON SILVER
SUBSTRATES USING A SPRAY PYROLYSIS TECHNIQUE
For the development of long conductors
required for high-current electric-power
transmission lines, considerable progress has
been made using bismuth-based HTS materials
through PIT processing. At the same time, the
PIT tapes based on YBa2Cu3O7 (Y-123) and
TlBa2Ca2Cu3O9 (Tl-1223) superconductors were
usually randomly oriented and characteristically
weak-linked. However, recent developments in
deposited conductors based on Y-123 and
Tl-1223 materials are very promising. For
example, the measured Jc of 3-µm-thick Tl-1223
films on polycrystalline YSZ substrates were
3.3 × 105 A/cm2 at 77 K and zero field, and
>104 A/cm2 at 60 K and 1 T with the magnetic
field applied parallel to the c axis.1,2 Also, a high
Jc of more than 1 × 106 A/cm2 at 75 K and zero
field was obtained on 1-µm-thick Y-123 films
on Ni-based alloys with IBAD-grown textured
YSZ buffer layers.3 Similarly, a high Jc of more
than 3.0 × 105 A/cm2 at 77 K and zero field was
recently demonstrated on >1-µm-thick films on
RABiTS samples coated with biaxially textured
Pd, CeO2, and YSZ buffer layers.4 High-quality
1-µm-thick (Tl,Bi)(Sr1.6Ba0.4)Ca2Cu3O9y films
were also grown on LaAlO3 (100) single crystal
substrates.5 Based on these properties,
TlBa2Ca2Cu3O9y or (Tl,M)(Sr1.6Ba0.4)Ca2Cu3O9y
(M = Bi,Pb) (both compositions are referred to
as Tl-1223) superconductors may have great
technological importance for applications over
40 K whereas Y-123 deposited conductors may
even be useful at 77 K. However, there is a need
for making highly aligned materials because
high Jc values with good in-field properties were
mostly observed on aligned materials. Hence,
we initiated our work on the processing of
TlBa2Ca2Cu3O9y films on metallic substrates
using open geometry through a two-step
process. In the first step, Tl-free precursor films
were deposited by a spray-pyrolysis technique
followed by thallination in a two-zone furnace.
The spray-pyrolysis equipment is relatively easy
to build, and high-density precursor films can be
deposited easily. Deposition of high-density
deposits is one of the important requirements for
making Tl conductors. In this paper, we report
our successful development of TlBa2Ca2Cu3O9y
thick films using the spray-pyrolysis technique
on Ag substrates. We have grown Tl-1223 films
with local biaxial alignment up to 5 mm in the
plane of the film on smooth Ag substrates.
Initially, thallium-free precursor films with the composition Ba2Ca2Cu3Ag0.37O7 were deposited on Ag substrates using the spray pyrolysis technique. The addition of Ag0.37 (5 at. wt %) aided the liquid-phase formation of the Tl-1223 phase. The Ag substrates were either used as received or were obtained by cold-rolling. The typical size of the Ag substrates was 5-mm wide and 1-in. long. Some of the substrates were mechanically polished on a Vibromet 2 using a felt polishing cloth and 0.5-µm grit size diamond paste. The precursor spray solution of Ba, Ca, Cu, and Ag nitrates with a stoichiometry of Ba:Ca:Cu:Ag = 2:2:3:0.37 (5 at. wt % Ag in fully reacted powder) and a molarity of 0.18 based on total cation concentration was prepared by dissolving BaCO3, CaCO3, CuO and AgNO3 in conc. HNO3 and distilled water. The pH of the solution was approximately 1. The spray pyrolysis apparatus was contained in a Plexiglas glove box maintained at ~1 atm pressure in a constant flow of dry air. About 10 mL of the spray solution was taken in a glass syringe fitted with a Teflon plunger and a stainless steel needle. Teflon tubing was used to connect the syringe with a stainless steel/Ti ultrasonic spray nozzle. This nozzle was operated at 120 kHz and produced droplets of a median diameter of 18 µm. The nozzle was mounted on a uniaxial translation table that was operated in a bidirectional mode. Substrates were mounted on an Inconel heater block that could heat substrates up to 700°C. The distance between the substrate and the nozzle tip was approximately 1 cm. The temperature of the substrate, the atomization rate of the nozzle, and the travel rate of the nozzle over the substrate were controlled by a computer. The Ag substrates were mounted on the substrate heater block, which was initially maintained at 180°C. After the first coat, the substrate was heated up to 600°C to pyrolyze the film, and was held at 600°C for 5 min and then cooled to 275°C. The lower temperature of 180°C was used for the first coat to increase the wetting of the film to the substrate. After the second coat at 275°C, the substrate was again heated to 600°C and held at 600°C for 5 min and then cooled back to 275°C. This procedure was repeated until the desired film thickness of ~3 µm was obtained. The total deposition time was less than 30 min. During each coat, a 0.75-µm-thick film was deposited. The samples were then transferred into a tube furnace with flowing oxygen. The films were rapidly heated to 500°C at the rate of 200°C/min and then heated to 850°C at the rate of 12°C/min. At that point, the furnace was turned off and the samples were cooled to room temperature under flowing oxygen. The precursor films obtained were smooth and continuous. The powder XRD of the precursor film showed the presence of BaCuO2, CaO, and CuO.
