Compound semiconductor heterojunction bipolar transistor

Abstract

The present invention provides a compound semiconductor heterojunction bipolar transistor, including a structure of an emitter layer and a base layer, in which, in a GaAsHBT using pseudo-grown InGaAs for the base layer, the influence of the conduction band energy barrier generated at the interface between the emitter layer and the base layer on the characteristics of a power amplifier is reduced. In the first invention, InGaP having a CuPt-type order is used for the emitter layer. In the second invention, the p-type impurity concentration of the part of the pseudo-grown InGaAs base layer on the emitter layer side is made lower than that on the collector layer side.

Description

Technical Field

[0001] The present invention relates to a compound semiconductor heterojunction bipolar transistor, comprising an emitter layer, a base layer and a collector layer epitaxially grown in sequence on a GaAs substrate, wherein the emitter layer, the base layer and the collector layer comprise semiconductor layers substantially free of relaxation of lattice strain caused by lattice mismatch with GaAs, and in particular, the base layer comprises at least an InGaAs layer obtained by pseudomorphic growth. Background Art

[0002] Heterojunction Bipolar Transistor (HBT) is a bipolar transistor that uses a semiconductor material with a band gap larger than that of the base region in the emitter region, and can simultaneously achieve high current amplification and low base resistance. Among them, the HBT (hereinafter referred to as GaAsHBT), which is composed of a compound semiconductor layer epitaxially grown on a GaAs substrate and has no relaxation of lattice strain caused by lattice mismatch with GaAs, is suitable for high-frequency applications because the mobility of conduction electrons, which are the main carriers of current flowing between the emitter and collector, is high, especially when used as an npn-type transistor. Therefore, GaAsHBT is widely used as a radio frequency (RF) power amplifier for mobile communications or wireless fidelity (WiFi).

[0003] In a conventional GaAsHBT, the base layer and the collector layer both use GaAs, which is the same as the substrate, as the main material, and the lattice constant is the same as the substrate, so there is no relaxation of the lattice strain caused by lattice mismatch. However, in the GaAsHBT of the present invention, InGaAs is used in the base layer (refer to non-patent document 1, non-patent document 2, and patent document 1). The lattice constant of InGaAs is larger than that of GaAs as the substrate, and the difference is roughly proportional to the In content, but good characteristics as a transistor are obtained, so pseudocrystal growth is performed while maintaining compressive strain. The pseudocrystal growth does not have relaxation of the lattice strain caused by lattice mismatch due to dislocation generation, that is, strain relaxation. If the thickness of the InGaAs layer exceeds the critical film thickness, strain relaxation is caused by the generation of dislocations, and lattice defects are generated in the crystal. Lattice defects act as recombination centers for conduction electrons (minority carriers) and holes moving in the base region. In addition, the energy released during recombination further proliferates lattice defects. Compared with transistors using majority carriers such as field effect transistors, in bipolar transistors that use minority carriers such as conduction electrons in the base layer for transistor operation, it is necessary to more strictly suppress the generation of lattice defects caused by strain relaxation. Therefore, the thickness of the InGaAs base layer is sufficiently thinner than the critical film thickness. For example, when the composition of the InGaAs base layer is described as In j Ga 1-j In the case of As, when the In content j is 0.1, the thickness of the base layer is set to be approximately 40 nm or less, and when the In content j is 0.07, the thickness of the base layer is set to be approximately 60 nm or less.

[0004] For GaAsHBTs that use pseudo-grown InGaAs for the base layer, although there are difficulties in device manufacturing or structural limitations, in addition to the generally known advantages such as high mobility of conduction electrons, there are also the following advantages obtained because the band gap of InGaAs is smaller than that of GaAs. In the operation of a bipolar transistor, the emitter region is grounded and a positive voltage is applied to the base region. The voltage between the base emitter and the emitter used to set the electron current from the emitter region through the base region to the collector region above a certain threshold is called the turn-on Vbe. The turn-on Vbe tends to become smaller as the band gap of the semiconductor used for the base region becomes smaller. Figures 36(a) and 36(b) respectively show the depth dependence of the conduction band energy and the valence electron band energy calculated by two-dimensional device simulation for a GaAsHBT whose base region includes a GaAs layer and an InGaAs layer. In the example shown in Figure 36(b), the composition of InGaAs is described as InGaAs. j Ga 1-jAs is 0.1. The horizontal axis is the depth in the direction of the base collector region measured with the inside of the emitter layer as the origin. In the GaAsHBT using InGaAs for the base layer shown in Figure 36(b), a composition gradient layer (represented as "composition gradient InGaAs collector layer" in the figure) in which the In concentration changes continuously from GaAs to the In concentration of the InGaAs base layer is inserted between the GaAs collector layer and the InGaAs base layer. The composition is described as In k Ga 1-k The In content k in the case of As changes continuously in a manner that it is approximately 0 on the GaAs collector layer side and approximately 0.1 on the InGaAs base layer side. The reason for this is to smoothly connect the energy of the conduction band from the collector region to the base region. As a result, the height of the energy peak between the base and collector regions of the conduction band shown in the figure is reduced, and the injection of conduction electrons from the base region to the collector region is not easily blocked. The p-type impurity concentration of the base layer is 4×10 19 cm -3 As shown by the bold downward arrow in FIG. 36( b ), in a GaAsHBT using InGaAs for the base layer, the conduction band energy of the base layer is 88 meV lower than when GaAs is used for the base layer, as the band gap decreases, and the on-state Vbe can be reduced. Reducing the on-state Vbe is preferred in terms of application to high-frequency power amplifiers. The reason for this is described below. Fig.37 The following is a diagram schematically showing the characteristics of the collector current (Ic) and the collector-emitter voltage (Vce) of the GaAsHBT for multiple different base currents (Ib) and the load curve (thick solid line). As the size of the transistor, the emitter area is assumed to be 360 ​​square microns. The so-called load curve is the trajectory of the voltage and current applied to the load on the output side of the transistor at the operating frequency during high-frequency operation, and depicts a circular curve. Fig.37 The three curves in the figure are respectively when the DC bias voltage Vce0 of the collector region is 1V, 3V, and 5V. Regarding the load curve, Ic is suppressed low in the high Vce region, while on the other hand, Ic is suppressed low in the low Vce region (Vce is about 0.5V). Fig.37 The load curve is realized by setting the DC bias voltage Vbe0 of the base region so that Ic becomes low during the rest period, thereby short-circuiting the second harmonic component of the output. Fig.37The load curve shown is the case where the output is saturated and PAE reaches the maximum. Improving PAE during saturated operation is important for power amplifiers that are based on operation at saturated output. In particular, in envelope tracking type power amplifiers, Vce0 is dynamically changed according to the amplitude change of the modulation signal, and Vce0 is increased at high output and reduced at low output. As a result, the transistor always operates in the saturation region where PAE reaches the maximum, so PAE can be improved not only at high output but also at low output. In order to improve the performance of envelope tracking type power amplifiers, it is necessary to obtain good characteristics at each point on the load curve when PAE is maximum, but it is particularly important to improve the characteristics of the transistor in the region indicated as the "low Vce region" in the figure. In the "low Vce region", Vce is about 0.5V. On the other hand, in order to turn on the transistor and flow the collector current Ic, the base-emitter voltage Vbe becomes a positive voltage much higher than 0.5V. For example, Vbe for Ic=0.01A is usually 1.25V or more. Therefore, the voltage (Vbc) applied to the pn junction between the base collectors becomes a high forward voltage. Therefore, the capacitance Cbc between the base collectors becomes very large, which becomes the main reason for the reduction of PAE and power gain in the output saturation region. However, when InGaAs is used for the base layer to reduce the on-state Vbe, the Vbe required to obtain the same Ic becomes smaller, so the forward voltage between the base collectors becomes smaller. As a result, the increase of Cbc is suppressed, which can suppress the degradation of the characteristics of the power amplifier when the output is saturated.

