JP2003332353A - Semiconductor device for communication equipment and equipment for communication system - Google Patents

Abstract

(57) [Problem] To simultaneously realize high speed and high breakdown voltage in a heterojunction field effect transistor which plays an important role in communication technology. SOLUTION: In a hetero-junction field-effect transistor, a band structure is optimized by applying tensile strain to a channel layer through which a two-dimensional electron gas flows and adjusting band discontinuity on a valence band side. Since a two-dimensional electron gas having a higher mobility and a higher carrier concentration than in the prior art is realized, the device operates at a higher speed, and a high breakdown voltage can be achieved by suppressing ionization due to ionization collision.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a heterojunction field effect transistor and a method for manufacturing the same.

[0002]

2. Description of the Related Art Conventionally, I that is lattice-matched to an InP substrate
Heterojunction field effect transistors (hereinafter referred to as HFETs) using nAlAs / InGaAs heterojunctions have been advanced in performance and are being applied to integrated circuits.

FIG. 11 is a sectional structural view of a typical conventional HFET. As shown in FIG. 11, semi-insulating InP
A buffer layer, undoped I, is formed on the substrate 1101.
The nAlAs layer 1102 (hereinafter, undoped is referred to as i-) is 2000 Å, i-In serving as a channel layer
The GaAs layer 1103 is 150 Å and the spacer layer is In
The AlAs layer 1104 is set to 20 Å, for example, 5 × 1012 cm
-2 is used as a barrier layer for the carrier supply layer 1105 having an atomic layer doping surface having Si with a surface density of -2.
As layer 1106 is 150 Å, 1 × 1019 cm-3 S
The n + -InGaAs layer 1107 having i is used as a cap layer and grown sequentially by the epitaxial method.

The semiconductor surface layer is, for example, AuGe /
Source electrode 1110 and drain electrode 11 made of Ni or the like
09 ohmic contact region is formed and electrically connected to the two-dimensional electron gas 1108. The region of the gate electrode 1111 is etched to form a cap layer 1107.
And the gate electrode 1111 is formed on the barrier layer 1106.

The concentration of the two-dimensional electron gas 1108 having high mobility is changed by the voltage applied to the region of the gate electrode 1111 so that the source electrode 1110 and the drain electrode 1109 can be changed.
Transistor operation is obtained by controlling the current flowing through.

[0006]

In the HFET having the above structure, InP is formed in the channel layer 1103 in which carriers travel.
InGaAs that is lattice-matched to is used. Also, I
InA so as to be adjacent to the nGaAs channel layer 1103
Although the 1As spacer layer 1104 is used, this layer has a conduction band discontinuity amount of 0.5 eV with InGaAs, which realizes strong confinement of the two-dimensional electron gas. In addition, the InGaAs layer has a small effective mass of electrons and can achieve high mobility, which is advantageous for the high-frequency characteristics of the transistor.

On the other hand, however, InGaAs lattice-matched with the InP substrate has a band gap of 0.77 eV.
Is small, collision ionization is likely to occur, and when the traveling carrier has large kinetic energy, it is more likely to transit to the upper level.

When impact ionization occurs, electron-hole pairs are generated. Of these, electrons flow to the drain electrode like other carrier electrons, but holes cannot flow to the source / drain electrodes due to the existence of the energy barrier. There is also an energy barrier for holes between the channel layer and the buffer layer, and holes generated in the channel layer by impact ionization remain accumulated in the channel layer. Further, as the frequency of impact ionization increases, the hole concentration increases, and some holes flow into the gate electrode, which is one of the causes of the gate leakage current.

Further, the holes accumulated in the channel change the potential distribution in the transistor, change the source resistance and the threshold voltage, and as a result, the output characteristics of the transistor become unstable. In addition, changes in these characteristics may induce an increase in drain current, which may lead to transistor breakdown by further increasing impact ionization. That is, the breakdown voltage of the transistor is significantly reduced due to the accumulation of holes.

On the other hand, even if the carrier does not have the energy required for collisional ionization to occur, if the traveling carrier has a large kinetic energy and transitions to an upper level, the trapping of the carrier becomes weak and the mobility is increased. There is a problem that

The problem to be solved by the present invention is to solve the above problems and to provide an HFET having excellent withstand voltage and stable characteristics, and having a reduced gate leakage current.

