Solid-State Electronics Pergamon Press 1960. Vol. 1, pp. 97-106.
Printed in Great Britain
TECHNOLOGY OF GALLIUM A R S E N I D E F. A. CUNNELL, J. T. E D M O N D a n d W. R. H A R D I N G
Services Electronics Research Laboratory, Baldock, Herts.
(]~eceived 21 September 1959) A b s t r a c t - - T h e steps in the preparation of the compound semiconductor gallium arsenide are described, from the treatment of the component elements to the zone purification and production of single crystals of the compound. The efficiency of zone refining and the influence of some impurities on the electrical properties of the material are discussed. The purest material prepared by the authors has an electron mobility of 6700 cm 2 V -1 sec -1 at room temperature. Techniques for making simple electrical measurements and etching-procedures are recommended. The production of p-n junctions by diffusion and some effects of heat treatment on the electrical properties are also included. R 6 s u m 6 - - O n d4crit les 4tapes dans la pr6paration du compos4 semiconducteur AsGa, depuis le traitement des 616ments cornposants, jusqu'~ la purification par zone fondue et la production des monocristaux du composd. On discute l'efficacit6 du raffinage par zone et l'influence de certaines impuret4s sur les propridt4s 61ectriques du matdriau. Le mat6riel le plus put pr6par4 par les auteurs a une mobilit4 de 6700 cm 2 V -1 sec -~ ~ la temp6rature ordinaire. On recommande des m6thodes de traitements et de mesures 4lectriques simples. Dans l'article est inclus aussi la pr6paration de jonctions p-n par diffusion et certains effets de traitements thermiques sur les propri~tds 61ectriques. Z u s a m m e n f a s s u n g - - E s werden die einzelnen Stufen der Pr~iparation der halbleitenden Verbindung Galtiumarsenid beschrieben, und zwar ausgehend yon der Behandlung der beiden Komponenten bis zur Zonenreinigung und Herstellung yon Einkristallen der Verbindung. Die Wirksamkeit der Zonenreinigung und der Einfluss einiger Verunreinigungen auf die elektrischen Eigenschaften des Materials werden besprochen. Das yon den Autoren hergestellte reinste Material hat eine Elektronenbeweglichkeit yon 6700 cm 2 V -1 sec -1 bei Zimmertemperatur. Hinweise f/Jr einfache elektrische Messungen und "A.tzverfahren werden gegeben. Ausserdem werden die Herstellung yon P-n-Uberg~ingen durch Diffusion sowie einige Auswirkungen des Temperprozesses auf die elektrischen Eigenschaften behandelt. 1. I N T R O D U C T I O N
purification of the elements and the preparation, zone purification and g r o w t h of crystals of the c o m p o u n d . I n the second section there is a discussion of the effect on the electrical properties of G a A s of doping w i t h certain i m p u r i t y elements and of the p r o b l e m s which arise in the course of m a k i n g m e a s u r e m e n t s on the material. T h i s includes etching and the practical details of m a k i n g low-resistance o h m i c contacts to n- and p - t y p e material. I n addition the effects of heat t r e a t m e n t on the c o m p o u n d and the p r o d u c t i o n o f p - n j u n c tions by i m p u r i t y diffusion are described.
