1-' -OL. 19 No. 5 OCTO BER 1964

Philips Research Reports1

EDITED BY THE 'RESEARCH LABORATORYOF N.V. PHILIPS' GLOEILAMPENFABRIEKEN. EINDHOVEN. NETHERLANDS

517 Philips Res. Repts 19, 407-421, 1964

THE SYSTEM VANADIUM-GALLIUM

by J. H. N. van VUCHT, H. A. C. M. BRUNING,

H.C. DONKERSLOOT and A. H. GOMES de MESQUITA

Abstract. ~It is found that vanadium and gallium form four intermetallic corn-pounds: V3Ga, VsGas, VOGa7, and V2GaS. Stability ranges both intemperature and composition of all phases present in the system werestudied by means of metallographic and X-ray-diffraction methods andby thermal analysis. Their superconductive critical temperatures werealso measured. It is shown that VeGas is structurally related to a com-pound NboSns. A compound TioSns is found to have two allotropieforms one of which is isomorphous to VeGas and the other to NboSns.A determination of this latter structure is given.

• Introduction

One of the most interesting superconductors today is undoubtedly the inter-etallic compound V3Ga. It was discovered in 1956 by Matthias 1), who alsoetermined its crystal structure to be of the Cr3Si (A15 or beta-tungsten) type.his structure has now long been recognized as the most favourable for highperconductive transition points. In fact the transition point of V3Ga is6,8, "K, Most promising, however, seems to be its critical magnetic field.rom measurements ofWernick et al. 2) on V2.95Ga, an extrapolation to 0 "Kads to a value of 350 or 500 kOe, dependent on whether one extrapolatesnearly or parabolically. .Apart from the above papers and a report on the variation of the transitionbint in vanadium-gallium alloys with gallium content between 20 and 35 at. %,o further published data about the system came to our knowledge until aftere contents of this paper, the results of a year's investigation, had been com-unicated at the Spring Meeting of the Institute of Metals in March 1964 in.ondon, Then, by a paper of Savitskii 3) our attention was drawn to two listsf data of intermetallic compounds given by Schubert and several coopera-ors 4,5), where the compounds described here already had been mentioned.n the meantime our preliminary results about these compounds were alreadyublished in compact form 6) without knowing - and therefore withoutentioning - Schubert's data.Since the aim of this paper is to report fully and in detail the results on the

ystem, all the literature on the subject which has become available up to now

I40S JoHoNo van VUCHT. HoAoCoMo BRUNING. HoCo DONKERSLOOT and AoHo GOMES de MESQUIT~

will be included and discussed. Nevertheless we first propose a phase diagrabased on our own observations and then shall discuss some of its featuresconnection with the findings of other authors.

2. Experimental

The starting materials were vanadium granules from Mackay Inc., containii(in per cent by weight) 0·2 Si, 0.025 Fe, 0·01 Mg, 0·06 AI, 0.005 Mn and galium from A.I.A.G. of 99·99% purity.

Weighed mixtures of vanadium and gallium were heated by high-frequentinduction in alumina crucibles in an atmosphere of purified argon (passethrough magnesium at 600 0C). It was possible to increase or decrease aut,matically the temperature of the specimen virtually linearly with time. In thway a thermal analysis could be performed of each sample. At the higher vandium contents W /W -30 % Re thermocouples were used, at contents lower ths60 at. % vanadium Pt/Pt-l0% Rh was used. It was feared initially that durirthe melting of the specimens at the higher V contents some aluminium migldissolve from the crucibleinto the metal and spectrochemical analysis of samphbefore and after melting in Ah03 seemed to corroborate this. However, sampbmelted in Th02 (less liable to react with V) showed no difference in roontemperature properties from those melted in A1203. We conclude therefore thaif Ah03 might dissolve in vanadium at higher temperatures, the result after mtoo rapid a cooling is a precipitated Ah03 in the unreacted vanadium. Stilwhere doubt might arise, thorium-oxide crucibles were always used, e.g. whelattice constants were important.

