WO2000079600A1

WO2000079600A1 - SINGLE HETEROJUNCTION InP-COLLECTOR BJT DEVICE AND METHOD

Info

Publication number: WO2000079600A1
Application number: PCT/US2000/017156
Authority: WO
Inventors: Daniel P. Docter; Miroslav Micovic; Mehran Matloubian
Original assignee: HRL Laboratories LLC
Current assignee: HRL Laboratories LLC
Priority date: 1999-06-22
Filing date: 2000-06-22
Publication date: 2000-12-28
Anticipated expiration: 2001-12-22
Also published as: AU5631900A

Abstract

A Single Heterojunction Bipolar Transistor employing a collector (3, 4) and base (5, 6) of InP, and an emitter (8, 9) of either a selected InxA11-xAsySb1-y compound, which preferably is lattice-matched to InP, or of a superlattice of ternary or binary compounds which mimic the selected InA1AsSb compound. The base-emitter junction is preferably graded, preferably by means of a chirped superlattice. Doping of the junction may be adjusted to include one or more delta doping layers to enhance the efficiency of the transistor.

Description

SINGLE HETEROJTJNCTION InP-COLLECTOR BJT DEVICE AND METHOD

FIELD OF THE INVENTION

The present invention pertains to bipolar junction transistors (BJTs), more particularly to heteroj unction BJTs (HBTs), and specifically to single heteroj unction BJTs (SHBTs).

BACKGROUND

High-speed, high-power BJTs are useful for high frequency power applications such as VHF and wideband radar systems. InP collectors are attractive for such BJTs due to the high saturation velocity and to the large bandgap and breakdown field which they provide. These characteristics translate into short collector transit time and high breakdown voltages

BV_CB0 and BV_CE0.

Double heterojunction BJTs (DHBTs) employing InP collectors are known, as disclosed, for example, in "Handbook of III-V Heterojunction Bipolar Transistors" by William Liu, March 1998 (John Wiley & Sons). The base of such devices is typically GalnAs doped with beryllium or carbon, while the collector is lightly Si-doped InP. These materials result in a conduction-band discontinuity in the collector-base junction, typically 0.25 eV for material grown by gas-source Molecular Beam Epitaxy (MBE). Such a discontinuity impairs the flow of electrons from the base to the collector, imposing a need for meticulous bandgap engineering, usually involving elaborate grading layers. It is believed that the design and growth of a graded collector-base junction is significantly more challenging, and the result more critical to device performance, than grading a forward-biased base-emitter junction.

DHBTs having a collector of InP and a base of GaAsSb are also known. For example, U.S. Patent 4,821,082 to Frank, et al., describes a DHBT having collector and emitter of InP, and a base of GaAs_{0 53}Sb_{0 47}. U.S. Patent 5,349,201 to Stanchina, et al. and Publication EPO

715357 Al to McDermott also describe DHBTs having collector and emitter of InP, and base of GaAsSb. The problem with this approach is that GaAs_{0 53}Sb_{0 47} has a conduction band energy level which is approximately 0.12eV higher than that of InP. Although the combination of a base of GaAsSb, lattice-matched to a collector of InP, eliminates the problem of electron accumulation at the base-collector interface of the DHBT, it instead causes injection of high energy electrons into the collector from the base, because the electrons gain energy equal to the conduction band offset at the base-collector junction. Such injection of high energy electrons into the collector reduces base-collector breakdown voltage by facilitating impact ionization.

It is well known that having a wide-gap emitter improves the gain and efficiency of a B JT, and it is believed that the growth of epitaxial layers is simplified, and functional reliability is enhanced, by lattice-matching the epitaxial layers of a device.

Consequently, it is desirable to develop a BJT device which avoids a difficult base- collector heterojunction by employing InP in both the ^-collector and the -base, while providing a wide-gap emitter which is lattice-matched to the InP base and collector.

Such a device has been invented. For example, U.S. Patents 5,610,086 and 5,612,551 to Liu, et al. describe a Single Heterojunction Bipolar Transistor (SHBT) having a collector- base homojunction and a base-emitter heterojunction of InP / A1P_{0 39}Sb_{0 6]} . Unfortunately, the A1P_{0 39}Sb_{0 61} of the emitter material has proven chemically unstable and extremely difficult to fabricate.

Grading the base-emitter junction has also been found to improve reliability and benefit device performance in certain heterojunction transistor structures, as described in U.S.