SINCE 1952 the g r o u p I I I - g r o u p V c o m p o u n d s have attracted an increasing a m o u n t of attention. T w o of these materials, G a A s and I n P , have energy gaps of 1 "37 eV and 1 "27 eV respectively and in v i e w of their high electron mobilities t h e y are considered as possible c o m p e t i t o r s of silicon for certain applications. T h i s paper concerns G a A s but m a n y of the technological p r o b l e m s w h i c h are discussed arise in the processing of o t h e r c o m p o u n d s possessing a volatile constituent e.g. InAs, I n P and GaP. T h e r e have been a n u m b e r of previous publications dealing w i t h various aspects of GaAs. (1-4) I n v i e w of the g r o w i n g interest in this c o m p o u n d a general account of t e c h n i q u e s required for preparation, purification, crystal growing and o t h e r p r o b l e m s seems timely. T h e account is divided into two sections. T h e first deals w i t h the
2. PREPARATION OF GALLIUM ARSENIDE
(a) Purification of the e/ement~ T h e application of zone refining to the c o m p o u n d G a A s has m e t with only limited success and will be discussed in detail below. C o n s e q u e n t l y it 97
F. A. C U N N E L L ,
J. T. E D M O N D
where(V, s) and to be present originally in the arsenic; equally it was thought that sulphur was likely to be present in GaAs and the arsenic was treated to remove this impurity. T h e method used is described by HARMANet a/.(s) ; the arsenic is dissolved in high-purity molten lead at 650-700°C and sublimed from the solution. Because of the affinity of sulphur for lead, the sublimate is relatively free of sulphur. This process is followed by a further sublimation in vacuum. GaAs prepared from arsenic which has had the above treatment has proved invariably to be purer than the compound made from untreated arsenic. The segregation coefficient of some impurities
is desirable to obtain starting elements of the highest available purity. In general, gallium and arsenic containing 3-30 p.p.m, total impurity, as indicated by spectrographic analysis, are readily available commercially. A problem arising in any purification work at the level of a few parts per million is the assessment of the impurity which remains. Spectrographic techniques are of limited application and in general the efficiency of the treatment has been judged by preparing batches of GaAs and measuring Hall constant (R) and conductivity (~).! '~ Methods for the purification of gallium have been described in the literature. (5,6) Although a
and W. R. H A R D I N G
(............ High-lernperoture furnace
Low-lernperoture furnoce l
I I Ii I
FIG. 1. Preparation of GaAs. number of attempts have been made by the present authors to improve the gallium, they find that gallium arsenide prepared from the best currently available metal without further treatment is at least as good as any other material they have measured. Arsenic is available in powder or metallic form, the powder form especially contains a quantity of arsenic trioxide, which can be removed by heating to 250°C in a continuously pumped vacuum system. T h e arsenic can then be sublimed at 500°C in vacuum and deposited in the metallic form at a lower temperature. Non-volatile residues such as carbon and traces of iron are left behind. A number of ingots of gallium arsenide were prepared with arsenic treated in this way. These showed an improvement on zone refining, indicating that some n-type impurities were segregating (2), but not all. Sulphur was believed to be a troublesome impurity in InAs prepared in this laboratory and else-
in GaAs has been measured by "~VHELANet al. (9~ They confirm that sulphur and also selenium and zinc show poor segregation. (b) Synthesis of gallium arsenide The pressure-temperature composition relationship for the system Ga-As has been investigated by BOOMGAARDand SCHOL(3). T h e y find that the dissociation pressure of the compound GaAs is 0"9 atm at the melting point (1237°C). T h e vapour pressure of arsenic rises very rapidly with temperature from 1 atm at about 610°C to 37 atm at 818°C, the critical point. As a result the direct synthesis of GaAs in a uniform temperature enclosure must be attempted with caution to avoid dangerously high arsenic pressures. It is possible, however, to prepare GaAs in a quartz tube in this way, provided the temprature is raised slowly so that near-equilibrium conditions are maintained. Since GaAs tends to "key" to quartz, it has been