A specimen melted as described above was always cooled rather slowlAfter cooling down, part of the specimen was crushed to powder (if possiblfor X-ray analysis and in some cases for the measurementofthe superconductiitransition temperature. The remaining pieces served for metallographic examnation and thermoresistometric measurements. The X-ray analysis was perforned by means of a Philips wide-range diffractometer PW 1050/30 fitted withproportional counter PW 1965/10, an electronic circuit panel PW 1051/30 andpulse height discriminator PW 4082/00R. For most of the samples copper Kradiation was used. Superconductive transition points were determined froithe change in self-inductance of a coil placed around the sample. The selinductance of this coil was measured in an a.c. bridge (4000 cis). Thermiresistometry was carried out up to 1400 oe in a molybdenum furnace in aatmosphere of purified argon. The specimen was placed between two tungstecurrent contacts. Thin tungsten wires, spot-welded to the specimen, served ~potential contacts. When the equilibrium state at higher temperatures was 1be determined, it was necessary to quench the specimens. For temperatures uto 1100 oe this was done in the following way: from a Nier furnace, where ti:annealing was performed in an atmosphere of hydrogen, the specimen was shr

THE SYSTÈM VANADIUM-GALLIUM

hto ice v.:ater.Itwas proved that the hydrogen did not affect our results, neithery dissolution into our specimens nor by forming hydrides with them. For highernnealing temperatures the h.f. induction apparatus was used. The specimena an aluminium-oxide crucible placed in an evacuated silica capsule was hung~ the coil. Quenching was performed by letting the capsule fall to pieces into icevater. In these cases the temperature measurement was relatively inaccurate.

• Results and discussion

As our main result, the equilibrium diagram of the V-Ga system, given intg. 1, is proposed. Special features of the diagram are:11) the large solid solubility of gallium in b.c.c. vanadium;

~

) the homogeneity range of the beta-tungsten type VaGa and its transforma-tion at 1300 oe to b.c.c. phase V-Ga solid solution;

) the occurrence of three intermetallic compounds in this system, additionalI to VaGa;14)the slope ofthe liquidus, which indicates a strong tendency to the occurrence

of a miscibility gap (possibly metastable) in the liquid at about 75 at. % Ga,and the negligible solubility of vanadium in the room-temperature galliumlattice.

In the next sections these points will be discussed separately.

2WOr-----------------------------~

ig. 1. Proposed phase diagram of the system V-Ga; X thermoanalytical results, EB X-rayiffractional results.

409

410 J.H.N. van VUCHT, H.A.C.M. BRUNING, H.C. DONKERSLOOT and A.H. GOMES de MESQUITA

]'0

],0

'Xi

54 /~ /• Beaiiy52 - ., Starting maierial V

Ea Quenched from 1400"C /50 -@ " " 12000C /48 1745 )

/41/I

40 /7I

'36 /12000C

0 10000C'34 V'32 7'J(J Ir28

]'0

]'0

~

jo ],0

]·0

]·04

],0

],038

]·0

],0

]·0

]·0

]'015 7lJ 25 30 35_at.%Ga

Fig. 2. Lattice constant a of b.c.c. vanadium-gallium solid solution against gallium conter

o 5 10

3.1. The solubility of gallium in b.c.c. vanadium

Preliminary experiments, whereby the argon-are furnace was used to melt tlspecimens, indicated a maximum solubility of more than 20 at. % galliunAnnealing at 1000 °C showed that the solubility at this temperature must 1about 13 at. %. Figure 2 shows a graph of the lattice constants of the b.c.c. pha:against gallium content, includin~ data that resulted from quenching experments up to 33·3 at. %Ga. Measurements ofthe lattice constants were performeon argon-are-melted samples where possible. They were homogenized allannealed in Ah03 crucibles in evacuated silica capsules, shielded by zirconiuifoil against possible contamination by SiO *). That contaminating elemenmay sometimes influence the results is demonstrated in fig. 2, where at the vellow gallium contents the curve deviates from a straight line. This is possib

*) Always when vanadium-containing specimens were treated in silica or Si02-containÏJmaterials at temperatures of 800 DC and higher in vacuum or reducing atmosphereespecially hydrogen, we observed them to be contaminated with Si. We presume this'be due to transport of silicon and oxygen via gaseous SiO. .