[0012]

The above-mentioned problem is that holes generated by impact ionization are accumulated in the channel, and it is necessary to operate the channel with a material having a low band gap energy. . A low bandgap energy is equivalent to a low effective mass of electrons, which is an essential property for high-speed operation. Therefore, it is necessary to use a material having a low energy gap as much as possible in the channel portion where electrons flow, and to strengthen the confinement of the two-dimensional electron gas to suppress impact ionization and prevent holes from being generated. Further, since confinement is required to be sufficient to generate a high-concentration and high-mobility two-dimensional electron gas, the amount of conduction band side band discontinuity existing at the interface between the channel layer and the spacer layer is It should be as big as possible. By creating a state of ultra-high mobility in this way, the acceleration electric field between the drain and source in the HFET can be made small, and at the same time, the mutual conductance can be made large, so that the noise figure can also be improved.

In order to achieve these objects, in the present invention, tensile strain is applied to the channel layer to increase the mobility of the two-dimensional electron gas. In addition, by introducing an Sb (antimony) -based material into the spacer layer and into the channel layer by using a material that positively uses a group V element such as P (phosphorus) and N (nitrogen), the energy bandgap is increased. The confinement of the two-dimensional electron gas is maintained without much change, and the energy barrier between the channel layer and the buffer layer is lowered.

[0014]

BEST MODE FOR CARRYING OUT THE INVENTION (Embodiment 1) A first embodiment of the present invention will be described with reference to the drawings.

[0015] Figure 1 is a relative material comprising a substrate or base having a _s becomes a lattice constant (a> a _s or a <a _s)
Shows how the lattice deformation occurs and how stress is applied when materials with different lattice constants are laminated. Of course, if the difference in the lattice constant is too large, the material itself will be broken, and the strain will be relaxed along with the generation of dislocations and defects. Therefore, the following explanations are all about the condition where the strain critical film thickness is not exceeded.

As shown in FIG. 1, when a material having a lattice constant larger than that of the underlying material is laminated, the material tries to align the lateral lattice constant with the underlying layer, so that the compressive stress is increased. On the other hand, epitaxial growth is performed by increasing the lattice constant in the vertical direction. Also, when trying to stack a material having a lattice constant smaller than that of the underlying material, the material tries to align the lateral lattice constant with the underlying layer, so tensile stress is applied, while the longitudinal lattice constant is increased. It is made smaller and epitaxial growth is performed.

Since the lattice is deformed by trying to make the lattice constant in the lateral direction uniform with the underlying material by changing the lattice constant in this way, the band structure also changes according to the deformation. Receive. This is shown in FIG.

In FIG. 2, as an example, In on an InP substrate is used.
It shows about GaAs. In _1-x Ga _x As is lattice-matched with the InP substrate when its composition is In _0.53 Ga _0.47 As,
The band gap energy at this time is 0.77 eV. Figure 2 shows the potential state at this time.
It is (b). It should be noted here that the heavy holes (HH) and the light holes (LH) in the valence band are condensed and have a momentum of 0.
In other words, at the point of, that is, at the point of k = 0, both are overlapping.

In contrast to this state, when the In composition is increased (x <0.47), the lattice constant of InGaAs becomes larger than that of InP, and the state of x <0.47 is obtained. The band gap energy should be small, as indicated by the dashed line in 2.

However, as just discussed, when epitaxially growing InGaAs having a composition having a lattice constant larger than that of InP, compressive stress is applied as described above,
The state shown in FIG. As shown in FIG. 2 (a), the energy band gap should be smaller than that in the case of lattice matching with InP as shown by the dotted line in FIG. 2, but the composition due to the compressive stress. It will have a larger bandgap than that calculated from. At the same time, the state functions of HH and LH are degenerated, and the potential top positions are separated,
When compressive stress is applied, HH will come above LH. It should be noted that δ is the amount of change in the band gap due to strain, and here, δ is allocated to the conduction band side by 2/3 and the valence band side by 1/3. Further, ξ is a value obtained by equally dividing the separated amount of HH and LH. The band gap in this case is to be seen between the potential maximum of the electron and the potential maximum of HH, and the magnitude is the band gap energy obtained from the composition Eg (x).
Then, Eg (x) −δEhy + ξ.

On the other hand, when the composition of In is reduced (x> 0.4)
7) Since the lattice constant of InGaAs is smaller than that of InP and x> 0.47, the band gap energy should be large as shown by the dotted line in FIG. 2 unless strain is taken into consideration. However, as I have just discussed, I with a composition such that the lattice constant is smaller than that of InP.
When nGaAs is epitaxially grown, tensile stress is applied as described above, and the state shown in FIG. 2C is obtained. Figure 2
As shown in (c), the energy band gap should be larger than that in the lattice matching with InP as shown by the dotted line in FIG. 2, but due to the tensile stress, the composition is different from the composition. It will have a smaller bandgap than the indexed bandgap. Also,
At the same time, the state functions of HH and LH are degenerated, and the respective potential peak positions are separated as in the case of compression, and when tensile stress is applied, LH is opposite to that of compression stress.
Comes over HH. Note that δ is the amount of change in the band gap due to strain, and here, δ is 2/3 on the conduction band side,
It is allocated at 1/3 on the valence band side. Ξ is HH and L
It is a value obtained by equally dividing the separated amount of H. The band gap energy in this case is to be seen between the potential minimum of the electron and the potential maximum of the LH.
The magnitude is Eg (x) −δEhy−ξ, where Eg (x) is the band gap energy obtained from the composition.

