Ag tapes

PHYSlCA ELSEVIER

Physica C 278 (1997) 1-10

Processing factors for high critical current density in Bi-2223/Ag tapes J.O. Willis *, R.D. Ray II, J.F. Bingert, D.S. Phillips, R.J. Beckman, M.G. Smith, R.J. Sebring, P.A. Smith, B.L. Bingham, J.Y. Coulter, D.E. Peterson Superconductivity Technology Center, Materials Science and Technology Dioision, MS-K763, Los Alamos National Laboratory, Los Alamos, NM 87545, USA Received 20 December 1996; revised 4 March 1997

Abstract

Ten factors relating to Bi-2223/Ag tape production and thermomechanical processing have been examined in a statistically designed study to determine optimal parameter values for high critical current density Jc in rolled tapes. The study indicates that low fill factor, narrow tapes processed for long times yield the best J¢ values, and that strain rate, as was expected, is not a significant factor. The major microstructural conclusion is that macrocracking, even at the early stages of tape processing, should be avoided, as these cracks cannot easily be healed. © 1997 Elsevier Science B.V. PACS: 74.72.Hs; 74.60.Jg; 85.25.Kx; 81.40.Ef Keywords: Bi-2223/Ag tapes; Critical current density; Thermomechanical processing

1. Introduction The production of high critical current density Jc conductors by the oxide-powder-in-tube technique employing the Bi-2223 phase superconductor is a complex process involving interactions among many different parameters. These include such factors as precursor powder stoichiometry, phase assemblage, particle size, packing density, deformation path to produce tape, thermal processing atmosphere, temperature and time of sintering, and type and number of intermediate deformations. There have been a number of earlier investigations of some of the parameters mentioned above.

* Corresponding author.

Gherardi and Caracino [1] investigated the effects of six parameters (powder source, stoichiometry, wire diameter before rolling, thickness of Bi-2223 core after rolling, annealing temperature and annealing time) on Bi-2223/Ag tape processing. They concluded that only four of the six factors were significant, with powder source being most significant, and that there were several significant interactions. Our group reported on a study [2] with precursor powder and first and second sinter temperatures as the three factors and using pressing for the intermediate mechanical deformation. Several powders yielded nearly equivalent Jc values but with different sensitivities to the sintering temperatures, that is, there were interactions among the variables. Lehndorff et al. [3] reported on an optimization study for Jc by independently varying the sintering temperature and the

0921-4534/97/$17.00 Copyright © 1997 Elsevier Science B.V. All rights reserved. PH S 0 9 2 1 - 4 5 3 4 ( 9 7 ) 0 0 109-3

2

J.O. Willis et al./Physica C 278 (1997) 1-10

number of intermediate deformations (both pressing and rolling) at fixed 50 h sintering times. This study concluded that the optimum values of Jc could be achieved with two rolling deformations and 150 h total sintering time or three presses and 200 h total sintering time at a sintering temperature of 840°C in air. Korzekwa et al. [4] investigated the effects of deformation path (pressing or rolling) and the effect of roll size on crack patterns. This study concluded that during rolling, the stress state tends to result in more transverse cracking compared to that during pressing. Greater geometric uniformity of the core can be achieved with small diameter rolls during the initial reduction of round wire to flat tape, but the details depend on friction, reduction per pass and other factors. This work was subsequently extended by Schoenfeld et al. [5] to investigate theoretically the effects of these various process variables in addition to those of sheath and core hardness and sheath work hardening rate. Fischer et al. [6] investigated the effects of partial pressure of oxygen ( p O 2) and that of roll diameter for intermediate deformation steps. They found that as the p O 2 was decreased, the rate of conversion from Bi-2212 and other phases to Bi-2223 increased. Larger diameter rolls for the intermediate deformation result in a 40% increase in Jc values. To better understand which factors are most important, and to examine interactions among these, we have designed and carried out an experiment to study the effects of ten of these process variables relating to the properties of the tube geometry and packing procedure, the deformation from tube to tape, and

the subsequent thermomechanical processing steps (time, temperature and size of intermediate reductions produced by rolling) leading to the production of Bi-2223 silver sheathed tapes. We have also examined the effects of the process variables on such material properties as phase purity, grain alignment and core morphology in the final tape. The rest of this paper is organized as follows. In Section 2, we motivate the factors selected for study, describe the statistical design of the experiment and finally detail the implementation of the experiment. In Section 3 we describe some of the results at intermediate stages of tape processing, followed by the statistical analysis of the critical current density performance at the final stage and end with a discussion of the microstructure and some of the possible mechanisms for the Jc dependencies observed. In Section 4, we draw some conclusions from this study and suggest future work.