OF G A L L I U M
found advisable to use a quartz boat or preferably an unsealed tube to contain the material. I n this way any risk of the vacuum wall being broken on cooling is avoided. T h e logical development of the above method of preparation which is now used is to employ a composite furnace system to provide the temperature profile shown in Fig. 1. At one end a constant temperature of 610°C is maintained over an appreciable length CD and at the other end a temperature of about 1250°C. The inner quartz tube initially containing only the gallium is located at the high-temperature end, and the arsenic pressure in the system determined by the temperature of the colder end. Sufficient excess arsenic is added to keep about 1 atm of pressure in the enclosure even after the stoichiometric compound has been formed. By moving the outer quartz tube relative to the temperature gradient AB, GaAs can be crystallised in a controllable manner from one end of the boat. It is essential to provide a uniform low-temperature region of sufficient extent to ensure that the arsenic reservoir remains at a constant temperature throughout the solidification. Variations in the temperature of the furnace surrounding the arsenic reservoir should not exceed 4-2°C. This method yields ingots containing large crystallites, Fig. 2(a). (c) Zone purification of the compound The problem of the zone purification of a compound in which one element is appreciably volatile has been discussed elsewhere. (4,1°) In all experiments in which GaAs is melted, whether it be zone refining, crystal growing or even a simple doping process, provision must be made for the required ambient pressure of arsenic. This involves the design of a suitable furnace. T h e production of a molten zone can be achieved either by radiation heating or by radio-frequency induction heating. In the first method accurate temperature control is necessary since the melting point of GaAs is close to the temperature at which quartz is easily deformed. With induction heating the hottest point is within the quartz tube, the walls of which remain at a much lower temperature, and this method is preferred. The material, together with excess arsenic, is placed in an unsealed quartz tube inside an evacuated sealed tube. This in turn is situated in a third quartz tube, much longer,
which is open at both ends and which can be moved through the coil and its two associated furnaces, placed one on either side of the coil (Fig. 3). The outermost tube serves only as a convenient, GaAs in quortz R.E coil
y//.//////~ o~ o v / 2 7 ~ / / / / / /
o o o V / / / / / / / / / / / / / / / / / / / / / , ~ ; ~; Y/.
Furnace 7 0 0 ° 0
FIo. 3. Apparatus for zone refining. transparent support for its contents. It is important to keep the gap between the furnaces to a minimum in order to avoid a cold spot in the temperature distribution. This requirement implies a coil with few turns. Radio-frequency generators operating at 400kc/s and 5 Mc/'s have been used. It is possible to produce a zone in relatively impure material using the lower frequency, but as the material is purified its resistivity increases and it is more difficult to couple sufficient power into the ingot to melt a zone. The reason for this can be understood by considering the variation of the skin depth (d) with resistivity (p) and frequency ( f ) ; dis proportional to "X/(p/y) and is numerically equal to 8 m m for 0.01 f2 cm material and 400 kc/s. For optimum coupling the skin depth should be less than half the radius of the ingot (assumed cylindrical); in a typical case the mean radius of an ingot would be about 4 mm. Consequently for material of the above resistivity and higher, 5 Mc/s is much better than 400 kc/s. It is possible to overcome this kind of difficulty when using a 400 kc/s generator by preheating the ingot to 1000°C or more, in one of the furnaces. This raises the conductivity and consequently the coupled power. Whenever radio-frequency heating is used it is found that a deposit of GaAs accumulates gradually on the container walls. It is believed that this transport from the liquid zone is due to the vigorous escape of arsenic from the centre of the zone where the temperature is well above the melting point of GaAs and the dissociation pressure exceeds the ambient. Small particles of gallium arsenide are ejected from the melt and impinge on the nearby walls.
