JP3168574B2 - Crystal growth method for high concentration compound semiconductor - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Overview] The present invention relates to lattice matching in hetero-epitaxial growth.
For the purpose of realizing more accurate lattice matching, in the epitaxial growth of the compound semiconductor of the present invention, a combination of the compound AB and the impurity C to be doped is used.
For example, assuming a compound of _CB , a lattice constant a _VC of a compound _CB crystal is calculated using an actually measured value of a lattice constant of compound AB doped with a known concentration of impurity C, and a compound assumed separately. When Vegard's law is applied between A'A "-B and the compound CB, the compound A'A" has the same lattice constant as the substrate crystal when the concentration of C is a desired impurity concentration. calculating the lattice constants a _VT of -B, defines the composition of the compound A'A "-B to have a lattice constant a _VT, and supplies the raw material such that the impurity concentration becomes the predetermined value in said set formed It is configured to perform epitaxial growth.

[Industrial applications]

The present invention relates to an epitaxial growth method of a multi-component compound semiconductor, and more particularly to an epitaxial growth method for correcting a lattice constant shift when an impurity concentration is increased and performing lattice matching.

When the emitter / base junction of the bipolar transistor is a homojunction, it is difficult to set a high impurity concentration in the base without lowering the injection efficiency because only a barrier exists due to a difference in impurity concentration between E and B. Therefore, there is a problem that the base resistance cannot be reduced.

Heterojunction bipolar transistor (HBT) is mainly III
A hole barrier is provided by making the emitter / base junction a heterojunction (in the case of an npn transistor). Even if the impurity concentration of the base is increased, the presence of the barrier causes the electron barrier. Is maintained at a high injection efficiency.

To increase the operation speed of the bipolar transistor, it is necessary to reduce the thickness of the base layer. At this time, in order to keep the base resistance from increasing, it is necessary to increase the impurity concentration of the base layer. As described above, this is difficult for a homojunction transistor, whereas for HBT, carrier injection is performed even if the impurity concentration is as high as possible, making it suitable for thinning the base layer and increasing the speed. ing.

When a heterojunction is provided in an element such as a transistor, it is generally required that the lattice constants be matched so as not to cause distortion at the junction due to lattice mismatch. For that purpose, the lattice constant and the band gap are adjusted to predetermined values by adjusting the composition of the compound semiconductor.

The combination of compound semiconductor materials commonly used for HBTs is AlGaAs, as well as InP / InGaAs, InAlAs / InGaA
s, InGaP / GaAs and the like. In this case, the substrate crystal for epitaxial growth is GaAs (lattice constant 0.5653 nm) or InP
(Lattice constant 0.5869 nm) is used, and the composition of the ternary compound lattice-matched to InP is In _0.53 Ga _0.47 As, In _0.52 Al
_0.48 As, In _0.51 Ga _0.49 P
To match.

Considering the case where impurities are doped in a combination in which the composition is set and lattice-matched in this way, if the impurity concentration is about normal, the change in the lattice constant is slight,
Lattice mismatch does not occur, but when impurity doping of 10 ¹⁹ cm ⁻³ or more, which is called ultra-high concentration doping, is performed, the lattice constant changes and the lattice matching with the substrate crystal is shifted.

Therefore, merely making the impurity concentration of the base of the HBT very high does not make the special production of the HBT satisfactory.

[Problems to be solved by conventional technology and invention]

Although the concentration of impurities in a semiconductor material has a limit known as solid solubility, in reality, the growth conditions during epitaxial growth, particularly the non-equilibrium state impurity concentration limit caused by the growth temperature is dominant.

For example, a paper by JLLievin & F. Alexandre (“Ultra-
high doping levels of GaAs with beryllium by molec
ular beam epitaxy ", Electronics Letters, Vol.21, No.1
0, pp. 413-415, 9th, May, 1985), when doping GaAs with Be, the maximum concentration of Be is 4 × 10 ¹⁹ cm ⁻³ when the growth temperature Ts is 600600 ° C. However, when Ts is 450450 ° C., this becomes 2 × 10 ²⁰ cm ⁻³ .

