SiGeC semiconductor crystals and the method producing the same

Abstract

A B-doped Si1−x−yGexCy layer 102 (where 0

Description

TECHNICAL FIELD

The present invention relates to an SiGeC semiconductor crystal applicable to a bipolar transistor or a field-effect transistor and a method for producing the same.

BACKGROUND ART

The present invention relates to an SiGeC semiconductor crystal, which is a Group IV mixed crystal semiconductor, and a method for producing the same.

Conventionally, attempts have been made at fabricating a semiconductor device which operates faster than known Si semiconductor devices by stacking a Si layer and a semiconductor layer containing Si as a main ingredient thereof so as to form a heterojunction. Si 1−x Ge x and Si 1−x−y Ge x C y , which are mixed crystal semiconductors each formed using a Group IV element that is in the same group as Si, are expected as candidates for a material for forming a heterojunction with the Si layer. Particularly, as for an Si 1−x−y Ge x C y mixed crystal semiconductor that is formed from three different elements, its band gap and lattice constant can be independently controlled by adjusting its composition, resulting in greater flexibility in semiconductor device design. Therefore, the Si 1−x−y Ge x C y mixed crystal semiconductor has attracted much attention. For example, a lattice matching between Si 1−x−y Ge x C y and Si crystals can be made by properly adjusting the composition of Si 1−x−y Ge x C y . A heterobarrier (band offset) can be also formed on both a conduction band edge and a valence band edge around the interface of the heterojunction between the Si and Si 1−x−y Ge x C y layers by properly adjusting the composition of Si 1−x−y Ge x C y . Japanese Unexamined Patent Publication No. 10-116919, for example, discloses a field-effect transistor in which a two-dimensional electron gas serves as a carrier and which can operate at a high speed by utilizing a heterobarrier formed on the conduction band edge near the interface of Si/SiGeC layers.

Meanwhile, for producing Si 1−x−y Ge x C y mixed crystals, use is now made of, for example, a chemical vapor deposition (CVD) process in which respective source gases of elements Si, Ge and C are dissolved so as to induce epitaxial growth of those elements on the Si or SiGe layers, or a molecular beam epitaxy (MBE) process in which respective source solids of the elements are heated and vaporized so as to induce crystal growth of the elements. In order to use an Si 1−x−y Ge x C y layer as a part of a semiconductor device, the Si 1−x−y Ge x C y layer is required to be doped with an impurity for generating a carrier, which will be a dopant so as to control the conductivity and specific resistance of the Si 1−x−y Ge x C y layer. In the Si 1−x−y Ge x C y layer, boron (B) and phosphorus (P) are used as a p-type dopant and an n-type dopant, respectively, in many cases. It is well known that the conductive type and specific resistance of a growth layer can be adjusted by doping the layer with a dopant during crystal growth.

Problems to be solved

FIG. 4 is a graph indicating the result of an experiment conducted by the inventors for the purpose of consideration as to doping of an Si 1−x−y Ge x C y layer and shows how the specific resistance of the Si 1−x−y Ge x C y layer changed depending on the C content thereof. The Si 1−x−y Ge x C y layer as a sample from which the data was collected is as-grown one obtained by being epitaxially grown by a CVD process with the use of Si 2 H 6 , GeH 4 and SiH 3 CH 3 as respective source gases of elements of Si, Ge, and C and B 2 H 6 as a source gas of boron (B) which is a p-type impurity (dopant) (i.e., through in-situ doping). In this experiment, the flow rates of Si 2 H 6 and GeH 4 and the temperature of the Si 1−x−y Ge x C y layer during the epitaxial growth thereof were kept constant and only the flow rate of SiH 3 CH 3 was changed. As shown in FIG. 4 , as for the sample having a C content of 0.45% or less, even when the C content was changed, the specific resistance of the sample stayed almost constant and relatively low. In contrast, as for the Si 1−x−y Ge x C y layer having a C content of 1.6%, the specific resistance thereof remarkably increased. That is to say, it was clearly shown that clearly shown that the specific resistance of the Si 1−x−y Ge x C y layer which had been epitaxially grown by this method increased to the level at which the layer would be no longer suitable for use as an active region of a semiconductor device (e.g., a channel region of FET, a base layer of a bipolar transistor).