[0005] Next, the prior art is described for the emitter region of GaAsHBT. Usually in HBT, a semiconductor with a band gap larger than that of the base layer is used in the emitter layer. As a result, the injection of holes existing in the base layer into the emitter layer is suppressed. Therefore, even when the holes in the base layer are set to a high concentration in order to reduce the sheet resistance of the base layer, the hole current flowing from the base layer to the emitter layer can be suppressed to a low level, and the ratio of the collector current to the base current, that is, the current amplification factor, can be maintained high. AlGaAs or InGaP are known as semiconductor materials with a small lattice mismatch with GaAs and a large band gap, but InGaP with excellent surface chemical stability has been widely used as the emitter layer of GaAsHBT in recent years. By describing the composition as In i Ga 1-iThe In content i at P is set to 0.48 to 0.5, so that the lattice constant is roughly the same as that of GaAs. The InGaP layer used for GaAsHBT is usually made by metalorganic chemical vapor deposition (MOCVD), and the band gap (Eg) calculated from the wavelength of light emitted by photoluminescence is about 1.89eV at room temperature (Non-patent Document 3). Compared with the Eg of GaAs being 1.42eV, the Eg of InGaAs is lower than 1.42eV, which is a sufficiently large value.

[0006] The Eg of InGaP previously used in the emitter region of GaAsHBT is 1.89 eV as described above, but as a prior art of the present invention, a special InGaP described below is known. In the InGaP usually obtained, In and Ga irregularly occupy the lattice positions of the Group III elements. However, by using special growth conditions or growth methods, it is possible to grow the following crystal, which has a CuPt-type order in which monolayers of GaP and InP are alternately arranged in the (111) direction of the crystal. The usual InGaP is roughly disordered, as described above, and when an In content that matches the lattice of GaAs can be obtained, that is, the composition is described as In. i Ga 1-i When the In content i at P is in the range of 0.48 to 0.5, Eg is about 1.89 eV at room temperature. Hereinafter, InGaP with Eg = 1.89 eV that matches the GaAs lattice will be referred to as "existing InGaP". On the other hand, Eg decreases as CuPt-type ordering progresses. Non-patent document 4 reports the following method: InGaP with CuPt-type ordering is grown by increasing the growth temperature in the MOCVD method. Figure 4 (FIG.4) shows the growth temperature dependence of Eg at room temperature. As the growth temperature rises, Eg decreases, and becomes Eg = 1.86 eV at a growth temperature of 650°C. However, if the growth temperature becomes higher, the CuPt-type order decreases again, and Eg rises. Furthermore, as a method for growing InGaP with CuPt-type order, the following InGaP growth method is reported. This InGaP growth method utilizes the so-called atomic layer epitaxial growth method (non-patent document 5) of alternately supplying In raw material gas and Ga raw material gas. According to the method, InGaP with order progressing to Eg = 1.76 eV can be obtained. Such InGaP with CuPt-type order has been known as a material before, but there is no example of its use as an emitter material for GaAsHBT using pseudomorphically grown InGaAs as a base layer.

[0007] [Prior art literature]

[0008] [Patent Document]

[0009] [Patent Document 1] Japanese Patent No. 6378398

[0010] [Non-patent literature]

[0011] [Non-patent document 1] PM Enquist, LR Ramberg, FE Najjar, WJ Schaff and LF Eastman, Applied Physics Letters, Vol. 49, No. 3, July 21, 1986, p. 179.

[0012] [Non-patent document 2] D.A. Ahmari, M.T. Fresina, Q.J. Hartmann, D.W. Barlage, P.J. Mares, M. Feng and G.E. Stillman, IEEE Electron Device Letters, Vol. 17, No. 5, May 1996, p. 226.

[0013] [Non-patent document 3] M. Zachau and WT Marcelink, Applied Physics Letters, Vol. 60, No. 17, April 27, 1992, p. 2098.

[0014] [Non-patent document 4] Q. Liu, S. Derksen, W. Prost, A. Lindner and F. J. Tegude, Journal of Applied Physics, Vol. 79, No. 1, January 1, 1996, p. 305.

[0015] [Non-patent document 5] N. El-Masry and SM Bedair, Optical and Quantum Electronics, Vol. 23, 1991, S909. Summary of the invention

[0016] [Problems to be solved by the invention]

[0017] As described above, GaAsHBTs using pseudo-grown InGaAs for the base layer can be expected to have improved characteristics for applications such as power amplifiers, compared to the case where GaAs is used for the base layer. However, there are the following problems with the emitter layer. InGaAs has a smaller band gap than GaAs. Therefore, when InGaP is used for the emitter layer, the discontinuity of the conduction band energy at the InGaP / InGaAs heterojunction interface is greater than that at the InGaP / GaAs heterojunction interface. In Figures 38(a) and 38(b), the depth dependence of the conduction band energy near the emitter-base junction calculated by two-dimensional device simulation is shown for GaAsHBTs using GaAs and InGaAs for the base layer, respectively. There is a triangular spike-shaped energy barrier on the emitter layer side of the emitter layer / base layer heterojunction interface, and its height is as low as 24meV in the case of the GaAs base layer. This value is less than the thermal energy unit kT (k is the Boltzmann constant, T is the absolute temperature) at room temperature (300K), which is 26meV. However, near the top of the triangular energy barrier, the penetration of electrons due to tunneling occurs, so the actual barrier height is further reduced. Therefore, in the case of the GaAs base layer, the peak-shaped energy barrier is so small that it can be ignored as an obstacle to electron conduction. However, in the case of the InGaAs base, the conduction band energy decreases as the band gap decreases, so as shown in Figure 38 (b), the height of the peak-shaped energy barrier at the emitter layer / base layer heterojunction interface is as high as 105meV. This value is equivalent to 4 times the thermal energy unit at room temperature. In order for the conduction electrons to flow from the emitter layer to the base layer and reach the collector region, they must cross the energy barrier. When there is no energy barrier, as described above, since the band gap of InGaAs is smaller than that of GaAs, the on-state Vbe is reduced, and good transistor characteristics are obtained. However, when InGaP is used for the emitter layer, the on-state Vbe is determined by the energy barrier, which greatly impairs the bandgap reduction effect brought by the InGaAs base layer. The object of the present invention is to provide a new emitter base semiconductor layer structure that solves the above-mentioned problem of GaAsHBT using pseudo-grown InGaAs for the base layer.

[0018] [Technical means to solve the problem]

[0019] In order to solve the problems of the GaAsHBT using pseudo-grown InGaAs for the base layer, the GaAsHBT of the first invention of the present application uses InGaP having a CuPt-type order for the emitter layer.

[0020] In a more preferred embodiment of the first invention of the present application, the band gap of the InGaP having the CuPt-type order at room temperature is in the range of 1.76 eV to 1.86 eV.

[0021] In a more preferred embodiment of the first invention of the present application, the composition of the InGaP having the CuPt type order is described as In i Ga 1-i The In content i in P is set in the range of 0.48 to 0.5.

[0022] In a more preferred embodiment of the first invention of the present application, the composition of the InGaAs base layer is described as In j Ga 1-j The In content j in the case of As is set in the range of 0.05 to 0.12.

[0023] In order to solve the problems of the GaAsHBT using pseudo-grown InGaAs for the base layer, the GaAsHBT of the second invention of the present application makes the p-type impurity concentration on the InGaP emitter layer side (hereinafter referred to as the second InGaAs base layer) lower than the p-type impurity concentration on the collector layer side (hereinafter referred to as the first InGaAs base layer) of the pseudo-grown InGaAs base layer.

[0024] In a preferred embodiment of the second invention of the present application, the p-type impurity concentration of the second InGaAs base layer is set to 1 / 2 or less compared to the p-type impurity concentration of the first InGaAs base layer.

[0025] In a more preferred embodiment of the second invention of the present application, the p-type impurity concentration of the first InGaAs base layer is set to 2×10 19 cm -3 above.

[0026] In a more preferred embodiment of the second invention of the present application, the thickness of the second InGaAs base layer is set to be in the range of 1 nm to 6 nm.

[0027] In a more preferred embodiment of the second invention of the present application, the first InGaAs base layer and the second InGaAs base layer both describe the InGaAs composition as In j Ga 1-j The average value of the In content j in the case of As is set in the range of 0.05 to 0.12.

[0028] In a more preferred embodiment of the second invention of the present application, an InGaP layer is used for the emitter layer.

[0029] [Effects of the Invention]

[0030] According to the present invention, in a GaAsHBT using pseudo-grown InGaAs as the base layer, the influence of the peaked conduction band barrier at the emitter / base junction interface can be reduced, thereby reducing the on-state Vbe, thereby realizing a high-frequency power amplifier with excellent basic characteristics such as PAE. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 FIG. 1 is a diagram showing the semiconductor layer structure of the main components of the GaAs HBT according to the first invention of the present application.

[0032] Figure 2 To indicate that Figure 1 FIG. 1 is a cross-sectional structure of a GaAs HBT according to the first invention of the present application, showing main components.

[0033] Figure 3 To express Figure 2 FIG. 1 is a graph showing the depth dependence of the conduction band energy, valence band energy, and Fermi level of the GaAs HBT according to the first invention of the present application.

