Optimized Heteroepitaxial Growth of Semiconductors
20230148397 · 2023-05-11
Inventors
Cpc classification
C30B29/48
CHEMISTRY; METALLURGY
H01L31/1828
ELECTRICITY
G02F1/3558
PHYSICS
C30B25/183
CHEMISTRY; METALLURGY
H01L21/02293
ELECTRICITY
H01L21/0262
ELECTRICITY
C30B29/46
CHEMISTRY; METALLURGY
International classification
Abstract
A method of performing HVPE heteroepitaxy comprises exposing a substrate to a carrier gas, a first precursor gas, a Group II/III element, and ternary-forming gasses (V/VI group precursor), to form a heteroepitaxial growth of a binary, ternary, and/or quaternary compound on the substrate; wherein the carrier gas is H.sub.2, wherein the first precursor gas is HCl, the Group II/III element comprises at least one of Zn, Cd, Hg, Al, Ga, and In; and wherein the ternary-forming gasses comprise at least two or more of AsH.sub.3 (arsine), PH.sub.3 (phosphine), H.sub.2Se (hydrogen selenide), H.sub.2Te (hydrogen telluride), SbH.sub.3 (hydrogen antimonide, or antimony tri-hydride, or stibine), H.sub.2S (hydrogen sulfide), NH.sub.3 (ammonia), and HF (hydrogen fluoride); flowing the carrier gas over the Group II/III element; exposing the substrate to the ternary-forming gasses in a predetermined ratio of first ternary-forming gas to second ternary-forming gas (1tf:2tf ratio); and changing the 1tf:2tf ratio over time.
Claims
1. A method of performing heteroepitaxy, comprising: exposing a substrate to a carrier gas, a first precursor gas, a Group II/III element, and ternary-forming gasses (V/VI group precursor), to form a heteroepitaxial growth of one of GaAs, AlAs, InAs, GaP, InP, ZnSe, GaSe (or Ga.sub.2Se.sub.3), CdSe, InSe, ZnTe, CdTe, GaTe (or Ga.sub.2Te.sub.3), HgTe, GaSb, GaSe, InSb, AlSb, AlAs, InAs, CdS, ZnS, GaN, AlN and some of their ternaries such as GaAsP, HgCdTe, etc. on the substrate; wherein the substrate comprises one of GaAs, AlAs, InAs, GaP, InP, ZnSe, GaSe (or Ga.sub.2Se.sub.3), CdSe, InSe, ZnTe, CdTe, GaTe (or Ga.sub.2Te.sub.3), HgTe, GaSb, InSb, AlSb, CdS, ZnS, GaN, AlN, Si, CaF.sub.2, BaF.sub.2, and LiNbO.sub.3 (including PPLN); wherein the carrier gas is H.sub.2, wherein the first precursor gas is HCl, the Group II/III element comprises at least one of Zn, Cd, Hg, Al, Ga, and In; and wherein the ternary-forming gasses comprise at least two or more of AsH.sub.3 (arsine), PH.sub.3 (phosphine), H.sub.2Se (hydrogen selenide), H.sub.2Te (hydrogen telluride), SbH.sub.3 (hydrogen antimonide, or antimony tri-hydride, or stibine), H.sub.2S (hydrogen sulfide), NH.sub.3 (ammonia), and HF (hydrogen fluoride); flowing the carrier gas over the Group II/III element; exposing the substrate to the ternary-forming gasses in a predetermined ratio of first ternary-forming gas to second ternary-forming gas (1tf:2tf ratio); and changing the 1tf:2tf ratio over time.
2. The method of claim 1, further comprising: flowing the ternary-forming gasses through the furnace at a 1tf:2tf ratio of about 1:0; heating the substrate to about 500° C.-900° C.; and gradually changing the 1tf:2tf ratio toward 0:1 over a time period of 1 min-10 hours.
3. The method of performing heteroepitaxy of claim 1, wherein the Group II/III element is Ga, the substrate is GaAs (gallium arsenide), the first ternary-forming gas is AsH.sub.3, the second ternary-forming gas is PH.sub.3 (phosphine), and the heteroepitaxial growth is GaP (gallium phosphide).
4. The method of performing heteroepitaxy of claim 1, wherein the Group II/III element is Ga, wherein the substrate is GaP (gallium phosphide), the first ternary-forming gas is PH.sub.3, the second ternary-forming gas is AsH.sub.3 (arsine), and the heteroepitaxial growth is GaAs (gallium arsenide).
5. The method of performing heteroepitaxy of claim 1, wherein the Group II/III element is Ga, wherein the substrate is GaAs (gallium arsenide), the first ternary-forming gas is AsH.sub.3 (arsine), the second ternary-forming gas is PH.sub.3 (phosphine), and the heteroepitaxial growth is GaAsP (gallium arsenide phosphide).
6. The method of performing heteroepitaxy of claim 1, wherein the Group II/III element is Ga, wherein the substrate is GaP (gallium phosphide), the first ternary-forming gas is PH.sub.3 (phosphine), the second ternary-forming gas is AsH.sub.3 (arsine), and the heteroepitaxial growth is GaAsP (gallium arsenide phosphide).
7. The method of performing heteroepitaxy of claim 1, wherein the Group II/III element is Zn, wherein the substrate is GaAs (gallium arsenide), the second ternary-forming gas is AsH.sub.3, the second ternary-forming gas is H.sub.2Se (hydrogen selenide), and the heteroepitaxial growth is ZnSe (zinc selenide).
8. The method of performing heteroepitaxy of claim 1, wherein the Group II/III element is Ga, wherein the substrate is ZnSe (zinc selenide), the first ternary-forming gas is H.sub.2Se, the second ternary-forming gas comprises AsH.sub.3 (arsine), and the heteroepitaxial growth is GaAs (gallium arsenide).
9. The method of performing heteroepitaxy of claim 1, wherein the Group II/III element is Zn, wherein the substrate is GaSb (gallium antimonide), the first ternary-forming gas is SbH.sub.3, the second ternary-forming gas is H.sub.2Te (hydrogen telluride), and the heteroepitaxial growth is ZnTe (zinc telluride).
10. The method of performing heteroepitaxy of claim 1, wherein the Group II/III element is Ga, wherein the substrate is ZnTe (zinc telluride), the first ternary-forming gas is H.sub.2Te, the second ternary-forming gas is SbH.sub.3 (hydrogen antimonide/stibine/antimony tri-hydride), and the heteroepitaxial growth is GaSb (gallium antimonide).
11. The method of performing heteroepitaxy of claim 1, wherein the Group II/III element is Zn, wherein the substrate is InAs (indium arsenide), the first ternary-forming gas is AsH.sub.3, the second ternary-forming gas is H.sub.2Te (hydrogen telluride), and the heteroepitaxial growth is ZnTe (zinc telluride).
12. The method of performing heteroepitaxy of claim 1, wherein the Group II/III element is In, wherein the substrate is ZnTe (zinc telluride), the first ternary-forming gas is H.sub.2Te, the second ternary-forming gas is AsH.sub.3 (arsine), and the heteroepitaxial growth is InAs (indium arsenide).
13. The method of performing heteroepitaxy of claim 1, wherein the Group II/III element is Ga, wherein the substrate is GaN (gallium nitride), the first ternary-forming gas is NH.sub.3, the second ternary-forming gas is H.sub.2Se (hydrogen selenide), and the heteroepitaxial growth is hexagonal ε-GaSe (gallium selenide).
14. The method of performing heteroepitaxy of claim 1, wherein the Group II/III element is Ga, wherein the substrate is hexagonal ε-GaSe (gallium selenide), the first ternary-forming gas is H.sub.2Se, the second ternary-forming gas is NH.sub.3 (ammonia), and the heteroepitaxial growth is GaN (gallium nitride).
15. The method of performing heteroepitaxy of claim 1, wherein the Group II/III element is Ga, wherein the substrate is GaP (gallium phosphide), the first ternary-forming gas is PH.sub.3, the second ternary-forming gas is H.sub.2Se (hydrogen selenide), and the heteroepitaxial growth is cubic phase zinc blende gallium selenide Ga.sub.2Se.sub.3.
16. The method of performing heteroepitaxy of claim 1, wherein the Group II/III element is Ga, wherein the substrate is cubic phase zinc blende gallium selenide Ga.sub.2Se.sub.3, the first ternary-forming gas is H2Se, the second ternary-forming gas is PH.sub.3 (phosphine), and the heteroepitaxial growth is GaP (gallium phosphide).
17. The method of performing heteroepitaxy of claim 1, wherein the Group II/III element is Ga, wherein the substrate is GaAs (gallium arsenide), the first ternary-forming gas is AsH.sub.3, the second ternary-forming gas is H.sub.2Se (hydrogen selenide), and the heteroepitaxial growth is cubic phase zinc blende gallium selenide Ga.sub.2Se.sub.3.
18. The method of performing heteroepitaxy of claim 1, wherein the Group II/III element is Ga, wherein the substrate is cubic phase zinc blende gallium selenide Ga.sub.2Se.sub.3, the first ternary-forming gas is H.sub.2Se, the second ternary-forming gas is AsH.sub.3 (arsine), and the heteroepitaxial growth is GaAs (gallium arsenide).
19. The method of performing heteroepitaxy of claim 1, wherein the Group II/III element is Ga, the substrate is AlN (aluminum nitride), the first ternary-forming gas is NH.sub.3, the second ternary-forming gas is H.sub.2Se (hydrogen selenide), and the heteroepitaxial growth is hexagonal ε-GaSe (gallium selenide).
20. The method of performing heteroepitaxy of claim 1, wherein the Group II/III element is Al, the substrate is hexagonal ε-GaSe (gallium selenide), the first ternary-forming gas is H.sub.2Se, the second ternary-forming gas is NH.sub.3 (ammonia), and the heteroepitaxial growth is AlN (aluminum nitride).
21. The method of performing heteroepitaxy of claim 1, wherein the Group II/III element is Al, the substrate is GaAs (gallium arsenide), the first ternary-forming gas is AsH.sub.3, the second ternary-forming gas is AsH.sub.3 (arsine), and the heteroepitaxial growth is AlAs (aluminum arsenide).
22. The method of performing heteroepitaxy of claim 1, wherein the Group II/III element is Zn, the substrate is GaP (gallium phosphide), the first ternary-forming gas is PH.sub.3, the second ternary-forming gas is H.sub.2S (hydrogen sulfide), and the heteroepitaxial growth is ZnS (zinc sulfide).
23. The method of performing heteroepitaxy of claim 1, wherein the Group II/III element is Ga, the substrate is ZnS (zinc sulfide), the first ternary-forming gas is H2S, the second ternary-forming gas is PH.sub.3 (phosphine), and the heteroepitaxial growth is GaP (gallium phosphide).
24. The method of performing heteroepitaxy of claim 1, wherein the Group II/III element is Cd, the substrate is InP (indium phosphide), the first ternary-forming gas is PH.sub.3, the second ternary-forming gas is H.sub.2S (hydrogen sulfide), and the heteroepitaxial growth is CdS (cadmium sulfide).
25. The method of performing heteroepitaxy of claim 1, wherein the Group II/III element is In, the substrate is CdS (cadmium sulfide), the first ternary-forming gas is H.sub.2S, the second ternary-forming gas is PH.sub.3 (phosphine), and the heteroepitaxial growth is InP (indium phosphide).
26. The method of performing heteroepitaxy of claim 1, wherein the Group II/III element is Ga, the substrate is InP (indium phosphide), the first ternary-forming gas is PH.sub.3, the second ternary-forming gas is H.sub.2Te (hydrogen telluride), and the heteroepitaxial growth is a cubic phase zinc blende gallium telluride Ga.sub.2Te.sub.3.
27. The method of performing heteroepitaxy of claim 1, wherein the Group II/III element is In, the substrate is GaSb (gallium antimonide), the first ternary-forming gas is SbH3, the second ternary-forming gas is AsH.sub.3 (arsine), and the heteroepitaxial growth is InAs (indium arsenide).
28. The method of performing heteroepitaxy of claim 1, wherein the Group II/III element is Ga, the substrate is InAs (indium arsenide), the first ternary-forming gas is AsH.sub.3, the second ternary-forming gas is SbH.sub.3 (hydrogen antimonide), and the heteroepitaxial growth is GaSb (gallium antimonide).
29. The method of performing heteroepitaxy of claim 1, wherein the Group II/III element is Cd, the substrate is InAs (indium arsenide), the first ternary-forming gas is AsH.sub.3, the second ternary-forming gas is H.sub.2Se (hydrogen selenide), and the heteroepitaxial growth is CdSe (cadmium selenide).
30. The method of performing heteroepitaxy of claim 1, wherein the Group II/III element is Al, the substrate is InAs (indium arsenide), the first and the second ternary-forming gas is AsH.sub.3 (arsine), and the heteroepitaxial growth is AlAs (aluminum arsenide).
31. The method of performing heteroepitaxy of claim 1, wherein the Group II/III element is Al, the substrate is InAs (indium arsenide), the first ternary-forming gas is AsH.sub.3, the second ternary-forming gas is SbH.sub.3 (hydrogen antimonide), and the heteroepitaxial growth is AlSb (aluminum antimonide).
32. The method of performing heteroepitaxy of claim 1, wherein the Group II/III element is Ga, the substrate is AlSb (aluminum antimonide), the first ternary-forming gas is SbH3, the second ternary-forming gas is SbH.sub.3 (hydrogen antimonide), and the heteroepitaxial growth is GaSb (gallium antimonide).
33. The method of performing heteroepitaxy of claim 1, wherein the Group II/III element is Al, the substrate is GaSb (gallium antimonide), the first ternary-forming gas is SbH3, the second ternary-forming gas is SbH.sub.3 (hydrogen antimonide), and the heteroepitaxial growth is AlSb (aluminum antimonide).
34. The method of performing heteroepitaxy of claim 1, wherein the Group II/III element is Al, the substrate is GaSb (gallium antimonide), the first ternary-forming gas is SbH3, the second ternary-forming gas is AsH.sub.3 (arsine), and the heteroepitaxial growth is AlAs (aluminum arsenide).