FIG. 5 is a graph indicating the result of the secondary ion mass spectroscopy on a sample formed basically in the same method as the sample from which the data shown in FIG. 4 was collected and shows how the boron concentration of the Si 1−x−y Ge x C y layer changed depending on the C content thereof. This is an experiment that was conducted to examine whether the specific resistance shown in FIG. 4 was affected by the boron concentration, because the doping efficiency of boron slightly changes, depending upon the C content of the Si 1−x−y Ge x C y layer, when boron is introduced into the Si 1−x−y Ge x C y layer by an in-situ doping process. Note that the sample from which the data of FIG. 5 was collected is not identical to the sample from which the date of FIG. 4 was collected. As shown in FIG. 5 , the B concentration of the Si 1−x−y Ge x C y layer did not largely depend on the C content thereof. In addition, as also shown in FIG. 5 , the B concentration of the Si 1−x−y Ge x C y layer tended to increase as the C content of the Si 1−x−y Ge x C y layer increased. That is to say, it was confirmed that the increase in the specific resistance of the sample having a B concentration of 1.6% shown in FIG. 4 was not caused due to the shortage of B concentration.

The inventors then assumed that an increase in specific resistance of regions having a relatively high C content in the Si 1−x−y Ge x C y layer would be caused by B having not sufficiently been activated during an epitaxial growth associated with an in-situ doping. Conventionally, with an in-situ doping of a dopant during an epitaxial growth of a semiconductor layer (e.g., a Si layer or an Si 1−x−y Ge x C y layer) using a CVD process, an annealing process for activating the dopant is considered as unnecessary because the dopant is activated concurrently with the epitaxial growth of the semiconductor layer, unlike an impurity doping by an ion implanting process. As shown in FIG. 4 , the Si 1−x−y Ge x C y layer, with its C content of 0.45% or less, has a relatively low specific resistance in the as-grown state. In such a case the Si 1−x−y Ge x C y layer as-grown can be therefore used for an active region of a semiconductor device. However, it is likely that when the C content of the Si 1−x−y Ge x C y layer increases, some phenomenon causing problems that cannot be solved by the conventional technology will appear. Particularly, it is empirically known that various properties of the layer are largely changed when the C content of the Si 1−x−y Ge x C y layer increases to over 1%. Therefore, around 1% of C content can be considered to be the critical value where the specific resistance of the Si 1−x−y Ge x C y layer starts increasing, although the data of FIG. 4 is not enough to confirm that.

DISCLOSURE OF INVENTION

It is an object of the present invention to provide an Si 1−x−y Ge x C y semiconductor crystal applicable as an active region of a semiconductor device and a method for producing the same by taking measures to activate boron (B), particularly for an Si 1−x−y Ge x C y layer having a relatively high degree of carbon (C) content which goes just over 1%.

A method for producing an SiGeC semiconductor crystal according to the present invention includes the steps of: a) epitaxially growing an SiGeC semiconductor crystal on a substrate, the SiGeC semiconductor crystal doped with a carrier generating impurity on a substrate the SiGeC semiconductor having a composition represented by Si 1−x−y Ge x C y (where 0<x<1, 0.01≦y<1); and b) performing an annealing process to activate the carrier generating impurity in the SiGeC semiconductor crystal.

In this method, it was empirically confirmed that the specific resistance of an SiGeC layer could be reduced. Conventionally, it has been considered that an in-situ doped impurity is activated during an epitaxial growth. However, this idea is not applicable to the SiGeC layer and thus it is assumed that even the in-situ doped impurity can be activated enough by an annealing process.

The temperature in the annealing process is within the range from 700° C. to 1000° C., both inclusive, and thereby the impurity can be particularly effectively activated.

If the epitaxially growing step includes a CVD process using, as a source, a hydride which is made of at least one material selected from the group consisting of Si, Ge, C and B, the present invention is of great significance.

A SiGeC semiconductor crystal of the present invention includes two or more alternately stacked sets of: an Si 1−z Ge z (where 0<z<1) layer which contains a carrier generating impurity; and an Si 1−w C w (where 0.01≦w<1) layer which has a carrier generating impurity higher in concentration than the Si 1−z Ge z layer, and functions as an SiGeC semiconductor crystal which has a composition represented by Si 1−x−y Ge x C y (where 0<x<1, 0.01≦y<1).

Thus, the Si 1−z Ge z layer and the Si 1−w C w layer are unitized to function as a single SiGeC semiconductor crystal. At the same time, in the Si 1−z Ge z layer, by utilizing activation of the carrier generating impurity as-grown without any special process performed, the SiGeC semiconductor crystal with a relatively low specific resistance which is suitable for an active region of a semiconductor device can be achieved.

If the Si 1−z Ge z layer and the Si 1−w C w layer are each smaller in thickness than the case where a discrete quantum state is generated in the layers, they can more reliably be unitized to function as a single SiGeC semiconductor crystal.

Specifically, each of the Si 1−z Ge z layer and the Si 1−w C w layer preferably has a thickness of 1.0 nm or less.