In the above description, the bandgap energy is changed by changing the composition, and at the same time, the bandgap energy is also changed by the effect of strain, which makes the story complicated, but if the bandgap energy is changed. Summing up the above assuming that there is no state, the band gap energy becomes larger than the original value when a compressive stress is applied, and becomes smaller than the original value when a tensile stress is applied.

Next, the phenomenon occurring at the hetero interface when strain is applied will be described with reference to FIG. When the In composition is increased, x <0.47 and the compressive stress is applied as described above. As described above, the band gap energy at this time becomes larger than the size determined from the composition. A bandgap energy E _g ^A indicated by a dotted line in FIG. 3 (a) is indexed from the composition, .DELTA.Ec in the conduction band by the compression strain, Delta] Ev only the position with the separation of the HH and LH occurs in the valence band side And the hetero interface is finally formed as shown by the solid line in FIG. At this time, δ is allocated to the conduction band side by 2/3 and the valence band side by 1/3, but the conduction band side simply increases by 2 / 3Ehy, while the valence band side is slightly complicated. Is. That is, the increase of only 1/3 Ehy is allocated to the valence band side,
On the valence band side, HH and LH are separated, so the band gap energy calculated from the composition rises by ξ toward the HH side and ξ toward the LH side when viewed from the position where the valence band side peak is lowered by 1/3 Ehy. Will go down. Therefore, the amount of change on the valence band side is “1 / 3Ehy−ξ”. This state is shown in both FIG. 2 (a) and FIG. 3 (a). Also,
From this, the amount of band discontinuity shown in FIG. 3A is determined from the composition, which is “ΔEc−2 / 3Ehy” for the discontinuity amount ΔEc on the conduction band side with InP which is determined from the composition. For the discontinuity amount ΔEv with InP on the valence band side, it is “ΔEv−1 / 3Ehy + ξ”.

As can be seen from this fact, when compressive stress is applied, the energy becomes larger than the band gap energy determined from the composition, and the rate of its spread is divided more toward the conduction band side. Become. That is, the heterojunction is formed in the direction in which the difference in the discontinuity amount ΔEv on the valence band side is reduced.

On the other hand, when the In composition is reduced,
Thus, x> 0.47 and tensile stress is applied. This and
Bandgap energy of mushrooms is determined from composition
As described above, it is smaller than the size.
E shown by the dotted line in FIG. _g ^AIs determined from the composition
Bandgap energy, depending on tensile strain
ΔEc is on the conduction band side, and HH and LH are separated on the valence band side.
As it occurs, the position shifts by δEv, and finally, as shown in FIG.
A hetero interface is formed as shown by the solid line in (b).
At this time, δ is 2/3 on the conduction band side and 1/3 on the valence band side.
It is allocated, but the conduction band side is simply 2 / 3E
While it is reduced by hy, the valence band side is slightly complicated.
That is, the decrease of only 1/3 Ehy is divided to the valence band side.
It is still shaken, but HH on the valence band side
Band that can be determined from the composition because separation of LH and LH occurs
The top of the gap energy on the valence band side is 1/3 Ehy
When viewed from the raised position, it rises by ξ to the LH side and then to the HH side.
It will be lowered by ξ. Therefore, the amount of change on the valence band side is
It becomes “1 / 3Ehy + ξ”. This is shown in Figure 2 (c)
Both are shown in FIG. Also, from this,
The band discontinuity amount shown in (b) is calculated from the composition.
The discontinuity amount ΔEc on the conduction band side with InP is
"ΔEc + 2 / 3Ehy", determined from the same composition
The discontinuity ΔEv on the valence band side with InP
This means “ΔEv + 1 / 3Ehy + ξ”.

As can be seen from this, when a tensile stress is applied, the energy becomes smaller than the bandgap energy determined from the composition, and the rate of its spread is different from that at the time of compressive stress. And the In composition. However, as a practical matter, giving a very large strain will significantly deteriorate the quality of the crystal, so it is not preferable to give a larger strain than necessary, so that the proportion of the bandgap expansion should be roughly equal. You can think.