Thallination of the precursor films was carried out using a two-zone thallination furnace, composed of a zone containing a Tl source (mostly Tl2O3 in a gold boat) and a sample zone in a flowing oxygen atmosphere, similar to that described by DeLuca et al.1,2 The sample-zone temperature was raised to 860°C while the source-zone temperature was raised to 600°C. After the sample temperature stabilized at 860°C, the source temperature was increased to a temperature between 725 and 730°C and was maintained at that temperature for 30 min. The partial pressure of thallous oxide (Tl2O) was calculated to be 1.1 × 103 atm at a Tl source temperature of 727°C in 1 atm oxygen. The furnace was then turned off and allowed to cool for a 3- to 4-h period with flowing oxygen. Quenching studies on similar samples have shown that the Tl-1223 phase grows from the substrate through a transient melting. The average thallium uptake was between 0.6 to 0.8 (relative to 1223), as determined from the weight gain of the sample. Previous studies have shown that a slight thallium deficiency is required to form single-phase Tl-1223.1,2 The resulting TlBa2Ca2Cu3O9y films were black and highly reflective. The film thickness was varied between 1 and 3 µm.
The films were analyzed by detailed XRD
studies. A Philips Model XRG3100
diffractometer with Cu Ka radiation was used to
record the powder diffraction pattern. A Rigaku
rotating anode X-ray generator was used for
and scans, with a graphite monochromator
selecting Cu Ka radiation, and slits defining a
2 × 2 mm incident beam. A four-circle
diffractometer was used to orient the 
textured film for diffraction from (1 0 14)
planes. Rotating analyzes the in-plane
alignment of the film. A 1 × 1 mm beam was
used to map local alignment, as described
earlier.6 The films were further characterized by
resistivity and transport Ic measurements using a
conventional four-probe technique. Either four
spring-loaded pin contacts or four Ag wire
contacts were made on top of the film. The
current leads were on the ends, and the voltage
leads were across the sample with a 0.5-cm
separation. Values of transport Jc were
calculated using a 1 µV/cm criterion. For
resistivity measurements, a constant current of
40 mA was used, and the temperature was
varied using a closed-cycle He refrigerator. For
Ic measurements, the sample was slowly inserted
into a liquid nitrogen container, which was
positioned inside a magnet capable of producing
a maximum field of 0.6 T. The magnetic field
was applied perpendicular to the plane of the
Results and Discussion
A typical room-temperature powder XRD pattern for a 1-µm-thick TlBa2Ca2Cu3O9y film formed on a Ag substrate is shown in Fig. 1.24. The diffraction pattern is characteristic of a predominantly c-axis-oriented Tl-1223 film. The addition of 5 at. wt % Ag in the precursor yielded a transient liquid phase in the presence of thallous oxide. The Tl-1223 phase has been shown to grow from the substrate up.7 The smooth Ag substrate apparently influenced the nucleation and/or kinetics of the Tl-1223 phase development. Atomic force microscope images taken on both smooth (polished) and rough (as-received or cold-rolled) Ag substrates were analyzed. The rough Ag had both smooth surfaces and large scratches along the rolling direction. The average surface roughness for the rough Ag is about 6 nm. The surface after polishing showed few irregularities, but a much smoother aspect (mirror-like finish) than for the rough Ag. The average surface roughness for the smooth Ag is about 2 nm. The Ag substrate had a (421) <210> biaxial texture. The rocking curves ( scan) for 1-µm- and 3-µm-thick TlBa2Ca2Cu3O9y films grown on smooth and rough Ag substrates, respectively, are shown in Fig. 1.25. The FWHM values for the (0010) c-axis peaks are 1.9° and 13° for Tl-1223 films grown on both smooth and rough Ag substrates. This clearly demonstrates that the smooth surface morphology is necessary for the substrates to achieve appreciable texture in the superconducting films. Thus, the substrate roughness dictates the nature of the films grown on it. The single scan of the (1 0 14) reflection for a 1-µm-thick TlBa2Ca2Cu3O9y film grown on a smooth Ag substrate is shown in Fig. 1.26. Also, a -scan contour map of the (1 0 14) reflection along the length of the film using a beam size of 1 × 1 mm is shown in Fig. 1.27. These results indicate the high degree of local in-plane alignment in the film for distances up to 5 mm. The colony microstructure was proposed earlier to explain similar features of local biaxial texture (the presence of small-angle boundaries) of Tl-1223 films with high current densities on polycrystalline YSZ substrates.7,8
Fig. 1.24. Room-temperature powder XRD pattern for a 1-µ-thick TlBa2Ca2Cu3O9-y film grown on a typical Ag substrate. The film shows preferential alignment along the c axis.
|Fig. 1.25. Rocking curves for 1- and 3-µ-thick TlBa2Ca2Cu3O9-y films grown on both smooth and rough Ag substrates, respectively, for a (0010) c axis orientation.|
Fig. 1.26. XRD scan of the (1 0 14) reflection for a 1-µ-thick TlBa2Ca2Cu3O9-y film grown on a smooth Ag substrates. The film shows preferential in-plane alignment.
Fig. 1.27. XRD scan of the (1 0 14) reflection along the length of the TlBa2Ca2Cu3O9-y films grown on a smooth Ag substrates using a beam size of a 1 × 1 mm. The film shows local in-plane alignment up to 5 mm.
Typical resistive transitions were fairly sharp for Tl-1223 films grown on both smooth and rough Ag substrates. The critical temperature (Tc), varied between 106 and 110 K. The magnetic field dependence of the Jc for 1- and 3-µm-thick TlBa2Ca2Cu3O9y films grown on smooth and rough Ag substrates, respectively, are shown in Fig. 1.28. The Jc values were 64,700 A/cm2 and 37,200 A/cm2 at 77 K and zero field for films grown on smooth and rough Ag substrates, respectively. For the film on the smooth Ag substrate, the Jc was measured within a single colony. At 0.5 T and 77 K, with the magnetic field applied perpendicular to the substrate, the Jc values were 9,500 A/cm2 and 3,200 A/cm2 for films grown on smooth and rough Ag substrates. This in-field behavior clearly shows the presence of high-quality Tl-1223 films on smooth Ag substrates. A typical SEM micrograph of a textured TlBa2Ca2Cu3O9y film grown on a smooth Ag substrate is shown in Fig. 1.29. This platelike morphology is similar to those observed previously for films grown on polycrystalline YSZ substrates.2
Fig. 1.28. The in-field dependence of Jc for H II c orientation of 1- and 3-µ-thick TlBa2Ca2Cu3O9-y films grown
Fig. 1.29. Typical SEM micrograph showing the platelike morphology of the TlBa2Ca2Cu3O9-y films grown on Ag substrates.