2. Experimental 2.1. E x p e r i m e n t a l d e s i g n

A complete analysis of the production process of Bi-2223/Ag tapes requires the inclusion of a large number of factors, many of which interact with other factors. Because the primary emphasis of this study is a better understanding and optimization of deformation processing, many of these factors, especially those related to powder processing, have to be held fixed. Practical considerations require that the results

Table 1 Factors and levels for Bi-2223/Ag tape study Factor Level 0 Pack Density(P) Core Thickness(CT) Strain Path (by roll diam)(SP) Strain Rate (SR) Temperature 1 (T1) Time 1 (tl) Reduction 1 (RI) Temperature 2 ( + 3) (T2) Time 2 (t2) Reduction 2 (R2)

Hand Pack Thin (20% FF) Narrow Tape _< 102/s 810°C 48h 5% 810°C ~h 5%

m

817°C 15% 817°C 15%

+ CIP Thick (60% FF) Wide Tape 103//s 825°C 96 h 25% 825°C 96 h 25%

J.O. Willis et al./Physica C 278 (1997) 1-10

be scalable to longer lengths; therefore, we use rolling, rather than pressing, for all deformation processing in the study. Because a secondary goal of the study is to determine thermal process values, we include time and temperature factors for the thermal treatment steps. The result is a set of 10 factors of two or three levels each, as shown in Table 1. The output variable of the study is the critical current density Jc in self field at 75 K. Secondary output for selected samples is the core morphology, phase fraction and grain alignment. The packing density can be varied by either cold isostatically pressing (CIP) or hand packing the powder into the Ag tube. The core thickness can be varied by adjusting the initial fill factor of the Ag tube. The strain path can be varied by using large diameter rolls, resulting in wide tape, or small diameter rolls, yielding narrow tape. The strain rate can be varied by controlling the roll speed. In the first heat treatment, Temperature 1 is set for Time 1. Then the tape is rolled by Reduction 1, based on the presinter thickness. It is heat treated again at Temperature 2 and Time 2, then rolled again by Reduction 2. Finally, it is heat treated a third time at Temperature 2 and 96 h. A full factorial study as shown in Table 1 would require more than 5000 different experiments to be performed. However, personnel and time constraints restrict this type of study to 100-200 tapes. Therefore, we adopt a fractional factorial type design based on the expected importance of the factors, on our interest in determining the effect of a factor, and

3

on the expected interactions. Table 2 lists this input to the design of the experiment. The Level of interest rankings again denote that the main interest of this study is in the tape fabrication and deformation processing factors. On the basis of the information in Tables 1 and 2, the experiment was configured as a fractional factorial and split plot design. This reduces the number of experiments from a full factorial study of > 5000 different conditions to only 178 experiments, including 58 for intermediate results after the first two heat treatment steps and replicates for analyzing the statistical significance of the results. The design uses 'blocking' of the factors into three sets: (1) tube fabrication: Pack Density, Core Thickness (4 tubes + 1 replicate); (2) processing to tape: Strain Path, Strain Rate (all 4 conditions/tube); and (3) subsequent processing: Temperature 1, Time 1, Reduction 1, Temperature 2, Time 2, Reduction 2 in a complex design. Intermediate results at the first (RS) ~ and second (RS) 2 roll-sinter stages were to come from the replicate tube, and these tapes all have the same four tube fabrication and tape processing factor levels, namely Packing Density (Hand Packing), Core Thickness (Thick), Strain Path (Narrow Tape) and Strain Rate (Low). There are strengths and weaknesses for all fractional factorial designs. This one allows for estimates of all the single factor effects and all of the interactions in Table 2 except for tl x t2 andT1 x R1, which is partially estimable, and where x denotes an interaction. However, all higher order interactions are 'confounded,' e.g. SR x T1 X R2 = CT x R1