To summarize the band discontinuity, when compressive stress is applied, the band gap energy expands and ΔEc and ΔEv decrease. The degree of contraction of the difference at this time becomes more remarkable on the conduction band side than on the valence band side. When a tensile stress is applied, the band gap energy becomes narrower and ΔEc and ΔEv become larger at the same time, but the extent of the difference at this time is almost equal.

Further, the effect of strain has the effect of suppressing the scattering of electrons regardless of whether it is tensile or compressive, and the electron mobility inevitably increases if strain is applied.

In the case of applying these things to the channel layer in the HEMT device, it is not preferable to increase the band gap because the mobility is better and the conduction band band discontinuity between the channel layer and the spacer layer is not preferable. It is important that the amount is large. Therefore, the strain applied to the channel layer must be tensile strain.

Based on the above, the embodiments of the present invention will be specifically described.

The cross-sectional structure of the HFET of this embodiment is shown in FIG.

In FIG. 4, 401 is a semi-insulating InP substrate, 402 is a buffer layer made of undoped InAlAs lattice-matched to the InP substrate 401, and 403 is, for example, I.
Lower channel layer 4 composed of undoped InGaAs lattice-matched to the nP substrate and forming a part of the channel layer, 4
Reference numeral 04 denotes a material having a lattice constant smaller than that of the InP substrate, for example, In having a composition ratio of In smaller than 0.53.
This channel layer made of GaAs, 405 is a spacer layer made of, for example, undoped InAlAs, 40
Reference numeral 6 denotes a carrier supply layer composed of an atomic layer doping surface having Si with an area density of 5 × 10 12 cm −2, for example, 407.
Is a barrier layer made of undoped InAlAs, 408 is an n + -InGaAs contact layer having Si of about 1 × 10 19 cm −3, 410 is a drain electrode, 411 is a source electrode, and 412 is a gate electrode.
On the semiconductor surface, ohmic contact regions of the source electrode 411 and the drain electrode 410 made of AuGe / Ni or the like are formed, and are electrically connected to the two-dimensional electron gas 409.

FIG. 5 shows a band structure diagram when the HFET having the above structure is viewed in the region XX 'shown in FIG. In FIG. 5, 501 is the aforementioned InGaAs lower channel layer 403, 502 is the aforementioned InGaAs main channel layer 404, 503 is the aforementioned InAlAs spacer layer 405, and 504 is the aforementioned carrier supply layer 406.
505 is the undoped InAlAs barrier layer 407 described above.
506 is a band structure diagram of the two-dimensional electron gas 409.

Now, when the In composition in the InGaAs main channel layer 502 is changed so as to be decreased, x>
0.47, and the InGaAs lower channel layer 501 and InGaA whose lattice constant is lattice-matched to the underlying InP.
Since the lattice constant is smaller than that of the InAlAs spacer layer 503 lattice-matched to InP on the s main channel layer 502, tensile stress is applied to the InGaAs main channel layer 502. As mentioned above, InG
The band gap energy of the aAs main channel layer 502 is larger than that in the case where lattice matching is estimated from the composition, but at the same time, tensile strain is applied, so that the band gap does not widen, while ΔE
c becomes large. For example, if x = 0.5, then E
Since g = 0.74 eV and ΔEc = 0.57 eV, the band gap energy can be increased only by slightly reducing the magnitude, and ΔEc can be increased. Further, since the electron scattering is suppressed by the effect of strain as described above, the mobility of the two-dimensional electron gas 506 becomes high. As a result, the mobility of the material itself of the channel layer 502 is increased and ΔEc is increased, so that the confinement in the two-dimensional electron gas 506 is strengthened and the mobility is increased, and the performance as a HEMT element can be improved.

In the first embodiment of the present invention, the lattice constant of the channel layer 502 is set smaller than that of InP to give tensile strain to the channel layer 502, and the movement of the material itself due to the reduction of the band gap. InAlAs lattice-matching with InP
Since the two main points are to increase the conduction band discontinuity ΔEc with respect to the spacer layer 503, even this requirement is satisfied, and the band gap energy is 1.0 eV.
Basically, any material may be used for the channel layer 502 as long as it is the following material.

In the present invention, the material used is III
Ga, In, and Al are used as group materials, and N, P, As are used as group V materials,
Considering III-V group compound semiconductors using Sb, examples of materials that meet the above requirements x>
InPN, InGaAsP, InGaPN, InGaA, other than 0.47 In _1-x Ga _x As
Examples include sN and InGaAsPN.

It should be noted that regardless of the presence or absence of distortion, basically by inserting P, the top position of the valence band is lowered and ΔEv
Becomes smaller, and by adding N, the conduction band minimum decreases and ΔEc increases without changing the position of the top of the valence band. Based on this, when tensile strain is applied to the channel layer 502 as in the present example, ΔEv is also increased. Therefore, by further reducing ΔEv and increasing ΔEc, a further device can be obtained. It can be expected to improve the characteristics.

