THE PHASE DIAGRAM OF THE Ga-Se SYSTEM - Philips Bound... · Philips J.Res. 37, 204-229, 1982 R1057 THE PHASE DIAGRAM OF THE Ga-Se SYSTEM by J.DIELEMAN, F. H. M. SANDERS and J.H. J.VAN

Philips J. Res. 37, 204-229, 1982 R1057

THE PHASE DIAGRAM OF THE Ga-Se SYSTEM

by J. DIELEMAN, F. H. M. SANDERS and J. H. J. VAN DOMMELEN

Summary

The phase diagram of the Oa-Se system is given in great detail. In agree.ment with the results of Suzuki et al. 18) a monoteetic is found between Gaand OaSe with a monoteetic temperature at 917°C and a miscibility gapwith edge compositions of 7.5 and 45.5 at.% Se at this temperature. Thegap closes at 994°C.Many of the results for the OaSe-Oa2SeS section are close to those reportedby Dieleman et al. 7). Melting points of OaSe, eutectic and Oa2SeS weredetermined at 937, 889 and 1007°C respectively. The existence region ofa-Oa2SeS was investigated and is shown to differ considerably from earlierpreliminary results. Vapour pressure minima were found in the Iiquid-vapour region of this section as well as in the solid-vapour region of thea-Oa2Ses section. An apprcximate determination of the I::iHevop• of thecomposition OaO.44SeO.56in the temperature region 960°C to 1080 DCyields 69.0 ± 4.4 kJ/mole. Approximate determination of the heats offusion, I::iHr, of Oao.s Seo.s and GaO.4SeO.6yielded values of 28 ± 1.75 kJ/mole and 16.8 ± 1.4 kJ/mole respectively.Earlier estimates of the section around Ga-Se, are shown to have beenplagued by kinetic problems. It is shown that these problems can be solvedby using a sufficient excess of liquid Se in contact with the crystalline mate-rials. It appears that a'-Oa2Ses prepared by quenching from the melt andthe y-phase prepared by heating the p-form are metastable phases. Thereversible transition of high-temperature a and its low-temperature orderedform p is located at 695°C. At this temperature P has a compositionOao.s9SSeO.405with an inaccuracy of ± 0.4 at.%. The three-phase linesp-a-L and p-a-OaSe are located at 645 and 625 oe respectively. A pre-liminary phase diagram of the p region is given. Much attention was paidto the X-ray difffraction pattern of a-Oa2Ses. Further studies are neces-sary to resolve the structure. The diffraction results of p-Oa2Ses are notinconsistent with the tetragonal unit cell proposed by Khan 29).The Iiquidus line of the Oa2SeS-Se section was determined unequivocally.The huge scatter in earlier estimates is shown to be due to insufficientreduction of idle volume and thus to varying composition because of thehigh vapour pressure. No monoteetic was found in this section.

1. Introduetion

In recent years there has been a marked growth of interest in the Ga-Sesystem. This is largely due to the occurrence of the compound GaSe with layerstructure, and the compound Ga-Se, with pseudo-zinc blende structure inwhich a third of the metal sites are unoccupied. The presence of compoundslike Ga2Se and GaSSe2 in the Ga-Se phase diagram has also been claimed.

Particularly extensive studies.have been made of GaSe to determine its elec-trical, optical, magneto-optical and photo-emissive properties and the aniso-tropy in these properties, and much attention has been paid to its (electro) lumi-

204 Phllips Journal of Research Vol.37 No.4 1982

The phase diagram of the Ga-Se system

nescence, photoconductivity, crystal structure and electronic band structure 1).Our interest in the Ga-Se system was stimulated by the relation between the

layer structure of the compound OaSe and the anisotropy in its physicalproperties (see for instance ref. 2), by the possibility of preparing amorphousthin films over an extended range of compositions, by the properties of thesefilms a), and by the application of these films in devices 4).

In order to obtain meaningful results the materials under study must be pre-pared in a well-defined way and should be sufficiently well characterized. Inthis context, accurate knowledge of the phase diagram is essential. Althoughseveral studies have been devoted to the Ga-Se system, there are still dis-crepancies that have to be resolved and important data are missing. The pres-ent investigations show what earlier results deserve preference and theyprovide much additional data. As tiny amounts of oxides in the selenides havea great effect on the phase diagram, due attention will be paid to their in-fluence. In order to ensure clear and convenient presentation of the results, theGa-GaSe, the "GaSe", the Gaêe-Gasêes, the "Ga-Se," and the Ga2Sea-Secomposition ranges will be dealt with in turn.

2. Experimental techniquesThe experimental results described in this paper were obtained by studying

several hundreds of samples. Experiments on a particular composition were atleast duplicated.

2.1. PreparationThe various compositions were prepared in a two-step procedure consisting

of the preparation of a large ingot of the eutectic composition followed by thepreparation of smaller samples of the various compositions required, byadding Ga or Se and remelting.The first step was performed by reacting a 5 gram bar of 6N Ga with the re-

quired amount of 5N5 Se pellets (both from MCP Alperton, Middlesex, Eng-land) in an evacuated, clean silica ampulla of the dimensions and form given infig.-la (the silica was from Quartz et Silice, France, quality Pursil453). The fol-lowing heating sequence was applied in a rocking furnace: at a rate of 800 °C/hrto 600 °C, then at 400 °C/hr to 850°C and finally at 800 °C/hr to 1050 °C, fol-lowed by rapid quenching in water to ensure microhomogeneity. The main reac-tion occurred in the second temperature interval. The third temperature intervalwas added for completion of the reaction and homogenizing in the melt.

The Ga pieces needed in the second step of the procedure were obtained bycooling a Ga bar wrapped in its polythene cover to liquid N2 temperature, atwhich temperature the bar is easily fractured to small clean pieces. The Se

Phlllps Joumul of Research Vol. 37 No.4 1982 205

J. Dieleman, F. H. M. Sanders and J. H. J. van Dommelen

10mmf----------1

I If

10mmf-------4

5mmf-------4

Fig. I. Dimensions of ampullae for al synthesis, bl equilibration, cl DTA experiments.

pellets showed a sufficiently large spread in weight to enable those we requiredto be easily selected.

The above procedure may seem somewhat involved, but it was necessary toavoid explosions during synthesis and to achieve the best purity. There is littlerisk of explosions if the above-mentioned quantities of starting materials, am-pulla size, wall thickness and furnace temperature program are closely followed.

2.2. Chemical analysis

Spectrochemical analysis of the products of the synthesis revealed thepresence of I ppm (by weight) of Mg and Bi each; all other metal impuritieswere below the detection limit. Flame photometric analysis indicated the ab-sence of alkali metals (detection limit a few O.I ppm). The maximum oxygencontent was 30 ppm and that of carbon 2 ppm.

