FR2833619A1 - Fabrication of crystalline semiconductor substrates involves using a molten layer of a second material for diffusion of atoms into a first material - Google Patents

Abstract

Crystalline growth method involves: forming a substrate incorporating an initial layer (20) of a first material presenting crystalline coherence and a layer (10) of a second material allowing, in the liquid state, the diffusion of atoms into the composition of the first material; and melting the layer of second material and growing a layer of the first material on the initial layer of the first material by subjecting the molten layer of the second material to a reactive gas incorporating at least one constituent of the first material. During the first stage the layer of second material is produced on the initial layer of the first material, without regard to crystalline coherence.

Description

<Desc / Clms Page number 1>

PROCESS FOR PRODUCING SEMICONDUCTOR SUBSTRATES
CRYSTAL.

Technical area
The present invention relates to a method for manufacturing crystalline semiconductor substrates, and in particular monocrystalline substrates. It is intended in particular to manufacture solid ingots of semiconductors such as SiC, CdTe, GaN, AlN or other materials for which obtaining ingots of good crystalline quality is difficult.

State of the art
The documents (1) to (3), the references of which are specified at the end of the present description illustrate techniques for manufacturing substrates.

 The manufacture of monocrystalline semiconductor substrates for use in the microelectronics industry generally includes the drawing of solid ingots, the slicing of the ingots and the polishing of the slices. This manufacturing technique, constantly improved, is particularly adapted to silicon. It makes it possible to obtain circular slices with a diameter of 300 mm and of very high crystallographic quality.

 For other semiconductor materials, on the other hand, it is almost impossible to control the crystalline quality over the entire length and over the entire diameter of an ingot. The thermodynamic parameters governing the bullion drawing process can not be maintained or controlled

<Desc / Clms Page number 2>

 in a way to obtain substrates of high diameter and good crystalline quality.

 In addition, for certain materials such as SiC, AlN, GaN, it is impossible to obtain, at the ordinarily accessible pressures and temperatures, a liquid phase suitable for drawing ingots.

 For the production of substrates made of these materials, other manufacturing techniques are envisaged. By way of example, for the manufacture of SiC substrates, a method known as Lely is used. This method essentially consists in sublimating a granular SiC feedstock at a temperature of the order of 2300 C and condensing it on a seed whose temperature is maintained greater than 2100 C.

 Other substrates, in materials such as GaN, are obtained by heteroepitaxy of a thin film on a heterosubstrate. Heterosubstrate means a substrate made of a given material and used as an epitaxial support for the formation of a layer of a material different from the material of the substrate. For example, sapphire can be used as a support material for GaN heteroepitaxy.

 During the manufacture of such substrates, difficulties are encountered related to the propagation of crystalline defects, such as dislocations. Dislocations may indeed propagate in the layer formed by heteroepitaxy, and alter the crystalline quality. Moreover, the heterosubstrate, or possibly adaptation layers with which it is provided, can constitute electrical barriers

<Desc / Clms Page number 3>

 parasites. These prohibit, for example, contacts made by the rear face.

 The document (1) mentioned above, indicates yet another method of manufacturing substrates, derived from the previous one, and aimed essentially at the production of substrates of silicon carbide (SiC).

This method essentially consists of growing a thin layer of SiC on a silicon substrate by heteroepitaxy and then using the SiC layer as a liquid phase epitaxial crucible. The silicon of the substrate is then used as a bath and as a source of growth material. However, to obtain a layer of SiC of acceptable quality, the process comprises complex and delicate operations consisting in particular in the realization of growth initiation tips. Obtaining a final substrate free of dislocations also remains random.

 Yet another technique for manufacturing substrates is to transfer by molecular bonding a thin layer of a given semiconductor material onto a substrate of a different material. This technique, however, does not allow the renewal of the semiconductor material of the donor substrate and therefore does not solve the problem of its manufacture.

Presentation of the invention
The aim of the invention is to propose a process for manufacturing crystalline substrates, and in particular monocrystalline substrates, which do not have the limitations of the processes mentioned above.

<Desc / Clms Page number 4>

 One aim is in particular to propose a method adapted to certain semiconductor materials whose physical properties do not allow the implementation of traditional drawing techniques.

