JP5032743B2 - Formation of relaxed useful layers from wafers without a buffer layer - Google Patents

Description

  The present invention relates to the formation of a useful layer from a wafer comprising a substrate and a strained layer, each selected from crystalline materials for microelectronics, optics or optoelectronics.

  In this specification, a layer is "relaxed" if it is formed by a crystalline material having a lattice constant that is substantially the same as the nominal lattice constant of the material that is in equilibrium in the bulk state. That's it.

  Conversely, a “strained” layer is a layer made of a crystalline material whose crystal structure is elastically distorted by stretching or compression during crystal growth such as epitaxial growth, and whose lattice constant is the nominal lattice constant of this material. It is substantially different.

  Within the same wafer, the first crystalline material layer is a substrate of a second crystalline material that has different lattice constants and maintains a crystalline structure that is at least partially relaxed and / or free of an excessive number of crystal defects. Forming on top is sometimes useful or effective.

  For this purpose, it is known to insert a buffer layer between the substrate and the formed layer.

  In this structure, “buffer layer” means a transition layer in which the lattice constant of the formed layer matches the lattice constant of the substrate.

  For this purpose, this type of buffer layer may have a composition that varies gradually with thickness, and the composition of the buffer layer varies gradually between the lattice constants of the substrate and the layer formed. This is directly related to the fact that the lattice constant of the buffer layer changes gradually.

  It can also be a more complex aspect such as, for example, variable-content compositional variation, inversion of content sign, or stepwise change in composition.

  Forming such a variable composition takes time and is complicated to implement.

  Furthermore, the buffer layer is usually thicker to reduce the crystal defect density, typically 1 to a few microns.

  Therefore, the manufacture of such a buffer layer often involves time-consuming and difficult and costly processes.

Other techniques that relieve elastic strain in the formed layer, require less processing and give similar results are described in B.C. Hollander et al. Of, for _{example, "Strain relaxation of pseudomorphic Si 1} -x Ge x / Si (100) heterostructures after hydrogen or helium ion implantation for virtual substrate fabrication" (in Nuclear and Instruments and Methods in Physics Research B 175-177 (2001) 357 -367).

  The described process relates to the relaxation of a SiGe layer distorted by compression, this layer being formed on a Si substrate.

  In the technique used, hydrogen or helium ions are implanted at a predetermined depth from the surface of the strained layer into the Si substrate.

  The SiGe layer is relaxed to some extent under heat treatment by the perturbation of the crystals that are produced by the implantation of ions and located within a single layer of the Si substrate lying between the implanted region and the SiGe layer.

  Thus, with this technique, a relaxed or pseudo relaxed layer is formed in the intermediate buffer layer by simple atomic or molecular injection into the substrate.

  Thus, this technique appears to be less time consuming, easier to implement, and less expensive than techniques involving the formation of buffer layers.

  It would be interesting to use this technique to later integrate this relaxation or pseudo relaxation layer into a structure to produce devices, particularly electronics and optoelectronic devices.

  The object of the present invention is, in order to promote the implementation of this layer integration, in a first aspect, from a wafer comprising a support substrate and a strained layer, each selected from a crystal material for microelectronics, optics or optoelectronics. A method for forming a useful layer, the step of forming a perturbation region at a predetermined depth capable of forming a structural perturbation of the support substrate, and supplying energy to at least relative to the elastic strain in the strained layer. There is provided a method comprising the steps of causing a target relaxation, and removing the portion of the wafer opposite to the strained layer so that the useful layer becomes a remaining portion of the wafer.

  A reuse method which is a further aspect of the present invention is given in claims 2 to 34.

  In a second aspect, the subject of the present invention is the application of the removal method of claim 35.

  In a third aspect, the subject of the present invention is a structure with a source wafer as provided in claims 36 to 39 which provides a thin layer by peeling.

  Further features, objects and advantages of the present invention will become apparent upon reading the detailed description with reference to the accompanying drawings.

  The invention includes a source substrate where the useful layer comprises a support substrate and a strained layer on the support substrate, and a receiving substrate that forms a support member for the formation of the useful layer.

  In this specification, and generally, the “useful layer” refers to the portion of the source substrate that is formed on the receiving substrate.

  The main object of the present invention is to form a relaxation or pseudo relaxation layer on the receiving substrate from the source substrate, the useful layer being at least partially contained within the strain layer of the source substrate.

  The strained layer has no buffer layer and has been relaxed or pseudo relaxed in advance.

  Shown in FIG. 1a is a source substrate 10 of the present invention.

  The source substrate 10 includes a support substrate 1 and a strained layer 2.

  In the first embodiment of the support substrate 1, the support substrate 1 is a pseudo substrate having an upper layer made of a crystalline material (not shown in FIG. 1) such as a semiconductor material, and the upper layer is an interface between the strain layer 2 and the upper layer. And has a first lattice constant at the interface with the strained layer 2.

