US 20070051299 A1
The present invention provides a method of forming a strained semiconductor layer. The method comprises growing a strained first semiconductor layer, having a graded dopant profile, on a wafer, having a first lattice constant. The dopant imparts a second lattice constant to the first semiconductor layer. The method further comprises growing a strained boxed second semiconductor layer having the second lattice constant on the first semiconductor layer and growing a sacrificial third semiconductor layer having the first lattice constant on the second semiconductor layer. The method further comprises etch annealing the third and second semiconductor layer, wherein the third semiconductor layer is removed and the second semiconductor layer is relaxed. The method may further comprises growing a fourth semiconductor layer having the second lattice constant on the second semiconductor layer, wherein the fourth semiconductor layer is relaxed, and growing a strained fifth semiconductor layer having the first semiconductor lattice constant on the fourth semiconductor layer. The method controls the surface roughness of the semiconductor layers. The method also has the unexpected benefit of reducing dislocations in the semiconductor layers.
16. A method comprising:
forming over a strained semiconductor layer having a lattice constant, a second strained semiconductor layer also having the lattice constant;
forming a sacrificial semiconductor layer having a second lattice constant on said second semiconductor layer;
annealing the sacrificial semiconductor layer and the second strained semiconductor layer in an etching ambient to remove the sacrificial semiconductor layer and reduce strain in a second semiconductor layer;
forming a relaxed semiconductor layer having the lattice constant on the second semiconductor layer; and
forming a third strained semiconductor layer having the lattice constant on the relaxed semiconductor layer.
17. The method of
18. The method of
19. The method of
providing a substrate having the second lattice constant; and
forming the strained semiconductor layer with a graded dopant profile on the substrate.
20. The method of
21. The method of
22. The method of
23. The method of
24. The method of
25. The method of
26. A method of controlling the surface roughness of a strained semiconductor material comprising:
forming a first strained semiconductor layer having a graded dopant profile on a wafer having a first lattice constant, the dopant imparting a second lattice constant to the first semiconductor layer;
forming a second strained semiconductor layer having said second lattice constant on the first semiconductor layer;
forming a sacrificial semiconductor layer having the first lattice constant on the second semiconductor layer; and
etch annealing the third and second semiconductor layer, wherein said third semiconductor layer is removed and said second semiconductor layer is relaxed.
27. The method of
28. The method of
29. The method of
30. The method of
31. The method of
32. The method of
33. The method of
34. A method of reducing dislocations in a semiconductor material comprising relaxing a strained semiconductor layer having a first lattice constant by etch annealing to remove an overlying sacrificial semiconductor layer having a second lattice constant, the strained semiconductor layer formed over a second strained second semiconductor layer having a second lattice constant imparted by a graded dopant profile.
35. The method of
36. The method of
37. The method of
38. The method of
This application claims the benefit of U.S. Provisional Application No. 60/484,181; filed Jun. 30, 2003, and is a continuation-in-part of U.S. application Ser. No. 10/264,393, filed Oct. 4, 2002.
Embodiments of the present invention relate to controlling the growth and morphology of surface roughness during an annealing and relaxation process of strained films, and more particularly to decreasing surface crystalline dislocations during a relaxation process.
In the conventional art, many semiconductor device fabrication processes utilize planar surfaces. Furthermore, as semiconductor fabrication technology progresses, increasing carrier mobility and decreasing lattice dislocation density become increasingly critical. Improving device yields by reducing dislocations provides for improved manufacturing efficiencies and cost.
In the conventional art, a silicon layer is used as the active device medium upon which semiconductor devices are fabricated. Single-crystal silicon has a specific carrier mobility value that is fundamental to the material. The mobility value is a key parameter in many active semiconductor devices. Often, it is desired to enhance or increase the device carrier mobility value to increase the switching speed and therefore the performance of the fabricated devices such as transistors. Because of the many fundamental and specific advantages in utilizing silicon as the semiconductor material, it is highly desirable to adopt methods to enhance silicon mobility instead of utilizing higher mobility materials that are harder to process such as Germanium or Gallium Arsenide.