35. The method of performing heteroepitaxy of claim 1, wherein the Group II/III element is Ga, the substrate is AlAs (aluminum arsenide), the first ternary-forming gas is AsH3, the second ternary-forming gas is SbH.sub.3 (hydrogen antimonide), and the heteroepitaxial growth is GaSb (gallium antimonide).
36. The method of performing heteroepitaxy of claim 1, wherein the Group II/III element is Cd, the substrate is InSb (indium antimonide), the first ternary-forming gas is SbH3, the second ternary-forming gas is H.sub.2Te (hydrogen telluride), and the heteroepitaxial growth is CdTe (cadmium telluride).
37. The method of performing heteroepitaxy of claim 1, wherein the Group II/III element is Hg, the substrate is InSb (indium antimonide), the first ternary-forming gas isSbH3, the second ternary-forming gas is H.sub.2Te (hydrogen telluride), and the heteroepitaxial growth is HgTe (mercury telluride).
38. The method of performing heteroepitaxy of claim 1, wherein the Group II/III element is Hg, the substrate is CdTe (cadmium telluride), the first ternary-forming gas is H.sub.2Te, the second ternary-forming gas is H.sub.2Te, and the heteroepitaxial growth is HgTe (mercury telluride).
39. The method of performing heteroepitaxy of claim 1, wherein the Group II/III element is Hg, the substrate is one of InSb (indium antimonide), CdTe (cadmium telluride), and HgTe (mercury telluride), the first ternary-forming gas is H.sub.2Te, the second ternary-forming gas is one of H.sub.2Te (hydrogen telluride) and SbH.sub.3, and the heteroepitaxial growth is HgCdTe (mercury cadmium telluride).
40. The method of performing heteroepitaxy of claim 1, wherein the Group II/III element is Ga, the substrate is CaF.sub.2 (calcium fluoride), the first ternary-forming gas is HF, the second ternary-forming gas is PH.sub.3 (phosphine), and the heteroepitaxial growth is GaP (gallium phosphide).
41. The method of performing heteroepitaxy of claim 1, wherein the Group II/III element is Ga, the substrate is CaF.sub.2 (calcium fluoride), the first ternary-forming gas is HF, the second ternary-forming gas is AsH.sub.3 (arsenide), and the heteroepitaxial growth is GaAs (gallium arsenide).
42. The method of performing heteroepitaxy of claim 1, wherein the Group II/III element is Al, the substrate is BaF.sub.2 (barium fluoride), the first ternary-forming gas is HF, the second ternary-forming gas is SbH.sub.3 (hydrogen antimonide), and the heteroepitaxial growth is AlSb (aluminum antimonide).
43. The method of performing heteroepitaxy of claim 1, wherein the Group II/III element is Ga, the substrate is BaF.sub.2 (barium fluoride), the first ternary-forming gas is HF, the second ternary-forming gas is SbH.sub.3 (hydrogen antimonide), and the heteroepitaxial growth is GaSb (gallium antimonide.
44. The method of performing heteroepitaxy of claim 1, wherein the Group II/III element is Ga, the substrate is one of LiNbO.sub.3 (Lithium niobate) or periodically poled lithium niobate (PPLN), the first ternary-forming gas is PH.sub.3, the second ternary-forming gas is PH.sub.3 (phosphine), and the heteroepitaxial growth is one of GaP (gallium phosphide) or OP—GaP.
45. The method of performing heteroepitaxy of claim 1, wherein the Group II/III element is Ga, the substrate is one of LiNbO.sub.3 (lithium niobate) or periodically poled lithium niobate (PPLN), the first ternary-forming gas is AsH.sub.3, the second ternary-forming gas is AsH.sub.3 (arsine), and the heteroepitaxial growth is one of GaAs (gallium arsenide) or OP—GaAs.
Description
BRIEF DESCRIPTION OF THE DRAWINGS
[0121] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments of the present invention and, together with a general description of the invention given above, and the detailed description of the embodiments given below, serve to explain the principles of the present invention. The patent or application file contains at least one drawing executed in color. Copies of this patent or patent application publication with color drawing(s) will be provided by the Office upon request and payment of the necessary fee.
[0122]
[0123]
[0124]
[0125]
[0126]
[0127]
[0128]
[0129]
[0130]
[0131]
[0132]
[0133]
[0134]
[0135]
[0136]
[0137]
[0138]
[0139]
[0140]
[0141]
[0142]
[0143]
[0144]
[0145]
[0146]
[0147]
[0148]
[0149]
[0150]
[0151]
[0152]
[0153]
[0154]
[0155]
[0156]
[0157]
[0158]
[0159]
[0160]
[0161]
[0162]
[0163]
[0164]
[0165]
[0166]
[0167]
[0168]
[0169] It should be understood that the appended drawings are not necessarily to scale, presenting a somewhat simplified representation of various features illustrative of the basic principles of the invention. The specific design features of the sequence of operations as disclosed herein, including, for example, specific dimensions, orientations, locations, and shapes of various illustrated components, will be determined in part by the particular intended application and use environment. Certain features of the illustrated embodiments have been enlarged or distorted relative to others to facilitate visualization and clear understanding. In particular, thin features may be thickened, for example, for clarity or illustration.
DETAILED DESCRIPTION OF THE INVENTION
[0170] A process for thick heteroepitaxial growth of semiconductor materials is presented below. The semiconductor structures of the growing layer(s) may be deposited in a horizontal or vertical direction, on plain substrates or on patterned templates, including orientation-patterned templates.
[0171] A further embodiment of the invention states that the heteroepitaxial growth is preceded by an in-situ pre-growth treatment of the substrate or the template followed by at least 300-500 .Math.m thick heteroepitaxial growth of one or more doped or undoped semiconductor materials or their binaries, ternaries, or quaternaries. (Note: An exception of the thickness range is when the purpose of the growth was growth of low-dimensional (LD) materials. In this case, thicknesses in the range of several nm to several .Math.m are acceptable.)
[0172] In order to be successful, each separate heteroepitaxial combination must meet particular requirements, including the lattice mismatch and the accumulated strain. Accordingly, the heteroepitaxy exhibited in any particular example below will be based on the degree and the sign of the lattice mismatch between substrate and growing layer, as well as on how the strain accumulated in the growing layer because of the lattice and thermal mismatch is released - in plastic or elastic strain release processes. While the elastic strain release process occurs through surface roughening, a typical example of a plastic strain release process is the periodic formation of the so-called “misfit dislocations” (MDs). Such dislocations may appear after a certain critical thickness h.sub.c of the so-called “pseudomorphous growth”, during which the layer is “forced” to grow with the lattice constant of the substrate. (Note: The words “mismatch” and “misfit” are almost identical, but their meanings are distinct in this context. However, in particular cases, one will be preferred to the other. For example, it is proper to say “lattice mismatch” and “misfit dislocations”.) In this invention, we teach that the sign and the degree of the lattice mismatch and the periodicity of the MDs may be used as criteria for one to determine in advance how successful a new case of heteroepitaxy would be.
[0173] The lattice mismatch f.sub.% is calculated using the formula:
where a.sub.0 and b.sub.0 are the lattice constants of the substrate and the layer material. The lattice constants of some traditional semiconductor materials are presented in
[0174] The periodicity τ of the misfit dislocations is determined by:
[0175] As an example, we can determine the lattice mismatch f.sub.% and the periodicity of the MDs τ in the particular heteroepitaxial case of growth of GaP on a GaAs substrate. The lattice constant (a.sub.o) of GaAs = 5.6532 Å, while the lattice constant (b.sub.o) of GaP = 5.4512 Å. According to equations (1) and (2), in this example the lattice mismatch f.sub.% is negative (-3.57%) because b.sub.o GaP < a.sub.o GaAs; we should expect the appearance of MDs at a periodicity τ of about 28 (i.e. 100/3.57) interatomic distances. Such lattice mismatch (-3.57%) may be considered as large. In general, lattice mismatches of 3-4% and more are considered as relatively large, while lattice mismatches under 1% may be considered as relatively small. Thus the lattice mismatch between GaP and GaAs (-3.57%) may be considered as large, while the lattice mismatch between ZnSe and GaAs (+0.26), for example, may be considered as small. However, whether a particular mismatch can be considered as large or small, it depends on other factors as well, for example, on the strength of the bonds (the bond dissociation energies) between the atoms of the substrate and those of the growing layer. For example, (Table 1) the bond energies of the bonds Ga—As and Ga—P are in the same order of magnitude. This means that the As and P atoms can easily replace each other, forming an intermediate GaAsP ternary transition layer between the substrate and the growing layer. Namely, because of this GaP and GaAs can grow successfully on each other even at the larger mismatch of 3.57%. As we discovered in the course of our study the formation of this intermediate ternary layer could be initiated still during the preheating stage of growth by treating the substrate with its non-native precursor, i.e. by exposing GaAs to phosphine (PH.sub.3) or GaP to arsine (AsH.sub.3). As Table 1 also shows, from this point of view GaSb, InSb and InP are also “easy” substrates due to the low bonding energies of their atoms. However, as one can see from Table 1, due to the high bonding energy of the Ge—Ge bond and, especially, of the Si—Si bond, these common substrates are hardly treatable. Thus, in the case of growth of Ge/Se, for example, namely because the atomic bonds are strong in both materials, the lattice mismatch of +3.96% between Ge and Si is considered as huge, no matter that as a number this mismatch is not much different from the lattice mismatch between GaAs and GaP:
TABLE-US-00001 Some bond dissociation energies (standard state enthalpy changes) at T = 298 K. Bond Bond dissociation energy ΔHf.sub.298 [kJ/mol] Ga—As 209 Ga—P 230 Ga—Sb 209 In—Sb 152 In—P 198 Ge—Ge 274 Si—Si 327
[0176] The sign of the lattice mismatch, minus (-) or plus (+), is also important. For example, it was determined that the thickness of the pseudomorphous growth, i.e. the critical thickness h.sub.c, is larger when the lattice mismatch is negative and the film is growing under tensile strain than in the case of a positive lattice mismatch when the layer is growing compressively strained. To clarify again that, according to eq. 1, we have a negative mismatch and a layer growing under tensile strain when the lattice constant of the layer material is smaller than the lattice constant of the substrate. In contrast, a positive mismatch means a larger lattice constant of the layer material - a case when the layer is growing under compressive strain. The significant difference in the mechanisms of dislocation nucleation (after the moment in which the pseudomorphous growth becomes energetically unfavorable and the accumulated elastic strain must be relieved somehow) in the case of tension vs. compression contributes to this difference as well. For example, while in the compression case the dislocations nucleate by squeezing out an atom at the base of surface depressions, in the tension case, the nucleation of misfit dislocations involves the concerted motion of a relatively large number of atoms, leading to insertion of an extra lattice (plane) row into an already continuous film. In addition to all that, the film morphology depends intimately on the sign of the misfit (+ or -), i.e., on the type of the strain (tensile or compressive). It is experimentally confirmed, for example, that growth under tensile strain (negative misfit) favors 2D growth, which usually results in smooth surface morphology, while compressive growth facilitates 3D growth, which typically results in hillock type, i.e., rougher surface morphology. In other words, plastic relaxation (negative misfit, tensile strain) is encouraged when the goal is to grow metamorphic buffers, while elastic relaxation (positive misfit, compressive strain), being associated with surface roughening, shall be used to drive quantum dot self-assembly processes. All this is, again, in support our notion that the growth under tensile strain, as it is in the case of growth of GaP on GaAs (negative misfit), should be more favorable for our coals than the opposite case of growth of GaAs on GaP (positive misfit).
[0177] The linearly increasing elastic strain accumulated during the pseudomorphous growth must be released at a certain point. The formation of MDs (misfit dislocations) is one of the possible mechanisms of strain relief. However, deeper crystallographic considerations are necessary to determine where they should be expected, or on which crystallographic plane it is most probable for the MDs to appear. In a zinc blende structure (this is the structure of many of the compound semiconductor materials presented herein, e.g. GaAs, GaP, ZnSe, etc.), for example, the biaxial strain σ accumulated during pseudomorphous growth may be resolved (see
[0178] Roughening of the surface may be the result of exposing the substrate to a non-native precursor during the preheating stage. This is supported by
[0179]
[0180] It is thought that the stronger influence of AsH.sub.3 on GaP than of PH.sub.3 on GaAs (compare the ranges of the distances between peaks and valleys on the scales that are left from
[0181] It was discovered that surface roughening (looking like pitting) is not only the result of exposing the semiconductor material to a non-native precursor during the preheating stage. Elemental analysis performed by Electron Dispersion Spectroscopy (EDS) of surfaces exposed to non-native precursors (see
[0182] The effect that the non-native precursors may have on the GaP and GaAs substrates (see
[0183] The proposed approach is to some extent universal because it may be applied to many different materials deposited one over another in a one-step epitaxial process, with or without the intentional assistance of an intermediate sub-lattice transition buffer layer (see
[0184] Because it is capable of controlling the thickness of the buffer layer, this invention allows one to extend the idea of the deposition of a ternary transition buffer layer to the growth of ternary layers hundreds of microns thick. This may be achieved by maintaining the ratio of the mixture of the native and the non-native precursors constant, which will ensure achieving the desired composition (x) in the growing ternary and keeping it constant during the entire growth process. As an example, the disclosed process has been used to grow up to 300 .Math.m thick GaAs.sub.xP.sub.1-x ternary layers on both GaAs and GaP substrates at the relatively high growth rate of about 100 .Math.m/h. The ability to grow thick ternaries by the proposed process is important because the tailoring of different compositions allows one to achieve the best combinations of properties, which may be suitable for a particular application. For example, it was discovered that in the particular case of GaAs.sub.xP.sub.1-x the following composition GaAs.sub.0.34P.sub.0.66 (x = 0.34) provides lower two-photon absorption (2PA) than the 2PA of GaAs but higher nonlinear susceptibility than GaP in the wavelength of interest (1-1.7 .Math.m); these parameters are of great importance for applications such as QPM frequency conversion. Another example is that by changing the x-composition of the GaAs.sub.xP.sub.1-x ternary one may “tune” its bandgap energy to an optimal value that fits to the bandgap of Si in a dual-junction solar cell panel made of the ternary with Si. This idea may be applied to other ternaries in heterostructures with Si or with other common substrates.