To conclude, we have demonstrated that
TlBa2Ca2Cu3O9y thick films grown on smooth
Ag substrates by a spray-pyrolysis technique
followed by thallination in a two-zone furnace
had local biaxial alignment up to 5 mm in the
plane of the film. High Jc values with good
in-field properties were obtained for these films.
It is not clear whether large colonies are
desirable from a Jc standpoint; however, the
surface morphology of the Ag substrates appears
to strongly influence the texture development in
the Tl-1223 films.
EFFECT OF Mg-DOPING ON THE PROCESSING AND CURRENT
CAPACITY OF Bi-2223 POWDER-IN-TUBE CONDUCTORS
Since the discovery of HTS, extensive
research effort has been devoted toward the
development of practical HTS conductors. To
date, PIT (Bi,Pb)2Sr2Ca2Cu3O10 (Bi-2223) tape
is one of only two proven HTS conductors that
can be fabricated in long lengths. Unfortunately,
even though these PIT tapes can be utilized at
77 K under self-field condition, the HTS
becomes nonsuperconducting in the presence of
a moderate magnetic field (0.5 T). Therefore, to
expand the range of applications of these tapes,
ways need to be developed to improved flux
pinning in these materials.
It has been shown in the melt-textured Y1Ba2Cu3O7 HTS system that flux pinning can be enhanced through the incorporation of inclusions.1 Even though the exact mechanism of flux pinning in this HTS is still unclear, it is generally accepted that the size of the inclusions must be submicron and that the dislocation density should be high. Because of the similarities between various HTS cuprates, the same beneficial effects should also be found in the Bi-2223 HTS as long as the inclusions do not react adversely with the compound. We reported last year on our effort in the investigation of Mg doping of Bi-2223 aerosol precursor powders. The initial study revealed that in sintered Bi-2223 with Mg doping, MgO particles of 50 to 200 nm were homogeneously distributed throughout the sintered HTS. In addition, the HTS/MgO interfacial regions contain a high density of dislocations, which is favorable for enhancement in flux pinning. Moreover, ac susceptibility measurements showed that there is little or no degradation in the superconducting transition temperature for up to about 10% of MgO doping.
Although no degradation was measured,
energy-dispersive spectroscopy (EDS) results
showed that the MgO particles contain a small
amount of copper. This indicates that unless the
dissolved Cu is compensated for, a possibility of
degradation exists in the superconducting
properties as a result of the tendency to form
low-copper-content secondary phases rather
than the HTS Bi-2223 phase. In this study,
aerosol precursor powders with various degrees
of Cu compensation were prepared, and PIT
Mg-doped Bi-2223 conductors were fabricated.
We will report on the effect of Mg-doping on
PIT processing parameters and current
capacity. In addition, the microstructure and
inhomogeneous MgO distribution within the
conductor are presented, and potential
solutions to this problem are discussed.
Aerosol precursors of 5 vol. % Mg-doped
Bi-2223 (Mg0.77) with various amount of Cu
compensation were prepared and are listed in
Table 1.2. These powders were packed into Ag
tubes, which were then swaged and drawn into
wires of about 1 mm overall diameter.
Following the initial drawing process, these
wires were rolled by smooth rolls to about
200 µm overall thickness at 6 µm reduction per
pass. After initial deformation, the
monofilamentary tapes were subjected to the
following thermomechanical treatment:
HT1 = 12 h, P1, HT2 = 12 h, P2, HT3 = 26 h,
P3, HT4 = 50 h with heat treatments (HT)
carried out at 810, 818, 825, and 832°C in 7.5%
oxygen--92.5% argon, and pressing (P)
performed at 1.5 GPa between polished steel
platens with 5-min holding time. Critical current
density measurements were performed on fully
processed tapes using the 1-µV/cm criterion.
After the Jc measurements, longitudinal cross
sections of selected tapes were polished and
were examined by SEM.
Results and Discussion
Variations in Jc with processing temperature of undoped and Mg-doped Bi-2223 with various amount of Cu compensation are shown in Figs. 1.30 through 1.33. It can be seen from Fig. 1.30 that the Jc of undoped Bi-2223 conductors (EX46) initially increases with sintering temperature (Ts), reaching a maximum value of 27,100 A/cm2 at Ts = 825°C, and decreases as the sintering temperature is further increased. Both the existence of an optimum processing temperature and the value of this temperature under a 7.5% oxygen processing atmosphere are in excellent agreement with previous studies. On the other hand, Figs. 1.31, 1.32, and 1.33 show that the Jc of Mg-doped Bi-2223 conductors increase monotonically with processing temperature, and the optimum processing temperature has not been reached at Ts of 832°C. This increase in processing temperature is analogous to that of Bi-2212 with MgO inclusions, where studies2,3 have shown that the optimum processing temperature was increased by 10 to 20°C. The maximum Jc attained by Mg-doped Bi-2223 conductors (EX48) investigated in this study is 28,200 A/cm2, a value that is only slightly higher than that of the undoped conductor. Unfortunately, the PIT tapes were exhausted and processing could not be performed at higher temperature to determine the optimum sintering temperature of Mg-doped conductors.