One practical method of enhancing silicon mobility is by straining the silicon layer. By placing the active silicon under tension, significantly higher mobility resulting in higher device switching speed and drive currents can be achieved.
A method of generating such tensile strained silicon involves growing the silicon layer epitaxially above a relaxed silicon germanium film of a specific composition. This effect occurs because the silicon lattice constant, about 5.43 Angstroms, is smaller than the lattice constant of a fully relaxed silicon germanium alloy film. Such alloys can be engineered to have a lattice spacing linearly varying from 5.43 Angstroms (100% silicon) to 5.65 Angstroms (100% germanium). For the pure germanium film, the lattice spacing is about 4% larger than pure silicon. Thus for example, a Si0.75Ge0.25 alloy (25% germanium content) would have a lattice constant about 1% larger than silicon.
The strained silicon film could therefore be advantageously fabricated by epitaxially growing the device silicon film on a relaxed silicon Germanium (SiGe) alloy film of the requisite composition.
A fundamental complication of this mobility enhancement approach is the requirement of a relaxed SiGe film. If the SiGe film is grown onto a base silicon wafer, the film will first grow in a lattice-matched manner as a compressive layer. This means that the SiGe alloy will be compressed to the natural silicon lattice spacing and will be strained. Since the function of the alloy film requires a relaxation of the compressive strain, there must be a step where the SiGe alloy is relaxed to its unstrained state. Such a step necessarily introduces numerous dislocations in the SiGe layer to accommodate the lattice spacing and volume increase. The film also usually “buckles” and roughens significantly during this relaxation process.
The major parameters characterizing a practical relaxed SiGe alloy film include the amount that the film has been relaxed from its strained state (i.e. 50% relaxation would mean that the film has relaxed half of its strain), the roughness of the film, and the dislocation defect density that would be affecting the subsequent growth of the strained silicon device film.
The surface dislocation density is a critical parameter affecting the electrical properties of semiconductor materials since they are highly dependent upon crystalline defects. Dislocations can comprise insertion of an extra half-plane of atoms into a regular crystal structure, displacement of whole rows of atoms from their regular lattice position, and/or displacement of one portion of the crystal relative to another portion of the crystal. Dislocations present on the device layer can tend to short-circuit p-n junctions and also scatter electrons in a uniform n-type crystal, impeding their motion and reducing their mobility. Dislocations also cause highly localized distortion of the crystal lattice leading to the formation of “trapping” sites where the recombination of positive (holes) and negative (electrons) carriers is enhanced. This may cause, for example, the electrons from the n-p-n transistor emitter to recombine with holes in the p-type base regions before they can be collected at the n-type collector region, reducing the transistor current gain. This electron “lifetime” may be significantly reduced by recombination when as few as one out of 1011 atoms/cm3 of silicon are removed from their normal lattice sites. Although some dislocations can be removed from a semiconductor material by thermal annealing, many dislocations are permanent and thermally stable. Many of the relaxation approaches are therefore tuned to minimize the defect density of the type that can be translated to the device layer and cause device performance degradation, failure and yield losses.
In one method according to the conventional art, the SiGe alloy is grown with a slowly varying grade from 0% germanium to the required alloy composition at a sufficiently low temperature to grow a dislocation free initial film and through subsequent annealing, the slow gradient helps to accommodate film relaxation through the generation of dislocations that are buried within the SiGe layer. This technology is explained in Legoues & al. (U.S. Pat. No. 5,659,187 “Low defect density/arbitrary lattice constant heteroepitaxial layers”). To limit the production of dislocations threading to the surface, the SiGe grade is usually less than 2% composition increase per 1000 Angstroms of SiGe film growth. This shallow gradient approach is lower in productivity due to its relatively thick SiGe layer composition and may require numerous growth/anneal cycles to achieve roughness and dislocation goals.