[0185] Thus, such a combination of material properties satisfies the requirements for using this new ternary material for frequency conversion devices as well as for many other applications. At the same time, ternaries may be grown on either one of the parenting materials (in this case on GaAs or on GaP substrates) depending on how close to the given substrate is the ternary composition. For example, in the case where the content of As is more than the content of P in the ternary, GaAs will match closer as a substrate, while in the opposite case GaP shall be the more suitable substrate.
[0186] Because of the smaller lattice mismatches that the ternary (GaAs.sub.xP.sub.1-x) has with each of the two substrates (GaAs and GaP) compared to the lattice mismatch between the original materials (GaP and GaAs) the growth of ternaries is also an easier, more favorable, heteroepitaxial task.
[0187] Modification of the material properties, as was already mentioned, is another opportunity that the proposed invention provides and the aforementioned examples demonstrate the usefulness of such modifications for the development of new frequency conversion devices, e.g. for laser sources in the case of using GaP, GaAs, and their GaAs.sub.xP.sub.1-x ternary. However, other material combinations may provide thick growths of other ternaries or quaternaries that may result in products that could support other research and development fields, e.g. optoelectronics, sensing (detectors), solar cell industry, etc.
[0188] This invention also allows to use the exposure of the substrate, the OP template, or the already grown OP material to a non-native precursor as an easy way to determine the crystal polarity of the material and, from here, to use this technique for polarity control during both the fabrication of the OP templates and the subsequent thick HVPE growth on them. This is possible because, in contrast to the case where the substrate is exposed to a non-native precursor and the shape of the observed pits are irregular and randomly distributed, in the case of exposing an OP template to a non-native precursor the shape of the pits are rather longitudinal. They are also predominantly oriented in two mutually perpendicular directions on the surface of domains with opposite crystallographic orientations (opposite polarity).
[0189]
of an OP—GaAs template before and after preheating the OP—GaAs template in the presence of PH.sub.3.
and along [011], on the (100) and
oriented domains. This phenomenon, from a practical point of view, is an important feature of this invention, because it gives us a tool to easily determine the presence of the opposite domain orientations on the OP-template. Moreover, this approach is also an easy way to estimate the quality of the OP-template prior to the growth, or the quality of the growth on the template QPM structure, which is a simple in-situ or post-growth nondestructive evaluation step. Until now, it was possible to determine crystallographic orientations only with great uncertainty. The estimation was based on the type of the surface texture or the specific shape of grown surface features that were experimentally associated more or less to one orientation or another. One may also judge the crystallographic orientation by the rate of etching (which is usually different for different crystallographic directions) or by the shape of the etch pits revealed after etching the sample. Etching, however, means damaging the surface, after which the surface is not usable for the subsequent thick HVPE growth. Heteroepitaxial growth, performed on OP—GaAs templates exposed to a non-native precursor (PH.sub.3), resulted in thick (up to 500 .Math.m) high quality OPGaP with rectangular domains having excellent fidelity (see
[0190] As already explained above, the proposed process may be used for the growth of metamorphic buffer layers or for the formation of quantum dots, quantum wells, or other nano- and micro-structures. This depends on which mode of heteroepitaxy is stimulated during growth: Volmer-Weber “island growth”, Frank-van der Merwe “layer-by-layer growth”, or the Stranski-Krastanov “layer-plus-island growth. In other words this depends on the sign and the magnitude of the of the lattice mismatch and, from there, the type of elastic (tensile or compressive) strain, and the mechanism of strain relief - elastic (surface roughening) or plastic (dislocation formation). By skillful use of these growth modes this process may combine similar or different materials, e.g. common electronic materials, e.g. Si or Ge, with some linear and nonlinear optical materials, e.g. GaAs, GaP, GaSe, ZnSe, ZnTe, ZnS, etc., or even with their ternaries or quaternaries. Thus, the described approach contributes to advances in the development of optoelectronic devices as well.
[0191] The disclosed process allows for one-step thick growth without the need for a preliminary deposition (usually using a thin growth technique) of another material with a smaller lattice mismatch with the growing material and/or with the substrate. For example, as it is known in the prior art, thick HVPE GaN layers can be grown on SiC substrates only after the deposition of a thin AlN layer by MBE or MOCVD. Similarly, thick HVPE GaN may be grown on thin GaN or AlN deposited in advance on sapphire, again, by MOCVD or MBE (this was already discussed in more details abover). The MBE and MOCVD on one hand, and the HVPE on the other are growth processes, which are quite different by their nature. Thus, while the first two (MOCVD and MBE) are far-from-equilibrium processes, the third one, HVPE, is a close-to-equilibrium process. This makes the MOCVD and MBE “less sensitive” to lattice mismatches than HVPE. However, although MOCVD and MBE can “handle” heteroepitaxial growths, in general, at larger lattice mismatches, they, in contrast to HVPE (which is a traditional technique for hundreds of microns of thick growth), can be used only for, typically, 1-2 .Math.m thin growths. This means that the old approach of thick heteroepitaxial growth, using as a first step MOCVD or MBE and after that HVPE, is a two-step growth process which needs more high-tech instruments and more equipment, i.e. greater investments. These limitations are not necessary with the disclosed approach, which focuses on the use of only HVPE.
[0192] The disclosed process also eliminates the need to grow (typically by HVPE) an intentionally deposited low temperature (LT) buffer layer on the substrate prior to the growth of the high temperature (HT) layer. This is known for the thick HVPE growth of GaN on a sapphire substrate, for example. The deposition of such intermediate transition layers aims to reduce the strain between the sapphire substrate and the growing GaN layer. The LT buffer layer can do this job, i.e. to accommodate the growing layer to the substrate. At the same time, however, the LT buffer layer is a highly defective area, a layer with extremely low crystalline quality and, thus, it is a source of a great number of different defects such as, for example, treading dislocations. All this means that the LT buffer layer does not provide always an optimal foundation to start the growth of the actual HT GaN layer. The proposed approach allows the formation of an intermediate transition layer naturally, in-situ, during the initial stage of preheating the substrate, and not through a growth process - it occurs prior to the start of growing the actual layer but also may continue during the initial stages of growth as well. According to the present invention, it is not necessary for the buffer layer to be a LT layer. Instead, the choice of the temperature of the buffer layer formation may be controlled, and thus its quality may be controlled.
[0193] We would like once again to emphasize another significant difference between the proposed buffer layer growth and the prior art by using the same example: an HVPE-grown LT GaN buffer layer. In the prior art case one may choose to change the V-III ratio (i.e. the ratio between the content of Ga and N) during the buffer layer deposition in order for this layer to accommodate both the substrate and growing layer. However, although these changes may change some material properties, like the balance in the carrier concentration, they may not provoke significant changes in the lattice mismatch and thus may not contribute much to accommodating the buffer layer (GaN) and the foreign substrate (SiC). Moreover, they not only may not help much in the buffer layer/substrate heteroepitaxy, but may aggravate the subsequent layer/buffer layer homoepitaxy. In summary, since the buffer layer and the subsequent layer are still from the same material, GaN, as a practical matter, we perform heteroepitaxy only in regards to the substrate (SiC or sapphire), while the HT GaN layer still grows homoepitaxially on the LT GaN buffer layer. In contrast, the inventive method, disclosed herein, is a heteroepitaxial growth of a buffer ternary layer (GaAsP in the chosen example) concerning both the substrate (GaAs) and the growing layer (GaP), since the lattice constant of the ternary (GaAsP) is always different from the lattice constants of the substrate and the growing layer. As an option, its composition may be continuously, gradually changed during the preheating stage but this may also continue during the initial stage of growth for a gradual replacement of the substrate material with the growing layer. This means that at the beginning the lattice constant of the buffer material is closer to the lattice constant of the substrate material, while at the end of its deposition the lattice constant of the ternary material is already close to the lattice constant of the layer material that will be grown on the ternary buffer layer (which means changes in the buffer layer composition make sense). However, there is another significant difference. In this case we do not change the V/III ratio of the material of the buffer layer, but only the ratio between the V-group atoms in the ternary, which are As and P in the given example. In addition, in the same example the ternary GaAsP is a combination of the two parenting materials - the substrate (GaAs) and the layer (GaP) material. However, this is not a strict requirement of this disclosure. The intermediate transition ternary layer may be from any suitable materials that accommodate the growing layer to the substrate. For example, if growth of GaSe on Si is attempted and the suggested non-native precursor, hydrogen selenide (H.sub.2Se), does not “pretreat” the Si substrate properly, another non-native precursor (for example PH.sub.3) may do a better job, at the same time forming on the Si surface an intermediate layer (GaP) with which GaSe has a fairly small lattice mismatch. All these options makes the proposed approach much more flexible than the approaches for growth of a transition buffer layer used until now.
[0194] Thus, the optimized growth approach presented herein avoids or solves many of the current problems and shortcomings of heteroepitaxy. The optimized growth approach presented herein demonstrates several advantages over the known and comfortable homo- and heteroepitaxial processes. The disclosed process clearly indicates that there are many particular cases when heteroepitaxy, as stated here, may be preferable even when homoepitaxy is possible.
[0195] Heteroepitaxy provides economic and quality advantages: for example, the GaP wafers (for 2-inch wafers) available on the market is 5-6 times more expensive than the corresponding GaAs. In addition, the commercially available GaP wafers have much lower quality with respect to the etch pit density (EPD) and wafer parallelism. This means that the quality of OP—GaP templates prepared from such wafers will also be low, and that is why we should expect poor quality of the HVPE growth on them. Accordingly, the performance of frequency conversion devices based on such templates will also be unacceptable, because of the expected poor domain fidelity. The ability to use GaAs substrates and OP—GaAs templates for growth of GaP and OP—GaP solves this problem.
[0196] In addition, heteroepitaxy, as stated in this disclosure, enables the use of techniques suitable for thick epitaxial growth, e.g. HVPE, and the corresponding practical applications that require thick epitaxial growths. At this moment, due to the complex growth mechanisms of heteroepitaxy, knowledge of such mechanisms is relatively limited despite the great deal of effort made over the last couple of decades. For example, on an atomic scale it is known how the growth proceeds only for the first few monoatomic layers, even for the homoepitaxial growth of only a few plain semiconductors, e.g. Si and Ge. That is why the semiconductor industry has adopted primarily thin growth techniques such as MOCVD and MBE, and only for a limited number of well-studied materials.
[0197] The disclosed heteroepitaxial approach relies on the most promising candidate for thick epitaxial growth, the HVPE technique. Some other existing techniques for thick growth are more bulk-growth than layer-growth techniques, and each of them suffers from its own disadvantages. For example, the aforementioned PVT process, used successfully in the industry for SiC bulk growth, continues to have material quality issues. This is, namely, the reason for seeking alternative approaches for the growth of SiC such as, for example, the Top Seeded Solution Growth (TSSG). PVT (and in some cases the Bridgman method) is the method of choice for the growth of other materials, already mentioned in the text, as well. Such materials include GaSe, ZnSe, ZnTe, ZnS, etc. However, looking at what is available on the market, one can easily figure out that the larger offered samples from such materials are either polycrystalline, or if they are crystalline, they are not large enough for device development. In addition, typically, they still consist of several domains with different crystallographic orientations, i.e. they are still not exactly “crystalline”. PVT has been used for the heteroepitaxy of OP—ZnSe on OP—GaAs templates, but the grown OP—ZnSe structures yielded such limited optical results that any further attempts in this direction were discontinued. Obviously, the PVT, by just mechanically delivering the raw material to the substrate surface, is less controllable than HVPE and incapable of providing the flexible options and the variety of chemical paths that the HVPE technique provides. The rest of the options for thick growth are even less competitive. For example, liquid phase epitaxy (LPE) is, practically, a solution growth technique, which suffers from numerous limitations, including relatively thin (microns) growths, difficulty in controlling the composition of the buffer layer, and edge effects (low quality of the layer edges, which are close to the crucible walls, etc. Thus, one of the best achievements of LPE - up to 150-200 .Math.m thick GaAs layers, grown in about 3 hours - is still not comparable with the millimeter thick growth of different materials deposited by HVPE with growth rates of up to 300 .Math.m/h. Solvothermal growth is another example for thick growth. Although quite successful in the bulk growth of quartz, this technique has never achieved anything even close to that performance with any other attempted materials. Yes, growth can be performed heteroepitaxially on multiple substrates but, once the autoclave is closed, nothing else can be done to correct the growth conditions, if necessary. In addition, the grower may wait for months to gain enough layer thickness on the samples that are still not large enough for device development.It was found that HVPE provides more options for controlling the growth process and more choices for growth chemistries for thick homo- and heteroepitaxy. This, however, does not mean that HVPE still does not have its own problems. Such are, for example, the severe parasitic nucleation on the inner quartz surfaces of the reactor that always accompanies the growth and competes with the deposition process that occurs on the substrate surface. Such parasitic nucleation slows down the process (reducing the growth rate on the substrate), depletes the precursor sources, and deteriorates the layer quality.
[0198] The disclosed process allows successful heteroepitaxial growth even at relatively large lattice and thermal mismatches, and without patterning the substrate, which in many cases is the standard procedure. The present invention is based on the accumulation of a significant amount of information about a great number of semiconductor materials and various growth processes used to grow them. This allowed us to realize the significance of the determination of important characteristics of heteroepitaxy, e.g. the thickness of the pseudomorphous growth, the periodicity of the misfit dislocations, and the mechanisms of strain relief, and to successfully make samples for several particular cases. In turn, the determination of these parameters and, in general, enriching our experience and knowledge related to these processes allowed us to develop criteria by which to predict other successful cases of heteroepitaxy, and thus to realize heteroepitaxial growth at mismatches that at first sight looked impossible. Pseudomorphous growth is not ternary or quaternary growth. Instead, it is the initial stage of growth during heteroepitaxy. During this stage of growth the grown layer is “forced” to grow with the lattice constant of the substrate. This growth occurs at the expense of building a linearly-increasing strain in the layer as a result of the lattice mismatch between layer and substrate. However, after reaching a thickness called critical thickness h.sub.c. the strain starts to be released by different energy absorbing mechanisms, e.g. formation of misfit dislocations, roughening of surfaces, formation of voids, etc.