Patent 5,365,077 to Metzger, et al. It is known that inserting delta doping planes at the ends of a graded base-collector junction allows transfer of bandgap discontinuities from the conduction band predominantly to the valence band, which improves device performance [Chanh Nguyen et al, IEEE

Proceedings Cat. No. 95CH35735, pp 552-62].

Accordingly, there is a need for a semiconductor device, and a method of forming such a device, having an InP collector and base, and having an emitter of a different material having a wider energy bandgap. There is also a need for useful variations upon the desired semiconductor device and method, such as grading the base-emitter junction, and employing doping planes for the base-emitter junctions in order to transfer bandgap discontinuities from the conduction band to the valence band. SUMMARY OF THE INVENTION

It is an object of the present invention to provide a semiconductor device, and a method of forming such a device, which is suited to high speed, high power applications.

This object is achieved by means of a semiconductor epitaxial structure which can be formed into discrete or integrated bipolar transistors, particularly SHBTs, using one of certain closely related layer structures. The preferred layer structure includes a collector of InP, a collector-base homojunction of InP, a base of InP, a base-emitter heterojunction, and an emitter which either includes Ir^Al_j^As Sb, , or else includes a superlattice of alternating materials which together mimic the electrical behavior of In_λAl₁__χAs Sb_j . The In-Al,, _χAs Sb, of the emitter should have x < .55 (preferably 0.2 < x < 0.52) and 0 < y < 1, where y is preferably a function of x such that the compound is lattice-matched to InP.

The base-emitter junction may employ an abrupt composition change (i.e. may be an abrupt junction), which is preferable for simplicity of fabrication, but a graded junction is preferable for device efficiency and reliability. Grading can be implemented through continuous compound change, but preferably uses a chirped superlattice employing a number of periodic layers, each having alternating sublayers of two materials, with the thickness of the sublayers of one material being increased from period to period while the thickness of sublayers of the other material are correspondingly decreased from period to period.

The emitter may be fabricated from a homogeneous quaternary compound in a desired formulation of In-.Al,_.χAs Sb_j , or may be formed as a superlattice with alternating layers of different binary or ternary materials which together mimic the electrical characteristics of the selected quaternary compound of In-,Al,__χAs Sb_j .

BRIEF DESCRIPTION OF THE DRAWINGS

Fig. 1 depicts the epitaxial structure of a SHBT according to the present invention. Fig. 2 depicts the structure of a chirped superlattice base-emitter junction grading layer.

Fig. 3 shows energy bandgaps and band lineups for abrupt base-emitter heterojunctions.

Fig. 4 shows energy bandgaps and band lineups for graded base-emitter heterojunctions with delta doping.

Fig. 5 depicts epitaxial structure details of a superlattice emitter. Fig. 6 depicts photoresist mask placement of an emitter contact. Fig. 7 shows the emitter and emitter contact after an etch using the emitter contact as a mask. Fig. 8 depicts photoresist mask placement of a base contact. Fig. 9 depicts photoresist mask for etching to the subcollector contact area. Fig. 10 depicts a device with a collector contact in place. Fig. 11 shows a device after isolation etching.

DESCRIPTION OF THE PREFERRED EMBODIMENTS

Fig. 1 shows a completed epitaxial structure for forming a SHBT according to the present invention. For convenience in supporting an unstrained InP collector structure, each of the preferred embodiments begins with InP substrate 1. Undoped InP buffer layer 2 is grown on InP substrate 1 to a thickness of about 100 A to enhance subsequent layer growth.

InP subcollector 3 is grown next, with heavy n doping to a density of about 10¹⁹ / cm³. Subcollector 3 is grown to a thickness of about 7000 A. InP collector 4 is grown next, with light n doping to a density of about 5 *10¹⁵ / cm³, and also to a thickness of about 7000 A.

InP base 5 is grown above collector 4, with very heavy ? doping to a density of about 2.6 *10¹⁹ / cm³, and to a thickness of about 500 A. InP base spacer layer 6 is preferably grown next to reduce diffusion of p dopant into n doped regions. Base spacer layer 6 is given more moderate p doping to a density of about 2 * 10¹⁸ / cm³, and is grown to a thickness of about 100 A. Base spacer layer 6 is omitted or modified in some delta-doped configurations. Base-emitter heterojunction grading layer 7 may optionally be formed next. Grading layer 7 is described in more detail below with reference to Fig. 2. Grading layer 7 is omitted in some of the preferred embodiments to create an abrupt base-emitter heterojunction.

