JP2013145906A - Semiconductor element, optical sensor device, and method of manufacturing semiconductor element - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a semiconductor element having a low dark current and whose light-receiving sensitivity is enlarged to a long-wavelength side of near infrared, and to provide an optical sensor device and a method of manufacturing the semiconductor element.SOLUTION: The semiconductor element 50 has: a type-2 (GaAsSb/InGaAs) MQW light-receiving layer 3 located on an InP substrate 1; and an InP contact layer 5 located on the MQW. In the MQW, with respect to GaAsSb, a composition x(%) is 44% or more, a film thickness z (nm) is 3 nm or more, and z(nm)≥-0.4x(atm.%)+24.6 is satisfied.

Description

  The present invention relates to a group III-V semiconductor device and a method for manufacturing the same, and more specifically, a type 2 multiple quantum well structure having a configuration that has light receiving sensitivity up to a long wavelength region in the near infrared ( The present invention relates to a semiconductor element, an optical sensor device, and a method for manufacturing a semiconductor element, including a multi-quantum well) in a light receiving layer.

A photodiode capable of obtaining a cutoff wavelength of 2 μm or more by forming an InGaAs / GaAsSb type II multiple quantum well structure on an InP substrate of a III-V compound semiconductor is disclosed (Non-patent Document). 1).
Further, an LED and a laser diode having an emission wavelength of 2.14 μm formed by forming InGaAs-GaAsSb type 2 MQW on an InP substrate as an active layer have been disclosed (Non-patent Document 2). This type 2 MQW GaAsSb has an Sb composition of 34 atm. % To 40 atm. %, And a strain compensation structure on the side lower than the lattice matching composition of InP. In the following description, “atm.%” Is simply referred to as “%”.
Further, without forming the type 2 MQW, a GaAsSb single phase with a changed Sb composition is grown on an InP substrate and the wavelength of PL is measured. Then, a light-emitting element using InP / GaAsSb type 2 MQW is considered.
Furthermore, a semiconductor laser element having a GaInNAsSb quantum well structure has been disclosed (Patent Document 1). This GaInNAsSb quantum well structure is a single quantum well structure (that is, the number of pairs = 1).

JP 2005-197395 A

R. Sidhu, et.al. "ALong-Wavelength Photodiode on InP Using Lattice-Matched GaInAs-GaAsSb Type-II Quantum Wells, IEEE Photonics Technology Letters, Vol.17, No.12 (2005), pp.2715-2717 M.Peter, et.al. “Light-emitting diodes and laser diodes based on a Ga1-xInxAs / GaAs1-ySbytype II superlattice on InP substrate” Appl. Phys. Lett., Vol.74, No.14 (5 April 1999 ), pp.1951-1953 M.Peter, et.al. “Band gaps and band offsets in strained GaAs1-ySby on InP grown by metalorganic chemical vapor deposition” Appl. Phys. Lett., Vol. 74, No. 3 (18 January 1999), pp. 410 -412

Since the field of application of the photodiode using the semiconductor element described above is widened, it is desired to increase the light receiving sensitivity to the long wavelength side as much as possible. For this reason, it can be considered that type 2 InGaAs / GaAsSbMQW has a strain compensation structure. In the MQW having a strain compensation structure using InGaAs / GaAsSb, compressive stress is generated in one of the paired InGaAs / GaAsSb, and tensile stress is generated in the other to prevent distortion in the pair. In order to increase the wavelength of light reception sensitivity, it is desirable to generate tensile stress in InGaAs and compressive stress in GaAsSb. In Non-Patent Document 2, the Sb composition of GaAsSb is lowered and pulled by InGaAs. It is assumed that the MQW having such a strain compensation structure first deteriorates in crystallinity. For this reason, there is no example used for the light reception use to which the deterioration of crystallinity is directly linked to the increase in dark current. That is, there is no example used for the light receiving element, only the laser diode in Non-Patent Document 2 described above.
When forming an MQW having a repetition number of 100 to 300, in the MBE (molecular beam epitaxy) method, the molecular beam can be instantaneously switched with a shutter. For this reason, automatic switching of microcomputer-controlled valves is possible, and it has been considered that deposition by the MBE method is almost inevitable for the growth of a steep interface and a high-quality multiple quantum well structure. In particular, when considering only the above-mentioned problem of crystal growth of a GaAsSb layer that easily undergoes phase separation, a crystal growth method with strong non-equilibrium is necessary for epitaxial growth while preventing phase separation. For this reason, the MBE method which is a crystal growth method with strong non-equilibrium is suitable. Actually, the MBE method is used to form a GaAsSb layer (Non-patent Document 1).
All of the above semiconductor devices target near-infrared long-wavelength light. In this case, the cap layer or contact layer located on the surface side of type 2 MQW absorbs light in the target wavelength region. Preferably, it is made of a material that does not. For this reason, InP is often used for the contact layer. When InP is grown by the MBE method, since solid phosphorus is used as a raw material, phosphorus (P) adheres to the inner wall of the growth tank. The adhered phosphorus may ignite when the growth tank is opened during maintenance and is in contact with air. Therefore, a manufacturing method is used in which MQW is grown by the MBE method and the InP contact layer is grown by a growth method other than the MBE method. When switching from the MBE method to the OMVPE method, for example, the intermediate product InP wafer is once exposed to the air. Such exposure to air is susceptible to contamination by impurities. Further, the switching of the growth method as described above inhibits the production efficiency.