Table 2 Expected impact on response and level of interest for the factors (1 = highest) and the expected interactions among the factors Factor

Expected impact on response

Level of interest

P CT SP SR T1 tl R1 T2 t2 R2

9 3 8 10 1 6 4 2 7 5

4 2 5 6 7 7 1 7 7 3

Expected interactions

P

CT

SP

SR

T1

tl

R1

T2

t2

R2

X

X

X

X

X

X

X

X

X

X X

X

X

X

X

X

X X

X

X

X X X

X X

X

4

J.O. Willis et al./Physica C 278 (1997) 1-10

particular higher order interaction is not unique and may actually be the result of an entirely different interaction (see for example [7]). The major advantage of the design is the drastic decrease in the number of experiments required.

2.2. Implementation of the design

I 0o=o

2.0 105

× T2. In other words, the apparent source of any

,.--"""-

E

60 ~ v

> 1.0 10 s

40 _~.

Differential..-' \

v°'uT/i

_-- 5.0 104

~ u

The powder, of nominal BiLsPbo.4Sr2Ca2Cu30, stoichiometry, was prepared using a variation of the two-component method [8] by blending Bit.sPbo.4Sr2CalCU20 x and sol-gel synthesized 'CaCuO 2' as described in detail in Ref. [9]. Commercial Bi-2212 powder (stoichiometry BiLsPb0.4Sr2CaCu20x) was fired for 24 h each in 10% O2/Ar at 800°C and 810°C with intermediate grindings. A powder with nominal stoichiometry CaCuO 2 was prepared by solgel techniques to obtain the phase assemblage C a 0 . 4 5 C u 0 . 5 5 0 2 - F CaO. These were milled together to obtain mixed powder of nominal stoichiometry Bi 1.8Pbo.aSr2 C a 2 C u 30.r. The mixed two-component powder was analyzed for stoichiometry, density and particle size distribution. Two independent ICP measurements on the powder (nominal stoichiometry BiLsPb0.aSr2Ca2Cu 3 and normalized to cation sum = 9.2) yielded results of 1.72 : 0.38 : 1.94 : 2.09 : 3.07 and 1.707:0.4: 1.942 :2.065:3.097 in very good agreement with each other and slightly deficient in Bi relative to the nominal value. The pycnometric density of 6.409 g / c m 3 is within the range of other powders of this composition that we have measured. The mean particle size is 5.1 i~m (see Fig. 1) and is dominated by the Bi-2212 platy phase; independent measurements of the second phase mixture indicate particle sizes from submicron to about 2 txm with soft agglomerates to significantly larger sizes. The cumulative fraction curve in the figure indicates that 90% by volume of the particles are smaller than 10 p,m. The size distribution is exactly what was desired for this study with a small mean particle size but with a range of sizes in order to promote good powder deformation during tape processing. Silver tubes were hand-packed with powder at ~ 40% theoretical density or filled with cold isostatically pressed (CIP'd), machined powder cores of ~ 80% theoretical density, and then sealed into the

b

"~.,.'"Bi-2212+ "CaCuOx" - 80 ~" ,..~ MeanDiam = 5.1 lam

"~---- - / /

1.5 105

--

o

L.

o

20 ~" 3

mulative:/.= L F ~ ~

0.0 10°

0 2

4

6 8 10 20 Particle Diameter (p.m)

30

Fig. 1. Particle size distribution for the mixed two powder precursor used in the experiment.

Ag tubes by electron-beam welding of the Ag plugs. The tubes were drawn from 12.7 mm to 1.25 mm diameter through a series of 50 dies. Depending on the strain path, tapes were rolled with 3.75 cm or 40.64 cm diameter rolls to produce Narrow or Wide tapes, respectively. All tapes were rolled to 145 Ixm thick at less than 10% reduction per pass at strain rates that gradually approached the target values from below. Finally, tapes were sectioned into 15 cm long samples for further thermomechanical processing. Samples were fired three times in 3-zone tube furnaces in flowing 10 -t- 0 . 5 % 0 2 - 9 0 % Ar ( p O 2 60 Tort). The central ~ 30 cm zone of the furnaces was adjusted to + I°C along its length. The sinter temperatures were based on values optimized previously by Smith et al. [9]. Subsequent deformation reductions R1 and R2 were based on presinter tape thicknesses and were all performed using the small (3,75 cm) diameter rolls. The final sinter time was fixed at 96 h. Finally, the critical current I c was determined using the standard 1 ixV/cm electric-field criterion for 2 regions per tape and 2 replicates per treatment condition. Optical microscopy was used to determine cross section areas for calculation of Jc and for determination of core morphology, phase fraction and grain alignment. 3. Results and discussion

