TW200308088A - Bipolar transistor with graded base layer - Google Patents

Abstract

A semiconductor material which has a high carbon dopant concentration includes gallium, indium, arsenic and nitrogen. The disclosed semiconductor materials have a low sheet resistivity because of the high carbon dopant concentrations obtained. The material can be the base layer of gallium arsenide- based heterojunction bipolar transistors and can be lattice-matched to gallium arsenide emitter and/or collector layers by controlling concentrations of indium and nitrogen in the base layer. The base layer can have a graded band gap that is formed by changing the flow rates during deposition of III and V additive elements employed to reduce band gap relative to different III-V elements that represent the bulk of the layer. The flow rates of the III and V additive elements maintain an essentially constant doping-mobility product value during deposition and can be regulated to obtain pre-selected base-emitter voltages at junctions within a resulting transistor.

Description

200308088 Description of the invention: [Technical field to which the invention belongs] The present invention relates to a transistor, and more particularly to a bipolar transistor having a graded base layer. [Previous technology] Integrated circuits (ICs) of bipolar junction transistors (BJT) and heterojunction bipolar transistors (HBT) have been developed into An important technology for a variety of applications, especially as a high-speed circuit (greater than 10 billion bits per second =) of gastric power amplifiers for wireless handsets, microwave instruments, and fiber optic communication systems. It is expected that future demand will require components with lower operating voltages, high-frequency performance, higher added power efficiency, and lower manufacturing costs. The BJT & synchronous turn-on voltage (1) is defined as the base-emitter voltage (vbe) required to reach a fixed collector current density (Jc). For low power applications where the supply voltage is limited by the power requirements of battery technology and other components, this turn-on voltage can limit the performance of the component. The emitter, base, and collector in… are made of one kind of semiconductor material, HBT is made of two dissimilar semiconductor materials, and the band gap of the emitter semiconductor material (bandgap, energy is: The band gap of the semiconductor material used to make the base is large. This results in a carrier injection efficiency to the collector that is better than hunger. This is due to the establishment of a barrier = a barrier for the injection of the carrier from the base back to the emitter. The 200308088 base with a smaller band gap can reduce the opening because the increase in the rate of carrier injection j from the base to the collector will increase the collector current density at a given base-emitter voltage. However, the band alignment of HBT semiconductor materials at heterojunctions has the disadvantage of sudden discontinuity, which can lead to spikes in the conduction band at the emitter-base interface of 纟 m. This conduction band bursts The effect of the peak is to prevent the electrons from transmitting from the base to the collector. Due to Λ, the electrons stay at the base for a longer time, resulting in an increase in the degree of recombination and a reduction in the collector current gain (/ 5dc). As discussed above, Due to the heterojunction of bipolar transistors The opening voltage is defined as the base-emitter voltage required to reach a fixed collector current density, so reducing the collector current gain effectively increases the forward opening voltage. Therefore, the manufacturing of HBT semiconductor materials needs further improvement. The voltage is turned on, thereby improving low-voltage operation elements. The present invention provides an HBT having an η㉟-doped collector, a base formed on the collector and composed of In-V group materials (including steel and nitrogen), And an n-type doped emitter formed on the base. The family material of the base layer has a carbon doping concentration of about LSxiPd3 to about 70 × 1019 cm_3. In a preferred embodiment, the base layer includes Gallium, indium, arsenic, and nitrogen elements. Compared to the band gap of GaAs, the presence of indium and nitrogen reduces the band gap of this material. In addition, the impurity concentration in this material is high, and the sheet resistance (Rsb ) Is low. Compared to HBTs with a similar doping concentration of the base layer, these factors lead to lower turn-on voltages. 200308088 In a preferred embodiment, the oscillating material system can: … MSl_A It is known that when a small amount of nitrogen is incorporated into ~ -x nxAs, the band gap of this material will substantially decrease. Furthermore: because nitrogen promotes the lattice constant in a direction opposite to indium, the Adding this material in the proper proportion of lice can grow a lattice-matching alloy 〇 ~ _χΙηχΑδι_Α. Therefore, it can eliminate the excess strain that leads to the increase of energy gap and the mismatch of materials. Therefore, the indium to nitrogen can be selected to reduce or eliminate strain In a preferred embodiment of the present invention, x = 3y in the base layer Ga ^ xInxAs and yNy of ηβτ. In a conventional HBT with GaAs, as the temperature increases, a hole is injected into the emitter. To a higher degree, the recombination current of the charge layer between $ is higher, and the diffusion length in the base may be shorter, so the current gain is typically reduced. In HBT with a GalnAsN base layer, as the temperature was increased, a significant increase in current gain was found (approximately 0.3% increase per rc). This result is explained as the diffusion length increases with increasing temperature. If the electrons at the bottom of the band are limited to a state where they are at least partially localized = π, this effect is expected, and as the temperature increases, they are thermally excited to other states where the electrons can diffuse more easily status. Therefore, constructing the base layer with a GalnAsN design improves the temperature characteristics of the β? Of the present invention and reduces the need for a temperature-compensated secondary circuit. Compared to the conventional Ηβτ with a GaAs base layer, HBT with a GalnAsN base layer has improved common emitter output characteristics. For example, the offset voltage and knee voltage of an HBT with a GalnAsN base layer are lower than that of a conventional HBT with a GaAs base layer. 200308088 In a specific aspect, the 'transistor is a double heterojunction bipolar transistor (DHBT). The semiconductor material constituting the base is different from the semiconductor material made of the emitter and collector. In a preferred embodiment, the GainAsN base layer can be represented by the chemical formula Ga ^ xIrixAsi-yNy, the collector is GaAs, and the emitter is selected from InGaP, AlInGaP, and AlGaAs. Another preferred embodiment of the present invention is HBT or DHBT, which reduces the height of the peak of the conduction band, and at the same time reduces the band gap (Egb) of the base layer. The conduction band spike is caused by a discontinuity in the conduction band at the base / emitter heterojunction or base / collector heterojunction. Reducing the lattice strain by matching the β θ to the emitter and / or collector layer reduces the conduction band spikes. This is typically controlled by the nitrogen and indium in the base layer. The concentration is done. The base layer preferably has the chemical formula 'χΙηχΑ ~ -Λ, where X is approximately equal to 3y. In a specific aspect, the base can be graded in composition to make a band gap at the collector The band gap is smaller and the band gap is larger at the emitter: The band gap layer of the ::. The band gap of the base layer is connected to the collector at the base layer, and it is better than the one in which the base layer is in contact with the emitter. The surface is about ZOmeV to about 120meV lower. The base-to-emitter changes linearly across the base layer and the listening gap is better. The band gap of semiconductor materials between bands is larger than that of GaAs and A poles. The band gap in the ㈣-mount ^ base layer is reduced at the collector than at the emitter. 200308088 ° However, compared to the band gap of GaAs across the base layer, the above energy gap typically decreases by about 10 on average. meV to about 300 meV. In a specific case, compared to the band gap of GaAs across the base layer, the above-mentioned monthly bf gap typically decreases by an average of about 80 meV to about 300 meV. In another specific complaint, In this case, the band gaps typically decrease by an average of about 10 meV to about 200 meV compared to the band gaps where GaAs crosses the base layer. This reduced band gap results in a GanIMShNy group with a composition hierarchy. The turn-on voltage (Vbe, ...) of the HBT of the electrode layer is lower than that of hBT with the base layer, because the main determinant in Vbeon is the intrinsic carrier concentration in the base. The intrinsic carrier concentration ( ni) is calculated by the following formula: ni = NcNvexp (-Eg / kT) In the formula above, Nc is a pass ..... w 匁 双 匁 反, text

The effective state of the energy band state is broken. P H At -Wt OB, J4 degrees, Eg band gap; T is the temperature; and k 疋 wave record and * number. It can be seen from the formula that the intrinsic carrier concentration in the base is largely controlled by the band gap of the material used in the base. The band gap of the base layer is changed from the larger energy gap T of the base-emitter interface to the smaller band gap of the base-collector interface, which introduces a quasi-field, which makes the electrons accelerate _ Base layer in a bipolar transistor. The electric field increases the speed of the electrons in the base, 竑 w, vv, and the collector transmission time, which improves the performance of the RF (radio frequency), and increases the collector's gain (also called dc). Current gain). In the case of UDrr ”in HBT with a heavily doped collector layer, the dc gain d) is subject to the overall recombination in a thief’ r-type base (η = 1). dc «current gain can be estimated by formula ⑴: / 5 dc« ^ ^ / Wb 10 200308088 In formula〉,> is the average speed of the secondary carriers in the base, and is the secondary carrier's average speed in the base. Lifetime; and Wb is the thickness of the base. Compared with the ungraded-base layer, the base layer in the gravitational base sound is appropriately classified. Due to the increase of the electron velocity material, there is a significant increase in stone dc.