[0034] Figure 4 To express Figure 2 FIG. 1 is a graph showing the depth dependence of the conduction band energy of the GaAs HBT according to the first invention of the present application.

[0035] Figure 5 To express Figure 2 FIG. 4 is a graph showing the relationship between the peak energy barrier height generated at the junction of the emitter region and the base region of the GaAs HBT according to the first invention of the present application and the band gap Eg of the InGaP emitter region layer.

[0036] Figure 6 To express in Figure 2 FIG. 2 is a graph showing the Vbe dependency of the collector current Ic when the same positive voltage Vbe is applied to the base and the collector in the GaAs HBT according to the first invention of the present application.

[0037] Figure 7 To express Figure 2 FIG. 4 is a graph showing the relationship between the on-state Vbe of the GaAsHBT according to the first invention of the present application and the band gap Eg of the InGaP emitter layer.

[0038] Figure 8 To express Figure 2 FIG. 1 is a graph showing the dependence of the base collector capacitance Cbc of the GaAs HBT according to the first invention of the present application on the collector current Ic.

[0039] Fig. 9 To express Figure 2 FIG. 1 is a graph showing the dependence of the power gain MSG / MAG of the GaAsHBT according to the first invention of the present application on the collector current Ic.

[0040] Fig.10 To express Figure 2 FIG. 1 is a diagram showing the relationship between the base collector capacitance Cbc and the band gap Eg of the InGaP emitter layer of the GaAsHBT according to the first invention of the present application.

[0041] Fig.11 To express Figure 2 FIG. 4 is a graph showing the relationship between the power amplification factor MAG of the GaAsHBT according to the first invention of the present application and the band gap Eg of the InGaP emitter layer.

[0042] Fig.12 This is a diagram showing a circuit for analyzing characteristics of a power amplifier including a GaAsHBT.

[0043] Fig.13 To indicate that Figure 2 The GaAsHBT power amplifier of the first invention of this application is shown in Fig.12 The characteristic analysis circuit shown is a graph showing the time variations of the collector current Ic and the collector-emitter voltage Vce during operation.

[0044] Fig.14 To indicate that Figure 2 The GaAsHBT power amplifier of the first invention of this application is shown in Fig.12 The characteristic analysis shown is a graph showing the output power dependency of the power added efficiency PAE obtained when the circuit is operated.

[0045] Fig.15 To indicate that Figure 2 The GaAsHBT power amplifier of the first invention of this application is shown in Fig.12 The graph shown is the output power dependence of the power amplification obtained when the characteristic analysis circuit is operated.

[0046] Fig.16 FIG. 1 is a diagram showing the semiconductor layer structure of main components of another embodiment of the GaAs HBT according to the first invention of the present application.

[0047] Fig.17 FIG. 1 is a diagram showing the semiconductor layer structure of the main components of the GaAs HBT according to the second invention of the present application.

[0048] Fig.18 To indicate that Fig.17 FIG. 1 is a cross-sectional structure of a GaAs HBT according to the second invention of the present application, showing main components thereof.

[0049] Fig.19 To express Fig.18 FIG. 1 is a graph showing the depth dependence of the conduction band energy, valence band energy, and Fermi level of the GaAs HBT according to the second invention of the present application.

[0050] Fig. 20 To express Fig.18 FIG. 1 is a graph showing the depth dependence of the conduction band energy of the GaAsHBT according to the second invention of the present application.

[0051] Fig.21 To express in Fig.18 FIG. 2 is a graph showing the Vbe dependency of the collector current Ic when the same positive voltage Vbe is applied to the base and the collector in the GaAs HBT according to the second invention of the present application.

[0052] Fig. 22 To express Fig.18 FIG. 2 is a graph showing the relationship between the on-state Vbe of the GaAsHBT according to the second invention of the present application and the thickness X of the second InGaAs base layer 240 .

[0053] Fig.23 To express Fig.18 Graph showing the relationship between the on-state Vbe of the GaAsHBT according to the second invention of the present application and the ratio Np2 / Np1 of the p-type impurity concentrations of the second InGaAs base layer 240 and the first InGaAs base layer 230 .

[0054] Fig.24 To express Fig.18 FIG. 1 is a graph showing the dependence of the base collector capacitance Cbc of the GaAs HBT according to the second invention of the present application on the collector current Ic.

[0055] Fig.25 To express Fig.18 FIG. 2 is a graph showing the dependence of the power gain MSG / MAG of the GaAsHBT according to the second invention of the present application on the collector current Ic.

[0056] Fig.26 To express Fig.18 FIG. 4 is a graph showing the relationship between the base collector capacitance Cbc and the thickness X of the second InGaAs base layer 240 of the GaAsHBT according to the second invention of the present application.

[0057] Fig. 27 To express Fig.18 FIG. 2 is a graph showing the relationship between the power amplification factor MAG of the GaAsHBT according to the second invention of the present application and the thickness X of the second InGaAs base layer 240 .

[0058] Fig.28 To express Fig.18 FIG. 2 is a graph showing the relationship between |H21|2 and the thickness X of the InGaAs base layer 240 of the GaAsHBT according to the second invention of the present application.

[0059] Fig.29 To express Fig.18 The GaAsHBT of the second invention of this application is shown in Fig. 20 A graph of the depth dependence of the conduction band energy for a different alternative embodiment.

[0060] Fig.30 To express Fig.18 1 is a graph showing the relationship between the base collector capacitance Cbc of the GaAsHBT according to the second invention of the present application and the ratio Np2 / Np1 of the p-type impurity concentration of the second InGaAs base layer 240 to the first InGaAs base layer 230.

[0061] Fig.31 To express Fig.18 FIG. 2 is a graph showing the relationship between the power amplification factor MAG of the GaAsHBT according to the second invention of the present application and the ratio Np2 / Np1 of the p-type impurity concentration of the second InGaAs base layer 240 to the first InGaAs base layer 230 .

[0062] Fig.32 This is a diagram showing the depth dependency of the conduction band energy of a GaAsHBT as an example in which the first invention of the present application and the second invention of the present application are combined.

[0063] Fig.33 This is a diagram showing the Vbe dependency of the collector current Ic when the same positive voltage Vbe is applied to the base and the collector in a GaAs HBT as an example combining the first invention of the present application and the second invention of the present application.

[0064] Fig.34 Graph showing the collector current Ic dependency of the base collector capacitance Cbc of a GaAs HBT as an example combining the first invention of the present application with the second invention of the present application.

[0065] Fig.35 Graph showing the collector current Ic dependency of the power gain MSG / MAG of a GaAsHBT as an example combining the first invention of the present application with the second invention of the present application.

[0066] FIG. 36( a ) is a diagram showing the depth dependence of the conduction band energy, valence electron band energy, and Fermi level of a conventional GaAs HBT using GaAs for a base layer.

[0067] FIG36( b ) is a diagram showing the depth dependence of the conduction band energy, valence electron band energy, and Fermi level of a conventional GaAs HBT using InGaAs for a base layer.

[0068] Fig.37The figure schematically shows a curve (thin solid line) representing the current-voltage characteristics of the collector current (Ic) and the collector-emitter voltage (Vce) of the GaAsHBT for a plurality of different base currents (Ib), and a load curve (thick solid line) when the transistor operates as a power amplifier.

[0069] FIG. 38( a ) is a diagram showing the depth dependence of the conduction band energy of a conventional GaAs HBT using GaAs for a base layer.

[0070] FIG38(b) is a diagram showing the depth dependence of the conduction band energy of a conventional GaAsHBT using InGaAs for the base layer.

[0071] [Explanation of Symbols]

[0072] 100, 200: GaAs substrate

[0073] 110, 210: Sub-collector layer

[0074] 120, 220: Collector layer

[0075] 121, 221: First collector layer

[0076] 122, 222: Second collector layer

[0077] 130: InGaAs base layer

[0078] 150: First emission layer

[0079] 151: Second emission layer

[0080] 160, 260: Emitting area contact layer

[0081] 161: High concentration GaAs emitter contact layer

[0082] 162, 261: GaAs emitter contact layer

[0083] 163, 262: InGaAs emitter contact layer

[0084] 170, 270: emitter

[0085] 180, 280: Base

[0086] 190, 290: Collector

[0087] 199, 299: Arrows indicating the width of the emission area

[0088] 230: First InGaAs base layer

[0089] 240: Second InGaAs base layer

[0090] 250: Emission zone layer. DETAILED DESCRIPTION

[0091] Hereinafter, embodiments of the present invention will be described with reference to the drawings.