For the same reason, the InAlAs spacer layer 503 lattice-matched with InP is not necessarily In.
_It does not have to be _0.48 Al _0.52 As, and it is also possible to introduce a mixed crystal in which Al, which is a group III element, and As, P, Sb of a group V element are mixed into the barrier layer and / or the spacer layer of the HFET element. Improvement of device characteristics can be expected.

The composition of the channel layer 502 is uniquely determined when the tensile strain is applied, and the epitaxial growth is performed on the lower channel layer 501 lattice-matched to the InP substrate. However, the composition of the channel layer 502 is as follows. The structure may be such that the lattice matching state with InP is continuously changed until the desired strain amount is obtained. At this time, since the channel layer 502 is lattice-matched to the InP substrate in the initial stage of growth, the above-described improvement in the characteristics of the HEMT device can be obtained without any problem even without the lower channel layer 501.

(Embodiment 2) A second embodiment of the present invention will be described with reference to the drawings. HFE of this embodiment
The sectional structure of T is shown in FIG. In FIG. 6, 601 is a semi-insulating InP substrate, and 602 is, for example, undoped InAl.
A buffer layer made of As, 603 is, for example, InGaAs
And a channel layer 604 made of, for example, undoped Al
A spacer layer made of AsSb, 605 is, for example, 5 ×
A carrier supply layer composed of an atomic layer doping surface having Si at an area density of 1012 cm −2, 606 is a barrier layer composed of, for example, undoped AlAsSb, and 607 is 1 ×.
N + -InGaA having Si of about 1019 cm-3
s contact layer, 609 is a drain electrode, 610 is a source electrode, and 611 is a gate electrode. On the surface of the semiconductor, ohmic contact regions of the source electrode 610 and the drain electrode 609 made of AuGe / Ni or the like are formed and electrically connected to the two-dimensional electron gas 608.

FIG. 7 shows a band structure diagram when the HFET having the above structure is viewed in the region XX 'shown in FIG. In FIG. 7, 701 is a channel layer 603 made of, for example, InGaAs, and 702 is, for example, undoped Al.
The spacer layer 604 made of AsSb is a barrier layer 6 made of, for example, undoped AlAsSb and the carrier supply layer 605 made of an atomic layer-doped surface having Si at a surface density of 5 × 10 12 cm −2.
06 is shown as a band structure diagram.

Now, consider a case where the As composition is changed in the AlAsSb spacer layer 702 so as to be reduced, that is, x ≦ 0.51 in AlAs _x Sb _1-x . At this time, AlAsSb has a larger lattice constant than InP. Therefore, the InGaAs channel layer 701 lattice-matched to the underlying InP is formed of AlAsSb.
Is subjected to tensile strain by the above, the electron scattering is suppressed and the band gap energy is reduced as described in the first embodiment of the present invention, and the minimum value of the conduction band is lowered, so that at the interface with the AlAsSb spacer layer 702. The band discontinuity ΔEc of the conduction band becomes large. As a result of the above phenomenon, the AlAsSb spacer layer 702 and the In
The mobility of the two-dimensional electron gas 705 existing at the interface with the GaAs channel layer 701 is increased, and the element performance of the HEMT device is improved.

The second embodiment of the present invention is characterized in that the material used for the spacer layer is a material having a lattice constant larger than that of InP. The material is basically arbitrary and can be improved. As an example, the channel layer 701 has an I
InGaAsP or InGaAsP lattice-matched to nP
N or the like is considered, and the spacer layer 702 has x ≦ 0.
In _1-x Al _x As set to 48 or Al set to x ≦ 0.51
AlAsPSb containing P in As _x Sb _1-x is considered, and InP is used for the carrier supply layer 703 and the barrier layer 704.
InAlAs which is lattice-matched to As described in the first embodiment of the present invention, the spacer layer 70
The composition of 2 does not have to be a single composition, and a desired strain amount of InP can be obtained from the channel layer 701 side toward the carrier supply layer side.
It may be a structure in which it is continuously changed so as to be in a lattice matching state with. In that case, the lattice mismatch may remain in the carrier supply layer 703 or the barrier layer 704.