Emitter 8 is grown next, preferably moderately «-doped to a density of about 8 *10¹⁷ / cm³ and grown to a thickness of about 1050 A, followed by heavily «-doped (to a density of about 10¹⁹ / cm³) emitter contact layer 9, which is grown to a thickness of about 350 A. Two general methods of forming emitter 8 and emitter contact 9 are preferred. The first method employs In Al₁__χAs _ySb. as a uniform quaternary compound. Te is generally preferred for n doping these Sb-containing compounds, though Class IV elements may be acceptable, particularly for lower Sb densities, as discussed elsewhere with regard to doping considerations. The resulting donor density is preferred to be about 8 *10¹⁷ / cm³ for emitter 8 and about 10¹⁹ / cm³ for emitter contact 9. The second general method of forming emitter 8 and emitter contact 9 is to form a superlattice having numerous periods, each with alternating sublayers of at least two binary or ternary compounds, such as AlAs and InSb or InAlAs and InAlSb. When the sublayers are sufficiently thin and of the correct proportion to each other, the superlattice mimics the electrical properties of In_χAl_]__χAs Sb, . According to this general method, it is preferred to n dope only the Sb-free sublayers, at a density which achieves the correct overall preferred density of about 8 *10¹⁷ / cm³ average for all of emitter 8 and about 10¹⁹ / cm³ average for all of emitter contact 9, as above.

The materials and methods of forming the emitter and the emitter contact layer are described in detail below with reference to Fig. 5.

Finally, contact layer 10 may be grown above emitter contact layer 9, preferably to about 1000 A thick. Contact layer 10 is preferably rc-doped similarly to emitter contact layer 9 (to a density of about 10¹⁹ / cm³), but is composed of an alternative material, such as InGaAs lattice-matched to InP, which is preferred for creation of the metal ohmic emitter contact.

The present invention has a number of preferred embodiments. The base-emitter heterojunction may be graded or abrupt, and may employ delta-doping techniques or not; and the emitter may be grown from a uniform quaternary compound, or as a superlattice comprising ternary compounds, or as a superlattice comprising binary compounds. These distinguishing features may be combined in various ways to yield the following twelve preferred embodiments, which employ:

1) a graded heterojunction and a uniform quaternary compound emitter;

2) a graded heterojunction and a superlattice emitter of ternary compounds;

3) a graded heterojunction and a superlattice emitter of binary compounds; 4) an abrupt heterojunction and a superlattice emitter of ternary compounds;

5) an abrupt heterojunction and a superlattice emitter of binary compounds;

6) an abrupt heterojunction and a uniform quaternary compound emitter;

7-9) as embodiments 1-3, but omitting the base spacer layer and adding a delta doping plane adjacent the emitter; 10-12) as embodiments 4-6, but replacing the base spacer layer with an undoped layer of emitter material and adding a delta doping plane thereafter. The above listing of preferred embodiments is not exclusive, and should not be construed as precluding other combinations and permutations of the features described herein.

Chirped Graded Base-Emitter Junction

Fig. 2 shows the preferred structure of optional graded layer 7. The grading may be done over a range of epitaxial thicknesses, but 300 A is a preferred thickness. Grading occurs over m steps, which requires (m-1) periods in grading layer 7. Fig. 2 depicts nine periods 21- 29, so that grading occurs over m = 10 steps. Nine periods are preferred, but the number may range from 3 to 50 or more. For convenience, each period is the same thickness, 33.3 A in this example. Each period has at least two sublayers 11, 12 or 13, 14, of differing materials. Except for doping, the material of sublayers 11 is preferably the same as that of sublayers 13, while sublayers 12 are similarly of the same material as sublayers 14. The material of superlattice 7 is graded stepwise, or chirped, by increasing the thickness of sublayers 11 (or 13) from period to period, while correspondingly decreasing the thickness of sublayers 12 (or 14), so that the total thickness of each period 21-29 remains the same. In Fig. 2, sublayer 13 is about 10% of the thickness of period 21 , or about 3.3 A, while sublayer

14 is correspondingly about 29.7 A of that period. In period 22, sublayer 13 is about 20% of the thickness, and in period 23 sublayer 11 is about 30%. Thickness of the sublayer 11/13 material continues to be increased in steps until it becomes 90% of the thickness of period 29, or 29.7 A. The thickness of material 12/14 correspondingly decreases from about 90% of period 21, to about 10% of period 29. Clearly, then, with shifts of 10% per period, the entire grading occurs over m - 10 steps in this example.