F. A. C U N N E L L ,
J. T. E D M O N D
The radiation losses from a zone heated by induction are quite large, so that the freezing front is concave towards the liquid. This applies especially to the arrangement of Fig. 3 and the effect is to make the material markedly polycrystalline. Radiation heating is generally an improvement in this respect at the expense of efficient stirring of the liquid. (d) Crystal growing, The techniques of crystal growing which have been used successfully in the case of germanium and silicon may in principle be applied to gallium arsenide with suitable modification to provide the ambient arsenic pressure. Crystals of GaAs and InAs have been grown by the Czochralski method. GREMMELMAIER Q1) has overcome the difficulty of controlling the movement of a seed crystal within a sealed system by the use of a magnetic lifting mechanism, while RmHARI)S(12) has accomplished the same thing in a rather different way by using an annular liquid-gallium trap separating the moving and the stationary portions of the equipment. This latter method is ingenious but there are some difficulties in the operation. Perhaps the most potentially attractive method for GaAs is the one described by CRESSELL and POWELL(t3) and BENNETT and SAWYER(14) for the particular case of germanium. Here, the crystal is grown in a horizontal boat from an orientated seed by the controlled traversal of the liquid-solid interface along the ingot. The apparatus required is essentially simple and no difficult manipulations are involved. Although a certain amount of success has been achieved with this method, there are difficulties encountered with GaAs which do not occur for germanium. The first of these arises in the choice of crucible material. Quartz would be the obvious choice but unfortunately there is a tendency for localized "wetting" to occur. A partial solution is obtained quite simply by grinding the interior of the quartz boat with fine carborundum and coating with a deposit of pure carbon. For example, "AnalaR" acetone, when burnt, provides a source of suitably pure carbon. This is rubbed into the ground quartz surface and vacuum-baked. It is found, however, that GaAs is contaminated whenever the boat is treated in the above manner. There is some evidence that the cause of the contamination is not any impurity in
and W. R. H A R D I N G
the carbon but the carbon itself. Four such ingots had n-type Hall constants of 15, 44, 7.4 and 6"3 cm 3 C -1 and one doped with spectrographically pure carbon had a Hall constant of 8"1 cm a (71. These are an order of magnitude worse than the values for the average GaAs prepared. HANNAy(t5} has reported the presence of 3 p.p.m, of carbon in an ingot of gallium arsenide after extensive zone refining in a graphite boat. This is an order of magnitude less than the concentration corresponding to the Hall constants mentioned above. A second difficulty which occurs is common to the crystallization of all compounds possessing a highly volatile component. Fluctuations of the ambient arsenic pressure around the value 0.9 atm can cause bubbles of arsenic vapour to form within the liquid. These can be trapped by the movement of the advancing interface and incorporated in the solid. The boat and contents, together with excess arsenic, are placed in a long quartz tube, evacuated and sealed (Fig. 4). The furnace system mounted Furnace Zone furnace 800-1200°C ~1260c'C -~ > ,~,:>;~-,, ,,. ~ Control f;rnace 610%/ L';:: ///)eL% -', : ~,~;; .
Sindanyo spacer (enlarged) rods
. ;: ~;/,>7/>g,:
-;~< ; , ;"
(oo 1 Zone furnace, wound ~(4,~./F-5~t o°/'-~°c~l [- - OnpyrophyllitehOrizontal rods [
Cross- section at position of zone FIG. 4. A p p a r a t u s for crystal g r o w i n g .
on a movable carriage consists of three separate parts. T h e first part is kept at 610°C, the minimum temperature of the system, and controls the pressure in the tube. At the beginning of the experiment one end of the ingot is situated in the second region at 1260°C and a molten zone is formed. T h e ingot is moved slowly into the third region in which the power input is set to produce a suitable temperature gradient at the freezing front. The centre furnace is wound in such a way that the power supplied to the upper and lower
OF G A L L I U M A R S E N I D E
