arX

iv:c

ond-

mat

/060

6303

v1 [

cond

-mat

.sup

r-co

n] 1

3 Ju

n 20

06

A Review of the Properties of Nb3Sn and Their Variation with A15 Composition,

Morphology and Strain State

A. Godeke∗

Ernest Orlando Lawrence Berkeley National Laboratory, Berkeley, CA 94720

(Dated: February 6, 2008)

Significant efforts can be found throughout the literature to optimize the current carrying capacityof Nb3Sn superconducting wires. The achievable transport current density in wires depends on theA15 composition, morphology and strain state. The A15 sections in wires contain, due to compo-sitional inhomogeneities resulting from solid state diffusion A15 formation reactions, a distributionof superconducting properties. The A15 grain size can be different from wire to wire and is also notnecessarily homogeneous across the A15 regions. Strain is always present in composite wires, andthe strain state changes as a result of thermal contraction differences and Lorentz forces in magnetsystems. To optimize the transport properties it is thus required to identify how composition, grainsize and strain state influence the superconducting properties. This is not accurately possible ininhomogeneous and spatially complex systems such as wires. This article therefore gives an overviewof the available literature on simplified, well defined (quasi-)homogeneous laboratory samples. Aftermore than 50 years of research on superconductivity in Nb3Sn, a significant amount of results areavailable, but these are scattered over a multitude of publications. Two reviews exist on the basicproperties of A15 materials in general, but no specific review for Nb3Sn is available. This articleis intended to provide such an overview. It starts with a basic description of the Niobium-Tinintermetallic. After this it maps the influence of Sn content on the the electron-phonon interactionstrength and on the field-temperature phase boundary. The literature on the influence of Cu, Tiand Ta additions will then be briefly summarized. This is followed by a review on the effects ofgrain size and strain. The article is concluded with a summary of the main results.

PACS numbers: 74.25.Dw, 74.62.Bf, 74.62.Dh, 74.81.Bd

Accepted Invited Topical Review for Superconductor Science and Technologyhttp://www.iop.org/EJ/journal/SUST c©IOPP 2006Provisionally scheduled for July 2006

I. INTRODUCTION

Superconductivity in Nb3Sn was discovered byMatthias et al. in 1954 [1], one year after the discovery ofV3Si, the first superconductor with the A15 structure byHardy and Hulm in 1953 [2]. Its highest reported criticaltemperature is 18.3 K by Hanak et al. in 1964 [3]. Eversince its discovery, the material has received substantialattention due to its possibility to carry very large cur-rent densities far beyond the limits of the commonly usedNbTi. It has regained interest over the past decade dueto the general recognition that NbTi, the communities’present workhorse for large scale applications, is oper-ating close to its intrinsic limits and thus exhausted forfuture application upgrades. Nb3Sn approximately dou-bles the available field-temperature regime with respectto NbTi and is the only superconducting alternative thatcan be considered sufficiently developed for large scaleapplications.

Intermetallic Niobium-Tin is based on the supercon-ductor Nb, which exists in a bcc Nb structure (Tc

∼=9.2 K), or a metastable Nb3Nb A15 structure (Tc

∼=

∗Previous address: Low Temperature Division, Faculty of Scienceand Technology, University of Twente, P.O. Box 217, 7500 AE,Enschede, The Netherlands

5.2 K) [4, 5]. When alloyed with Sn and in thermody-namic equilibrium, it can form either Nb1−βSnβ (about0.18 ≤ β ≤ 0.25) or the line compounds Nb6Sn5 andNbSn2 according to the generally accepted binary phasediagram by Charlesworth et al. [6] (figure 1). The solidsolution of Sn in Nb at low concentrations (β < 0.05)gradually reduces the critical temperature of bcc Nb fromabout 9.2 K to about 4 K at β = 0.05 (Flukiger in [7]).Both the line compounds at β = 0.45 and 0.67 are su-perconducting with Tc < 2.8 K for Nb6Sn5 [8, 9] andTc < 2.68 K for NbSn2 [9, 10] and thus are of negligibleinterest for practical applications.

The Nb-Sn phase of interest occurs from β ∼= 0.18 to0.25. It can be formed either above 930 C in the pres-ence of a Sn-Nb melt, or below this temperature by solidstate reactions between Nb and Nb6Sn5 or NbSn2. Someinvestigations suggest that the nucleation of higher Snintermetallics is energetically more favorable for lowerformation temperatures [11, 12, 13, 14, 15, 16]. This isindicated by the dashed line within the Nb1−βSnβ sta-bility range in figure 1. The critical temperature forthis phase depends on composition and ranges approx-imately from 6 to 18 K [11]. At low temperatures andat 0.245 < β < 0.252 it can undergo a shear transforma-tion at Tm

∼= 43 K. This results in a tetragonal struc-ture in which the reported ratio of the lattice parametersc/a = 1.0026 to 1.0042 [17, 18, 19, 20, 21, 22, 23, 24].

2

0 10 20 30 40 50 60 70 80 90 1000

500

1000

1500

2000

2500

16 18 20 22 24 26 280

10

20

30

40

50

? Nb3Sn + Nb6Sn5

231.9 °C

NbSn2 + Sn

NbSn2 + Liquid

NbSn2

Nb6Sn5 + NbSn2

Nb6Sn5

Nb6Sn5 + Liquid

845 ± 7 °C930 ± 8 °C

2130 ± 30 °C

α−Nb + Liquid

α−Niobiumand Nb3Sn

α−Nb

Nb3Sn

Nb3Sn + Liquid

Liquid

Tem

pera

ture

[ °C

]

Atomic Sn content [ % ]

Tetragonal

Cubic

Atomic Sn content [ % ]

Tem

pera

ture

[ K

]

FIG. 1: Binary phase diagram of the Nb-Sn system after Charlesworth et al. [6]†, with an optional modification to suggestedpreference for Sn rich A15 formation [11, 12, 13, 14, 15, 16]. The inset depicts the low temperature phase diagram after

Flukiger [17]‡, which includes the stability range of the tetragonal phase.†

c©1970 Chapman and Hall Ltd. Adapted with kind permission of Springer Science and Business Media. There are instances where we havebeen unable to trace or contact the original authors. If notified the publisher will be pleased to rectify any errors or omissions at the earliestopportunity.‡ c©1982 Plenum Press. Adapted with kind permission of Springer Science and Business Media and R. Flukiger.

This transformation is schematically depicted in the lowtemperature extension of the binary phase diagram asshown in the inset in figure 1.

Nb1−βSnβ exists in the brittle A15 crystal structurewith a cubic unit cell, as schematically depicted in fig-ure 2. The Sn atoms form a bcc lattice and each cubeface is bisected by orthogonal Nb chains. The importanceof these chains is often emphasized to qualitatively un-derstand the generally high critical temperatures of A15compounds. An excellent review on this subject is givenby Dew-Hughes in [18] and can be summarized specifi-cally for the Nb-Sn system. In bcc Nb the shortest spac-ing between the atoms is about 0.286 nm, starting froma lattice parameter a = 0.330 nm [25]. In the A15 lat-tice, with a lattice parameter of about 0.529 nm for thestoichiometric composition [11], the distance between theNb atoms is about 0.265 nm. This reduced Nb distancein the chains is suggested to result in a narrow peak inthe d-band density of states (DOS) resulting in a veryhigh DOS near the Fermi level. This is in turn believedto be responsible for the high Tc in comparison to bcc

Nb. Variations in Tc are often discussed in terms of long-

range crystallographic ordering [4, 18, 26, 27, 28] sincedeviations in the Nb chains will affect the DOS peak.Tin deficiency in the A15 structure causes Sn vacancies,but these are believed to be thermodynamically unsta-ble [4, 29]. The excess Nb atoms will therefore occupySn sites, as will be the case with anti-site disorder. Thisaffects the continuity of the Nb chains which causes arounding-off of the DOS peak. Additionally, the bcc sitedNb atoms cause there own broader d-band, at the cost ofelectrons from the Nb chain peak.

The above model is one of many factors that are sug-gested to explain effects of disorder on the superconduct-ing transition temperature [28]. It is presented here sinceit intuitively explains the effects of Nb chain atomic spac-ing and deviations from long range crystallographic or-dering and also is mostly cited. In the Nb3Nb systemwith a lattice constant of 0.5246 nm [5] the distance be-tween the chained Nb atoms of about 0.262 nm is, as inthe A15 Nb-Sn phase, lower than in the bcc lattice. Thiswould suggest an increased Tc compared to bcc Nb. Itslower Tc of 5.2 K [4] could possibly be explained in theabove model by assuming that ’Sn sited’ Nb atoms cre-

3

FIG. 2: Schematic presentation of the Nb3Sn A15 unit cell.The light spheres represent Sn atoms in a bcc lattice. The darkspheres represent orthogonal chains of Nb atoms bisecting thebcc cube faces.

ate their own d-band at the cost of electrons from thechained Nb atoms, thereby reducing their DOS peak andthus degrading the expected Tc gain.

