Considerations on critical currents and stability of MgB2 wires made by different preparation routes

Physica C 401 (2004) 80–86 www.elsevier.com/locate/physc

Considerations on critical currents and stability of MgB2 wires made by different preparation routes Wilfried Goldacker a

a,*

, Sonja I. Schlachter a, Bing Liu a, Bernhard Obst a, Evgenie Klimenko b

Forschungszentrum Karlsruhe, Institut f€ur Technische Physik, P.O. Box 3640, Karlsruhe D-76021, Germany b Kurchatov Institute, 123182 Moscow, Russia

Abstract MgB2 wires and tapes are attractive for the application in persistent mode coils, due to the absence of inter-granular weak links, sharp critical current transitions with high n-values and the chance to prepare superconducting contacts. High current densities are achieved for a row of different conductor concepts. A common bottleneck in most conductors is the still non-sufficient thermal stabilisation at high transport currents which leads to early quenches and limits the achievable transport currents. The evaluation of n-values from EðIÞ transitions can therefore be influenced by thermal effects from the dissipated energy, especially in the high current regime at lower fields or in self field. The microstructure of MgB2 , i.e. the phase purity and grain size distribution, differs significantly for in situ and ex situ preparation routes. It depends on the precursor constitution, heat treatments and the conductor geometry and influences percolation path and flux pinning and the characteristics of the EðIÞ transitions. We present EðIÞ data, measured at different background fields for MgB2 wires from different preparation routes and extract information about the thermal instability of the conductors. The results are correlated and discussed with the specific features of the MgB2 microstructure in the different conductors. Ó 2003 Published by Elsevier B.V. PACS: 74.60.Jg; 74.62.Bf Keywords: Critical currents; Thermal stability; MgB2

1. Introduction Since the discovery of superconductivity in MgB2 , a variety of conductor concepts, carrying high transport currents, were developed and investigated [1–10]. Quite different sheath composi-

* Corresponding author. Tel.: +49-7247-824179; fax: +497247-825398. E-mail address: [email protected] (W. Goldacker).

0921-4534/$ - see front matter Ó 2003 Published by Elsevier B.V. doi:10.1016/j.physc.2003.09.014

tions and materials, as Ta, Nb, Fe, Ni, Cu and steel, and preparation techniques like ex situ and in situ phase formation [11] were applied in tape and wire geometries (see review in [12]). However, common to most concepts so far is a non-sufficient thermal stabilisation of the conductor which leads to spontaneous local quenches in the high current– low field regime. As a consequence the measured values for the transport critical currents at low fields are underestimated compared to the current carrying potential of the material. The quench usually leads to an irreversible damage of the

W. Goldacker et al. / Physica C 401 (2004) 80–86

sample due to overheating. The n-values resulting from EðIÞ transitions may be affected strongly by the contribution from such heating effects and have to be evaluated with care. The necessary separation of the MgB2 core into a multifilamentary structure with fine filaments for an improved external thermal stabilisation without losses in the critical current density is presently limited through the precursor homogeneity, the final filament microstructure and the specific problems of the applied PIT-technique to achieve a regular conductor geometry with minimized sausaging of the filaments for a large deformation ratio. An external copper stabilisation in the sheath cannot solve the intra-filament stabilisation problem at this stage, but can serve to avoid the damage of the wire composite as already shown in Ref. [8] for a steel reinforced wire with Ta(Nb)–Cu composite sheath. It is well known from various investigations on the critical current transitions of LTS materials that the shape of the EðIÞ characteristics is strongly determined through the distribution of the local microscopic critical currents in the conductor material [13–15]. These investigations showed that inhomogeneity of the phase, the presence of secondary phase particles, a non-regular grain structure [16] and sausaging of the filaments [17] along the sample strongly influence this current distribution and consequently the shape of EðIÞ. The characterisation becomes more complex if the dissipated energy is not sufficiently captured by the normal conducting material of the sheath and hot spots form as a consequence in microscopic filament sections, as it is the case for MgB2 filaments. The aim of this paper is to analyse EðIÞ transitions of round MgB2 monofilamentary wires made by different preparation techniques which lead to a different filament microstructure. Information about the energy dissipation behaviour will be extracted from the EðIÞ dependence. The consequence for the interpretation of n-values will be discussed.