(Embodiment 3) A third embodiment of the present invention will be described with reference to the sectional view of the element shown in FIG.
In FIG. 8, 801 is a semi-insulating InP substrate, 802
Is a buffer layer made of undoped InAlAs,
803 is, for example, I having a composition having a larger lattice constant than InP.
A lower channel layer made of nGaAs, and 804 is, for example, I
The present channel layer, which is made of undoped InGaAs lattice-matched to the nP substrate and is a part of the channel layer, 80
5 is, for example, a spacer layer made of undoped InAlAs, 806 is a carrier supply layer made of an atomic layer-doped surface having Si with a surface density of 5 × 10 12 cm −2, 807 is a barrier layer made of undoped InAlAs, and the like. 808 is S of about 1 × 1019 cm−3
An n + -InGaAs contact layer having i, 810 is a drain electrode, 811 is a source electrode, and 812 is a gate electrode. On the semiconductor surface, for example, AuGe / Ni
And the like, ohmic contact regions of the source electrode 811 and the drain electrode 810 are formed, and the two-dimensional electron gas 8
09 is electrically connected.

FIG. 9 shows a band structure diagram when the HFET having the above structure is viewed in the region XX 'shown in FIG. In FIG. 9, 901 is the above-mentioned InGaAs lower channel layer 803, 902 is the above-mentioned InGaAs main channel layer 804, 903 is the above-mentioned InAlAs spacer layer 805, and 904 is the above-mentioned carrier supply layer 806.
905 is the undoped InAlAs barrier layer 807 described above.
906 shows the two-dimensional electron gas 809 as a band structure diagram.

Now, the InGaAs lower channel layer 901
When the In composition is changed to increase in x, x <
0.47, the lattice constant becomes larger than that of the underlying InP substrate.
The main channel layer 902 made of InGaAs lattice-matched to InP on the first layer is pulled. Since the strain at this time is mainly applied to this main channel layer 902,
The effect of strain as described above is strongly exerted in this channel layer 902. That is, scattering of electrons is suppressed, and ΔEc formed between the spacer layer 903 and the spacer layer 903 is increased, so that the mobility of the two-dimensional electron gas 906 flowing through the channel layer 902 is improved.
For this reason, the confinement in the two-dimensional electron gas 906 becomes stronger, the mobility is increased, and the performance as a HEMT element can be improved.

By the way, the purpose of this embodiment is to make the lower channel layer 901 have a lattice constant larger than that of the InP substrate so as to give tensile strain to this channel layer 902 and increase the mobility of the two-dimensional electron gas flowing therethrough. Therefore, the material used for the lower channel layer may be any material as long as it has a lattice constant larger than that of InP. Basically, Ga, In, and Al are used as group III raw materials and N, P, As, and Sb are used as group V raw materials.
III-V group compound semiconductors made of InG.
aAsP, InPSbN, InPAsN, InAlAs
N etc. can be used.

At this time, the thickness of the main channel layer film 902 may be 2 nm or more, and the lower channel layer 90 may be formed.
It is also possible to treat 1 as a film thickness of 2 nm or more like an insertion layer. As described in the first embodiment of the present invention, the lower channel layer 901 does not have to be a layer composed of a single composition.
The structure may be such that the lattice matching state with P is continuously changed so as to obtain a desired strain amount. In that case, the lattice mismatch may start from the underlying layer of the lower channel layer.

If the channel layer 902 is to be strained, the lower channel layer 901 may be omitted and the buffer layer made of InAlAs existing thereunder may have a lattice mismatch. The materials to be used and how to provide the lattice mismatch are also as described above.

(Embodiment 4) A fourth embodiment of the present invention will be described with reference to the drawings. HFE of this embodiment
The sectional structure of T is shown in FIG. In FIG. 10, 100
1 is a semi-insulating InP substrate, 1002 is a buffer layer made of, for example, undoped InAlAs, and 1003 is I, for example.
A lower channel layer made of InGaAsP having a lattice constant larger than nP, 1004 is a main channel layer made of InGaAs having a lattice constant larger than InP, 1005, for example.
Is a spacer layer having a lattice constant larger than that of InP and made of undoped AlAsSb, 1006 is a carrier supply layer made of an atomic layer doping surface having Si with a surface density of 5 × 10 12 cm −2, and 1007 is undoped InAlAs, for example. A barrier layer composed of 10
08 is n + − having Si of about 1 × 1019 cm−3
InGaAs contact layer, 1009 is drain electrode,
Reference numeral 1010 is a source electrode and 1012 is a gate electrode.
On the semiconductor surface, ohmic contact regions of the source electrode 1010 and the drain electrode 1009 made of AuGe / Ni or the like are formed, and the two-dimensional electron gas 1011 is formed.
Is electrically connected to.

In each of the first to third embodiments of the present invention, a layer having a different lattice constant is prepared for only one region in each case, and a tensile strain is finally applied to the channel layer. However, there are cases where the strain cannot be sufficiently given due to the consideration of the strain critical film thickness and the energy band gap. In that case, it is possible to avoid that the strain is applied to only one region that is complexly distorted as in the above configuration. The combination in that case should just combine at least 2 or more types out of 3 types, and should just select suitably according to the conditions, such as the material used. In addition, the effect is synergistic by providing a complex distortion, and further improvement in device characteristics can be expected.