BRIEF DESCRIPTION OF DRAWINGS

FIGS. 1A and 1B are cross-sectional views schematically illustrating respective process steps for producing B-doped Si 1−x−y Ge x C y semiconductor crystals according to a first embodiment of the present invention.

FIG. 2 is a graph showing the data on how the specific resistance of the B-doped Si 1−x−y Ge x C y semiconductor crystals changed with the temperature in RTA according to the producing method of the present invention.

FIG. 3 is a cross-sectional view schematically illustrating the structure of the Si 1−x−y Ge x C y semiconductor crystals according to a second embodiment of the present invention.

FIG. 4 is a graph indicating the dependency of the specific resistance of a B-doped Si 1−x−y Ge x C y layer upon the C content thereof.

FIG. 5 is a graph indicating the dependency of the B concentration of the Si 1−x−y Ge x C y layer upon the C content thereof, which have been obtained as the result of a secondary ion mass spectroscopy.

BEST MODE FOR CARRYING OUT THE INVENTION

Hereinafter, embodiments of an SiGeC semiconductor crystal according to the present invention and a method for producing the same will be described with reference to the accompanying drawings.

First Embodiment

FIGS. 1A and 1B are cross-sectional views schematically illustrating respective process steps for producing a B-doped Si 1−x−y Ge x C y semiconductor according to a first embodiment of the present invention. FIG. 2 is a graph indicating the data on how the specific resistance of B-doped Si 1−x−y Ge x C y semiconductor crystals changed with the temperature in RTA according to the producing method of the present invention.

First, in the process step shown in FIG. 1A , a B-doped Si 1−x−y Ge x C y layer 102 with a thickness of approximately 300 nm is epitaxially grown, on a Si substrate 101 whose principal surface is the {001} surface, by an ultra high vacuum chemical vapor deposition (UHV-CVD) process. In the meantime, Si 2 H 6 , GeH 4 and SiH 3 CH 3 are used as respective source gases of elements Si, Ge and C which constitute the Si 1−x−y Ge x C y layer, and B 2 H 6 is used as a source gas of boron (B) which is a p-type impurity (dopant) (i.e., these gases are in-situ doped). In this case, the respective flow rates of Si 2 H 6 and GeH 4 and the temperature of Si 1−x−y Ge x C y layer during an epitaxial growth (at approximately 490° C.) are kept constant and only the flow rate of SiH 3 CH 3 is changed. The growth pressure is approximately 0.133 Pa (=1×10 −3 Torr) and the growth temperature is 490° C. According to the results of the evaluation of the composition of the B-doped Si 1−x−y Ge x C y layer 102 by an X-ray diffraction analysis, the contents of Si, Ge and C thereof were 82.5%, 13.2% and 1.6%, respectively. Also, According to the result of the evaluation of the Si 1−x−y Ge x C y layer 102 by an secondary ion mass spectroscopy, the B concentration was 2.6×10 18 atoms cm −3 .

Next, in the process step shown in FIG. 1B , the Si 1−x−y Ge x C y layer 102 is annealed by rapid thermal annealing (RTA), thereby forming a B-doped Si 1−x−y Ge x C y crystalline layer 103 . Here, annealing is made at, e.g., approximately 950° C. at highest, for an approximately 15 seconds duration at the highest temperature, in an atmosphere of nitrogen (N 2 ) at 1 atmospheric pressure.

FIG. 2 , is a graph showing the annealing temperature (highest RTA temperature) dependency of the specific resistance of the B-doped Si 1−x−y Ge x C y crystalline layer 103 when the RTA temperature (highest temperature) was changed. Note that the data shown in FIG. 2 was collected under the condition in which the highest RTA temperature was changed from 700° C. through 1050° C., both inclusive, and the annealing time was constantly kept at 15 seconds. In the FIG. 2 , the data marked with (&Circlesolid;) indicates the specific resistance of the B-doped Si 1−x−y Ge x C y crystalline layer 103 . The data marked with (▪) indicates the annealing temperature dependency of the specific resistance of a B-doped Si 1−x Ge x layer (Ge content of 13%) which was formed for comparison with the inventive crystalline layer. As shown in the FIG. 2 , the specific resistance of the B-doped Si 1−x−y Ge x C y crystalline layer 103 (&Circlesolid;) monotonously decreases as the annealing temperature increases in the range from 700° C. to 900° C., both inclusive. When the annealing temperature is in the range from 900° C. to 1000° C., both inclusive, the specific resistance of the B-doped Si 1−x−y Ge x C y crystalline layer 103 stays approximately constant and then when the annealing temperature becomes over 1000° C., the specific resistance starts increasing. It can be seen that the B-doped Si 1−x−y Ge x C y crystalline layer 103 has a smaller specific resistance at any temperature when annealed after having been epitaxially grown than when being in the as-grown state. On the other hand, the specific resistance of the B-doped Si 1−x Ge x layer (▪) stays low and approximately constant whether or not the Si 1−x Ge x layer is subjected to the RTA. That is to say, it is not significant at all to anneal the B-doped Si 1−x Ge x layer for activation after the layer has been epitaxially grown.