  An object of the present invention is to provide a semiconductor element, an optical sensor device, and a method for manufacturing a semiconductor element, in which dark current is low and the light receiving sensitivity is increased on the long wavelength side of the near infrared.

The semiconductor element of the present invention is formed on a group III-V semiconductor substrate. This semiconductor element includes a light-receiving layer of type 2 multiple quantum well structure positioned on a III-V group semiconductor substrate, and the multiple quantum well (hereinafter referred to as MQW: Multi-Quantum Well) is at least Ga, As, Sb. And a layer containing at least In, Ga, As, and a layer containing at least Ga, As, Sb has an Sb composition x (atm.%) Of 44 atm. % And the film thickness z (nm) is 3 nm or more and satisfies the following condition: z (nm) ≧ −0.4x (atm.%) + 24.6.
Here, the layer containing at least Ga, As, and Sb and the layer containing at least In, Ga, and As in MQW have substantially the same film thickness within a range of variation of ± 1.0 nm. For this reason, in the following description, when the film thickness z of one layer (for example, GaAsSb layer) is referred to, it is appropriate to interpret it as the film thickness z of the one layer or the other layer (for example, InGaAs layer).

The above-described invention uses an increase in the thickness of the quantum well (factor (F1)) in order to expand the light receiving region to the long wavelength side. According to said structure, the range which can light-receive can be expanded to the long wave side, maintaining a low dark current, without increasing a lattice defect density.
Regarding the film thickness z of the layer including at least Ga, As, and Sb, the film thickness z (nm) is set to 3 nm or more, and in relation to the Sb composition x (≧ 44 atm%) of the layer including at least Ga, As, and Sb, z (nm) ≧ −0.4x (atm.%) + 24.6 is satisfied. Thus, increasing the film thickness z is effective for expanding the light receiving wavelength region to the longer wavelength side. In the xz range limited by z ≧ −0.4x + 24.6, the long wavelength range where light can be received can be 2.4 μm or more. That is, the boundary z = −0.4x + 24.6 is an xz line formed by the Sb composition x and the film thickness z with the longest receivable wavelength being 2.43 μm.

The III-V semiconductor can be InP. As a result, the MQW of the type 2 InP compound semiconductor can be formed using an InP substrate that has been used for a long time, and a semiconductor element having sensitivity on the long wavelength side in the near infrared region can be easily obtained.

MQW has a strain compensation structure, and an Sb composition x of a layer containing at least Ga, As, and Sb and an In composition y (atm.%) Of a layer containing at least In, Ga, As are expressed as 100 ≦ x + y ≦ 104. , Distribution can be satisfied.
In the above invention, in order to expand the light receiving region to the longer wavelength side, the strain compensation structure (factor (F2)) is applied in addition to the above-described increase in the film thickness of the quantum well (factor (F1)). Good. According to the above configuration, in the MQW forming the strain compensation structure (factor (F2)), the Sb composition x of the layer containing at least Ga, As, and Sb is increased to increase the Sb composition x of the layer containing at least In, Ga, and As. When the In composition y is lowered, the valence band of the layer containing at least Ga, As, and Sb and at least In, Ga, As are maintained while maintaining lattice matching with InP (hereinafter referred to as “InP lattice matching”). The energy difference from the conduction band of the containing layer can be reduced. As a result, it is possible to expand the light receiving range to the long wave side while maintaining a low dark current without increasing the lattice defect density.
The relationship of 100 ≦ x + y ≦ 104 between the Sb composition x of the layer containing at least Ga, As, and Sb and the In composition y of the layer containing at least In, Ga, and As is such that the Sb composition x is increased from 44%. Therefore, it is necessary to reduce the lattice defect density while maintaining InP lattice matching.

  An InP contact layer may be provided on the MQW. As a result, the contact layer can be formed of a crystal layer that is highly transmissive to light on the long wavelength side in the near infrared region. The InP layer has many achievements as a contact layer and is excellent in surface smoothness. Therefore, the hygroscopic property is small and the durability is excellent.