  The first lattice constant of the top layer is effectively the nominal lattice constant of the material from which it is formed, and therefore this material is in a relaxed state.

  Furthermore, the upper layer has a thickness sufficient to give the lattice constant to the strained layer 2 without the upper strained layer 2 significantly affecting the crystal structure of the upper layer of the support substrate 1.

  In the second embodiment of the support substrate 1, the support substrate 1 is made of only a crystalline material having a first lattice constant.

  In another effective embodiment, the support substrate 1 is a single crystal substrate.

  Whatever mode is selected as the support substrate 1, the support substrate 1 effectively has a crystal structure with a low density of structural defects such as dislocations.

  In the first embodiment of the strained layer 2, the strained layer 2 is composed of only a single crystal material such as a semiconductor material.

  The material chosen to form this strained layer 2 has a second nominal lattice constant that is substantially different from the first lattice constant.

  The formed strained layer 2 is compressed or stretched by the support substrate 1 and elastically strained, that is, has a lattice constant that is substantially different from the nominal lattice constant of the material of which it is formed. It is close to one lattice constant.

  The material chosen to form this strained layer 2 effectively has a second nominal lattice constant that is substantially greater than the first lattice constant and is therefore distorted by compression.

  The strained layer 2 more effectively has a substantially constant atomic composition.

  In the second embodiment of the strained layer 2, the strained layer 2 is composed of a plurality of layers of materials, and each single layer is composed of a crystalline material such as a semiconductor material.

  Each single layer of the strained layer 2 more effectively has a substantially constant atomic composition.

  The material thickness of the strained layer 2 that is in direct proximity to the interface with the support substrate 1 has substantially the same characteristics as the strained layer 2 of the first embodiment.

  The reduced thickness of the relaxed material of the strained layer 2 has at least one of the following effects.

  It constitutes at least part of an active layer formed on the receiving substrate to achieve certain material properties.

  It constitutes a stop layer during the selective material removal performed by the selective material removal means, in particular to protect the neighboring layers from material removal such as selective chemical etching by the etching solution.

  It can be done by selective material removal means such as selective etching, and material removal can be substantially greater than the neighboring layer that becomes a stop layer during selective material removal and is therefore protected from material removal. The

  The relaxed material may combine these functions in a single layer and may have other functions.

  In all cases, the strained layer 2 has the usual structure of strained material, but it may further include a relaxed material or layers of material having a much thinner accumulated thickness than the strained layer 2. Thus, the strained layer 2 is maintained in a generally distorted state.

  Whatever mode is selected for the strained layer 2, for example, the strained layer 2 is effective by crystal growth such as epitaxial growth using known techniques such as CVD (Chemical Vapor Deposition) and MBE (Molecular Beam Epitaxy). Formed on the support substrate 1.

  For example, in order to obtain such a strained layer 2 without an excessive number of crystal defects such as extended defects such as point defects and dislocations, the support substrate 1 and the strained layer 2 (in the vicinity of the interface with the support substrate 1) include In the crystal material to be formed, it is effective to select a crystal material having a sufficiently small difference between the first and second nominal lattice constants.

  This difference in lattice constant is typically between about 0.5% and about 1.5%, but may be higher.

  For example, in Group IV-IV materials, Ge has a nominal lattice constant of about 4.2% greater than the nominal lattice constant of Si, so SiGe containing 30% Ge is more than the nominal lattice constant of Si. It has a large nominal lattice constant of about 1.15%.

  Furthermore, it is preferred that the strained layer 2 has a substantially constant thickness so that it can have a certain intrinsic property and / or adhere to the receiving substrate 5 (as seen in FIG. 1c).

  The thickness of the strained layer 2 must further be less than the critical elastic strained thickness so that no kind of plastic relaxation or internal stress appears in the strained layer 2.

  This critical elastic strain thickness mainly depends on the material chosen for the strained layer 2 and the difference in lattice constant between the support substrate 1.

  A person skilled in the art can know the value of the critical elastic strain thickness of the material used for the strained layer 2 formed on the material used for the support substrate 1 in this field.

  Once formed, the strained layer 2 has a lattice constant substantially similar to the lattice constant of the growth substrate 1 and has an internal elastic strain due to compression or extension.

  Referring to FIG. 1 b, when the wafer 10 including the strained layer 2 is manufactured, the perturbation region 3 is formed at a predetermined depth in the support substrate 1, and the transition layer 4 is substantially formed by the perturbation region 3 and the strained layer 2. Be bound.

  The perturbation region 3 is characterized as a region having an internal strain that causes structural perturbations in the surrounding portion.

  This perturbation region 3 is actually effectively formed on the entire surface of the support substrate 1.

  This perturbation region 3 is effectively formed as a whole parallel to the surface of the support substrate 1.

  One process for forming such a weak region 3 includes implanting atomic species into the predetermined depth in the support substrate 1 with a predetermined implantation energy and a predetermined atomic species dose.

  In one method of performing the implantation, the implanted atomic species includes hydrogen and / or helium.