It can be understood from the above description that the specific resistance of an Si 1−x−y Ge x C y layer can be reduced by forming the Si 1−x−y Ge x C y layer through epitaxial growth associated with in-situ doping and then annealing the Si 1−x−y Ge x C y layer. Specifically, as for the Si 1−x−y Ge x C y layer which has been epitaxially grown concurrently with in-situ doping to contain a 1% or more dopant therein, it has a problem of high specific resistance. However, annealing the Si 1−x−y Ge x C y layer after its epitaxial growth can prevent an increase in the specific resistance.

As shown in FIG. 2 , annealing (RTA) particularly in the range from 700° C. to 1020° C., both inclusive, provides the effect of ensuring that the specific resistance of the Si 1−x−y Ge x C y layer is reduced.

Furthermore, it is also shown that when the Si 1−x−y Ge x C y layer is annealed at a temperature in the range from 900° C. to 1000° C., both inclusive, the effect of reducing the specific resistance thereof is remarkable.

In this embodiment, the case of using boron (B) as a dopant for generating a carrier has been described. However, the present invention is not limited to this embodiment but is applicable to the case of using phosphorus as a dopant for generating a carrier.

In a paper (“Epitaxial growth of Si 1−x−y Ge x C y film on Si(100) in an SiH 4 -GeH 4 -CH 3 SiH 3 reaction”, A. Ichikawa et. al., Thin Solid Film 369(2000) 167-170) published after a prior application (Japanese Application No. 2000-086154) based on which this application claims for priority had been filed, it was reported that a P-doped Si 1−x−y Ge x C y layer which had been formed though epitaxial growth associated with in-situ doping had an increased specific resistance even in the as-grown state, but it was not mentioned how the specific resistance could be prevented from increasing. That is to say, the reduced specific resistance of the Si 1−x−y Ge x C y layer having a relatively high C content was not achieved. In contrast, the present invention is of great significance in that, according to the present invention, the specific resistance of an impurity-doped Si 1−x−y Ge x C y layer can be reduced by a relatively easy process.

When a device such as a field-effect transistor and a bipolar transistor is fabricated using Si 1−x−y Ge x C y crystals, the above-described annealing process also provides the same effects with the use in combination with another annealing process such as, e.g., a process for activating source and drain regions or a process for activating polysilicon electrodes.

In this embodiment, the Si 1−x−y Ge x C y layer is epitaxially grown by a UHV-CVD process. However, the present invention is not limited to this embodiment, the same effects can be obtained for an Si 1−x−y Ge x C y crystalline layer which is formed by another process including a low pressure chemical vapor deposition (LP-CVD) process.

In this embodiment, the annealing process is effected by RTA. However, the same effects can also be achieved by any other annealing process including a furnace annealing process.

Second Embodiment

Next, the structure of a B-doped Si 1−x−y Ge x C y crystalline layer according to a second embodiment of the present invention and a method for producing the same will be described.

FIG. 3 is a cross-sectional view schematically illustrating the superlattice structure of a semiconductor according to the embodiment of the present invention. Epitaxially grown on a Si substrate 111 by a UHV-CVD process was a superlattice structure 104 in which a B-doped SiGe layer 113 (1 nm thick) containing boron (B) as a carrier generating impurity and having a composition represented by Si 1−z Ge z (where 0<z<1l), and a non-doped SiC layer 112 (1 nm thick) containing a carrier generating impurity lower in concentration than the B-doped SiGe layer 113 and having a composition represented by Si 1−w C w (where 0.01≦w<1) have been stacked 150 times. In the meantime, Si 2 H 6 , GeH 4 , and SiH 3 CH 3 were used as respective source gases of elements Si, Ge, and C, and B 2 H 6 as a source gas of boron (B) which is a p-type dopant (i.e., these gases were in-situ doped). In this case, the growth pressure was approximately 0.133 Pa (=1×10 −3 Torr) and the growth temperature was 490° C. As a result of evaluation of the composition of the B-doped SiGe layer 113 by an X-ray diffraction analysis, the Ge content was 26.4%. As a result of evaluation of the composition of the non-doped SiC layer 113 by the X-ray diffraction analysis, the C content was 3.2%. As a result of evaluation of the composition of the B-doped SiGe layer 112 by a secondary ion mass spectroscopy, the B concentration was 5.2×10 18 atoms cm −3 . The non-doped SiC layer 113 was not intentionally doped with B, but contains B at a very low concentration because of residue or diffusion of the impurity doping gas.

