GB2118860A - Crystal growth - Google Patents

Abstract

The temperature range of operation of liquid encapsulated Czochralski crystal growth from the melt is extended to a new lower temperature range by the use of fluoride glasses. The use of these glasses to permit liquid phase epitaxy from volatile solutions is also described, as is their use to protect seed-substrates from decomposition before they become covered by the growth melt, covering the surface of the melt or solution with a layer of an inert fluoride glass having a viscosity of not more than 60 poise at the temperature of the melt or solution to inhibit volatilisation.

Description

SPECIFICATION Crystal growth This invention relates to the growth of crystalline material from the melt or from solution in circumstances in which the vapour pressure of the melt, or the solute or solvent, is significant enough to present problems in obtaining adequate control of the composition of the grown material.

In the context of growth from a melt this problem has been solved for a limited number of semiconductor materials by the liquid encapsulated Czochralski (LEC) technique in which a semiconductor crystal is pulled through a "blanket" of boric oxide as encapsulating medium floating on the surface of the melt. Above the encapsulating medium is maintained an inert gas at a pressure exceeding the vapour pressure of the melt. In the case of growing gallium phosphide and indium phosphide the dissociation of the molten material provides a vapour pressure at its surface of some tens of atmospheres; in the case of gallium arsenide the vapour pressure is about one atmosphere. Diffusion of volatile species through the encapsulating "blanket" is found to be negligible and the nett effect is as if the dissociation were suppressed.This is a matter of considerable practical significance, since otherwise a balancing pressure of the actual dissociated component (e.g. phosphorus or arsenic vapour) would have to be applied to the melt, and this is difficult to do.

Ideally the encapsulating medium should be a glass. One useful attribute of a glass is that it has no sharp freezing point which would be liable to induce abrupt mechanical strain on a growing or grown crystal. The viscosity of a glass changes smoothly as function of temperature. Once any portion of the growing crystal rises above the level of the encapsulant, it too needs to be protected against the effects of dissociation, and this protection is conveniently provided by ensuring that the encapsulant has a high enough viscosity for an adequate thickness of layer of encapsulant to be carried up with the growing crystal. This, and the need to avoid convective heat transfer problems associated with a low viscosity, sets a lower limit to the viscosity that is rather greater than that of typical inorganic materials in their liquid phase.On the other hand the viscosity must not be so high as to disrupt the growth process.

In addition to these viscosity requirements for the encapsulant, any candidate substance should be chemically inert with respect to the underlying melt. This limits the choice to substances with great stability, i.e. high free energy of formation. It must also be available in very great chemical purity and for use in growing semiconductor crystals it must not contain constituents which have undesirable effects upon the physical, particularly electronic, properties of material pulled from the underlying melt. Thus alkali metals and transition metals such as manganese, iron, chromium, and gold must be avoided. The only practical encapsulant has been boric oxide. This is available in extremely high purity but, despite'its high free energy of formation, does not appear to be entirely inert, there being evidence that very small amounts of boron and oxygen contaminate LEC grown crystals.Another problem is that it becomes unuseable at temperatures significantly beneath 1 1000C because the viscosity is too great. This is demonstrated by the fact that there is considerable evidence that when LEC is used to grow indium phosphide (melting point about 10700C) the sharply increasing viscosity of boric oxide with falling temperature exerts mechanical forces at the solid liquid interface which encourage twinning. This means for instance that it has not been possible to grow for instance gallium antimonide (melting point 7100C) or mercury telluride melting point 6900 C) by LEC.

The present invention is concerned with the use of fluoride glasses as alternative encapsulants to boric oxide. A feature of these fluoride glasses is that in comparison with boric oxide, their temperature viscosity characteristics are displaced towards lower temperatures so that it is possible to use the encapsulation technique at lower temperatures than is possible with boric oxide.

According to the present invention there is provided a method of crystal growth performed in an inert atmosphere, which growth is from the melt or from solution by liquid phase epitaxy, wherein a layer of an inert fluoride glass having a viscosity of not more than 60 poise at the temperature of the melt or solution covers a surface from which significant volatilisation would otherwise occur to inhibit volatilisation by isolating that surface from said inert atmosphere which is maintained at a pressure in excess of that in equilibrium with that surface.

There follows a further description of the invention in which reference is made to the accompanying drawing which is a graph showing the dependence of viscosity upon temperature of three glasses.

It is found that fluoride glasses being developed for mid infra-red optical fibre communication purposes also have suitable properties for LEC use. Chemically these glasses are formed from extremely stable compounds such as fluorides of zirconium, hafnium, thorium, barium, lanthanum, and aluminium. Moreover techniques evolved for obtaining these glasses in suitably pure form for optical communications use is capable of providing material of more than adequate purity for LEC. It is noted that these glasses do not need to be stabilised with alkali metal constituents which as previously noted are undesirable impurities to be in contact with a semiconductor.

By way of example the accompanying graph shows the dependence of viscosity upon temperature of two particular fluoride glasses and compares them with that of boric oxide. One of these glasses is a fluorozirconate glass having the composition 57 mole % ZrF4, 36 mole % BaF2, 3 mole % LaF3, 4 mole %AIF3, while the other is an equivalent fluorohafnate glass in which all the 57 mole % ZrF4 is replaced with HfF4. These are well behaved glasses, and provided oxygen contamination does not occur, do not recrystallise on prolonged heating in the lower viscosity range.