II. VARIATIONS IN LATTICE PROPERTIES

Variations in superconducting properties of the A15phase are throughout the literature related to variationsin the lattice properties through the lattice parameter(a), atomic Sn content (β), the normal state resistivityjust above Tc (ρn) or the long range order (LRO). Thelatter can be defined quantitatively in terms of the Bragg-Williams order parameters Sa and Sb for the chain sitesand the cubic sites respectively. These can be expressedin terms of occupation factors ([4], Flukiger in [7]):

Sa =ra − β

1 − βand Sb =

rb − (1 − β)

1 − (1 − β), (1)

where ra and rb are the occupation factors for A atomsat the chain sites and B atoms at the cubic sites respec-tively in an A15 A1−βBβ system. At the stoichiometriccomposition S = Sa = Sb = 1 and S = 0 representscomplete disorder. Qualitatively, the more general term’disorder’ is used for the LRO since any type of disor-der (e.g. quenched in thermal disorder, off-stoichiometry,neutron irradiation) reduces Tc [28]. A review of theliterature suggests that the aforementioned lattice vari-ables (a, β and S) and ρn are, at least qualitatively, in-terlinked throughout the A15 phase composition range.This makes it possible to relate them back to the mainavailable parameter in multifilamentary composite wires,i.e. the atomic Sn content.

The lattice parameter as function of composition wasmeasured by Devantay et al. [11] and combined with ear-lier data from Vieland [13]. These data are reproduced

18 19 20 21 22 23 24 250.5280

0.5282

0.5284

0.5286

0.5288

0.5290

0.5292

Flükiger 1981:Nb3Nb extrapolation:a( ) = 0.0176 + 0.5246

Devantay 1981 Vieland 1964

Devantay 1981:a( ) = 0.0136 + 0.5256

Latti

ce p

aram

eter

[ nm

]

Atomic Sn content [ % ]

FIG. 3: Lattice parameter as function of atomic Sn contentafter Devantay et al. [11]†, including proposed linear depen-

dencies after Devantay et al. and Flukiger in [7]‡.†

c©1981 Chapman and Hall Ltd. Adapted with kind permission ofSpringer Science and Business Media and R. Flukiger.‡ c©1981 Plenum Press. Adapted with kind permission of SpringerScience and Business Media and R. Flukiger.

0.0 0.2 0.4 0.6 0.8 1.00

25

50

75

100

125

150

Nölscher 1985 Flükiger 1987

n(S) = 147 (1 − S 4 )

Resi

stiv

ity [

µΩcm

]

Order parameter

FIG. 4: Resistivity as function of order parameter includinga fourth power fit after Flukiger et al. [27]†.†

c©1987 IEEE. Adapted with kind permission of IEEE andR. Flukiger.

in figure 3. The solid line is a fit to the data similar asgiven by Devantay et al.:

a (β) = 0.0136β + 0.5256 [nm] . (2)

The dotted line is an alternative line fit proposed byFlukiger in [7]:

a (β) = 0.0176β + 0.5246 [nm] , (3)

in which the lattice parameter is extrapolated from0.529 nm at β = 0.25 to a Nb3Nb value of 0.5246 nm [5]

4

18 19 20 21 22 23 24 250

20

40

60

80

100

Devantay 1981 Hanak 1964 Orlando 1981

n( ) = 91 [1 − (7 − 0.75)4 ] + 3.4

Resi

stiv

ity [

µΩcm

]

Atomic Sn content [ % ]

FIG. 5: Resistivity as function of atomic Sn content afterFlukiger [4, 32]†. The solid curve is a fit to the data accordingto (5).†Adapted with kind permission of R. Flukiger.

at β = 0. The argument for doing so is that from analysisof the lattice parameter of a wide range of Nb1−βBβ su-perconductors as function of the amount of B element, itappears that most lattice parameters extrapolate to theNb3Nb value for β = 0. It is interesting to note that therise in lattice parameter with Sn content is apparentlycontrary to all other A15 compounds, for which a re-duction in lattice parameter is observed with increasingB-element. For the analysis in this article, (2) is pre-sumed an accurate description for the lattice parameterresults as measured by Devantay et al. and Vieland forthe Nb-Sn A15 stability range.

The amount of disorder, introduced by irradiation orquenching is often related to changes in resistivity. Fur-thermore, the amount of disorder is important in dis-cussions on the strain sensitivity of the superconductingproperties in the Nb-Sn system in comparison to otherA15 compounds (section VIII). A relation between ρn

and the order parameter S in Nb3Sn was establishedby Flukiger et al. [27] after analysis of ion irradiationdata obtained by Nolscher and Seamann-Ischenko [30]and data from Drost et al. [31]. This is reproduced infigure 4. Flukiger et al. proposed a fourth power fit todescribe the ρn(S) data:

ρn (S) = 147 (1 − S)4 [µΩcm] , (4)

represented by the solid curve through the data points.The superconducting properties of the Nb-Sn system

are mostly expressed in terms of either resistivity oratomic Sn content. Resistivity data as function of com-position were collected by Flukiger [4, 32] from Devantayet al. [11], Orlando et al. [33, 34], and Hanak et al. [3] andare reproduced in figure 5. The relation between the twoparameters from various sources is consistent. Since Sndeficiency will result in anti-site disorder (section I) it isassumed here that ρn(β) will behave similarly to ρn(S),

i.e. a fourth power fit, identical to (4) will be appro-priate. The solid curve in figure 5 is a fit according to:

ρn (β) = 91[

1 − (7β − 0.75)4]

+ 3.4 [µΩcm] , (5)

which accurately summarizes the available data.

III. ELECTRON-PHONON INTERACTION AS

FUNCTION OF ATOMIC TIN CONTENT

The Nb3Sn intermetallic is generally referred to asa strong coupling superconductor. Any connection tomicroscopic formalisms, however, requires some level ofunderstanding of the details of the interaction, since itchanges the way in which physical quantities can be de-rived from measurements. More specifically, it requires adescription for the interaction strength that varies withcomposition.

The BCS theory [35] provides a weak coupling approx-imation for the energy gap at zero temperature:

∆0∼= 2~ωc exp

[

−1

λep

]

, (6)

in which λep is a dimensionless electron-phonon interac-tion parameter [36]. ~ωc is a cutoff energy away from theFermi level (EF) outside which the attractive electron-electron interaction potential (V0) becomes zero. Equa-tion 6 is valid for λep ≪ 1. Evaluation of the temperaturedependency of the gap ∆(T ) and requiring that the gapbecomes zero as T → Tc yields the following descriptionfor the critical temperature in the weak coupling approx-imation:

Tc (0) ∼=2eγE

πkB~ωc exp

[

−1

λep

]

, (7)

in which γE∼= 0.577 (Euler’s constant). The ratio be-

tween the zero temperature gap (6) and the zero fieldcritical temperature (7): 2∆0/kBTc = 3.528 is a con-stant and represents the BCS weak coupling limit. Forstrong coupling (6) and (7) are no longer valid since theybecome dependent on the details of the electron-phononinteraction. This effectively results in a rise of the ratio2∆0/kBTc above the weak coupling limit of 3.528.

Moore et al. [37] have analyzed the superconductinggap and critical temperature of Nb-Sn films as function ofatomic Sn content by tunneling experiments. Their mainresult is reproduced in figure 6. In the left plot the in-ductively measured midpoint (i.e. halfway the transition)critical temperatures, and from tunneling current-voltagecharacteristics derived gaps are reproduced as function ofcomposition. The dotted curves are fits to the data usinga Boltzmann sigmoidal function:

y (β) =ymin − ymax

1 − exp(

β−β0

dβ

) + ymax, (8)

5

12 16 20 24 28 320

2

4

6

8

10

12

14

16

18

16 18 20 22 24 26 282

3

4

5

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

Mid

poin

t crit

ical

tem

pera

ture

[ K

]

Atomic Sn content [ % ]

BCS weak coupling limit

2∆ / k BT c

Atomic Sn content [ % ]

Boltzmann function

∆ [ meV

]

FIG. 6: Critical temperature and superconducting gap as function of composition (left plot) and the ratio 2∆0/kBTc as functionof composition (right plot). The ratio indicates weak coupling for compositions below 23 to 24 at.% Sn and strong coupling forcompositions close to stoichiometry. After Moore et al. [37]†.† c©1979 The American Physical Society. Adapted with kind permission of the American Physical Society and M.R. Beasley.

where y represents Tc or ∆, β is the atomic Sn contentand ymin, ymax, β0 and dβ are fit parameters. Both Tc(β)and ∆(β) are accurately described by (8).