[0092] Figure 1 The semiconductor layer structure of the main components of the GaAs HBT of the first invention of the present application is shown. On the GaAs substrate 100, a sub-collector region layer 110, a collector region layer 120, an InGaAs base region layer 130, a first emitter region layer 150 and an emitter region contact layer 160 are formed by epitaxial growth. The surface orientation of the GaAs substrate 100 is (100). The sub-collector region layer 110, the collector region layer 120, the first emitter region layer 150 and the emitter region contact layer 160 are n-type semiconductors, and the InGaAs base region layer 130 is p-type. The sub-collector region layer 110, the collector region layer 120, the InGaAs base region layer 130 and the first emitter region layer 150 constitute the intrinsic part of the npn type transistor, and these layers all use semiconductor layers without relaxation of lattice strain caused by lattice mismatch with GaAs. For the sub-collector layer 110 and the collector layer 120, for example, GaAs, which is the same material as the substrate, is used. Alternatively, other semiconductors that have little lattice mismatch with GaAs, such as InGaP, AlGaAs, or a stacked structure of these semiconductors, may also be used. The InGaAs base layer 130 is grown by pseudomorphic growth without strain relaxation while maintaining compressive strain. For the first emitter layer 150, InGaP with CuPt-type order, which is a feature of the first invention of this application, is used. The so-called CuPt-type order, as described above, is a structure in which monolayers of GaP and InP are alternately arranged in the (111) direction of the crystal. The band gap (Eg) of existing InGaP that is lattice-matched with GaAs is 1.89 eV, but for InGaP with CuPt-type order, Eg is lower than 1.89 eV and becomes smaller as the order increases. The composition of the fully ordered ideal crystal is recorded as In i Ga 1-i When the In content i is 0.5, it becomes a compound represented by InGaP2, but in practice, there is no problem as long as i is in the range of 0.48 to 0.5. If i is in the above range, the lattice mismatch with GaAs is almost eliminated, so it is easy to grow a layer without strain relaxation. The effect obtained by using InGaP with CuPt type order for the first emitter layer 150 will be described below.

[0093] Figure 2 The cross-sectional structure of the GaAsHBT shown in FIG. Figure 1The GaAsHBT shown as the main component is used as a typical example of a power amplifier. In this embodiment, the emission region is divided into three regions: Figure 2 The width of each emission region indicated by arrow 199 is 3 micrometers. In addition, the emission region size (length) in the direction perpendicular to the paper surface is 40 micrometers. Therefore, the emission region area is 360 square micrometers. Figure 2 In the GaAs HBT shown, an emitter 170 is provided on the emitter contact layer 160, a base 180 is provided on the InGaAs base layer 130, and a collector 190 is provided on the sub-collector layer 110, and electrical connections with various regions of the transistor are obtained through these electrodes. Figure 2 In the embodiment, the collector layer 120 includes a two-layer stacked structure, and the emitter contact layer 160 includes a three-layer stacked structure. Table 1 shows the conductivity type, impurity type and concentration, and layer thickness of each semiconductor layer including these stacked structures. The first collector layer 121 includes GaAs. For the second collector layer 122, a gradient is provided for the In content in such a way that the composition of the semiconductor becomes GaAs on the first collector layer 121 side and becomes InGaAs on the InGaAs base layer 130 side. In other words, the composition described as In k Ga 1-k The In content k in the case of As changes continuously so that it becomes approximately 0 on the first collector layer 121 side and approximately 0.1 on the InGaAs base layer 130 side. This is because the conduction band can be smoothly connected from the first collector layer 121 to the InGaAs base layer 130. The p-type impurity of the InGaAs base layer 130 is carbon (C) with a concentration of 4×10 19 cm -3 In the embodiment shown in Table 1, the In content of the InGaAs base layer, that is, the composition is recorded as In j Ga 1- j As, the In content j is 0.1 and the thickness is 35nm. As described in the prior art, the thickness of the InGaAs base layer is selected to a preferred value according to the In content. The first emitter layer 150 is InGaP with CuPt type order as a feature of the present invention, and the composition is described as In i Ga 1-iThe In content i at P is 0.48. InGaP with CuPt-type order has spontaneous polarization. When used for the first emitter layer 150, negative charges caused by spontaneous polarization are generated on the surface of the emitter contact layer side. The high-concentration GaAs emitter contact layer 161 is a layer that increases the concentration of Si as an n-type impurity in order to offset the negative charges caused by spontaneous polarization, thereby suppressing the conduction band energy from rising due to the negative charges caused by spontaneous polarization. In this embodiment, except for the uppermost InGaAs emitter contact layer 163, all semiconductor layers on the GaAs substrate 100 are grown without substantially relaxing the strain caused by the lattice mismatch with GaAs. As a result, an element with excellent characteristics and no lattice defects caused by strain relaxation can be obtained. On the other hand, the InGaAs emitter contact layer 163 is used to obtain a lattice mismatch with GaAs. Figure 2 The content of In is increased sufficiently to provide good ohmic contact to the emitter 170 shown in FIG. Therefore, strain relaxation occurs in the InGaAs emitter contact layer 163 to generate lattice defects, but by increasing the thickness of the underlying GaAs emitter contact layer 162 to 110 nm, the influence of the lattice defects is prevented from affecting the intrinsic region of the element and degrading the characteristics.

[0094] The effects of the first invention of this application are described below. Figure 2 The structure of the band end energy of the GaAsHBT calculated by two-dimensional device simulation shown in Table 1 is explained. Figure 3 The depth dependence of the conduction band energy, the valence electron band energy and the Fermi level from the first emitter layer 150 to the InGaAs base layer 130 , the second collector layer 122 and the first collector layer 121 is shown. Figure 4 This is a diagram showing the depth dependence of the conduction band energy at the junction part. No voltage is applied to the transistor, so Figure 3 As shown in FIG. 1 , the Fermi level is constant in the depth direction. Figure 3 , Figure 4 In the figure, the portion representing the first emitter layer 150 is shown with a dotted line and a solid line, but the dotted line is the case where the first emitter layer 150 is replaced with an emitter layer obtained by conventional InGaP, and the height of the spike-shaped energy barrier is 105 meV as shown in FIG38(b) as a conventional example. On the other hand, the solid line is the case of the GaAsHBT of the present invention, where InGaP having a CuPt-type order with a band gap (Eg) of 1.81 eV at room temperature is used for the first emitter layer 150. First, Figure 3The valence band energy shown can be explained. The valence band energy of the portion represented as the first emitter layer 150 is almost the same when it is replaced by the existing InGaP layer and the InGaP layer with CuPt-type order of the present invention, so it is shown overlapping. This is a characteristic of the CuPt-type order of InGaP. The reduction in the band gap caused by ordering is almost entirely caused by the reduction in conduction band energy, and the energy of the valence band is unchanged (Non-Patent Document 4). The use of a barrier formed by the valence band of the emitter layer to suppress the injection of holes in the base layer into the emitter layer is a characteristic of HBT, but the InGaP emitter layer of the present invention can also obtain the same effect as the case of using the existing InGaP emitter layer. Next, focus on the conduction band energy. The height of the spike-shaped energy barrier at the interface between the first emitter layer 150 and the InGaAs base layer 130 is shown in Figure 4 As shown, when the first emitter layer 150 is replaced with conventional InGaP (dashed line), it is 105 meV. In contrast, in the present invention in which InGaP having a CuPt-type order is used for the emitter layer, it is significantly reduced (solid line). Figure 4 When Eg is 1.81 eV, the height of the spike energy barrier is 25 meV, which is the same height as the spike energy barrier of the GaAs base layer shown in FIG38(a), and the influence of the barrier almost disappears. Figure 5 To express Figure 4 The graph of the relationship between the peak barrier height and the Eg of InGaP with CuPt type order used for the first emitter layer 150 is shown. The point where Eg is 1.89 eV corresponds to the case where the existing InGaP is used for the emitter layer. When InGaP with CuPt type order is used for the emitter layer, Eg decreases according to the degree of ordering, and the peak barrier height decreases accordingly. Figure 4 In the data points where Eg is 1.86 eV or less, the data points are indicated as "the present invention". The reason for this is that when Eg is 1.86 eV or less, a particularly significant improvement in characteristics can be obtained, which will be described in detail in Fig.11 Described in the instructions.