3.1. Intermediate stage results The as-drawn wires were, as designed, all the same diameter and had two distinct fill factors (21

J.O. Willis et al./Physica C 278 (1997) 1-10

Fig. 2. As rolled wires that will result in tapes with (left) thin cores and (right) thick cores.

and 43%), Fig. 2. Therefore, the change in densification path between the low density hand-packed and high density CIP'd cores had been calculated well in the experimental design. One problem in the experimental design is also found here, however. The two powder ingredients were not as well mixed as desired in the as-drawn wire, with the 'CaCuO 2' additive clumped on a ~ 20 ~zm scale. In terms of subsequent processing, this means that the kinetics of forming the Bi-2223 phase would suffer from the need to transport material over unnecessarily long distances to attain the needed stoichiometry. The fill factors of the as-rolled tapes do not change significantly during tape rolling, although the

5

variance in the tape measurements is larger as a result of measurement sensitivity to edge-rounding on the thickness. All of the samples rolled along the positive strain path (large lateral spreading), regardless of wire packing or strain rate, show inclined shear cracking in longitudinal section (Fig. 3, bottom). These cracks are frequently observed in transverse sections as well (Fig. 3, center), suggesting that at least the later stages of deformation along the positive strain path approximate uniaxial compression. After cracking, the HTS core blocks are prone to localized shear along the fault zones. Optical micrographs of tapes after the first heat treatment are shown in Fig. 4 for those that had Time 1 + ( = 96 h). It is observed that Bi-2223 grain growth and the size of second phase (lighter color) particles increase with increasing sintering temperature. The alignment is better at the higher temperatures but is accompanied by some grain growth into the silver sheath. These microstructure results are corroborated by the superconductivity measurements: the low field susceptibilities of the tapes show sharper transitions at the higher temperatures, and the transport I c values are higher. However, this does not necessarily indicate that Temperature 1 + is the optimum factor level for the fully processed tape. Statistical analysis of the data after the first sinter

Core Thickness Strain Path Core Thickness + Strain P a t h Strain Path + Strain R a t e Strain Path + Strain Rate + Strain Path + Strain Rate + (Long. Sect.) Fig. 3. As rolled tapes for several combinations of core thickness, strain path and strain rate. The top four images are all transverse sections; the bottom is a longitudinal section.

6

J.O. Willis et al./Physica C 278 (1997) 1-10

confidence level in the other conclusions drawn from the study. 3.2. Final stage results

Fig. 4. Transverse section of tapes after the first sinter for 96 h at (a) 810°C, (b) 817°C and (c) 825°C. Both Bi-2223 and second phase grain growth and grain growth into the silver sheath increase with sintering temperature.

(RS) 1 stage shows a strong main effect for both Temperature 1 and a slightly weaker effect for Time 1 with performance increasing with both increasing temperature and time. This response is expected for tapes in this early state of processing because both these parameters correlate positively with the degree of phase conversion from Bi-2212 to Bi-2223 and thus with Jc at 75 K. At the (RS) 2 stage the strongest main effects are Temperature 1 and Temperature 2, and for both of these factors the performance improves as the temperature is lowered. In addition, there is a strong interaction between Temperature 1 and Reduction 1; the best performance is achieved for T1 0 and R1 0; next best is T1 - and R1 + . The results of statistical analysis of Jc as a function of serial position along the original tape length for tapes from the replicate tube after the first sinter (RS) 1 and after the second sinter (RS) 2 show no systematic variation in Jc along the length of the tape. These tapes all came from the same hand packed, thick core tube made into a narrow tape at the low strain rate. These results thus raise the