The A15 composition range as used by Moore et al. isslightly wider than generally accepted on the basis of thebinary phase diagram of Charlesworth et al. (figure 1,[6]). Nevertheless, following [37] the ratio 2∆0/kBTc canbe derived from these data sets and is reproduced in theright graph in figure 6. The BCS weak coupling limit of3.528 is indicated by the solid line. The weak couplingvalue of about 3 for low Sn content A15 was attributedby Moore et al. to the finite inhomogeneity in their sam-ples of 1 to 1.5 at.% Sn, in combination with an inductivemeasurement of Tc which preferably probes the highestSn fractions. These Tc values could thus be higher thanrepresentative for the bulk. It is also known that in tun-neling experiments the gap at the interface can be lowerthan in the bulk, effectively lowering the measured value.Also, the gap measurement was performed at finite tem-perature, which also reduces its value. A fourth possi-ble origin is the existence of a second gap in Nb3Sn, aswas recently postulated by Guritanu et al. [38]. Nev-ertheless it is clear that the Nb-Sn system shows weakcoupling for most of its A15 composition range and onlybecomes strong coupled for compositions above 23 to 24at.% Sn. The general statement that Nb3Sn is a strongcoupling superconductor therefore only holds for compo-sitions close to stoichiometry. Strong coupling correc-tions to microscopic descriptions thus become relevantfor compositions approaching stoichiometry and shouldbe accounted for in any theory that is applied to describethe entire A15 composition range.

IV. Tc AND Hc2 AS FUNCTION OF ATOMIC

TIN CONTENT

Compositional gradients will inevitably occur in wiressince their A15 regions are formed by a solid state diffu-sion process. It is therefore important to know the vari-ation of the critical temperature and upper critical fieldwith composition. The most complete data-sets of Tc(β)that exist in the literature are those of Moore et al. ([37],figure 6) and Devantay et al. [11]. The data of Mooreet al. suggest, as mentioned in the previous section, aslightly broader A15 composition range than is generallyaccepted to be stable according to the binary phase di-agram by Charlesworth et al. ([6], figure 1). The datafrom Devantay et al., however, are in agreement with theaccepted A15 stability range and are therefore assumedto be more accurate.

The results from Devantay et al. are reproduced inthe left plot in figure 7. Some points are reproducedfrom Flukiger [32], who credited these (partly) additionalresults also to Devantay et al.. The linear Tc(β) fit, in-dicated by the dashed line, was originally proposed byDevantay et al. to summarize their results:

Tc (β) =12

0.07(β − 0.18) + 6. (9)

The dotted curve summarizes the results of Moore et al.

using (8). The general tendency of the data sets of De-vantay et al. and Moore et al. is similar, although thelatter covers a wider A15 stability range and its Tc val-ues are slightly lower. The solid curve represents a fit tothe data of Devantay et al. according to a Boltzmann

6

17 18 19 20 21 22 23 24 25 260

2

4

6

8

10

12

14

16

18

20

18 19 20 21 22 23 24 25 260

5

10

15

20

25

30

35

Moore 1979

Devantay 1981 Devantay 1981 (after Flükiger 1981)

Tc( ) linear

Tc( ) Boltzmann function

Crit

ical

tem

pera

ture

[ K

]

Atomic Sn content [ % ]

tetra−gonal

cubic

(t)

(c)

Devantay 1981 Orlando 1981 Arko 1978 Foner 1981 Jewell 2004

0Hc2( ) function

Upp

er c

ritic

al fi

eld

[ T ]

Atomic Sn content [ % ]

FIG. 7: Literature results for the critical temperature (left pot) and upper critical field at zero temperature (right plot) asfunction of Nb-Sn composition. The Tc(β) Boltzmann function (10) and the µ0Hc2(β) function (11) are empirical relationsthat summarize the available literature results.Adapted from [11] c©1981 Chapman and Hall Ltd., [12] c©2004 The American Institute of Physics, [32], [37] c©1979 The American PhysicalSociety. With kind permission of Springer Science and Business Media, the American Physical Society, the American Institute of Physics,R. Flukiger, T.P. Orlando, M.R. Beasley and M.C. Jewell.

function identical to (8):

Tc (β) =−12.3

1 + exp(

β−0.220.009

) + 18.3. (10)

Equation (10) assumes a maximum Tc of 18.3 K, thehighest recorded value for Nb3Sn [3].

The right plot in figure 7 is a reproduction of a Hc2(β)data collection that was made by Flukiger [32] with somemodifications. After Flukiger, it represents a collectionof results from Devantay et al. [11], Orlando et al. [33, 34]and a close to stoichiometric single crystal from Arko et

al. [39]. The dotted line in the plot separates the cubicand tetragonal phases at 24.5 at.% Sn. The Foner andMcNiff results in figure 7 are, in contrast to Flukiger’scollection, here separated in the cubic and the tetragonalphase with assumed identical composition. The compo-sition that was used here was taken from the Flukigercollection, which showed a single µ0Hc2 point at a valueof about 27 T (the average between the cubic and tetrag-onal phase). It is unclear where this compositional valueis attributed to or whether the cubic single crystal in facthas a lower Sn content.

In addition to the results collected by Flukiger alsorecent Hc2(β) measurements from Jewell et al. [12] areadded. A slightly different route was followed than givenin [12] for the analysis of the Hc2(T ) results. The com-positions of the homogenized bulk samples in [12] werecalculated from the Tc values assuming a linear Tc(β)dependence as given by (9). The resulting Hc2(β) us-ing these compositions is linear and deviates from thecollection by Flukiger, as shown in [12]. Here, the com-positions were recalculated based on the Tc values usingthe Boltzmann fit on the Devantay et al. results as given

by (10). The Jewell et al. bulk results then become con-sistent with the earlier Hc2(β) data. The highest µ0Hc2

value of 31.4 T represented, although extrapolated [12], arecord for binary Nb-Sn, but has very recently been sur-passed by an (extrapolated) µ0Hc2(0) = 35.7 T in highlydisordered bulk samples [40]. The existing Hc2(β) datacan be summarized by a summation of an exponentialand a linear fit according to:

µ0Hc2 (β) = −10−30 exp

(

β

0.00348

)

+577β − 107. (11)

The solid curve in the right graph in figure 7 is calculatedusing (11). It is interesting to note that a linear extrapo-lation of the cubic Hc2(β) results yields around 35 T forthe stoichiometric composition, in agreement with the re-ported record value by Cooley et al. [40]. Unfortunately,however, compositional information is not available forthis bulk sample, a high degree of uncertainty is statedfor their reported resistivity value of 300 µΩcm, and thecorresponding Tc value of 15.1 K would place the Sn con-tent around 23%. Nevertheless, (10) and (11) summarizethe consistent existing literature data on well defined,homogeneous laboratory samples for the main supercon-ducting parameters Tc and Hc2 as function of composi-tion.

V. CHANGES IN Hc2(T ) WITH ATOMIC TIN

CONTENT

To describe the critical currents in wires, not only theupper critical field at zero temperature and zero fieldcritical temperature are important. Also an accurate

7

description of the entire field-temperature phase bound-ary is required. The temperature dependence of the up-per critical field as function of temperature [Hc2(T )] iswell investigated up to about 0.5Hc2, mainly since thefield range up to about 15 T is readily available usingstandard laboratory magnets. Behavior at higher fieldsis often estimated by using an assumed Hc2(T ) depen-dence or by using so called ’Kramer’ [41] extrapolationsof lower field critical current data that rely on an as-sumed pinning mechanism. Estimates of µ0Hc2(0) orKramer extrapolated critical fields [µ0HK(0)] in wiresthat are derived in this way range from 18 T to > 31 T[42, 43, 44, 45, 46, 47, 48, 49, 50, 51]. It is obvious thatthis wide range of claimed Hc2(0) values in wires com-plicates understanding of their behavior. It is thereforerequired to analyze data over the full field range in bet-ter defined samples. Three data-sets which are measuredover nearly the full field-temperature range on well de-fined laboratory specimen exist in the literature whichwill be outlined below. Recent investigations on Hc2(T )over the entire magnetic field range in wires emphasizedon the compositional dependence and the shape of thephase boundary [52, 53]. It was shown that all availableHc2(T ) can be described by one single relation, indicatingthat the shape of the field-temperature phase boundaryis constant and independent of composition, morphol-ogy or strain state. Knowledge of Hc2(0) and Tc(0) thusmeans that the entire phase boundary is known. It wasconcluded that the highest Sn fractions in wires repre-sent an upper phase boundary of µ0Hc2(0) ∼= 29 T andTc(0) ∼= 18 K for ternary wires. The critical currentscales with a lower phase boundary which is an effectiveaverage over the compositional range that is present in awire.