THE SYSTEM VANADIUM-GALLIUM

ue to a scavenging action of gallium in the vanadium lattice but mayalso beFcribed to a purification by evaporation during the melting process.I. Thermo-analytical results are shown in the phase diagram in fig. 1. The effectt 14·3 at. % Ga was very weak. : .

[As is seen in fig. 1 we concluded that the b.c.c. region at higher' temperaturerrounds the existence region of the V3Ga phase. This is in disagreement with.lavitskii'Sphase diagram 3),where V3Ga is seen to decompose peritectically at525°C. Savitskii mentions, however, when describing alloys with the nominalbmposition of 37·5 at. % Ga, that "In cast alloys and at 1050 "C annealedIloys with a concentration gradient we observed the presence of a phase withe a-Fe type structure" and he concludes that evidently two separate regionsb.c.c. V-Ga solid solution exist.The superconductive properties of the b.c.c. phase, with varying galliumntents, were measured down to a temperature of 2·1 "K, As is shown in fig. 3e transition point decreases immediately upon adding gallium, and soon dropsbelow measurable values.

1Br-----------------------------,• Wernick et al.x van Vucht et Cf"

ig. 3. Superconductive transition point Tern, measured magnetically as a function of galliumntent of the b.c.c. and beta-tungsten (A15)-type alloys with- vanadium.

2. The homogeneity range ofV3Ga and its transformation

According to the variation of the lattice constant of the b~ta-tungsten-typeit cell ofV3Ga (see fig. 4) a considerable deviation from the stoichiometriemposition is possible. The lattice dimensions change between about the com-fsitions 20 and 33·5 ~t.% Ga. If the specimens are considered to be binary,~at is to say if possible contaminations are neglected, this range must belentified with the homogeneity range.How the superconductive transition point changes in the above range is shown

411

/

412 J.H.N. van VUCHT, H.A.C.M. BRUNING, H.C. DaNKERSLaaT and A.H. GaMES de MESQUITA

4·840

4·810

I-a (A) II- t ~Ga-pOOse •l-

T T

t- /j.A' -r ,

l- VI-

I- ./10000 c

I--

I-

lPYy("

t-

I-- );

t-. I I. I I I I I I I I I I I I I

4·830

4·820

20 25 JO 35_at.% Ga

Fig. 4. Variation of the lattice constant of the beta-tungsten-type unit cell of V3Ga witgallium content. Samples quenched from 1000 oe.

in fig. 3. For comparison, the data ofWernick et al. 2) are added. The maximurtransition temperature lies near the stoichiometrie composition of 25 at. % G~but we find it to be lower than Wernick et al. The discrepancy might be due tVaSi at the surface; this compound has a higher transition point.

Thermal arrests at about 1300and 1700 oe in each of the specimens led to thexpectation that at the lower temperature decomposition takes place, while tbhigher one has to be associated with the melting process of b.c.c. material. Ifact we found in the X-ray' diagram of a VaGa specimen, which had beequenched from 1350 oe, only very weak lines ofthe beta-tungsten type but verstrong b.c.c. lines. Initially we considered a periteetic decomposition, but wcould not observe any sign of melting.The quenching of b.c.c. material from temperatures below 1300 oe ~

37·5 at. % Ga convinced us (no incipient melting) that the transformation takeplace in the solid state. This result is in agreement with Wernick 7), and ianalogous to what has recently been found for (Nb,V)aAu by Bucher et al. 8Apart from the periteetic decomposition of VsGa at 1525 oe Savitskii et al. :give in their diagram a width of the VaGa homogeneity range of abor3 at. % (1000 0C), which is in total disagreement with our results.

