RU2369669C2 - Substrate for growing of epitaxial layers of gallium nitride - Google Patents

Abstract

FIELD: electric engineering.

SUBSTANCE: invention is related to electronic engineering, namely, to technology of materials for creation of information display and processing devices. As material of substrates used for growing of epitaxial layers of gallium nitride, a row of compounds is suggested - single crystals of intermetallides, selected from group that includes silicide of magnesium (MnSi), palladium silicide (Pd₂Si), manganese stannate (Mn₃Sn), iron stannate (Fe₃Sn), vanadium phosphide (VP), aluminium zirconide (Zr₃Al) with moderate melt temperatures.

EFFECT: advantages of this class of compounds compared to available ones consist in higher quality of gallium nitride films grown on substrates from mentioned compounds.

Description

The invention relates to electronic equipment, in particular to the technology of materials designed to create semiconductor devices for displaying and processing information. Recently, the scope of work on creating powerful semiconductor devices based on wide-gap semiconductors, LEDs, and semiconductor lasers emitting in the blue and blue spectral regions, in particular, on the basis of AlN, GaN, has sharply increased. However, for the widespread use of these devices, a number of problems associated with the technology of materials for their manufacture should be overcome.

It is known that the high excitation threshold and short life of lasers based on AlN, GaN, InN are associated with low crystalline perfection of the layers used for the manufacture of lasers based on them. Their low quality is primarily due to the low crystalline perfection of the materials used as substrates for growing epitaxial layers of these materials [1]. To date, dielectric or wide-gap semiconductor materials such as sapphire Al ₂ O ₃ , aluminum magnesium spinel MgAl ₃ O ₃ and silicon carbide SiC, galate and lithium aluminate LiGaO ₂ , LiAlO ₂ are used for these purposes.

The successful use of these materials for use as substrates is hindered by the presence of common disadvantages inherent in these materials. One of the reasons for their low crystalline perfection is related to the fact that they are all high-temperature materials (sapphire has a melting point of 2030 ° С, magnesium spinel - 2050 ° С, silicon carbide ~ 2600 ° С, lithium galate - 1600 ° С, lithium aluminate - 1900 ° C). As a rule, they are grown under nonequilibrium conditions with a significant number of defects in the form of block structures with small-angle boundaries. The epitaxial layer grown on their surface inherits all these defects. Another important drawback of these materials as single-crystal substrates is the significant mismatch of their crystal lattices with aluminum nitride. It is known that the mismatch of the crystal lattices of the growing layer and the substrate leads to mechanical stresses at the boundary of the epitaxial layer and the substrate with the formation of small-angle boundaries. For gallium nitride on sapphire, this mismatch is ~ 14%, on magnesium-aluminum spinel ~ 9% and the smallest on silicon carbide ~ 3.5%. Lithium galate and lithium aluminate generally belong to orthorhombic and tetragonal syngony, respectively, and a comparison of the mismatch of the periods of the crystal lattices in the epitaxy plane can be made only on one of two axes lying in the epitaxy plane.

In [2], in order to eliminate the high density of defects associated with the high temperature of growth of these single crystals and the large mismatch between the periods of the crystal lattices, it was proposed to use a new class of materials as the materials for epitaxial growth of gallium nitride: monosilicides of metals of the IV period of the periodic system and solid solutions on their basis (conductive). The melting point of iron silicide of 1410 ° C should be considered moderate (for comparison, the melting point of silicon is 1430 ° C, of pure iron is 1535 ° C). The fact that this compound consists of two components with similar melting points and their vapor pressure [3], made it possible to simply solve the problem of growing stoichiometric, large, crystallographically sufficiently perfect crystals of iron silicide.

When growing single crystals of iron silicide in vacuum (p ~ 1 · 10 ^-4 mm Hg) by the Czochralski method, it was found that during their growth on the surface of the iron silicide melt, insoluble films and solid particles are formed and gradually accumulate in the crystallization zone, contributing to polycentric growth and transformation of a single crystal into a polycrystal. It was shown that the molten surface of FeSi is slowly but constantly oxidized due to the residual oxygen of the vacuum chamber and the desorption of oxygen from the crucible during the entire growth process of the SiFe single crystal, since it (molten liquid) is chemically very reactive. The effect of this effect increases with increasing crystal size and the resulting increase in the exposure time of the melt surface in the environment. Due to the rather high growth temperature of iron silicide, particles of iron carbide can also form on the surface of the melt. The most primitive form of cleaning such a surface during the growth of a single crystal was the periodic assembly of the formed oxide film and solid particles from the melt surface and their removal with some kind of fork or broom analog. The formation of oxides on the surface of the iron silicide melt is tolerable in the case of single-piece growing of small single crystals and is highly undesirable in the mass production of large single crystals, lowering their quality and decreasing the yield of high-quality single crystals.