2. Experimental Monofilamentary MgB2 -wires with Fe-Sheath and 1.1 mm final diameter were fabricated via the

81

powder-in-tube technique as described elsewhere in detail [10,11]. Three different precursor techniques were applied to achieve different microstructures after the final heat treatment, which was performed at various temperatures between 850 and 980 °C to get a dense filament material via a partial phase decomposition, reformation and sintering. The precursors are commercial powder from Alfa Aesar with original particle size distribution (precursor AAE), the same powder milled in a propanol suspension by means of an attrition mill technique (precursor AAE-AT) and as a reference a Mg–B powder mixture for in situ formation of MgB2 (precursor IN-SI) with a nonstoichiometric composition of Mg0:5 B2 . Particle sizes in the precursor were analysed by means of a Coulter–Beckman LS230 analyser. The particle size distributions of the AAE and AAE-AT precursors in Fig. 1 show the milling effect. The shape of the distributions of both precursors is similar, but centred at different particle size levels. This is an indication that mostly agglomerates were cracked. The AAE precursor has a significant number of particles >100 lm, whereas all particles in the AAE-AT precursor are smaller than 10 lm with a maximum at 7 lm and a gravity centre at about 2 lm. Transport critical currents were measured by means of the usual 4-point method on short sample sections of 10 mm at 4.2 K in background fields

Fig. 1. Particle size distributions of ex situ precursors: sample AAE is Alfa Aesar MgB2 as delivered, AAE-AT is the same powder milled as alcoholic suspension by means of an attrition technique.

82

W. Goldacker et al. / Physica C 401 (2004) 80–86

up to 9 T applying a 1 lV/cm criterion. EðIÞ values were detected in the range 0.1 lV/cm to 10 mV/cm. The sweep ramp for the currents was typically 0.2–2 A/s depending on the expected critical current level. The measured EðIÞ dependences were analysed with the following relationships [13–15]. Fitting a power law behaviour at I ¼ Ic leads to the commonly used n-values
n I EðIÞ ¼ Ec
Ic The electrical field EðIÞ can also be expressed in the following equation Z I R ðI  ic Þ
/ðic Þ dic EðIÞ ¼
l 0 /ðic Þ is the distribution of the critical currents, usually assumed as a Gaussian distribution [15]. It follows from this equation that in principle the current distribution can be extracted from the EðIÞ graph via the second derivative /ðic Þ ¼

l d2 E
R dI 2

But in practise very often noise on the EðIÞ curve and a necessary smoothing of the graph limits a quantitative and reliable analysis. In our case of MgB2 wires, the application of this method failed. A quite favourable method to separate and visualize thermal effects from the EðIÞ transition is given regarding a function GðIÞ [18]
1 d lnEðIÞ GðIÞ ¼ dI The linear part in the plot of GðIÞ with positive slope at lower currents corresponds to the potential law, the pass through a maximum and the part with negative slope at higher currents indicates contributions from thermal effects.

3. Results 3.1. Filament microstructure In the reference wire with in situ reacted filament (IN-SI), the main secondary phase consists

Fig. 2. Microstructure in the filament of the in situ processed sample IN-SI. The black grains with light channels (MgB2 ) are not fully reacted boron grains and represents the boron rich, non-superconducting secondary phase. Particle sizes range up to about 20 lm.

of not fully reacted boron grains or boron rich particles up to a size of approximately 20 lm. Despite of the very inhomogeneous microstructure, the reacted filament sections between the boron grains are filled with high transport current carrying stoichiometric MgB2 phase (Fig. 2). The wires with original commercial (AAE) precursors are characterised by an inhomogeneous grain size distribution and inclusions of secondary phases, MgO and B (Fig. 3). Single very large MgB2 grains of up to 50 lm are characteristic for these conductors. In contrast, the milled precursor (AAE-AT) leads to a much more homogeneous distribution of both, grain sizes and fine distributed boron rich secondary phase inclusions with small particle sizes well below 10 lm and a centre of gravity at about 2 lm. 3.2. Transport critical current densities For the MgB2 wires with AAE and AAE-AT precursors, the optimum heat treatment temperature for high critical currents was investigated as is shown in Fig. 4. For the AAE-wire a plateau in the range T ¼ 900 to 965 °C was found, whereas for the AAE-AT wires a more narrow heat treatment interval of T ¼ 950 to 965 °C was found for optimised currents. This was not expected since a

W. Goldacker et al. / Physica C 401 (2004) 80–86

83

Fig. 3. Cross sections of wires (diameter 1.1 mm) AAE (left side) and AAE-AT (right side). Sample AAE-AT shows a homogeneous microstructure with grains <10 lm, whereas sample AAE is characterised by quite large MgB2 grains of up to about 50 lm.