The homogeneity in the composition of the starting ingots with eutectic corn-position was checked by neutron activation and X-ray fluorescence 5). The re-sults of these analyses were within at least 0.03070 of the calculated composition.

2.3. Annealing

In order to determine the boundaries of the single-phase region of Ga2Se3,samples were equilibrated in ampullae as sketched in fig. lb, following the

206 Philips Journalof Research Vol.37 No.4 1982


method described by Àlbers 6). The other annealing experiments were alsoperformed in evacuated silica ampullae. In those cases where high volatilitycould cause a shift in the composition, the idle volume was carefully filled withsilica rod. If necessary, this was also done in the differential thermal analysisexperiments (sec. 2.5).

2.4. X-ray studiesX-ray powder diffraction data were measured on standard Philips equip-

ment using CUKa radiation.

2.5. Differential thermal analysisDifferential thermal analysis (DTA) experiments were performed on a

home-built apparatus, the central unit of which is shown in fig. 2. It consistsof a furnace 1with an Alundum liner 2. The DTA block 3 is made of stainlesssteel. It has two cylindrical holes for the two DTA ampullae 4. These DTAampullae are made of silica in a form and with dimensions as shown in fig. le.After filling with the sample to be investigated the sample ampulla is evacuatedand sealed. The reference is filled with an alumina rod and also evacuated andsealed. When they have been placed in the holes of the block the idle space isfilled with fine alumina grains to ensure good thermal contact with the DTAblock. This block has three additional narrow, cylindrical bores for the ther-mocouples. One contains the thermocouple used for furnace control, theremaining pair is used for the pair of DTA thermocouples 6 which protrudeinto the well in the bottom of the DTA ampullae. A silica cap envelops theDTA block; the idle space in this cap is filled with alumina wool. The centrecontact of the two DTA thermocouples is automatically kept at 20°C. Theunit is fully automatic and runs unsupervised even overnight and at weekends.The furnace control system consists of a programming unit coupled to amV-mA transmitter, a PID controller, a thyristor unit and control thermo-couple of the furnace. All units are standard Philips products. The DTAsignal is automatically recorded using a suitable two-channel Philips recorder.

Samples of 100 mg were analysed with a minimum signal-to-noise ratio of3 : 1 and mostly 100: 1 or better, depending on the magnitude of the heat effects.If necessary, idle volume in the sample ampulla was filled with silica rod (sec.2.3). The accuracy of the composition of all samples was within ± 0.1%.Thermocouple calibration was performed as described earlier"). The accuracyof the temperature measurements was in all cases at least within ± 1°C.

The heats of fusion of OaSe and Oa2Ses were estimated from similar DTAexperiments by comparing the areas under the melting peak of these com-pounds to those of standards which melted in the same temperature range.

Philips JournnI of Research Vol. 37 No.4 1982 207

208 Phlllps Journal of Research Vol.37 No.4 1982


_0 ~III---'

1

~ ~~

~ 2, ~~ /I

3~Nt: "/

~~

4~~ ~~ ~~/Ir-

5

0 --, ~III---' '11 I-- 6

L.- -

~

~ ~~0 I /7I---'

Fig. 2. Furnace for DTA experiments: 1. furnace wall, 2. Alundum liner, 3. stainless steel DTAblock, 4. DTA ampullae, 5. silica cap, 6. thermocouples, 7. base.

The standards used were Cu, Au, Ag, Oe and InAs. Their heats of fusion canbe taken from the literature'':"). In these experiments one DTA ampulla con-tained OaSe or Ga2SeS and the reference ampulla was filled with the standard.

2.6. Vaporization experimentsTo determine the congruently vaporizing composition as a function of tem-

perature, samples were partly vaporized in a Balzers BAS10 vacuum unit,using a silica-lined furnace. Estimates of the vapour pressure were made usingan effusion hole with an area of 20 mm! in the silica liner, and by observingthe temperature at which a cylindrical lid with known area and weight andcontaining the effusion hole started floating on the vapour.


3. Results

3.1. The Ga-GaSe sectionBelow 917 DCand above 30 DConly liquid Ga and solid OaSe were observed .

. As the Ga normally remained supercooled at room temperature, samples hadto be cooled with liquid N2 for Ga to solidify, otherwise the melting peak ofGa at 30 DCwas not observed in the DTA experiments. The next thermal effectoccurred at 917 DC. Visual inspection of samples annealed at just above917 DCand quenched to room temperature showed the presence of a miscibilitygap in the liquid and proved the 917 DCtransition to be monotectic. X-rayfluorescence analyses of the two layers of samples with overall compositionGaO.75SeO.25,annealed for 10 days at 930 DCand quenched to room tempera-ture, showed that the lower layer contained 9.2 at. % (= atom %) Se and theupper 44.6 at.%. Se. The miscibility gap closed at 994 DC.

Combining the results of similar annealing experiments and chemical anal-ysis with those of DTA experiments yields the Ga-GaSe section of the phasediagram given in fig. 3. The boundaries of the miscibility gap at the mono-

/_-~900 /

IIIIIIII

700 fI

~

L1100

t

L + GaSe

200

Ga + GaSe

~o 10 20 30 40 50- Se (at.%)

Fig. 3. Ga-GaSe section of the phase diagram.

Phillps Journalof Research Vol. 37 No.4 1982 209


teetic temperature were found to be 7.5 at.OJoSe at the Ga side and 45.5 at.%at the OaSe side, with an accuracy of 0.5 at. %.

3.2. Polytypes of GaSe

No research into the various polytypes of GaSe and their existence regionwas performed in the work reported here. A detailed review on this subjecthas been published recently 1) and the reader is referred to it.

3.3. The GaSe-Ga2Se3 section

3.3.1. Liquidus curve and eutectic

The liquidus and eutectic lines are given in fig. 4 ..The difference between thedata given here and those published earlier 7) is that the precision has been im-proved slightly. The melting points of OaSe, Ga2Ses and the eutectic are937 oe, 1007 oe and 889 oe respectively. The eutectic composition is still55.1 ± 0.1 at. % Se. This greater precision will be used later on in compositionanalysis (see secs 3.3.4 and 3.4.1).

1100

1000

55-Se (at.%)

Fig. 4. GaSe-Ga2Ses section of the phase diagram.

210

60

PhllIps Journalof Research Vol.37 No.4 1982


3.3.2. Influence of oxygenSamples of gallium selenides which had been stored for periods of several

months or longer in polythene, glass or silica bottles in air showed changes inthe phase diagram. Similar changes were observed when we analyzed residuesof evaporation runs in which fresh gallium selenides had been evaporated bymore than 800/0.Chemical analyses proved this to be due to oxygen concentra-tions about one order of magnitude above the maximum of 30 pprn present inthe freshly prepared bulk materials (see sec. 2.2). The same changes could beinduced by adding minute amounts (from 0.1 to 1 at.% of 0) of Ga20s tofreshly prepared gallium selenide samples or to samples that had been storedin vacuum. Evidently storage in air, particularly moist air, causes oxidationand, as expected, gallium oxide evaporates only slightly at the vaporizationtemperatures normally used for gallium selenides 10). The changes whichoccur as compared to the phase diagram of the pure material are illustrated infig. 5 for an oxygen content of 0.2 at. %. As can be seen, the melting points of

t-- ..........