Also, when doping InGaAs with a composition lattice-matching to InP with Be, when Ts ≒ 500 ° C., the maximum concentration is 2 × 10 ¹⁹ cm ^−3.
However, it is also known that a higher concentration of doping can be achieved by lowering Ts. In this regard, RAHamm, MBPanish, RNNottenburg, YKChen &
DAHumphrey; “Ultrahigh Be doping of Ga _0.47 In
_0.53 As by low-temperature molecular beam epitax
y ", Appl.Phys.Lett., Vol. 54, No. 25, pp. 2586-2588, 19t
h, June, 1989).

If the doping is excessively performed beyond the concentration limit as described above, the epitaxial growth surface does not become a mirror surface but becomes a rough surface which indicates island growth. In this situation, a p-type impurity such as Be cannot be completely substituted at a lattice position, and dopant diffusion occurs to lower the doping efficiency.

With such recognition, some efforts have been made to realize an HBT having an ultra-highly doped base layer.

One of them is to change the growth temperature.
Lievin et al. Formed an AlGaAs / GaAs-based HBT. When growing a GaAs base layer, Ts was set at 450 ° C.
s = 620 ° C. (JLLievin, CDCheval-lier, G.
Leroux, J. Dangla, F. Alexander & D. Ankri; “Ga _0.72 Al
_0.28 As / Ga _0.99 Be _0.01 As heterojunc-tion bipolar tra
nsistors with ultra low baser esistance grown by M
BE ", Proc.11th Int'1S ymp.on GaAs and Related Compo
unds, Karuizawa, Japan, 1985, Int.Phys.Conf.Ser.No.79,
pp. 595-600).

In the case of InP / InGaAs-based HBT, 5 × 10 ²⁰ cm at Ts = 365 ° C.
There is also an example in which 1 × 10 ²⁰ cm ^-3 Be doping is performed on an InGaAs base layer by a low-temperature growth technique capable of ^-3 doping (RAHamm et al, Appl. Phys. Lett., 1989, supra).

As described above, ultra-high concentration doping is possible by lowering the growth temperature, but this method requires a process of changing the growth temperature Ts when epitaxially growing the emitter layer following the base layer.

This is because Ts is optimally about 700 ° C. in epitaxial growth of AlGaAs as an emitter layer, and it is difficult to obtain a high-quality epitaxially grown film at 600 ° C. or less. The reason that low-temperature growth of AlGaAs is difficult is that the migration rate of Al is low. In addition, InAlAs / InGaAs-H
The optimum Ts of InAlAs, which is the emitter of BT, is about 500 ° C.
If s is less than 400 ° C, it is not good. This is also due to the low migration speed of Al.

For this reason, after growing the base layer as described above, the growth process must be interrupted to adjust the temperature, and then the emitter layer must be grown. However, if there is an interruption time during the growth, the substrate is constantly exposed to the gas remaining in the growth environment during this time, and if the substrate is contaminated by CO, CO ₂ , O _2, etc. Defects are introduced at the emitter / base interface, which leads to deterioration of the device characteristics of the formed HBT.

As another method for realizing an ultra-high concentration doping base layer, a method of doping an isoelectronic impurity together with a p-type impurity is also known. One example is an AlGaAs / GaAs HBT.
In the GaAs base layer, the group III element I
Doping n.

KCWang et al. Doped In with 10% In in a GaAs base layer and doped Be with a concentration of 1 × 10 ²⁰ cm ⁻³ (KCWang,
PMAsbeck, MFChang, GJ Sullivan &DLMiller;
“A 20-GHz Frequency divider implemented with het
erojunction bi-poler transisters ", IEEE Electron D
evice Lett., Vol.EDL-8, No. 9, pp. 383-385, Sep., 198
7). In addition, D. Barker et al. Dope Be to a concentration of 6 × 10 ¹⁹ cm ^-3 by mixing In with 10% in a GaAs base layer (D. Barker, Y. Ashizawa, P. Tasker, B. et al. Tadayon & LFEa
stman; “Extremely high peak specific transconductu
nce AlGaAs / GaAs hetero-junction bipolar transisto
rs ", IEEE Electron Device Lett., Vol. 10, No. 7, July, 19
89).

Although ultra-high-concentration doping is possible with this method, the fact that the material to be doped exists in addition to the impurity for controlling the conductivity, even if it is an isoelectronic impurity, does not require the supply system of the source gas or the control. It requires an extra system.
As the apparatus becomes more complicated, the chances of contamination due to leakage and the like increase, and the inconvenience due to residual gas and the like increases.