  Such a perturbation region 3 formed by implantation has internal strains or even crystal defects given by the implanted atomic species on the crystal lattice near this region.

  Therefore, crystal perturbation can be generated in the portion lying on the wafer 10 by these internal strains.

  In particular, for this purpose, it is advantageous to perform an appropriately parameterized process for:

  Promotes the appearance of perturbations in the transition layer 4.

  These perturbations in the transition layer 4 are moved from the fragile region 3 toward the interface with the strained layer 2.

  At least relative relaxation occurs in the strained layer 2 after the appearance and movement of the perturbation.

  Therefore, the main purpose of such treatment is to produce at least a relative relaxation of strain in the strained layer 2 in order to form the relaxed strained layer 2 '.

  Thus, if appropriately parameterized, a heat treatment is performed to cause these structural changes.

  This heat treatment must be performed at a temperature or temperatures that are well below the critical temperature above which a significant number of implanted atomic species are degassed.

  Therefore, local crystal perturbation occurs in the fragile region 3 due to internal strain.

  In particular, these perturbations shifted in the direction of the interface between the transition layer 4 and the strained layer 2 along the path defined by the specific crystal plane are transitioned because the elastic energy in the strained layer 2 is suppressed. Appears in layer 4.

  When reaching the interface between the transition layer 4 and the strained layer 2, these perturbations cause relative relaxation of elastic strain in the strained layer 2, which is mainly due to the material of the strained layer 2 and the supporting substrate. It is distortion due to lattice mismatch between the nominal lattice constants of one material.

  Such relaxation of elastic strain in the strained layer 2 is often accompanied by crystal perturbation around the strained layer 2, which is probably due to dislocations with different strains at the interface and atoms in the free plane. It is a step.

  However, relaxation of the strained layer 2 is also accompanied by inelastic crystal defects within the thickness of this layer, such as transverse dislocations.

  Appropriate processing can be performed to reduce the number of these defects.

  For example, an appropriate process for increasing the dislocation density between two limit values is performed, and these two limit values determine a dislocation density range in which at least a part of the dislocations disappears.

  For this purpose, heat treatment suitable for the material used may be effectively performed so that perturbation occurs in the dislocation layer 4.

  In all cases, the final result is a relaxed or pseudo relaxed layer 2 ′ without an intermediate buffer layer, whose nominal lattice constant is substantially different from the nominal lattice constant of the growth substrate 1.

  However, a single or more elastic strain material may be present in the relaxation strain layer 2 '.

  These multiple layers of materials are provided in the strained layer 2 before the elastic relaxation of the strained layer 2 occurs, and the lattice constant thereof is substantially different from the lattice constant of the remaining strained layer 2.

  Such multiple layers of material are naturally relaxed, for example as described in the second embodiment of the strained layer 2.

  During the overall relaxation of the strained layer 2, these multiple layers of material are distorted under the elastic strain produced by the surrounding relaxation material.

  However, since the accumulated thickness of these multiple layers of materials is considerably smaller than the thickness of the strained layer 2, the strained layer 2 generally maintains a relaxed or pseudo relaxed state after the elastic relaxation process.

  Referring to FIG. 1c, the receiving substrate 5 is placed on the surface of the wafer 10 on the relaxation strain layer 2 'side.

  The receiving substrate 5 comprises a mechanical support member that is strong enough to support the useful layer to be formed and to protect the useful layer from mechanical stresses that may be applied from the outside world.

  The receiving substrate 5 may be made of, for example, silicon, quartz, or other material.

  The receiving substrate 5 is brought into close contact with the wafer 10 and adhered by a bonding process, effectively, wafer bonding (molecular adhesion) between the receiving substrate 5 and the wafer 10.

  This deposition technique as well as other methods are described, for example, in Q.I. Y. Tong, U. It is described in the document “Semiconductor Wafer Bonding” (Science and Technology) by Gosele and Wiley.

  Appropriate pretreatment and / or thermal energy supply of each surface to be deposited is performed as needed along with the deposition.

  Therefore, for example, if heat treatment is performed during the adhesion, the adhesion is increased.

  If an adhesion layer is inserted between the wafer 10 and the receiving substrate 5, the adhesion force is further increased.

  This adhesion layer is applied to at least one of the two surfaces to be adhered.

Silicon dioxide (also called silica or SiO ₂ ) may be selected as such an adhesion layer, which can be formed by oxidative deposition, thermal oxidation or other techniques.

  Before and / or after deposition, the surface finish may be performed, for example, by etching, chemical mechanical polishing CMP, heat treatment or other planarization treatment.

  Once the receiving substrate 5 is attached, the portion of the wafer 10 opposite the relaxed strain layer 2 ′ is removed and the useful layer 6 becomes the remaining portion of the wafer 10.

  Various material removal techniques are used.