[0095] Table 1

[0096]

[0097] Next, Figure 2 The electrical direct current characteristics (DC characteristics) of the GaAsHBT shown in Table 1 calculated by two-dimensional device simulation will be described. Figure 6 Indicated in Figure 2The Vbe dependence of the collector region current Ic when the emitter 170 is grounded and the same positive voltage Vbe is applied to the base 180 and the collector 190 in the GaAsHBT shown is a characteristic of Ic obtained by measurement using a so-called Gummel Plot. Figure 3 , Figure 4 Similarly, the dotted line represents the case where the first emitter layer 150 is replaced with an emitter layer obtained by conventional InGaP. On the other hand, the solid line represents the case where the GaAsHBT of the present invention uses InGaP having a CuPt-type order with a band gap of 1.81 eV at room temperature for the first emitter layer 150. The Vbe at which Ic reaches a certain predetermined value is the on-state Vbe. Figure 6 The two Ic curves shown are horizontally shifted in parallel by 75mV. That is, the on-state Vbe of the present invention is independent of the Ic that defines the on-state Vbe and is 75mV lower than before. This change is caused by Figure 4 The height of the spike-shaped energy barrier shown is caused by a decrease of 80 meV from 105 meV to 25 meV. Figure 7 Graph showing the relationship between on-state Vbe and Eg of InGaP. The point where Eg is 1.89 eV is the case where the first emitter layer 150 is replaced with an emitter layer obtained from the existing InGaP. As Eg decreases, that is, the CuPt order increases, on-state Vbe decreases. Figure 7 In the data points where Eg is 1.86 eV or less, the data points are indicated as "the present invention". The reason for this is that when Eg is 1.86 eV or less, a particularly significant improvement in characteristics can be obtained, which will be described in detail in Fig.11 Described in the instructions.

[0098] Next, Figure 2 The high frequency characteristics of the GaAsHBT calculated by two-dimensional device simulation shown in Table 1 are described. Figure 8 This is a graph showing the dependence of base-collector capacitance Cbc on collector current Ic. The collector-emitter voltage Vce is fixed at 0.5 V, and the base-emitter voltage Vbe is varied to obtain Ic dependence. Cbc is calculated using the following formula from the Y parameter at frequency f0 = 2 GHz.

[0099] Cbc=-Im(Y12) / (2πf0)

[0100] Here, Im(Y12) is the imaginary part of the Y parameter component Y12, and π is the circumference of a circle. The dotted line is the case where the first emitter layer 150 is replaced with an emitter layer obtained by conventional InGaP, and the solid line is the case of the present invention, that is, the InGaP emitter having a CuPt type order, indicating that Eg is 1.81 eV at room temperature. Figure 8The data points shown have an Ic of 0.01A or more, all entering Fig.37 As shown in the "low Vbe region". Figure 6 As we know, when Ic becomes 0.01A or more, Vbe exceeds 1.25V. On the other hand, Vce is 0.5V, so Vbc is 0.75V or more, that is, the pn junction between the base and collector is in a high forward bias state. Therefore, Figure 8 In the case of the conventional GaAsHBT shown by the middle dotted line, as Ic increases, that is, as the forward bias increases, Cbc becomes a high value exceeding 1pF. In contrast, the Cbc of the GaAsHBT of the present invention is significantly reduced compared to the past, and the higher the Ic, the more obvious the reduction. The reason is that the forward bias between the base and collector regions becomes smaller due to the reduction in the on-state Vbe. Fig. 9 The maximum power gain at frequency f0 = 2 GHz is dependent on Ic. Figure 8 Similarly, the collector voltage Vce is 0.5V. The vertical axis is expressed as MSG / MAG, but this is the abbreviation of Maximum Stable Gain (MSG) and Maximum Available Gain (MAG). MAG is the maximum power gain defined when the transistor meets the stable condition without oscillation. On the other hand, in the area where the transistor does not meet the stable condition, the maximum power gain is defined as the value obtained when forced stabilization is achieved by external components (MSG). Fig. 9 In the figure, the slope of the curve changes sharply at Ic=0.05A, but below 0.05A is the MSG region, and above 0.05A is the MAG region. Fig. 9 In the figure, the solid line is the GaAsHBT of the present invention, which shows that the Eg of the InGaP emitter layer at room temperature is 1.81eV. Compared with the existing GaAsHBT shown by the dotted line, MSG / MAG is higher at any current. This is due to Figure 8 The reduction of Cbc shown in FIG. 1 is caused by the reduction of Cbc. Cbc acts as a feedback capacitor in the power amplification of the transistor and reduces the power amplification factor. In particular, in the "low Vce region", the pn junction between the base and collector regions becomes a strong forward bias state, so Cbc becomes very large, which becomes the dominant factor for reducing the power amplification factor. According to the present invention, Cbc can be greatly reduced in the "low Vce region", and the reduction of the power amplification factor in this region can be suppressed. Fig.10 represents the relationship between Cbc and Eg, Fig.11 It shows the relationship between the maximum power gain and Eg. Vce is 0.5V. In addition, the black circle is the value when Ic is 0.072A, and the black triangle is the value when Ic is 0.144A. Ic=0.072A, 0.144A both enter Fig.37The “low Vce region” shown in the figure, but the area of ​​the emitter region is 360 square microns, so if converted to the average current density of the emitter region, it is equivalent to 20kA / cm 2 , 40kA / cm 2 . Fig.11 In the case of Ic=0.072A and 0.144A, the maximum power amplification factor is in the MAG region, so the vertical axis is represented by MAG. Fig.10 and Fig.11 The data point indicated as "existing" is the case where the first emitter layer 150 is replaced by the existing InGaP GaAsHBT. As the CuPt-type ordering of InGaP progresses, Eg decreases, Cbc decreases, and MAG increases. In order to obtain a significant effect of the present invention, it is preferred that Ic = 0.072A (20kA / cm 2 ) when MAG improves by more than 1dB, Ic=0.144A(40kA / cm 2 ) when MAG improves the existing by more than 2dB. Fig.11 The Eg range that satisfies the above conditions, i.e., 1.86 eV or less, is indicated as "the present invention". Figure 5 , Figure 7 , Fig.10 Similarly, the region where Eg is 1.86 eV or less is indicated as "the present invention". In addition, the growth of InGaP with CuPt-type ordering progressing to Eg = 1.76 eV is reported in Non-Patent Document 5. The first invention of this application also can obtain sufficient effects if 1.76 eV ≤ Eg ≤ 1.86 eV.