" ,,"-"-"""",,

1000 2 x 10-1at. % 0

950

900

- Se(at.%)

Fig. 5. Influence of a small amount of oxygen on the phase diagram in the GaSe-Ga2Ses section.

Phlllps Journalof Research Vol.37 No.4 1982 211


OaSe and Oa2Ses are depressed considerably, whereas the liquidus lines arepartly enhanced and partly depressed. Furthermore an additional transitionappears at 877°C, while in the region near Gaêe, the eutectic line too is de-pressed. Changes of similar magnitude were also observed in other parts ofthe phase diagram.As will be described in secs 3.3.4 and 3.4.1, the above information was used

for composition analysis by DTA. The largest amount of oxygen present inthe freshly prepared materials had no measurable effect on the phase diagram.

3.3.3. Heats of fusion of Ga Se and Ga-Se,

The heats of fusion ofthe compounds OaSe and Oa2Ses were determined asdescribed in sec. 2.5. The heat of fusion of OaSe is 56 ± 3.5 kJ/mole and thatof Ga-Se, 84 ± 7 kJ/mole.

3.3.4. Congruent vaporization of liquid gallium selenides

Partial vaporization of molten OaSe or molten Ga-Se, yielded a residuewith a composition between these two compounds in both cases. This meansthat there is a congruently vaporizing composition between them. As was also .indicated by the rapid decrease of the-evaporation rate with time in the aboveexperiments, the p-x diagram for this composition range shows a minimum inthe liquid-vapour loop as sketched in fig. 6. The compositions of these minimawere determined by vacuum vaporization in the temperature range from 930tb 1120 °C, starting with compositions OaSe or Oa2Sea. The line XL = Xv infig. 4 represents these results.

p

t

v

--xFig. 6. Sketch of liquid-vapour loop projected on a p-x plane in the GaSe-Ga2SeS section.

212 Philip. Journalof Research Vol.37 No.4 1982 .

A fair estimate of the vapour pressure of the congruently vaporizing compo-sitions Xc in the temperature range 961 to 1079 oe was made by measuring theweight loss per unit of orifice area and time. The results are given in table I foran effusion hole area of 20 mm". Although the value of Xc changes slightly with

_____________ -=-T.:.:.'he:.....!:.p.:..:h=asediagram of the· Ga-Se system

TABLE I

Vapour pressure of the congruently vaporizing composition as a function oftemperature

Temperature Effusion rateVapour pressure

T(K) W (kg/m2/sec)(Pa)

PI (0asSe) P2 (Se2) P

1234 0.012 4. 1 2.7 6.8

1274 0.028 10. 0 6.6 16.6

1300 0.048 17. 3 11.4 28.7

1334 0.121 43. 9 28.9 72.8

1352 0.146 53. 1 34.9 88.0

temperature, a temperature independenee will be assumed and x; = 56 at.%Se will be used. A further assumption is that the vapour contains only mole-cules of Oa2Se and Se2, as has been found for the compound Ga2Sesll,12).This implies that in the vapour phase Oa2 Se and Se2 molecules occur in theratio 22/17. We write the well-known Knudsen formula 13) for a mixture ofgases, as

PI = wd21t RT/M, (1)

where Pi is the partial pressure of the ith component in Pa, Wi the rate of effu-sion ofthe ith component in kg/m2/sec, R the gas constant in l/K/mole, Tab-solute temperature and M the molecular weight in kg/mole, with i as 1or 2 forGa2Se or Se2 respectively. For the ratio of the rates of effusion of Ga2Se andSe2 we find

Wl 22Ml

w;= 17M2'

hence Wl = 0.64 (Wl + W2) = 0.64 w,

so, it follows that

(2)

(3)

(4)

The resulting values of PI. P2 and P are given in table 1. We have calculatedthat the mean free path at the low side of our temperature range has about the

PhllIps Journal of Research Vol. 37 No.4 1982 213


same value as the diameter of the effusion holes, which means that we couldhave left the validity range of the Knudsen method 14).However, we foundexperimentally that in all the evaporation studies reported here w was propor-tional to the effusion hole area up to and including the largest area used. Thismeans that the Knudsen method is still approximately valid.As mentioned in sec. 2.6, at the higher end of the pressure range investigated

the pressures were checked independently by observing levitation of suitable,loose lids of the Knudsen cell. The pressure values obtained in this way coveredthose obtained from the effusion rate experiments within a factor of two.

Our results allowan estimate to be made of the enthalpy of evaporation ofx: In order to permit an easy comparison with values for GasSe, 11,12)andGaSe 15)we give the evaporation enthalphy for one mole of the hypotheticalcompound GaO.44SeO.56.This value is 69.0 ± 4.4 kJ/mole.

3.4. The section around Ga2Se33.4.1. The high-temperature, single-phase existence region of

"Ga2Ses"The starting composition of most of the samples used for determining the

solidus curve of "Ga2SeS·" was the composition Ga2SeS. The exceptions werethe samples employed for determining the Se-rich boundary. Because thisboundary of "Ga-Se;" could be close to or even at the Ga side of the com-position Ga2Ses, in the latter case we started with a Ga-rich composition. Theabove starting compositions were put in compartment I of the equilibrationcell (see fig. lb). The composition ofthe sample in compartment II was alwayschosen at or very close to the liquidus composition at the annealing tempera-ture, and at the same side of the existence region at which the value of thesolidus composition was required. These liquidus compositions were takenfrom the results given in sec. 3.3.1. Various values for the ratio of weights ofthe two samples were tried for each annealing temperature, until a preliminaryestimate of the solidus point was obtained. This ratio was then taken in theright range and annealing was continued until no further change of composi-tion occurred. At all temperatures above 900°C a few weeks of annealingsufficed to reach equilibrium. The composition of the equilibrated sampleswas analyzed using DTA and the phase diagram data of secs 3.3.1 and 3.5.The uncertainty in the compositions thus obtained was ± 0.2 at. 0/0. The de-tailed data, are shown in fig. 7.