halves can be regulated separately. This makes crystallite in a horizontally cast ingot. The axis provision for adjusting the inclination of the was found to be a (321) direction. freezing front to the vertical. It is possible to obtain This method has the following advantages. very large crystals even in the absence of a seed, Firstly there is no contact between the molten Fig. 2(b). material and the silica container. Secondly, the The following conditions are recommended for the successful application of this technique : (1) A constant arsenic vapour pressure of 0.9 atm should be maintained. (2) The freezing front should be planar and normal to the growth direction. A small temperature gradient assists in obtaining a planar interface. (13~ (3) The temperature of any point in the liquid zone should not be much in excess of the melting point. This prevents significant unbalance between the dissociation pressure of the liquid and GoASrod the ambient. (4) A slow rate of growth should be used. The most satisfactory method used by the authors for growing crystals of GaAs has been the floating-zone technique commonly used in silicon technology. The application of this technique to furnaces Control~ l ~f~-qR.Ecoil gallium arsenide has been described by WHELAN and WHEATLE¥(4). Fig. 5 shows schematically the equipment which the present authors have used. A square-section rod of GaAs is cut from a cast ingot and fitted to a quartz chuck by careful grinding of its edges. A seed crystal is fitted to the lower chuck and held in position by a constriction in the quartz, while the GaAs rod is supported by a silica pin sealed through the upper chuck. A small separation of seed and rod is necessary to allow expansion of the material when heated to the temperature of the ambient furnaces. Excess arsenic is added to the system before evacuation. Silica rods are fused to either end of the assembly, the lower one being supported by a vertical pillar Fro. 5. Floating-zone equipment. motor-driven at a rate of between 4 and 5 in/hr and the upper one passing through a guide to keep the assembly vertical. The furnaces are separated agitation produced within the zone by the circuby a gap of approximately 0.75 in. in which a lating r.f.-~currents leads to homogeneous material three-turn work coil of an 8 Mc/s r.f. generator is free from arsenic vapour "pockets". The principal disadvantage is the limitation in situated. Annular baffles, and a cylindrical mica window between the furnaces, are employed to the diameter of the crystal which may be grown. reduce convection currents through the furnaces. The maximum length of a stable zone has been The zone is initially formed at the butt joint calculated by HEYWANG(16). This depends on both between the rod and the seed. Having produced a the surface tension of the liquid and its density. zone such that the tangent to the lower solid- Increasing the diameter of the rod while preservliquid interface is vertical, it is traversed up the ing a zone length less than the maximum value rod. Fig. 6 shows an example of a crystal grown leads to the practical problem of obtaining a comin this way. The seed crystal was cut from a large plete melt-through. The surface tension of liquid
F. A. C U N N E L L ,
J. T. E D M O N D
GaAs has not been measured but the greater density of GaAs compared with silicon makes it rather less suited than silicon to this method. In addition to the use of this technique for the growth of single crystals, zone purification by repeated passes has also been attempted. While this operation has obvious advantages over the more conventional zone purification described previously, the present apparatus is not ideally suited to it for the following reasons. Firstly, there is no facility for adjustment of the length of the zone and slight variations in the diameter of the zone tend to be accentuated by successive passes. It has been
and W. R. H A R D I N G
ways prior to making the compound, the results may be considered, broadly, under two headings: (1) for GaAs, prepared and directionally frozen as described in Subsection (b). (2) for GaAs, which has been zone-refined. (1) A number of ingots of GaAs have been made recently using gallium produced by AluminiumIndustrie-Aktien-Gesellschaft. This was used without further treatment while the arsenic from the American Smelting and Refining Co. was sublimed and treated with lead as described earlier. Specimens cut from these ingots at the first end to freeze have had a Hall constant (R) from --100 to
7000 6000 500~ I
:~ ~ooo -
I IOOOI O" IO00
"iRI, cm3 C< FIG. 7. Mobility of electrons in GaAs of various purities at 20"C. shown that this difficulty can be overcome in a more recent equipment by introducing a cylindrical sliding-joint fabricated in quartz. This consists of a quartz rod approximately 16 m m in diameter which is ground and polished over a length of 6-7 cm to be a close fit (approximately 10 F clearance on the diameter) in a quartz tube of similar internal surface finish. This improvement makes the equipment demountable and at the same time a negligible amount of arsenic is lost during operation. Secondly, there is a gradual deterioration in visibility through the silica envelope owing to material depositing on the inner wall. T h e greatest number of passes accomplished has been five.