Furthermore, in the above example, when Be is doped, Be is diffused in the growth layer, so it is necessary to distribute In correspondingly. It is necessary to adjust the timing.

As described above, two methods are known as treatment methods for ultra-high concentration doping. However, in the method of changing the growth temperature Ts, there is a possibility that the epitaxial crystal is contaminated during the growth interruption period for raising the temperature. And the method of adding isoelectronic impurities adds complexity to equipment and control conditions.

SUMMARY OF THE INVENTION An object of the present invention is to provide an epitaxial growth method in which ultra-high concentration doping is performed by an epitaxial growth method without changing the growth temperature or complicating the apparatus.

[Means for solving the problem]

The basic measures of the present invention for achieving the above object are as follows:
The fluctuation of the lattice constant caused by ultra-high concentration doping,
The composition is compensated by adjusting the composition of the growth layer in advance, so that the doped impurity atoms are easily replaced with predetermined lattice points.

The process of setting the compound composition by predicting the change in the lattice constant due to doping is described below in steps.

In the first stage, it is assumed that the crystal in which the dopant impurity atoms have replaced all the atoms at the substitutional positions of the parent lattice is assumed. Determine the constant. This is the work of actually measuring the lattice constant of a crystal doped with the same impurity and extrapolating it to the high concentration side.

Next, a virtual lattice constant to be possessed by the compound crystal is determined on the assumption that the Vegard rule is satisfied between the virtual crystal and the parent compound crystal to be subjected to ultra-high concentration doping. Here, the condition that the lattice constant becomes the same as the lattice constant of the epitaxially grown substrate crystal when the doping is performed at a predetermined concentration is a condition to be satisfied, and the lattice constant is calculated by extrapolation as in the previous step.

When the lattice constant of the host crystal is determined in this manner, the corresponding compound composition is also determined, and accordingly, a source gas is supplied and epitaxial growth is performed. that time,
A predetermined amount of the dopant impurity material is also supplied.

[Action]

As described above, if the lattice constant of the epitaxial growth layer is biased, the substitution of the atoms of the host lattice with the impurity atoms realizes lattice matching, and approaches a stable state. Therefore, substitution by impurity atoms is promoted, and ultra-high concentration doping is performed.

One example of a change in lattice constant with an increase in impurity concentration is the seventh.
It is shown in the figure. In the figure, the change in the lattice constant when the Vegard rule is applied by regarding the impurity element as one of the constituent elements of the multi-component system is plotted by a solid line. It shows that the application of Vegard's rule is reasonable.

〔Example〕

The calculation of how much the lattice constant is shifted is performed as follows.

First, the case where the substitutional impurity C is doped into the crystal XA will be considered. Assuming a crystal XC in which all substitution site lattice points in the crystal XA are substituted by C atoms, a state in which XA is doped with C can be indicated by a position on the XA / XC pseudo-binary system.

Now, assuming that C is doped at a concentration of n (cm ⁻³ ), its lattice constant d is obtained by applying the Vegard rule to the XA / XC pseudo-binary system. Assuming that the molecular density of XA is n _A , the lattice constant is a, the molecular density of XC is n _C , and the lattice constant is c, the following equation (1) holds.

d = (a × n _A ^1/3 + c × n _C ^1/3 ) / (n _A ^1/3 + n _C ^1/3 ) (1) Here, the 1/3 power of n _A and n _c is used. The reason is that the molecular density, which is the number of atoms per volume, is matched with the lattice constant represented by the length, and the molecular density of XC is the concentration n of C.
Is equivalent to

The lattice constant of crystal XA not doped with impurities is known, and the lattice constant d when crystal XA is doped with impurity C can be obtained by actual measurement. Furthermore a value above n _A and n _C may also be calculated from the known or measured values. Therefore, the lattice constant c of the virtual crystal XC can be obtained by substituting these values into the above equation (1).

This will be described with a specific example. When the crystal XA is a GaAs crystal and is doped with Be as the impurity C, Be
Since a Ga atom is substituted, a compound called BeAs is considered as a virtual crystal XC. GaAs has a lattice constant a = 5.653 °, and when Be is doped to a concentration of n = 2 × 10 ²⁰ cm ⁻³ , the lattice constant is d.
= 5.647 ° (JLLievin et al., Inst. Phys. Conf., 198, supra).
Five). Note that n _A = 4.4 × 10 ²² cm ⁻³ .