 Another aim is to propose a simple process making it possible to obtain substrates of large diameter and of good crystalline quality.

 To achieve these aims, the invention more specifically relates to a crystalline growth process in which: a) a substrate is formed comprising an initial layer (20) of a first material having crystalline coherence, and a layer (10) of a second material, the second material allowing, in the liquid state, the diffusion of atoms used in the composition of the first material, b) the second material layer is melted and a layer of the first material is grown on the layer; initial material first (20), by subjecting the molten second material layer to a reactive gas comprising at least a component of the first material.

 According to the invention, during step a), without regard to the crystalline coherence, the second material layer is reported on the initial layer of first material.

 The second material is used to form a liquid bath for the growth of the first material. This is similar to growth in the liquid phase. The filler materials for the growth of the first material can in all or

<Desc / Clms Page number 5>

 part be supplied by the reactive gas. However, the second material, forming the liquid bath, can also be used as a filler material. It is then consumed as the first material grows.

 The initial layer of first material, which has a crystal coherence, for example which is monocrystalline, is used to initiate the growth of the first material in step b).

 More precisely, the face of the thin layer that is in contact with the second material layer has the particularity of acting as a growth germ for the first material. This feature is used in the method of the invention and thus avoids additional steps such as peak formation or sprouting steps.

 According to a particular implementation of the method, it is possible to relate the second material layer, for example, by bringing a block of the second material into contact with the layer of first material.

The second material block may be a slice of said second material, obtained, for example, by sawing or cleavage of a donor substrate, and transferred to the initial layer of first material.

 The donor substrate may, for example, be cleaved according to a technique known under the name Smart-Cut. This technique involves implanting light ions into a donor substrate before the donor substrate is transferred to a target substrate. The thickness of the future thin layer is fixed by the depth

<Desc / Clms Page number 6>

 implantation. After bonding between the implanted face of the donor substrate and the target substrate, heat treatments, and / or the exercise of mechanical forces, allow coalescence of the implanted species and weaken the donor substrate by promoting cleavage. The cleavage finally makes it possible to detach the remaining portion of the donor substrate while leaving the thin layer on the target substrate. In this regard, reference may be made to document (2) mentioned above.

 The cleavage technique makes it possible to use relatively thin layers of the first material, thereby saving the donor substrate. The thin layer of the first material has a thickness of the order of a few microns, for example of the order of 1 to 10 m for SiC. Moreover, the subsequent growth of the first material, according to the method of the invention, makes it possible to produce new substrates of the first material. Part of these may optionally be used later as donor substrates.

 The thickness of the new substrates obtained depends essentially on the thickness of the layer of the second material when this material is used as a filler material for growth. For example, it is possible to use, in the case of silicon, layers with a thickness of the order of 100 to 1000 m.

 The contacting of the block of second material with the initial layer of first material may comprise direct bonding, by molecular adhesion, or indirect bonding by means of a film of

<Desc / Clms Page number 7>

 bonding. The bonding film is, for example, an SiO 2 -Si 3 N 4 oxide or silicon nitride. It may incidentally have the function of improving the quality of the bonding between the assembled layers.

 When the bonding is a direct bonding by molecular adhesion, that is to say without the addition of material, the molecular adhesion can be carried out at ambient temperature after an adequate cleaning of the surfaces to be brought into contact. The binding forces are Van der Walls with interatomic distances of a few nanometers. These bonds are much weaker than those which would be obtained by heteroepitaxy. By heteroepitaxy, we obtain covalent bonds with atomic distances of the order of a few tenths of a nanometer.

 According to another possibility of implementing the invention, it is also possible to relate the second material layer, not by bonding a block, but by depositing said material on the initial layer of first material. The deposition of material differs here from an epitaxy. While an epitaxy agrees with the crystal coherence of the medium on which it is carried out, the deposit remains incoherent. The deposition can take place according to any technique such as sputtering, chemical vapor deposition (CVD), electroplating, etc.

 According to a particular and advantageous implementation of the method, the first and second materials may be chosen so that the first material is a binary compound of the second material. AT

<Desc / Clms Page number 8>

 As an illustration, the first material may be selected from SiC, InP, GaP, GaAs, CdTe, AlN, GaN, and the second material, respectively from Si In, Cd, Te, Al, Ga and their compounds.