The dependencies observed earlier for the (RS) l and (RS) 2 stages are further modified at the final stage (RS) 3. Fig. 5 summarizes the statistical analysis results for the four tube- and tape-fabrication factors by means of 'Box' plots. As noted earlier, two pieces of tape were subjected to each of the factor levels in the design, and Jc was determined on two sections of each tape, giving a total of four independent values of Jc. These values were then averaged to give one Jc value per condition. Each variable in the plots shown in Fig. 5 (and also in Fig. 7 below) represents the Jc values of a minimum of 39 different treatment conditions (factor levels), and thus at least 156 independent determinations of Jc. The Strain Path and Core Thickness (SP and CT) are significant factors, that is, there is a strong main effect in that, statistically, narrow (SP - ) tapes performed better than wide (SP + ) tapes and thin core (CT - ) tapes perform better than thick core (CT + ) tapes. The best results ( ~ 10 k A / c m 2) were obtained for Strain Path - and Core Thickness - . Two significant interactions also occur among these 11~

1(~

Narrow Wide Strain Path

i!

Thin Thick Core Thickness

!o: Hand Packing

CIP

Low

Strain Rate High

Fig. 5. Jc data for tape production variables. For each variable, the bar across the box is the median. The ends of the boxes are the 25th and 75th percentile. The solid lines run to the last data point that is within 1.5 times the length of the box from the 25th or 75th percentile. The points outside these limits are "known as 'outliers' or extreme data points and are plotted individually.

J.O. Willis et al./Physica C 278 (1997) 1-10

7

8[ .... --

Thin --

....

.

--

_

~

Thin Thick

....

5

--

Thin -- Thick

.=.

~4~

~4~ ~3

--

Thick

.°

-..: 2

11

Hand CIP Packing

Narrow Wide Strain Path

Small Medium Large Reduction 2 6

....

5

~

-"'..

Thin

.... Thin -- -- Thick

5

--Thick

.... Thin -- -- Thick

5

°

• ,

4

°

-° %.

°" "°°"

%.

3

,, %.s

t~

1

2

2

2

1

Low Medium High Temperature 1

'~

Low Medium High Temperature 2

SmalIMediumLarge Reduction 1

Fig. 6. Transport critical current density Jc as a function of factor values illustrating the interaction of the Core Thickness with the Strain Path, Packing, Reduction 2, and Temperature 1 variables. An interaction is when the response of one variable depends strongly on the level of another variable.

four tube- and tape-fabrication factors: Packing × Core Thickness and Strain Path × Core Thickness. These can be seen in the upper left of Fig. 6; in both cases, Jc has the opposite dependence for thick

10

10 o

8 6

(CT + ) and thin (CT - ) cores. The best results are for a thin core ( C T - ) and the poorest for a thick core (CT + ) that had been CIP'd. Strain Rate is found to not be a significant variable. Packing den-

o

10

o

~_8

8

o

8 v

2 0

Low Medium High Temperature 1

10

8

2

2

o

SmallMediumLarge Reduction 1 10

10

8

~_8

56

6

2 0

Low Medium High Temperature 2

2 Small Medium Large Reduction 2

0

Low High Time 1

0

Low High Time 2

Fig. 7. Jc data for within tape variables. Box plots are described in the caption of Fig. 5.

8

J.O. Willis et al./Physica C 278 (1997) 1-10 2.0 E

%Narrow ow

1.5

t~ "" 1.0 O ~ "

Main Effects of Factors on Jc

yhin h~,, Low Small "~.,-L~ Hand-Pack

0.5

O=

and Std. Dev. for (RS) 3 Tape

~-....

.¢

*'""

o.o

7=~l.~°W_Sma II Short

,- "'"

-'~~~

cO -0.5

[

i o.

•-

I

[

i

~

1

o.

"o

.~

~

i

o

I.-

~

.~

i

i

I

-o

Factor co

Fig. 8. Main effects o f the factors on the critical current density Jc and the standard deviation o- for fully processed (RS) 3 tape.