[0101] Next, Figure 2 The improvement in characteristics when the GaAsHBT according to the first invention of the present application is used in a power amplifier as shown in Table 1 will be described. Fig.12This is a circuit for analyzing the characteristics of a power amplifier including GaAsHBT, which is a so-called load-pull / source-pull circuit. A ballast circuit including a resistor and a capacitor is connected to the base 180 connected to the RF input terminal to stabilize the transistor against oscillation and temperature rise. Furthermore, on the RF input terminal side, an active tuner is inserted through a DC block capacitor that blocks the DC voltage and only allows the RF signal to pass. The source tuner is used to obtain impedance matching with the input impedance of the transistor to increase the power amplification factor. On the other hand, a load tuner is inserted between the collector 190 and the RF output terminal through a DC block capacitor. The load tuner is used to obtain impedance matching with the output side impedance of the transistor to increase the peak value of the power added efficiency (PAE) when the RF output is saturated. Furthermore, in order to suppress the power loss caused by the second harmonic (2f0 when the fundamental frequency is f0), the impedance of 2f0 ( Fig.12 Recorded as Z 2f0 ) is set to approximately zero. The impedance of other harmonic components is set to 50 ohms. A DC bias voltage Vbe0 and a DC bias voltage Vce0 are applied to the base 180 and the collector 190 respectively through an RF block inductor that blocks the RF voltage and only allows the DC voltage to pass. The results described below are obtained using Figure 2 The device with an emission area of ​​360 square micrometers shown in Table 1 is used as a GaAsHBT. Vbe0 is adjusted so that Ic becomes 6 mA when there is no RF input power. The operating frequency f0 is set to 2 GHz, and the impedance of the source tuner and the load tuner is matched at Vce0 = 3V. Fig.13This is a graph in which Ic and Vce of the GaAsHBT of the present invention are plotted with time as the horizontal axis during RF operation. The band gap of InGaP as the first emission layer is 1.81eV. The solid line is Ic and the dotted line is Vce. Both the solid line and the dotted line represent three curves, but they are curves when Vce0 is 1V, 3V, and 5V respectively. In addition, these curves are signals during power saturation operation when the power added efficiency PAE reaches the maximum. The overlapping part between Ic and Vce is small, and the power loss is small. This is an effect obtained by setting the impedance of the second harmonic 2f0 to zero in the load tuner. The value of Ic at the peak is 0.05A when Vce0=1V, 0.17A when Vce0=3V, and about 0.28A when Vce0=5V. On the other hand, Vce when Ic reaches the peak is about 0.5V. In other words, when Ic is at its peak, the GaAsHBT is Fig.37 The "low Vce region" operation is described. Fig.14 represents the output power dependence of PAE, Fig.15 Represents the output power dependence of the power amplification ratio. Fig.14 , Fig.15 The characteristics are shown for the case where the DC bias voltage Vce0 of the collector region is 1V, 2V, 3V, 4V, and 5V. In any case of Vce0, the dotted lines are the case of the existing GaAsHBT in which the first emitter layer 150 is replaced with the existing InGaP layer. On the other hand, the thin solid line and the thick solid line are the case of the GaAsHBT of the present invention, the thin solid line is the case of using the CuPt-type order InGaP with a band gap Eg of 1.85eV for the first emitter layer, and the thick solid line is the case of using the CuPt-type order InGaP with an Eg of 1.81eV. Fig.14 In the figure, a line connecting the peaks of the power added efficiency is also shown. The peak value of the power added efficiency of the GaAsHBT of the present invention is higher than that of the existing GaAsHBT at any Vce0, and the maximum difference is 4% when Eg is 1.81eV. This is due to the reduction of Cbc in the "low Vce region" of the GaAsHBT of the present invention and the increase in power amplification caused by this. The lower the Vce0, the more obvious the improvement in power added efficiency. The reason is that, as shown by Fig.37 As known, if Vce0 is low, the rate at which the load curve enters the "low Vce region" increases. As also described in the prior art, an envelope tracking type power amplifier dynamically changes Vce0 according to the amplitude change of the modulation signal, increases Vce0 at high output, and reduces Vce0 at low output. Thus, the transistor is always operated in the saturation region where the power added efficiency is maximized. Fig.14As shown, the peak value of the power added efficiency of the GaAsHBT of the present invention is higher than that of the conventional GaAsHBT at any Vce0. Fig.15 The power gain ratio shown is also higher than that of the conventional GaAsHBT. Therefore, by using the GaAsHBT of the present invention, the characteristics of a power amplifier operating in the power saturation region such as a power amplifier, especially an envelope tracking amplifier, can be greatly improved.

[0102] In the embodiment of the first invention of the present application described above, the CuPt type order of InGaP used for the first emitter layer 150 is uniform within the layer and has a constant band gap. As another embodiment of the first invention of the present application, the CuPt type order can also be changed in a manner that it is high near the interface with the InGaAs base layer 130 and decreases in the area away from the interface. In this way, the polarization charge can be dispersed and the influence of the conduction band energy rise caused by the negative polarization charge can be reduced. Alternatively, as Fig.16 As shown, a second emitter layer 151 is provided between the first emitter layer 150 and the emitter contact layer 160, the first emitter layer 150 is set to InGaP with CuPt type order, and the second emitter layer 151 is set to existing InGaP. Thus, it is possible to maintain other characteristics of the GaAsHBT using existing InGaP for the emitter layer, and reduce the height of the spike-shaped energy barrier generated at the interface with the InGaAs base layer 130 to reduce the on-state Vbe. In addition, Fig.16 In the embodiment, the second emitter layer 151 may also use semiconductors other than InGaP such as AlGaAs. In addition, the n-type impurity concentration near the interface between the first emitter layer 150 and the second emitter layer 151 may be increased to reduce the influence of negative polarization charges generated on the side surface of the second emitter layer 151 of the first emitter layer 150.

[0103] In the embodiment of the first invention of the present application described above, the In content of the InGaAs base layer 130, that is, the composition is described as In j Ga 1-j Although the In content j in As is described as 0.1, the same effect can be obtained as long as j is in the range of 0.05 to 0.12. Alternatively, the average value of j in the film can be set in the range of 0.05 to 0.12, and a gradient can be added to the value of j in the direction perpendicular to the film.

[0104] Fig.17The semiconductor layer structure of the main components of the GaAsHBT of the second invention of the present application is shown. On the GaAs substrate 200, a sub-collector layer 210, a collector layer 220, a first InGaAs base layer 230, a second InGaAs base layer 240, an emitter layer 250 and an emitter contact layer 260 are formed by epitaxial growth. The surface orientation of the GaAs substrate 200 is (100). The sub-collector layer 210, the collector layer 220, the emitter layer 250 and the emitter contact layer 260 are n-type semiconductors, and the first InGaAs base layer 230 and the second InGaAs base layer 240 are p-type. The sub-collector region layer 210, the collector region layer 220, the first InGaAs base region layer 230, the second InGaAs base region layer 240 and the emitter region layer 250 constitute the intrinsic part of the npn type transistor, and these layers all use semiconductor layers that do not have relaxation of lattice strain caused by lattice mismatch with GaAs. For the sub-collector region layer 210 and the collector region layer 220, for example, GaAs that is the same as the substrate material is used. Alternatively, other semiconductors that have almost no lattice mismatch with GaAs, such as InGaP, AlGaAs, or a stacked structure of these semiconductors, can also be used. The first InGaAs base region layer 230 and the second InGaAs base region layer 240 are grown by pseudomorphic growth without strain relaxation while maintaining compressive strain. Here, as a feature of the present invention, when the p-type impurity concentration in the first InGaAs base layer 230 is set to Np1 and the p-type impurity concentration in the second InGaAs base layer 240 is set to Np2, a relationship of Np2<Np1 holds.

[0105] In a preferred embodiment of the second invention of the present application, Np2≦0.5×Np1.

[0106] In a preferred embodiment of the second invention of the present application, Np1≧2×10 19 cm -3 .

[0107] In a preferred embodiment of the second invention of the present application, the thickness of the second InGaAs base layer 240 is in the range of 1 nm to 6 nm.

[0108] In a preferred embodiment of the second invention of the present application, the first InGaAs base layer 230 and the second InGaAs base layer 240 are both described as InGaAs. j Ga 1-j The average value of the In content j in As is in the range of 0.05 to 0.12.

[0109] Fig.18 The cross-sectional structure of the GaAsHBT shown in FIG. Fig.17The GaAsHBT shown as the main component is used as a typical example of a power amplifier. In this embodiment, the emission region is divided into three regions: Fig.18 The width of each emission region indicated by arrow 299 is 3 micrometers. In addition, the emission region size (length) in the direction perpendicular to the paper surface is 40 micrometers. Therefore, the emission region area is 360 square micrometers. Fig.18 In the GaAsHBT shown, an emitter 270 is provided on the emitter contact layer 260, a base 280 is provided on the second InGaAs base layer 240, and a collector 290 is provided on the sub-collector layer 210, and electrical connections with various regions of the transistor are obtained through these electrodes. Fig.18 In the embodiment, the collector layer 220 and the emitter contact layer 260 each include a two-layer stacked structure. Table 2 shows the conductivity type, impurity type and concentration, and layer thickness of each semiconductor layer including these stacked structures. The first collector layer 221 includes GaAs. For the second collector layer 222, a gradient is provided for the In content in such a way that the composition of the semiconductor becomes GaAs on the first collector layer 221 side and becomes InGaAs on the first InGaAs base layer 230 side. In other words, the composition is described as In k Ga 1-k The In content k at As changes continuously in a manner that it becomes approximately 0 on the first collector layer side and approximately 0.1 on the first InGaAs base layer 230 side. The reason for this is that the conduction band can be smoothly connected from the first collector layer 221 to the first InGaAs base layer 230. The p-type impurities of the first InGaAs base layer 230 and the second InGaAs base layer 240 are carbon (C), and the concentrations are Np1 and Np2, respectively. As a feature of the present invention, Np2 is less than Np1. Its effect will be described below. The thickness of the entire InGaAs base layer is 35nm, of which the thickness of the second InGaAs base layer 240 is X. X is in the range of greater than 0nm and less than 35nm, but as will be explained below, particularly significant improvement in characteristics can be obtained when 1nm≦X≦6nm. The In content of the first InGaAs base layer 230 and the second InGaAs base layer 240, that is, the composition is recorded as In j Ga 1-j As, the In content j is 0.1. As described in the prior art, the thickness of the InGaAs base layer as a whole (35 nm in this embodiment) is selected to be a preferred value according to the In content. For the emitter layer 250, a common InGaP having an Eg of 1.89 eV at room temperature is used. The composition is described as In i Ga 1-iThe In content i at P is 0.48. In this embodiment, except for the uppermost InGaAs emitter contact layer 262, all semiconductor layers on the GaAs substrate 200 are grown without substantially relaxing the strain caused by the lattice mismatch with GaAs. Thus, an element with excellent characteristics and no lattice defects caused by strain relaxation can be obtained. On the other hand, the InGaAs emitter contact layer 262 is Fig.18 The content of In is sufficiently increased by making good ohmic contact with the emitter 270 shown. Therefore, strain relaxation occurs in the InGaAs emitter contact layer 262 to generate lattice defects, but by setting the thickness of the underlying GaAs emitter contact layer 261 to 120 nm, the influence of the lattice defects is prevented from affecting the intrinsic region of the element and degrading the characteristics.