It is generally preferred top dope periods 21-22 of grading layer 7 to the same density as base spacer 6, and to n dope periods 23-29 to the same density as emitter 8. Thereby, periods 21-22 of grading layer 7 act with base spacer 6 to minimize the effects of diffusion of the heavy concentration of p dopant in base 5. The p-n junction is thus positioned partway through grading layer 7. Thus, sublayers 11 differ from sublayers 13 in that the former are n- doped, while the latter are -doped, and sublayers 12 similarly differ from sublayers 14. This is the generally preferred method; however, an alternative method of doping a graded junction, delta doping, is discussed later. The simplest preferred arrangement of materials for chirped superlattice grading layer 7 employs the same material structure as that of emitter 8 for sublayers 11 and 13, and employs the same material as base spacer 6 for sublayers 12 and 14. However, an alternative to this simple material arrangement is described next. To best understand this alternative method of selecting the composition of sublayers

1 1/13 and 12/14, it should be realized that whether emitter 8 is a uniform quaternary compound, or is a superlattice of dual ternary or binary compounds as described in a later section, it mimics and may be considered as in a sense as having a quaternary compound compositional formula, In_χAl,__χAs Sb, defined by values of x and y. With that premise, the following method can be applied to any embodiment of an emitter according to the present invention.

Given a selected emitter 8 quaternary compound defined by x and y, and base spacer 6 material of InP, one may grade from base to emitter by using sublayers of ternary compounds IILAI^AS and Ir^Al_j^Sb. Grading occurs over (m-1) periods of grading layer 7; Fig. 2 is shown with 9 periods, hence m = 10.

In grading from the base to the selected emitter composition, we are primarily concerned with grading the energy bandgap, and to a lesser extent the lattice constant. For these purposes, a near equivalent to InP in an In- l^As Sb_j formulation occurs when x =

.5 and y = 1. Thus, the grading is performed as if the base composition had that formulation. That formulation is then shifted to the desired emitter formulation, which as an exemplary preferred embodiment is In_{0 35}A1_{0 65}As_{0 85}Sb_{0 15} (x = .35, y = .85). Accordingly, since we are grading in this example in m = 10 steps, each step should shift x by (.35-.5)/m, or -.015, and y by (.85 - l)/m, which is also -.015.

To accomplish this grading, period 21, which is adjacent base spacer 6, may have sublayer 13 composed of In_{Q 485}A1_{0 515}Sb for 1.5% of a 20 A thickness of period 21, or 0.3

A, and sublayer 14 composed of In_{0 485}A1_{0 515}As for the remainder, or 19.7 A. The effective quaternary formula for the entire period is thus In_{Q 485}A1_{0 515}As_{0 985}Sb_{0 015}, or In- -l^ _χAs Sb, where x = 0.485 and y = 0.985. Progressively, then, in period 29 x is (.5 - 9 *

0.015 = .365), while y is (1 - 9 * 0.015 = .865), so sublayer 11 would be In_0-365Al_0ι635Sb for 13.5% of 20 A, or 2.7 A, while sublayer 12 would be In_{0 365}A1₀₆₃₅As for the remaining 17.3 A. The effective quaternary formula averaged over period 29 would thus be

^In0 365^A10 635^As0 865^0 135"

It is preferred, but not essential, that grading occur in approximately equal steps. Such a grading arrangement has several advantages. The Sb-containing sublayers need not be doped, solving the issue of Te versus Si doping, and a quaternary compound of

InAlAsSb is not necessary. Furthermore, it avoids a difficulty which would arise in the case that emitter 8 is itself a superlattice with sublayers of at least two materials. If emitter 8 is a superlattice, the usual method of having sublayers 11/13 of emitter material would inconveniently require that sublayers 1 1/13 be composed of further sub-sublayers, as in the emitter.

The grading arrangement also has disadvantages. Periods 21-29 grown according to this method are somewhat strained. Moreover, a variety of ternary compounds are necessary to grow the varying sublayers, since the material itself, and not merely the thickness, varies from period to period. This embodiment is accordingly not generally the most preferred, but is preferred for certain purposes and processing capabilities.