Called Smart-cut ^(R), known to those skilled in the art (and its description is found in many studies covering a technique for thinning the wafer) first material removal technique, the receiving substrate In order to form a fragile region at a depth close to the implantation depth before deposition of 5, in order to implant atomic species (such as hydrogen or helium ions) and to peel the wafer 10 into two parts in the fragile region Energy supply to vulnerable areas, such as heat and / or mechanical processing or other energy supply.

  In order to further weaken the fragile region, it is effective to heat-treat the wafer 10 during or after the implantation.

  In the first method of carrying out this material removal, a fragile region is formed between the support substrate 1 and the relaxation strain layer 2 'or in the relaxation strain layer 2'.

  In the second method of carrying out this material removal, a fragile region is formed in the support substrate 1.

  The fragile region may be formed during or after the formation of the perturbation layer 3.

  In the special case of the second method for carrying out the material removal, and in the case of forming the transition layer 4 by forming the perturbation region 3, atomic seed implantation with a predetermined implantation energy and a predetermined atomic seed dose. The fragile region may be formed at substantially the same position as the perturbation region 3 by substantially the same technique as described above.

  In this special case, the fragile region can be formed in substantially the same time as the perturbation region 3 is formed.

  Before or after implantation, the wafer 10 may be further subjected to a heat treatment having two functions of weakening the further weakened region and further relaxing the strained layer 2.

  Therefore, a weak region as if having a dual function of weakening the support substrate 1 and relaxing the strained layer 2 is formed therein.

  A second material removal technique includes the formation of at least one porous layer by atomic species implantation or anodization by other porous forming techniques such as those described in document EP0849788A2, and the wafer 10 in the fragile layer. Energy supply to the fragile layer, such as mechanical processing or other energy supply to peel into two parts.

  In the first method for carrying out this material removal, a fragile layer is formed between the support substrate 1 and the relaxation strain layer 2 'or in the relaxation strain layer 2'.

  In the second method of carrying out this material removal, a fragile layer is formed in the support substrate 1.

  In order to form the fragile layer in the support substrate 1, the porous layer is effectively formed on the thin piece of single crystal material, and the nonporous crystalline material layer having substantially the same lattice constant as the thin piece is formed on the porous layer. The substrate 1 is grown and consists of a flake, a porous layer and a non-porous Si layer.

  A substantial portion of the wafer 10 is quickly removed by the first and second non-limiting material removal techniques.

  With these techniques, the removed portion of the wafer 10 can be used again in other processes such as the process of the present invention.

  Accordingly, the strained layer 2 and, in some cases, the support substrate 1 and / or other layers are preferably re-formed after the surface of the support substrate 1 is polished.

  A third known technique consists of a chemical and / or chemical mechanical material removal process.

  For example, any selective etching of the material of the donor wafer 10 to be removed may be performed using an “etchback” type process.

  In this technique, the wafer 10 is etched “from behind”, that is, from the free surface of the support substrate 1, in order to finally leave a portion of the wafer 10 to be left on the receiving substrate 5.

  Wet etching using an etching solution that removes the material may be performed.

  In order to remove the material, dry etching such as plasma etching or sputtering may be performed.

  The etching process or etching processes may be performed purely chemically, electrochemically or photoelectrochemically.

  The etching process or the plurality of etching processes may be performed before or after the mechanical attack on the wafer 10 such as lapping, polishing, mechanical etching, or sputtering of atomic species.

  The etching process or the plurality of etching processes may optionally be accompanied by a mechanical attack such as polishing in which the action of a mechanical abrasive is combined in the CMP process.

  Therefore, the portion of the wafer 10 to be removed is completely removed by pure chemical means or chemical mechanical means.

  In the first method of performing the material removal, the etching process or the plurality of etching processes are performed so that only at least a part of the relaxed strain layer 2 ′ is left on the wafer 10.

  In the second method of removing the material, the etching process or the plurality of etching processes are performed so that a part of the supporting substrate and the relaxed strain layer 2 ′ are left on the wafer 10.

  In particular, the third technique can maintain the high surface quality and constant thickness of the strained layer 2 obtained during crystal growth of the strained layer 2.

  These three techniques are proposed as examples in this document, but are not limited to these, and the invention extends to any type of technique that can remove material from a wafer according to the process of the invention. is there.

  Whatever technique is selected from these three techniques or other known techniques, any surface treatment technique such as selective chemical etching, CMP polishing, heat treatment, or other planarization treatment is applied to the active layer. Done effectively.

  In the special case where a part of the support substrate 1 remains after one of these three techniques is performed, and when the remaining part of the support substrate 1 is not required, It is effective to perform a finishing process including selective etching of the remaining portion.

  In the latter special case, a moderately strained layer 2 ′ is obtained in which the thickness is constant and / or the surface finish is good, and many defects such as work-hardened regions that sometimes occur in mechanical finishing can be prevented.

  A part of the relaxed strain layer 2 ′ having a constant thickness and a good surface finish can be obtained by performing selective etching on the relaxed strain layer 2 ′ serving as a stop layer for the etching performed.