(e) Purity of gallium arsenide Although gallium and arsenic have been obtained from several sources and treated in a variety of
--1200 cm a C -1 and an electron mobility (R~) from 4000 to 6700 cm 2 V -1 see -J, the best value, at room temperature. A typical example had an electron mobility of 9000 cm 2 V -1 sec -1 at 90°K. Specimens from the last end to freeze usually had higher resistivity, sometimes by several orders of magnitude. It is believed, however, that in these cases the number of carriers was not an indication of the total impurity content (probably ,-~1017/rcma). h is likely that the high resistance arises either through the compensation or over-compensation of donor by acceptor impurities with relatively large ionization energy, possibly oxygen or copper. (2) In the case of horizontally refined ingots the values of R were all negative and smaller than those described above in (1). This was taken to mean that the GaAs contained donor and acceptor
OF G A L L I U M A R S E N I D E
impurities and that the latter were swept more readily to the last end to freeze. As a result the total impurity content would be lower over the majority of the ingot than if treated by method (1). One serious inconsistency, which has not yet been resolved, is that material made by process (1) shows higher values of electron mobility than that made by process (2). This might happen if impurities in material (1), donor and acceptor, were associated to make uncharged pairs with a small scattering cross-section. The results obtained on ingots which were refined by the floating-zone method show some similarity to those discussed above [both (1) and (2)]. The apparent segregation of impurities is better in this experimental arrangement than when the ingot is zone-refined horizontally, but it is believed that this is due to less contamination from boat and walls. One significant advantage of any method employing r.f. heating is the improved homogeneity which results from the agitation of the melt. A curve is shown in Fig. 7 correlating Hall constant and electron mobility for a variety of specimens measured. While it is believed that the intrinsic Hall constant in the range 102-10acm3 C -1 cannot be used as a measure of the total impurity content in the specimens, it is significant that Ra increases with R. I f the total impurity content is in the region of 1017 cm -3 or greater it seems unlikely that the compensation of donors and acceptors could be a random process. It is possible that some form of automatic self compensation takes place. During the work, a considerable amount of spectrographic information on impurities in gallium and arsenic has been gathered. Copper, silver, magnesium, calcium and lead have been most frequently detected in gallium; and copper, silver, magnesium, calcium, sodium, iron and silicon in arsenic. Such elements as oxygen, sulphur and selenium, which are not detected by spectrographic techniques, are believed to be present in one or the other.
3. MEASUREMENTS A N D TECHNIQUES
(a) Electrical properties of impurities As is well known, the electrical behaviour of semiconductors depends on the amount and nature
of the impurity content. Observations on this behaviour are often useful in the identification of impurities and indicate which elements may be suitable for the purpose of doping. Consequently, a study was made using the known impurities in GaAs, i.e. those detected by spectrograph in gallium and arsenic, and certain other elements such as Zn, Sn, Si and Te. Silicon was of special importance since the possibility of pick-up of this element from quartz-ware must always be borne in mind. Work on similar lines involving impurities in InSb, InAs, GaSb and GaAs has been described by EDMOND(17). The method employed to prepare the material was to melt about 5 g of relatively pure GaAs in a small evacuated quartz tube with a few milligrammes of impurity and a little excess arsenic. The material was kept in a liquid state at 40°C above the melting point for 1 hr and then cooled slowly. Single-crystal sections were usually formed of sufficient size to provide rectangular specimens for measurements of R and a. The results are shown in Table 1. The seventh column is calculated from the weight of impurity and of the compound on the basis of one carrier per impurity atom and assuming that R = 1/ne where n is the number of carriers per cm 3 and e is the value of the electronic charge. The measured and calculated values of R differ not only because of solubility considerations but also as a result of some segregation during solidification. The results show some interesting features: (1) Copper is an acceptor impurity, as found also by FULLER and WHELAN(ls). Silver and gold seem to be much less soluble than copper. (2) Magnesium behaves as a donor impurity. This rather unexpected behaviour could be due to magnesium atoms occupying an interstitial position in the lattice. (3) Silicon is a donor impurity. Recently KOLM eta/. (19) have shown that silicon is soluble in GaAs up to 0.5 per cent and they suggest that silicon atoms substitute for nearest-neighbour pairs. The fact that the present work shows that silicon is a donor impurity implies that silicon atoms may also replace single gallium atoms. (4) Iron is an acceptor impurity. Measurement of resistivity over a range of temperature shows an ionization energy of 0-37 eV.