 The second material may advantageously be composed of elements of which at least one does not participate directly in the development of the layer. For example, it may be preferable to deposit AsGa as a second material on a GaN or GaP layer rather than pure gallium. During the heating, arsenic evaporates to leave only the molten gallium in which the nitrogen or phosphorus diffuses.

 The first material remains solid during the growth stage. In other words, the second material layer is selectively melted.

This can be done by appropriate heating or, more simply, by using a first material whose melting temperature is higher than that of the second material. This is often the case for the materials targeted by the invention, and for which the conventional drawing methods are difficult to implement.

 The melting of the second material can take place in a crucible, or without crucible. When the wetting of the second material on the first material is well adapted, the layer can in fact be maintained on the layer of the first material.

 The use of a crucible, however, allows to give the final substrate a desired shape. For example, a crucible with flared walls makes it possible to increase the surface of the substrate.

<Desc / Clms Page number 9>

 It should be noted that the method can be implemented with additions of the second material to compensate for its consumption.

 It may also be advantageous to deposit on the free surface of the substrate a layer of additional material chosen for a particular function such as temperature control or surface quality control. For example, a Ge film on Si can play the role of surfactant.

 Other features and advantages of the invention will emerge from the description which follows, with reference to the figures of the accompanying drawings.

This description is given for purely illustrative and non-limiting purposes.

Brief Description of the Figures - Figure 1 is a schematic sectional view of an initial substrate formed of a layer assembly.

 - Figures 2 and 3 are schematic sections of crucibles containing substrates comparable to the substrate of Figure 1.

 FIG. 4 is a diagrammatic section of the crucible of FIG. 2, containing a substrate, and illustrates a step of material growth.

 - Figure 5 is a schematic section of a substrate as obtained at the end of the growth process.

 - Figure 6 is a schematic section of another crucible containing a substrate and illustrates a particular implementation of the method.

<Desc / Clms Page number 10>

 FIG. 7 is a schematic cross-section of yet another crucible containing a substrate and illustrates yet another particular embodiment of the process.

Detailed description of modes of implementation of the invention.

 In the following description, identical, similar or equivalent parts of the different figures are identified by the same reference signs to facilitate the transfer between the figures. Moreover, and for the sake of clarity of the figures, not all elements are represented in a uniform scale.

 FIG. 1 shows a substrate comprising a thick silicon layer on which a thin layer of SiC 20 has been reported. The Si layer 10 has, for example, a thickness of the order of 300 μm, and the SiC layer 20, a thickness of the order of 3 .mu.m.

 The thin layer 20 is transferred to the thick layer 10 by direct bonding. A dashed line 12 represents any intervening layers such as oxide or nitride layers. These intermediate layers are not shown in the following figures for the sake of clarity.

 FIG. 2 shows the substrate of FIG. 1 which has been put into place in a crucible 30 of suitable dimensions. The crucible is, for example, a graphite crucible. The thin layer 20 of SiC is oriented towards the bottom of the crucible 30 and is maintained

<Desc / Clms Page number 11>

 against the bottom. The maintenance can result from a collage.

Direct bonding can be envisaged. Bonding with graphite lacquer, or any other carbon-containing compound, can also be retained.

 The crucible, provided with the substrate, is placed in a reactor which makes it possible to reach temperatures suitable for melting the material of the thick layer 10 under a controlled atmosphere. For example, for silicon, the melting temperatures are of the order of 1450-15000C. To simplify the drawing, the reactor is not shown.

 FIG. 3 illustrates a variant in which the crucible has dimensions greater than the substrate. Maintaining the substrate in the crucible 30 is provided by a guard ring 32, also in graphite, which bears on a flange that the thin layer 20 of the substrate forms with respect to the thick layer 10.

 In a particular case, the crucible can be removed or replaced by a support without side walls.