The superlattice structure 104 formed by stacking the Si 1−z Ge z /Si 1−w C w layers according to this embodiment has such a thin superlattice structure that an SiC layer and an SiGe layer exhibit no quantum effect, and thus no discrete quantum state will be generated. Properties of the SiC layer 112 and the SiGe layer 113 are averaged (in other word, united) and thus the whole superlattice structure 104 functions as a single Si 1−x−y Ge x C y layer (where 0<x<1, 0.01≦y<1). In the Si 1−x−y Ge x C y layer (Si 1−z Ge z /Si 1−w C w short-period superlattice), the average concentrations of Si is approximately 85.2%, the average concentrations of Ge is approximately 13.2%, the average concentrations of C approximately 1.6% and the average concentration of B is approximately 2.6×10 18 cm −3 , and therefore the Si 1−x−y Ge x C y layer can be considered to have the same composition as the Si 1−x−y Ge x C y layer shown in FIG. 1 .

Therefore, the same effects as described in the first embodiment can be achieved according to this embodiment. Furthermore, this embodiment has the advantage that in the process steps of producing the SiGeC semiconductor crystals of this embodiment, an annealing process is not required unlike in the first embodiment and therefore reduction in resistance can be realized in an as-grown state.

In the superlattice structure 104 , only the SiGe layer 113 is doped with B and the SiC layer 112 contains very little B. As shown in FIGS. 2 and 4 , B in the SiGe layer 113 is activated as a dopant in the as-grown state. A minute amount of C could be mixed into the SiGe layer 113 in the course of production. However, the existence of C up to approximately 0.45% of the SiGe layer 113 would not increase the specific resistance of the SiGe layer as shown in FIG. 4 .

In this embodiment, the SiC layer 112 and the SiGe layer 113 are both 1 nm in thickness and a superlattice structure obtained by stacking these two layers 150 times is used. However, the present invention is not limited to this embodiment. By stacking an Si 1−x−y Ge x C y layer which includes a carrier generating impurity at a high concentration and an Si 1−w C w layer which contains almost no carrier generating impurity, the carrier generating impurity can be reliably activated and therefore a multilayer structure which functions as a single Si 1−x−y Ge x C y layer can be achieved. Note that, when a quantum state is generated in the Si 1−z Ge z layer or the Si 1−w C w layer, some other characteristics may possibly appear in the layers. Therefore, the Si 1−z Ge z layer and the Si 1−w C w layer are preferably formed as thin as no quantum state will be generated.

Particularly, if the Si 1−z Ge z layer and the Si 1−w C w layer are 1 nm or less in thickness, this ensures that each layer has no quantum state. Therefore, the thickness of each layer is preferably 1 nm or less. Note that the thicknesses of the Si 1−z Ge z and Si 1−w C w layers may be different from each other.

In this embodiment, the Si 1−x−y Ge x C y layer is formed by epitaxially growing the Si 1−z Ge z and Si 1−w C w layers in an alternate manner by the UHV-CVD process. However, the present invention is not limited to this embodiment and the same effects can be achieved in the Si 1−x−y Ge x C y layer formed from the Si 1−z Ge z and Si 1−w C w layers which have been formed by another growing process including a low-pressure chemical vapor deposition (LP-CVD) process.

INDUSTRIAL APPLICABILITY

This invention is applicable to a field-effect transistor or a bipolar transistor which include an Si/SiGeC or SiGe/SiGeC hetero structure.

Claims

1. An SiGeC semiconductor crystal, comprising two or more alternately stacked sets of:an Si1−zGez (where 0<z<1) layer which contains a carrier generating impurity; and an Si1−wCw (where 0.01≦w<1) layer which has a carrier generating impurity lower in concentration than the Si1−zGez layer, wherein the SiGeC semiconductor crystal functions as an SiGeC semiconductor crystal which has a composition represented by Si1−x−yGexCy (where 0≦x<1, 0.01<y<1).

2. An SiGeC semiconductor crystal according to claim 1, characterized in that the Si1−zGez layer and the Si1−wCw layer are each smaller in thickness than the case where a discrete quantum state is generated in the layers.

3. An SiGeC semiconductor crystal according to claim 1, characterized in that each of the Si1−zGez layer and the Si1−wCw layer has a thickness of 1.0 nm or less.