Fig. 4. Critical current densities (normalized) at 6 T, 4.2 K for different annealing temperatures for the ex situ samples AAE and AEE-AT.

mechanical homogenisation should decrease and not raise the reaction temperature. An explanation for this behaviour and the role of the milling technique is still under investigation. Samples heat treated at 965 °C were selected for further characterisation. In Fig. 5, the field dependence of the transport critical current densities of all three kinds of samples are given. All conductors show thermal instability at lower fields of 3–5 T and burn through. The most stable conductor is sample AAE-AT, mainly due to the lower critical current density level. The reasons for the decreased critical current densities are presently still under investigation. There is room for some speculation. On one hand a contamination with oxygen during milling cannot be fully excluded. On the other hand, a possibly very successful phase homogenisation could have the effect that good percolation

Fig. 5. Field dependence of the critical current densities and the corresponding n-values (calculated at EðIÞ ¼ 1 lV/cm) for the samples IN-SI (in situ) and AAE and AE-AT (annealed at T ¼ 965 °C).

paths are averaged out to a reduced MgB2 performance. Or as an alternative possibility a homogeneous contribution of ‘‘good’’ and ‘‘bad’’ sample section reduces macroscopic transport currents. A similar behaviour was shown in Ref. [12] and explained with segregation of impurities at the grain boundaries. The Jc level for occurring thermal instabilities is different for the different wire types, already indicating an influence of the microstructure and phase quality.

84

W. Goldacker et al. / Physica C 401 (2004) 80–86

The n-values determined at Ic (1 lV/cm) are similar for the high current wires IN-SI and AAE and much lower and with a narrower field dependence for the sample AAE-AT. This picture correlates well with the comparison of the current densities. The small grained microstructure of sample AAE-AT with homogeneously distributed secondary phases obviously provides less good conditions for the energy dissipation, indicated by the lower current density level for the quench. The nðl0 H Þ graphs of sample AAE and AAE-AT are only slightly curved as the data in Ref. [12] and reach values close to 50 at B ¼ 5 T. A curvature of the nðl0 H Þ graph can be interpreted as an indication of extrinsic limitations as sausaging, a straight line correlates to intrinsic limitations [13,14]. All three samples are very close to a straight line. 3.3. E(I) transitions A detailed investigation of the EðIÞ transitions was limited to the ex situ wires, since despite of the excellent current densities, the IN-SI wires quench very early above the critical current density and an evaluation of the EðIÞ characteristics could not be performed. The reason is an obviously broad local current density distribution around the boron rich inclusions (see Fig. 2). For the other two samples AAE and AAE-AT, the EðIÞ curves are given in Fig. 6 in a double logarithmic presentation. In high fields the calculated n-value depends sensitively on the position on the curve for both samples, at lower fields and higher currents the curves become more linear and closer to a power law. This behaviour is quite similar to LTS conductors. At currents of >10 A quench effects begin to occur represented by the sudden step of EðIÞ. The step size increases with higher currents, the onset finally crossing the Ic criterion of 1 lV/cm. At higher currents the thermal stabilisation becomes rapidly so poor that the wires burn through. Surpassing the local bad critical currents generates enough heat to quench the entire filament before an average critical current is reached. Wire AAE-AT differs from AAE only slightly in a lower level where the quenches occur. This is an indication for a broader distribution of the critical currents and

Fig. 6. EðIÞ graphs for different background fields for the wires AAE and AAE-AT given in a double-logarithmic presentation. The current sweep was 0.2–2 A/s.

is consistent with the picture of a worse percolation path for the transport critical current. The same EðIÞ data in the half-logarithmic representation, given in Fig. 7, are suitable to recognize changes in the slope of the EðIÞ graphs and to identify turning points. Turning points are observed for both samples in the high current regime and indicate the onset of contributions from thermal effects. The function GðIÞ as defined above (see Ref. [18]) helps to quantify this onset and is given in Fig. 8. For sample AAE, the almost linear part with positive slope represents the power law behaviour. The maximum and the section with the negative slope show the influence of thermal effects. The large scattering of EðIÞ in the curves at low currents can be interpreted as thermal fluctuations, probably due to the inhomogeneous microstructure and MgB2 grain size in the AAE filament. For sample AAE-AT, with the much

W. Goldacker et al. / Physica C 401 (2004) 80–86

85

Fig. 7. EðIÞ graphs for different background fields for the wires AAE and AAE-AT given in a single-logarithmic presentation. The current sweep was 0.2–2 A/s.

Fig. 8. Plot of function GðIÞ for samples AAE and AAE-AT (for the definition of GðIÞ see text) for different background fields at 4.2 K.

more homogeneous microstructure, these fluctuations are nearly absent, but thermal effects contribute more to some curvature of the linear part and the maximum of GðIÞ is broader. This is again an indication for a broader distribution of critical currents and the contribution of thermal effects.