[0199] The disclosed method is based on our understanding of the complex chemistry and growth mechanisms of heteroepitaxy of wideband semiconductor materials. The method secures a smooth transition between two distinct materials, for example GaAs and GaP, or GaAs and ZnSe, or GaP and GaSe, or GaSb and ZnTe, etc., not through a forced-growth process but while preheating the substrate and during the initial stages of growth. During the initial growth stages, the process directs the gradual replacement of substrate atoms, for example V-group atoms, with V-group atoms from the growing material. (Note: The later, however, is not a strict requirement meaning that the replacing atoms can by not only V-group atoms of the growing material but also any other suitable atoms that could assist the smooth substrate-to-layer transition). Thus the process, according to the given example, may direct the replacement of As atoms in the crystal cell of a GaAs substrate with P atoms during the preheating process, which may be conducted in a phosphine (PH.sub.3) atmosphere or in the flow of a mixture of phosphine (PH.sub.3) and arsine (AsH.sub.3). The operator may also make timely decisions to optimize the process, e.g the temperature (if the growth is still in the preheating stage) at which to initiate such replacement, i.e. at which temperature to start exposing the substrate to the non-native precursor), and whether this temperature should be kept constant during the deposition of this buffer layer, or whether it should be increased at some rate until achieving the growth temperature, etc. The user also must decide whether to keep the arsine/phosphine ratio constant or to gradually change this ratio from arsine only to phosphine only in order to achieve the smoothest transition between substrate material (e.g. GaAs) and grown layer material (e.g. GaP). These possibilities point to the great flexibility of the proposed method in controlling the quality of the buffer layer.
[0200] From another point of view, the disclosed approach is universal also because it may be applied to a wide range of different materials having wide ranges of differences in their lattice constants and in their thermal properties expressed by the differences in their thermal expansion coefficients and their thermal conductivities. In spite of all these differences, by using this technique, these materials still may be deposited one over another in a one-step epitaxial growth process, with or without the intentional deposition of the aforementioned intermediate sub-lattice transition buffer layer between them. This approach has been demonstrated to be extremely successful in the growth of GaP on GaAs and in the opposite case, i.e. in the growth of GaAs on GaP although these two cases, from the perspective of lattice mismatches, are not highly favorable (see
[0201] Regarding the particular case of the growth of ZnSe on GaAs, the GaAs substrate (or template) is preheated in hydrogen selenide (H2Se) mixed with H.sub.2 or in a H.sub.2Se:AsH.sub.3:H.sub.2 mixture in order to partially and gradually replace the V-group atom (As) in the GaAs crystal cell with Se (which is a VI-group atom) and thus form a GaAs.sub.xSe.sub.1-x ternary buffer layer. After this step, the growth may continue with the introduction of the Zn—precursor, which may be either metallic Zn overflowed by HCl (or an HCl+H.sub.2 mixture) to form zinc chloride (ZnCl.sub.2), or simply ZnCl.sub.2 overflowed by H.sub.2, or even a Zn—rich ZnCl.sub.2 solution overflowed by HCl+H.sub.2 mixture. (Note: The latter one may be the better choice due to the relatively high vapor pressure of zinc.) Table 2 compares the lattice mismatch of the realized GaP/GaAs heteroepitaxy with the lattice mismatches of some other examples (incl. ZnSe/GaAs) for prospective heteroepitaxial cases as more examples and details are provided in
TABLE-US-00002 Some favorable cases of heteroepitaxy based on their lattice mismatch. Heteroepitaxy Lattice mismatch [%] GaP on GaAs - 3.57 ZnSe on GaAs + 0.24 ZnTe on GaSb + 0.08 ZnS/GaP - 0.57 ZnTe on InAs + 0.70 AlAs/GaAs + 0.13 GaP/Si + 0.37
[0202] As one can see in some of the cases (e.g. ZnSe/GaAs, ZnTe/GaSb, etc.) the V-group atom will not be replaced by a V-group atom but by a VI-group atom (Se, Te, etc.). This means that the replacement of atoms shall be considered as flexible as more factors such as, e.g. the ionic radius or other technological limitations are taken into account when choosing the replacement options. This also means that it is possible to choose for pre-growth treatment of the substrate a chemical that it is not a native precursor for either the substrate or the layer material.
[0203] From Table 2 one may also see that all other given examples by provide much smaller (less than 1%) lattice mismatches, which are more favorable than GaP/GaAs, and which give them a better chance for success. In addition, to make the “right” choice one should take into account also what the desired application might be and some other related properties of the particular material candidate. For example, a brief comparison of ZnSe and ZnTe shows that ZnTe has about the same transmission range as ZnSe but lower 2PA and 3 times higher nonlinear susceptibility at the desired pumping wavelength of about 1 .Math.m, i.e. ZnTe may be a better choice for nonlinear frequency conversion devices. Of course, one also should pay attention to the specific technological limitations related to the growth of a particular material. For example, for the growth of ZnTe, according to the suggested chemistry, we shall use H.sub.2Te for both the pre-growth treatment of the substrate (GaSb) and the actual growth of ZnTe. However, the worldwide supply of H.sub.2Te is relatively limited as, in addition, this gas is relatively unstable - even light assists in its decomposition - which means that the use of alternative Te precursors and chemistry paths is preferable. In contrast, H.sub.2S is readily available and a well-known precursor, and the lattice mismatch of ZnS with the readily available GaP substrates (and OP—GaP templates) is negligible (- 0.569%). Regretfully, the nonlinear properties of ZnS are not as good as the NLO properties of ZnTe, which does not mean that heteroepitaxially grown ZnS (large area ZnS substrates are also unavailable for homoepitaxy) cannot be used for applications other than nonlinear frequency conversion. Another example of a technological limitation is that as of today GaP still cannot be grown by HVPE directly on Si, but it can be grown at high quality by MOCVD or MBE. From this point of view, although that the disclosed method gives better chances for the growth of other materials (including some that have never been grown epitaxially in a monocrystalline shape, and in a size large enough for device development) such as (ZnSe, ZnTe, ZnS, etc.), we continued to explore more options. For example, PH.sub.3 does not exhibit a strong effect on the Si surface during the pre-growth treatment, however, H.sub.2Se dramatically attacks Si wafers with 4-degree miscut during the preheating stage (see
[0204] With regard, again, to the GaP/Si growth (see Table 2) the poorer impact of some non-native precursors, and especially of PH.sub.3, is probably due to the strong Si—Si bond. In this case, bearing in mind that Si can handle higher temperatures, instead of using an alternative precursor, the Si—substrates may be preheated only in H.sub.2 to provoke some thermal decomposition (as shown in
[0205] All the new heteroepitaxial cases (see Table 2 and
[0206] Another direction that may be taken in developing this idea is to make the lattice constants between substrate and growing layer closer by doping the growing layer, while keeping in mind that doping with dopants that are different in size (i.e. smaller or larger ionic radius) will change more or less the lattice constant of the doped material. On the other hand, the dopant concentration may be gradually changed during growth, which will form a transition layer with a gradually changing lattice constant. This approach may secure an even smoother transition between the substrate and the growing layer. Finally, the suggested in-situ doping procedure conducted during the pre-growth treatment or during the initial stages of growth may be replaced by some prior growth doping procedure such as, for example, ion implantation. Indeed, ion implantation is known as being capable of changing the properties, including the lattice constant, of a thin area of the target (in our case, the substrate) near its top surface. Each of these variations of the proposed substrate pre-growth pretreatment shall be followed by growth that aims to deposit a thin or thick epitaxial layer.
[0207] The following examples illustrate particular properties and advantages of some of the embodiments of the present invention. Furthermore, these are examples of reduction to practice of the present invention and confirmation that the principles described in the present invention are therefore valid but should not be construed as in any way limiting the scope of the invention.
Experimental
[0208] As illustrated in
[0209] The second precursor, or ternary-forming gas, is usually a hydride of a V or VI group element (AsH.sub.3, PH.sub.3, H.sub.2Se, H.sub.2S, SbH.sub.3, etc.) diluted to the desired extent by the carrier gas (usually H.sub.2). The second precursor (ternary-forming gas), which is actually the precursor of the V or VI group element, is delivered to the mixing zone, making it available to participate in the growing process. We call the second precursor “ternary-forming” because the reactions between the first precursor gas and the ternary-forming gas on the foreign substrate may result in the formation of ternary islands on the substrate, which may eventually coalescence to form a continuous ternary transition buffer layer.
Example 1 - Growth of GaP on GaAs Substrates, GaAs on GaP Substrates and GaAs.SUB.x.P.SUB.1-x on Either GaAs and GaP Substrates, or All of Them on the Related OP Templates
[0210] These embodiments of the invention are based on the hydride vapor phase epitaxy (HVPE) process, which results in thick heteroepitaxial growth of GaP on GaAs substrates (or of OP—GaP on OP—GaAs templates)(see
[0211] With regard to
##STR00001##
However, forming gallium mono-chloride (GaCl) or gallium dichloride (GaCl.sub.2) when HCl pass over the molten Ga is also possible in similar reactions as shown in the following equations 3b and 3c):
##STR00002##
##STR00003##
Another peripheral flow, a mixture of hydrogen and phosphine (PH.sub.3), as a second precursor gas, in the case of GaP growth, or a mixture of hydrogen and arsine (AsH.sub.3) in the case of GaAs growth, or their mixture (PH.sub.3+AsH.sub.3) is mixed, again, with H.sub.2 as a carrier gas and a diluter, in the case of GaAs.sub.xP.sub.1-x growth, is introduced in the reactor 10 to mix with the GaCl.sub.3 in the second reactor zone 20, called “mixing zone”, with the intention the gases in the mixture to react in such a way that to form on the surface of the substrate 22, respectively, GaP, GaAs, or a GaAs.sub.xP.sub.1-x ternary layer. The same basic hardware is used in all variations of the disclosed method depicted in
##STR00004##
##STR00005##
These reactions are based on the assumption that when HCl is passing over the molten Ga it forms gallium tri-chloride GaCl.sub.3. However, as was shown in eq. 3b and eq. 3c, forming gallium mono-chloride GaCl or gallium dichloride GaGl.sub.2 is also probable. In such cases, the above equations shall be differently balanced, as is shown below in equations 4c and 4d for the case of forming GaP and in equations 4e and 4f for the case of forming GaAs:
##STR00006##
##STR00007##
##STR00008##
##STR00009##
The schematic of the process and its chemistry during the growth of the GaAs.sub.xP.sub.1-x ternaries are shown in
[0212] As one can see all other heteroepitaxial cases discussed further in the text - some of which are illustrated in the following examples - have distinct but similar chemistry.
[0213] Some typical values for the inner flows of H.sub.2 and HCl, and the outer flows of PH.sub.3 and/or AsH.sub.3 related to the HVPE growth of GaP, GaAs, and/or GaAs.sub.xP.sub.1-x are provided in Table 3 below, as an example. However, these numbers are strictly correlated to growths of GaAs, GaP, or their ternaries and to the particular configuration of the HVPE reactor shown in
TABLE-US-00003 Some typical values for the inner and outer flows (sccm) of H.sub.2, HCl, PH.sub.3, and AsH.sub.3 during growths of GaP, GaAs, or their ternaries Inner Flow Outer Flow H.sub.2 HCl H.sub.2 HCl AsH.sub.3 PH.sub.3 65-75 30 - 35 100 - 110 30 - 50 70-80 50 - 60
[0214] With respect to the rest of the growth parameters, all experiments were conducted with parameters within the following ranges: pressure < 10 Torr, substrate temperatures 720-740° C. for the growth of GaP (respectively 680-700° C. for the growth of GaAs), and V/III ratios in the range of 2 - 3. These ranges provided conditions for a close-to-equilibrium process at relatively low supersaturation (typically between 0.5 - 1.0), which is the nature of the HVPE growth.
[0215] Growth experiments were conducted homoepitaxially (GaP/GaP and GaAs/GaAs) and heteroepitaxially (GaP/GaAs and GaAs/GaP) on plain “on-axis” (100) GaAs and GaP substrates and on the same (100) substrates but misoriented with 4° towards (111)B. As it was already mentioned, the growths of the GaAs.sub.xP.sub.1-x ternary at different x ratios were performed on both GaP and GaAs substrates (
[0216] An important step in this process, strongly correlated to this invention, is related to the way of protecting the substrate 22 (
Example 2 - Growth of ZnSe on GaAs Substrates and OP-GaAs Template, as Well as the Opposite Growth of GaAs on ZnSe Substrates
[0217] This embodiment of the invention is based on hydride vapor phase epitaxy (HVPE) and the heteroepitaxial growth of ZnSe on GaAs substrates (
[0218] ZnCl.sub.2 to the mixing zone. Another peripheral flow, a hydrogen + hydrogen selenide (H.sub.2:H.sub.2Se) mixture is introduced as a second precursor in the reactor 10 to mix with the ZnCl.sub.2 in the second reactor zone 20 (the “mixing zone”). The intention is to have the II and VI group (Zn and Se) species in the mixture react and to form a ZnSe monocrystalline layer on the surface of the substrate 22. With regard to
[0219] Growth of ZnSe on GaAs substrates and on OP—GaAs templates is a reasonable option for the growth of ZnSe because large area (enough large for device development) ZnSe monocrystalline substrates are not commercially available - the largest available “monocrystalline” ZnSe samples on the market are typically 5 mm × 5 mm, and which still consist of several domains with different orientations. Consequently, OP—ZnSe templates are unavailable, as well.
[0220] Heteroepitaxy of ZnSe on GaAs is also reasonable due to the very small lattice mismatch (+0.238%) between ZnSe and GaAs. Fortunately, the availability of high quality GaAs substrates, their reasonable price, and the maturity of two OP—GaAs template preparation techniques (the wafer bonding and the MBE assisted polarity inversion technique) provide great opportunities for the growth of high quality crystalline ZnSe on non-native GaAs substrates, as well as OP—GaSe on the high quality OP—GaAs templates.
[0221] The advantage of ZnSe over GaAs and GaP is its wide and smooth IR transparency and its smaller refractive index (see
[0222] Gas flow parameters for the growth of ZnSe on GaAs are similar to those presented in example 1. As depicted in
##STR00010##
##STR00011##
(Note: We assume that when HCl pass over the molten Zn the formed zinc chloride is zinc dichloride (ZnCl.sub.2). However, mono (ZnCl) or tri-chloride (ZnCl.sub.3) may be formed; the chemical equations must be properly balanced, similar to the cases of gallium chloride discussed above (eq. 4c - 4f).)