In yet another method according to the conventional art, the surface roughness or the SiGe alloy layer can be reduced using a chemical mechanical polishing (CMP) process such as taught by Fitzgerald (U.S. Pat. No. 6,291,321 “Controlling threading dislocation densities in Ge on Si using graded GeSi layers and planarization” and U.S. Pat. No. 6,107,653, “Controlling threading dislocation densities in Ge on Si using graded GeSi layers and planarization”). CMP utilizes a combination of vertical force between a wafer and an abrasive pad as well as a chemical action of a slurry, to polish the surface of the wafer to a highly planar state. The roughness of the resulting semiconductor surface can typically be reduced to approximately 1 Angstrom RMS when measured by an Atomic Force Microscope (AFM). However, CMP is relatively costly as a result of the slurry and the amount of time it takes to perform the process. Furthermore, the CMP process does not generally reduce the dislocation density in the wafer. Finally, this linear growth/anneal/CMP sequence is costly as it requires numerous sequential process and wafer handling steps.
Another method uses miscut wafers to help the grown film to relax as much as possible and accommodate the lattice mismatches. See for example Fitzgerald & al. (U.S. Pat. No. 6,039,803, “Utilization of miscut substrates to improve relaxed graded silicon-germanium and germanium layers on silicon”) that teaches the improvement of using base wafers having 1 to about 8 degrees of miscut from a true  orientation to help grow a less defective, relaxed layer of a second semiconductor material. Although the base substrate miscut can improve the relaxed defect density to some extent, the improvements are generally considered insufficient for leading edge applications.
Referring now to
As depicted in
As depicted in
Thus, the conventional art is disadvantageous in that planarizing processes are relatively costly and time-consuming processes. The conventional art also suffers from relatively high levels of dislocations. A better and less costly approach that can fully relax strained SiGe alloy films while controlling surface roughness and dislocation defect levels is highly desirable.
A method for etch annealing a semiconductor layer is disclosed. In addition to its significant effect in controlling the roughness increases during the relaxation process, the method has the unexpected benefit of substantially reducing dislocations. The reduced dislocation density is advantageous in that carrier mobility and yield are increased.
In one embodiment, a method of forming a strained semiconductor layer is provided. The method comprises growing a strained first semiconductor layer, having a graded dopant profile, on a wafer, having a first lattice constant. The dopant imparts a second lattice constant to the first semiconductor layer. The method further comprises growing a strained boxed second semiconductor layer having the second lattice constant on the first semiconductor layer and growing a sacrificial third semiconductor layer having the first lattice constant on the second semiconductor layer. The method further comprises etch annealing the third and second semiconductor layer, wherein the third semiconductor layer is removed and the second semiconductor layer is relaxed. The method may further comprises growing a fourth semiconductor layer having the second lattice constant on the now relaxed second semiconductor layer, wherein the fourth semiconductor layer is relaxed, and growing a strained fifth semiconductor layer having the first semiconductor lattice constant on the fourth semiconductor layer. The method controls the surface roughness of the semiconductor layers. The method also has the added benefit of reducing dislocations in the semiconductor layers.
The present invention is illustrated by way of example and not by way of limitation, in the figures of the accompanying drawings and in which like reference numerals refer to similar elements and in which:
Reference will now be made in detail to the embodiments of the invention, examples of which are illustrated in the accompanying drawings. While the invention will be described in conjunction with the preferred embodiments, it will be understood that they are not intended to limit the invention to these embodiments. On the contrary, the invention is intended to cover alternatives, modifications and equivalents, which may be included within the spirit and scope of the invention as defined by the appended claims. Furthermore, in the following detailed description of the present invention, numerous specific details are set forth in order to provide a thorough understanding of the present invention. However, it will be obvious to one of ordinary skill in the art that the present invention may be practiced without these specific details. In other instances, well-known methods, procedures, components, and circuits have not been described in detail as not to unnecessarily obscure aspects of the present invention.
In one implementation, the strained graded first semiconductor layer comprises a strained SiGe alloy having a graded doping profile. The stained graded SiGe layer of approximately 4000-20,000 angstroms (Å) is formed by an epitaxial deposition process. The doping profile of the germanium (Ge) increases from approximately 0% at the wafer to 25% at the surface of the strained SiGe layer. In one implementation, the epitaxial deposition process is performed in a hydrogen-chloride (HCl) ambient. The as-grown strained graded SiGe layer has a roughness of approximately 2 Angstroms root-mean-square (RMS), and less than approximately 1 dislocations/cm2.