3.3. New intermetallic compounds

Two intermetallic compounds were found to be stable at room temperatunThe first one seems to have the ideal composition V6Ga5 and apparently heno appreciable region of homogeneity at room temperature. From thermcanalytical and resistometric measurements it is concluded that this compoundecomposes at 1105 ± 10 oe. Quenching experiments indicate that the stablcomposition range of the compound shifts to higher vanadium content at ten

THE SYSTEM VANADIUM"GALLIUM 413

rratures just below 1105 °C. The phases which are formed on decompositionere identified by X-ray diffractometry as b.c.c. vanadium-gallium solid solu-~nand still another phase, referred to as V6Ga7,which will be described below.re X-ray powder diagram of V6Gas could be indexed on the basis of a hexag-

~

l unit cell with à = 8·496 ± 0·001 A, c = 5·176 ± 0·001 A, cja = 0·61,not too good agreement with Schubert et al. 5)who found a = 8·459 A and5·185 A. The diagram is tabulated in table I. Single crystals of the com-nd have not been found, so that a full structure determination is very

cn-cult.Efforts in this direction are described in sec. 4. The compound V6Gasd not become superconducting at temperatures down to 1·1 OK.

m temperature was found in specimensg needles (up to 6 mm) were easily grown

The other new phase stable at roontaining more than 50 at. %Ga. Lon

,TA

X-ray powder d

~kl do(A) deCA) 10(arb. units)

~:3·675 3·676 12·994 2·997 8

fo~ 2·778 2·779 112·584 2·587 9

0 2·451r 2·445 2·448 222·440

~~!~2·2132·215

222·211

201 2·1182·121

100Oil 2·116no 2·03921 1·964

)!2~ 1·8941·893

7!jU 1·89700 1·837 1·838 2!j02 1·779 1·779 21>22 1·641 1·641 I,10~ 1·603

1·6053

B21~ 1·603~03 1·561 1·561 1I

BLE I

iagram of V6Ga5

hkl do(A) deCA)ID

(arb. units)

411 1·532 1·532 3402 1·499 1·498 12500 1·471 1·471 1213 1·466 1·465 4

330~ 1·4151·415 .....2

501 1·414303 1·411 1·410 5420 1·389 1·389 <1

421~ 1·3401·342

1223 1·339313 1·312 1·317 2004 1·292 1·293 5

511l 1·2801·280 5

502~ 1·278332 1·242 1·241 1

422~ 1·2251·224

15600 1·224413 1·176 1·175 2214 1·173 1·173 2521 1·148 1·148 3224 1·104 1-104 7

414 J.H.N. van VUCHT. H.A.C.M. BRUNING. H.C. DONKERSLOOT and A.H. GOMES de MESQUIT.I

TABLE 11

X-ray powder diagram of V2Ga5

hkl do(A)10

(arb. units)

110 6·325 . 3200 4·478 <1210 4·004 <1220 3·167 10310 2·831 41001 2·688 4

320 ~ 2·485 27111201 2·306 4400 2·240 1211 2·234 16410 2·178 100330 2·112 44221 2·051 5420 2·004 15311 1·951 15321 1·826 7401 1·718 3411 1·691 2520 ~ 1·676 <1331421 1·620 2440 1·584 1530 1·537 3600 1·494 1

hkl do(À)10

(arb. units)

5111·473 1

610620

1·417 4521540 1·399 <1002 1·345 1630 1·336 7601 1·307 <1611 1·292 1

550 ~ 1·268 12710621 1·254 7541 1·242 6 "312 1·215 <1531 1·197 4322 1·183 1

65°1551 1·147 ......3711412 1·144 ......3332 1·135 <1800 1·121 3422 1·118 <1

81°1 1-112 4740820 1·087 2

in a gallium-rich (90 at. %) liquid, while cooling it from 1200 "C. Excess galliuwas removed by washing with hydrochloric acid. Wet chemical analysis of tl

, ,

isolated needles yielded 22·50 ± 0·02 weight % V, which corresponds to tlcomposition V2Ga5. X-ray powder diagrams' (see table 11) as well as Weisseiberg diagrams were made. They indicated a tetragonal unit cell with