The formation of iron carbides due to the interaction of molten iron with crucible graphite can be effectively suppressed by first coating the surface of the crucible with carbon silicide. The process of oxidizing the melt surface during the growth of single crystals can be reduced only by removing oxygen from the atmosphere surrounding the crucible with the melt, creating a very high vacuum (above 1 · 10 ^-9 mm Hg) or a reducing medium without traces of oxygen.

The first seems quite expensive, and in the second there is the possibility of partial dissolution of hydrogen in the melt and single crystal SiFe.

It is known that the rate of oxidation of melts having chemically active components, such as FeSi, exponentially depends on the temperature of the melt, and its reduction may be the most effective way to reduce the rate of oxidation of the surface of the melt.

Melts of solid solutions based on iron silicide described in [2], used in the preparation of substrates for GaN epitaxy, are less prone to oxidation compared to iron silicide. A serious drawback of these solid solutions was that the grown single crystal is not uniform in composition: during their growth, the chemical composition of the single crystal along the length of the ingot continuously changes. At the beginning of the ingot, the composition of the single crystal contains an excess of the refractory component, the tail of the single crystal contains an excess of the more fusible component of the solid solution. Currently, there are methods that make it possible to fairly accurately maintain the chemical compositions of solid solutions along the entire ingot during their growing, but they are quite complex and significantly complicate the process of growing single crystals of solid solutions, especially when growing large ingots.

The technical problem solved in the present invention is to simplify and improve the technology of growing single crystals, leading to an increase in the quality of single crystals used for epitaxy and to improve the properties of epitaxial films of gallium nitride. This is achieved through the use of intermetallic compounds with a lower melting point, namely manganese silicide (MnSi), palladium silicide (Pd ₂ Si), manganese stannate (Mn ₃ Sn), iron stannate (Fe ₃ Sn), vanadium phosphide (VP) , zirconide aluminum (Zr ₃ Al).

The transition metal intermetallics for epitaxy of gallium nitride proposed in this application crystallize in two structural types: cubic — MnSi, Zr ₃ Al, and hexagonal — all others.

The magnitude of the mismatch between the periods of the crystal lattices of the substrates of manganese silicide and aluminum zirconide in the (111) orientation and the epitaxial layer of gallium nitride (0001) in the epitaxy plane is 1.33% and 3%; for hexagonal crystals of palladium silicide - 2.04%; for iron stannide - 0.9%, for manganese stannide - 4.45%, for vanadium phosphide - 0% for the (0001) plane.

The features of these intermetallic compounds are as follows.

1. Their melting points are significantly lower than the melting points of the intermetallic compounds described in the prototype. For example, manganese silicide melts at a temperature (1275 ° C) close to the temperature of gallium arsenide (1248 ° C). The proximity of the melting points of manganese silicide and gallium arsenide turned out to be fundamentally important. This made it possible to grow single crystals of manganese silicide according to the technology developed for growing single crystals of gallium arsenide, namely, to grow single crystals of manganese silicide under a protective layer of flux - boron oxide. The latter dissolves in itself all the oxides (both manganese and silicon), thereby protecting the surface of the growing crystal from oxide films and oxide crusts formed on the melt surface. This made it possible to grow high-quality single crystals of MnSi and other intermetallic compounds listed above, eliminating traces of oxides on the surface, often leading to the formation of twins and other growth defects. Their lower melting temperature also makes it possible to significantly reduce the likelihood of carbides forming in the melt and their contamination of the melt and single crystals, which also reduce the quality of single crystals.

2. The proposed alloys are stoichiometric compounds (except for Mn ₃ Sn, which has a narrow, ~ 2%, homogeneity region), this excludes the possibility of a noticeable change in the composition of the single crystal along and across the entire ingot, and therefore the technology for growing single crystals of uniform composition is significantly simplified. Their subsequent use as substrates significantly increases the uniformity of the epitaxial layers of gallium nitride grown on the surface of these single crystals.