At 320, a strained boxed second semiconductor layer is grown on the strained graded first semiconductor layer. The strained boxed second semiconductor layer is grown having the second lattice constant. The term “boxed” is intended to indicate that the doping profile of the layer is substantially constant. The roughness of the strained boxed second semiconductor layer is low because the as-grown semiconductor is close to being or is fully strained and has not been subjected to thermal cycles that would have started the relaxation process. In one implementation, the strained boxed second semiconductor layer comprises a strained SiGe layer having a constant Ge doping profile. The strained boxed SiGe layer of approximately 500-5000 Å is formed by an epitaxial deposition process. The doping profile of the Ge is approximately 25% throughout the first strained boxed SiGe layer. In one implementation, the epitaxial process if performed in an HCl ambient.
At 330, a sacrificial third semiconductor layer is grown on the strained boxed second semiconductor layer. The sacrificial third semiconductor layer is grown having the first lattice constant. In one implementation, the sacrificial third semiconductor layer comprises a silicon (Si) layer. The sacrificial Si layer of approximately 100-300 Å is formed by an epitaxial deposition process. It is also appreciated that the growth of the sacrificial third semiconductor layer may be omitted.
At 340, an etch anneal is performed upon the sacrificial third semiconductor layer and the boxed second semiconductor layer. The surface is “etch annealed” utilizing, for example, an epitaxial chamber subjecting the wafer to a high temperature anneal in an etching ambient. The etch anneal etches away the sacrificial third semiconductor layer, formed at 330, and relaxes the boxed strained second semiconductor layer, formed at 320. The etch anneal may also relax the graded strained first semiconductor layer, formed at 310. This etch anneal produces a smoother surface than a simple anneal performed in a non-etching ambient.
A similar process has been shown to smooth unstrained films in a process referred to as an “epi-smoothing” process. The epi-smoothing process is disclosed in U.S. Pat. No. 6,287,941, granted Sep. 11, 2001, entitled “Surface Finishing of SOI Substrates Using an EPI Process,” which is incorporated by reference herein. Since the etching process applied to relaxing strained films controls the overall roughness rather than smooths the surface, it will hereinafter called “etch annealing”.
In one implementation, the etch annealing process comprises subjecting the sacrificial Si layer to an etchant including a halogen bearing compound such as HCl, HF, HI, HBr, SF6, CF4, NF3, CCl2F2, or the like. The etch annealing process is performed at an elevated temperature of 700-1200° C., or greater. For example, the surface of the sacrificial Si layer and the strained boxed SiGe layer is exposed to a HCl containing gas, at an elevated temperature of 700-1200° C., in an etch annealing process, such that:
Accordingly, the etchant removes the sacrificial Si layer. Furthermore, the concurrent use of the etchant and a temperature sufficient to relax the strained boxed SiGe or the strained boxed SiGe and strained graded SiGe layers has been found to help reduce or eliminate the generation of dislocations with a concurrent reduction in relaxation roughening of the surface. This favorable effect is believed linked to a reduction of the stress inducing cycloidic cusp tips present during a non-etch anneal (H. Gao & W. D. Nix, “Surface Roughening of Heteroepitaxial Thin Films”, Annu. Rev. Mater. Sci. 1999, 29, pg. 173-209). In the work by Gao and Nix, it is explained that the strain caused by the lattice mismatch drives the generation of an undulating profile on the surface that has periodic sharp cusp tips that favor the creation of dislocations at these highly stressed locations. The concurrent etching process during lattice relaxation is believed to significantly blunt or round the cusp tips that reduce the stress concentration and thus reduces the surface dislocation density by affecting its creation kinetics. The surface roughening is also disfavored by the etch ambient.