THE SYSTEM VANADIUM-GALLIUM 415

a = 8·9732 ± 0·0002 A,c = 2·6895 ± 0·0003 A,

c]a -·0·30.

e density was found to be 6·90 ± 0·05 g/cm3, which leads to Z = 2. Weund in the handbooks 9.10) one other compound with a co~rable unit11, namely Mn2Hg5, and ascertained thatït is isomorphous with V2Ga5. Allis turned out to be in rather good agreement with the work of Schubert et.4). De Wet 11) has recently determined the structure of Mn2Hg5, which ishematically given in fig. 5. Reddy et al. 12) confirmed the data on V2Ga5·

Unit cell of V2Ga5projection afong c-axis

~ Vatom at Z=r2

o Go atom at z=O

Fig. 5. Structure of V2Ga5; schematical.

d have carried out a full structure determination. For the sake of completenesseir data are added here:

Space group D~h- P4fmbm (no. 127).

in (h):x,t+x,t; x,t-x,t; t+x,x,t; t-x,x,t; with x = 0·3197.

a in (i): ± (x,y,O);±(t+x,!-y,O); ±(ji,x,O);±(t+y,t+x,O);with x = 0·29368 and y = 0·56030.

a in (d): 0,1-,0; t,o,O.

he compound V2Ga5 decomposes peritectically at 1085 ± 5 oe into a gal--rich liquid and a phase which we called V6Ga?, and which has been de-

minated VGa by Savitskii et al. 3); V2Ga5 showed, even after prolongedealing at temperatures from 700 to 900 oe, a broad superconductive tran-

ion from 3·4 to 2.·3 "K.he 'phase V6Ga? is not stable at room temperature. On cooling it decomposes1085 ± 5 oe into V6Ga5 and V2Ga5 in sharp contrast to Savitskii's phasea which according to its X-ray diagram, has to be identified with our V6Ga?,t which is stable at room temperature. On heating, our V6Ga? decomposes

416 J.H.N. vnn VUCHT. H.A.C.M. BRUNINg. fI.C .. DONKE~SLOOT nnd A.H. GOMES de MESQUlTA

TABLE III

X-ray powder diagram of V6Ga7

at 1150 ± 10 oe into b.c.c. vanadium-gliquid. X-ray powder diagrams, suggestition with a superlattice, could be indexewith a = 9·19 A (see table Ill). From the change m lattice dimensions (f~a = 9·197 ± 0·002 A atthe V6Gas side to a = 9·182 ± 0·002 A at the V2uIside) and by estimates from powder diagrams and micrographs of the amouof different phases occurring in samples of different compositions, we concludthat this phase remained homogeneous from about 50 to 54 at. % gallium.was not possible to obtain single crystals of this phase; a structure determinatitherefore was notfeasible, We measured the density ofthe powdered phase to6·6 ± 0·1 gfcm3• In this region of the phase diagram this is an indication th

hkl do(A) deCA) 10(arb. units)

110200 ",4,58 4·58 1211 '" 3·74 3·74 1220 3·238 3·240 2310 2·900 2·901 3222 2·644 2·644 7321 2·447 2·448 8400 2·295 0411( 2·159 2·159 100330)420 2·047 2·047 2332 1·954 1·954 8422 1·870 1·870 3

510~ 1·796 1·796 4431521 1·674 1·673 1440 1·622 0

530~ 1·574 1·574 1433

6oo~ 1·527 1·528 4442

hkl do(A) deCA) 10(arb. units

611~ 1·487 1·487 2532620 1·451 0541 1·416 0622 1·384 1·383 <1 I

631 1·353 1·353 3444 1·323 1·323 4

71°1550 1·298 1·298 2543640 1·272 0721

1633 1·249 1·249 9552642 1·226 1·226 2730 1·204 1·205 <1732< 1-166 1·165 2651)800 1·147 0

8111741 1·129 1·129 5554

allium solid solution and gallium-ring a disordered b.c.c. V-Ga solid sold on the basis of a b.c. cubic unit c