Fig. 1.30. Variation in Jc with processing temperature
of undoped (EX46) PIT Bi-2223 conductors.
| Fig. 1.31. Variation in Jc with processing temperature
of Mg-doped (EX47) PIT Bi-2223 conductors.
Fig. 1.32. Variation in Jc with processing temperature
Fig. 1.33. Variation in Jc with processing temperature
Figure 1.34 shows the magnetic field dependency of Ic of undoped (EX46, 27,100 A/cm2) and Mg-doped (EX48, 28,200 A/cm2) conductors with similar Jc. It can be seen from the normalized Ic comparison that flux pinning has not improved in the Mg-doped PIT. This lack of pinning enhancement may be caused by a number to factors, such as lower Tc, excessive MgO particle size, insufficient doping, or inhomogeneous distribution of MgO inclusions. Transport resistivity measurements and detailed microstructural examination were performed to determine the reason for the lack of flux pinning improvement. Figure 1.35 shows the variations in normalized resistance of undoped and Mg-doped PIT conductors; both the onset and zero-resistance transition temperatures of the two tapes are identical, and the lack of pinning enhancement is not caused by a lowering of Tc through Mg doping.
Figure 1.36 shows a SEM image of the interior of a Mg-doped PIT conductor; the MgO inclusions are less than 1 µm in diameter with a mean separation distance of about 600 nm. Although the question as to whether sufficient doping is available for pinning cannot be answered, the figure shows that submicron particles are present within the HTS core, which should provide a high density of dislocation for pinning. On the other hand, detailed examination of the HTS/Ag-sheath interfacial areas revealed that MgO particles are rarely seen within 4 µm of the interface (Fig. 1.37). Numerous reports have shown that the bulk of the supercurrent in Bi-2223 PIT conductors is typically transported via this well-aligned interfacial region. If this is the case, then the lack of MgO particles within this well-aligned region can certainly lead to zero flux pinning enhancement in the Mg-doped PIT conductors.
Fig. 1.34. Comparison of Ic field dependency of
undoped and Mg-doped PIT Bi-2223 conductors.
| Fig. 1.35. Variations in normalized resistance of
undoped and Mg-doped PIT Bi-2223 conductors.
Fig. 1.36. SEM image of the interior of the HTS core of Mg-doped (EX48) PIT Bi-2223 conductor.
Fig. 1.37. Bi-2223/Ag sheath interfacial region of the Mg-doped PIT conductor. Note the lack of MgO inclusions
The observation that no MgO particle is incorporated within the growing Bi-2223 grains from the HTS/Ag-sheath interface has significant implication in the effort of flux pinning enhancement of Bi-2223 PIT conductors. That is, other types of dopants in the form of inclusions may also be pushed away from the well-aligned interfacial region. The problem of particle pushing has been studied during the solidification of simple systems.46 Whether the particles are entrapped within a growing matrix or are simply pushed along the growth front has been shown to depend on the critical velocity of the system, which in turn is affected by numerous factors. Among these are particle size, interfacial energies, liquid viscosity, relative densities, interface curvature, and thermal conductivity. For example, particle entrapment is favored if the particle size is increased. This, however, is contrary to what is required for pinning enhancement in HTS (i.e., the inclusions have to be as small as possible). Other dominant parameters such as liquid viscosity and interfacial energies may be altered by critically adjusting the characteristics of the liquid phase. One possibility is to dope the precursor powder with Ag, which may alter the viscosity of the liquid as well as provide an abundance of liquid throughout the bulk of the HTS core. Another possibility is by drastically reducing the HTS core thickness such that the liquid phase is not confined to the HTS/Ag-sheath interface but is available throughout the entire core.
Preliminary work on these two potential solutions has been initiated at ORNL. Aerosol precursor powder with 10% Mg and 5% Ag doping was prepared, and was enclosed within a Ag foil in the form of a "folded" tape. In addition, PIT Bi-2223 with 5% Mg doping was carefully rolled to an overall tape thickness of 60 µm. These two tapes were subjected to the following thermomechanical treatment: HT1 = 12 h, P1, HT2 = 88 h with heat treatments (HT) carried out at 837°C in air, and pressing (P) performed between polished steel platens with 5-min holding time (0.2 GPa for folded tape and 1 Gpa for PIT tape). Polished longitudinal cross sections of the folded tape with Mg and Ag doping were examined by SEM. Silver inclusions with a particle size of about 1 µm were found to be distributed throughout the HTS core. More importantly, MgO particles could be seen at or near the HTS/Ag-sheath interface (Fig. 1.38). Similarly, MgO particles can be found near the HTS/Ag sheath interface in the PIT with reduced core thickness. Figure 1.39 shows the SEM image of the fracture surface of a peeled PIT tape. This Bi-2223 fracture surface represents a 30° tilted plan view of the HTS roughly 0.2 to 0.5 µm from the HTS/Ag sheath interface. A sizable number of MgO particles (<0.5 µm) were found to be embedded within the well-aligned Bi-2223 grains, whereas the holes are believed to be prior locations of MgO particles that were torn away during tape peeling. From these preliminary results, it appears that the problem of lack of MgO particles at the well-aligned HTS/Ag sheath region may be circumvented by a combination of Ag doping and/or fabricating fine-filament PITs (such as that found in multifilamentary conductors). Effort is now under way to fabricate Mg- and Ag-doped fine-filament PITs to examine the microstructural and superconducting characteristics.
Fig. 1.38. Bi-2223/Ag sheath interfacial region of Mg-
and Ag-doped Bi-2223 sample. MgO particles can routinely
be seen at the interface.
Fig. 1.39. Fracture surface of the Bi-2223/Ag sheath interfacial region of a tape that has been rolled to about 60 µ thick. MgO particles are embedded within the well-
aligned Bi-2223 grains at the interfacial region.