A. Hc2(T ) of cubic and tetragonal phases in single-

and polycrystalline samples

The articles by Foner and McNiff [54, 55] contain dataover the entire magnetic field range on single- and poly-crystals in the cubic and tetragonal phase. In addition,single crystals are investigated in the [100] and [110] di-rections to analyze the anisotropy in Hc2(T ). Theseresults are reproduced in the left plot in figure 8, to-gether with the results of Foner and McNiff on the ap-proximately stoichiometric Arko et al. single crystal [39].Their characterizations were performed using a RF tech-nique and the resulting Hc2(T ) are claimed to correspondto a 50 to 90% resistive criterion. The curves are fitsto the data (in the [100] directions) according to micro-scopic theory assuming a dirty Type II superconductorand no Pauli paramagnetic limiting. The anisotropy inHc2(T ) is 3 to 4%, both in the cubic and in the tetrago-nal phase. The shape of the Hc2(T ) phase boundary andTc(0) are similar despite the 9 T difference in µ0Hc2(0)between the three crystals. The µ0Hc2(0) value for thetetragonal phase is about 4.5 T lower than for the cu-

bic phase. A slightly smaller difference in µ0Hc2(0) ofabout 3.5 T was observed by Foner and McNiff for poly-crystalline sample material in the cubic and tetragonalphase. The reduced Hc2(0) of the (partly?) tetragonalpolycrystalline sample was combined with a 0.1 to 0.2 Krise in Tc(0). Their cubic polycrystalline sample averagedthe [100] and [110] directions of the cubic single crystal.The single- and poly-crystal results of Foner and McNiffcan explain differences in Hc2(0) values for Sn rich A15compounds in terms of differences between the cubic andtetragonal phases.

B. Hc2(T ) of thin films with varying resistivity

Thin film Hc2(T ) results on films with varying resis-tivity are available from Orlando et al. [33, 34] for fieldsup to 23 T. They used a resistive technique with a 50%normal state criterion. Their results are reproduced inthe right plot in figure 8. The curves are fits to the dataaccording to microscopic theory in the absence of Pauliparamagnetic limitations. Orlando et al. quote the re-sistivities of the films. In figure 8 also the compositionsare given, which were calculated from the resistivity val-ues using (5). Lowering the resistivity, or increasing theSn content, increases the field-temperature phase bound-ary. Low resistivity (9 µΩcm), close to stoichiometric(24.8 at.% Sn) A15 has a Tc(0) of close to 18 K, buta reduced µ0Hc2(0) of about 26.5 T. An apparently op-timal dirty thin film (35 µΩcm, 23.6 at.% Sn) reachesµ0Hc2(0) ∼= 29 T at the cost of a reduced Tc(0) ∼= 16 K.The lowest phase transition that was measured occurredfor ρn = 70 µΩcm (β = 21 at.% Sn) at µ0Hc2(0) ∼= 15 Tand Tc(0) ∼= 10.5 K. The shape of the phase boundariesare again similar and independent of composition. TheOrlando et al. results yield explanations for the place-ment of the phase boundary in terms of resistivity andcomposition.

C. Hc2(T ) of bulk materials

New Hc2(T ) for the full field range has recently be-come available on bulk samples from Jewell et al. [12].Their results are reproduced in figure 9. The curves arefits to the data according to microscopic theory usingthe Maki-de Gennes approximation [52]. The depictedcompositions and resistivities were calculated from theTc(0) values using (10) and (5). All samples were mea-sured in a Vibrating Sample Magnetometer (VSM) us-ing the first observable diamagnetic deviation from re-versible normal state paramagnetic behavior as criterionfor Hc2 (which approximately corresponds to a 90% re-sistive criterion [52]), except the β = 23.8 at.% Sn sam-ple which was measured resistively, using a 90% normalstate criterion. All magnetically measured bulk sam-ples were homogenized but not the resistively measuredsample. Its measured resistivity value (extrapolated to

8

0 2 4 6 8 10 12 14 16 18 200

5

10

15

20

25

30

0 2 4 6 8 10 12 14 16 18 200

5

10

15

20

25

30

Arko single crystal

RCA single crystals [100] cubic [110] cubic [100] tetragonal [110] tetragonal

(t)

(c)

Upp

er c

ritic

al fi

eld

[ T ]

Temperature [ K ]

9 µΩcm 24.8%

3523.6

6021.9

n = 70 = 21.0 *

Temperature [ K ]

FIG. 8: The upper critical field versus temperature for single crystals after Foner and McNiff [55]† at approximately a 50 to90% resistive criterion (left plot), and for thin films with varying resistivity after Orlando et al. [34]‡ at a 50% resistive criterion(right plot). *Compositions were calculated from the resistivity values using (5).† c©1981 Pergamon Press Ltd. There are instances where we have been unable to trace or contact the copyright holder or the original authors.If notified the publisher will be pleased to rectify any errors or omissions at the earliest opportunity.‡ c©1981 IEEE. Adapted with kind permission of IEEE and T.P. Orlando.

zero applied magnetic field from ρn(H) data) amounted22 µΩcm. That this value is lower than the calculatedvalue of 29 µΩcm (from Tc) can be attributed to inaccu-racies in voltage tap separation length which were about30% and/or percolation of the measuring current. Theseresults are in qualitative agreement with the thin filmdata of Orlando et al. in the previous section, i.e. thefield-temperature phase boundary increases dramaticallyif the A15 Sn content is increased. The shift of the bound-ary with composition appears proportional, i.e. with-out the tilt that occurred in the Orlando et al. resultsbetween the low resistivity and optimal dirty thin film.Again, the shape of the phase boundary is independentof composition.

D. Concluding remarks

The three Hc2(T ) data sets on defined laboratory sam-ples are highly consistent. Bulk sample materials are ar-guably the closest representation of A15 layers in wires.More experiments on bulk materials will therefore be wel-come to clarify the inconsistency between the Tc(β) datasets of Moore et al. and Devantay et al. and to gener-ate a larger quantity and more accurate Hc2(β) databaseusing the these days readily available high field magnetsystems. The existing data can be summarized by plot-ting their extremes [Hc2(0) and Tc(0)] as function of com-position as was done in figure 7. This is valid since theshape of the phase boundary is constant and indepen-dent of composition (and thus independent of the electronphonon interaction strength), or whether the material isin the cubic or the tetragonal phase [52, 53]. The entire

0 2 4 6 8 10 12 14 16 18 200

5

10

15

20

25

30

35

20.7%(1)

72 µΩcm(2)

21.7%(1)

63 µΩcm(2)

23.8%(1)

29 µΩcm(2) / 22 µΩcm(3)

= 24.5%(1)

n = 15 µΩcm(2)

U

pper

crit

ical

fiel

d [ T

]

Temperature [ K ]

FIG. 9: Upper critical field as function of temperature on ho-mogenized bulk samples with varying composition after Jew-ell et al. [12]† at approximately a 90% resistive criterion. (1)Compositions calculated from Tc(0) using (10). (2) Resistivi-ties calculated from compositions (1) using (5). (3) Resistivityextrapolated from ρn(H) data.†

c©2004 The American Institute of Physics. Adapted with kind per-mission of the American Institute of Physics and M.C. Jewell.

field-temperature dependence of the Nb-Sn can thus besummarized to depend solely on composition using (10)and (11) as was recently postulated [52].