When these numerical values are substituted into the above equation (1), the lattice constant of BeAs, which is a virtual crystal, is 5.56 °. Note that all the lattice constants are values at room temperature, and the same applies to the description later in this specification.

As another example, when the crystal XA is GaAs and the virtual crystal is SiAs, a lattice constant of 5.652 ° is obtained as an actually measured value when the concentration of Si is n = 5.5 × 10 ¹⁹ cm ^−3. Is calculated by substituting the numerical value of
5.643Å. The measured values of the above lattice constants were obtained from Tokunaga et al., "Evaluation of ultra-highly doped p-type GaAs for HBT", Tokyo Tech Ultra-High Speed Electronics 2nd Research Report Meeting sample (May, 198
8).

Furthermore, carbon (C) in GaAs becomes an acceptor by substituting As, as described in Yamada et al., “MOMB of metallic p-type GaAs.
When the same calculation is performed using the data of "E growth", a sample of the Tokyo Institute of Technology ultra-high-speed electronics second research report meeting, June, 1989), the lattice constant of virtual crystal GaC is 5.578Å.

The process described above is the first step of determining the composition of the compound semiconductor to be epitaxially grown in the present invention, and in the subsequent second step, the lattice constant a _VC of the virtual crystal XC obtained here is determined.
From the lattice constants a _SB of the epitaxial growth substrate crystal compound crystal X which in lattice constant to a predetermined impurity concentration of a _SB
The lattice constant of B is determined. That is, this is a process for determining how much the lattice constant should be shifted when no impurity is doped. Hereinafter, the processing of the second stage will be described in detail.

At this stage, what is fundamental is that the state in which the compound XB is heavily doped with the impurity element C is indicated by the position in the XB / XC pseudo-binary system. When the Vegard rule is applied to the XB / XC pseudo-binary system, a lattice constant d 'at a certain composition in the system, a molecular density n _{B of} XB, a lattice constant b, a molecular density n _{C of} XC, a lattice constant c ′, d ′ = (b × n _B ^1/3 + c × n _C ^1/3 ) / (n _B ^1/3 + n _C ^1/3 ) (2) The following relationship holds.

Lattice constant c of the crystal XC is given as d 'condition to be satisfied that matches the lattice constant a _SB of already sought, the molecular density of XC showing an impurity concentration n _C and the substrate crystal in the first stage . Further, since n _B >> n _C , n _B may be regarded as constant regardless of the impurity concentration, even when the composition of the compound XB slightly varies. Therefore, b can be obtained by substituting the measured values into the equation (2).

Here is a specific example. In the case where the virtual crystal XC is BeAs, the lattice constant c is 5.56 ° as determined above, and d ′ is 5.869 ° because d ′ matches the lattice constant of the epitaxial growth substrate InP. XB for epitaxial growth is I
Regardless of nGaAs or InAlAs, the molecular density is n _B = 4.0 × 10 ²² cm ⁻³ irrespective of the composition. By substituting these values into equation (2), the doping concentration of Be n
Fig. 1 shows the relationship between the lattice constant of the crystal XB and the undoped crystal XB. Where n in the figure is above
corresponding to n _C.

That is, when growing InGaAs or InAlAs of ultra-high concentration doping on an InP substrate, the lattice constant of the mother crystal lattice corresponding to the impurity concentration is determined from the graph of FIG. The equivalent x value of In _x Ga _1-x As or In _x Al _1-x As is calculated separately, and the source gas is adjusted to achieve the x value and the impurity concentration, and the epitaxial growth is performed. That is good.

Next, the virtual crystal XC is SiAs and the epitaxial growth layer is
In the case of InGaAs or InAlAs, the relationship between the impurity concentration and the lattice constant of the undoped crystal XB by the same treatment is as follows.
It is determined as shown in FIG.

Furthermore, the virtual crystal XC is GaC, and the epitaxial growth layer is G
In the case of aAsSb, the relationship between the impurity concentration and the lattice constant of the undoped crystal XB is obtained by the same processing as shown in FIG. In this case, unlike the first two examples, the impurity element C is replaced with a group V element, but the concept is exactly the same.

In the case where a lattice mismatch occurs between the epitaxially grown layer and the substrate crystal due to the ultra-high concentration doping, a matrix lattice biased by the value obtained by the above-described processing is used as the mother lattice. By doping the impurities, lattice matching is realized as a result.