  The latter two finishing steps including selective etching are particularly effective when it is desired to finally obtain a very thin relaxed strain layer 2 '.

  In all cases, the final result is a structure 20 with a receiving substrate 5, an active layer 6, and optionally an insertion adhesion layer.

  In the first method of material removal, only at least a portion of the relaxed strain layer 2 'is left.

  Therefore, the active layer 6 comprises at least a part of the relaxed strain layer 2 '.

  In the second method of carrying out this material removal, only a part of the support substrate 1 and the relaxation strain layer 2 'are left.

  Therefore, the active layer 6 comprises the remaining portion of the support substrate 1 and the relaxed strain layer 2 '.

  In this case, the remaining portion of the support substrate 1 is at least partially distorted by the lower relaxation strain layer 2 '.

  In one particular method using structure 20, one or more crystal growth processes are performed on structure 20.

  Once the final structure is obtained, for example, a finishing process such as a finishing process such as an annealing process may be optionally performed in order to enhance the adhesion interface between the useful layer 6 and the receiving substrate 5.

  One special method using structure 20 and any structure 20 may cause one or more epi layers to be grown on wafer 10.

  The remainder of this document gives some examples of the materials that make up the structure obtained by carrying out the process of the invention.

  In particular, layers of Si and SiGe type materials are considered.

  As described above, SiGe containing 30% Ge has a nominal lattice constant that is approximately 1% greater than the nominal lattice constant of Si.

  A strained layer 2 of SiGe having a specific Ge concentration and formed on a Si support substrate is suitable for carrying out the process of the present invention.

  A preferred process for forming the useful layer of this invention is illustrated in the following examples.

  Example 1: Referring to FIG. 1, this shows that a wafer 10 has a Si support substrate 1 and a specific Ge concentration, and the thickness is from the critical end-of-strain thickness (described above). And a strained layer 2 made of thin SiGe.

  The strained SiGe layer 2 has a typical Ge concentration greater than 15%.

The strained SiGe layer 2 effectively also has a defect density such as dislocations less than about 10 ⁷ cm ⁻² .

  Typical thicknesses of the strained layer 2 containing 15% Ge and the strained layer 2 containing 30% Ge are about 250 nm and about 100 nm, respectively, and are therefore below the critical strain termination thickness, respectively.

  Referring to FIG. 1b, a perturbation region 3 is formed in the Si support substrate by implantation with atomic species such as H or He.

  The range of H or He implantation energy used is typically between 12 and 25 KeV.

The implanted H or He dose is typically between 10 ¹⁴ and 10 ¹⁷ cm ⁻² .

Therefore, for example, for the strained layer 2 containing 15% Ge, it is preferable to use H with an energy of about 25 KeV and a dose of about 3 × 10 ¹⁶ cm ⁻² as an implant.

Therefore, for example, for the strained layer 2 containing 30% Ge, it is preferable to use He with an energy of about 18 KeV and a dose of about 2.0 × 10 ¹⁶ cm ⁻² as an implant.

  The implantation depth of atomic species is typically between 50 nm and 100 nm.

  After the perturbation region 3 is formed, an appropriate heat treatment is performed, in particular, the perturbation in the transition layer 4 is moved, and dislocations are eliminated in the relaxed strain layer 2 '.

  Heat treatment is performed in an inert atmosphere.

  However, the heat treatment may be performed in another atmosphere, for example, an oxidizing atmosphere.

  Thus, this type of wafer 10 is subjected to a heat treatment, typically at a temperature between 600 ° C. and 1000 ° C., typically for a time between about 5 minutes and about 15 minutes.

Further details of such experimental techniques can be found in B.C. Hollander et al. Has been reviewed by, in _{particular, "Strain relaxation of pseudomorphic Si 1} -x Ge x / Si (100) heterostructures after hydrogen or helium ion implantation for virtual substrate fabrication" (in Nuclear and Instruments and Methods in Physics Research B 175-177 ( 2001) 357-367).

In other cases forming the perturbation region 3 of the present invention, hydrogen or helium is implanted at a dose of about 10 ¹⁷ cm ⁻² .

Based on the formation of the fragile region using the Smart-cut ^(R) process, both the perturbation region 3 and the fragile region can be formed by the predetermined dose amount.

  This fragile region thus has the dual function of creating a crystal perturbation in the upper transition layer 4 and creating an internal stress that is weak enough for the wafer 10 to delaminate into two parts after energy application.

  In one particular embodiment, the subsequent heat treatment has the dual function of relieving strain in the strained layer 2 and further weakening the fragile region.

  Whatever implantation method is selected to form the transition layer 4, the SiGe strained layer 2 is at least partially relaxed to form the SiGe relaxed strained layer 2 ′.

  Referring to FIG. 1c, the receiving substrate 5 attached to the wafer 10 may be formed of any material such as silicon or quartz.