9

VI. COPPER, TANTALUM AND TITANIUM

ADDITIONS TO THE A15

A. Copper additions

Wires are different from pure binary Nb-Sn systemssince they are always at least ternary due to the presenceof Cu. The addition of Cu lowers the A15 formation tem-perature, thereby limiting grain growth and thus retain-ing a higher grain boundary density which is required forpinning (see section VII). Understanding of how Cu cat-alyzes the reaction requires detailed knowledge of ternaryphase diagrams. Although ternary phase diagrams of theNb-Sn-Cu system are sparsely available in the literature(mostly at fixed temperatures) [56, 57, 58, 59, 60, 61],they do not complement and in regions contradict eachother. It can therefore be argued that the exact influ-ence of Cu additions is not thoroughly known. It is clearthat the presence of Cu destabilizes the line compoundsNbSn2 and Nb6Sn5 [7, 62]. It was further suggested thatthe presence of the Nb-Sn melt is extended to far below930 C, enabling rapid A15 formation [59]. Very low tem-perature A15 formation has been recorded at tempera-tures down to about 450 C in a ternary system contain-ing 5.4 at.% Cu [59]. Although Cu can be detected inthe A15 layers [63, 64], it is generally assumed to existonly at the grain boundaries and not to appear in theA15 grains [65, 66]. The absence of Cu in the A15 grainsmight explain why the binary A15 phase diagram canoften be used to, at least qualitatively, interpret compo-sitional analysis in wires. Also, to first order, the additionof Cu does not dramatically change the superconductingbehavior of wires as compared to binary systems [52].Finally, the presence of Cu was recently stated to be apossible origin for Hc2(0) suppression in wires [12].

B. Titanium and Tantalum additions

Possibilities to improve the superconducting proper-ties of the Nb-Sn system by the use of third and fourthelement additions have been extensively investigated(see [31, 46, 47, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76,77, 78]) and reviewed [4, 45]. Of a wide range of possibleternary additions Ta and Ti are most widely applied inwires. Both Ta and Ti occupy Nb sites in the A15 lat-tice [79]. The effect of these third elements is to suppressthe low temperature cubic to tetragonal transition for≥ 2.8 at.% Ta or for ≥ 1.3 at.% Ti (see e.g. [4] and the ref-erences therein). Increasing Ta or Ti content has shownto increase deviations from stoichiometry which in itselfis sufficient for stabilization of the cubic phase (figure 7).Retaining the cubic phase results in an increased Hc2(0),while retaining Tc(0) ([55], figure 8). Also the resistiv-ity increases with increasing Ta or Ti additions [4] whichis often referred to as the main cause for an increase inHc2(0). The latter is, according to the GLAG theory inthe dirty limit, proportional to ρn through (Tinkham [80],

p. 162, modified using 1/νFℓ = 1/3e2N(EF)ρn):

µ0Hc2 (0) ∼= kBeN (EF) ρnTc (0) =3e

π2kBγρnTc (0) ,

(12)where νF, ℓ and e represent the velocity of the electronsat the Fermi level, the electron mean free path and theelementary charge quantum respectively and γ is the elec-tron specific heat constant or Sommerfeld constant. Or-lando et al. [34] showed for pure binary films, that Hc2(0)[and Hc2(4.2 K)] both increase with resistivity, peak atρn

∼= 25 µΩcm and then start to decrease. Similarly,Hc2(4.2 K), as measured in alloyed wires [45] initially in-creases with Ti or Ta additions, peaks at about 1.5 at.%Ti or 4 at.% Ta, and then reduces with increasing alloycontent. The Tc(0) in the pure binary films is about 18 Kin the clean limit and initially not influenced by increas-ing resistivity. However, Tc(0) starts to decrease aboveρn

∼= 30 µΩcm. At ρn = 35 µΩcm, µ0Hc2(0) is still highat 29 T, but Tc(0) has lowered to 16 K (figure 8). Thecritical temperature in alloyed wires peaks only slightlyat about 1 at.% Ti and 2 at.% Ta. A clear reduction inTc(0) occurs above about 1.5 at.% Ti and 3.5 at.% Ta.

The above summary shows that the effects of Ta andTi additions are similar but Ti is approximately a factortwo more effective in increasing the resistivity. Indeed1.3 at.% Ti addition results in ρn

∼= 40 µΩcm whereas2.8 at.% Ta addition results in ρn

∼= 30 µΩcm [4]. Con-sistent overall behavior emerges when the initial increasein Hc2(0) [while retaining a constant Tc(0)] with alloy-ing, can be attributed to a suppression of the tetragonalphase, and above ρn

∼= 30 µΩcm (or about 1.5 at.% Tiand 3.5 at.% Ta) the increased resistivity suppresses bothHc2(0) and Tc(0). Note that this suggestion is contraryto the general assumptions in the literature that Hc2(0)rises as a result of increased resistivity due to its propor-tionality with ρn. It can be argued that such an assump-tion represents an oversimplification, since it neglects theoccurrence of the ternary phase and contradicts combina-tion of the well established datasets of ρn(β), Hc2(β) andHc2(T ): Increasing the resistivity at compositions below24.5 at.% Sn strongly reduces Hc2(0) as follows from rightplot in figure 7, since increasing ρn is equivalent to reduc-ing the Sn content (figure 5). The latter can be physicallyunderstood in terms of a reduction of the d-band peaksand thus in N(EF) in (12), which influences Hc2(0) di-rectly, as well as through λep in Tc(0) [see (7)]. This cancounteract the increase of Hc2(0) through ρn. The binarysummary (figure 7) remains consistent, at least qualita-tively, with Ta or Ti ternary additions if it is assumedthat their main effect is to push the composition increas-ingly off-stoichiometric and simultaneously increasing theresistivity and preventing the tetragonal transformation.It should be emphasized that this argument is not wellsupported, since all knowledge on ternary systems withTa and Ti stems from inhomogeneous wires as opposedto homogenized laboratory samples.

10

VII. GRAIN SIZE AND THE MAXIMUM

BULK PINNING FORCE

A15 Nb-Sn behaves as a Type-II superconductor. Itscurrent carrying capabilities for any practical field (aboveµ0Hc1

∼= 38 mT [38]) thus rely on its capability to pin theflux-lines. The bulk pinning force at Jc can be calculatedfrom its critical balance with the Lorentz force:

FP ≡ −Jc × B. (13)

The bulk pinning force depends on the magnetic field andmostly has a maximum which is positioned roughly be-tween 0.1 to 0.4Hc2 [81, 82]. The magnetic field value atwhich the maximum occurs depends on the details of theflux-line to lattice interactions [41, 81, 83]. From (13) it isclear that the bulk pinning force determines the conduc-tors’ current carrying capabilities. The general assump-tion is that the grain boundaries in the A15 phase repre-sent the primary pinning centers. This is supported bystrong experimental evidence in which the maximum pin-ning force is related to the reciprocal grain size and thusthe grain boundary density [84, 85, 86, 87, 88]. Theseresults were summarized by Fischer [89], who also ob-tained additional results for the maximum bulk pinningforce versus grain size from VSM characterizations. Areproduction after Fischer, including his own additionsis shown in figure 10. It remains unresolved how grainboundary pinning in Nb3Sn occurs in detail. It is ob-vious that minima are present in the superconductingwave-function at the grain boundaries but it is unclearwhat physical principle (the presence of Cu, morphology,etc.) constitutes to this. Investigating what is present atthe grain boundary might resolve some of this questionand recent efforts are made in this respect [90].

The calculated value for FP from transport current de-pends on the area for Jc, which can be normalized to e.g.the non-Cu area or the A15 area. Most pinning forcedata from the literature are calculated from the non-CuJc and figure 10 shows that all pre-2002 results are con-sistent. They follow a single line on a semi-logarithmicplot summarized with:

FP,max = 22.7 ln

(

1

dav

)

− 10[

GN/m3]

, (14)

in which dav represents the average grain size in µm. Ad-ditionally the maximum bulk pinning force data from Fis-cher, calculated from VSM derived non-Cu Jc values areincluded in the graph (black squares). These results ap-proximately correspond to the earlier literature data. If,however, the values obtained by Fischer are calculatedfrom Jc values which are normalized to the A15 area,the calculated maximum pinning force becomes higherdue to the increased Jc values. The slope in the semi-logarithmic plot effectively increases and Fischer’s A15area results can thus be summarized by:

FP,max = 39.2 ln

(

1

dav

)

− 10[

GN/m3]

, (15)

100 1010

10

20

30

40

50

60

70

80

901000 100

39.2 ln(1 / dav ) − 10

22.7 ln(1 / dav ) − 10

Scanlan 1975 Shaw 1976 Marken 1986 Schauer 1981 West 1977 Fischer non-Cu 2002 Fischer A15 2002

Max

imum

pin

ning

forc

e [ G

N/m

3 ]

Reciprocal grain size [ µm-1 ]

25 Average grain size [ nm ]

FIG. 10: Maximum pinning force as function of reciprocalgrain size after Fischer [89].Adapted with kind permission of C.M. Fischer.

in which again dav is in µm.The most objective inter-wire comparisons can be

made when the pinning force is normalized to the avail-able grain boundary length resulting in an effective pin-ning force QGB [89]:

QGB =

(

1

η

) (

FP

GBl/A

)

, (16)

in which η represents an efficiency normalization (as-sumed 1/3 for equiaxed grains) and GBl/A representsthe grain boundary length per area. The effective pinningforce can be interpreted as a measure of the effective pin-ning strength of a grain boundary. Calculation of (16),however, requires an accurate determination of GBl/A

which, although thoroughly analyzed by Fischer [89], isoften not available.