In order to achieve a graded band gap in the thickness of the base layer, the base layer is prepared so that the concentration of indium and / or nitrogen at the first surface of the base layer (near the collector) is higher than that of the base layer. Comes high at the second surface (near the emitter). The content of indium and / or nitrogen is preferably changed linearly across the base layer, resulting in a linearly graded band gap. The concentration of dopants (e.g. carbon) is preferably maintained constant throughout the base layer. In a specific aspect, the% xiMs Λ base layer (for example, the deleted base layer) is classified into X and 3y at the collector, which is approximately equal to 0.00, and at the emitter, it is classified into X and approximately Equal to zero In another specific example, the Gai_ »hNy base layer ranges from a value of about χ to about ⑽ at the surface where the base layer is in contact with the collector, and is graded at the surface where the base layer is in contact with the emitter. The value of χ is in the range of about 0 ″ to zero, provided that the value of χ at the surface where the base layer is in contact with the collector is larger than at the surface where the base layer is in contact with the emitter. In this specific aspect, y can be kept constant throughout the base layer, or it can be linearly graded. When y is linearly graded, the base layer is graded into a surface in which the base layer is in contact with the emitter from a value in the range of about 0.2 to about 0.02 at the surface where the base layer is in contact with the collector. If the value of y is in the range of about 0.1 to zero, then the value of y at the surface where the base layer is in contact with the collector is greater than at the surface where the base layer is in contact with the emitter. In a preferred embodiment, 11 200308088 x is approximately 0.06 at the collector and linearly graded to approximately 0.01 at the emitter. In a better specific aspect, X is about 0.06 at the collector and linearly graded to about 0.001 at the emitter, while y is about 0.001 in the entire base layer. In another specific aspect, the present invention is a method for forming a graded semiconductor layer. The graded semiconductor layer has a substantially linear graded band gap from the first surface through the layer to the second surface and is substantially fixed. Instead, the product of X mobility is doped. This method includes: (a) comparing the dopant x mobility product of multiple correction layers, where each layer is selected from the group consisting of an organometallic compound of a III or V atom of the periodic table or a carbon tetracarbon compound of deposited carbon; One is formed at a clearly differentiated flow rate, thereby determining the relative flow rates of the organometallic compound and carbon tetraoxide required to form a substantially constant doped χ shift product; and (b) in At this relative flow rate, an organometallic compound and a tetradentate carbon compound are flowed on a surface to form a substantially fixed product of doped χ mobility, and the flow rate is changed during deposition, thereby forming There is a substantially linearly graded band gap in the entire graded semiconductor layer. The base layer can also grade the dopants, so that the concentration of the dopants near the collector is relatively high, and gradually decreases to the base-emitter heterojunction through the thickness of the base. Another way to minimize conduction band spikes is to include one or more transition layers at the heterojunction. A set back layer with a low band gap, a graded band gap layer, a doped spike, or a combination thereof can be used to minimize the conduction band spikes ...…, or more A lattice matching layer may exist between the base and the emitter or between the base and the collector to reduce the lattice strain of the material at the heterojunction. Ben Ming also provides a method of making bribe # DHBT. This method includes growing a base layer consisting of Su, Yin, Shishen, and Ni on a GaAs collector that is heterogeneous. The base layer can be grown using internal and external carbon sources to provide a carbon-doped base layer. An n-doped emitter layer is then grown on the base layer. Internal and external carbon sources are used to provide carbon dopants for the base layer ... to help form materials with a higher concentration of carbon dopants. Typically, the method of the present invention achieves a large dopant level of about UxlO ^ W to about.纟 —In a preferred embodiment, this method can achieve about 30 × 1019cm_3 to about 7.0 × 1019 ^ 3: the privacy of multiple debris. Higher dopant concentrations in the material will reduce the sheet resistance and gap of the material. Therefore, the higher the surface area of the impurities / chronicity in the base layer of 'HBT and DHBT, the lower the turn-on voltage of the device. The present invention also provides a material 4 represented by the chemical formula Gai xIMsi A, where X # y is independently about i. ㈣G_4 to about Uxio'x is preferably approximately equal to 3 "" port 3y is more preferably approximately equal to .0 In a specific aspect, the material is doped with carbon at a concentration of about ΐ5χΐ〇ΐ9Μ_3 to about 7.0x101W. In a specific embodiment, the concentration of the carbon dopant is about 3.〇xl0lw to about 7.〇χΐ〇ι ^ _3. Lower the power of the radio to better manage the voltage budget of GaAs-based wired and wireless RF circuits. These circuits are subject to standard fixed voltage Supply or subject to battery output. Lowering the power generation can also be changed. 13 200308088 Changing the relative size of various base current components in HBT based on GaAs. The dc current gain has been shown to be stable as a function of both junction temperature and applied stress. 'Degree' is heavily dependent on the relative size of the base current component. The reduction in reverse hole injection caused by the low turn-on voltage is conducive to the temperature stability and long-term reliability of the device. Because of & Debris concentration Relatively speaking, the corresponding GaixInxAShNy base material can significantly improve the RF performance of GaAs-based HBT and DHBT. [Embodiment] From the more specific description of the preferred embodiment of the present invention, the foregoing and other aspects of the present invention The purpose, features, and advantages of this will become apparent. The better specific aspects are illustrated in the attached drawings, where similar words are used to refer to the same parts in the same view. The drawings may not be It is based on the ratio ㈣, but emphasizes the principle of exemplifying the present invention. The materials of group ΙΠ-ν are crystal lattices including at least one X * prime selected from Zhou Yi and half of the elements selected from V (A) of the periodic table. In a specific aspect, the IU-V group material is a crystal lattice including 录, 锢, Shishen, and nitrogen. Η Group I-V; γ ^ The material can be better by the chemical formula Gai_xIMsi_ χ θ is better, ⑭ is approximately equal to ⑭ . In-the best specific aspect τ, X and 3y is about 〇1. The "transition sound is wide ..."