The relationship between the x value and the lattice constant is shown in FIG. 4 for In _x Ga _1-x As, FIG. 5 for In _x Al _1-x As, and FIG.
_x Sb _1-x is shown in FIG. 6 respectively. These are described in publicly known documents, and FIGS. 4 and 6 show “Landort-Boernstein Numerical Data and Functiona”.
l Relationships in Sience and Tech-nology, New Ser
ies "Vol.17, p.619 & p.631 (1982, New York), and Fig. 5 is Makoto Konagai," Introduction to Semiconductor Superlattice ", p.31
(1987, Baifukan) is a redrawn part of Figure 2.27 (b).

As described above, the HBT whose base is formed by the epitaxial growth method of the present invention has excellent characteristics without deterioration of device characteristics due to lattice mismatch. The HB
The specifications and the manufacturing method of T are the same as those of the ordinary one except for the formation of the hetero junction.

〔The invention's effect〕

As described above, according to the epitaxial growth method of the present invention, lattice matching is realized in a state where ultra-high concentration doping is performed. As a result, the residual strain is reduced, and the device characteristics of the HBT are improved.

[Brief description of the drawings]

FIG. 1 is a diagram showing the relationship between the impurity concentration and the lattice constant of the mother crystal in the first embodiment, and FIG. 2 is the relationship between the impurity concentration and the lattice constant of the mother crystal in the second embodiment. FIG. 3 is a diagram showing the relationship between the impurity concentration and the lattice constant of the mother crystal in the third embodiment. FIG. 4 is a diagram showing the relationship between the lattice constant and the composition of the first mother crystal. 5 is a diagram showing the relationship between the lattice constant and the composition of the second mother crystal, FIG. 6 is a diagram showing the relationship between the lattice constant and the composition of the third mother crystal, and FIG. 7 is the impurity concentration and the lattice constant. FIG.

────────────────────────────────────────────────── ─── Continuation of the front page (56) References JP-A-63-217672 (JP, A) JP-A-1-166568 (JP, A) Nagai Haruo (Two others), "III-V" Group Semiconductor Mixed Crystals ”(63-10-25) Corona's first edition, first printing (second printing), issued July 30, 1993, page 30 (58) Fields investigated (Int. Cl. ⁷ , DB Name) H01L 21/20 H01L 21/331 H01L 29/205 H01L 29/73

Claims (3)

(57) [Claims] 1. A multi-component compound obtained by doping a compound AB composed of A and B, which are single or plural homologous elements, with a high concentration of an impurity element C, having a lattice constant of a _SB In the process of epitaxially growing on a certain substrate crystal, the multi-system is regarded as (a) a pseudo-binary system composed of compound AB and virtual compound CB when element C has an attribute of substituting element A. Or (b) if element C has the attribute of substituting element B, consider it as a pseudo-binary system consisting of compound AB and virtual compound AC, and apply Vegard's law to the pseudo-binary system By
From the measured lattice constant of the multi-component compound, the known concentration of C and the known lattice constant of compound AB, CB
Or calculating the lattice constant a _VC of the virtual compound _AC ,
Next, the lattice constant of the multi-component compound obtained by doping the compound AB with C at a desired concentration is changed to the lattice constant a _{SB of the} substrate crystal.
In order to be equal to the lattice constant a _VT virtual should have said Compound A-B, by applying the law of Vegado considers multi source system and the pseudo binary system, the lattice constant a _VC , Calculated using the desired concentration of C and the lattice constant a _SB , wherein A or B of compound AB is A ′ and A ″ or B ′, each of which is an element analogous to A or B. And when A ′ _X A ″ _1-X or B ′ _X B ″ _1-X consisting of B ″, the value of x such that the lattice constant of the compound AB becomes the a _VT is determined according to Vegard's law. Calculating, while containing the desired concentration of C, wherein A or B is
The multi-component compound layer having the calculated x value of A ′ _X A ″ _1-X or B ′ _X B ″ _1-X is epitaxially grown on a substrate crystal having a lattice constant of a _SB. To grow a high concentration compound semiconductor. 2. The method for growing a high-concentration compound semiconductor according to claim 1, wherein said AB compound is a III-V compound. 3. A method for manufacturing a heterojunction bipolar transistor, wherein a base region forming a heterojunction with an emitter region is formed by the method for growing a high-concentration compound semiconductor crystal according to claim 1 or 2. .