F. A. C U N N E L L ,
J. T . E D M O N D
and W. R. H A R D I N G
Table 1 BEFOREDOPING R (cm3/C -1) p X 104(~cm) (290°K) (290°K) --38 --48 --120 --86 --86 --86 --152 --86 --75 --86 --86 --35 --80 --72 --53 --88 --310
300 280 280 280 280 479 280
AFTER DOPING Impurity added Ca
Au Ag Mg Mg Mg Ca Zn
Si Si Sn S Te Te Pt ~:i"e
R (cma/C -1) (290°K) +883 +94 --78 --89 --5 --2'3 --2"7 --17-1 +1"1 --2"5 --1 '5 --13 --1-4 --2 "8 --1 '8 --10 --301 -]-9800
(b) Heat treatment of GaAs I n the course of measurements of impurity diffusion in GaAs it has been found that the electrical properties of the material, normally n-type, may suffer a pronounced change as a result of heat treatment. T h e treatment has consisted of heating mostly at 1000°C for 1-60 hr in a small evacuated capsule of quartz (Vitreosil) and cooling rapidly to room temperature. T h e modification to the electrical properties has always been such that the electron concentration has been reduced and in some cases the type of conductivity has changed from n-type to p-type. A n u m b e r of experiments have been carried out to investigate these changes and the conclusion reached is that in a n u m b e r of cases they are due to contamination by copper. T h e copper has been identified in the GaAs by means of a square-wave polarograph and by comparing resistivities as a function of temperature of heat-treated specimens and copper-doped specimens, all p-type. Furthermore it is found that the changes are an order of magnitude smaller if the purest quartz (Spectrosil) is used instead of Vitreosil, e.g. in Spectrosil the change in carrier concentration in the extrinsic range Ap is estimated
p x 104(flcm) R (cma/'c -1) Calculated R (290°K) (90°K) (cma/C -*) 68,300 3400
P P ? ?
16"5 61 110
0.52 0.12 0.11 0.1 0"23 0"39 0'13 0'16 0"62 0'24 0"86 0"045 0'80 0"29 0"11
(n?) p 1l
(n?) n )t n
to be about 1016(cm -3 for 15 hr at 1000°C and in Vitreosil 1-5 × 1017 cm -3 for the same time and temperature. T h e effect also depends on the duration of the heating, e.g. Ap is equal to 1017 cm -a for 1 hr and 7"5× 1017 cm -3 for 60 hr at 1000°C (Vitreosil) for specimens from the same ingot. A tendency for the specimens to return towards their original state on annealing has been noted. This effect would be consistent with the precipitation of copper at dislocations as suggested by DIXON and ENRIGHT(2°) in the case of InAs. It is hoped to publish these results later ill more detail. (c) Preparation of contacts Low-resistance ohmic contacts to both n-type and p-type GaAs are required for routine electrical measurements and in the fabrication of devices. T h e method to be employed is determined partly by the area of contact required. T h e following techniques have been used successfully. (1) Pure tin or indium heated to between 400 and 500°C in contact with clean n-type material in an inert atmosphere produces a strong mechanical join. This is suitable for end contacts to Hal[
specimens or base contacts to rectifiers. Indium, doped with a small percentage of zinc (4"8 atomic per cent zinc), is suitable for p - t y p e material. (2) Silver electroplating can be used for end contacts to both p - t y p e and n-type material. F o r p - t y p e samples one layer of silver is usually sufficient but for n-type specimens the technique employed is to plate, discharge a condenser (say 0.25/z F at a suitable voltage) between the ends of the specimen, and plate again. (3) Gold wires " f o r m e d " either by a condenser discharge or by the passage of an a.c. current are useful for small-area Hall or resistivity probes on both n-type and p - t y p e material. Bulk contamination of the specimen with gold is unlikely since it is believed that the diffusion of gold is slow in GaAs (D~-~2"5 x 10 -11 cm 2 sec 1 at 1000°C(21)). (4) Fine gold wires can be bonded to a clean GaAs surface by the combined application of pressure and heat. T h e temperature required is well below the melting point of gold. T h i s type of contact is especially useful where the contact has to be very restricted in area. Alternatively, fine gold wires (0.002-0.003 in. dia.) can be alloyed with a clean GaAs surface heated to a temperature of 550-600°C in an inert atmosphere. T h e range of application of each of these methods could be extended by suitable modifications of the experimental technique. (d) Etching SCHELL(22) has discussed preferential and nonpreferential etches for GaAs, and their application to the display of dislocations. W h e n it has been necessary to clean the surface of GaAs the authors have found that a warm mixture of 3 volumes concentrated sulphuric acid, 1 volume of water and 1 volume of hydrogen peroxide leaves a polished finish;* a layer of GaAs about 0-0005 in. thick is removed in 2 min. A good alternative is a hot mixture of concentrated hydrochloric acid and nitric acid (approximately five to one by volume). A less reactive mixture than the above, consisting of 1 volume sulphuric acid, 8 volumes of water and 1 volume of hydrogen peroxide, is valuable for showing grain structure. * This etch was suggested by N. RAIN of this laboratory.