The etch annealing process may be performed at a range of elevated temperatures that would favor the concentration of the dislocations away from the surface to relax the lattice structure of the boxed SiGe layer. The thermal treatment may be from a resistance heater, RF heater, high intensity lamps, or the like. The thermal treatment means should be capable of heating the semiconductor material at a rate of approximately 10-20° C./sec, or more.
Because the strained graded SiGe layer, strained boxed SiGe layer, and the sacrificial Si layer are performed within an epitaxial reactor and the etch annealing is also performed within the same system, repetition of these steps is straightforward and the general economy of the process can be fully appreciated since no cleans, external anneals or CMP planarization steps are needed.
The etch annealing process removes strained semiconductor material in a manner that blunts sharpening (roughening) features that form on the surface upon film relaxation. The rate of etching is a function of time, temperature, and the etchant type and concentration. Therefore, controlling these parameters during the etch annealing process controls the amount of etching. The etch annealing process is performed until the process reduces surface roughness by approximately fifty percent or more compared to an anneal without the etchant. Thus, the etch annealing process acts to control surface roughening during film relaxation sufficient for subsequent semiconductor device fabrication processes. The method also provides the added benefit of reducing dislocations by up to two orders of magnitudes or more.
Furthermore, unlike convention CMP processes that are limited to removing a few tens of nano-meters or less, the etch annealing process can be used to remove as much as a few hundreds of nano-meters or more of semiconductor material.
It is also appreciated that the etch anneal may be performed partway through the growth of the strain graded first semiconductor layer formed at 310. Furthermore, the etch anneal may be performed more than once at during various processes, such as 310, 320 and/or 330.
At 350, a boxed fourth semiconductor layer is grown on the relaxed boxed second semiconductor layer. The boxed fourth semiconductor layer is grown having the second lattice constant and has a relaxed structure. In one implementation, the boxed fourth semiconductor layer comprises a relaxed SiGe layer having a constant Ge doping profile. The boxed SiGe layer of approximately 1000-10,000 Å is formed by an epitaxial deposition process. The doping profile of the Ge is approximately 25% throughout the boxed SiGe layer. In one implementation, the epitixial process if performed in an HCl ambient.
At 360, a fifth semiconductor layer is grown on the boxed fourth semiconductor layer. The fifth semiconductor layer is grown having the first lattice constant. Accordingly, the fifth semiconductor layer is formed as a strained semiconductor layer. In one implementation, the fifth semiconductor layer comprises a silicon (Si) layer. The second Si layer of approximately 100-200 Å is formed by an epitaxial deposition process.
The fifth semiconductor layer may be utilized for fabricating additional layers thereupon or device regions therein. The reduced dislocation density of the resulting strained fifth semiconductor layer advantageously results in higher carrier mobility. The higher carrier mobility improves characteristics of devices, such as field effect transistors, bipolar transistors, and the like.
Referring now to
As depicted in
The etch annealing process yields a less costly alternative than the conventional art by allowing for a usable strained silicon device layer without costly external planarization steps. Furthermore, one skilled in the art would expect that etch annealing would not affect the generation of the dislocation process. However, the etch annealing process results in the unexpected benefit of decreasing the dislocation density to approximately 105 dislocations/cm2, as compared to an anneal made without an etch ambient of approximately 107 dislocations/cm2. The etch annealing process is also highly effective in fully relaxing the film, rendering it suitable as a base to grow a strained silicon device film.
The resulting strained fifth semiconductor layer 460 can be utilized for fabricating semiconductor layers thereupon or device regions therein. The reduced dislocation nature of the semiconductor layer advantageously results in higher carrier mobility. The high carrier mobility in the fifth semiconductor layer 460 improves characteristics of the devices formed therein.
The foregoing descriptions of specific embodiments of the present invention have been presented for purposes of illustration and description. They are not intended to be exhaustive or to limit the invention to the precise forms disclosed, and obviously many modifications and variations are possible in light of the above teaching. The embodiments were chosen and described in order to best explain the principles of the invention and its practical application, to thereby enable others skilled in the art to best utilize the invention and various embodiments with various modifications as are suited to the particular use contemplated. It is intended that the scope of the invention be defined by the Claims appended hereto and their equivalents.