EFFECTS OF INITIAL COLD-WORK CONDITIONS ON THE DEFORMATION
AND CURRENT CAPACITY OF MONOFILAMENTARY AND MULTIFILAMENTARY Bi-2223 CONDUCTORS
Among the various processing techniques used in HTS fabrication, the PIT method is currently one of the most promising for applications on an industrial scale. To obtain high Jc in PIT (Bi,Pb)2Sr2Ca2Cu3O10 (Bi-2223) conductors, the tapes have to be subjected to repeated heating and deformation cycles (thermomechanical treatment). While many of the factors affecting the final superconducting properties of the conductors are still not clear, numerous experimental results have shown that thermomechanical treatment leads to the following important characteristics: (1) a high fraction of Bi-2223 phase, (2) a high degree of texture, and (3) a high final density of the HTS core.15 Even though precise scheduling and mode of deformation utilized in the thermomechanical treatment are extremely important, there is now ample evidence that the final properties of the conductor depend on every fabrication step, from precursor powder selection3,6,7 to the cooling rate following the final heat treatment step.8,9
Because final properties strongly depend on previous processing history, deformation conditions and the resultant characteristics of the precursor core even before the initiation of thermomechanical treatment can significantly influence the superconducting characteristics of fully processed Bi-2223 conductors. For example, although the Jc of a Bi-2223 PIT conductor increases with decreasing core thickness, work instability can occur when the initial reduction by cold rolling is large, leading to variations in core layer thickness (sausaging) and an abrupt decrease in Jc.10 In addition, studies have shown that with an increase in either the initial packing density11,12 or the core density by careful cold work,12 both the final core density and Jc are enhanced. The final density of the tape, which has been shown to be intimately related to current-carrying capability,13,14 can be improved by initial deformation despite the occurrence of retrograde densification during heat treatment.
Typically, a Bi-2223 PIT conductor is
fabricated by packing precursor powder into a
silver tube, which is then swaged, drawn, and
rolled at room temperature into tape form.
Ideally, it is desirable for the powder-packed Ag
tube to deform homogeneously through the
various working stages into a tape with uniform
microstructure and properties. However,
because of constraint factors such as tool-stock
configuration or complexity of the stock shape,
plastic deformation is typically inhomogeneous,
leading to localized instability and flow that
result in nonuniform tape characteristics.
Friction has long been known to be one of the
most important constraint factors in plastic
deformation. An increase in working friction
usually leads to effects such as inhomogeneous
deformation, increased deformation resistance,
and shearing stresses beneath the contact
surfaces. Consequently, friction between the
tools and the work piece will undoubtedly have
a significant effect on the extent of deformation
as well as the core homogeneity of PIT tapes.
Moreover, the influences of working friction on
the deformation of mono- and multifilamentary
tapes are not expected to be identical because of
the difference in core geometry. We have
previously reported on the effects of friction and
reduction per pass on the inhomogeneous
deformation of mono- and multifilamentary PIT
Bi-2223 tapes, and we have shown the existence
of dead zones in the conductors. In this report,
the current-carrying capability of fully
processed tapes will be correlated with initial
deformation conditions, and the implication of
these conditions on the design and processing of
PIT tapes will be discussed.
Experimental Details: Monofilamentary Tapes
In the fabrication of monofilamentary PIT tapes, two precursor powders having different Pb contents with the overall compositions of Bi1.84Pb0.32Sr1.91Ca2.03Cu3.06Ox (powder A) and Bi1.84Pb0.42Sr1.91Ca2.03Cu3.06Ox (powder B) were produced by an aerosol process. The detailed description of the aerosol process is given elsewhere.15 After treating these precursor powders at 700°C for 4 h in air to obtain the desired phases, the powders were packed into Ag tubes having a 6.25-mm outer diameter and a 0.75-mm wall thickness. Powder packing was carried out in a dry box with the aid of a hand press, and the typical packing density was approximately 35% of theoretical. The openings of the Ag tubes were plugged and then closed by swaging, and the wires were drawn to 0.91 mm overall diameter at 50 to 100 µm reduction per pass.
The monofilamentary wires were initially rolled to 245 µm overall thickness at 6 µm reduction in thickness per pass (increasing from 0.7% to 2.4% per pass with successive reductions) using a 4-in.-diam rolling mill with an exit velocity of 25 mm/s. Following initial reduction, deformation by rolling under normal-friction (smooth rolls) and high-friction conditions (roughened rolls; a portion of the roll was roughened by grinding the roll faces with a 60-grit SiC paper) was carried out by deforming the two monofilamentary wires down to about 150 µm overall thickness, again at 6 µm per pass (increasing from 0.7% to 3.8% per pass with successive reductions) with an exit velocity of 25 mm/s (Table 1.3). Sections of the 150-µm-thick tapes were retained for thermomechanical treatment and Jc measurements while the remaining tapes were deformed down to a thickness of approximately 50 µm to detect the onset of work instability.
Experimental Details: Multifilamentary Tapes
The multifilamentary wire used in this investigation was fabricated at American Superconductor Corporation (ASC). The precursor powder was packed into Ag tubes, and the tubes were drawn into monofilamentary wires; 85 of these wires were packed into a Ag tube and then drawn to 1.3 mm diam. Sections of these wires were deformed to approximately 290 µm overall tape thickness between smooth (normal friction) and roughened (increased friction) rolls at 6 µm reduction per pass (increasing from 0.5% to 2.1% per pass with successive reductions). In addition to the different rolling surfaces, deformation was also performed at 20% reduction per pass so that the effect of strain rate on powder flow could be examined. The various deformation conditions of multifilamentary samples investigated in this study are summarized in Table 1.4.