[0110] Table 2

[0111]

[0112] The effects of the second invention of the present application are described below. Fig.18 The structure of the band end energy of the GaAsHBT calculated by two-dimensional device simulation shown in Table 2 is explained. Fig.19 The depth dependence of the conduction band energy, the valence electron band energy and the Fermi level from the emitter layer 250 to the second InGaAs base layer 240 , the first InGaAs base layer 230 , the second collector layer 222 and the first collector layer 221 is shown. Fig. 20 This is a diagram showing the depth dependence of the conduction band energy at the junction part. No voltage is applied to the transistor, so Fig.19 The Fermi level shown is constant in the depth direction. Fig.19 , Fig. 20 In the figure, the dotted line and the solid line are shown in the emitter layer 250 and the second InGaAs base layer 240. The dotted line is the case where X=0nm and the second InGaAs base layer 240 does not exist, that is, the case of the conventional GaAsHBT. On the other hand, the solid line is the case of the GaAsHBT of the present invention, and the thickness X of the second InGaAs base layer 240 is 2nm. In addition, the p-type impurity concentration Np1 of the first InGaAs base layer 230 is 4×10 19 cm -3 The p-type impurity concentration Np2 of the second InGaAs base layer 240 is 1×10 19 cm -3 If we focus on Fig. 20 The conduction band energy shown in the figure is as follows: Figure 4As shown in the conventional example, the peak barrier height is 105 meV. In contrast, in the case of the GaAs HBT of the present invention shown by the solid line, the peak barrier height is 42 meV, which is 63 meV lower. The reason for this is explained below. The p-type impurity concentration of the first InGaAs base layer is as high as 4×10 19 cm -3 . Generally, if the concentration of impurities in a semiconductor increases and the concentration of conductive carriers increases, the Fermi level enters the band and becomes a so-called degenerate semiconductor. Compared with GaAs, InGaAs has a lower state density of the valence band, so p-type semiconductors tend to become degenerate. Even with the same p-type impurity concentration, the Fermi level enters deeper into the band than GaAs. For example, in the conventional examples shown in Figures 36(a) and 36(b), the concentration of carbon as a p-type impurity is 4×10 19 cm -3 , but the Fermi level of the InGaAs base region shown in FIG36(b) is deeper into the valence band than the GaAs base region shown in FIG36(a). In the second invention of the present application, this feature of p-type InGaAs is used. The valence band energy measured by the Fermi level varies depending on the concentration of the p-type impurity (carbon in this embodiment). Fig.19 , Fig. 20 In the embodiment shown, the carbon concentration of the second InGaAs base layer 240 is 1×10 19 cm -3 , which is one-fourth of the carbon concentration of the first InGaAs base layer. Fig.19 As can be seen from the second InGaAs base layer 240, the energy from the Fermi level of the valence band is lowered. Therefore, the energy of the band including the conduction band is pulled downward as a whole, and as a result, the spike-shaped energy barrier generated near the interface between the emitter layer 250 and the base layer is also pulled downward. This feature is that the p-type impurity concentration in the first InGaAs base layer is 2×10 19 cm -3 becomes apparent when above.

[0113] Next, Fig.18 The electrical direct current characteristics (DC characteristics) of the GaAsHBT shown in Table 2 calculated by two-dimensional device simulation will be described. Fig.21 Indicated in Fig.18 The Vbe dependence of the collector current Ic when the emitter 270 is grounded and the same positive voltage Vbe is applied to the base 280 and the collector 290 in the GaAs HBT shown is a characteristic of Ic obtained by measuring what is called a Gummi plot. Fig.19 , Fig. 20Similarly, the dotted line represents the case where X=0 nm and there is no second InGaAs base layer 240, which is the case of the conventional GaAsHBT. On the other hand, the solid line represents the case of the GaAsHBT of the present invention. In this embodiment, the thickness X of the second InGaAs base layer 240 is 2 nm, and the p-type impurity concentration Np2 is 1×10 19 cm -3 In addition, the impurity concentration Np1 of the first InGaAs base layer 230 is 4×10 19 cm -3 The Vbe at which Ic reaches a certain predetermined value is called the on-state Vbe. Fig.21 The two Ic curves shown are horizontally shifted in parallel by 49mV. That is, regardless of the Ic that defines the on-state Vbe, the on-state Vbe of the present invention is 49mV lower than before. This change is caused by Fig. 20 The height of the spike-shaped energy barrier shown is due to the fact that the height of the spike-shaped energy barrier in the case of the present invention is reduced by 63 mV (ie, the height of the spike-shaped energy barrier in the present invention is reduced by 63 mV). Fig. 22 It is a graph showing the relationship between the on-state Vbe and X. Np1 is 4×10 19 cm -3 , Np2 is 1×10 19 cm -3 As X increases, that is, as the thickness of the second InGaAs base layer 240 increases, the on-state Vbe decreases. Fig. 22 In the present invention, the range of X from 1 nm to 6 nm is indicated as "the present invention". The reason is that when 1 nm ≤ X ≤ 6 nm, a particularly significant improvement in characteristics can be obtained, which will be described in detail in Fig. 27 Described in the instructions. Fig.23 This figure shows the relationship between the on-state Vbe and the ratio of Np2 / Np1. This figure shows that Np1 is fixed to 4×10 19 cm -3 Therefore, Np2 / Np1=0.25 is equivalent to Np2=1×10 19 cm -3 X is 2 nm. Np2 / Np1=1 corresponds to the case where the second InGaAs base layer 240 is substantially absent, that is, the case of the conventional GaAsHBT. The on-state Vbe decreases as Np2 / Np1 decreases. Fig.23 The point where Np2 / Np1 is 0.5 or less is indicated as "the present invention". The reason is that when Np2 / Np1 ≤ 0.5, a particularly significant improvement in characteristics can be obtained. The specific description will be given in Fig.31 Described in the instructions.