Band Energy Lineups and Delta Doping

Fig. 3 shows the band lineups for an exemplary emitter 8 of In_{0 35}A1_{0 65}As_{0 85}Sb_{0 ] 5}, in an abrupt junction with InP base spacer 6. Conduction band energy in emitter 8 is represented by line 41, with the conduction band energy of base spacer 6 at a different level represented by line 42, such that conduction band transition 46 is a sudden shift or discontinuity in conduction band energy at the abrupt junction between emitter 8 and base spacer 6, the discontinuity having a calculated value of 0.5 eV. Valence band energy level 47 in emitter 8 is 0.03 eV lower than valence band energy level 48 in base spacer 6, such that valence band transition 45 is a discontinuity of only .03 eV. Bandgap 43 of emitter 8 between conduction band energy 41 and valence band energy 47 is calculated to be 1.88 eV, while bandgap 44 of base spacer 6 is calculated to be 1.35 eV.

Employing delta doping in grading layer 7 may produce the band diagram shown in Fig. 4. Two effects occur gradually across grading layer 7, and superimpose to result in the band diagram of Fig. 4. There, conduction band transition line 46 and valence band transition line 45 show the band energy changing gradually across grading layer 7, with the alignment such that most of the shift occurs in the valence band, shown as lines 47, 45 and 48. The first effect is a gradual shift in band gap, which is a result of the shift in materials of grading layer 7. The second effect is an approximately 0.5 eV shift which also occurs across the width of grading layer 7, but which is due to an electrostatic field between delta doping planes 51 and 52. The net result of the superposition of these two effects is that the conduction band transition 46 seen in Fig. 3 between base spacer 6 and emitter 8 is effectively "transferred" to the valence band. Therefore, valence band transition 45 displays the majority of the total .53 eV bandgap difference between base 5 and emitter 8.

To accomplish such delta doping, base spacer 6 is omitted, and heavy doping of base 5 is continued up to grading layer 7. Due to the heavy doping of base 5, and the depletion effects of the junction, the transition plane 51 between base 5 and grading layer 7 appears as a plane of negative charge, which serves as base-side delta doping plane 51. Grading layer 7, preferably 300 A thick, is left undoped. Most of emitter 8 is ordinarily doped to about 8 * 10¹⁷ / cm³. However, the first 10 A of emitter 8, adjacent grading layer 7, is especially heavily doped to become emitter-side delta doping plane 52. Delta doping plane 52 is made of the same material as emitter 8, but is heavily n doped to a density of 1.2 * 10¹⁹ / cm³. Since the charges are depleted in this region, delta doping plane appears as a plane of positive charge. Thus, positive delta doping plane 52 faces negative delta doping plane 51 300 A away on the opposite side of grading layer 7. The electrostatic field between these planes causes a shift between the band energy lineup between base 5 and emitter 8, as described above.

Delta doping can also be implemented with an abrupt junction. In that case, most of the layers are preferably as described above with reference to Fig. 2. However, base spacer 6 is changed to 300 A of undoped emitter material, and layer 7 is changed from a grading layer to 10 A of emitter material which is heavily doped to a density of 1.2 * 10¹⁹ / cm³, to become the emitter-side delta doping layer. As in the graded junction case, then, appropriate planes of charge face each other across a certain thickness of the junction, effecting a shift in band lineup such that the discontinuity effectively moves from the conduction band to the valence band. Formation of Emitter

Emitter 8 and emitter contact 9 of a SHBT according to the present invention may include a uniform quaternary compound of In-^Al.^As Sb, . The quaternary compound is formulated preferably with 0.2 < x < 0.52, and with y selected as a function of x such that the resultant crystal is lattice-matched to InP. With higher Al content, the Sb content must also be increased to lattice-match the crystal with InP. In_{0 35}A1_{0 65}As_{0 85}Sb₀ ,₅ is a specific example of such a quaternary compound. For some purposes, x may be at least 0.55.

The present invention also contemplates formulations where 0 < x < 0.2. Indeed, since the bandgap of the emitter generally increases as the concentration of Al increases, leading to higher gain and efficiency, a ternary compound AlAs Sb, (x = 0) offers potentially the highest efficiency of emitters according to the present invention. However, it is believed that compounds with high mole densities of Al, i.e. with x < .2, will be unstable. Accordingly, such compounds are less preferred.