J (transitional layer) —The word refers to the layer between the base / emitter heterojunction °, between, or between the base / collector heterojunction, and temporarily pulls the right side of the The peak of the conduction energy band reaches a minimum of 14 200308088. Two = The minimum is reached-a method is to make the inter-band two: the most: =! / Collector heterojunction_, the energy of the transition layer, ":, the m-like layer gradually The layer that is reduced to the closest to the base is similarly 'heterojunction at emitter / base = gap = near-emitter transition layer is gradually reduced to the nearest ::: crossing layer. Minimize the peak of conduction band Another kind of grading band gap transition layer. If you use this method, you can increase the gradual gap by grading the concentration of the material. For example, 过 can be higher near the base and gradually lower near the collector or emitter. Alternatively, a lattice strain can be used to provide a transition layer with graded band gaps. For example, the transition layer can be graded on the substrate to minimize the lattice strain on the surface contacting the base and increase the lattice strain on the surface contacting the collector or emitter. Another way to minimize the spikes in the conduction band is to use a transition layer with spikes in dopant concentration. One or more of the above methods for minimizing the peak of the conduction energy band can be used in the present invention. The BTT β suitable transition layers for the BTT of the present invention include InGaAs and InGaAsN. The lattice matching layer is grown on materials with different lattice constants. Layer. The lattice matching layer typically has a thickness of about 500 people or less and conforms to the lattice constant of the underlying layer. If the lattice matching material is not strained, this results in a band gap between the band gap of the underlying layer and the band gap of the lattice matching material. The method of forming the lattice matching layer is familiar to those skilled in the art and can be found on pages 303-328 of Gallium Arsenide Technology by Ferry et al. (1985 15 200308088, Indiana) Published by Howard W. Sams, Inc. of Indianapolis), which is taught and incorporated herein by reference. An example of a material suitable for the lattice matching layer of the present invention is InGaP. HBT and DHBT of the composed base layer The HBT and DHBT of the present invention can be prepared by using an appropriate epitaxial growth system of metal organic chemical vapor deposition (MOCVD). Suitable examples of MoCVD epitaxial growth systems are the AIXTR0N 2400 and AIXTR0N 2600 platforms. In HBT and DHBT prepared by the method of the present invention, typically, an undoped GaAs buffer layer can be grown after removing the adsorbed oxide on the spot. For example, a secondary collector layer containing a high concentration of an n-type dopant (eg, a dopant concentration of about 1 × 1018cnr3 to about 9 × 1018αιΓ3) can be grown at a temperature of about 70 ° (rc). The collector layer of a dopant (eg, a dopant concentration of about 5 × 10 5 cm_3 to about 5 × 10 16 cm-3) can be grown on the sub-collector layer at a temperature of about 700 ° C. The sub-collector and the collector are most Well, the typical thickness of the secondary collector layer is about 4000A to about 6000A, and the typical thickness of the collector layer is about 3000A to about 5000 people. In a specific aspect, the secondary collector and / or the dopant in the collector are The impurity is silicon. The lattice-matched InGaP tunneling layer can be selectively grown on the collector under typical growth conditions as needed. The thickness of the lattice-matching layer is generally about 500 A or less, preferably about 200 A or less. It is smaller, and its dopant concentration is about 1x1016cnT3 to about 1x1018cm-3. One or more transition layers can be selectively grown on the lattice matching layer or the collector under typical growth conditions as needed (If not using Lattice 16 200308088 ). The transition layer can be prepared from n-doped GaAs, n-doped InGaAs, or n-doped InGaAsN. The transition layer can be selectively graded on the composition or dopant as required, or Can contain spikes of dopants. The typical thickness of the transition layer is about 75A to about 25A. If no lattice matching layer or transition layer is used, the doped InGaAsN base layer is grown on the collector. The base layer is Grows at a temperature below about 750 ° C, and is typically about 400Λ to about 1500Λ thick. In a preferred embodiment, the base layer is grown at a temperature of about 50 ° C to about 60 ° C The carbon-doped InGaAsN base layer can be selectively grown on the transition layer or the lattice matching layer (if no transition layer is used). The base layer can use an appropriate gallium source (such as trimethylgallium) Or triethylgallium), arsenic sources (such as europium, tributylphosphonium, or trimethylphosphonium), indium sources (such as trifluorenylindium), and nitrogen sources (such as ammonia or difluorenylhydrazine). Use a low molar ratio of arsenic source to gallium source. In type, the molar ratio of the arsenic source to the gallium source is less than about 3.5. The ratio is more preferably about 2.0 to about 3.0. The degree of nitrogen and indium sources is adjusted to include about 0.01% To about 20% indium and about 0.01% to about 20% nitrogen. In a preferred embodiment, the 'base layer's indium content is about three times the nitrogen content. In a more preferred embodiment In a specific aspect, the content of indium is about 1%, and the content of nitrogen is about 0.3%. In the present invention, the high carbon dopant concentration of about 1.5x1019cm-3 to about 7.0xl019cnT3 The InGaAsN layer can be obtained using an external carbon source or an organic metal source (especially a gallium source). A suitable example of an external carbon source is% desertification. Tetragasified carbon is also an effective external carbon source. 17 200308088 One or more transition layers can optionally be grown between the base and the emitter from n-doped GaAs, n-doped inGaAs, or n-doped InGaAsN, as needed. The transition layer between the base and the emitter is relatively lightly doped (for example, about 5.0x1015cnT3 to about 5.0x1016αι3), and it can be seen that the dopants are selectively included in the spikes. The transition layer is preferably about 25 to about 75A thick. The emitter layer is grown on the base at a temperature of about 700 ° C, or can be selectively grown on the transition layer as needed, and is typically about 400A to about 1500A thick. The emitter layer includes, for example, inGap, Ai in (jap, or AlGaAs. In a preferred embodiment, the emitter layer includes InGap. The emitter layer may be n-type doped to about 1.0 × 1017cnr3 to about 9.0 × 1017〇ττ3 The concentration of the emitter contact layer containing a high concentration of n-type dopants (for example, about 1.0 × 1018cnT3 to about 9 × 1018〇η-3 ^ GaAs can be selectively grown on the emitter at a temperature of about 70 ° C. if necessary The 'emitter contact layer is typically about 1000 A to about 2000 people thick. It has a progressive indium composition and a high concentration of n-type dopants (eg, about 5x10i8cnr3 to about 5x1019cm-3) The InGaAs layer is grown on the emitter contact layer. This layer is typically about 400A to about ιοοοΑ thick. Example 1 To demonstrate the reduction of the band gap of the base layer and / or the conduction energy of the emitter / base heterojunction The peak effect is minimized, and three different GaAs-based bipolar transistor structures are compared: GaAs emitter / GaAs base BJT, InGaP / GaAs HBT, and the InGaP / GalAsAs DHBT of the present invention. It is used below Experimental InGaP / GaInAsN Training 18 200308088 Representative is the general structure illustrated in Figure 1. Since the base and collector electrodes are formed of GaAs, so that only a heterojunction at the emitter / base interface.

The band gap of the GaAs base layer of InGaP / GaAs HBT is larger than that of the base layer of InGap / GalnAsN DHBT. Since the emitter, collector, and base of GaAs / GaAs bjt are made of GaAs, there is no heterojunction. Therefore, the GaAs BJT structure is used as a reference to determine the impact of the conduction band spike at the base / emitter interface on the collector current characteristics of the InGaP / GaAs HBT (if any). In the DHBT in Figure i, hG # is selected as the emitter and the base is Gai x InxAsi yNy. This is because the band gap of InGap is wide and its conduction band is aligned with the conduction band of the GaiXA base. The comparison between inGaP / GaInAsN DHBT and InGaP / GaAs HBT in Figure 1 is used to determine the effect of the base layer with lower band gap on the collector current density. The GaAs elements used for the discussion below all have a base layer of doped carbon grown by MOCVD, where the dopant concentration varies from about 15 × 1019cnr3 to about 6.5 × 1019cnr3, and the thickness varies from about 500 people to about 15 °. 〇 入 ′ so that the base sheet resistance (Rsb) is between 100Ω / □ to 400ω / □. Large-area components (L = 75 // mx75 // m) are fabricated using a simple wet etch method and tested in a common base architecture. A relatively small amount of indium (X ~ 1%) and nitrogen (y ~ 0.3%) are added one by one to form two different sets of InGaP / GainAsN DHBT. The growth of each group has been optimized to maintain a high and uniform carbon dopant level (> 2.5 > < 1019cnr3), good mobility (~ 85 cm2 / Vs), and high dc current gain (Rsb ~ 30〇ω / □ when it is > 6〇).