To examine the evolution of precursor core
microstructure, small sections were cut from the
various mono- and multifilamentary tapes at
different stages of rolling and were examined by
optical microscopy. At the conclusion of the
initial cold-rolling procedure, the overall
thickness and width of individual tapes were
measured with the aid of an optical microscope
equipped with micrometer-controlled x-y stage. These tapes were then subjected to
thermomechanical treatment, and the Jc of each
tape was determined at 77 K using the 1-µV/cm
Results and Discussion: Monofilamentary Tapes
In the FY 1995 annual report we showed that decreased friction under uniaxial pressing conditions resulted in crack initiation and propagation in monofilamentary tapes. The ideal case would be zero interface velocity for high-friction coefficient. However, as the friction coefficient between Ag sheath and loading fixtures decreases, the ductile Ag at the interface is free to flow. Consequently, the hard and brittle ceramic core of nonuniform density can flow inhomogeneously, and cracks appear perpendicular to the elongation direction at regions of lower density. Under increasing load, the cracks will continue to propagate until the Ag sheath is sheared between the high-density core sections, and any further spreading of the tape is accomplished by widening of the cracks. In the case of rolling, we have shown that under normal smooth-rolling conditions, sausaging became noticeable when the core thickness was reduced to about 25 µm. The appearance of sausaging has been explained by Osamura et al.10 in terms of work instability, and the core thickness where instability occurs as determined in this study is in excellent agreement with the results of those investigators. Furthermore, when lubrication was provided by a soap coating and there was not front or back tension, deformation of the precursor core was found to be extremely nonuniform and sausaging occurred at an earlier stage where the core thickness was still in excess of 50 µm. As the tape thickness was reduced further, the degree of sausaging became progressively worse. In addition, transverse cracks started to appear in the Ag sheath, which corresponded to a periodic appearance of the core.
It appeared that an increase in deformation friction may provide an avenue toward a more homogeneous and denser core in monofilamentary tapes. This is in contrast to traditional rolling of metals and alloys, during which lubrication is employed to reduce the power requirement of the machinery and to facilitate metal flow under the action of the load. When the microstructures of the tapes initially deformed between roughened rolls were examined, neither the general appearance of the precursor cores nor the onset of work instability could be distinguished from those deformed between smooth rolls. The only geometrical difference that can be observed is in the greater amount of lateral spread experienced by the tapes deformed between roughened rolls. As seen in Table 1.3, the width of the tape containing powder A deformed between roughened rolls is larger than that of the smooth rolls even though the former tape is not yet as thin as the latter one, a situation which is consistent with an increase in rolling friction.16
Following initial deformation, the monofilamentary tapes were subjected to the following thermomechanical treatment: HT1 = 25 h, P1, HT2 = 25 h, P2, HT3 = 25 h, P3, HT4 = 50 h with heat treatments (HT) carried out at 825°C in 7.5% oxygen--92.5% argon, and pressing (P) performed at 1.5 Gpa between polished steel platens with a 5-min holding time. Pressing instead of rolling was selected in the thermomechanical treatment to minimize the occurrence of sausaging, which would have invalidated any correlation between initial deformation conditions and final superconducting properties. In addition to the difference in lateral spreading, Jcs of the tapes deformed between smooth and roughened rolls were found to differ throughout the thermomechanical treatment. The variation in Jc with accumulated processing time for powders A and B are shown in Figs. 1.40 and 1.41, respectively. Regardless of the tape composition, the Jcs of the tapes that were initially deformed by roughened rolls are consistently higher than those processed by smooth rolling surfaces.
|Fig. 1.40. Variation in Jc with total sintering time of monofilamentary Bi-2223 PIT tapes containing powder A initially deformed between smooth and roughened rolls.|
|Fig. 1.41. Variation in Jc with total sintering time of monofilamentary Bi-2223 PIT tapes containing powder B initially deformed between smooth and roughened rolls.|
| Fig. 1.42. Optical micrographs of the transverse cross sections of multifilamentary Bi-2223 PIT tapes initially deformed by cold rolling to roughly 600 µm under the following conditions: (a) smooth rolls, 6 µm reduction per pass; (b) smooth rolls, 20% reduction per pass; (c) roughened rolls, 6 µm reduction per pass; (d) roughened rolls, 20% reduction per pass (bar = 200 µm).
|Fig. 1.43. Schematic of the transverse cross section of a metal cylinder deformed by simple compression at room temperature under high-friction condition. The cone-shaped dead zones at the top and bottom of the sample are separated from the deformed bulk by narrow regions of high deformation.|
Results and Discussion: Multifilamentary Tapes
Transverse cross sections of the four sets of multifilamentary tapes at various stages of initial deformation were examined by optical microscopy; Fig. 1.42 shows the filament distribution when these tapes were reduced to a thickness of about 600 µm. It is fortunate that, after wet polishing with Al2O3 and lapping oil, the degree of initial core densification in these non-heat treated multifilamentary tapes can be determined qualitatively: gray areas represent denser precursor material; black areas represent porous cores. This relative difference in powder density combined with individual core geometry provide a means to probe the internal deformation characteristics of Bi-2223 PIT tapes. The contrasts in Fig. 1.42 indicate that the mechanical deformation of these composite tapes is inhomogeneous. Moreover, the degree of inhomogeneity increases with friction (from smooth rolls to roughened rolls) and strain rate (from small to large reduction per pass). In fact, the deformation pattern of these tapes closely resembles that of the "dead metal zones," or "dead zones" frequently encountered in metal forming. A cross-sectional schematic of the deformation pattern of a cylinder deformed in simple compression under high friction condition is shown in Fig. 1.43. Because of friction between the tools and the work piece, the material adjacent to these areas (region I) remains almost stationary. These dead metal zones undergo little or no deformation, and have been shown to increase in size with increasing friction, although material at the free surface (region III) deforms as a result of axial compression and secondary hoop tension, resulting in a bulged surface. With increasing frictional constraint, the tensile strain becomes larger while the compressive strain decreases, resulting in increased bulge severity. The most severe deformation is concentrated in regions just outside the dead zones near each contact surface (region II) in the form of shear bands.17 These shear bands are initiated along the velocity discontinuities of the slip-line (zero-extension-direction) field and typically appear as two complementary bands forming an X-shaped region in the deforming sample.