THE SYsJlVvARlIJIJRS:r£~E' l A M PEN FA BR IEKitH

teuni! cell must contain less than the 54 atoms of a CsCl-type cell W.ithtripledtice parameter: For 52 atoms per unit cell (as in the phases of types D8l-S)ich means 4 units ofV6Ga7, the calculated density is 6·81 g/cm3• It may beted that also for this phase the Hume-Rothery rule for this kind of structure

!a = 21/13) holds, ifthe valence of vanadium is taken to be zero, as is usuallyne for other transition metals. As for the structure of the phase, we may say

tOCUlativelYthat its unit cell is based on a cube consisting of 27 normal b.c.c,It cells, the whole lacking 2 atoms (possibly in the centre and at the cornersthe large cell). Ordering and shifting of the vanadium and gallium atoms willn be responsible for the superstructure. Schubert et al. 5) found a phaseich they called VGa, apparently with a homogeneity region, with CU5Znse and a = 9·14A. Obviously this phase has to be identified with our V6Ga7.for superconductivity, V6Ga7 did not show any transition down to 1·1 "K.n the X-ray diagrams of samples with compositions between 25 and 50 at. %, annealed at 1090 °C and quenched to room temperature, we initially foundriabie amounts of another unknown phase. After several repeated experimentsconviction grew that contamination was responsible for its occurrence. Onlyen the rate of flow of hydrogen between the sample and the silica tube wallthe furnace at high temperature was too low was the phase formed. Itsiffraction peaks in the powder diagrams could be indexed on the basis of aagonal cell with a = 5·13 A and c = 2·37 A. We presumed that this phasetains contaminating elements, which stabilize this particular kind of struc-e. Results with specimens with deliberate addition of silicon and Si02,wever, did not agree with the presumption that the elements Si and 0 areponsible. However, now we are certain that the phase in question is identicalh what Schubert et al. call V5GasOZ'This phase has the hexagonal Mn5Sisucture, which is supposed to be stabilized by traces of oxygen. Schubertes the dimensions a = 7.244 A, cja = 0,674, which agree very well with thesition of the six diffraction lines we found.

t The solubility of vanadium in gallium

hermo-analytical experiments revealed clearly the position of solidus anduidus in the phase diagram. The liquidus curve rises very steeply upwardsm a eutectic point, which is indistinguishable from the melting point oflium. This and the fact that we observed no shifts of the diffraction peakssolid gallium in vanadium-containing specimens, leads to the result that theubility of vanadium in solid gallium is virtually nil. Even in liquid gallium,to a temperature of 700°C the solubility of vanadium is very small. This isagreement with Savitskii's.ê) work.t higher vanadium contents the liquidus is concave seen from above, which

practice is often an indication that a miscibility gap may occur. This gap,wever, if it is stable, can only be about 15at. % broad according to our results.

418 J.H.N., van VUCHT. H.A.C.M. BRUNING. H.C. DONKERSLOOT and A.H. GOMES de MESQUIT

Our numerous thermo-analytical data in the region of 75 to 85 at. % Ga 5somewhat confusing, but do not contradiet the existence of such a gap. Savit~et al. 3) have found a totally different course of the solidus and liquidus liin the diagram and no miscibility gap. .

4. The structure of V6GaS; its relation to Nb6SnS

. The unit cell ofV6Gas appears from X-ray powder diagrams to be hexagowith the dimensions a = 8·496 A, c = 5·176 A, cla= 0·61. In the ha~books 9.10)we encountered only one intermetallic compound with a comparajunit cell, namely Ti6Sns, discovered by Pietrokovsky and Prink 13).After argoare melting of this compound we compared its X-ray powder diagram with tof V6GaSand we found that the two structures are probably isomorphous.undertook the determination of the structure of Ti6Sns, this compound beimore easily obtained in the form of single crystals (judging from the phase dgram of Pietrokovsky and Frink) than V6GaS. The crystals were grown frosolution of 10 at. % titanium in tin, by cooling slowly from 1200 °C to ratemperature. Excess tin was removed by dissolving in HN03 of such a cacentration that formation of the white insoluble hydrate of Sn02 did not occ~In this way we obtained largely needle-like crystals. Occassionally, howev_1thin plate-like crystals had grown upon them. Weissenberg diagrams of btypes of single crystals were made and it was ascertained that the needles whexagonal with the unit-cell dimensions a = 9·22 A, c = 5·69 A, cja = O-space group P63mc, P63/mmc or P62c. The plates, however, were of a strture with a b.c. orthorhombic cell with edges a = 16·930 ± 0·003 A, b= 9·144 ± 0·002 A, c = 5·735 ± 0·001 A *).The first conclusion from this must be that the phase diagram of the Ti-j