sity/method did not change the results except for the interaction with Core Thickness. For the within-tape factors (thermomechanical processing), reductions RI and R2 could be achieved within + 2 % of target values. Temperature 1 and Reduction 1 (T1 and R1) are found to be significant variables (Fig. 7). High T1 clearly degrades Jc, and the best values are obtained for R1 at + or values. There are also many interactions observed in this part of the study, and many of these involved the core thickness. The significant interactions are with Reduction 2 and Temperature 1 (see Fig. 6), in addition to the Strain Path and Packing sensitivities noted earlier. The largest response for all of these factors is for a thin Core. There are also significant interactions for R1 × R2, T2 × R2, T1 × T2, and T1 × R1 (not shown). Fig. 8 shows the main effects of each factor on both arc and the standard deviation or of (RS) 3processed tapes. The main effect as calculated here is equal to the sum of the responses (either Jc or or) for the ( ' + ' level minus ' - ' l e v e l ) / ( 1 / 2 the total number of runs). Note that the designation of ' + ' and ' - ' levels is arbitrary, and the analysis shown in Fig. 8 takes advantage of this to keep all J~ main effects positive. The designations next to the J~ data points reflect the ' + ' level designation for that factor, i.e. the level that resulted in the higher Jc. The middle level was ignored for the 3-level factors.

The o- main effect measures the influence of the level of a factor on the variance of Jc response. Most of these values scale with the Jc main effect. For example, the value of + 0.42 for the main effect of Strain Path on or signifies that although the narrow path results in higher Jc responses, this set of responses also contains more variance than those with lower J~ values (wide path). The most significant exceptions occur with: (1) Temperature 2 and Packing, where improved Jc response is associated with less variance, and (2) Reduction 2, where a modest effect on J¢ for the small level is associated with relatively larger variances. The analysis of Fig. 8 also emphasizes the most significant responses to the factors: narrow tape, a low Temperature 1, thin core, and low Temperature 2 appear to be the most important. Finally, it must be emphasized that because of the complexity of the physical processes taking place in the experiment, interactions must not be neglected in interpreting the results. In particular, while Strain Path - (Narrow) is the largest main effect shown in Fig, 8, this value is an average over both thick and thin core tapes. The upper left plot of Fig. 6 shows a very strong SP × CT interaction. Because of this interaction, the main effect would be very small considering Strain Path + (Wide tapes) only. It is only because of the very strong dependence of Jc on Strain Path for Narrow tapes that there is a net strong main effect as shown in Fig. 8.

3.3. Microstructural observations The major second phases observed in the 12% of the fully processed tapes that were examined by optical microscopy are CuO and (Cal_xSrx)2CuO 3, the 2:1 alkaline earth cuprate phase. Ca2PbO 4 may also be present, but the grains are usually too small to see by optical microscopy. In some of the more fully reacted CT + (thick core) tapes the second phase is predominantly CuO of grain size 1-2 i.zm and just a few percent of the core by volume. However, for other tapes that might either not be as fully reacted, or that may have reacted more rapidly because they have CT - (thin cores) or were reacted at higher temperature, etc., both CuO and (Cal_.,.Sr.,.)2CuO 3 are present in roughly equal proportions, and the grain size and phase fraction may

J.O. Willis et al. / Physica C 278 (1997) 1-10

be as large as 10 Ixm and greater than 5%, respectively. Some of these larger second phase grains might have been eliminated or reduced in size with better mixing of the two powder components before tube loading. Mechanical mechanisms for some of the observations have also been investigated by optical microscopy. Prior examination of as-rolled tapes showed gross shear-cracking in both longitudinal and transverse sections of all 'wide' tapes, and no such cracking in any 'narrow' tape. This is a likely consequence of the difference in the amount of longitudinal strain for the two different roll diameters used to generate the different strain paths during the initial processing of wire to tape. There is significantly more longitudinal strain for wide tapes, made using large diameter rolls, which tend to promote transverse cracks. It is natural to hypothesize that such cracks never fully heal throughout the subsequent process cycle, so that examination of any finished 'wide' tape should show some logical current limiting defect remaining in the prior crack positions. This examination was performed and showed coarse, mistextured grains to decorate the prior crack planes, often with remnant cracking associated (Fig. 9). We conclude that these cracks are the reasons for the low current densities of the wide strain path tapes. Parrell et al. [10] have recently used magneto-optic imaging to examine crack patterns in pressed and rolled Bi-2223/Ag tapes, and also find that residual transverse cracks in rolled tapes lead to degradation in Jc.