[0223] The opposite case of heteroepitaxy of GaAs on ZnSe substrates is less attractive for the same reasons and mostly because of the market unavailability of large-area, good quality monocrystalline ZnSe substrates, and the much higher prices of the small dimension ZnSe samples that are available. However, large area polycrystalline ZnSe substrates are available and growth on them for other purposes still might be useful. Growths of GaAs on small monocrystalline ZnSe substrates may be useful for some other applications, bringing the advantage that in the case of GaAs/ZnSe the lattice match is the same but negative, i.e. the GaAs layer will be tensely strained, which according some considerations is more favorable.
[0224] As illustrated in
[0225] The growths discussed in Example 2 are presented schematically in
Example 3 - Growth of a Cubic Phase Zinc Blende Gallium Selenide Ga.SUB.2.Se.SUB.3 on GaP and GaAs Substrates (and the Related OP—GaP and OP—GaAs Templates) and Hexagonal ε—GaSe on GaN (or AlN) Substrates (and the Related OP—GaN Templates)
[0226] This embodiment of the invention is based on hydride vapor phase epitaxy (HVPE) and the heteroepitaxial growth of cubic phase zinc blende gallium selenide (Ga.sub.2Se.sub.3) on GaP and GaAs substrates (
[0227] From point of view of transparency range, GaSe is as good as ZnSe but with nearly 3-times higher nonlinear susceptibility (at λ = 1 .Math.m), making it close to one of the leading materials for frequency conversion devices, GaAs. At the same time using molten Ga instead, the fast evaporating Zn allows runs with longer durations, i.e. eventually growing thicker layers with larger optical apertures. Regretfully, monocrystalline GaSe substrates larger than 10 mm×10 mm are unavailable, and even if they were available, GaSe is too soft (as soft as gypsum with hardness only of 2 by Mohs) to be able to handle the heavy bonding, polishing, and etching procedures performed in the fabrication of OP templates. Fortunately, GaSe has several different crystallographic phases, two of which, a cubic phase zinc blende gallium selenide Ga.sub.2Se.sub.3 and hexagonal ε-phase GaSe have relatively close lattice matches (
[0228] As illustrated in
##STR00012##
However, forming GaCl or GaCl.sub.2 instead GaCl.sub.3 or as fractions of the total gallium chloride flow is not excluded and in such cases eq. 6a shall be differently balanced as is shown below:
##STR00013##
##STR00014##
In these three chemical equations, however, the assumption is that the formed gallium selenide is the hexagonal ε-phase GaSe. For completeness, if we assume that the formed gallium selenide is the cubic phase zinc blende Ga.sub.2Se.sub.3, for completeness equations 6a - 6c shall be rewritten after replacing GaSe with Ga.sub.2Se.sub.3, as it follows:
##STR00015##
##STR00016##
##STR00017##
To be clear, while equations 6a - 6c shall be used for the growth of gallium selenide on the III-Nitride substrates (GaN or AlN), equations 6d - 6f shall be associated with the growths on GaP and GaAs substrates. It is worth also to mention that during all three growths the chemistry is the same because the same material, GaSe, is growing. However, since the substrates are different, the optimal growth temperatures for each one of the substrates could be also different, but 500-900° C. or even 720-850° C. is acceptable. Similarly, during the pre-growth treatment all substrates shall be exposed to the same non-native precursor, H.sub.2Se, with the expectation, however, that the composition of the intermediate buffer layer will be different. Thus, the buffer layer on the GaP substrate shall be from GaPSe, while the buffer layers in the other two cases of growth on GaAs and GaN (or AlN) shall be, from GaAsSe and, respectively, from GaNSe (or AlNSe). Of course, exposing the substrates during the pre-growth stage or some earlier stages of growth to a mixture of the non-native and the native precursors (H.sub.2Se+PH.sub.3, or H.sub.2Se+AsH.sub.3, or H.sub.2Se+NH.sub.3) is always an option. As described above, all these variations are related to either the growth of a cubic phase zinc blende gallium selenide Ga.sub.2Se.sub.3 on zinc blende GaP or GaAs substrates, or hexagonal ε-phase GaSe (GaSe) on hexagonal substrates (GaN or AlN). Although from the point of view of lattice mismatch the most favorable case was initially the growth of a cubic phase zinc blende gallium selenide Ga.sub.2Se.sub.3 on GaP (lattice mismatch of only -0.607%). As discussed in the results, the growth of a cubic phase zinc blende gallium selenide Ga.sub.2Se.sub.3 on GaAs turned out to be most successful, even with the much larger lattice mismatch (-4.181%). The poor results of the GaSe/GaN is not surprising, bearing in mind the large positive lattice mismatch (~17%) between ε—GaSe and the III-Nitride substrates. However, the poor surface morphology and crystalline quality obtained after homoepitaxy of GaSe on a cubic phase zinc blende gallium selenide Ga.sub.2Se.sub.3 substrate is quite surprising. Obviously, all this leaves room for further improvements related to this material.
[0229] In the opposite cases, which may be less reasonable, growths of the hexagonal wurtzite structure GaN or AlN shall be performed on the hexagonal ε-phase GaSe. One of the most probable chemical reactions for the growth of GaN is:
##STR00018##
while for the growth of AlN, the most probable chemical reaction is:
##STR00019##
Respectively, the growth of the cubic zinc blende GaP and GaAs shall be performed on the cubic phase zinc blende gallium selenide (Ga.sub.2Se.sub.3) as the most probable chemical reaction for the case of GaP growth is:
##STR00020##
while for the growth of GaAs the most probable chemical reaction is:
##STR00021##
Although equations 7a-7d are similar or the same as some of the equations given in Example 1 (e.g. eq. 3a, 4a, and 4b) they are presented here again for more clarity. In general, the growth of the same material follows similar chemical schemes no matter what kind of substrate is used.
[0230] In three of the four growths described in Example 3, the growths of GaP, GaAs, and GaN need molten Ga as a source of the III-group element precursors. Only the growth of AlN needs molten aluminum as a source of the III-group element. From the examples provided in this Example 3 one can also see that different materials (e.g. GaN, AlN, GaP) can be grown on different phases of the same material (cubic phase zinc blende gallium selenide Ga.sub.2Se.sub.3 and hexagonal phase gallium selenide ε—GaSe). This is another alternative variation of the proposed in this disclosure growth approach. In other cases, however, different materials may be grown on the same crystallographic phase of the same material but on surfaces with different crystallographic orientations. This is possible because the lattice mismatch in a particular direction may be smaller for a particular material. Thus, it is known that the growth of cubic GaN is more favorable on (100) GaAs substrates, while the growth of hexagonal GaN is more favorable on (111) GaAs substrates. This is another alternative variation of the approach proposed in this disclosure.
[0231] Examples of heteroepitaxial growth of other selenides that are favored by the proposed approach are, e.g. CdSe/InAs (lattice mismatch - 0.139%) or even growths of other layered materials such as, e.g. InSe. In the latter case the lattice mismatch and from here the substrate are not that important as far as the most probable growth of a layered material is through the van der Waals heteroepitaxy. Any substrate is acceptable. In these experiments, the II or III group material in the compound (in this case Cd and In) determines the first precursor, which may be just molten metal in the designated quartz boat 14. In contrast, the second precursor (the ternary gas) in all these cases may be hydrogen selenide H.sub.2Se, which is also the non-native precursor for the pre-growth treatment of these substrates. The chemistry in these two cases as also presented in the table in
##STR00022##
for the growth of CdSe, and:
##STR00023##
for the growth of InSe. (Note: The large variety of possible chemical paths shown in this Example 3 is applicable to some extent in the rest of the given examples but for simplicity, they are not shown in such details. For the same reason the
[0232] The growths given in this Example 3 are presented in
Example 4 - Growth of ZnTe on GaSb and InAs Substrates
[0233] This embodiment (see
##STR00024##
The different substrate materials may require different temperatures for pre-growth treatment and growth, depending on the material properties. At the same time, although the substrates are pre-growth treated with the same non-native precursor, H.sub.2Te, the formed ternary transition buffer layer shall be different. Thus, in the case of a GaSb substrate, the treatment shall result in the formation of a GaSbTe ternary buffer layer, while in the case of an InAs substrate the treatment shall result in the formation of an InAsTe ternary transition buffer layer.
[0234] The advantage of growing ZnTe comes from the fact that ZnTe has the widest and smoothest IR transparency among all known nonlinear optical materials for frequency conversion, which starts conveniently in the visible region and goes all the way down to 20 .Math.m. At the same time, the ZnTe nonlinear d-coefficient is almost as large as the d-coefficient of GaAs. ZnTe, in addition, has one of the lowest two-photon absorption (
[0235] However, there is one important inconvenience that one should take into account when planning to use our approach for growth of ZnTe or other tellurides - the poor availability of H.sub.2Te. As most of the hydrides, hydrogen telluride is corrosive, flammable, and poisonous. However, the real reason for its very limited presence on the market is that H.sub.2Te is an unstable gas, which decomposes easily; light facilitates this process. To solve this issue, we propose instead of using H.sub.2Te, to produce it in-situ within the reactor volume as a part of the growth process. This approach, i.e. producing in-situ of a needed chemical, which is hard to provide or, which is a subject of serious hazardous limitations, shall be considered as another variation of this disclosure, as far as we believe that it can be applied in other similar complicated cases.
[0236] H.sub.2Te may be produced in-situ by using several commonly available Te-precursors. The first idea was to use some of the metal-organic tellurium precursors such as dimethyl or diethyl tellurium, which are already used in the MOCVD growth of ZnTe. However, while these metal-organics are relatively expensive, their chemistry is relatively complex. The solution we propose in this disclosure is based on simple chemical compounds which are readily available at a reasonable price, which in this particular case are tellurium tetrachloride (TeCl.sub.4), sodium telluride (Na.sub.2Te), and N.sub.2TeO.sub.3. These chemicals, if not exactly environmentally friendly, are at least not strongly poisonous. As hydroscopic materials, we shall protect them from the environment rather than the environment from them. However, most importantly, they can relatively easily produce H.sub.2Te under rather ordinary conditions in one the following simple chemical reactions:
##STR00025##
##STR00026##
##STR00027##
(Note: In contrast to the other two chemicals, the price of Na.sub.2Te is relatively high. That is why we may choose to avoid the reaction shown in eq. 10b, although it is relatively simple.)
[0237] Some other heteroepitaxial growths of tellurides (not shown schematically but included as favorable cases in
##STR00028##
##STR00029##
##STR00030##
##STR00031##
However, since the substrate materials are different for each of these cases, each of them will have its own specific composition based on the material properties of the related substrate. Accordingly, the composition of the ternary or quaternary buffer layer formed as a result of the pre-growth treatment in each individual case will be different as well, but will be some combination of 3 or 4 of the elements, Ga, Te, In, P, As, and Sb. On the other hand, the decision as to how reasonable a particular heteroepitaxial growth is may be based on how “common” (i.e. how available and expensive) a given substrate is and how important from practical point of view the growing layer is. This is one of the reasons why one might pay less attention to the opposite growths, including those related to the last four examples.
[0238] Some doped substrate materials may be a good solution for the growth of ZnTe substrates (and for other materials, as well). In general, many commercially available substrate materials are undoped (or unintentionally doped), while others are n- or p-doped, even co-doped. The doping may occur during the growth of the crystal boule, when the dopant is distributed relatively uniformly in the entire volume of the material. However, the introduction of the dopant atoms may also be done after the boule is sliced into substrates by using techniques such as ion implantation. In such cases the dopant atoms are introduced only to a small depth in the area near the substrate surface and do not present in the entire substrate volume, which in some cases is enough. While the major role of the doping is, in general, to change the material properties, it may also be used as another way of pretreating the sample prior to growth. This, in another way, may facilitate the process of heteroepitaxy from the point of view of ensuring a smoother transition between the substrate and the growing layer and more efficient relief of the initial stress built at the growing interface due to the lattice and thermal mismatches. This is because the dopant atom, depending on its ionic radius, may change the lattice constant of the substrate material, making it closer to the lattice constant of the growing material. For example, GaSb substrates are offered on the market as undoped and doped substrates, including some heavily doped with Zn. In this case the Zn—rich GaSb—Zn substrates shall match closer to the growing ZnTe layer than a regular undoped GaSb substrate. The degree to which the dopant atom will change the crystal cell of the substrate in the ‘right’ direction and how much it will contribute to the formation of the ternary transition layer shall be a subject of study in any particular case. Doping or using properly doped (and eventually co-doped) substrates for the following heteroepitaxial growth is another variation of our approach presented in this disclosure.
[0239] Some of the cases presented in Example 4 are illustrated in
Example 5 - Growth of ZnS on GaP and CdS on InP Substrates
[0240] This embodiment (see
[0241] The most likely chemistry for growing ZnS and, respectively, CdS (and some other sulfides), by the approach proposed here may be expressed by the following chemical equations:
##STR00032##
and, respectively:
##STR00033##
[0242] In some cases, sulfides offer great advantages compared to other nonlinear optical materials. Let us compare, for example, ZnS with ZnTe. Next to the convenience of using one well-known, readily available, and relatively cheap precursor (H.sub.2S), ZnS, itself, offers several attractive features, even compared in one direction only - as a nonlinear optical material for frequency conversion devices. The ZnS transparency does not go very far into the infrared than that of ZnTe, but it starts at shorter wavelengths (see
[0243] Some of the cases presented in Example 5 are illustrated in
Example 6 - Growth of Some Other Antimonides Such as GaSb and AlSb on InAs Substrates, and AlSb Also on GaSb Substrates
[0244] As it was already discussed in Example 4 the opposite to the ZnTe/GaSb growth, i.e. the growth of GaSb on ZnTe is less reasonable due to the commercial absence of large area monocrystalline ZnTe substrates. However, due to the small lattice mismatch of GaSb with InAs (+0.620%) and the availability of InAs substrates GaSb may be easily grown by the disclosed technique on available InAs substrates, although GaSb substrates are available at about the same price and quality. However, there are a number of other reasons to prefer hetero- before homoepitaxy. An example is the case where the device development requires a structure combining two or more materials with different bandgap energies. For the same reason the more exotic material AlSb (relatively small lattice mismatch of +1.273%) may also be grown on InAs substrates, although the lattice mismatch of AlSb with GaSb is smaller (+ 0.650%) and that is why the growth of AlSb shall be more favorable on GaSb than on InAs substrates.