Draw typical Gummel diagrams of GaAs / GaAs BJT, InGaP 19 200308088 / GaAs HBT, and InGaP / GalnAsN DHBT with equivalent base sheet resistance, and overlay them in Figure 2. The collector currents of InGaP / GaAs HBT and GaAs / GaAs BJT are indistinguishable by more than five orders of magnitude (five 10) until the difference in effective series resistance affects the current-voltage characteristics. On the other hand, the collector current of InGaP / GalnAsN DHBT is twice that of GaAs / GaAs BJT and InGaP / GaAs HBT in a wide bias range, which corresponds to a collector current density of 1.78 A / cm2 (jc The on voltage at) is reduced by 25.0mV. The observed increase in the low-bias base current (n = 2 component) in BJT is consistent with the increase in recombination of the space charge driven by the band gap. The neutral base recombination component of the collector current in InGaP / GalnAsN DHBT is driven higher than InGaP / GaAs HBT because the collector current is increased and the secondary carrier life is reduced or the carrier speed is reduced. (Inb, Icwb / ur) increased. InGap /

For GalnAsN DHBT components, components with a base sheet resistance of 234 Ω / □ have reached a peak dc current gain of 68, which corresponds to a decrease in the turn-on voltage of 11.5 mV, while components with a base sheet resistance of 303 Ω / □ have A peak dc current gain of 66 is reached, which corresponds to a reduction of 25 mV of the turn-on voltage. For these several structures, this represents the highest known boost to the United States 0.3) 0 &amp; 1.χΙηχΑδιθγN ^; The decrease in the band gap is the reason for the decrease in the opening voltage observed, J. (77. 〇 Luminescence is demonstrated. DCXRD measurement shows that the base electrode has a minimum degree of mismatch (<250arcsec). In the limits of popularization, 'the base-emitter voltage (v, inflammation be) is The ideal collector current density of a function climbing electrode can be approximated as: (2) 200308088

Jc = (qDnn2ib / pbwb) exp (qVbe / kT) where

Pb and wb base doping and width;

Dn diffusion coefficient

The intrinsic carrier concentration in the Rib base represents nib as a function of the band gap (Egb) of the base layer, and the product of the base doping and width is rewritten with the base sheet resistance (Rsb). Logarithmic function expressed as base sheet resistance (3)

Vbe = -A ln (Rsb) + V (and A = (kT / q) (4) and V〇 = Egb / q- (kT / q) ln (q2 // NcNvDn / Jc) (5) where Nc and Nv is the effective density of states in the conduction and covalent energy bands, and # is the major carrier mobility in the base layer. Figure 3 plots multiple inGaP / GaAs hbt, GaAs / GaAs BJT, and InGaP / GalnAsN DHBT When Jc = l · 78 A / cm2, the opening voltage is a function of the resistance of the base sheet. InGaP / GaAs HBT and GaAs / GaAs BJT do not have any conduction band spikes, and their opening voltages are qualitatively exhibited for the base sheet resistance. Has the same log-dependent relationship as expected by equation (2). Quantitatively, the change in base-emitter voltage (Vbj with the resistance of the base sheet is not as dramatic as represented by equation (3) (a = 〇 〇i74mV instead of 0.252mV). However, the reduction observed by A is consistent with the quasi-ballistic (qUasiballistic) transmission through a thin base GaAs bipolar element. It is derived from a comparison with the characteristics of GaAs / GaAs BJT The conclusion is: the effective height of the conduction band spike of InGaP / GaAs HBT can be zero, 21 200308088 and the collector current shows ideal (n = 1 ) Behavior. Therefore, InGaP / GaAs HBT can be designed to have substantially no conduction band spikes. Previous studies on AlGaAs / GaAs ΗΒΤ also found similar results. In order to further reduce these components under fixed base sheet resistance Turn-on voltage, you need to use a base material with a lower band gap and still maintain the continuity of the conduction band. GahlrixAs ^ yNy can be used to reduce Egb while maintaining close to lattice matching. As seen in Figure 3, two The turn-on voltage of the group InGaP / GaInAsN DHBT follows the logarithmic dependence on the resistance of the base sheet, which shows that the conduction can be approximately zero. In addition, compared to the observed InGap / GaAs HBT and GaAs / GaAs BJT, The turn-on voltage of one group was shifted down by 11.5mV and the other group was shifted by 25 mV (two dashed lines). The above experiments show that the turn-on voltage of HBT based on GaAs can be reduced to below using the InGaP / GalnAsN DHBT structure. GaAs BJT turn-on voltage. The low turn-on voltage is achieved through two key steps. By selecting the base and emitter semiconductor materials (the conduction band of At the same energy (P white), first optimize the base-emitter interface to suppress the conduction band spikes. This is successfully done using InGaP or AlGaAs as the emitter material and plutonium as the base material. Here. Then, by reducing the band gap of the base layer, the hitting is further reduced. This is achieved by adding both indium and nitrogen to the base layer while still maintaining the lattice matching of the entire Satoshi structure. With proper growth parameters, it is possible to triple the collector current density without significantly sacrificing the base doping or the lifetime of the secondary carriers (υ4Ω / port M-68). These results show that making Gai JMsi-A materials provides a way to reduce the turn-on time based on GaAs # ΗΒΤ and DHBT. Since 22 200308088 incorporation of indium and nitrogen in GaAs reduces the band gap of the material, a larger percentage of indium and nitrogen is added to the base. If high p-type doping concentrations are maintained, HBT and DHBT based on GaAs can be expected There is a large decrease in the turn-on voltage.

The decrease in the band gap of the GalnAsN base is assumed to be the cause of the observed decrease in the opening voltage, which has been confirmed by the low temperature (77. Erbium luminescence. Figure 4 compares the emission spectra of InGaP / GalnAsN DHBT and the traditional InGaP / GaAs Ηβτ. From The base layer signal of InGaP / GaAs HBT is at a lower energy than the collector (1.455eV vs. 1.507eV), which is due to the narrowing effect of the band gap caused by the high doping level. From InGaP / The signal of the base layer of GaInAsN DHBT appearing at L 408eV has been reduced, which is due to the band gap narrowing effect and the reduction of the base layer band gap caused by adding indium and nitrogen to the base layer. Compare here However, compared with the band gap of the GaAs base, a reduction of 47 meV in the signal position of the base layer can be equivalent to the reduction of the band gap of the base layer of the GalnAsN base. The displacement of the luminous signal correlates well with the measured decrease of the on-voltage of 45mv. In the absence of a conduction band spike, the decrease in the on-voltage can be directly related to the reduction of the band gap of the base layer.

The DCXRD spectrum shown in Figure 5 demonstrates the effect of adding a carbon dopant and indium to a GaAs semiconductor. Figure 5 shows DCXRD spectra from both inGaP / GaInAsN DHBT and standard InGaP / GaAs HBT with equivalent base thickness. In InGaP / GaAs HBT, the base layer can be regarded as the shoulder on the right-hand side of the GaAs substrate spike, which corresponds to the position of + g〇arcsec. This is due to the high concentration of 4x1 〇19cm-3 carbon dopants. The resulting tensile strain. Due to the addition of indium, the base layer spike of this particular InGaP / GaInAsN DHBT junction 23 200308088 structure is at -425 arcsec. One

The peaks of the GalnAsN base are similar to 罟, and they are related to the position of the large peak 疋 锢, a function of nitrogen and carbon concentrations. Indium to GaAs increase the compressive strain, which is compensated for by the mouth. 1 both with tensile strain

When indium (and nitrogen) is added to the carbon-doped GaA ^ ^ ^ Λ, the P-type doping degree is returned, but the growth must be optimized carefully. ^ ^ α ^ J Λ The combination of the base sheet resistance and the thickness of the base plate to obtain a useful degree of doping. The base doping can also be confirmed by first selectively selecting ′ to the base layer and then obtaining the M PGlarcm port profile. Figure 6 comparison

Such p-laron c-v doping of the GaAs base layer and the GaInAsN base layer = profile. In both cases, the degree of doping exceeded 3 × 1019α_3. Figure 7A shows another alternative DHBT structure with a fixed composition GalnAsN base layer that uses a transition layer between the emitter / base and collector / base junctions. In addition, a lattice-matched InGaP penetrating layer is used between the transition layer and the collector.