In deformation by rolling, the main variables that control the process are (1) the friction between the rolls and the sample, (2) the roll diameter, (3) the deformation resistance of the sample, and (4) the presence of front and back tensions.16 The friction condition is particularly important because it varies with roll diameter, changes the flow stress by increasing the constraint factor, and determines the amount of front or back tension needed for significant reduction in rolling load.18 In addition to affecting the other processing variables, friction is important in rolling because it pulls the metal into the rolls, and it influences the magnitude and distribution of the roll pressure. Typically, theoretical as well as experimental analyses of rolling deformation are concerned with the forces acting perpendicular to the rolled surfaces and along the rolling direction (i.e., a plane strain condition is assumed and lateral spreading is ignored). From those investigations, the presence of a friction hill in the rolling direction with its peak location corresponding to the neutral point has been well established.18 In particular, the friction hills that were developed by deformation between smooth and roughened rolls have been measured.19 From these results, it is now known that with roughened rolls, the rolling load will be greater, the roll face pressure will be higher, and the neutral point will move toward the entrance of the rolls.
Although deformation along the longitudinal direction is extremely important, different points of a rolled tape will be subjected to the same loading in this direction because these points have to pass through the identical friction-hill profile at various times. Consequently, techniques such as front or back tension can be employed to alter the roll pressure such that the onset of work instability can be delayed. The deformation pattern exhibited in the transverse cross sections of the multifilamentary tapes, however, clearly demonstrates that the plane strain assumption is incorrect in the case of PIT (finite width) rolling because deformation varies from filament to filament. We have already seen from Fig. 1.42 that as either the working friction or the strain rate is increased, areas of limited deformation (i.e., the dead zones) penetrate deeper toward the interior of the tape. Typically, the formation of these dead zones occurs early in the deformation process, and the dead zones tend to not deform at all until the overall reduction is sufficient to bring them into contact. Because the amount of overall tape reduction is dictated by the roller gap, higher strain in the tapes is localized to within the central portion along the diagonals (region II in Fig. 1.43), and the extent of this flow localization becomes more severe with higher friction because of the larger dead zones. Figure 1.42 clearly shows that under the conditions of increased friction or strain rate, filaments at the central portion of the tapes are highly densified, become thinner, and elongate more in the lateral direction. In contrast, the filaments near the tape surfaces remain thick and porous and are subjected to little or no deformation. Moreover, the filaments at the edges of the tapes experience only limited densification because these regions (region III in Fig. 1.43) are not subjected to direct compression. Instead, the material at the edges spreads and elongates under the influence of a secondary tensile stress system generated by the adjacent elastic material outside the roll gap.17
The different extent of lateral spreading can also be seen in the overall dimension of the various tapes. As listed in Table 1.4, the lateral spreading of the tapes deformed between both roughened and smooth rolls increases with the reduction per pass. In addition, the width of the tape deformed between roughened rolls at a small reduction per pass is equal to that deformed between smooth rolls even though it has been reduced by a lesser amount along the thickness direction. In other words, lateral spread is greater for the roughened rolls under these deformation conditions. On the other hand, lateral spread of the tape deformed between roughen rolls at a large reduction per pass is less than that of the smooth rolls. One possible reason may be that, at a small reduction per pass, the influence of strain rate on the flow stress and strain hardening of Ag is minimal, and lateral spreading is dominated by the friction effect. In the case of a large reduction per pass, the flow stress of Ag is likely to increase from both strain rate sensitivity and strain hardening. Under this condition, a higher load can be transferred to the powder core because the deformation is not necessarily localized to the previously ductile Ag. Thus, multifilamentary tapes deformed between smooth rolls at a large reduction per pass should be deformed more homogeneously but to a larger extent because of the limited size of the dead zones. Moreover, longitudinal sections of this tape revealed that the degree of sausaging is much less than expected even though the reduction per pass is large and the filament thickness is less than 10 µm, a value which is lower than the typical onset of work instability. This lower-than-expected sausaging is believed to be caused by a combination of the increased flow stress and strain hardening of the Ag, interaction between the ceramic filaments, and the low rolling speed employed,20 which has been shown to reduce sausaging.
Following the microstructural examination, 30-mm sections of the various multifilamentary tapes were subjected to the following thermomechanical treatment HT1 = 5 h, P1, HT2 = 5 h, P2, HT3 = 40 h, P3, HT4 = 50 h. Critical current densities of these tapes are listed in Table 1.4. The Jc values indicate that unlike monofilamentary tapes, initial deformation of a multifilamentary PIT tape under high-friction conditions does not necessarily result in the best current capacity. Rather, Jc values of multifilamentary PIT tapes depend on the densification and homogeneity of the cores. This information may be elucidated from the changes in the width-to-thickness ratio (w/t) as well as from the fill factor; a larger ratio and a lower fill factor correspond to better densification. The variation in Jcs with the changes in width-to-thickness ratio is shown in Fig. 1.44. It can be seen from this figure that Jc increases with the absolute ratio value; i.e., Jc increases with the core density. Figure 1.45 shows Jc as functions of fill factor before the initial heat treatment (as deformed) and after the final heat treatment (fully processed). In both cases, the current capacity increases monotonically with decreasing fill factor, again indicating that Jc is enhanced with improved core density. In addition, the consistency of the ordering of fill factors both before and after heat treatment reaffirms that the beneficial effect of high initial core density will remain despite the occurrence of retrograde sintering.