system by Pietrokovsky and Prink needs some correction or addition. The t1~types of crystals may belong to two compounds of different composition, or tmay belong to one composition, but then one has to be metastable at ratemperature. Prom other experiments on the preparation of Ti6Sns we gotfurther indication that two compounds of different composition exist in tiregion. Consequently we are inclined to believe that two allotropie formsTi6Sns exist, one of which is hexagonal, designated a-Ti6Sns and one of whiis orthorhombic, designated as ,8-Ti6Sns.As soon as we have assembled enoucrystals of both types for a chemical analysis, we shall be able to prove or dprove this assumption.The importance ofthe above facts will become clear, when one considers t

in the niobium-tin system a compound Nb6SUS exists, which has a strikisimilarity to the orthorhombic form ,8-Ti6Sns. The dimensions of the unit

*) In a previous publication 6) the a- and c-axis of the unit cell were interchanged.description given here is chosen because of the relationship with the hexagonal c-Tinstructure. '

I" THE ,mEM VANADIDM·GALUUM 419

tNb6Sn5'derived by us from powder diagrams, are a = 16·814± 0·001 A,9·2037 ± 0·0004 A *), c = 5·6549± 0·0005 A, in good agreement with

e results of Ellis 14)for a phase which he calls Nb3Sn2. The space groupsmpatible with the systematic extinctions are Immm, Imm2, 1222 and 1212121.The structures ofboth forms ofTi6Sn5 were determined. Except for the space-oup assignment (P63/mmc rather than P31c) the results for the a-phase weregood agreement with those of Schubert et al. 5), who derived the followingomic coordinates:

ki in (h): ±(x,x,!); ±(x,2x,!); ±(2x,x,t); with x = 0'165.~i in (g): G-,O,O); (O,t,O)i (t,t,O); H-,O,t); (O,t,t); (t,t,t).~n in (h): (see above) with x = 0'795. -'~n ~n (b): (0,0,0); (O,O,t). , .Fn m (c): (t,t,t); (t,t,i)·[The striking analogy between the hkO intensity data for both phases immedi-ely led to the correct structure for ,B-Ti6Sn5:by abolishment of the 6-foldmetry and replacement of one c-glide plane with. a mirror plane, a b.c.

thorhombic cell is obtained, space group Immm. The agreement betweenserved and calculated X-ray intensities seems to justify the choice of thisace group; however, Imm2, 1222 and 1212121would be equally in accordanceth the symmetry and the systematic absences of the diffraction pattern.The intensities of209 non-equivalent reflections ofthe types hOland hkOwerecorded, the former with CuK,B radiation and counter techniques, the latterth MoKa radiation and a Weissenberg camera. Least-squares refinement ofe data led to conventional reliability indices R(FlIoz) = 10·6% andFlIkO) = 7·9%. Absorption (f-Ld IV 1) and dispersion effects were not takento account; one over-all temperature factor was applied to both sets ofta. Possible deviations from the stoichiometrie composition of the alloyre not considered.The final atomic positions are:

See footnote on the preceding page.

i in (n) : (x,y,O) etc.; with x = 0·410 ± 0·002 and y = 0·275 ± 0·004.i in (k) : (ht.t) etc.i in' (e) : (x,O,O)etc.; with x = 0·156 ± 0·004.i in (j) : (t,O,z) etc.; with z = 0·26 ± 0·01.n in (n) : (x,y,O) etc.; with x = 0·108 ± 0·001 and y = 0·305± 0·002.n in (f) : (x,t,O) etc.; with x = p·289 ± 0·002:n in (e) :: (x,O,O)etc.; with x = 0·322 ± b·002.n in (i) : (O,O,z)etc.; with z = 0·249 ± 0·005.