A contribution of such a successive energy dissipation to the shape of the EðIÞ transition is very difficult to detect especially for small E-values, the observed heat fluctuations however indicate their presence. The comparison of the presented MgB2 monofilamentary wires with 0.8 mm thick filaments shows that both a homogeneous microstructure and a high current carrying percolation path have a significant influence on the intra-filamentary thermal stabilisation, the occurrence of hot spots and the dissipation of the energy reaching Ic . The distribution of the local critical currents and current density, a consequence of the microstructure, especially inclusions of secondary phases, plays an important role for the ignition of the complete quench at high transport currents. This was especially the case for the IN-SI wire with boron-rich inclusions in the filament. If the complete quench occurs at low E-values, it is very

4. Discussion and conclusions A detailed investigation of the whole EðIÞ transition in MgB2 conductors gave some information about the onset and contribution of thermal effects. Such effects were found at all current levels and background fields and depend on the transport current density level. At high transport currents, thermal effects can significantly increase the slope of EðIÞ. As a consequence the calculated n-values become larger and may be overestimated.

86

W. Goldacker et al. / Physica C 401 (2004) 80–86

likely that the n-values are affected by local thermal effects inside the filament, which become effective through the limited heat transfer to the sheath. Similar mechanisms were considered for Nb3 Sn-strands in Ref. [19]. An improvement of the filament quality improves on one hand the internal thermal stabilisation of the conductor in the low field–high current regime, but on the other hand raises the performance level for the critical current density. For a final statement about the true intrinsic nvalues of MgB2 filaments and for the development of a stabilised technical conductor, a multifilamentary structure with thin filaments and homogeneous microstructure and a sufficient external thermal stabilisation is absolutely necessary to eliminate the contribution of intra-filamentary thermal effects. Acknowledgements We thank Prof. Dr. Klaus G. Nickel and his coworkers of the Eberhard-Karls-Universit€ at T€ ubingen, Institute for Geosciences, Applied Mineralogy for the expertised attrition milling of the MgB2 and fruitful discussions about specific material aspects.

References [1] P.C. Canfield, D.K. Finnemore, S.L. Budko, J.E. Ostenson, G. Lapertot, C.E. Cunningham, C. Petrovic, Phys. Rev. Lett. 86 (2001) 2423.

[2] H.L. Suo, C. Beneduce, M. Dhalle, N. Musolino, J.Y. Genoud, R. Fl€ ukiger, Appl. Phys. Lett. 79 (2001) 3116. [3] B.A. Glowacki, M. Majores, M. Vickers, J.E. Evetts, Y. Shi, I. Mcdougall, Supercond. Sci. Technol. 14 (2001) 193. [4] S. Soltanian, X.L. Wang, I. Kusevic, E. Babic, A.H. Li, H.K. Liu, E.W. Collings, S.X. Dou, Physica C 361 (2001) 84. [5] G. Grasso, A. Malagoli, D. Marr_e, E. Bellingeri, V. Braccini, S. Roncallo, N. Scati, S. Siri, Physica C 378–381 (2002) 899. [6] W. Goldacker, S.I. Schlachter, H. Reiner, S. Zimmer, B. Obst, H. Kiesel, A. Nyilas, in: A. Narlikar (Ed.), Studies of High Temperature Superconductors, vol. 45, Nova Science Publishers, New York, 2002. Available from . [7] A. Serquis, L. Civale, D.L. Hammon, J.Y. Coulter, X.Z. Liao, Y.T. Zhu, D.E. Peterson, F.M. Mueller, Appl. Phys. Lett., in press. Available from . [8] W. Goldacker, S.I. Schlachter, S. Zimmer, H. Reiner, Supercond. Sci. Technol. 14 (2001) 787. [9] H. Kumakura, A. Matsumoto, H. Fujii, K. Togano, Appl. Phys. Lett. 79 (2001) 2435. [10] W. Goldacker, S.I. Schlachter, Physica C 378–381 (2002) 889. [11] W. Goldacker, S.I. Schlachter, S. Zimmer, H. Reiner, Adv. Cryog. Eng. 48 (2002) 864. [12] R. Fl€ ukiger, H.L. Suo, N. Musolino, C. Beneduce, P. Toulemonde, P. Lezza, Physica C 385 (2003) 286. [13] W.H. Warnes, D.C. Larbalestier, Cryogenics 26 (12) (1986) 643. [14] W.H. Warnes, J. Appl. Phys. 63 (5) (1988) 1651. [15] D.P. Hampshire, H. Jones, Cryogenics 27 (1987) 608. [16] R. Kimmich, F. Hornung, A. Rimikis, Th. Schneider, P. Lee, IEEE Trans. Appl. Supercond. 11 (1) (2001) 3675. [17] J. Ekin, Cryogenics 27 (1987) 603. [18] R. Kimmich, Thesis, University of Karlsruhe, FZK-report no. FZKA 6562, 2001. [19] M.D. Sumption, E.W. Collings, in: ICMC Topical Conference, Twente, 25–28 May, Physica C (2004), these Proceedings.