The optimum grain size to achieve the ideal limit ofone pinning site per flux-line can be estimated, sincethe grain boundaries are the primary pinning centers inNb3Sn. This limit will be approached if the grain sizeis comparable to the flux-line spacing at a given field.The flux-line spacing can be calculated by assuming atriangular flux-line lattice [80]:

a (H) =

(

4

3

)1/4 (

φ0

µ0H

)1/2

, (17)

which results in 14 nm at 12 T. The smallest grain sizesin figure 10 are about 35 nm but commercial wires havegrain sizes of 100 to 200 nm and are thus far from opti-mized. In NbTi in contrast, the ideal limit of one flux-line per pinning site has been closely approached through

11

specific heat treatments that introduce a fine distribu-tion of normal conducting alpha-Ti precipitates that actas pinning sites [91]. NbTi is therefore (under presentunderstanding) close to fully optimized.

VIII. VARIATIONS OF THE

SUPERCONDUCTING PROPERTIES WITH

STRAIN

A. Microscopic origins of strain dependence

The superconducting properties of all materials are toa certain extent sensitive to strain, although the effectcan be marginal as for example for NbTi. The strain in-fluences on the critical current in the Nb-Sn system origi-nate from shifts in the field-temperature phase boundary,i.e. Jc varies through a change in Hc2(T ) and possibly adirect influence on the maximum bulk pinning force, asis detailed elsewhere [82]. Strain influences can be sepa-rated into two regimes. Obviously, large strain can evokeserious and irreversible damage to the superconductor inthe form of cracks in the brittle A15 material [92]. Lowerstrain can vary the superconducting properties throughstrain induced lattice instabilities that result in a tetrag-onal distortion [93, 94] and by influencing the electron-electron interactions, which can both be reversible. Thelatter can, in the case of A15 Nb-Sn, be discussed interms of lattice distortions that influence the Nb chainintegrity and thus N(EF), or in terms of changes in lat-tice vibration modes which influence the electron-phononinteraction spectrum [which includes changes in N(EF)].

Since distorting the lattice will modify N(EF), it canbe argued that the net effect is similar to an increasein the amount of disorder. This would render strain ef-fects to be qualitatively similar to irradiation damage,or also similar to Sn deficiency if ρn(S) (figure 4) andρn(β) (figure 5) can be considered of identical origin (sec-tion II). Snead and Suenaga [95] have performed irradia-tion experiments on A15 filaments with and without pre-strain and concluded that irradiation increases the effectof pre-strain on Tc. This supports the above hypoth-esis, since the fact that pre-strain and irradiation am-plify each other could imply that these effects are basedon a similar microscopic principle [i.e. a modificationof N(EF)]. Reversing this line of thought it can be ar-gued that increasing the amount of disorder in a highlyordered system will have a relatively larger effect on itssuperconducting properties compared to a system whichis already substantially disordered. Flukiger et al. col-lected strain sensitivity data of the critical current forvarious A15 wires with a varying degree of ordering andsuggested to attribute changes in strain sensitivity to thedegree of ordering [26]. His collection is reproduced in fig-ure 11. Specifically, when comparing Nb3Sn (S = 1) andNb3Al (Sa = 0.95), the latter being always decidedly off-stoichiometric due to the instability of the stoichiometricphase at lower temperatures, the differences are large,

-0.5 0.0 0.5 1.0 1.50.4

0.5

0.6

0.7

0.8

0.9

1.0 0.57 ≤ H / Hc2* ≤ 0.68

V3Si (S = 1)Nb3Sn (S = 1)

Nb3Sn + (Ni, Zn)

V3Ga (S = 0.98)Nb3Al (Sa = 0.95)

Nb3Au (Sa = 0.93)

Nor

mal

ized

crit

ical

cur

rent

Intrinsic axial strain [ % ]

FIG. 11: Strain sensitivity of the critical current for differentA15 superconductors with varying amounts of disorder afterFlukiger et al. [26]†.† c©1984 Plenum Press. Adapted with kind permission of SpringerScience and Business Media and R. Flukiger.

despite roughly comparable Hc2 and Tc values. If theabove argument is correct, it can be expected that strainsensitivity of the A15 Nb-Sn system depends on composi-tion. Supporting this hypothesis is the fact that in axialstrain experiments on wires, the strain sensitivity of thecritical current increases with increasing magnetic field ortemperature (see e.g. [82] and the references therein). Athigher magnetic field and temperature, the lower Sn A15fractions are no longer superconducting. With increas-ing magnetic field or temperature, the observed straindependence is more and more determined by the higherSn A15 fractions in the wires. These results thereforealso represent a suggestion for enhanced strain sensitiv-ity at higher Sn content. Whether this indeed is trueis important for practical applications, since the presenttrend to improve the performance of commercial wiresby Sn enrichment in the A15 could lead to an undesiredincreased strain sensitivity.

Strain sensitivity in terms of variations in electron-phonon interaction strength can be explained within theweak coupling BCS theory and also in the more gen-eral Eliashberg formalism by a pressure dependence ofTc(0) [96]. However, strong indications exist throughoutthe literature that hydrostatic strain has only a negligi-ble influence in comparison to non-hydrostatic or distor-tional strain components [96, 97, 98]. Following theseinitial publications, which were based on strong cou-pling renormalized BCS theory, a recent attempt hasbeen made to couple a microscopic full strain invari-ant analysis to Eliashberg-based relations for the criticaltemperature through strain-induced modifications in theelectron-phonon spectrum [99, 100, 101]. This analysis isdiscussed in more detail elsewhere [53, 82]. These recentdescriptions are based on Tc variation through a changein the electron-phonon interaction constant by a changein the lattice vibration modes. The latter is modeled by

12

strain induced modifications to the frequency dependentelectron-phonon interaction (in which N(EF) is implicit)and phonon DOS. These models were verified using liter-ature results for Tc versus hydrostatic strain on a singlecrystal and Tc versus axial strain on wires and tapes. Itshould be emphasized, however, that for a precise de-scription of the strain influences also the strain depen-dence of Hc2(T ) should be correlated to the microscopictheory.

Strain induced changes in N(EF) in the full invariantanalysis for Tc are only accounted for through a strain-modified frequency dependence of the electron-phononinteraction, i.e. the change in N(EF) is not directly cal-culated, but this apparently yields sufficient accuracy.For a microscopic description of the strain dependence ofHc2(T ), however, the exact change in N(EF) with strainshould be known, as is evident from (12). This direct in-fluence of a change in N(EF) on Hc2(0) would hypothet-ically result in a higher strain sensitivity for Hc2(0) com-pared to Tc(0). This is verified by experimental resultswhere the relative strain dependence of Hc2(0) is roughlya factor 3 times higher than for Tc(0) (see e.g. [82] andthe references therein). This enhanced strain sensitiv-ity of Hc2(0) compared to Tc(0) was not satisfactory ex-plained using strong coupling corrected BCS theory [98].Promising recent work by Oh and Kim [102], that utilizesthe interaction strength independent Eliashberg theory,yields a more satisfactory explanation, although prefer-ably this should be combined with Markiewicz’s his fullinvariant analysis to arrive at a complete description forthe strain dependence of the superconducting propertiesof Nb3Sn.

B. Available strain experiments on well defined

samples

The previous sections showed that the main supercon-ducting properties (Tc and Hc2) are well investigated onlaboratory samples with defined composition. The avail-able literature presents a clear and consistent compo-sition dependence which differs only in details betweenthe various results. Strain sensitivity experiments on de-fined laboratory samples are, in contrast to this, seri-ously lacking. Abundant results are available on techni-cal wires which focus on the influences of axial strain onthe critical current density (see e.g. the pioneering workby Ekin [103] and the references in [82]). Wires are,however, due to their fabrication processes, always de-cidedly inhomogeneous in composition as well as strain.The foregoing discussions on the possible microscopicorigins of strain dependence emphasized the need forstrain dependence results, obtained from samples witha fixed and known composition, since both composi-tion and strain influence Hc2(T ). Results on labora-tory specimen with defined composition are limited tohydrostatic experiments on single crystals or polycrys-talline layers (e.g. [21, 104]) and one uniaxial experi-

ment, also on a single crystal [105]. These result indTc/dP = −0.14 and −0.22 K/GPa for a single crys-tal [21] and a polycrystalline layer [104] respectively. Oneresult exists for the pressure dependence of Hc2 on aninhomogeneous tape conductor [96], which resulted indµ0Hc2/dP = −1.2 ± 0.2 T/GPa.