AJLM composition-level base layer DHBT All layers in DHBT with a composition-graded base layer can be grown similar to DHBT with a fixed-layer base layer, with the exception of joining from one of the transistors through the base The electrode layer is bonded to another as the base layer of the graded band gap. For example, if a crystalline terbium matching layer or a transition layer is not used, a GalnAsN base layer doped with carbon and capable of band gap classification can be grown on the collector. Heterocarbon and graded (jalnAsN base layer can be selectively grown on the transition layer or crystal matching layer if not used). The base layer can be at a temperature below about 750 ° C It grows below 24 200308088 and is typically about 400A to about 1500A thick. In a specific aspect, the base layer is grown at a temperature of about 500 ° C to about 60 ° C. The base layer may use appropriate gallium Sources (such as trimethylgallium or triethylgallium), arsenic sources (such as thorium, di (secondary butyl) rhenium, or trimethylphosphonium), indium sources (such as trimethylindium), and nitrogen sources (such as ammonia , Dimethyl hydrazine or tertiary butyl amine) to grow. Prefer to use low molar ratio of arsenic source to gallium source. Typically, the molar ratio of arsenic source to gallium source is less than about 3.5. This ratio It is more preferably large, 々2 · 0 to about 3.0. The degree of nitrogen and indium sources can be adjusted to obtain a group III element in which about 0.01% to about 20% indium and a group y element are about 0.01% to about 20% nitrogen material. In a specific embodiment, the group III The content of elemental indium varies from about 10% to 20% of the base / collector junction to about 0.01% to 5% of the base / emitter junction, and the content of the V group το element nitrogen is basically It is fixed at about 0.3%. In another specific aspect, the indium content of the base layer is about three times the nitrogen content. As previously discussed, the InGaAsN base layer with a fixed composition is believed to have about 1.5 × 10 The InGaAsN layer with a high carbon dopant concentration from% πτ3 to about 7.0 × 1019 cm-3 can be obtained by using an external carbon source (such as tetrahalide inversion) in addition to the gallium source. The external carbon source used can be, for example, It is carbon tetrabromide. Tetragasified carbon is also an effective external carbon source. The amount of carbon dopants contributed to the InGaAsN base layer by the organic indium compound used as an indium source gas is different from that used as a gallium source gas. Organic gallium compounds to be used, so during the growth of the base layer, the flow rate of the carbon dopant source gas is typically adjusted, so that a constant carbon doping concentration is maintained in the composition-graded InGaAsN base layer 25 200308088. In one In the body shape, the change in the flow rate of the carbon source gas on the grading base layer is determined using the method described below. — Correction of the flow rate of the methyl indium source for the 1nGaAsN and semiconductor layers Prepare at least two sets of HBTs for calibration, each of which contains at least two members (DHBT can be used instead of HBT). The thickness of the base layer for all calibrations Ηβτ is ideally the same, but this is not a requirement. And each HBT has a fixed composition, such as a fixed composition # LAW or MaAs base layer and a fixed carbon dopant concentration in this layer. Each group is different from the other-group of In or ¥ Additives (such as indium for the Li family! Group V uses nitrogen) to grow at a source gas flow rate such that members of each group have gallium, arsenic, and nitrogen compositions that are different from those of other groups. For example: 'Indium is used as an additive that affects band gap classification. Each member of a particular group grows at a different external carbon source (such as four desertified carbon or four gasified carbon) at a flow rate ', so each member of a particular group has a different degree of carbon dopant. The product etch of the doping X mobility of each member is determined and the flow rate of the carbon source is plotted. The product of the doped X mobility of the members of each group changes in proportion to the flow rate of the carbon source gas. The flow of the doped X mobility of the five groups to the flow of the four desertified carbon is due to Figure 8. In addition, You can choose between each group of ΗΒΤ or shaped by a carbon source gas (such as tetragenetic carbon) that maintains a fixed flow rate ^: Individual samples in the group may be different at different dynamic rates relative to other source gases M⑴ or V additive flow rate 26 200308088 The flow rate of the heterogeneous x mobility source gas obtained from the solid source is the slave source gas required for the second product. The indium is drawn across the line-Figure 8 (for example One: = fixed doping x mobility product ^ line of thousands of lines on the X axis). This line is related to: a set of straight? Intersections represent the indium source gas f set in the set The external carbon source flow rate required to obtain the product of this doping x mobility at the dynamic rate. For the product of the fixed doping x mobility, the external carbon source flow rate is plotted against the indium source gas flow rate. Figure 9. Multiplication of different doped X mobility A similar line can be derived in the same way. The collector current of each HBT is plotted as a function of the base-emitter electricity M (Vbe) 'and the resulting curve is compared with a GaAs base layer and everything else: compared The members of the group are compared for graphs of the same t HBT (for example, with the same dopant concentration, the same base, emitter and collector layer thickness, etc.). The voltage difference is the change in the base-emitter voltage vbe (uvbe), which is due to the lower band gap of the base layer caused by the indium and nitrogen added during the formation of the base layer. Figure 丨 〇 shows The graph of the collector current of HBT with GalnAsN base layer and HBT with GaAs base layer is a function of vbe. The horizontal arrow drawn between the two curves is △ Vbe. △ vbe of each member of all groups Determine and plot the carbon source gas flow rate. The ΔVbe versus carbon tetrabromide flow rate used to form each member of the five groups of HBTs in Figure 8 is plotted in Figure u. Note the delta to a group of members vbe spans a range of △ Vbe in this group, this range can find the most suitable These straight lines are used to determine (interpolate) or allow the use of the same indium source gas flow rate for a particular group, but the carbon source gas # kinetic velocity differs from the other members of the group and grows △ Vbe. 27 200308088 、 r ride: the product of the doping X mobility of * 疋, linearly changes with the function of indium to the movement rate as a function of the interpolation of the fixed doping mobility product △ Vbe is plotted as the pin source It can be seen as a function of the gas flow rate. Figure 12 shows such a graph for the five groups of Figure u. The @shape shown in Figure 12 is used to determine the base / emitter and base / set = combination. To obtain the required indium source gas flow rate for the desired ΔL. -Once the flow rate of the indium source gas is determined, use Figure 9 to determine the flow rate of the carbon source gas at the source * L body flow rate, which is the product of the mobility of the desired dopant X. Follow the same procedure to determine the required indium source gas flow rate and carbon source emulsion flow rate at the base / collector junction, so that the composition-classified GalnAs or GalnAsN layer holds the dopants to be fixed. x mobility product. When the base layer grows from the base / collector junction to the base / emitter junction, the indium source gas flow rate and carbon hafnium gas flow rate change linearly with respect to the extent of gallium and thorium to the junctions. These source gases determine values to obtain a linear graded base layer with the desired band gap rating. /, Example 2 The GaAs elements used for the discussion below all have a doped base layer grown by M0CVD, where the dopant concentration is changed from about 30 × 1019 cm_3 to about 5.0 × 10 × 9 cm_3, and the thickness is From about 500 people to about π⑽ people, so that the base sheet resistance (RSb) is between 100Ω / □ to 650 Ω / □. Large-area components (L = 75 // mx75 / z m) are manufactured using a simple wet etch method and tested in a common base architecture. A relatively small amount of indium 28 200308088 (x ~ 1% to 6%) and nitrogen (y ~ 0.3%) were added in small amounts to form two different sets of / GalnAsN DHBT. The growth of each group has been optimized to maintain a high &amp; uniform carbon exchange level (&gt; 2.5xl0i9cm-3), good mobility (~ milk cmVV-s), and high Dc current gain (Rsb ~ 300 Ω / □ is> 6〇). The dhbt structure with a compositionally graded GaInAsN base layer used in the following experiments is shown in FIG. An alternative DHBT structure with a compositionally graded base layer is shown in Figures ⑺ and 7C. The DHBT structure of the GalnAsN base layer with a fixed composition used in the bottom experiment is shown in Fig. 14 for comparison. Fig. 15 shows the fixed base DHBT and the graded base DHBT with equivalent open voltage and base sheet resistance. —Figure. The neutral base component of the base current in the graded base junction is significantly lower, and the peak dc current gain it exhibits can be more than twice that of the fixed base structure. Figure 16 compares dc current gains of similar fixed and graded surface structures with varying thicknesses as a function of base sheet resistance. The increase in the ratio of the gain to the base sheet resistance is obvious. The ratio of the base line resistance of the increasing leg of the D leg depends on the growth conditions used and the specific details of the overall structure, but it has been observed that the dc current gain of DHBT with a hierarchical base layer is greater than that of a fixed base layer DHBT caused a 50% to 100% increase. Figures 17 and 18 compare the structure of a hierarchical base structure with two fixed base structures.