|Fig. 1.44. Variation in Jc of multifilamentary Bi-2223 PIT tapes with the changes in width-to-thickness ratio.|
|Fig. 1.45. Critical current density of multifilamentary Bi-2223 PIT tapes as functions of tape fill factor before initial heat treatment and after final heat treatment.|
|Fig. 1.46. Schematic of dead zones within (a) monofilamentary and (b) multifilamentary tapes deformed under high friction condition. The single core in monofilamentary tape is expected to lie outside the dead zones, whereas a significant number of filaments in the multifilamentary tape will be located within the areas of limited deformation.|
The different responses of mono- and multifilamentary tapes to friction can be explained (disregarding the influence of strain rate) with the aid of dead-zone development as shown in Fig. 1.46. The precursor core in a monofilamentary tape is located within the central region of the tape. Consequently, even though an increase in working friction leads to larger dead zones, substantial portions of these zones of limited deformation are located within the silver sheath. Therefore, the strain is localized to the precursor and results in higher initial core density, which has been shown to lead to high final core density and high Jc.12 This result is analogous to that of Utsunomiya et al.21 who found higher Jc in tapes that were deformed between rolls of larger diameter. In their study, rolling was employed as the deformation process in the thermomechanical treatment. The effect on the partially densified HTS core in their study, however, should be similar to the influence on the unreacted precursor investigated in this study. As pointed out by Dieter,16 a large roll diameter results in a larger area of contact and therefore greater frictional forces. In contrast, the same investigators reported that monofilamentary tapes rolled with mineral oil as lubricant exhibited higher Jcs than those of unlubricated samples.21 This discrepancy may be caused by rolling during thermomechanical treatment in that particular experiment or by the utilization of front tension in their work, a technique that minimizes slipping and moves the neutral point toward the roll entrance.18 As mentioned previously, increased working friction has the same effect on the neutral point. In deformation by rolling, the stock at the entry side of the neutral point tends to be pushed back out of the rolls, whereas it is extruded from the rolls at increasing speed at the exit side. Consequently, the partially adhered brittle core suffers a reversal of flow direction at the neutral point, which may lead to core fracture at the low-density sections. By moving the neutral point forward or even to the roll entrance, the compression zone is lengthened, and partial separation of the core material may be able to be healed under the longer compression zone of the roll gap.
The core in monofilamentary PIT tape is
confined to the central region of the conductor;
however, precursor filaments in
multifilamentary tapes are distributed
throughout the entire cross section. Thus an
increase in friction means that a larger number
of filaments will reside within the dead zones
and will experience only limited deformation.
Because these filaments are porous, important
characteristics such as phase purity, c-axis
texture and grain-to-grain connectivity will
remain inferior after heat treatment and will not
be able to contribute significantly to the current-carrying capability of the tapes. Consequently,
deformation under high friction in
multifilamentary tapes of the present geometry
does not lead to enhancement in Jc. In fact, the
highest Jc in the multifilamentary tapes
examined in this study is exhibited by the PIT
tape initially deformed between smooth rolls at
a large reduction per pass. As previously
mentioned, a combination of limited dead zones
and high strain rate in this tape resulted in the
largest amount of lateral spreading and therefore
thin, dense cores and the most homogeneous
deformation characteristics. Consequently, most
of the filaments in this tape should be able to
carry supercurrent resulting in the highest Jc.
Interestingly, the Jcs of these multifilamentary
PIT samples increase with lateral spreading, as
shown in Fig. 1.44. In fact, the highest Jc values
are exhibited by the tapes with the largest
amount of lateral spreading in both mono- and
multifilamentary configurations, which may be a
characteristic to strive for in the continuing
refinement of the deformation processing of
Bi-2223 PIT tapes.
The effects of initial cold-work conditions on the deformation of Bi-2223 PIT tapes have been studied in this investigation. The results revealed that deformation of PIT conductors by rolling is three-dimensional (plane strain condition is not followed), and the transverse cross sections of the rolled tapes resemble those of samples deformed under a simple compression condition (i.e., dead zones develop). In monofilamentary tapes, the precursor powder core is located in the central region of the tape and is therefore outside the areas of limited deformation. Under high-friction conditions, most of the deformation is sustained by the monocore, which leads to higher core density and enhanced Jc. In contrast, many filaments in multifilamentary tapes are located within the dead zones. Therefore, when the working friction is increased, the dead zones penetrate deeper into the tape and envelop more filaments. Because these filaments sustain little or no deformation, fewer filaments are active in current transport and high friction does not necessarily result in the highest Jc in multifilamentary tapes. Instead, the optimum Jc found for the deformation conditions utilized in this study is by deformation between smooth rolls at a large reduction per pass. It is believed that the combination of reduced dead zones and increased flow stress result in large but more uniform deformation of the multifilamentary tape.
It is interesting, however, that although only
a limited number of filaments are expected to
transport supercurrent in the tape deformed
under high-friction and large reduction-per-pass
conditions, the Jc is only slightly lower than that
of the tape with the most homogeneous
deformation. This result indicates the
importance of high initial core density as well as
the benefit of thin filament cores and suggests
various strategies that may increase the overall
performance of multifilamentary tapes. One
possible modification is the utilization of a thick
outer sheath so that the filaments are located
outside the dead zones under high-friction
conditions. By doing so, the higher strain
limited to the central region can be experienced
by most of the filaments. An opposite approach
may involve the utilization of lubricant to limit
the dead zone size. This strategy, however, will
need to be accompanied by other modifications,
such as the imposition of front and back tension
to minimize the occurrence of sausaging.