[0052]

As described above, at least one of the buffer layer, the spacer layer, the channel layer, or the channel layer in the HFET has a lattice constant different from that of the conventional InP substrate, and finally, By giving tensile strain to the channel layer in which the two-dimensional electron gas flows and optimizing the band gap and band lineup of the materials used at the same time, the mobility of the channel layer itself is increased, and a large conduction band band is obtained. It is possible to provide a HFET device that realizes discontinuity and realizes a two-dimensional electron gas having higher mobility and higher carrier concentration than before, and that suppresses ionization due to ionization collision.

[Brief description of drawings]

FIG. 1 is a diagram showing the relationship between strain and lattice constant.

FIG. 2 is a diagram showing how strain is applied and changes in energy band gap.

FIG. 3 is a diagram showing how distortion is applied and the amount of band discontinuity.

FIG. 4 is a side view of the HFET according to the first embodiment of the present invention.

FIG. 5 is a diagram showing the relationship between the HFET structure and the energy band according to the first embodiment of the present invention.

FIG. 6 is a side view of an HFET according to a second embodiment of the present invention.

FIG. 7 is a diagram showing the relationship between the HFET structure and the energy band according to the second embodiment of the present invention.

FIG. 8 is a side view of an HFET according to a third embodiment of the present invention.

FIG. 9 is a relational diagram of the HFET structure and the energy band according to the third embodiment of the present invention.

FIG. 10 is a side view of an HFET according to a fourth embodiment of the present invention.

FIG. 11 is a relationship diagram of energy bands of a conventional HFET.

[Explanation of symbols]

401 substrate 402 buffer layer 403 Lower channel layer 404 Channel layer 405 spacer layer 406 Carrier supply layer 407 barrier layer 408 Cap layer 409 Two-dimensional electron gas 410 drain electrode 411 Source electrode 601 board 602 buffer layer 603 channel layer 604 spacer layer 605 Carrier supply layer 606 barrier layer 607 Cap layer 608 Two-dimensional electron gas 609 drain electrode 610 Source electrode 611 gate electrode 801 substrate 802 buffer layer 803 Lower channel layer 804 channel layer 805 spacer layer 806 Carrier supply layer 807 barrier layer 808 Cap layer 809 Two-dimensional electron gas 810 drain electrode 811 Source electrode 1001 substrate 1002 buffer layer 1003 Lower channel layer 1004 Channel layer 1005 spacer layer 1006 Carrier supply layer 1007 barrier layer 1008 cap layer 1009 Two-dimensional electron gas 1010 drain electrode 1011 source electrode 1101 substrate 1102 buffer layer 1103 Channel layer 1104 Spacer layer 1105 Carrier supply layer 1106 Barrier layer 1107 Cap layer 1108 Two-dimensional electron gas 1109 drain electrode 1110 Source electrode 1111 Gate electrode

Claims (22)