Gummel plot and gain curve. The base composition of the first fixed base structure corresponds to the base layer 分级 of the graded base at the base-emitter junction. The base composition of the second 2 疋 base structure corresponds to the base layer composition of the hierarchical base at the base / collector junction. The opening of the hierarchical base structure is between the two end structures, which are not vortexed and weighted towards the base / emitter end point. The dc electrical throw of the grading base structure is 9% higher than the end structure by 50% to 95%. This shows that most of the increase in the current gain is due to the increase in electron velocity.

An HP 8510C parameter analyzer was used to perform a test on a wafer on a 2-fin, 4 / zrax4 / zm emitter area το piece. Use open and short structures to remove the buried pad parasitics, and use a small signal current gain (then — a slope of 20dB / 10 to extrapolate the cutoff frequency of the current gain (b). Figure η synthesizes ft on the two structures Dependence on the collector current density 。. Figure 囷 illustrates the relationship between the small signal gain and frequency under the Schottky bias point. As Jc increases and the base transition time (η) starts to deduct / In the restricted role, although the base thickness of the hierarchical base structure is relatively large (the fixed base layer is thicker and the hierarchical base layer is 80 thicker), the ft of the hierarchical base structure still becomes The structure is significantly larger than the fixed composition. The spike of the fixed composition GaInAsN base at 60nm is 53 GHz, and the spike of the graded GaInAsN base at 80nm is 60GHz. Therefore, the cutoff frequency of the current gain is increased by 13%.

In order to properly compare the RF results of DHBT with a fixed and graded GaInAsN base layer with traditional GaAs HBT, the ft value in Figure 19 is plotted as a function of the zero input current breakdown voltage (BU that can be applied to the transistor. This figure and quote The ft values of the traditional GaAs HBT are compared in the literature. The ft values of the traditional GaAs HBT are expected to be quite broad because this information is compiled from many groups using different stupid crystal structures, component sizes, and test conditions, and is only intended to Give an overview of current industry standards. BVce. Often must be estimated from the quoted collector thickness, assuming a relationship between 30 200308088 collector thickness (Xc), BVcb◦, and BVce◦ as shown in Figure 21. See also Three simple calculations of the expected dependence of 疋 ft on BVce0 shown in Figure 21 assume that the transition time (r _) through the space charge layer of the collector is pure by the electron saturation (drift) velocity (vs). Ground is related to people. In the baseline calculation, it is like:, 〇A GaAs base layer disorder. The calculation is expected, "assuming 1.115 ps, and the total of the remaining emitter and collector transition time (R e + τ c) is regarded as 0.95 ps. Examination view 21 shows that although the fixed composition of the dagger is not completely outside the expected range of the traditional GaAs-based body, it is clearly at the lower end of the distribution The graded base structure has a significant improvement: '. The second calculation "b reduces the baseline of A 2/3) Suggestion: When the base transition _ for a structure with a fixed composition is reduced by about 5⑽. Compared to the base layer with a fixed composition, t, this shows that the carrier speed has been increased twice in the graded base layer. This is because the carrier speed doubles (2χ) and the base thickness increases by 33. % Is the expected reduction in rb to 1/2 χ 4/3 = 2/3. The third calculation (... salty 1/3 and (r, rc) reduced to 1/2 baseline) approximates the use of a thin and / or hierarchical base structure and a modified layout and size of 7 ( In order to minimize the value of… and ^) Example 3 In order to improve the efficiency of the amplifier and thus reduce the operating voltage and extend the battery life, it is desirable to reduce the offset voltage (VcEsat) and the corner voltage ⑹. The method is to minimize the asymmetry 打开 of the turn-on voltage of the base / emitter and base / collector diode pairs. _ Collector Energy 31 200308088 Wide DHBT has been shown to produce low # u, but in fact this This results in higher Vk and reduced efficiency because it is difficult to control the potential barrier at the base / collector heterojunction. Insertion of a thin layer (tunneling collector) with a two-band gap allows simultaneous reduction of vCE, sat and vk, which improves the efficiency of the device. Figure 22 shows a diagram of a graded GalnAsN base layer and a tunneling collector DHBT. The base layer system is graded such that there is a band gap between the emitter and collector junctions There is an energy difference of about 40 meV. Between the base and collector A 100-thick-thick tunneling collector was made, which is composed of a high-band gap material A. ". The figure shows the band gap diagram of DHBT in Figure 22. The simple wet method is used. A large area element (L = 75〆mx75 &lt; &quot;) was made by etching, and tested in a common base and common emitter architecture. Figure 24 shows its graph, and Figure 25 shows the common emitter characteristics of the DHBT in Figure 22 As can be seen in Figure 24 and the drawing, the element has a low offset voltage of about 0.12V. Although the equivalent has been particularly shown and described with reference to the preferred specific aspects, the person skilled in the art is familiar with it It will be understood that, without departing from the scope of the present invention, which is encompassed by the scope of the appended patent application, various changes can be made in type and detail. [Simplified description of the drawings] (I) Schematic illustration of the drawings InGaP / GalnAsN DHBT structure of a preferred embodiment of the present invention, where X is approximately equal to 3y. 32 200308088 Figure 2 is a Gummel diagram illustrating the InGaP / GalnAsN DHBT of the present invention and the prior art InGaP / GaAs HBT and GaAs / GaAs B JT Base and collector currents as a function of turn-on voltage. Figure 3 illustrates the turn-on as a function of the resistance of the base sheet for the InGaP / GalnAsN DHBT of the present invention and the prior art InGaP / GaAs HBT and GaAs / GaAs BJT. Voltage (when Jc = 1.78 A / cm2). Figure 4 illustrates the luminescence of the InGaP / GalnAsN DHBT of the present invention and the InGaP / GaAs HBT of the prior art measured at a nominal base thickness of 1,000A and 77 ° K. read. #Etch the InGaAs and GaAs coatings and selectively terminate at the top of the InGaP emitter before measuring the luminescence. The band gap of the η-type GaAs collector of both InGaP / GaAs HBT and InGaP / GalnAsN DHBT is 1.507eV. The band gap of the p-type GaAs base layer of InGaP / GaAs HBT is 1. 455eV, while the band gap of the p-type GalnAsN base layer of InGaP / GalnAsN DHBT is 1. 408eV. Fig. 5 illustrates a double crystal x-ray diffraction (DCXRD) spectrum of the InGaP / GalAsAs DHBT of the present invention and the prior art InGaP / GaAs HBT at a nominal base thickness of 1500A. The location of the base layer spikes is indicated. Figure 6 is a Polaron C-V profile diagram illustrating the carrier concentration across the base layer thickness in the InGaP / GalnAsN DHBT of the present invention and the InGaP / GaAs HBT of the prior art. The nominal base thicknesses of both InGaP / GalnAsN DHBT and InGaP / GaAs HBT are 100 OA. The Polaron contours of the two were obtained after selectively etched down to the top of the base layer. FIG. 7A illustrates a better InGaP / GalnAsN DHBT structure. 33 200308088 has a transition layer between the emitter and the base, and 隹 ^ ^ ^ has a transition layer and a lattice match between the collector and the base. Floor. Figures 7B and 7C illustrate an alternative, InGap / GaInAsN DHBT structure &apos; having a compositionally graded base layer. Figure 8 is a graph of the product of doped X mobility in a carbon-doped GalnAsN base layer grown at a constant indium source gas flow rate as a function of carbon tetrabromide flow rate ("TMIF" is three Flow of methyl indium). Figure 9 is a graph of the flow rate of if-to-tetrabromide required for obtaining a GainAsN base layer with a graded composition of the mobility product while growing a heterogeneous dopant. Figure 10 (a) shows a graph in which the turn-on voltage of InGaP / GalnAsN HBT is lower than that of InGaP / GaAs HBT. FIG. 11 is a graph of the ΔVbe vs. carbon tetrabromide flow rate in a carbon-doped GalnAsN base layer grown under a fixed TMIF. Figure 12 is a graph of Δ TMIF. Fig. 13 is a DHBT structure having a composition-graded base layer used in the experiment of Example 2. Fig. 14 is a DHBT structure with a fixed base layer used in the experiment of Example 2. FIG. 15 is a Gummel diagram comparing a DHBT having a fixed composition GaInAsN base layer and a DHBT having a compositionally graded GalnAsN base layer. Figure 16 is a graph comparing the DC current gain of a dhbt with a fixed composition of the GalnAsN base layer and a DHBT with a compositionally graded GalnAsN base layer as a function of the base sheet resistance. 34 200308088 Figure 17 is a Gummel diagram comparing a DHBT with a compositionally graded GalnAsN base layer and two DHBTs with a fixed composition of the GalnAsN base layer. Figure 18 is a graph comparing the DC current gain of a DHBT with a GalnAsN base layer with a composition hierarchy and two DHBTs with a GalnAsN base layer with a fixed composition as a function of collector current density. Figure 19 is a graph comparing the extrapolated current gain cut-off frequency of a DHBT with a fixed composition of the GalnAsN base layer and a DHBT with a compositionally graded GalnAsN base layer as a function of the collector current density. Fig. 20 is a graph comparing the small signal current gain of the G a I n A s N base layer with a fixed composition to D Η B T with a compositionally graded GalnAsN base layer as a function of frequency. FIG. 21 is a comparison between the DHBT with a fixed composition of the GalnAsN base layer and the DHBT with a compositionally graded GalnAsN base layer. For a conventional HBT with a GaAs base layer, the peak f t is BVce. Graph as a function. Figure 22 is a table showing the composition of a DHBT with a hierarchical GalnAsN base layer and a tunneling collector. FIG. 23 is a diagram of a band gap of the DHBT described in FIG. 22. FIG. 24 is a Gummel diagram of the DHBT described in FIG. 22. FIG. 25 shows common emitter characteristics of the DHBT described in FIG. 22. (Two) the symbol of the component (none)