[0245] This embodiment (see
##STR00034##
for the case of the GaSb growth, and:
##STR00035##
for the case of the AlSb growths.
[0246] Since AlSb is a material that is more exotic, and large area monocrystalline AlSb substrates are unavailable, the opposite growth of GaSb on AlSb is less reasonable (if not impossible at all), but the opposite growth of InAs on GaSb is as reasonable as the GaSb/InAs growth. Moreover, in this case the sign of the lattice mismatch is negative, which means that the GaSb layer grows under tensile strain, which according to some studies is a more favorable case than those grown under compressive strain (positive lattice mismatch). In this case, the source of the III-group element is Ga, while the second precursor, the ternary gas, is SbH.sub.3, which during the pre-growth treatment of the InAs substrate shall form an intermediate InGaAs ternary buffer layer. The probable chemical reaction for this growth is:
##STR00036##
[0247] Some of the cases presented in Example 6 are illustrated in
Example 7 - Growth of AlAs on GaAs Substrates
[0248] The growth of AlAs on GaAs is an illustrative example of the usefulness of the approach proposed in this disclosure. First, AlAs substrates are not available, which comes with two conclusions. First, heteroepitaxy is the only way to grow AlAs layers and, second, the opposite growth, i.e. GaAs/AlAs is unreasonable and, in general, impossible. The growth of AlAs on GaAs is highly reasonable for the following reasons: First, the relatively small lattice mismatch (+0.127%) between AlAs and GaAs. Second, as a substrate material GaAs is almost as common as Si, i.e. it is widely available, with high quality and with relatively low cost. Third, OP—GaAs templates with high quality are readily available, which means that growth of OP—AlAs may also be immediately attempted. Fourth, the band gap of AlAs is larger (different) than the bandgap of GaAs. Fifth, it is known that oxygen is the most common and least desirable impurity in Al-containing semiconductor materials. By using metal aluminum, which has the highest possible purity (much higher than the purity of any other Al-precursor) and by pre-making in-situ the first precursor (aluminum chloride) within the reactor chamber, the proposed approach ensures the lowest possible oxygen levels in growing AlAs.
[0249] This embodiment (see
##STR00037##
for the AlGaAs ternary layer formation, and:
##STR00038##
for the HVPE growing AlAs layer.
[0250] Some of the cases presented in Example 7 are illustrated in
Example 8 - Growth of Multilayered Structures
[0251] The disclosed method has another advantage - it may be used for the growth of multilayer structures in pursuing different applications. For example, growth may start on a GaAs substrate pretreated with PH.sub.3 followed by the growth of a GaP layer. In the next step, the grown GaP is pretreated with H.sub.2S followed by the growth of ZnS taking advantage from the small negative lattice mismatches between GaP and ZnS (- 0.57 %). Thus we could have in one structure three layers (ZnS/GaP/GaAs) with increasing bandgap energies (see
[0252] Another example is the growth of a ZnSe/GaAs/GaP/Se multilayered structure. In this case, after the deposition of GaP on a Si substrate, the as grown GaP layer shall be pretreated in AsH.sub.3. After the deposition of a GaAs layer on the GaP layer, as a next step the as grown GaAs layer shall be pretreated in the flow of H.sub.2Se after which to continue with the ZnSe/GaAs growth.
[0253] These examples are only two of many other options for growth of multilayered structures for various practical applications. Growth of multilayered structures by using the approached provided in the text is another variation of the disclosure proposed here. For simplicity, the examples provided within the text of Example 8 are not illustrated on separate figures.
Example 9: Growths After Prior Growth (in-situ or Non-in-situ) Treatment of the Substrate With a Non-Native Material That Is Also Non-Related to the Substrate or the Layer Material Precursor
[0254] it was discovered that in some heteroepitaxial cases that the non-native precursor typically associated with the growing material has less impact on the foreign substrate than another not related to the growing material non-native precursor. For example, while PH.sub.3 has almost no effect on the surface of a Si wafer exposed prior to the intended direct HVPE heteroepitaxy of GaP on Si, the exposure of a (100) Si wafer with 4° miscut to H.sub.2Se is visibly attacked by this non-native (to Si) material, which is also not related to the GaP layer precursor (see
[0255] Si is, in general, inert to acids, except hydrofluoric acid (HF), which is the aqueous solution of the gas H.sub.2F. HF has the unique ability to react with the naturally formed thin silicon oxide (SiO.sub.2) layer (called also silica) when the material is even briefly exposed to air or water. HF forms with SiO.sub.2 a hydrogen terminated layer of Si—H bonds in the reaction:
##STR00039##
The hydrogen from this top surface layer may be used to form bonds with other chemical elements by treating the surface with other precursors, for example with H.sub.2Se or PH.sub.3. If the surface is, however, saturated with fluorine (F) atoms through the formation of SiF.sub.4 (see eq. 18) such a treated Si—surface may be used for the growth of some fluorides, e.g. CaF.sub.2 or BaF.sub.2, which are also excellent optical materials. The pre-growth treatment in this case may be done before introducing the Si substrate in the reactor chamber in HF by following some standard or optimized procedures for treating Si in HF. The pre-growth treatment of the Si substrate may also be done in-situ within the reactor chamber by exposing the Si—wafer to HF gas flow. This option - pre-growth treatment of the substrate in a suitable solution before introducing it into the reactor chamber is another variation of our approach along with the aforementioned “in-situ” pre-growth treatment of the substrate in a non-native related or not related to the layer material precursor.
[0256] The preliminary deposition of a thin buffer transition layer from the same layer material by a different, typically, far-from-equilibrium growth technique, such as MOCVD or MBE, that was already described in the text could be accepted as another alternative variation of a pre (HVPE) growth treatment of the substrate,
[0257] For simplicity, the examples provided within the text of Example 9 are not illustrated in separate figures.
Example 10 - Growths of Nonlinear Optical and Other Single or Compound Semiconductor Materials on Halides
[0258] To this point the heteroepitaxy examples provided were related mostly to either some well known As, P, or Sb-based semiconductor materials, e.g. GaAs, GaP, GaSb, etc., which may possess optical nonlinearity, or similar chalcogenide materials, i.e. materials containing S, Se, or Te, e.g. GaSe, ZnS, ZnTe, etc. However, there are still many other areas and different kinds of materials to match where heteroepitaxy can be a great help. Such materials are, for instance, the halides, i.e. materials that contain F, Cl, Br, or I. Examples for using halides as substrates are two well-known optical materials with a broad range of transparency that starts in the ultraviolet (UV) range and continues all the way to the mid- and long-wave IR (see
[0259] For simplicity, the examples provided within the text of Example 10 are not illustrated in separate figures.
Example 11 - Growths of Non-Ferroelectric Materials on Ferroelectric Substrates (GaP/LiNbO.SUB.3.) and OP Templates (OP—LiNbO.SUB.3.)
[0260] This example of the growth of a non-ferroelectric material (GaP) on a substrate or an OP template from a ferroelectric material (LiNbO.sub.3) is, at the same, time an example of the growth of a material with one crystal symmetry (cubic) on another material with a different crystal symmetry (trigonal). Furthermore, this is also an example for the growth of a traditional III-V semiconductor material (GaP) on a traditional oxide material (LiNbO.sub.3).
[0261] Although the in-plane (a-plane) lattice mismatch between GaP and LiNbO.sub.3 is relatively large (+ 5.87%), there are good reasons for choosing such combinations, among which is the extremely easy way for in-situ preparation of ferroelectric OP materials by growing them in a periodically alternating electrical field. Such periodically polled Lithium Niobate (PPLN) may be easily grown from melt by the conventional CZ growth technique and eventually sliced into OP—LiNbO.sub.3 templates. As a following step, thick HVPE growth of GaP is performed on these templates using the aforementioned chemistry for the growth of phosphides (see equations 4). The unfavorable growth of GaP on the large matching LiNbO.sub.3 may be dramatically facilitated by the growth of either an intermediate transition buffer layer or by a thin GaP layer on the LiNbO.sub.3 substrates (respectively, the OP— LiNbO.sub.3 templates). In the latter case, the deposition shall be performed by a far-from-equilibrium growth technique, such as MOCVD or MBE. The reason for this is that these techniques have demonstrated in practice that, although they are favorably less sensitive to the lattice mismatch, they can still transfer the polarity of the material underneath unchanged.
[0262] All of these examples of growth of one material on a substrate with different crystal symmetries, or the growth of a non-ferroelectric material on a ferroelectric substrate or template, or the growth of a semiconductor material on an oxide material, are variations of the method disclosed herein. Another variation is the HVPE growth of a thick layer on a thin layer (from the same or from a different material) deposited in advance on the substrate or on the template by a process that is less sensitive to lattice mismatch, such as a far-from-equilibrium growth process, e.g. MOCVD or MBE.
[0263] For simplicity, the examples provided within the text of Example 11 are not illustrated in separate figures. It should be also clear that by illustrating these examples with only one particular pair of materials, i.e. GaP and LiNbO.sub.3, we do not put any limitations to other suitable materials that may be suggested in each of the above particular cases. GaN, for example, is another material where the fabrication of OP templates is easy, due to the great deal of effort dedicated to polarity control in III-Nitrides. This makes GaN and, in general, the III-Nitrides candidates as substrates or templates for heteroepitaxy in cases similar to these provided in example 11.
Some Basic Crystal Growth Considerations
[0264] Without being bound by theory, during heteroepitaxy the relation between the forces that keep the atoms of the substrate in place and the atoms of the growing layer, Ψ.sub.AA and Ψ.sub.BB from one side and the interfacial force Ψ.sub.AB from the other side, is important. Thus, in the case when Ψ.sub.AB » Ψ.sub.BB and Ψ.sub.AB ≅ Ψ.sub.AA the interfacial force Ψ.sub.AB is strong enough to produce pseudomorphous growth. As a result, during pseudomorphous growth the lattice of the growing crystal B (e.g. GaP) will be, initially, homogeneously strained to fit to the lattice of the substrate crystal A (e.g. GaAs), which occurs at the expense of a linearly-increasing elastic strain. This, depending reciprocally on how large the lattice mismatch is, may typically continue to the deposition of no more than about 10-15 monoatomic layers. After this critical thickness h.sub.c (the thickness of the pseudomorphous growth), according to the misfit dislocation (MDs) concept, the pseudomorphous growth will become energetically unfavorable and the homogeneous strain will be released in the formation (in the ideal case) of MDs with a periodicity τthat should depend on the difference between the two lattice constants a.sub.0 - b.sub.0. (Note: Interfacial force, Ψ.sub.AB , is the force across the interface between two faces that keep them together.)
[0265] The critical thickness h.sub.c is, in general, larger when the lattice mismatch in a particular heteroepitaxial case is smaller, i.e. for different cases h.sub.c increases with the decrease of the lattice mismatch. However, the critical thickness does not depend linearly on the linearly increasing elastic strain. Many other factors, such as the sign of the lattice mismatch, the mechanisms of MD formation or other alternative mechanisms of strain relief, etc., must be considered in order to predict this thickness. Most of these factors have been incorporated into several models related to the particular case of stress relaxation that occurs through misfit dislocation nucleation. Thus, taking into account the sign and the degree of the lattice mismatch between GaP and GaAs (-3.57%) (See eq. 1) and the periodicity τ of the MDs (~28 interatomic distances) (see eq. 2), it was attempted, theoretically to determine the expected thickness of the pseudomorphous growth h.sub.c in this particular case of GaP/GaAs growth (see
[0266] When searching for the appearance of MDs, however, one should bear in mind that all theoretical works and simulations (such as those graphically expressed in
Characterizations
[0267] Each pretreated or grown sample was characterized with regard to its surface morphology and crystalline and optical quality by at least several of the following characterization techniques: cross section and top layer surface Nomarski optical imaging, x-ray diffraction (XRD), scanning electron microscopy (SEM), atomic force microscopy (AFM), tunnel electron microscopy (TEM), electron dispersion spectroscopy (EDS), optical transmission, and linear and nonlinear optical absorption. Each of these material characterizations was performed in order to reveal the mechanisms of the formation of defects near the interface between the substrate and the growing layer, and also how these defects propagate in the layer and how they impact the final layer quality, taking into account the impact of the applied pretreatment conditions and the applied growth parameters during the growth stage.
[0268] Characterizations related to the surface morphology, the crystalline layer quality, and other electrical and optical parameters related to some specific practical applications were used also as a feedback to the growth process that allowed the determination of the optimal parameters for pretreating and growth, e.g. substrate and mixing zone temperatures and the rates of their increase or decrease, reactor pressure, V-III ratios, gas flow regimes, etc., for a number of different material cases.
[0269] From the dark field TEM images of GaP grown on GaAs sample (see
[0270] Our measurements indicate that we have determined good control parameters for engineering the transition buffer layer, particularly with regard to its thickness, composition, and quality, which allows a smooth transition between two mismatched materials.