3.4.2. Congruent vaporization in solid" Gas Ses "Congruent vaporization was also observed in the solid-vapour region of

"Ga2Ses". These congruently vaporizing compositions were determined in the

214 Phlllps Journni of Research Vol.37 No.4 1982


1000

950

IIIIII

, I

r-l-x -x1----------1' I S - v, I, I, I, I, I, IIII

t

900

8505'~6~-----5~8-------6~O-------j62

-Se{at.%}

Fig. 7. Single-phase existence region of Ga2Ses at higher temperatures and projections on a T-xplane of the vapour pressure minima in the liquid-vapour region, XL = Xv, and in the solid-vapourregion, xs = Xv. Two of the solid-liquid-vapour equilibria are also drawn (see the two thick hori-zontallines).

temperature range 910 to 935 oe, starting with either the liquid eutectic com-position or with a solid of composition Ga-Ses. In the latter case very small(about 10 urn diameter), very loosely packed Ga-Se, particles were used. If,for example, compact sintered Ga2Sea ingots were tried as a starting materialwe were faced with the development of steady state evaporation as describedby Huyer et al. 16).Evaporation of two thirds of the source filling sufficed in all cases to reach

the same final compositions, starting either from a Ga-rich or Se-rich corn-position. Again the final compositions were determined using DTA and thedata given in secs 3.3.1 and 3.3.2. The uncertainty in the compositions thusobtained was ± 0.2 at. 070. These results are shown in fig. 7 in the form ofprojections on a T-x plane of the congruently vaporizing compositions with

Phillps Journol of Research Vol.37 No.4 1982 215


minimum vapour pressure as a function of temperature. This yields the curveindicated by Xs = Xv. The curve indicated by XL = Xv holds for the liquid-vapour region and was discussed in sec. 3.3.4. We have also drawn two ofthevapour-liquid-solid three-phase lines in this region (the two thick horizontallines).

3.4.3. Polytypes of "Ga2Sea"

3.4.3.1. Quenching from the melt: metastable a '-Ga2SeJ.When Ga-Se, was prepared by quenching a melt with a composition

between that of Ga-rich Ga-Se, and that of a dilute solution of Ga-Se, in Se,in the temperature range between the melting point of Ga2Sea at 1007 oeand a temperature of 700 oe (see also sec. 3.5), the a'-form of Ga-Se, wasobtained. A crude analysis of the X-ray powder diffraction pattern of thisa'-Ga2Sea suggested that the compound roughly had a cubic, and possiblyzinc blende, crystal structure. As shown by fig. 8, where the a-values of thevarious diffraction lines of a'-Ga2Se3 have been plotted as a function of

1144 35 27 1943 36 20

5.470 0

a (A)

t 'ji-.,..o<-xx

5.4500

0 0

0

0

oo

~ __ o JJ o O O ._

5.430

4840 32 24 16 8

I I

900 50° 40° 30° 25°-f{e)

Fig. 8. Plot of a-values of the various X-ray powder diffraction lines versus the function j'(éi) todetermine an "ao"-value for a'-Ga2Ses and an ao-value for y-Ga2Ses, respectively.

20°

216 Philips Journalof Research Vol.37 No.4 1982


f«(9) = icos2(9/(sin(9 + (9) in the usual procedure for determining an ao-value, only the lines with :E h2 + k2 + [2 = 8n (the octuples) fit well, yielding anao (a'-act) = 5.433 ± 0.001 Á. All other diffraction lines deviate so much andso erratically that we can safely state that whatever crystal structure a'-Ga2Seamay have, it is certainly not purely cubic, hence also not purely zinc blende.The latter diffraction lines had a mean value of a = 5.445 ± 0.002 Á.

Other remarkable features of the diffraction pattern are that almost all dif-fraction lines are more diffuse than is usually observed for well-crystallizedpowders and that many of them possess a hump at low and/or high angles(LAH, HAH respectively). A survey of the width of the various diffractionlines of stoichiometrie melt-quenched Ga2Sea relative to those of y-Ga2Ses 17)

(see also sec. 3.4.3.4) and of the humps on these lines that are clearly dis-cernible, is given in table Il. The y-phase has been chosen as a reference be-

TABLE Il

Survey of X-ray diffraction line widths, expressed as full width at half maxi-mum (FWHM) of a'-Ga2Ses relative to the corresponding lines of y-Ga2Ses.The low- and/or high-angle humps (LAH, HAR) observed on the lower angle

lines of both forms are also given.

:E h2 + k2 + [2FWHM(a') HumpsFWHM(y) a' y

3 2.7 LAH + HAH LAH + HAR4 2.7 - -8 1.0 LAH LAH + HAR

11 2.3 - LAH + HAR12 2.1 - -16 1.3 LAH HAR

19 2.5 - LAH + HAR20 2.0 HAH -24 1.4 LAH (LAH) + RAR

27 2.5 - HAR32 1.6 LAR LAH + HAR

35 2.7 LAH LAH + HAR36 2.8 - -40 1.7 LAR LAH + HAR

Philip. Journalof Research Vol. 37 No.4 1982 217


cause, apart from the presence of humps on the diffraction lines, the crystalstructure is cubic and these lines possess widths as observed for other well-crystallized, cubic materials. This comparison shows that all diffraction linesof a'-Ga2Sea except the first octuple appear to be wider than those of they-phase. The octuples are the least diffuse, but their width nevertheless in-creases from a relative value of 1.0 for the first to 1.7 for the fifth octuple. Thenon-octuples have much larger relative widths: from 2.0 to 2.8. All octuples ofa'-Ga2Sea are accompanied by a low-angle hump; of the non-octuples onlythose with L h2 + k2 + 12= 3, 20 and 35 show clear humps. a I-Ga2Ses ob-tained by quenching of Se-rich melts generally shows even more diffuse dif-fraction lines than a'-Ga2Ses prepared by quenching of a stoichiometrie or aGa-rich melt.The last peculiar feature of the X-ray diffraction pattern to be mentioned

here is that the intensities agreed with those predicted for the zinc blendestructure with a reliability factor R < 100/0,where R = L I~II /L I and I is theintensity of a line, including humps.As will be shown in the next section a'-Ga2Sea is a metastable form ofthis

compound.