[0114] Next, Fig.18 The high frequency characteristics of the GaAsHBT calculated by two-dimensional device simulation shown in Table 2 are described. Fig.24 This is a graph showing the dependence of the base collector-to-collector capacitance Cbc on the collector current Ic. The collector-to-emitter voltage Vce is fixed at 0.5V, and the base-to-emitter voltage Vbe is varied to obtain the Ic dependence. The definition of Cbc is as described above. The dotted line is the case where X=0nm and there is no second InGaAs base layer 240, which is the case of the existing GaAsHBT. On the other hand, the solid line is the case of the GaAsHBT of the present invention. In this embodiment, the thickness X of the second InGaAs base layer 240 is 2nm, and the p-type impurity concentration Np2 is 1×10 19 cm -3 In addition, the p-type impurity concentration Np1 of the first InGaAs base layer 230 is 4×10 19 cm -3 . Fig.24 The data points shown have an Ic of 0.01A or more, all entering Fig.37 As shown in the "low Vbe region". Fig.21 As we know, when Ic becomes 0.01A or more, Vbe exceeds 1.25V. On the other hand, Vce is 0.5V, so Vbc is 0.75V or more, that is, the pn junction between the base and collector is in a high forward bias state. Therefore, Fig.24 In the case of the conventional GaAsHBT shown by the middle dotted line, as Ic increases, that is, as the forward bias increases, Cbc becomes a high value exceeding 1pF. In contrast, the Cbc of the GaAsHBT of the present invention is significantly reduced compared to the past, and the higher the Ic, the more obvious the reduction. The reason is that the forward bias between the base and collector regions becomes smaller due to the reduction in the on-state Vbe. Fig.25 The maximum power gain at frequency f0 = 2 GHz is dependent on Ic. Fig.24 Similarly, the collector voltage Vce is 0.5 V. The vertical axis is expressed as MSG / MAG, but its meaning is as described above. Fig.25 In the figure, the slope of the curve changes sharply at Ic=0.05A, but below 0.05A is the MSG region, and above 0.05A is the MAG region. Fig.25 In the figure, the solid line is the GaAsHBT of the present invention, the thickness X of the second InGaAs base layer 240 is 2 nm, and the p-type impurity concentration Np2 is 1×10 19 cm -3 The p-type impurity concentration Np1 of the first InGaAs base layer 230 is 4×10 19 cm -3 Compared with the existing GaAsHBT shown by the dotted line, MSG / MAG is higher at any current. Fig.24The reason for this is the reduction in Cbc shown in FIG. 1 . Cbc acts as a feedback capacitor in the power amplification of the transistor and reduces the power amplification factor. In particular, in the "low Vce region", the pn junction between the base and collector regions is in a strong forward bias state, so Cbc becomes very large, which becomes a factor in reducing the power amplification factor. According to the present invention, Cbc can be greatly reduced in the "low Vce region", and the reduction in power amplification factor in this region can be suppressed. Fig.26 is a graph showing the relationship between Cbc and X, Fig. 27 It is a graph showing the relationship between the maximum power gain and X. Np1 = 4 × 10 19 cm -3 , Np2=1×10 19 cm -3 Vce is 0.5V. In addition, the black circle is the value when Ic is 0.072A, and the black triangle is the value when Ic is 0.144A. Ic = 0.072A, 0.144A both enter Fig.37 The “low Vce region” shown in the figure, but the area of ​​the emitter region is 360 square microns, so if converted to the average current density of the emitter region, it is equivalent to 20kA / cm 2 , 40kA / cm 2 . Fig.25 In the case of Ic = 0.072A and 0.144A, the maximum power gain is in the MAG region, so Fig. 27 The vertical axis is represented by MAG. Fig.26 and Fig. 27 The data point X=0 nm indicated as "existing" in FIG. 2 is the case of an existing GaAs HBT without the second InGaAs base layer 240. As X becomes larger than 0 nm, Cbc decreases and MAG increases. In order to obtain a significant effect of the present invention, it is preferred that Ic=0.072A (20 kA / cm 2 ) when MAG improves by more than 1dB, Ic=0.144A(40kA / cm 2 ) when MAG improves the existing 2dB or more. Fig. 27 The range of 1 nm≦X≦6 nm satisfying the above condition is indicated as “the present invention”. Fig. 22 , Fig.23 , Fig.26 Similarly, the range of 1nm≦X≦6nm is expressed as "the present invention". Fig. 27 The X dependence of the power amplification factor (MAG) of FIG. 1 is described in more detail. MAG initially increases as X increases, but decreases again when X is above 6 nm. Fig.28 Representation and Fig. 27A graph showing the relationship between |H21|2 and X under the same conditions. H21 is one of the H parameters and represents the current amplification factor during RF operation. The black circle is the value when Ic=0.072A, and the black triangle is the value when Ic=0.144A. Both decrease as X increases. Fig.29 The depth dependence of the conduction band energy when X = 8 nm is shown in Fig. 19 cm -3 , Np2 is 1×10 19 cm -3 . The conduction band energy shows a sharp rise at the interface between the second InGaAs base layer 240 and the first InGaAs base layer 230, which becomes an energy barrier for electrons in the second InGaAs base layer 240. Due to the energy barrier in the base layer, electrons accumulate in the second InGaAs base layer 240, and the flow of current is blocked. This leads to a decrease in |H21|2 and even a decrease in power amplification. Therefore, even if the thickness X of the second InGaAs base layer is too large, the characteristics cannot be improved. As a condition to meet the improvement conditions of MAG described above, in addition to 1nm≦X, X≦6nm is required. Finally, the dependence of the characteristic improvement on Np2 / Np1 is explained. Fig.30 , Fig.31 The graphs show the relationship between Cbc, MAG and Np2 / Np1. X is 2 nm, Np1 is 4×10 19 cm -3 , Vce is 0.5V. In addition, the black circle is the value of Ic = 0.072A, and the black triangle is the value of Ic = 0.144A. Np2 / Np1 = 1 is the case where the second InGaAs base layer 240 does not exist substantially, which is equivalent to the existing GaAsHBT. As Np2 / Np1 becomes less than 1, Cbc decreases monotonically and MAG increases monotonically. Again, MAG is at Ic = 0.072A (20kA / cm 2 ) becomes 1dB higher than the existing one, and at Ic=0.144A(40kA / cm 2 ) becomes 2dB or more than the existing one, and the conditions for obtaining the significant improvement obtained by the present invention become Fig.31 The range of 0≦Np2 / Np1≦0.5 is indicated as "the present invention". The above conditions are also shown in Fig.30 .

[0115] When the GaAsHBT of the second invention of the present application is used in a power amplifier, the power added efficiency PAE and other characteristics of the power amplifier can be improved according to the degree of improvement in the power amplification factor in the "low Vce region" in the same way as the first invention of the present application. In particular, the characteristics can be greatly improved in a power amplifier operating in a power saturation region such as an envelope tracking amplifier.

[0116] The first invention of the present application may also be combined with the second invention of the present application. Fig.32 For Fig.18 , Table 2 shows the depth dependence of the conduction band energy obtained by two-dimensional device simulation when the emitter layer 250 is replaced by the first emitter layer 150 of the first invention of the present application in the second invention of the present application. The first emitter layer is InGaP with Eg=1.85eV and CuPt-type order. Np1=4×10 19 cm -3 , Np2=1×10 19 cm -3 , X = 2nm. The dotted line is Figure 4 , Fig. 20 In the case of the conventional GaAs HBT shown in , the spike barrier height is 105 meV. In contrast, in the present embodiment combining the first invention of the present application with the second invention of the present application, the spike barrier height is substantially zero as shown by the solid line. Fig.33 For Fig.21 The same plot is shown. Compared with the conventional GaAsHBT shown by the dotted line, in the present embodiment shown by the solid line, Ic becomes larger and shifts to the low voltage side, achieving an 83mV reduction in the on-state Vbe. Fig.34 , Fig.35 The Ic dependence of Cbc and power gain (MSG / MAG) is shown respectively. Vce is 0.5 V. The definitions of Cbc and MSG / MAG are as described above. Compared with the existing GaAsHBT shown by the dotted line, in this embodiment, the MSG / MAG is greatly improved by greatly reducing Cbc.

[0117] When the GaAsHBT obtained by combining the first invention of the present application and the second invention of the present application is used in a power amplifier, the power added efficiency PAE and other characteristics of the power amplifier can be improved according to the degree of improvement in the power amplification factor in the "low Vce region" in the same way as the first invention of the present application. In particular, the characteristics can be greatly improved in a power amplifier operating in a power saturation region such as an envelope tracking amplifier.

[0118] [Industrial Applicability]

[0119] The GaAsHBTs of the first and second inventions of the present application can be widely used as transistors for high-frequency amplifiers, such as power amplifiers for wireless communication.

Claims

1. A compound semiconductor heterojunction bipolar transistor, which is an npn-type bipolar transistor comprising a collector layer, a base layer and an emitter layer formed in sequence on a GaAs substrate, characterized in that: The collector region layer, the base region layer, and the emitter region layer all include compound semiconductor layers that are epitaxially grown on the GaAs substrate and have substantially no relaxation of lattice strain caused by lattice mismatch with GaAs. The base layer at least comprises a pseudo-crystal grown InGaAs layer, and The emitter layer at least comprises a first InGaP layer having a CuPt-type order and a second InGaP layer on the first InGaP layer on the opposite side of the base layer, and The second InGaP layer is substantially free of CuPt-type order, The n-type impurity concentration of the first InGaP layer, the second InGaP layer, or both near the interface between the first InGaP layer and the second InGaP layer is higher than the n-type impurity concentration in any portion of the first InGaP layer and the second InGaP layer except a portion near the interface.

2. The compound semiconductor heterojunction bipolar transistor according to claim 1, characterized in that: The band gap of the first InGaP layer having the CuPt-type order at room temperature is in the range of 1.76 eV to 1.86 eV.

3. The compound semiconductor heterojunction bipolar transistor according to claim 2, characterized in that: The composition of the first InGaP layer having CuPt-type order is described as In i Ga 1-i The In content i at P is in the range of 0.48 to 0.

5.

4. The compound semiconductor heterojunction bipolar transistor according to claim 1, characterized in that: The composition of the InGaAs layer is described as In j Ga 1-j The average value of the In content j in As is in the range of 0.05 to 0.12.

Images