Single crystals of manganese silicides and other intermetallic compounds listed above were grown from pyrocarbon coated with silicon silicide, alundum or nitride-boron crucibles with a volume of 150-250 mm ³ under a melt layer of anhydrous boron oxide and had the following dimensions: diameter 25 mm, length up to 100-180 mm .

The growing of manganese silicide and aluminum zirconide single crystals by the Bridgman method from a crucible with a conical seed nose is difficult due to the appearance of many crystallization centers in the embryonic channel, which is very characteristic of the growth of cubic syngony crystals, where there is no preferred growth direction. Therefore, we were growing single crystals of manganese silicide and zirconide aluminum by turning the melt onto nose mounted seeds made in advance from MnSi and Zn ₃ Al single crystals, which were extracted from coarse-grained polycrystals previously grown by the Bridgman method in a crucible with a conical nose. Usually they were a single crystal with linear dimensions of 2 × 2 × 15 mm ³ .

The grown single crystals were metallographically and radiographically tested for their crystalline perfection: axial and radial inhomogeneity, the presence of voids in the volume of the single crystal, and the inclusion of excess compound components in them. The optimal growth rate of single crystals was determined. She was in the range of 2-6 mm / h. The dislocation density in the grown single crystals was checked. It turned out that their density on average was in the range of 1-5 × 10 ³ / cm ² .

Plates were cut out of such single crystals, the normal to the surface of which is parallel to the axis of the third order of the crystal. The plate thickness was ~ 1 mm. The plates were polished according to standard technology for silicon, polished mechanically and chemically (in a mixture of HF + HNO ₃ ).

Epitaxial layers of gallium nitride were grown by molecular beam epitaxy (MPE) in an MPE apparatus using pure gallium (99.999) and compressed high-purity compressed ammonia balloon (5-30 l / h) with ionization of an ammonia beam in a Penning cell as starting materials by analogy with [4]. The growth rate of gallium nitride layers varied in the range of 0.01-0.05 μm / h. The crystal structure of the growing gallium nitride layer was controlled by an integrated electron diffractometer with an accelerating voltage of up to 50 kV.

The mutual crystallographic orientation of the epitaxial layers of gallium nitride relative to substrates of cobalt silicide, aluminum zirconate (cubic syngonium) and other intermetallic compounds used was established: Pd ₂ Si, Mn ₃ Sn, Fe ₂ Sn, VP (related to hexagonal syngonium):

/ 110 / a (MnSi) = 0.443√2 = 0.6276 nm || / 1010 / a (GaN) = 3.180 × 2-0.6224 nm

(111) (MnSi) || (0001) (GaN),

/ 1010 / a (Mn ₃ Sn) = 0.566 nm || / 1120 / a (GaN) = 0.3l8 · √3 = 0.558 nm

(0001) (Mn ₃ Sn) || (0001) (GaN)

On the (111) MnSi surface, epitaxial films of gallium nitride grow from 450 ° С to 1000 ° С. Approximately the same conditions (the beginning of epitaxial growth) are satisfied for the other substrates listed above. The dislocation density in the grown films is ~ 10 ^–3 / cm, ² it substantially depends on the dislocation density in the substrate and the growth conditions of the GaN epitaxial layers. This gives grounds to expect their substantial decrease with the improvement of the entire technological chain: increasing the purity of the starting materials, single crystal growth technologies, improving the quality of surface treatment of substrates, optimizing the growth conditions of epitaxial gallium nitride films, and increasing the overall level of work.

Literature

1. N. M. Johnson et al. Physics today. v. 53, 2000, No. 10, p. 312.

2. S.A. Aitkhozhin. Substrate for growing epitaxial films and layers of gallium nitride. RF patent No. 2209861 dated 06/15/2001, IPC С30В 19/12.

3. A.N. Nesmeyanov. Steam pressure of chemical elements. M .: Academy of Sciences of the USSR, 1961, p. 374.

4. X. Y. Liu et al. Growth of GaN and GaN / AlN multiple quantum wells on sapphire, Si and GaN template by molecular beam epitaxy. Jour. Crys. Growth, v. 300, 2007, No. 1, p. 45-49.

Claims (1)

A substrate for growing epitaxial layers of gallium nitride made of a single crystal of an intermetallic compound, characterized in that the single crystal of an intermetallic compound is selected from the group consisting of manganese silicide (MnSi), palladium silicide (Pd ₂ Si), manganese stannate (Mn ₃ Sn), iron stannate (Fe ₃ Sn), vanadium phosphide (VP), aluminum zirconide (Zr ₃ Al).