The high temperature limit of application of these glasses will be determined by the volatilisation of 7.of4 or HfF4, which will be lower with the latter on account of its higher molecular weight (and also because of the stronger bonding from 5d-orbitals than from 4d-orbitals). Volatilisation is also reduced by the addition of other constituents, in this case the Bay2, LaF3 and AIR3, which lower the chemical activity of the tetrafluoride. Further limitations are set by the rather high densities of these glasses: ZrF4 and HfF4 band glasses have densities in the region of 4 and 6 respectively, and melts of metals or semiconductors to be protected must clearly have higher densities. These glasses can be used for LEC to about 7500C and perhaps even above that.They must be used in an inert atmosphere in order to suppress oxidation, but 'this presents no great problem since argon or another inert gas atmosphere is invariably used in LEC. The lower temperature limit of useful operation is set by viscosity considerations and comparing the viscosity curves for these glasses with that of boric oxide it is seen that the lower limit set by the 60 poise criterion is about 4250C for the fluorohafnate glass and slightly lower for the fluorozirconate glass. Thus the useful temperature range is seen to cover in particular the LEC growth of gallium antimonide (melting point 7100C) and mercury telluride (meiting point 6900C).

While the foregoing description has concentrated on the use of these glasses for LEC, i.e. pulling crystals from the melt, it should be clearly understood that the invention has another application in the enabling of crystal growth by liquid phase epitaxy from solutions with relatively high vapour pressure. In conventional liquid phase epitaxy apparatus volatility of the solution is liable to present problems in control of composition when the vapour pressure of the solution exceeds a few Torr, but by covering the solution with a "blanket" of encapsulant after the manner used in LEC it is possible to achieve satisfactory epitaxial growth at much higher vapour pressures, thereby permitting for instance the growth of cadmium telluride or mercury cadmium telluride from solution in tellurium, and gallium arsenide or gallium aluminium arsenide from solution in arsenic.

A related volatility problem in liquid phase epitaxy concerns the effects of loss by volatilisation of material from the surface of the seed substrate prior to its being covered by the melt. Thus for instance, the surface of an indium phosphide seed-substrate becomes impaired when held above about 4000 C. As a result it has been the practice to precede epitaxial growth on an indium phosphide surface with a short etch in indium to remove the damaged surface layer. This approach to the problem presents difficulties in thickness control and surface morphology. This is avoided by using a glass 'blanket' to cover the seed substrate until it is presented to the melt.

Boric oxide cannot be used for protecting an indium phosphide surface because it is too viscous in the relevant temperature range of about 400-5000C. However, fluoride glasses are satisfactory for this purpose, provided that the density of the glass is not as large as that of the epitaxy solution so that when the seed is introduced to the solution the glass becomes displaced from the substrate by the solution. The glass is applied to the surface of the seed substrate in the form of a frit which will coalesce to form an impervious blanket before significant deterioration of the seed substrate surface has occurred.

The fluoride glasses required for the application of this invention can be prepared in adequately pure form by preparing the constituent fluorides from their oxides by heating those oxides with excess ammonium bifluoride. It is desirable to minimise the risk of oxygen contamination, and the residual hydroxyl contamination resulting from use of the ammonium bifluoride route can be reduced by processing the glass in a chlorine atmosphere or by bubbling carbon tetrachloride through it. For application to the growth of semiconductor material the final purity needs to be comparable with, a slightly better than that of the semconductor material which is typically of the order of parts per million. However, these glasses are required in still purer form, typically in parts per billion, for optical fibre applications, and the routes currently being developed for the preparation of such glass is likely to supersede the ammonium bifluoride route and to be attractive to meet the lower purity requirements for the crystal growth application of the present invention.

Claims (12)

1. A method of crystal growth performed in an inert atmosphere, which growth is from the melt or from solution by liquid phase epitaxy, wherein a layer of an inert fluoride glass having a viscosity of not more than 60 poise at the temperature of the melt or solution covers a surface from which significant volatilisation would otherwise occur to inhibit volatilisation by isolating that surface from said inert atmosphere which is maintained at a pressure in excess of that in equilibrium with that surface.

2. A method of crystal growth by the liquid encapsulated Czochralski technique, wherein the encapsulating medium is a fluoride glass having a viscosity of not more than 60 poise at the growth temperature.

3. A method of crystal growth from solution by liquid phase epitaxy, wherein the growth is performed in an inert atmosphere at a pressure exceeding the solution vapour pressure from which inert atmosphere the solution surface is isolated by a layer of an inert fluoride glass having a viscosity of not more than 60 poise at the growth temperature.

4. A method of crystal growth from solution by liquid phase epitaxy, wherein the growth is performed in an inert atmosphere maintained at a pressure exceeding the vapour pressure of the seed substrate at the growth temperature and wherein, immediately prior to being brought into contact with the solution, the surface of the seed substrate upon which material to be epitaxially grown is covered by a layer of an inert fluoride glass which has a viscosity of less than 60 poise at the growth temperature and has a density less than that of the solution such that when the seed substrate is introduced to the epitaxy solution the glass is displaced from contact with the surface by the solution.

5. A method of crystal growth as claimed in claim 4, wherein the glass is initially applied to the seed substrate surface in the form of a glass frit.

6. A method of crystal growth as claimed in any preceding claim, wherein the crystal is a semiconductor.

7. A method of crystal growth as claimed in claim 1 or 2, wherein the melt is gallium antimonide.

8. A method of crystal growth as claimed in claim 1 or 2, wherein the melt is mercury telluride.

9. A method of crystal growth as claimed in claim 3, wherein cadmium telluride or mercury cadmium telluride is grown from solution in tellurium.

1 0. A method of crystal growth as claimed in claim 3, wherein gallium arsenide or gallium aluminium arsenide is grown from solution in arsenic.

11. A method of crystal growth as claimed in claim 4 or 5, wherein growth is performed upon an indium phosphide seed substrate.

12. A method of crystal growth as claimed in claim 1, and substantially as hereinbefore described with reference to the accompanying drawing.