3.4.3.2. Annealing ofa'-Ga2Se3 at or above 700 oeWhen a stoichiometrie melt of Ga2Sea was quenched inadequately a low-

intensity, sharp X-ray diffraction peak was clearly discernible as a high-angleshoulder on each of the a I-octuples. The slower the cooling from the melt, themore these higher-angle peaks grew at the expense of the a I-octuples, till at acooling rate of 5 oe per hour all a I diffraction peaks had disappeared. TheX-ray diffraction pattern ofthis final stage, which we propose to indicate witha-Ga2Ses, strongly resembles that of a'-Ga2Ses, but with three differen-ces: the first was the shift of the octuples to higher angles resulting in anao (a-oct) = 5.418 ± 0.001 Á; the second, (some) slight changes in the relativeint ensities of the various diffraction peaks; and the third, slight shifts of someof the non-octuples to lower angles. The relative widths of the variousdiffraction lines and the presence of humps were the same as observed for a'.The intensities of the a-lines agreed with those predicted for a zinc blendestructure with an R value < 10%.Annealing of "stoichiometrie" a I for one week under its natural pressure in

. the temperature range between 900 and 950 oe converted about one half of a I

to a. The presence of an excess Ga from 0.1 to 1.3 at.% enhanced the con-version rate slightly, while similar excess amounts of Se caused a minor reduc-tion of this rate. When samples with a composition in the single-phase regionof Ga2Se:a (see sec. 3.4.1) were annealed at temperatures above 900 oe it was

218 Phillps Journalof Research Vol.37 No.4 1982


observed that, independently of composition, the final stage was always thea-form. However, the "cell constant" decreased up to a couple of units in thethird decimal when samples at the Ga-rich boundary were studied and in-creased by about one unit in the third decimal when samples at the Se-richboundary were investigated. When a' was annealed at temperatures below900 oe, either as the stoichiometrie compound or with a few tenths of atompercents excess of Ga or Se, the time needed to get observable conversion in-creased to over more than a few days. At temperatures below 800 oe evenannealing times of 2 weeks caused no or barely perceptible conversion. Itwasfound that in this temperature range rapid conversion of a' could be obtainedif, and only if, the crystallites were in direct contact with a melt of Se. Forexample, if an evacuated silica sample tube containing a I and a sufficientexcess of Se to keep some of it liquid at the annealing temperature was in a ver-tical position during annealing, the lower part of a I which had been in contactwith liquid Se converted, whereas the upper part of a I that had been in contactwith saturated Se vapour showed no or at best a minor degree of conversion. Itseems that conversion via the Se melt is much more rapid. However, completeconversion at these temperatures was not obtained, because during quenchingGasSe, dissolved in the Se melt precipitated as a' (see sec. 3.4.3.1).When using the above recipe for conversion via the Se solution it was found

that at all temperatures above 700 oe a-Ga2Sea is the final form obtained,even for annealing times of up to a month.

3.4.3.3. Annealing of a r : or a-Ga2Se3 between 400 and 700 oeWhen stoichiometrie a'- or a-Ga2Sea is annealed between 625 ± 20 oe and

695 ± 5 oe for periods of weeks a small part of it is converted to the fJ crystalstructure 17). Even for prolonged annealing times the amount of fJ formedremains small. Between 400 and 625 ± 20 oe no formation of fJ was observed.Annealing of a i: or a-Ga2 Sea containing an excess of 0.3 at. 070 Ga does not

yield any fJ in the whole temperature range between 400 and 700 oe. The sameobservation is made for higher excess amounts of Ga.

Annealing of a'- or a-Ga2Sea with an excess of 0.3 at.% Se enhances theconversion to the fJ-form. It appeared that conversion was higher when liquidSe was in direct contact with a'- or a-Ga2Sea, i.e. for the lower part of theannealed samples. Also, the conversion rate was higher in this part. Slightfurther increases in the excess amount of Se caused increasing conversion to fJin the part of a'- or a-Ga2Sea that had been in contact with liquid Se. Above645 ± 20 oe the percentage of fJ formed with increasing excess amounts of Seshowed a maximum and, for example, at an excess of 2 at. 070 Se fJ was absentagain. The only phase observed for long annealing times and for Ga-Se, in

Phillps Journalof Research Vol.37 No.4 1982 219


direct contact with liquid Se was a-Ga2Ses. At large excess amounts of Se,a'-Ga2Ses was also observed in an amount equal to that dissolved in liquid Seat the annealing temperature. Below 645 ± 20°C prolonged annealing yieldedP as the final form, except for larger excess amounts of Se, where a'-Ga2Seswas again formed in an amount equal to that dissolved in liquid Se at the an-nealing temperature. Another observation made was that, when a sufficientexcess amount of Se was used (say 1 at. %), the conversion rate increased firstwith increasing temperature difference at 695°C, went through a maximum atabout 500 °C and decreased again. At 500°C, one week of annealing of such asample sufficed to ensure complete conversion to p. At 400 "C the conversionrate was already impractically low.A preliminary phase diagram in this region, accounting for all our present

observations, is sketched in fig. 9. The a-Ga2Ses has a transition point to pat695 ± 5°C. P is slightly more Se-rich than stpichiometric Ga2Ses. The three-

I \I a. \

700~ \.... ..,..,~.... \I -: / \', \I »" I ,,\

: / / \ \\\/: \ 1 _'L I I---I I

I (3 II II II I, II II II II I

f3 + GaSe\ :I II II II II II II II :I II II II II II II II I

t600

500

400..;

a.+L

(3+L

I

60 61 62

Fig. 9. The "medium"-temperature region of the section around Ga2Ses.- Se (at.%)

220 PhllIps Journalof Research Vol.37 No.4 1982


phase line /)-a-L lies at 645 ± 20 oe, the three-phase line /)-a-GaSe at625 ± 25 oe and the composition boundaries indicated in fig. 9 have an in-accuracy of ± 0.4 at. % Se.

3.4.3.4. Annealing pure /) above 695 °CIn addition to the usual dependence on annealing time and temperature, the

results of annealing pure /) above the transition point at 695 ± 5 oe are foundto be determined by the excess amount of Se in contact with /).

Annealing pure /) with composition GaO.395SeO.406for one week at 800 oecaused complete conversion to pure )I17). Annealing this )I for one week attemperatures of up to 950 oe caused no further change.

Annealing pure /), containing a sufficient excess of Se to make the overallcomposition GaO.39SeO.6hat temperatures above 695 oe for a period of oneweek showed that just above this temperature part of the /) was converted to )I.At temperatures above about. 800 oe a was found in addition to )I. Theamount of )Idecreased and that of a increased with increasing annealing tem-perature and time. At the higher annealing temperatures, e.g. 950 oe, only aand an amount of a' equal to the quantity of melt formed at the annealingtemperature were found.

Annealing pure /) with excess amounts of Se between 3 and 10 at'.% in thetemperature range between 700 and 750 oe for annealing times of between aweek and a month showed that under these conditions /) was quite rapidly con-verted to )I and a. The amount of )I decreased and that of a increased with an-nealing time. The conversion rate increased with an increasing excess amount ofSe. For example, one week of annealing pure /) with an excess of 10 at. % Se at700 oe sufficed to convert it completely to a and a minor fraction of )I.

In the whole temperature range between 695 oe and the melting point at1007 oe the a-form is the final product of annealing /). The )I-form is a meta-stable, intermediate form between /) and a. Whereas below 695 oe, a can beconverted to /), especially when using sufficient excess amounts of Se, conver-sion under equivalent conditions of )I to /) seems impossible or at least veryslow.