Next to inhomogeneities, also the three dimension-ality of strain complicates the analysis in wires. TenHaken [96] made attempts to overcome this problemby switching to deformation experiments on two dimen-sional tape conductors. Although this has resulted inmany new experimental insights that emphasize the im-portance of non-hydrostatic strain components, the tapeconductors used in his research still suffered from com-positional imperfections.

The lack of extensive strain experiments on well de-fined laboratory samples thus limits the ability to de-velop a description for Tc and Hc2 [or more specificallyHc2(T )] versus strain, due to the possible different strainsensitivity with composition. Such results are, however,required to accurately analyze wire behavior. The datasets needed for this would preferably be analyzed in sim-plified model samples for which the three dimensionalstrain can be calculated and which separate the straindependencies for well defined compositions.

IX. CONCLUSIONS

Superconductivity in the A15 Nb-Sn intermetallicis microscopically well described in general terms ofelectron-phonon coupling of Cooper pairs. Strong cou-pling corrections to the microscopic theory become nec-essary above 23 at.% Sn. Below this concentration thematerial is weak coupling. The A15 lattice contains Nbchains in which the atoms are closely spaced resulting inan increased density of states around the Fermi level,which is believed to be the main origin for the highcritical temperature. The binary phase diagram is wellestablished although a preference for the formation ofclose to stoichiometric A15 can be argued. Effects ofoff-stoichiometry can be discussed in terms of lattice pa-rameter, long range ordering, resistivity or atomic Sncontent. These terms are equivalent and consistent be-havior is observed throughout the literature. The field-temperature phase boundary varies substantially with Sncontent, but the observations can be explained in termsof off-stoichiometry and the occurrence of the tetragonalphase close to stoichiometry. The dependence of the crit-ical temperature and upper critical field on compositionalvariations varies only in detail throughout the literatureand is well established for the binary system. This consis-tent behavior arguably also remains valid when ternaryadditions such as Cu, Ti or Ta are present.

Consistent ternary or quaternary phase diagrams arenot available but the pure binary phase diagram can beused, at least qualitatively for the A15 phase, to describecompositional variations in higher order systems. The

13

addition of Cu allows for lower temperature A15 forma-tion. It is present at A15 grain boundaries but does notappear to exist in the A15 grains. Additions of Ti andTa occupy Nb sites and apparently result in distortions ofthe Nb chain integrity, cause off-stoichiometry, increasethe resistivity and prevent the low temperature tetrago-nal transformation.

Abundant and consistent data on the bulk pinningforce as function of reciprocal average grain size representsolid experimental evidence that the grain boundaries arethe main pinning centers. The exact mechanism that de-termines the pinning strength at the grain boundaries re-mains unresolved. A consistent relation emerges betweenthe maximum bulk pinning force, calculated from thecritical current density normalized to the non-Cu area,and the average grain size. This maximum bulk pinningforce, combined with the field-temperature phase bound-ary and strain effects, determines the attainable criticalcurrent density.

The influence of hydrostatic deformations can be qual-itatively predicted from microscopic theory and limiteddata from hydrostatic deformation experiments are avail-able. Distortional (non-hydrostatic) deformations, how-ever, dominate the strain sensitivity. Although defor-mation experiments on wires are abundant, no generalstatements can be made. This is due to the lack of non-hydrostatic deformation experiments on homogeneous,

and spatially simplified samples since it is plausible thatstrain sensitivity varies with composition. Very recentwork opens new possibilities towards a full, strong cou-pling theory based description of strain sensitivity of thesuperconducting properties in the Nb-Sn system.

Acknowledgments

Most of the material in this article was collected for theintroduction chapters of my PhD thesis during my stayat the Applied Superconductivity Center, University ofWisconsin-Madison, USA and at the Low TemperatureDivision, the Special Chair for Industrial Application ofSuperconductivity at the Faculty of Science and Technol-ogy, University of Twente, Enschede, The Netherlands.I would like to thank Bennie ten Haken, Herman tenKate and David Larbalestier for their rigorous correc-tions to the original text. This work was supported by theNetherlands Organization for Scientific Research (NWO)through the Technology Foundation STW and by the Di-rector, Office of Science, High Energy Physics, U.S. De-partment of Energy under Contracts DMR–9632427 andDE–AC02–05CH11231. All contributions are greatly ac-knowledged.

[1] B. Matthias, T. Geballe, S. Geller, and E. Corenzwit,Phys. Rev. 95, 1435 (1954).

[2] G. Hardy and J. Hulm, Phys. Rev. 89, 884 (1953).[3] J. Hanak, K. Strater, and R. Cullen, RCA Review 25,

342 (1964).[4] R. Flukiger, Atomic ordering, phase stability and su-

perconductivity in bulk and filamentary A15 type com-pounds, Technical report, Kernforschungszentrum Karl-sruhe, 1987.

[5] G. Stewart, L. Newkirk, and F. Valencia, Phys. Rev. B21, 5055 (1980).

[6] J. Charlesworth, I. MacPhail, and P. Madsen, J. Mater.Sci. 5, 580 (1970).

[7] S. Foner and B. Schwartz, editors, Superconductor Ma-

terial Science, Plenum Press, New York, 1981.[8] R. Enstrom, J. Appl. Phys. 37, 4880 (1966).[9] J. Charlesworth, Phys. Lett. 21, 501 (1966).

[10] D. van Ooijen, J. van Vucht, and W. Druyvesteyn, Phys.Lett. 3, 128 (1962).

[11] H. Devantay, J. Jorda, M. Decroux, J. Muller, andR. Flukiger, J. Mater. Sci. 16, 2145 (1981).

[12] M. Jewell, A. Godeke, P. Lee, and D. Larbalestier, Adv.Cryo. Eng. (Materials) 50B, 474 (2004).

[13] L. Vieland, RCA Review 25, 366 (1964).[14] C. Toffolon, C. Servant, and B. Sundman, J. Phase

Equil. 19, 479 (1998).[15] C. Toffolon, C. Servant, J. Gachon, and B. Sundman,

J. Phase Equil. 23, 134 (2002).[16] R. Schiffman and D. Bailey, High Temp. Sci. 15, 165

(1982).

[17] R. Flukiger, Adv. Cryo. Eng. (Materials) 28, 399(1982).

[18] D. Dew-Huges, Cryogenics 15, 435 (1975).[19] R. Mailfert, B. Batterman, and J. Hanak, Phys. Lett.

A 24, 315 (1967).[20] R. Mailfert, B. Batterman, and J. Hanak, Physica Sta-

tus Solidi 32, k67 (1969).[21] C. Chu and L. Vieland, J. Low Temp. Phys. 17, 25

(1974).[22] L. Vieland, J. Phys. Chem. Solids 31, 1449 (1970).[23] A. Junod, J. Muller, H. Rietschel, and E. Schneider, J.

Phys. Chem. Solids 39, 317 (1978).[24] L. Vieland and A. Wicklund, Phys. Lett. A 34, 43

(1971).[25] M. Straumanis and S. Zyszczynsk, J. Appl. Cryst. 3, 1

(1970).[26] R. Flukiger, R. Isernhagen, W. Goldacker, and

W. Specking, Adv. Cryo. Eng. (Materials) 30, 851(1984).

[27] R. Flukiger, H. Kupfer, J. Jorda, and J. Muller, IEEETrans. Magn. MAG-23, 980 (1987).

[28] M. Beasley, Adv. Cryo. Eng. 28, 345 (1982).[29] D. Welch, G. Dienes, O. Lazareth Jr., and R. Hatcher,

J. Phys. Chem. Solids 45, 1225 (1984).[30] C. Nolscher and G. Seamann-Ischenko, Phys. Rev. B

32, 1519 (1985).[31] E. Drost, W. Specking, and R. Flukiger, IEEE Trans.

Magn. MAG-21, 281 (1985).[32] R. Flukiger, in Superconductivity in d- and f-Band

Metals, edited by W. Buckel and W. Weber, Kern-

14

forschungszentrum Karlsruhe, 1982.[33] T. Orlando, E. McNiff Jr., S. Foner, and M. Beasley,

Phys. Rev. B 19, 4545 (1979).[34] T. Orlando et al., IEEE Trans. Magn. MAG-17, 368

(1981).[35] J. Bardeen, L. Cooper, and J. Schrieffer, Phys. Rev.