[0271] For example, we demonstrated that by extending the idea of the transition buffer layer to the growth of a thick ternary layer, we were able to achieve hundreds of microns thick layer of GaAs.sub.xP.sub.1-x with a changing composition. In a particular series of experiments we grew 300-700 .Math.m thick GaAs.sub.xP.sub.1-x with compositions within the range of x = 0.1-0.91 as the growth on plain (GaAs) substrates (
[0272] Another advantage of ternaries, and specifically of GaAsP, can be easily seen on the phase-matching curves for GaAs (
[0273] By demonstrating that heteroepitaxy is possible and successful in some less favorable cases (e.g. GaP/GaAs and GaAs/GaP) we have opened the doors wide for other heteroepitaxial cases (e.g. ZnSe/GaAs; ZnTe/GaSb; ZnTe/InAs; AlAs/GaAs; GaSe on GaP, GaAs, or GaN; ZnS/GaP; and even GaP/Si) that often provide closer, more favorable lattice and thermal matches. As provided in the Experimental results above, many of these cases have already been successfully grown by our technique in large size substrates (halves or quarters of 2-inch wafers) with high surface and crystalline quality. For example, up to 500 .Math.m thick ZnSe with smooth surface morphology and FWHM of about 60 arcsec of XRD 2 theta-omega scan was grown with growth rate of up to 105 .Math.m/h as it is shown in
[0274] In this series of experiments, cubic phase gallium selenide with zinc blende symmetry (Ga.sub.2Se.sub.3) was grown on GaP (
[0275] Although ε—GaSe has been used for a long time for phase matching frequency conversion in the mid-IR, GaSe, has never achieved the ubiquity of other NLO materials in commercial and industrial applications due to the difficulties associated with the growth process and sample preparation. This comes from the fact that GaSe has a layered structure with weak interlayer van der Waals bonding and, further, hardness close to zero by the Mohs’ scale. The numerous phases in which this material can exist (even co-exist in a single sample) brings additional complexity to the growth process. In contrast to the hexagonal ε-phase GaSe (suitable for phase matching frequency conversion) the cubic phase zinc blende Ga.sub.2Se.sub.3 could be an alternative material for quasi-phase matching frequency conversion. However, although Ga.sub.2Se.sub.3 has been grown as single crystals, large enough crystalline gallium selenide is not practically available, plus, such a soft material cannot handle the heavy polishing and etching procedures associated with the preparation of an OP—template. These two reasons make heteroepitaxy the only option for the growth of gallium selenide for QPM frequency conversion. Fortunately, the zinc blende Ga.sub.2Se.sub.3 has about the same lattice mismatch with GaAs (— 4.02%) as GaP with GaAs (— 3.57%). This supports for favorability of Ga.sub.2Se.sub.3/GaAs heteroepitaxy and, eventually, the growth of OP—Ga.sub.2Se.sub.3 on OP—GaAs templates, similar to GaP/GaAs and OP—GaP/OP—GaAs heteroepitaxy. Realizing HVPE growth of continuous layer zinc blende Ga.sub.2Se.sub.3 with smooth surface morphology and excellent crystalline is a good opportunity to grow this excellent NLO material for QPM frequency conversion in the MLWIR. However, the even closer lattice mismatch with GaP (-0.42%) supports the heteroepitaxy of Ga.sub.2Se.sub.3/GaP. Moreover, other applications of that would find heteroepitaxy of the cubic zinc blende gallium selenide on the close lattice matched Si are also attractive. For example, in contrast to some traditional 2D materials, wherein zero bandgap energies restrict them from their use in logic electronics or for field-effect transistors (FETs), the low-dimensional GaSe has the advantage that its bandgap energy may be tuned by the number of the deposited monolayers, which are easily controlled by growth duration. In addition, GaSe has a strong photo-response, which makes this material suitable for photodetectors. The growths of GaSe presented herein are classic examples of van der Waals heteroepitaxy, which is another variation of the invention.
[0276] According to
[0277] From point of view of lattice mismatch, many of the heteroepitaxial cases presented in
[0278] In one of the example in
[0279] As taught above, however, the lattice mismatch is not the only important criterion when matching two materials in a growth process. First, to be practical, the substrate material should be available in a relatively large size (e.g. at least 2-inch wafers), at a reasonable price, and with high surface and crystalline quality; the so-called “epi-ready” surface. The availability of such substrates indicates a mature growth and wafer preparation technology, as well viable OP template preparation techniques. Next to the well-known, common substrates, e.g. Si, Ge, GaAs, or GaP, some other materials, e.g. InAs, InP, InSb, GaSb, and CdTe are also available as substrates for subsequent epitaxial growth. From this point of view, it makes perfect sense to grow materials that are either more expensive or not available in a large size and good crystalline quality, e.g. ZnSe, ZnTe, or GaSe, on common or high quality substrates, e.g. growth of crystalline ZnSe/GaAs, ZnTe/GaSb, or GaSe/GaAs, etc., as well on the related OP templates, when they are available. In the same way, it is much more reasonable to grow CdSe, which is also not available as large crystalline substrates, on InAs, which is available at high quality and at a reasonable price of about $100 per 2-inch wafer. Similarly, it is preferable to grow CdS (about $2,000 per 2″ wafer) on the cheaper InP (about $400 per 2-inch wafer) than to perform the opposite growth, i.e. InP/CdS. Also to be considered are the growths of zinc-blende (cubic) materials on zinc-blende (cubic) substrates (e.g. GaP/GaAs), or wurtzite (hexagonal) materials on wurtzite (hexagonal) substrates (e.g. GaN/sapphire) rather than, for example, a zinc-blende material, which has a cubic symmetry, on a wurtzite substrate, which has a hexagonal symmetry. One also should bear in mind that different crystallographic orientations might provide a closer lattice match to different phases of one material. Thus hexagonal GaN may be successfully grown on (111) GaAs substrates, while cubic GaN may be grown on (100) GaAs. The opposite arrangement, i.e. that two different phases of the same material could be grown successfully on completely different materials is also possible in many particular cases. Thus, as described above, Ga.sub.2Se.sub.3 can be grown on GaP, while ε— GaSe can be grown on GaN (for more examples see
[0280] Many other factors important to the growth process shhhould also be taken into account. For example, the lack of a H.sub.2Te precursor on the market and its easy decomposition to H.sub.2 and Te (most probably before to have the chance to react with the other chemicals involved in the growth) require a search for alternative chemical approaches in the growth of tellurides, for example, in the growth of ZnTe. One of these approaches is to produce in-situ H.sub.2Te within the reactor volume using other chemicals and chemical reactions, e.g. thermal decomposition of TeCl.sub.4 in the flow of H.sub.2 or H.sub.2+HCl mixture (see eq. 10a), or other options shown in eq. 10b and eq. 10c (see also
[0281] Next to the already described compound semiconductor substrates we may also add some more details about using some traditional optical (CaP.sub.2 or BaF.sub.2) or ferroelectric (LiNbO.sub.3) or non-ferroelectric materials as substrates, including some perovskites (e.g. BaTiO.sub.3), some of which were already mentioned above and in
[0282] One should also bear in mind that, as described above, from a crystallographic point of view, growth at a negative mismatch that results in a tensile strained growing layer is more favorable than growth performed at a positive mismatch, i.e. under compressive strain. One simplified explanation of such a preference is that a tensile growing layer can compensate to some extent the strain of the naturally compressed substrate surface, as well as the fact that the tensile growth provides conditions for thicker pseudomorphous growth that are also more favorable for 2D (layer) growth, which is the reason for observing smooth surface morphology after such growths. In contrast, when the growth occurs under compressive strain, such growth conditions allow a smaller critical thickness and favor 3D (island) growth instead, which results in rougher surface morphology. That is why, for example, growth of CdS/InP (-0.624% lattice mismatch) should be preferred to the growth of InP/CdS (the same but positive, +0.624%, lattice mismatch). However, such considerations may depend on what we want to achieve with the growth, whether metamorphic buffer layers or the formation of quantum dots, wells, or other nano or microstructures.
[0283] It should also be remembered that the thermal mismatch between the growing layer and the substrate, i.e. the difference between the thermal expansion coefficients and their thermal conductivities, starts to play a more and more important role with increasing the layer thickness, which can lead to cracking of the growing layer. For example, the 3 times smaller thermal conductivity of ZnSe (18 W.m-1K-1) but larger thermal expansion coefficient (compare 7.1 vs. 5.7 10.sup.-6 . K.sup.-1) (see
[0284] In general, the best ternary for the buffer layer will be the one that is formed by the two parenting materials. For example, GaAsP sounds like the best buffer layer material for the GaP/GaAs or GaAs/GaP growths. For the purpose, the initial growth shall be initiated under the flow of a mixture of AsH.sub.3 and PH.sub.3. In this case, the ratio of these two precursors in the mixture may be changed from only arsine (AsH.sub.3) to only phosphine (PH.sub.3) in order to build up a graded GaAsP ternary buffer layer. However, if we want to grow a thick GaAsP ternary layer (not only a ternary buffer layer), we shall keep a desired precursors’ ratio constant, to obtain a constant layer composition. One should also keep in mind that phosphine is much more volatile than arsine, which means that to achieve equal amounts of phosphorus and arsenic in the ternary composition, the amount of phosphine in the AsH.sub.3+PH.sub.3 ratio shall be much greater than arsine. In this example, the III-group element (Ga) is the same in the substrate and the layer material. Similar to this example is the growth of GaSe on GaAs. In this case, the ternary GaAsSe shall be formed by using one boat of molten Ga over flowed by a mixture of arsine (AsH.sub.3) and hydrogen selenide (H.sub.2Se). Similar is also the case of growth of GaSb on GaAs, where a mixture of arsine and hydrogen antimonide (called also stibine), AsH.sub.3+ SbH.sub.3, flows over the boat with molten Ga, to form the GaAsSb ternary. The cases of growths of layers on substrates when the III-group element of the layer and substrate are different, however, are more complicated. Thus, according to one study GaSb has been grown (using a different growth technique, MBE) not only on GaAs substrates but also on InAs and AlSb buffer layers deposited in advance on the GaAs substrate. In these two cases (InAs and AlSb buffer layers) the III-group elements, when growing the initial buffer layers on the GaAs substrates, are different — Ga and In in the InAs/GaAs case and, respectively, Ga and Al in the AlSb/GaAs case. Towards this purpose, an additional boat with molten In or, respectively, with molten Al is introduced into the reactor chamber. As for the V or VI group element precursors, only one V-group element precursor, AsH.sub.3, is needed for the growth of the InAs buffer layer, while in the case of the AlSb/GaAs growth we will need not only a V-group element precursor, AsH.sub.3, to flow over the boat with the molten Ga but also one VI-group element precursor, SbH.sub.3, to flow over the boat with the molten Al. In addition, making it even more complicated, after the deposition of the buffer layer we proceed with the growth of the desired GaSb layer on the grown InAs or, respectively, AlSb buffer layer. For this purpose, in order to initiate the GaSb/AlSb growth, the VI-group element precursor, SbH.sub.3, shall be kept flowing over both the Ga and the Al boats, while the AsH.sub.3 flow shall be turned off. Respectively, in the case of GaSb/InAs growth, the VI-group element precursor, SbH.sub.3, shall be turned on to overflow the boat with molten Ga, while the AsH.sub.3 flow shall be kept flowing over the boat with molten In. Thus, the growth of GaSb on GaAs with the assistance of an InAs buffer layer (GaSb/InAs/GaAs) will start with the growth of an intermediate GaInAs ternary layer at the InAs/GaAs interface. This with the involvement of SbH.sub.3 will convert gradually through a GaInAsSb quaternary phase within the InAs buffer layer to a GaInSb ternary layer near the growing GaSb/InAs interface which, eventually, by reducing the In content (with reducing the AsH.sub.3 flow over the molten In) gradually will convert in the desired GaSb binary layer. Similarly, the growth of GaSb on GaAs with the assistance of an AlSb buffer layer (GaSb/AlSb/GaAs) will start with the growth of an intermediate AlGaAsSb quaternary buffer layer at the AlSb/GaAs interface, which will convert into an AlGaSb ternary at the GaSb/AlSb interface, which eventually will convert in the desire GaSb binary layer.
[0285] Another example with similar complexity is the growth of ZnSe on a GaAs substrate. The growth in this case occurs with the involvement of one II and one III-group element precursors (Zn and Ga) and two V-group element precursors (AsH.sub.3 and H.sub.2Se). However, this process could be simplified by not involving AsH.sub.3, if (similarly to the growths of GaP/GaAs or GaAs/GaP) the non-native VI-group element precursor (H.sub.2Se) “attacks” during the pretreatment stage the surface of the GaAs substrates, with the expectation that this ternary transition layer will gradually convert into ZnSe during the following ZnSe growth.
[0286] Other good examples of heteroepitaxy are the growths of ZnTe/GaSb (
[0287] In summary, the examples provided in the previous paragraphs consider two different approaches for facilitating heteroepitaxial growth. The first one is by “in-situ” growing of a ternary or quaternary buffer layer between the substrate and the growing layer. The second one is by growing the intended layer on an already deposited buffer layer (it could be by a different growth technique) from a material that, hopefully, has a close lattice and/or thermal match with at least one of the substrate or the growing layer. In many cases, multiple buffer layers from different materials may be used to secure the desired smooth gradual transition between two materials that, at first sight, look completely incompatible. One good example of that is the growth of InSb on a GaAs substrate. The first step in this growth is to grow an intermediate layer of InP (lattice constant 5.8668 Å) on the GaAs substrate (lattice constant 5.6533 Å). The second step is to grow a second intermediate layer of GaSb (lattice constant 6.0959 Å) on the first InP buffer layer. The next step is to grow a third intermediate layer of ZnTe (lattice constant 6.1010 Å) on the second GaSb intermediate layer, and then to finish with the growth of a thick growth of InSb (lattice constant 6.4794 Å) layer. This was our initial goal, which, however, was not possible by a direct InSb/GaAs growth, due to the huge positive lattice mismatch (+10.74%) between these two materials. Of course, the success of such efforts will be greatly improved if the intermediate layers are grown in mixtures of the related precursors (in the case of the first InP/GaAs transition, an AsH.sub.3+PH.sub.3 mixture in the presence of Ga and In overflowed by HCl, etc.), which will support the growth of ternary or quaternary intermediate layers with a gradually-changing composition, ensuring a smooth transition between the two materials. Thus, materials that are otherwise completely incompatible may be grown on each other, even with differences in the lattice constants of 10 Å or more.