3.5. The Ga2Se3-Se sectionThe liquidus curve in this section could only be obtained reliably if the idle

volume of the silica sample cell was rigorously reduced with the aid of a snuglyfitting silica rod (see sec. 2.3 and fig. Ie). Our results are reproduced in fig. ID.This figure also contains a few data points for DTA runs performed without

, reduction of idle volume. As expected, the "liquidus line" then shifted tohigher temperatures. The eutectic (or peritectic) temperature ofthe degenerate

Phillps Journalof Research Vol. 37 No.4 1982 .221


o no reduced ampulla volume

._.-._.-._._._._.-._.-._._._.-.

60 70 80 100

- Se (at.%)

Fig. 10. The Ga-Ses-Se section of the phase diagram.

eutectic (or peritectic) between Ga-Se, and Se was found to be 220 ± 1°C.The eutectic composition was not determined exactly, but it contains at least99.5 at. 0J0 Se.

4. Discussion

4.1. The oe-ass« section

Results of former studies and those of this investigation are compared intable Ill. If care is taken to solidify the supercooled Ga as mentioned in sec.3.1, melting of Ga (or the degenerate eutectic or peritectic) at 30°C is ob-served up to and including 49.5 at. % Se. This result, combined with the ob-

222 Philip. Journalof Research Vol.37 No.4 1982


TABLE IIIComparison of results for the Ga-GaSe section

valuesproperties Rustamov Suzuki and

et al. 21) Mori 18) this study

eutectic temperatureGa-GaSe 29.8°C 29.8°C 30 ± 1°C

compound Ga2Se claimed absent absentcompound GaSSe2 absent absent absentmonoteetic tempera-

ture L1-L2-GaSe 920°C*) 915°C 917 ± 1 °CL I-composition 4.5 at.OJoSe 5.5 at. 0J0 Se 7.5 ± 0.5 at. % SeL2-composition 17.5 at.% Se 44 at.% Se 45.5 ± 0.5 at. % Semaximum temperature

of miscibility gap not given 995°C 994 ± 1°C

*) This author claims a monoteetic with Ga-Se.

servations that between this temperature and the monoteetic at 917°C noother compound except GaSe is found and the fact that this monoteetic ispresent, seems to exclude the formation of solid Ga2Se up to this temperature.Nor were any indications of Ga2Se obtained above this temperature, except inthe gas phase. Even the annealing of mixtures of Ga and GaSe with an overallcomposition of Ga2Se for one month at temperatures of up to 917°C did notreveal other components than solid GaSe and liquid Ga with some Se dis-solved. Thus, in agreement with a number of earlier investigations 18,19,20)thecompound Ga2Se is not present in the (equilibrium) phase diagram except inthe gas phase. This does not mean that this compound cannot be formed insolid form 21,22),but if formed it is a non-equilibrium substance. As alreadysuggested 20), the same holds true for the solid compound GaSSe2. Wechecked this again by annealing as described for the overall composition ofGa2Se. The compositions at the boundaries ofthe miscibility gap at the mono-teetic temperature are in fair agreement with the values reported by Suzukiand Mori 18),but in complete disagreement with the preliminary values givenby Rustamov et al. 21). The values for the monoteetic temperature and thevalue of the temperature at the top of the miscibility gap agree closely withearlier values 18,21).Our results for this section agree very well with those ofSuzuki and Mori 18). For those values that differ slightly we prefer our own

Phllips Journalof Research Vol. 37 No.4 1982 223


values, because of precision both in temperature calibration and in compositionanalyses. We expect a vapour pressure maximum to be present in the liquid-vapour equilibrium at a composition equal or close to Ga2Se; see also sec. 4.2.

4.2. The GaSe-Ga2Se3 sectionA comparison of the values for the characteristic temperatures, and com-

positions in this region is given in table IV. The results of this study are inexcellent agreement with those of Suzuki and Mori 18), in fair agreement withthe values of Palatnik and Belova 23), whereas the studies of Rustamov etal. 21)give values deviating from those of the other three papers. In view of

TABLE IVComparison of results for the GaSe-Ga2Se3 section

valuesproperties

Rustamov Palatnik and Suzuki andet al. 21) Belova 23) Mori 18) this paper

congruent meltingpoint of GaSe 960 oe (fig.) 950 oe (fig.) 938 ± 3 oe 937 ± IOC

congruent meltingpoint of Ga2Se3 1020 oe (fig.) 1010 oe (text 1003 ± 3 oe 1007 ± 1 oe

melting point of +fig.)

eutectic 912 oe (text 780 °C(text) 884 ± 3 oe 889 ± 1 oecomposition of +fig.) 880 oe (fig.)

eutectic 55 at. % Se 55.4 at.% Se 55 at.OJoSe 55.1 ± 0.1(text) (text) at.% Se

our greater precision and the possible influence of small amounts of oxygenimpurities not accounted for in the work of Palatnik and Belova 23) andSuzuki and Mori 18),we believe that our values deserve preference. In the caseof the melting point of Ga2Se3 insufficient reduction of idle volume (see alsosecs 4.3 and 4.4) also causes the composition of Ga2Se3 to deviate to theGa-rich side by the development of a significant pressure of Se-rich vapour.In some earlier studies this might have resulted in too Iowa melting point.

Our observation that partial vaporization of liquid GaSe yielded a residuewith a Se content higher than the starting composition is similar to observationsof Piacente et al. 15). The observation of the congruently vaporizing com-

224 PhllIps Journalof Research Vol.37 No.4 1982


position in the liquid and the determination of its composition as a function oftemperature is a new fact. Our observation that this azeotropic compositioncorresponds to a vapour pressure minimum in the liquid-vapour equilibrium isquantitatively confirmed by comparing the vapour pressures measured forliquid GaSe 16), liquid GaO.4S2SeO.668thisstud

y) and of solid GaeSes 12) at thesame temperature. For example, at 960 oe these equilibrium vapour pressuresare estimated at 45, 7 and 19 Pa respectively. The results of Piacente et al.:")and those of this study support the occurrence of a vapour pressure maximumat or around the composition Ga2 Se, a compound which appears to be stableonly in the vapour phase.

4.3. The section around Ga2Se3Earlier estimates of the high-temperature part of the single-phase existence

region of a-Ga2Sea 7,17)can only be considered as preliminary and no morethan a good>starting point for the more refineddetermination as a function oftemperature in the range between the melting point of Ga2Ses at 1007 oe andthe eutectic temperature at 889 oe, described in sec. 3.4.1. The large width ofthis region, up to 1.3 at. 070, at temperatures around the transition point at932 ± 2 oe, is clearly related to the disturbances caused by the vapour pres-sure minima present in this part of the system. Down to 930°C the width atthe Se-rich side is found to be :;:;;;0.1 at. %. Larger values of this width have.been reported earlier 17), but the report referred to did not take sufficientaccount of the amount of Se in the vapour phase, particularly when large idlevolumes were present.