108, 1175 (1957).[36] G. Eliashberg, Zh. Eksperim. i Teor. Fiz. 38, 966 (1960),

[Sov. Phys. JETP 11, 696 (1960)].[37] D. Moore, R. Zubeck, J. Rowell, and M. Beasley, Phys.

Rev. B 20, 2721 (1979).[38] V. Guritanu et al., Phys. Rev. B 70, 184526 (2004).[39] A. Arko et al., Physical Review Letters 40, 1590 (1978).[40] L. Cooley, Y. Hu, and A. Moodenbaugh, Enhancement

of the upper critical field of Nb3Sn utilizing disorderintroduced by ball milling the elements, Submitted toAppl. Phys. Lett., 2006.

[41] E. Kramer, J. Appl. Phys. 44, 1360 (1973).[42] M. Suenaga and A. Clark, editors, Filamentary A15

Superconductors, Plenum Press, New York, 1980.[43] D. Kroeger, D. Easton, A. DasGupta, C. Koch, and

J. Scarbrough, J. Appl. Phys. 51, 2184 (1980).[44] K. Hechler, G. Horn, G. Otto, and E. Saur, J. Low

Temp. Phys. 1, 29 (1969).[45] M. Suenaga, D. Welch, R. Sabatini, O. Kammerer, and

S. Okuda, J. Appl. Phys. 59, 840 (1986).[46] R. Akihama, K. Yasukochi, and T. Ogasawara, IEEE

Trans. Magn. MAG-13, 803 (1977).[47] K. Tachikawa, T. Asano, and T. Takeuchi, Appl. Phys.

Lett. 39, 766 (1981).[48] J. Lindenhovius, E. Hornsveld, A. den Ouden, W. Wes-

sel, and H. ten Kate, IEEE Trans. Appl. Supercond.10, 975 (2000).

[49] B. ten Haken, A. Godeke, and H. ten Kate, J. Appl.Phys. 85, 3247 (1999).

[50] N. Cheggour and D. Hampshire, Cryogenics 42, 299(2002).

[51] A. Godeke, M. Jewell, A. Golubov, B. ten Haken, andD. Larbalestier, Supercond. Sci. and Techn. 16, 1019(2003).

[52] A. Godeke et al., J. Appl. Phys. 97, 093909 (2005).[53] A. Godeke, Performance Boundaries in Nb3Sn Super-

conductors, PhD thesis, Univ. of Twente, Enschede,The Netherlands, 2005.

[54] S. Foner and E. McNiff Jr., Phys. Lett. 58A, 318 (1976).[55] S. Foner and E. McNiff Jr., Solid State Commun. 39,

959 (1981).[56] R. Hopkins, G. Roland, and M. Daniel, Metallurgical

Trans.: Phys. Metallurgy and Mat. Sci. 8A, 91 (1977).[57] U. Zwicker and L. Rinderer, Zeitschrift f 66, 738 (1975).[58] W. Neijmeijer and B. Kolster, Zeitschrift f 78, 730

(1987).[59] G. Lefranc and A. Muller, J. Less Common Metals 45,

339 (1976).[60] N. Ageeva and L. Petrovoj, in Diagrammy Sostoyaniya

Metallicheskikh Sistem, edited by N. Ageeva, volume 28,page 26, Viniti, Moscow, 1982.

[61] W. Neijmeijer, Microstructural and Kinetic Studies of

the Manufacturing of Superconducting Nb3Sn, PhD the-sis, Univ. of Twente, Enschede, The Netherlands, 1988.

[62] D. Smathers, in Metals Handbook, volume 2, page 1060,ASM International, Metals Park, OH, 1990.

[63] J. Livingston, Phys. Status Solidi A 44, 295 (1977).[64] D. Smathers and D. Larbalestier, Adv. Cryo. Eng. (Ma-

terials) 28, 415 (1982).[65] M. Suenaga and W. Jansen, Appl. Phys. Lett. 43, 791

(1983).[66] M. Suenaga, 2004, Private communication on unpub-

lished results.[67] J. Livingston, IEEE Trans. Magn. MAG-14, 611

(1978).[68] E. Springer, M. Wilhelm, H. Weisse, and G. Rupp, Adv.

Cryo. Eng. (Materials) 30, 747 (1984).[69] M. Suenaga, IEEE Trans. Magn. MAG-21, 1122

(1985).[70] K. Tachikawa, T. Takeuchi, T. Asano, Y. Iijima, and

H. Sekine, Adv. Cryo. Eng. (Materials) 28, 389 (1982).[71] D. Dew-Hughes, IEEE Trans. Magn. MAG-13, 651

(1977).[72] K. Togano, T. Asano, and K. Tachikawa, J. Less Com-

mon Metals 68, 15 (1979).[73] M. Suenaga, T. Luhman, and W. Sampson, J. Appl.

Phys. 45, 4049 (1974).[74] T. Takeuchi, T. Asano, Y. Iijima, and K. Tachikawa,

Cryogenics 21, 585 (1981).[75] H. Sekine et al., IEEE Trans. Magn. MAG-19, 1429

(1983).[76] I. Wu, D. Dietderich, J. Holthuis, W. Hassenzahl, and

J. Morris, IEEE Trans. Magn. MAG-19, 1437 (1983).[77] H. Kurahashi et al., IEEE Trans. Appl. Supercond. 15,

3385 (2005).[78] V. Abacherli et al., IEEE Trans. Appl. Supercond. 15,

3482 (2005).[79] J. Tafto, M. Suenaga, and D. Welch, J. Appl. Phys. 55,

4330 (1984).[80] M. Tinkham, Introduction to Superconductivity,

McGraw-Hill, New York, 2nd edition, 1996.[81] L. Cooley, P. Lee, and D. Larbalestier, Adv. Cryo. Eng.

(Materials) 48B, 925 (2002).[82] A. Godeke, B. ten Haken, H. ten Kate, and D. Lar-

balestier, A general scaling relation for the critical cur-rent density in Nb3Sn wires, Submitted to Supercond.Sci. and Techn., 2006.

[83] D. Dew-Hughes, Phil. Mag. 30, 293 (1974).[84] R. Scanlan, W. Fietz, and E. Koch, J. Appl. Phys. 46,

2244 (1975).[85] B. Shaw, J. Appl. Phys. 47, 2143 (1976).[86] K. Marken, Characterization Studies of Bronze-Process

Filamentary Nb3Sn Composites, PhD thesis, Univ. ofWisconsin-Madison, 1986.

[87] W. Schauer and W. Schelb, IEEE Trans. Magn. MAG-

17, 374 (1981).[88] A. West and R. Rawlings, J. Mater. Sci. 12, 1862 (1977).[89] C. Fischer, Investigation of the relationships between

superconducting properties and Nb3Sn reaction condi-tions in powder-in-tube Nb3Sn conductors, Master’sthesis, Univ. of Wisconsin-Madison, 2002.

[90] R. Flukiger, Presented at the MT-17 conference, Genoa,Italy, 2005.

[91] P. Lee and D. Larbalestier, Wire Journal International36, 61 (2003).

[92] M. Jewell, P. Lee, and D. Larbalestier, Supercond. Sci.and Techn. 16, 1005 (2003).

[93] E. Savitskii, M. Bychkova, V. Baron, B. Dzevitskii, andN. Savateev, Phys. Status Solidi A 37, K165 (1976).

[94] R. Flukiger, W. Schauer, W. Specking, B. Schmidt,and E. Springer, IEEE Trans. Magn. MAG-17, 2285(1981).

15

[95] C. Snead and M. Suenaga, Appl. Phys. Lett. 37, 659(1980).

[96] B. ten Haken, Strain effects on the critical properties of

high-field superconductors, PhD thesis, Univ. of Twente,Enschede, The Netherlands, 1994.

[97] L. Testardi, Phys. Rev. B 3, 95 (1971).[98] D. Welch, Adv. Cryo. Eng. (Materials) 26, 48 (1980).[99] W. Markiewicz, Cryogenics 44, 767 (2004).

[100] W. Markiewicz, Cryogenics 44, 895 (2004).[101] W. Markiewicz, IEEE Trans. Appl. Supercond. 15, 3368

(2005).[102] S. Oh and K. Kim, J. Appl. Phys. 99, 033909 (2006).[103] J. Ekin, Adv. Cryo. Eng. (Materials) 30, 823 (1984).[104] K. Lim and J. Thompson, Phys. Rev. B 27, 2781 (1983).[105] J. McEvoy, Physica 55, 540 (1971).