 An important step in the process is illustrated in FIG. 4 which shows the growth of the first material, in this case SiC, on the thin layer 20. The crucible is heated to a temperature sufficient to melt the thick silicon layer but low enough not to melt the SiC. The thick layer 10 is subjected in the reactor to a reactive gas containing the species necessary for the growth of the first material. The reactive gas is symbolized by arrows in Figure 4. Here it

<Desc / Clms Page number 12>

 it is a gas such as propane or silane which contains carbon or silicon which can be mixed with a carrier gas such as hydrogen. The supply of silicon is, in this case, provided essentially by the bath of the thick layer melted. The pressure of the reactant gas is preferably maintained at a value below a limit corresponding to SiC formation on the surface of the silicon bath. It is indeed necessary to avoid a passivation of the surface. This could block the diffusion of the carbon element to the thin layer 20. The partial pressure of propane is for example 1000 Pa.

 The growth of the first material takes place from the contact interface between the thin layer and the thick layer, which, as previously indicated, behaves like a seed. Crystalline growth of SiC from a Si liquid phase is faster than growth in the gas phase. It can reach 100 GB per hour.

 In FIG. 4, the reference 22 indicates solid SiC which has grown on the thin layer 20.

 During growth, the silicon of the thick layer 10 is consumed until exhaustion. We then obtain a final substrate as shown in Figure 5, entirely of SiC. The layer 22 of SiC obtained by growth has a crystalline quality comparable to the initial layer 20.

 A variant of the process is illustrated in FIG. 6. This shows a crucible 30 with a carbon insert 34 making it possible to confer on the crucible

<Desc / Clms Page number 13>

 side walls flared. The flared walls lead to an increase in the diameter of the substrate as the first material grows. It may be advantageous to open the crucible to an angle of 1800 so as to obtain a consequent gain in the size of the crystal. This variant is shown in Figure 7. It allows to use initial slices of the first material with a smaller diameter than the first material finally obtained.

 Figure 6 also shows symbolically a contribution of the second material, in this case silicon, to compensate for the consumption of the material during growth. A silicon block 14 is added to the surface of the thick silicon layer 10 initially present. The addition can take place before or after the complete transformation of the silicon layer into SiC.

 If the silicon initially present is exhausted, a new molecular bonding can take place to fix the silicon block 14 on the substrate obtained, entirely of SiC. The silicon block 14 can also be placed on the remaining silicon of the thick layer 10, in the molten state or not.

CITES DOCUMENTS (1)
WO-00/31322 (2) Silicon carbide on insulator formation by the
Smart-Cut process by L. Di Cioccio et al.,

<Desc / Clms Page number 14>

Materials Science and Engineering B46 (1997) pp.

 349-356.

(3) wafer semiconductor Bonding: Science and
Technology of QY TONG and U. GÖSELE, A WILEY
INTERSCIENCE PUBLICATION, JOHN WILEY & SONS, INC.

Claims (12)

A crystalline growth process in which: a) a substrate is formed comprising an initial layer (20) of a first material having a crystalline coherence, and a layer (10) of a second material, the second material allowing, at the liquid state, the diffusion of atoms entering the composition of the first material, b) the second material layer is melted and a layer of the first material is grown on the initial layer of first material (20), subjecting the layer of second material, melted, with a reactive gas comprising at least one component of the first material, characterized in that, during step a), the layer of second material is reported, without respecting the crystalline coherence, the initial layer of first material.  2. The method of claim 1, wherein the second material layer is brought by contacting a block of the second material with the layer of first material.  3. The method of claim 2, wherein the contacting comprises a direct bonding by molecular adhesion. <Desc / Clms Page number 16>  The method of claim 2, wherein the contacting comprises bonding via a bonding film (12).  5. The method of claim 1, wherein the second material layer is deposited by depositing said material on the initial layer of first material.  The method of claim 1, wherein the first material is a compound of the second material.  7. The method of claim 6, comprising an addition of the second material offsetting a consumption of this material during the growth of the first material.  The method of claim 1, wherein the first material has a higher melting temperature than the second material.  The method of claim 1, wherein the first material is selected from SiC, CdTe, AlN, InP, GaP, GaAs, GaN, and wherein the second material is selected from Si, Cd, Te, Al, Ga and their compounds.  The method of claim 1, wherein the second material comprises at least one component different from the components of the first material. <Desc / Clms Page number 17>  11. The method of claim 1, wherein the second material layer is melted in a crucible (30) with flared walls.  12. The method of claim 1, wherein is further deposited at least one additional layer of a third material to the free surface of the first material.