Combination of the observations made by Berger et al. 12),who noted thatthe vapour in equilibrium with solid Ga2Sea just above the eutectic tempera-ture consisted of almost pure Se2, and those of Piacente et al.16) who remarkedthat in this temperature region the vapour in equilibrium with solid Ga2Se2consisted of almost pure Ga2Se and a comparison of the equilibrium pres-sures, e.g. at 907 oe 2 Pa for GasSe, and 5 Pa for Ga2Se2, strongly suggeststhat a vapour pressure minimum with a composition between that of these twocompounds also exists in the temperature region below 932°C. This is in ex-cellent agreement with our results, which indicate a similar vapour pressureminimum in the equilibrium solid a-Ga2Sea with vapour (see fig. 7).

Comparison of earlier results on the value of the cell constants of the variousstable and metastable phases of "Ga-Se;" with one another and with thepresent results as shown in table V is seriously hampered by lack of sufficientdata on preparation and insufficiently accurate description of how "cell con-stants" were determined. For example, although we quenched molten Ga2Sea

. by dropping the evacuated silica container from the furnace into cooled brine

Phlllps Journolof Research Vol. 37 No.4 1982 225

~~

TABLE VComparison of some crystal structure data of various forms of Ga2Ses

...ê;~...o=3aSo

~ri.,.

phasea' IJref. a y

Hahn et al. 24) ? am = 5.418 ± 0.005 Á 0) - -Woolley et al.25) a« (oct.) = 5.435 Á - - -

a« (rest) = 5.445 ÁPalatnik et al. 17,23,26) a (even) = 5.411 ± 0.002 Á a (even) = 5.416 ± 0.002 Á a = 7.76Á a« = 5.463 ± 0.003 Á

a (odd) = 5.44 Á a (odd) = 5.43 Á b = 11.64Ác = 1O.822Áorthorhombic

mixture?Popovié et al.27) a (sharp) = 5.426 Á - -

a (broad) = 5.453 Á

Ghémard et al.28) ao = 5.446 ± 0.005 Á 00) - a= b = 6.66Á -c = 11.65Áy = 108°12'monoclinic

Khan 29) - . - a = 7.760Á -b = 11.640Ác= 1O.822Átetragonal

this work a« (oct.) = 5.433 ± 0.001 Á a« (oct.) = 5.418 ± 0.001 Ä not ao =5.4595 ± 0.0005 Äam (rest) = 5.445 ± 0.002 Á inconsistent

with Khan'sanalysis

~tJ~.~~~~!J::~CI)'