Whatever quaternary compound In-Al^As Sb, , defined by x and y, is selected as described above, it may be mimicked by a superlattice of ternary compounds. Referring to

Fig. 5, emitter 8 may be formed of alternating sublayers 55 and 57 organized into periods, each period containing one sublayer 55 and one sublayer 57. The material for each sublayer 57 is preferably

and its thickness is y times the period thickness; while each sublayer 55 is preferably In_χAl.__χSb with a thickness which is (1-y) times the period thickness. Accordingly, since the exemplary preferred emitter material is

In_{0 35}A1_{0 65}As_{0 85}Sb₀ ,₅ (x = .35, y = .85), the thickness of sublayers 57 is 85% of the period thickness, while the thickness of sublayers 55 is 15% of the period thickness.

The same composition and thicknesses apply to sublayers 54 and 56 of emitter contact 9 as apply to sublayers 55 and 57 of emitter 8. The preferred difference between emitter 8 and emitter contact 9 is doping level, and the materials of the sublayers are otherwise preferably the same as those of emitter 8.

These sublayer materials are not individually lattice-matched to InP. However, across each full period the overall lattice mismatch will be balanced to zero by making the average composition equal to a quaternary compound which is lattice-matched to InP. That is, one sublayer will be in tensile strain, and the other sublayer will have an equal but opposite compressive strain. In order to minimize any difficulties caused by the strained material, the sublayers are preferably very thin, with the overall period thickness being preferably between 2 A and30 A, and most preferably about 4 A.

Thus, to mimic In_{0 35}A1_{0 65}As_{0 85}Sb₀ ,₅, sublayers 57 would be In_{0 35}A1_{0 65}As at a thickness 85% of the period thickness, or 3.4 A, while sublayers 55 would be In_{0 35}A1_{0 65}Sb at a thickness of about 15% of the period, or 0.6 A.

Since one sublayer contains Sb while the other does not, it is preferred to dope only the Sb-free sublayer. In order to achieve the desired overall doping effect, for example 8 * 10¹⁷ / cm³ as preferred for emitter 8, Sb-free sublayers 57 are doped to a density inverse to their proportional thickness within the period. Thus, in the example of mimicking

In_{0 35}A1_{0 65}As_{0 85}Sb₀ ,₅, sublayers 57 would preferably be doped to a density of 8 * 10¹⁷ / cm³ / 0.85, or 9.4 *10¹⁷ / cm³.

The preceding discussion in regard to the superlattice for emitter 8 applies also to emitter contact 9, except that sublayers 55 and 57 of emitter 8 are replaced by sublayers 54 and 56 of emitter contact 9. Sublayers 54 and 56 of emitter contact layer 9 are preferably the same materials as sublayers 55 and 57, respectively, and have the same thicknesses, but differ in doping level. Sublayers 56 are doped to the level desired for emitter contact layer 9 divided by their proportion in each period, that is preferably to a density of (1 *10¹⁹ / cm³⁾ / .85, or 1.18 * 10¹⁹ / cm³ in the present example. Quaternary formulations of In_ Al,__xAs Sb, in which x = (1 - y) may be mimicked by a superlattice of binary compounds InSb and AlAs. A quaternary compound In_χAl,_ _χAs Sb, which is lattice-matched to InP and in which x = (1 - y) occurs at approximately x = .25, y = .75. In that case, therefore, a superlattice in which each period is 25% InSb and 75% AlAs will have an average formula over each period of In_{Q 25} I_{0 5}N^so ₇₅^_{0 25}-' ^an<^ accordingly will mimic the electrical behavior of that quaternary compound. Referring again to Fig. 5, such a superlattice would preferably have sublayers 57 (56) of AlAs at a thickness of 3 A, and sublayers 55 (54) of InSb at a thickness of 1 A, for a total period thickness of 4 A. Only the AlAs sublayers would be doped, and to a density increased by 1/.75 to compensate for the undoped InSb portion. Thus, in this example sublayers 57 of emitter superlattice 8 would be n doped to 8 *10¹⁷ / cm³ / 0.75, or 1.07 *10¹⁸ / cm³, while sublayers 56 of emitter contact superlattice 9 would be n doped to 1 *10¹⁹ / cm³ / 0.75, or 1.33 *10¹⁹ / cm³.

When delta doping is incorporated, as described above in the Band Energy Lineup and Delta Doping section, the first emitter period adjacent grading layer 7 becomes delta doping layer 52. Accordingly, only the Sb-free layer is doped, and the doping density is adjusted to compensate for the proportion which is not doped. Because of the high doping density of delta doping layer 52, Te is preferred as a dopant due to its efficiency in proximity to Sb.