[0288] In some cases, the substrate pretreatment may be successful with a non-native precursor that has nothing to do with the subsequent growth. An example of that is the pretreatment of a Si substrate with H.sub.2Se in order to prepare the Si surface for the following thick HVPE growth of GaP. In this case, PH.sub.3 had to be the first choice of a non-native precursor for pretreatment of the Si substrate with the idea PH.sub.3 to pit the Si surface, starting to form GaPSi islands, which to coalesce eventually in a GaPSi buffer layer. The effect of PH.sub.3 on the Si substrate, however, was not as strong as expected, as compared to the unexpectedly strong effect of H.sub.2Se (
[0289] In summary, the major criteria for choosing the substrate and the growing material pair are: [0290] 1. The sign and the magnitude of the lattice mismatch between the substrate and the growing layer: This determines the type of the elastic strain built in the growing layer and the mechanisms of its relief, as well the thickness of the pseudomorphous growth (h.sub.c) and which type of growth - 2D (layer growth) or 3D (island growth) is favored. [0291] 2. The difference in the thermal expansion coefficients and the thermal conductivities of the substrate and the growing material: The importance of this factor increases with increase of the growing layer thickness. [0292] 3. The crystallographic structure (symmetry) and the chemical bonds of both the substrate and the growing material and how compatible they are to each other. [0293] 4. The maturity of the substrate growth technology, wafer processing, and template preparation techniques. [0294] 5. The price, quality, and availability of the particular substrate or patterned template. [0295] 5. The maturity of the growing technique for the growth of the related layers. [0296] 6. The availability, toxicity, and flammability of the chemicals used and how corrosive they are. In other words, are they environmentally friendly and how dangerous they are for humans? [0297] 7. The range and importance of the expected practical applications and, in general, are there alternative approaches for preparing these materials and, if there are any, how competitive are they to the approaches presented here.
Precursor Gases and Ternary-Forming Gases
[0298] The first precursor gas is usually hydrogen chloride (HCl) diluted to the desired extent by the carrier gas (usually H.sub.2). The role of the precursor gas is to pick-up a II or III group element (e.g. Ga, Al, Zn, Cd, Hg, etc.) from an open boat or from a bubbler, and with it to form a metal-chlorine compound, called II or III group element precursor or called just “precursor”, which is delivered to the mixing zone, making it available to participate in the growing process.
[0299] The second precursor gas, called ternary-forming gas, is usually a hydride or halide in which there is a V or VI-group element (AsH.sub.3, PH.sub.3, H.sub.2Se, SbH.sub.3, H.sub.2S, HF, NH.sub.3, etc.) diluted to the desired extent by the carrier gas (usually H.sub.2). The ternary-forming gas, which is actually the precursor of the V or VI-group element, is to be delivered to the mixing zone, making it available to participate in the growing process, reacting with the first precursor gas. We call this precursor “ternary-forming” because the chemical reactions between the precursor gas and the ternary-forming gas, which hopefully will occur on the surface of the foreign substrate rather than in the gas stream, resulting in the formation of ternary islands which, eventually coalescence to form a continuous ternary intermediate buffer layer.
[0300] Alternative names: The precursor gas may be called “precursor” of “the II or III group element” or “first precursor”. The ternary-forming gas may be called “precursor of the V or VI group element” or “second precursor”.
[0301] By demonstrating that heteroepitaxy is possible and successful at larger lattice mismatches without using a specially-patterned template in a one-step growth process (preceded by substrate pre-growth pretreatment), we have eliminated the need for growth on patterned substrates at larger mismatches, or the preliminary deposition of a thin MOCVD or MBE layer, or even the HVPE deposition of a low-temperature lower quality buffer layer. The disclosed parameters of heteroepitaxy, e.g. the thickness of the pseudomorphous growth and the periodicity of the misfit dislocations, for some particular cases have established clear criteria by which additional cases of heteroepitaxy may be deemed favorable.
[0302] The invention described herein is an innovative approach for pre-growth in-situ treatment of a substrate and the subsequent optimized thick HVPE heteroepitaxial growth on the substrate as a continuation of the pretreatment. The inventive approach applies discovered process parameters that secure a smooth transition between two different materials, and the process is flexible enough to adapt these parameters even at relatively large lattice and thermal mismatches. The evidence provided herein regarding the successful heteroepitaxial growths of materials which are disfavored according to the known prior art or conventional wisdom supports the application of the inventive process over a wide range of semiconductor pairs of materials (or pairs of a semiconductor material with another type of material - for example optical, ferroelectric, etc.) having differing degrees of lattice or thermal mismatches.
[0303] As mentioned above, the invention consists broadly of two steps: (1) pre-growth treatment of the substrate in order to initiate the replacement of V (or VI) group atoms of the substrate with V (or VI) group atoms of the layer intended to be grown, and (2) heteroepitaxial growth on the pretreated substrate. The proposed approach allows plenty of opportunity for applying different process parameters, using different pretreatment and growth temperatures and regimes of their achieving, pressures, durations of substrate pretreatment and growth, flow rates, and flow regimes, etc. The invention allows one to freely adjust the process parameters with the major goal to accommodate the substrate and the layer material or two or more subsequent layer materials to each other.
Alternative Variations
[0304] 1. Growths of one material or different phases of one material on different crystallographic orientations of another material, as the substrate and the layer material could be with the same or with different symmetry: In some cases one crystallographic plane of one material could match closer, i.e. have smaller lattice mismatch, with one or another material or different phases of one and the same material.
[0305] Example: Hexagonal GaN can be successfully grown on (111) GaAs substrates, while cubic GaN can be grown on (100) GaAs.
[0306] 2. Two or more different phases of the same material can be grown successfully on completely different materials and, wise-versa, two different materials (e.g. GaN, AlN, GaP, GaAs, etc.) can be grown on different phases of one and the same material (a cubic phase zinc blende gallium selenide Ga.sub.2Se.sub.3 and hexagonal gallium selenide ε-GaSe).
[0307] Example: Ga.sub.2Se.sub.3 can be grown on GaP and on GaAs substrates, while GaSe may be grown on GaN substrates (see also
[0308] 3. A layer from one material can be grown on a buffer layer comprising a third material that has a lattice constant between those of the substrate and the growing layer: In this case, the substrate and the layer materials are not “parenting” to the buffer layer materials.
[0309] Example: GaSb can be grown on GaAs by using not only GaSb, but also InAs or AlSb buffer layers.
[0310] 4. Multiple buffer layers from multiple distinct materials can be used for a gentle and gradual transition between two completely incompatible materials.
[0311] Examples: 1) Growth of a InSb/ ZnTe/GaSb/InP/GaAs multilayered structure - Grow, first, an intermediate InP layer (lattice constant 5.8668 Å) on a GaAs substrate (lattice constant 5.6533 Å). Then continue with the growth of a GaSb layer, (lattice constant 6.0959 Å) on the already deposited InP layer. After that grow a third, ZnTe layer, (lattice constant 6.1010 Å) on the GaSb layer, and then finish with a thick growth of InSb (lattice constant 6.4794 Å) on the already grown ZnTe layer; 2) Growth of a ZnSe/GaAs/GaP/Se multilayered structure - In this case, after the deposition of GaP on a Si substrate, the as grown GaP layer shall be pretreated in AsH.sub.3. As a second step, GaAs is grown on the pretreated GaP layer. As a third step the as grown GaAs shall be pretreated in the flow of H.sub.2Se after which growth of ZnSe shall be performed the pretreated GaAs layer.
[0312] 5. Growth of multilayered structures not only for fitting one material to another through heteroepitaxy but also in support of developments related to various applications.
[0313] Examples: 1) Successful growth of optical on electronic materials, for example, GaP on Si may have huge impact on developments in Optoelectronics; 2) Subsequent growths of multiple layers from materials with different bandgap on the same substrate could accelerate developments in both multicolor detectors and large and small surface highly efficient solar cells.
[0314] 6. Heteroepitaxial growth of low dimensional (LD) and two dimensional (2D) semiconductor materials through van der Waals heteroepitaxy where the lattice mismatch does not play that important role in contrast to the case of the classic heteroepitaxy.
[0315] Examples: 1) Growth of LD Ga.sub.2Se.sub.3 on GaAs substrates (or OP—GaAs templates); 2) Growth of 2D elemental (e.g. graphene) or compound semiconductor materials such as InSe and other chalcogenides, phosphides, arsenides, iodides, and oxides.
[0316] 7. The proposed approach is applicable not only for III-V compounds (e.g. GaAs, InAs, AlAs, GaSb, InSb, AlSb, GaP, InP, GaN, AlN, etc.) but also for II-VI (e.g. ZnSe, CdSe, CdTe, HgTe, ZnTe, CdS, etc.) and even III-VI (GaSe, InSe, GaTe) semiconductor compounds. At the same time, it is not a strict rule to grow one group’s materials, on substrates from the same group (e.g. an III-V group material on a III-V group substrate, etc.), because some mixed cases may be more favorable from point of view of, for example, lattice and thermal mismatch.
[0317] Example: The already realized case of growth of ZnSe (II-VI group) grown heteroepitaxially on the III-V group GaAs substrate.
[0318] 8. Growth of a semiconductor material on a semiconductor substrate with different crystal symmetry.
[0319] (Note: A particular case from variation 1, purposely extracted to emphasize on its importance) Example: 1) Growth of hexagonal GaN on cubic GaAs substrates
[0320] 9. Growth of semiconductor materials on oxide (including perovskites) or non-oxide (including CaF.sub.2 and BaF.sub.2) optical or non-optical materials including with different chemical composition and symmetry.
[0321] Examples: 1) GaP and GaAs can be grown on CaF.sub.2 (lattice mismatches of -0.21% and +3.50%, respectively); 2) growths of AlSb/BaF.sub.2 and GaSb/BaF.sub.2.
[0322] 10. Growth of non-ferroelectric semiconductor materials on oxide or non-oxide ferroelectric materials including on patterned ferroelectric materials with different chemical composition and symmetry.
[0323] Examples: 1) Growth of GaP on plain LiNbO.sub.3 at the large lattice mismatch of + 5.87%, which can be facilitate with the preliminary deposition of a thin MOCVD or MBE GaP layer on the substrate; 2) Growth of OP—GaP on orientation-patterned periodically pulled Lithium Niobate (PPLN).
[0324] 11. Growth of semiconductor materials on optical materials, which are neither oxide nor ferroelectric materials.
[0325] 12. The combination of one close-to-equilibrium growth technique (HVPE) with one far-from-equilibrium growth technique (MBE, MOCVD), makes it possible to grow heterostructures which appear impossible according to the prior art, including growths on common substrates, such as Si or Ge: The preliminary deposition of a thin MOCVD or MBE layer that facilitate the HVPE growth can be accepted as a pre-growth pretreatment of the substrate prior the HVPE growth
[0326] Examples: 1) Growth can start with thin MOCVD or MBE growth of GaP on Si and continue with thick HVPE growth of GaP on the GaP intermediate layer, which may be continued with thick HVPE growth of GaAs on the HVPE GaP layer, and then continued further with thick HVPE growth of ZnSe on the HVPE grown GaAs, etc.; 2) Growth can start with thin MOCVD or MBE growth of GaAs on Ge and continued with thick HVPE growth of GaAs on the GaAs intermediate thin layer, which may be continued with thick HVPE growth of GaP on the HVPE GaAs layer, and then continued further with thick HVPE growth of Ga.sub.2Se.sub.3 on the HVPE grown GaP, etc.; 3) The unfavorable effect of the large lattice mismatch during the growth of GaP or OP-GaP on a LiNbO.sub.3 substrate or on an orientation-patterned periodically pulled Lithium Niobate (PPLN) can be greatly facilitated by the preliminary deposition of a thin MOCVD or MBE GaP or GaAs (or from other suitable material) layer on the LiNbO.sub.3 substrate or OP template.
[0327] 13. Seeking for alternative chemical solutions such as using alternative precursors or alternative way to for delivering them or in-situ preparing of unavailable precursors.
[0328] Examples: 1) Due to the high vapor pressure of Zn, the boat with molten Zn during the growth of ZnSe could be replaced by prepared in advance or in-situ Zn—rich ZnCh-solution; 2) Instead of using an open boat with the molten II or III group element (e.g. Ga, Al, etc.), the supply of this chemical element can be done by using a bubbler and even another II or III group element precursor (e.g. TMG (tri-metal gallium), TMA (tri-metal aluminum, etc.); 3) If a precursor is either unavailable or unstable, or it is a subject of serious hazardous limitations (e.g. H.sub.2Te), it could be prepared in-situ within the reactor volume as a part of the very growth process.
[0329] 14. Pretreatment of the substrate surface with a non-native precursor for which is not expected to form a ternary or quaternary buffer layer if this can facilitate the following growth.
[0330] Example: If a direct HVPE growth of GaP on Si is intended, the low EPD of Si (10.sup.2 EP/cm.sup.2 -compared to the much higher numbers for GaAs and GaP provided in [0004]) is a problem. In this case, PH.sub.3 shall be the first choice of a normative precursor for pretreatment the Si substrate with the idea PH.sub.3 to pit the Si surface forming GaPSi islands which, eventually, to coalesce in a GaPSi buffer layer. The effect of PH.sub.3 on the Si substrate, however, is not strong enough. Fortunately, H.sub.2Se has a significant impact on the Si surface, etching it, i.e. making it rougher. The newly formed etch pits after the H.sub.2Se pretreatment, by providing more sites for the atom approaching the surface to adhere there, facilitate the HVPE GaP/Si growth, even without initially forming of any ternary islands and, eventually, ternary buffer layer.
[0331] 15. Replacing the in-situ pre-growth treatment of the substrate with a non-native precursor (or a mixture of precursors) can be replaced by in-situ thermal or/and chemical treatments of the substrate or through wet or dry etching (e.g. ion beam etching) before introducing the substrate in the reactor chamber for the following growth.
[0332] Example: It is known that with a certain etching procedure hydrofluoric acid (HF) can remove the oxides layers from the surface of a Si substrate. Using optimized etching procedures HF may etch deeper the Si surface as the newly formed itch pits can facilitate the following HVPE growth. In addition, a number of other chemicals and procedures can also have the desire impact on many other common substrate materials.
[0333] 16. Doping or co-doping of the substrate prior or during the first stage of growth could also be used to fit two different materials during the HVPE heteroepitaxy, bearing in mind that dopants can change dramatically the lattice constant of a material, making in this way the lattice mismatch between growing layer and substrate more acceptable. This “in-situ” procedure could be replaced by doping the substrate prior to introduce it in the reactor chamber using techniques such as ion implantation. Other techniques (e.g., ion beam spotter deposition) may also facilitate the initials stage of the HVPE growth.
[0334] While the present invention has been illustrated by a description of one or more embodiments thereof and while these embodiments have been described in considerable detail, they are not intended to restrict or in any way limit the scope of the appended claims to such detail. Additional advantages and modifications will readily appear to those skilled in the art. The invention in its broader aspects is therefore not limited to the specific details, representative apparatus and method, and illustrative examples shown and described. Accordingly, departures may be made from such details without departing from the scope of the general inventive concept.