Hydrochloric acid and nitric acid, cold, in the ratio of five to one, with or without a small amount of water, can be used to detect the position of a p - n junction. (e) Diffusion p - n junctions have been prepared by diffusion of zinc from the vapour phase into n-type GaAs. A t t e m p t s to measure the diffusion constant of zinc in GaAs by comparison of junction depths in material of two different electron concentrations have been unsuccessful. Subsequently radio-tracer techniques have established that the zinc diffusion process does not follow the simple concentrationindependent laws at the concentrations examined and have thereby explained the failure to obtain satisfactory results from the electrical method. (23)
Acknowledgements--The authors wish to thank Mr. N. A. C, THOMPSONand Mrs. I. S. PINN0CKfor making some of the measurements, Miss J. GOODSONfor preparing some of the material, Dr. 0. SIMPSON-for assistance in preparing the manuscript, and the Admiralty for permission to publish the work.
1. O. G. FOLBERTHand H. WEISS, Z. Naturf. 10a, 615 (1955). 2. J. T. EDMOND, R. F. BROOM and F. A. CUNNELL, Report of the Meeting on Semiconductors, Physical Society, London, April 1956, p. 109. 3. J. VANDEN BOOMGAARDand K. SCHOL, Philips Res. Rep. 12, 127 (1957). 4. J. M. WHELAN and J. H. WHEATLEY,J. Phys. Chem. Solids 6,169 (1958). 5. J. R. RIeHAROS,Nature, Lond. 177, 182 (1956). 6. D. P. DETWlLERand W. M. Fox, J. Metals, N. Y. 7, 205 (1955). 7. E. SeHILLMAN, Z. Naturf. lla, 463 (1956). 8. T. C. ~IARMAN, E. P. STAMBAUGH and H. L.
11. 12. 13. 14.
GOERING, Semiconductor Symposium of the Electrochemical Society, Cleveland, October, 1956. J. M. WHELAN, J. D. STRUTHERSand J. A. DITZENBERGER.Private communication. J. VAN DEN BOOMGAARD, F. A. KROGER and H. J. VINK, J. Electronics 1,212 (1955). R. GREMMELMAIER,Z. Naturf. l l a , 511 (1956). J. L. RICHARDS,J. Sci. Instrum. 34, 289 (1957). I. G. CRESSELLand J. H. POWELL,Progressin Semiconductors, Vol. II, p. 37. Heywood, London (1957). D. C. BENNETTand B. SAWYER,Bell Syst. Tech. J. 35, 637 (1956).
J. T .
15. N. B. HANNAY, Semiconductors p.413. Reinhold, New York (1959). 16. W. HEYWANO, Z. Naturf. l l a , 238 (1956). 17. J. T. EDMOND,Proc. Phys. Soc. Lond. 73,622 (1959). 18. C. S. FULLER and J. M. WHELA2q,J. Phys. Chem. Solids 6, 173 (1958). 19. C. KOLM, S. A. KULIN and B. A. AVERBACH,Phys. Rev. 108, 965 (1957).
R. H A R D I N G
20. J. R. DIXON and D. P. ENRIGHT,J. Appl. Phys. 30, 573 (1959). 21. H. B. CLARKE. Unpublished work at Services Electronics Research Laboratory (1959). 22. H. A. SCHELL, Z. Metalk. 48, 158 (1957). 23. J. W. ALLEN and F. A. CU.WNELL,Nature, Lond. 182, 1158 (1958).