Fabrication Techniques

All of the embodiments of the present invention are achievable by employing either Molecular Beam Epitaxy (MBE) or Metal Organic Vapor-Phase Epitaxy (MOVPE) techniques which are well-known in the art, and one skilled in the art will understand that any technique may be used to achieve the described layer structure. The dopant materials and densities identified below are merely examples, and one skilled in the art will recognize that many other dopants will perform similarly and that the doping density may be varied. One skilled in the art will also appreciate that the epitaxial layer thicknesses given are also merely examples, and that the SHBT structure of the present invention is expected to work well with a wide variety of epitaxial layer thicknesses.

Doping to n type is preferably accomplished using Si, Sn or other common group IV elements, except for materials containing significant amounts of Sb. It is known that compounds lattice-matched to InAs or GaSb, which also contain high mole fractions of Sb, may cause Class IV dopants to act as acceptors rather than donors. Although the effect has not yet been observed in the InAlAsSb materials lattice-matched to InP which are employed in the present invention, it is expected that at high Sb mole fractions it would be preferable to rø-dope with Class VI elements, such as Te. Thus, for y < about .3, Te or another Class VI element is preferred for H-doping the material. If y > about .3, however, Si or other Class IV element are acceptable. Even at these lower Sb mole densities, Te is preferred, if readily available, to avoid uncertainty about the behavior of the dopant.

Doping to p type is preferably accomplished with Be, C or Zn, unless otherwise noted. Metallization may be accomplished by any standard compatible technique, such as sputtering, unless otherwise noted. The preferred metal is actually a combination of separate layers of titanium (about 100 A), platinum (about 300 A), and gold (about 1000 A), which is generally denoted as simply Ti/Pt/Au. Etches may be any compatible standard dry or wet etch, with testing of the depth achieved in those instances when the layer to be reached is thin and a selective wet etch is not available.

Figs. 6 - 11 show an implementation of an integrated SHBT according to the present invention. First, an epitaxial structure according to one of the preferred embodiments described above is grown. Fig. 6 shows substrate 1, subcollector 3, collector 4, base 5, emitter

8, emitter contact 9 and contact layer 10. For simplicity, buffer layer 2, base spacer 6 and optional grading layer 7 are not shown. Fig. 6 shows emitter contact and emitter contact An oxide cap layer of about 50 A, not shown, may be placed above contact layer 10 after growth of the epitaxial structure to protect the structure until further processing takes place. Fig. 6 also shows photoresist 52A, which has been patterned to define the emitter contact area. Any oxide cap in that area is first removed, and then emitter contact metallization 53 A is deposited onto the surface to achieve a thickness of about 4000 A. As is well known in the art, the side walls of the photoresist layer are controlled to have negative or vertical slope, so that the metallization does not attach thereto. Photoresist 52A will then be removed, and will lift off that portion of metal 53 A which is above photoresist 52A.

Fig. 7 shows the device after the "lift off," and after a subsequent etch which used emitter metallization 55 as a mask. The etch may be any standard, compatible dry or liquid etch, and removes the unused portion of the emitter layers 8 (which is understood to include layers 9 and 10 when appropriate) to expose base 5. Fig. 8 shows second photoresist 52B, which has been patterned and upon which second metal layer 53B has been deposited. Metal 53B deposited upon base 5 preferably forms a single contact 54 (Fig. 9) surrounding emitter 8, but contact 54 appears as two pieces in cross section.

After lift off of metallization 53B above photoresist layer 52B, third photoresist layer 52C is patterned with void 56 which will define the collector contact. After the photoresist is patterned as shown in Fig. 9, it is used for two subsequent steps. Fig. 10 shows the results of those two steps. Base 5 and collector 4 have been etched away to expose subcollector 3; and metallization has been deposited to form collector contact 58. Fig. 10 shows the device after these steps, and after the photoresist has been removed along with the unused metallization. Fig. 11 shows the device after a further photoresist has been patterned to allow a mesa etch to isolate the entire device structure down to the substrate. Substrate 1 supports subcollector 3, usually with a buffer layer between (not shown). Collector contact metal 58 is positioned upon subcollector 3. Collector 4 supports base 5, upon which base contact metal 54 is positioned, and emitter 8 (which typically includes emitter contact 9 and emitter contact layer 10, not shown). Emitter contact metal 55 is defined above emitter 8. A final passivation of the device, not shown, may be performed on the device shown in Fig. 11, as is well known in the art.