A SiO ₂ adhesion layer is effectively inserted between the relaxed strain layer 2 ′ and the receiving substrate 5, finally (see FIG. 1 d) to form a structure 20 of SGOI or Si / SGOI type. The insulator is an SiO ₂ layer.

  Referring to FIG. 1d, one or more known techniques for material removal may be implemented.

In particular, selective etching of Si is performed using an etching solution that is substantially selective to SiGe, and such a solution is at least KOH, NH ₄ OH (ammonium hydroxide), TMAH, EDP or HNO. A solution comprising at least one compound of ₃ or HNO ₃ , HNO ₂ , H ₂ O ₂ , HF, H ₂ SO, as currently under discussion, as described on page 9 of document WO 99/53539 ₄ , a solution in which chemical agents such as CH ₃ COOH and H ₂ O are mixed.

  In the first state, the latter selective etching can remove the remaining portion of the support substrate 1 that should be removed from the relaxed strained layer 2 ′ so that it remains on the structure 2. It consists of a relaxation strain layer 2 '.

  In the second state, a Si etching stop layer is provided in the support substrate 1 to protect the Si layer on the stop layer from etch-back type selective chemical etching, and the active layer 6 in this case is a relaxed strain layer 2 ′. And a Si layer on the stop layer.

  The stop layer may be made of, for example, SiGe, where the selective chemical etching uses one of the above etching solutions.

  Referring to FIG. 1d, a structure 20 comprising a receiving substrate 5 and an active layer 6 is obtained.

  Depending on the removal method used, the active layer 6 comprises at least a part of the SiGe relaxation strain layer 2 ′ and optionally a Si layer and a remaining part of the support substrate 1.

  Example 2: Referring to FIG. 2, this example relates to a wafer 10 that is substantially the same as Example 1, but further comprising a Si layer substantially relaxed on the strained SiGe layer.

  The strained layer 2 is thus composed of a strained SiGe layer 2A and a relaxed Si layer 2B.

  If this strained layer 2 is thicker than this, it is thinner than the critical thickness of SiGe where SiGe relaxes.

  The strained layer 2A has substantially the same characteristics as the strained SiGe layer 2 of Example 1.

  The thickness of the relaxed Si layer 2B is much thinner than the entire thickness of the strained layer 2, and the strained layer 2 maintains the entire strained structural characteristics.

  The thickness of the relaxed Si layer 2B is several tens of nanometers.

  The implementation of the removal process is substantially the same as in Example 1.

  Formation of transition layer 4 and further effective heat treatment, substantially the same as in Example 1, elastically relaxes strained layer 2A to form relaxed strained layer 2A ′ (not shown) and ( The strain relaxation layer 2B ′ is elastically strained to form the strain relaxation layer 2B ′ (not shown), and the strain relaxation layer 2B ′ has the effect of having a lattice constant close to that of the lower relaxation SiGe.

  When the wafer 10 is attached to the receiving substrate 5 with or without an intermediate adhesion layer in the strain relief layer 2B ', the material can be removed using one or more of the known techniques described above.

  In the first method of material removal, it is desirable to leave at least a portion of the relaxed strained layer 2A 'and the strained Si layer 2B', and material removal is substantially the same as described in Example 1.

  The final result is a structure 20 (as in FIG. 1d) comprising a receiving substrate 5 and an active layer 6, which is at least part of a strained Si layer 2B ′ and a relaxed SiGe layer 2A ′ (further Depending on the removal method used, it is optionally composed of the Si layer or the remaining part of the support substrate 1).

  In the second method of performing the removal process, it is desirable to leave at least a portion of the strained Si layer 2B ′, the material removal is substantially the same as described in Example 1, and the relaxed SiGe layer 2A ′ is further removed. An additional step of removal is added.

For this purpose, SiGe may be selectively etched using a solution for selective etching of SiGe with respect to Si, for example, a solution containing HF: H ₂ O ₂ : CH ₃ COOH (selectivity ratio of about 1: 1000). it can.

  Therefore, in the second method for carrying out this treatment, the relaxed SiGe layer 2A 'becomes a sacrificial layer.

  In order for the relaxed SiGe layer 2A ′ to be a sacrificial layer, structural defects such as dislocations with different strains disappear, which may be included on the surface, and before the deposition, the transition layer After the perturbation in 4 propagates, it may appear near the interface with the transition layer 4.

  The relaxed SiGe layer 2A 'thus protects the strained Si layer 2B' from structural defects that may be caused by the results of certain relaxation methods used in the process of the present invention.

  Therefore, this sacrificial technique is particularly suitable for finally obtaining the strained Si layer 2B ′ having very few structural defects.

  The final result is a structure 20 (such as the structure of FIG. 1d) comprising a receiving substrate 5 and an active layer 6 consisting of a strained Si layer 2B '.

  Example 3 Referring to FIG. 3, this example relates to a wafer 10 that is substantially the same as Example 2 but further includes a substantially relaxed SiGe layer on the relaxed Si layer.

  The strained layer 2 therefore comprises a strained SiGe layer 2A, a relaxed Si layer 2B, and a relaxed SiGe layer 2C.