Claims (1)

200308088 Scope of patent application ·· 1 · —A heterojunction bipolar transistor including: (a) n-type doped collector; (b) formed on the collector and includes a base of a Π-ν family material , Where the IV material includes indium and nitrogen, and where the base is doped with carbon having a concentration of about 1.5 × 10i9cnr3 to about 70 × 1019cm_3, and (c) an in-type doped emitter formed on the base pole. 2. The transistor as described in the patent application, wherein the base includes gallium, indium, arsenic and nitrogen. 3. The transistor according to item 2 of the patent application, wherein the collector is GaAs, the emitter is InGaP, AUnGap, or A1GaAs, and the transistor is a double heterojunction bipolar transistor. 4. As for the transistor in the second item of the patent application, the band gap of the base layer is at the surface where the base layer is in contact with the collector, and the band gap of the base layer is in contact with the emitter at the base layer. The surface at the surface is about 20 meV to about 120 meV lower. 5. The transistor according to item 4 of the patent application, wherein the band gap of the base layer is linearly graded from the surface where the base layer is in contact with the collector to the surface where the base layer is in contact with the emitter. 6. The transistor according to item 5 of the patent application, wherein the average band gap reduction of the graded base layer is in the range of about 20 mev to about 300 meV less than the band gap of GaAs. 7. The transistor according to item 6 of the patent application, wherein the average band gap reduction of the graded base layer is in the range of about 80 mev 36 200308088 to about 300 meV less than the band gap of GaAs. 8. The transistor according to item 6 of the patent application, wherein the reduction in the average band gap of the hierarchical base layer is in the range of about 20 meV to about 200 meV less than the band gap of GaAs. 9. The transistor as claimed in claim 3, wherein the base layer includes a layer having a chemical formula of Gai-xIrixAspyNy, where X and y are each independently about 1.0 × 1 (Γ4 to about 2.0 × 10_1. 10 · For example, the transistor in the ninth scope of the patent application, where X is approximately equal to 3y 〇11 · The transistor in the ninth scope of the patent application, where the X value is in the range of about 0.2 to about 0.02 at the collector, And graded to the range of about 0.1 to about zero at the emitter, provided that X is larger at the collector than at the emitter. 12. · As for the transistor in the 11th scope of the patent application, where χ is in the set At the pole, it is about 0. 06, and at the emitter, it is about 0.0j. 13. For example, the transistor in the patent application No. 10, wherein the base layer is about 400A to about 500A thick, and its sheet resistance is It is about 100 Ώ / | square to about 400 Ω / square. 14. As the transistor of the patent application No. 13, the concentration of the n-type dopant present in the emitter is about 3 5x1017 cm_3 To about 4.5XWW 'and the n-type dopant concentration range present in the collector is: about 9 &gt; &lt; 1015cn r3 to about 2x1016cnr3 15. For the transistor in the scope of patent application No. 14, in which the emitter and collector are doped with Shi Xi. 37 200308088 16. In the transistor in scope of patent application No. 15, where the emitter is It is about 500 people to about 750 mm thick, and the collector is about 3500 mm to about 4500 A thick. It includes n-type doped materials consisting of a group of free GaAs, InGaAs, and InGaAsN. The transistor of scope 16 further includes a ㈣-transition layer disposed between the base and the collector, the first-transition layer having a first surface adjacent to the first surface of the base, wherein the first transition layer includes 18. The transistor as claimed in claim 7 of the patent scope, further comprising a second transition layer having a first surface adjacent to the first surface of the emitter and a second surface having a second surface connected to the base. Two surfaces, where the second transition layer includes an n-type doped material selected from the group consisting of GaAs, InGaA ". InGaAsN. 19. For a transistor as claimed in item 18 of the patent application scope, further comprising a lattice matching layer, Adjacent to collector The first surface of the first surface and the second surface having a second surface adjacent to the first transition layer, wherein the lattice matching layer is a material with a wide band gap. 20. The transistor as claimed in item 9 of the patent application The lattice matching layer is selected from the group consisting of InGaP, AlInGaP, and A1GaAs. 21. The transistor of claim 18, wherein the first and second transition layers are about 40A to about 60A thick. 2 2. The transistor of claim 19, wherein the first and second transition layers are about 40A to about 60A thick, and the lattice matching layer is about 150A to about 250A thick. 38 200308088 23 · A method for manufacturing a heterojunction bipolar transistor, comprising the following steps: (a) growing on a n-type doped GaAs collector layer from gallium, indium, silicon and nitrogen sources including gallium, indium, A base layer of arsenic and nitrogen, wherein the base layer is p-type doped with carbon from an external carbon source; and (b) a ^ -type doped emitter layer is grown on the base layer. Where the external carbon comes from where the source of gallium is where the source of nitrogen is where the source of arsenic is where the base is in which the base is in which the base layer is covered independently. 24. For the method in the 23rd scope of the patent application, the source is four Carbon bromide or tetragas. 25. The method of claim 24, which is selected from the group consisting of trimethylgallium and triethylgallium. 26. The method of claim 25, such as gas, dimethylhydrazine or tertiary butylamine. 27. The method according to item 26 of the patent application, wherein the ratio of the gallium source is about 2.0 to about 3.5. 28. The method according to item 27 of the scope of the patent application grows at a temperature below about 750 ° C. 29. The method according to item 28 of the scope of patent application. Growth at a temperature of about 500 ° C to about 600 ° C. 30. A method as claimed in item 28 of the scope of patent application, including a layer with a chemical formula of GahInxASbNy, where χ sum is approximately 1.0 × 1 (Γ4 to approximately 2. 