The SHBT device described herein is merely exemplary. Those skilled in the art will understand that many approaches for building a device according to the present invention, whether now known or hereafter developed, are encompassed within the following claims.

For example, one might build a device laterally using Lateral Epitaxial Overgrowth, might add or delete some layers, might alter doping levels and thicknesses, and could fabricate by a wide range of techniques. The scope of the invention is defined only by the following claims.

Claims

CLAIMSWhat is claimed is:

1. A method of forming a bipolar junction transistor comprising the steps of: forming a collector including InP; forming, adjacent said collector, a base including InP; and forming an emitter opposite said base from said collector, said emitter including either a quaternary compound of In_xAlι__xAs_ySb,__y where 0 < y < 1 and x < 0.55, or a superlattice having an average composition with mole fractions of In, Al, As and Sb the same as in the quaternary compound In_xAl,_.xAs_ySb,._y where 0 < y < 1 and x < 0.55.

2. The method of forming a bipolar junction transistor according to claim 1 wherein said In_xAl,__xAs_ySb₁__y is lattice-matched to InP.

3. The method of forming a bipolar junction transistor according to claim 1 wherein 0.2 < x < 0.52.

4. The method of forming a bipolar junction transistor of claim 1 further comprising the step of forming a grading layer between the base and the emitter.

5. The method of forming a bipolar junction transistor of claim 1 including a further step of forming a chirped superlattice layer which at least partially grades a transition between said InP base and said emitter.

6. The method of forming a bipolar junction transistor of claim 1 further comprising the step of forming delta doping planes between the base and the emitter such that a valence band difference between the base and the emitter is increased and a conduction band gap between the base and the emitter is reduced.

7. The method of forming the bipolar junction transistor of claim 1 wherein the emitter includes a superlattice having an average composition in which the mole fractions of In, Al, As and Sb are the same as in the quaternary compound In_xAlι__xAs_ySb,__y where 0 < y

< 1 and 0.2 < x < 0.55.

8. The method of forming the bipolar junction transistor of claim 7 wherein the emitter superlattice includes ternary compounds InAlAs and InAlSb.

9. A bipolar junction transistor comprising: a collector including InP; a base including InP; and an emitter opposite said base from said collector, said emitter including either a quaternary compound of In_xAl,_.xAs_ySb,_.y where 0 < y < 1 and x < 0.55, or a superlattice having an average composition in which the mole fractions of In, Al, As and Sb are the same as in the quaternary compound In_xAl₁__xAs_ySb,__y where 0 < y < 1 and x < 0.55.

10. The bipolar junction transistor of claim 9 wherein said emitter is lattice- matched to InP.

11. The bipolar junction transistor of claim 9 wherein said emitter includes a superlattice composed primarily of InAs and AlSb.

12. The bipolar junction transistor of claim 9 further comprising at least one heavily doped layer less than 100 A thick disposed within less heavily doped material and positioned within 600 A of a junction between the emitter and the base.

13. The bipolar junction transistor of claim 9 further comprising a grading layer between the base and the emitter.

14. The bipolar junction transistor of claim 13 wherein said grading layer includes a chirped superlattice layer which at least partially grades a transition between said InP base and said emitter.

15. The bipolar junction transistor of claim 9 wherein the emitter includes a superlattice having sublayers of InAlAs and InAlSb or having sublayers of InAs and AlSb.

16. The method of forming a bipolar junction transistor of claim 1 comprising the further steps of: disposing an undoped spacer layer between said base and said emitter; and disposing a very heavily doped delta doping plane layer less than 100 A thick between said spacer layer and said emitter; such that compared to an absence of said delta doping plane layer, alignment of a conduction band energy of said base is shifted with respect to a conduction band energy of said emitter, and alignment of a valence band energy of said base is shifted with respect to a valence band energy of said emitter.

17. The method of forming a bipolar junction transistor of claim 5 wherein said emitter includes a superlattice of binary or ternary compounds.

18. A single heterojunction bipolar transistor comprising: a collector region including InP, a base region including InP adjacent said collector, a homojunction between said base and said collector, and an emitter region adjacent a side of said base opposite said collector, said emitter region including ιn_xAl,__xAs_ySb₁__y where 0.2 < x < 0.52 and 0 < y < 1.

19. The single heterojunction bipolar transistor of claim 18, wherein said In_xAl,_ _xAs_ySb,._y is lattice-matched to InP.