  If this strained layer 2 is thicker than this, it is thinner than the critical thickness of SiGe where SiGe relaxes.

  The strained layer 2A has substantially the same characteristics as the strained SiGe layer 2 of Example 1.

  It is effective to select a thickness of the layer 2A that is greater than or equal to a typical thickness such that structural defects appearing in the vicinity of the interface with the transition layer 4 after the perturbation in the transition layer 4 propagates.

  Such a strained SiGe layer 2A thus protects the relaxed Si layer 2B and the strained SiGe layer 2C from any structural defects while the strained layer 2 is totally relaxed.

  Therefore, this sacrificial technique is particularly suitable for finally obtaining the Si layer 2B having very few structural defects.

  Since the thickness of the relaxed Si layer 2B is much thinner than the entire thickness of the strained layer 2, the strained layer 2 maintains the overall strained structure characteristics.

  The thickness of the relaxed Si layer 2B is about several tens of nanometers.

  The strained SiGe layer 2C has substantially the same characteristics as the strained SiGe layer 2A.

  However, the strained SiGe layer 2C is effectively thicker than the strained SiGe layer 2A.

  The strained SiGe layer 2C becomes a single major part of the strained layer 2 in certain special situations.

  The implementation of the removal process is substantially the same as in Example 2.

  The formation of the transition layer 4 and the further effective heat treatment as in Example 1 elastically relaxes the strained layer 2A to form a relaxed strained layer 2A ′ (not shown), and strain relaxation (not shown). The relaxation layer 2B is elastically strained to form the layer 2B ′, the strain relaxation layer 2B ′ has a lattice constant close to that of the lower relaxation SiGe, and the relaxation strain layer 2C ′ (not shown) is In order to form it, there is an effect that the strained layer 2C is elastically relaxed.

  When the wafer 10 is attached to the receiving substrate 5 with or without an intermediate adhesion layer in the relaxed strain layer 2C ', the material can be removed using one or more of the known techniques described above.

  In the first method of performing material removal, it is desirable to leave at least a portion of the relaxed strained layer 2A ′, the strained Si layer 2B ′, and the relaxed SiGe layer 2C ′, the material removal being described in Example 1. Is substantially the same.

  The final result is a structure 20 (as in FIG. 1d) comprising a receiving substrate 5 and an active layer 6, which comprises a relaxed SiGe layer 2C ′, a strained Si layer 2B ′ and a relaxed SiGe layer. It consists of at least a part of 2A ′ (and optionally the remaining part of the Si layer or the support substrate 1 depending on the removal method used).

  In the second method of performing the removal process, it is desirable to leave at least a portion of the strained Si layer 2B 'and only the relaxed SiGe layer 2C', and material removal is substantially the same as described in Example 2.

  The final result is a structure 20 (as in FIG. 1d) comprising a receiving substrate 5 and an active layer 6, which consists of at least a portion of a strained Si layer 2B ′ and a relaxed SiGe layer 2C ′. .

  In the third method for carrying out the removal process, it is desirable to leave at least a part of the relaxed Si layer 2C ′, the material removal is substantially the same as in the second method, and the strained Si layer 2B ′. An additional step of removing is added.

For this purpose, a solution containing at least one compound of KOH, NH ₄ OH (ammonium hydroxide), TMAH, EDP or HNO ₃ H, or HNO ₃ , H ₂ O ₂ , HF, H ₂ SO ₄ , a combination of chemical agents such as H ₂ SO ₂ , CH ₃ COOH, and H ₂ O can be used to selectively etch the strained Si layer 2B ′ using a solution such as the solution currently under consideration. .

  Since the relaxed SiGe layer 2C ′ is an etching stop layer, this method finally yields a layer having a uniform thickness with particularly little surface roughness.

  Thus, in particular, a high quality and very thin layer can be obtained.

  The final result is a structure 20 (as in FIG. 1d) comprising a receiving substrate 5 and an active layer 6, the active layer 6 comprising a relaxed SiGe layer 2C ′.

  One special method of using the structure 20 and any structure 20 to form a multilayer structure, epitaxial growth of SiGe layers or strained Si layers, or stacking of SiGe layers or strained Si layers One or more epi layers may be grown on the wafer 10, such as the epi layer by

  The semiconductor layer indicated in this document contains other components, for example carbon whose carbon concentration in the semiconductor layer is substantially lower than or equal to 50%, or in particular lower than or equal to 5%. Can be added.

  The present invention is not limited to strained SiGe layer 2 and Si support substrate, but can be used by the process of the present invention (two-component, three-component, four-component, or more) III-V or II-VI It applies to other materials such as atomic genera.

  The structure finally obtained after peeling is not limited to the SGOI, SOI or Si / SGOI type.

It is a figure which shows the process process of this invention. It is a figure which shows the process process of this invention. It is a figure which shows the process process of this invention. It is a figure which shows the process process of this invention. It is a figure which shows the wafer of this invention from which a useful layer is peeled. It is a figure which shows the other wafer of this invention from which a useful layer is peeled.