〇χΐ〇-ι. 31. Such as the scope of application for item 30 Method, where the collector is approximately equal to 3y 〇x, where the collector includes 32. For example, method 30 of the scope of patent application 39 200308088 GaAS, the emitter includes a material selected from the group consisting of 1nGaP, AlInGaP, and AlGaAs: The crystal is a method of double heterojunction bipolar electricity = ·, * applicable patent scope of 30 items. The method of growing a base and attracting progress includes growing a second layer of an n-type doped first transition layer on a collector layer. Step 'where the base layer is grown on the n-type doped first transition layer, and the -transition layer has a band gap smaller than the collector band gap. The method of the 331th range, in which a group consisting of the first transition, free GaAs, InGaAs, and InGaAsN. 35. The method of the "Scope of Patent" item, which is further included on the base before the long =: polar layer Grow the second transition layer— The transition layer has a second surface that is adjacent to the base surface and a second surface that is connected to the emitter surface. The doping X of the Xu-Le transition layer is at least one order of magnitude lower than the doping concentration of the emitter. 6 The method from the 351th patent range, wherein the second transition k is a group consisting of GaAs, InGaAs, and InGaAsN. The method from the 36th patent range, wherein the two formed have doped spikes. ㈣—The transition layer and the second transition layer are heterogeneous: the third method of the scope of patent application is as follows: before the η-type layer is grown, the Γ transition layer further includes a lattice-matching second-second crystal grown on the collector. The lattice matching layer has a surface adjacent to the first surface of the collector and a second surface adjacent to the second surface of the -transition layer. 40 200308088, as in the patent application No. 3 ... method, the layer includes InGaP. 40.- A kind of material including gallium, fine, kun and nitrogen, wherein the doping is at a carbon concentration of about 15 × 1 η19 per material MW Cm to about 7.0 × 1019 cm-3. 凊 Patent Range Item 40 Material, which can be composed of chemical formula, In xAsl_yNy to represent, where 2 independently is about ihio'4 to about 2.0x10-ι. Different from the two "please apply for the material of the 41st item, where x is about 43. If the patent application range 42 _ # ^ about 0 01. 々 外 大 44. If the scope of application for patent No. 43 is at least about 3.0x100W, the material with a concentration of 22, ㈣, and nitrogen, wherein the second surface of the material a . — '' To represent 'and a7 independently and linearly grades the smaller value at one surface from the larger value at the surface of the material's younger brother. Hui "invites the material in the scope of patent 45, which is doped with carbon. T 7 * The material in the scope of patent application 46, where χ is linearly graded from about 0.01 at the second surface of the material to the One is about 0.06. The surface composition of 48 · —including gallium, indium, arsenic, and nitrogen—is represented by chemical formulas A-A, and the 2: 200: 088 of the larger value at the surface is linear The ground is graded to a smaller value at the second surface of the material, and y remains substantially constant throughout the material. 49. For example, the material of scope 48 of the patent application, wherein the material is doped with carbon. 50 · For example, for the material in the 49th area of the patent application, χ is linearly graded from about 0.01 at the second surface of the material to about 0.06 at the first surface of the material, and y is about 0.001. 51. One form A method for a graded semiconductor layer having a product of a substantially linear graded band gap and a substantially constant doped 乂 mobility from a first surface through the layer to a second surface, the method comprising The following steps: (a) The product of the doping x mobility of more than one correction layer, where each layer is a clearly distinguished flow of any one of an organometallic compound of an IU or v group atom of the periodic table of the deposition or a tetra-carbon compound of the deposited carbon Formation at a constant rate, thereby determining the formation of a substantially constant doping χ mobile 丨 the relative flow rate of the organometallic compound and carbon tetrahalide required for production and production; and (b) the relative flow rate Next, the organometallic compound and the tetrahalogen compound are flowed on a surface to form a product of a substantially constant doping χ mobility, and the flow rate is changed during deposition, thereby forming a semiconductor that is graded throughout. The layer is basically a linearly graded gap. 52. The method of claim 51, further comprising the step of depositing the graded layer on the second semiconductor layer during the manufacturing of the bonded 7L piece. 42 200308088 53 · The method according to the scope of patent application No. 52, wherein the second semiconductor layer is a collector layer. 54 · The method according to the scope of patent application No. 52, wherein The semiconductor layer is an emitter layer. 55. The method of claim 51, wherein the graded semiconductor layer includes gallium, indium, and arsenic, and determines the deposition rate of the carbon tetrahalide to form a substantially constant doping χ The organometallic compound of the mobility product includes an organic indium compound. 56. The method of claim 55, wherein the carbon tetrazide is CBr4. Wherein the organometallic deposit is further divided into steps. Wherein the third semiconducting u * · Gekou Jiahua mouthpieces further include nitrogen source gas. 58 · If the method of the patent application No. 52, and the second semiconductor layer of the semiconductor layer includes GaAs 乂 59 · If the application of the patent scope No. 58 A third semiconductor layer is deposited on the blade-level semiconductor layer. 60. For example, the rectangular solid layer of the 59th scope of the patent application is InGaP. For example, the doped X mobility of the 58th layer of the patent scope is multiplied by #, where the -th and second / plane of each correction: are related to the band gap, thereby the band at the surface of the grading layer Gap fit = The relative flow rate of the multiplier and tetragonal carbon that will be corrected for the deposition of this stepped semiconducting wide mobility. h. The organic metallization 62. If the scope of the patent application is 帛 61j, the gap between the band 20032003088088 layer as the base layer is corrected relative to GaAs to become a base emitter using the correction 0 component. Extremely electric dust. 63. The base layer in the bipolar transistor formed in the graded bumper / / in the scope of the patent application No. 62. b4. For example, the ancient 4 compound ττα # in the scope of the patent application. Method 'where the flow rate of the organometallic tetrahalide causes M-M pi' ^ λ *. »^, A λ _ the energy of the graded base layer ▼ the gap from the heterojunction bipolar transistor / collector junction reduce. ^ W emitter junction to base 65. Conductor material One, schema: as the next page 〇 44