9

4. Summary and conclusions The experiment confirmed many of the expected main effects, specifically, the response of the tape production factors Core Thickness, Packing, Strain Path and Strain Rate; this was one of the main goals of the study. Strain Path and Core Thickness (SP and CT) are significant variables. Strain Rate, as anticipated, is not a significant variable. The best results are obtained when the Strain Path is Narrow and the Core Thickness is Thin. Significant interactions occurred for Packing × Core Thickness and Strain Path x Core Thickness. The best results here are for a Thin Core and the poorest for a Thick Core that had been CIP'd. Temperature 1 and Reduction 1 (T1 and R1) are found to be significant variables for the within tape factors. Furthermore, the quadratic responses of Reduction 1 and Reduction 2 were not expected. There are a significant number of interactions in this part of the experiment as well. The major microstructural finding is that macrocracking of the core during the initial deformation to tape appears to be unrecoverable and should be avoided at all costs. A secondary finding is that good initial mixing of the powders is important to control second phase grain size. In addition, times that are too long or temperatures too high may also contribute to second phase grain growth. Based on the factor levels that gave the best performance in this study, future experiments can now be designed with fixed Core Thickness (Thin),

(a) As-rolled, narrow strain path

--~ Post (RS) 3 processed, Jc = 8.7 kA/cm 2

(b) As-rolled, wide strain path

--~ Post (RS) 3 processed, Jc = 2.0 kA/cm 2

Fig. 9. As rolled and fully processed narrow (a) and wide (b) tapes. Note the extensive cracking in (b) which never fully heals after processing and is believed to be the primary reason for the difference in performance for tapes processed with these different strain paths.

10

J.O. Willis et al. / ehysiaa C 37~ {lqqT) !-1o

Packing (either, with hand packing slightly better), Strain Path (Narrow), and Strain Rate (either, with slow values slightly better), and Time 1 and Time 2 can be tentatively fixed near the long value. The tape production factor levels are expected to be relatively independent of the powder employed whereas the details of the within tape processing are likely to be highly powder dependent, Thus the Time, Temperature and Reduction factors may need to be reevaluated for each new powder examined.

Acknowledgements This work was performed under the auspices of the United States Department of Energy, Office of Energy Management. We thank the members of the BSCCO Wire Development Group, including Argonne, Los Alamos, and Oak Ridge National Laboratories, the National Institute of Standards and Tech. nology-Gaithersburg, the University of Wisconsin and American Superconductor Corporation, for fruitful discussions during the course of this work.

References [1] L. Gherardi, P. Caracino, in: H.W. Weber (Ed.), Proceedings Of the 7th International Workshop on Critical Currents in ~Jl~p~rconductors, World Scientific, Singapore, 1995, p. 545. [2] J,O, Willis~ I~..P. Ray II, D.S. Phillips, K.V. Salazar, J.F. BirLgert, ,I,I~, Brem~er, D.E. Peterson, J. Electron. Mater. 24 (1995) 1760, [3] B. Lehndorff, D, Busch, R. Eujen, B. Fischer, H. Piel, R. Theisejans, IEEE Trans. Appl. Supercond. 5 (1995) 1251. [4] D.A. Korzekwa, J.F. Bingert, EJ. Podtburg, P. Miles, Appl. Supercond. 2 (1994) 261. [5] S.E. Schoenfeld, S. Ahzi, R.J. Asaro, J,F. Bingert, J.O. Willis, Phil. Mag. A 73 (1996) 1591. [6] K. Fischer, M. Schubert, C. Rodig, P. Verges, H.-W. Neumiiller, M. Wilhelm, B. Roas, A. Jenovelis, IEEE Trans. Appl. Supercond. 5 (1995) 1259. [7] G.E.P. Box, W.G. Hunter, J.S. Hunter, Statistics for Experimenters: An Introduction to Design, Data Analysis, and Model Building, Wiley, New York, 1978. [8] S.E. Dorris, B,C. Prorok, M.T. Lanagan, S. Sinha, R.B. Poeppel, Physica C 212 (1993) 66. [9] M.G. Smith, J.O. Willis, D.E. Peterson, J.F. Bingert, D.S. Phillips, J.Y. Coulter, K.V. Salazar, W.L. Hults, Physica C 231 (1994) 409. [!o] J,A, P~.pg!!, A.A. Polyanskii, A.E. Pashitski, D.C. Larbalestier, Supercond. Sci. Technol. 9 (1996) 393.