Claims (27)

A method of forming a useful layer from a wafer comprising a support substrate and a strained layer, each selected from a semiconductor material, and no buffer layer between the support substrate and the strained layer ,
A perturbation region is formed at a predetermined depth of the support substrate by implantation of atomic species, and the perturbation region has an internal strain that can cause a structural perturbation in the support substrate, and the perturbation region and the strain layer The substrate portion is a transition layer, and
Providing energy to cause at least relative relaxation of the elastic strain in the strained layer to form a relaxed strained layer, wherein the relative relaxation of the strained layer occurs beyond the transition layer;
A receiving substrate is attached to the relaxation strain layer of the wafer, a portion of the wafer is removed from a free surface side of the support substrate, and the useful layer comprising the relaxation strain layer and the remaining portion of the support substrate is received. Forming on a substrate ;
Forming at least one layer on the useful layer;
With
The method wherein the relaxed strain layer comprises a chemical etch stop layer and the removing step comprises performing a selective chemical etch of the relaxed strain layer to remove a portion covering the stop layer.   The method for forming a useful layer according to claim 1, wherein the atomic species implanted to form the perturbation region comprises at least partially hydrogen and / or helium.   The method for forming a useful layer according to claim 1, wherein the energy supplied during the supplying step comprises thermal energy for promoting relaxation of strain in the strained layer.   The method for forming a useful layer according to any one of claims 1 to 3, wherein the attaching step is performed after providing an adhesion layer on at least one of the two surfaces to be adhered.   The method of forming a useful layer according to claim 4, wherein the adhesion layer is made of silica.   The method for forming a useful layer according to any one of claims 1 to 5, wherein at least one of the two surfaces to be adhered is subjected to a planarization treatment as a surface finishing treatment.   The method for forming a useful layer according to any one of claims 1 to 6, further comprising a heat treatment for enhancing adhesion. Before the removing step, comprising the step of forming a fragile region in the support substrate,
The method for forming a useful layer according to claim 1, wherein the removing step includes a step of supplying energy into the fragile region in order to peel the useful layer from the wafer.   9. The method of forming a useful layer according to claim 8, wherein the fragile region is formed by implanting atomic species.   10. The method of forming a useful layer according to claim 9, wherein the implanted atomic species at least partially comprises hydrogen and / or helium. The method for forming a useful layer according to claim 9 or 10, wherein the fragile region and the perturbation region are located at the same point in the wafer. The method of forming a useful layer according to claim 11, wherein the fragile region is formed by the same means as the perturbation region .   9. The method for forming a useful layer according to claim 8, wherein the fragile region is formed by making a layer in the wafer porous.   14. The removing step includes a step of performing selective chemical etching of a part of the support substrate in the vicinity of the relaxation strain layer, and the relaxation strain layer forms a stop layer of the etching. The method of forming the useful layer in any one.   15. The method for forming a useful layer according to claim 1, wherein the useful layer comprises at least a part of the relaxation strain layer.   The method for forming a useful layer according to any one of claims 1 to 15, further comprising a planarization treatment as a surface finishing treatment performed on the surface of the useful layer after the removing step.   The method for forming a useful layer according to any one of claims 1 to 16, wherein at least one layer formed on the useful layer has a lattice constant distorted by the relaxation strain layer.   The method for forming a useful layer according to claim 1, wherein the support substrate is made of silicon, and the strained layer is made of silicon germanium.   At least one layer formed on the useful layer comprises at least one of relaxed or pseudo relaxed silicon germanium having a germanium concentration equal to the germanium concentration of the strained layer and strained silicon. A method for forming a useful layer according to claim 18.   20. The method for forming a useful layer according to claim 1, wherein the receiving substrate is made of silicon or quartz.   The method for forming a useful layer according to any one of claims 1 to 19, wherein at least one of the layers used in the method further includes carbon having a carbon concentration of less than or equal to 50%.   The method for forming a useful layer according to any one of claims 1 to 21, wherein at least one of the layers used in the method further comprises carbon having a carbon concentration of less than or equal to 5%.   A method for producing a semiconductor-on-insulator structure using the method for forming a useful layer according to any one of claims 1 to 22, wherein the semiconductor layer of the structure includes the useful layer. how to. A wafer used in the method for forming a useful layer according to any one of claims 1 to 22,
It said support substrate having a first lattice constant, further comprising a the relaxed strained layer whole having a second lattice constant is relaxed or pseudo relaxation,
Does not include the buffer layer
The wafer, wherein the relaxation strain layer comprises a chemical etching stop layer.   25. The wafer of claim 24, further comprising a perturbation region in the support substrate. The wafer according to claim 24, further comprising a fragile region in the support substrate . 27. A structure comprising a receiving substrate attached to the relaxed strain layer of a wafer according to any of claims 24 to 26.

Images