[Claims] 1. A substrate of a III-V group compound semiconductor, a buffer layer formed on the substrate and formed of a III-V group compound semiconductor, and a buffer layer provided on the buffer layer and more than the buffer layer. In a channel layer formed of a III-V group compound semiconductor having a low potential at the conduction band edge, the channel layer body has a lattice constant different from that of the substrate and / or another layer has a lattice constant different from that of the substrate. A spacer layer formed of a III-V group compound semiconductor, which has a tensile strain by, is provided on the channel layer, and has a conduction band edge potential higher than a conduction band edge potential of the channel layer; An impurity formed of a III-V group compound semiconductor provided on the spacer layer and having a conduction band edge potential higher than the conduction band edge potential of the channel layer. Is provided on the electron supply layer, and is formed of a III-V group compound semiconductor that is provided on the electron supply layer and has a conduction band edge potential higher than the conduction band edge potential of the channel layer. And a barrier layer formed on the semiconductor device. 2. The semiconductor device for communication equipment according to claim 1, wherein the tensile strain of the channel layer is exhibited because the lattice constant of the channel layer body is smaller than the lattice constant of the substrate. A characteristic semiconductor device for communication equipment. 3. The semiconductor device for a communication device according to claim 2, wherein the substrate is InP, and the material forming the channel is at least one of Group III raw materials selected from In, Al, and Ga. A semiconductor device for communication equipment, which is a III-V group compound semiconductor configured to use at least one of N, As, P, and Sb as a group raw material. 4. The semiconductor device for communication equipment according to claim 3, wherein the material forming the channel layer is In _1-x Ga _x A.
A semiconductor device for a communication device, wherein s and the value of x satisfy x ≧ 0.47. 5. The semiconductor device for communication equipment according to claim 3, wherein the material forming the channel is InGaAsP,
A semiconductor device for communication equipment, which is one of InGaAsPN. 6. The semiconductor device for communication equipment according to claim 3, wherein the composition of the material forming the channel is lattice-matched on the substrate side, and the lattice constant is continuous or steps toward the spacer layer side. A semiconductor device for a communication device, which is configured to be smaller in size. 7. The semiconductor device for communication equipment according to claim 1, wherein the tensile strain of the channel layer is exhibited by making the lattice constant of the spacer layer larger than the lattice constant of the substrate. Semiconductor device for communication equipment. 8. The semiconductor device for communication equipment according to claim 7, wherein the substrate is InP, and the material forming the spacer is at least one of Group III raw materials selected from In, Al, and Ga. A semiconductor device for communication equipment, which is a III-V group compound semiconductor configured to use at least one of N, As, P, and Sb as a group raw material. 9. The semiconductor device for communication equipment according to claim 8, wherein the material forming the spacer layer is AlAs _x Sb.
_A semiconductor device for communication equipment, wherein _1-x , and the value of x is x ≦ 0.51. 10. The semiconductor device for communication equipment according to claim 8, wherein the material forming the spacer layer is InAlAS.
b, AlGaAsSb, or a semiconductor device for communication equipment. 11. The semiconductor device for a communication device according to claim 8, wherein the composition of the material forming the spacer layer is set to be larger than the lattice constant of InP on the channel layer side, and goes toward the carrier supply layer side. A semiconductor device for communication equipment, characterized in that its lattice constant is configured to approach InP continuously or stepwise. 12. The semiconductor device for a communication device according to claim 1, wherein the tensile strain of the channel layer is exhibited by making the lattice constant of the buffer layer larger than the lattice constant of the substrate. Semiconductor device for communication equipment. 13. The semiconductor device for communication device according to claim 12, wherein the substrate is InP, and the material forming the buffer layer is at least one of Group III raw materials selected from In, Al, and Ga. III-V configured to use at least one of N, As, P and Sb as a group V raw material
A semiconductor device for a communication device, which is a group compound semiconductor. 14. The semiconductor device for communication equipment according to claim 13, wherein the material forming the buffer layer is AlAs _x.
A semiconductor device for communication equipment, wherein Sb _1-x , and the value of x is x ≦ 0.51. 15. The semiconductor device for a communication device according to claim 13, wherein the material forming the spacer layer is InAlA.
s or AlGaAsSb, a semiconductor device for communication equipment. 16. The semiconductor device for communication equipment according to claim 13, wherein the composition of the material forming the buffer layer is In.
The semiconductor device for a communication device is characterized in that the P substrate side is matched with the lattice constant of InP, and the lattice constant becomes larger than InP continuously or stepwise toward the channel layer side. . 17. The semiconductor device for communication equipment according to claim 1, wherein the tensile strain of the channel layer is exhibited by making the lattice constant of the layer inserted in the channel larger than the lattice constant of the substrate. A semiconductor device for communication equipment characterized by the above. 18. The semiconductor device for communication equipment according to claim 17, wherein the substrate is InP, and the material forming the intra-channel insertion layer is at least one of In, Al, and Ga as a group III raw material. The group V raw material as N, As, P, S
II configured to use at least one of b
A semiconductor device for communication equipment, which is a group IV compound semiconductor. 19. The semiconductor device for communication equipment according to claim 17, wherein the material forming the intra-channel insertion layer is In.
_1-x Ga _x As _y P _1-y , and the value of x is x ≦ 0.2, y
The value of is y ≦ 0.6. A semiconductor device for a communication device. 20. The semiconductor device for communication equipment according to claim 13, wherein the material forming the spacer layer is InPSb.
A semiconductor device for a communication device, which is any one of N, InPAsN, and InAlAsN. 21. The semiconductor device for a communication device according to claim 13, wherein the composition of the material forming the intra-channel insertion layer is matched with the lattice constant of InP on the InP substrate side.
A semiconductor device for a communication device, characterized in that the lattice constant thereof is continuously or stepwise made larger than that of InP toward the channel layer side. 22. The semiconductor device for a communication device according to claim 1, wherein the tensile strain of the channel layer is (1) the lattice constant of the channel layer body is smaller than the lattice constant of the substrate, and (2) the spacer layer. The lattice constant of is larger than that of the substrate, and (3) the lattice constant of the buffer layer is larger than that of the substrate,
(4) At least 2 of making the lattice constant of the layer inserted in the channel larger than the lattice constant of the substrate
A semiconductor device for a communication device, which is expressed by using one or more means.