The atoms at z = °and z IV t,projected along the c-axis, are shown infig. 6b;

(J-r'fjSn5orthOrfumbicImmma=16.930±O.003.8.b= 9.144 ±O.OO2.8.c = 5.735 t.a.OOt.8.

420 J.H.N. van VUCHT. H.A.C.M. BRUNING. H.C. DONKERSLOOT and A.H. GOMES de MESQUIT

0""""" _T'O"""cx.-7i6Sn5hexagonalprwm mea=9.22.8.c=5.69 ,8.

Fig. 6a. The projection of one half of a unit cell of a-TioSn5 along its c-axis, The z paramlis 0 unless indicated otherwise. The other half of the cell is found by application of the e-glplane. Our model differs only from Schubert's model in the position of the origin of the cFig. 6b. Schematic models of the lattices of a- and f3-TioSn5 in projection, demonstratingrelationship of the two structures.

the relationship with the structure ofthe a-phase is clearly seen. Dr J. R. Ogkindly sent us the manuscript of a paper by himself and co-workers onstructure of NbsSns prior to its publication 15)., It ,corroborates that this Icompound is isomorphous .with {3-Ti6Sns, as put forward by us at the afomentioned London conference. For a more detailed description of this nstructure type reference should be made to Ogren's paper.Finally we want to point out that the relationship between the phases occu

ring in the systems Nb-Sn and V-Ga, which might have been expected on ~grounds of the chemical relationship of niobium and vanadium, extends beyoJthe {3-tungsten-type structure, which they have in common, to the side of ti

Eindhoven, September 1964

THE SYSTEM VANADWM·GALLIUM 421

lcsmd component in both systems. In Ti6Sns we seem to have found' a kindi link between the corresponding compounds Nb6Sns and V6GaS. '

REFERENCES) B. T. Matthias,:ç. A. Wood, E. Corenzwit and V. B. Bals, J. Phys, Chem. Solids 1,

188-190, 1956. ..J. H. Wernick, F. J. Morin, S. L. Hsu, D. Dorsi, J. P. Maisa and J. E. Kunzier,High magnetic fields, ed. by Kolm, Lax, Bitter- and Miles, John WHey and Sons,New York, 1962, pp. 609·614.

) E. M. Savitskii, P. I. Kripyakevitch, V. V. Baron and U. V. Efimov, Zhurnalneorg, Chimii (Russian) 9, 1155·1157, 1964.

) K. Schubert, H. G. Meissner, M. Pötschki, W. Rossteutscher and E. Stolz,Naturwiss. 49, 57, 1962.

) K. Schubert, K. Frank, R. Gohie, A. Maldonado, H. G. Meissner, A. Ramanand W. Rossteutscher, Naturwiss. 50, 41, 1963.

) J. H. N. van Vucht, H. A. C. M. Bruning and H. C. Donkersloot, Physics Letters 7,297, 1963. .

) J. H. Wernick, private verbal communication.) E. Bucher, F. Laves, J. MüIIer and H. van Philipsborn, Physics Letters 8, 27·28,1964.

) M. Hansen, Constitution of binary alloys, McGraw-HiII, New York, 1958.) W. B. Pearson, A handbook of lattice spacings and structures of metals and alloys,Pergamon Press, London, 1958.

) J. F. de Wet, Acta cryst.14, 733-738, 1961.) J. M. Re ddy, A. R. Strom and K. Kn ox, to be published.) P. Pietrokovsky and E. P. Frink, Trans. A.S.M. 49, 339-358, 1957.) T. G. Ellis and H. A. Wilhelm, J. less-common Metals 7, 67-83, 1964.) J. R. Ogren, T. G. Ellis and J. F. Smith, Acta cryst., to be published.