§~è:l§t:).

~~~§tJClss~~;::s .

~~z?...~:ii:... 0) a; = mean value of a derived from the a-values of the various diffraction lines; since no details on synthesis were given it is uncertain whether this

value is valid for a.00) metastable phase of exact zinc blende structure obtained by extremely fast quenching of melt.


,we never obtained the pure zinc blende structure mentioned by Ghémard etal. 28).As these authors do not describe their recipe for rapid quenching it is

. difficult to assess this difference in results. In our case we always obtainedmetastable a' with pseudo zinc blende structure. The only quantitative com-parison possible in the case of a' is that with the results of Woolley et al. 25).As table V shows, the agreement with .our results is quite good. Comparisonwith work of Palatnik et al. 17,2S,26)or Popovié et al. 27)is hardly possible orleads to deviating results, e.g. the data of Popovié et al. 27)on the presence ofsharp components and humps in the various diffraction lines largely deviatefrom our results (compare table IIwith ref. 27).For stable a-Ga2Se3 the situation is even worse. For instance Hahn et al. 24)

do not describe their preparation procedure. Even comparison with the resultsof Palatnik et al. seems useless because they did not notice that only the octuplesbehaved well in the determination of ao for a zinc blende structure. Furthermorethese authors state that for a and a', diffraction lines with4h2 + k2 + [2 = evenare sharp, while "odd" lines are diffuse. Our results in table II clearly showthat only octuples are relatively sharp, while on average the remaining evenlines are broadened to almost the same extent as the odd diffraction lines.Our observation that the reversible transition of a to fJ is located at 695± 5 °C

is in disagreement with the 790°C published by Palatnik et al. 23) and the800°C claimed by Ghémard et al. 28).As shown in this paper, the latter resultsare in error because of kinetic reasons. An analysis of the diffraction lines of fJshowed that the conclusions of Ghémard et al. 28) and Palatnik et al. 23)dis-agree with our results and that the present data are not inconsistent withKhan's analysis 29). So fJ could be an ordered form of a with possibly a tetra-gonal, body-centred cell, space group 14dacd with a = 23.235 ± 0.005 Á andc = 10.828 ± 0.002 Á. If fJ is an ordered form of a then the transition at695 ± 5 °C is an order-disorder transition where the disorder in a is reflectedin the form of the diffraction lines. A quantitative explanation of the lineforms of a cannot be given at present. In all probability these line forms aredetermined by something different from just stacking faults as suggested byPalatnik et al. 23,26).

In disagreement with Ghérnard et al. 28)and Khan 29),who concluded thatthe disorder decreases with decreasing temperature and is uniquely related totemperature, we found that P is completely ordered below 695 ± 5°C. A suf-ficient excess of Se, a sufficiently high temperature and sufficient time seem tobe all that is required to reach pure fJ.Evidently y is a metastable phase of "Gae.Ses". The ao values of the exact

zinc blende structure of y as determined in this work and by Palatnik et al.!")agree within the accuracy of the experiments. We completely agree with the

Philips Journalof Research Vol.37 No.4 1982 227


observations ofthe latter authors thatp can be converted to y and y to a, but ycannot be reconverted to P other than via a. /'Because of our more thorough investigations and our much better defined

experiments we prefer our conclusions to the divergent results of earlierstudies. This preference is supported by the good agreement between ourresults and those of earlier well-defined studies. Further studies are necessaryto define the existence region of P better than our preliminary results indicate.

4.4. The Ga2Se3-Se section

When our results given in fig. 10 are compared with earlier results for thissection 18,21,23,30) it is immediately obvious that a stringent reduction of idlevolume is a prerequisite for obtaining reliable results at higher vapour pres-sures of incongruently vaporizing compositions. No doubt, failure to realizethis in the past has been the main cause of the scatter in data in this section ofthe phase diagram. In contrast to conclusions drawn in earlier work 30) we didnot observe a monoteetic in this section. Again, because of our more thoroughinvestigations and our more careful experiments we judge that our resultsdeserve preference. The transition at about 220 oe is connected with themelting of amorphous Se and as such does not belong to the equilibrium phasediagram.

Acknowledgement

Our sincere thanks are due to J. Ponsioen and C. Damen for their coopera-tion during the preliminary determination of the vapour pressure minimum ..The continuous support of the Departments of Analytical Chemistry andX-ray diffraction as well as the help of various workshops is gratefully ac-knowledged. .

Philips Research Laboratories Eindhoven, April1982

REFERENCES1) E. Mooser (ed.), Physics and chemistry of materials with layered structures, Reidel Pub!.

Comp., Dordrecht, The Netherlands, 1977.2) J. D. Wasscher and J. Dieleman, Anisotropy of the optical constants of GaSe near the

band edge, Phys. Lett. 39A, 279-280, 1972.3) C. Wood, L. R. Gilbert, C. M. Garner and J. C. Shaffer, Amorphous group Ill-chalco-

genides, in J. Stuke and W. Brenig (eds.), Proc. 5th Int. Conf. on amorphous and liquidsemiconductors, Taylor and Francis Ltd., London, England, pp. 285-295, 1974.

4) E. T. J. M. Smeets, J. Dieleman, F. H. M. Sanders and D. de Nobel, Passivation ofsilicon p-n junctions by slightly conductive chalcogenide films, J. Electrochem. Soc. 124,1458, 1977.

228 PhllIpsJournal of Research Vol.37 No.4 1982


5) E. Bruninx, The accurate determination of Ga in Ga-Se, by means of neutron activation: acomparison with X-ray fhiorescence and wet chemical methods, J. Radioanal. Chem. 19,47-53, 1974.

6) W. Albers, Physical chemistryofdefects, in M. AvenandJ. S. Prener(eds.), Physics andchemistry of U-VI compounds, North Holland Publ. Comp., Amsterdam, The Netherlands,p. 204, 1967.

7) J. Dieleman and A. R. C. Engelfriet, The phase diagram of the system Gal_xSex for0.5 ~x~ 0.6 and 300 K ~ T~ 1500 K, J. Less-Common Metals 25, 231-233, 1971.

8) O. Kubaschewski, E. LL. Evans and C. B. Alcock, Metallurgical therrnochemistry, inG. V. Raynor (ed.), Metal Physics and Physical Metallurgy, Pergamon Press, Oxford, Eng-land, 1967.

9) D. Richmann and E. F. Hockings, The heats of fusion of InSb, InAs, GaAs and InP, J.Electrochem. Soc. 112,461-462, 1965.

10) S. A. Shchukarev, G. A. Semenov and 1. A. Rat' kovskii, A thermal investigation oftheevaporation of gallium and indium oxides, Russ. J. Inorg. Chem. 14, 1-5, 1969.

") O. M. Uy, D. W. Muenow, P. J. Ficalora and J. L. Margrave, Mass spectrometricstudies at high temperatures, Part 30. - Vaporization of Ga2Ss, Ga2Ses and Ga2Tes, andstabilities of the gaseous gallium chalcogenides, Trans. Far. Soc. 64, 2998-3005, 1968.

12) L. J. Berger, S. S. Strel'chenko, S. A. Bondar, A. D. Molodyk, A. E. Balanevskayaand V. V. Lebedev, Dissociation pressures of binary semiconductor compounds of theB~"C~Itype, Inorg. Mater. 5, 743-745, 1969.

IS) F. Daniels and R. A. Al b er t y, Kinetic theory, in F. Daniels and R. A. Alberty (eds.),Physical chemistry, Wiley, New York, U.S.A., p. 311, 1967.

14) R. Speiser and H. L. J ohnston, Methods of determining vapor pressure of metals, Trans.A.S.M. 42, 283-305, 1950.

15)V. Piacente, G. Bardi, V. Di Paolo and D. Ferro, The vapour pressure over Ga2S2 andGa2Se2, J. Chem. Thermodynamics 8, 391-401, 1976.

16)P. Huyer, W. Th. Langendam and 1. A. Lely, Vacuum deposition of resistors, PhilipsTechn. Rev. 24, 144-149, 1962/63.

17) L. S. P alatnik and E. K. Belova, A study ofthe polymorphism of the selenide Ga2Ses withvariable composition, Inorg. Mater. 1, 1703-1707, 1965.

18)H. Suzuki and R. Mori, Phase study on binary system Ga-Se, Jap. J. Appl. Phys. 13,417-423, 1974.

19)K. Schubert, E. Dörre and M. Kluge, Zur Kristallchemie der B-Metalle, lIl. Kristallstruk-tur von GaSe und InTe, Z. Metallkde 46, 216-224, 1955.

20) J. C. J. M. TerhelI and R. M. A. Lieth, Structures and compounds in the system Gal_xSex,Phys, Stat. Sol. (a) 10, 529-535, 1972.

21) P. G. M. Rustamov, B. K. BabaevaandN. P. Luzhnaya, Theinteractionofgallium withselenium, Inorg. Mater. 1, 775-776, 1965.

22)W. Klemm and H. U. von Vogel, Messungen an Gallium- und Indium-Verbindungen.X. Ueber die Chalkogenide von Gallium und Indium, Z. Anorg. AlIg. Chern. 219, 45, 1934.

2S) L. S. Palatnik and E. K. Belova, Investigation of the Ga-Se phase diagram, Inorg. Mater.2, 657-659, 1966.

24) H. Hahn and W. Kli ng ler, Ueber die Kristallstrukturen von Ga2Ss, Ga2Ses und Ga2Tes,Z. Anorg. AlIg. Chem. 259, 135-142, 1949.

25) J. C. Woolley and P. N. Keating, Solid solubility of In2Ses in some compounds of zincblence structure, J. Less-Common Metals 3, 194-201, 1961.

26) L. S. Palatnik, E. K. Belova and A. A. Koz'ma, Anomalous effects on radiographs ofgallium selenide and its alloys, Sov. Phys. Doklady 9, 1014-1017, 1965.

27) S. Popovié, B. Celustka, Z. Ru äié-Tor o s and D. Broz, X-ray diffraction study andsemiconducting properties of the system Ga2Ses-In2Ses, Phys. Stat. Sol. (a) 41, 255-262,1977. .

28) G. Ghémard, R. Ollitrault-Fichet and J. Flahaut, Etude cristallographique du séléniurede gallium Ga2SeS en fonction de la température: formes désordonnées. Surstructure mono-c1inique. C.R. Acad. Sci. Paris 282C, 831-834, 1976. .

29) M. Y. Khan and S. Z. Ali, Crystal daa for p-Ga2Ses, J. Appl. Cryst. 10,70-71, 1977.SO)R. Ollitrault-Fichet, J. Rivet and J. Flahaut, Etude du système ternaire Ga-As-Se.

Diagramme de phases partiel. Materiaux vitreux: formation, propriétés, Ann. Chim. Fr. 3,287-303, 1978.

PhllIps Journal of Research Vol. 37 No.4 1982 229

Documents

THE PHASE DIAGRAM OF THE Ga-Se SYSTEM - Philips Bound... · Philips J.Res. 37, 204-229, 1982 R1057 THE PHASE DIAGRAM OF THE Ga-Se SYSTEM by J.DIELEMAN, F. H. M. SANDERS and J.H. J.VAN