The film thickness z (nm) of the layer containing at least Ga, As, and Sb in MQW depends on the Sb composition x.
(A1) 44 atm. % ≦ x <56.8 atm. %, Z <10 nm,
(A2) 56.8 atm. % ≦ x ≦ 68 atm. It is preferable that z <−0.625x + 45.5 in the range of%.
As described above, (F1) increase in the thickness of each layer of MQW, and (F2) increase in Sb composition in a layer containing at least Ga, As, and Sb and decrease in In composition in a layer containing at least In, Ga, As ( The distortion compensation structure is effective for expanding the light receiving wavelength range. In particular, regarding the increase in the film thickness of each layer of MQW in (F1), when the film thickness z of the layer containing at least Ga, As, and Sb is increased, the following may be performed in relation to dark current or light receiving sensitivity. Good.
(A1) 44 atm. % ≦ x <56.8 atm. In the range of%, it is not so difficult to maintain InP lattice matching between the layer containing at least Ga, As and Sb and the layer containing at least In, Ga and As. For this reason, even if the film thickness is increased and the deviation from strict lattice matching is enlarged, the deterioration of crystallinity is small. For this reason, by increasing the film thickness, it is possible to promote a longer wavelength while maintaining a low dark current. However, when the film thickness is increased, the overlap of the wave function between the holes confined in the layer containing at least Ga, As and Sb and the electrons confined in the conduction band of the layer containing at least In, Ga and As is caused. Get smaller. As a result, the quantum efficiency may decrease, and the light receiving sensitivity may deteriorate. When the film thickness z of the layer containing at least Ga, As, and Sb is 10 nm or more, the light receiving sensitivity is greatly reduced. In order to secure the light receiving sensitivity, the film thickness z is preferably less than 10 nm.
(A2) 56.8 atm. % ≦ x ≦ 68 atm. In the range of%, the Sb composition x deviates higher from the range in which the InP lattice matching is obtained only by the layer containing at least Ga, As, and Sb. A decrease in the In composition y of the layer containing at least In, Ga, As can be obtained, and InP lattice matching can be obtained in MQW. For this reason, roughly, the characteristics of the crystal film are not the best, and the InP lattice matching is barely maintained. When the thickness z of the layer containing at least Ga, As, and Sb is increased, the deviation from the InP lattice matching of the layer containing at least Ga, As, and Sb in the MQW becomes clear (proportionally the film The same applies to a layer containing at least In, Ga, and As that increases the thickness), and the lattice defect density increases (deteriorates) as the film thickness z increases. As a result, the dark current increases as the film thickness z increases. Therefore, in such a high Sb composition x range, the dark current depends on the film thickness z as well as the Sb composition x of the layer containing at least Ga, As, and Sb. The range of the Sb composition x and the film thickness z that brings the dark current within a practically acceptable level is z <−0.625x + 45.5.

  The film thickness z (nm) of the layer containing at least Ga, As, and Sb in MQW depends on (b1) 44 atm. % ≦ x <54.3 atm. %, Z ≦ 7 nm, and (b2) 54.3 atm. % ≦ x ≦ 68 atm. % ≦ z ≦ −0.27x + 21.7. As a result, it is possible to pursue improvement in sensitivity on the long wavelength side in the near infrared region while placing importance on good crystallinity and on reducing dark current.

At least In, Ga, lattice mismatch of the layer containing As and [Delta] [omega ₁ in MQW, at least when the Ga, As, lattice mismatch of the layer containing Sb and [Delta] [omega _2, of the whole multiple quantum well structure lattice mis Consistency Δω = {Σ (Δω ₁ × layer thickness including at least In, Ga, As + Δω ₂ × layer thickness including at least In, Ga, As} / {Σ (layer including at least In, Ga, As) Δω defined by “thickness + thickness of a layer including at least Ga, As, and Sb)}” is preferably set to satisfy −0.2% or more and 0.2% or less.
Here, “a is a lattice constant of a substrate, and a ₁ is a lattice constant of a layer containing at least In, Ga, As (eg, InGaAs), and a lattice mismatch of a layer containing at least In, Ga, As (eg, InGaAs). The degree Δω ₁ is defined as Δω ₁ = {(a−a ₁ ) / a} × 100% ”. For even [Delta] [omega _2, is the same. In the following description, the degree of lattice mismatch is the same.
With the above configuration, a layer containing at least Ga, As, and Sb and a layer containing at least In, Ga, and As cannot satisfy InP lattice matching, but both deviate from the lattice constant in the opposite direction. As a result, it is possible to obtain InP lattice matching by integration. As a result, the lattice defect density can be lowered and the low dark current can be maintained while changing the band structure so as to be advantageous for expansion to the longer wavelength side.

The layer containing at least Ga, As, and Sb can be GaAs _1-x Sb _x (hereinafter referred to as GaAsSb). This makes it possible to obtain a crystal layer having a relatively high level of conduction band and valence band, which is easy to achieve lattice matching with the InP substrate and is a favorable condition for constituting one layer of type 2 MQW. be able to.

The layer containing at least In, Ga, and As can be In _y Ga _1-y As (hereinafter referred to as InGaAs). This makes it possible to obtain a crystal layer having a relatively low level of conduction band and valence band, which is easy to achieve lattice matching with the InP substrate and is favorable for forming the other layer of the type 2 MQW. be able to.

  The longest wavelength with which the light receiving layer has light receiving sensitivity can be 2.4 μm or more. By satisfying the above xz relationship, the energy difference between the valence band of GaAsSb and the conduction band of InGaAs can be reduced to a wavelength equivalent to 2.4 μm or less. As a result, a substance having an absorption band of 2.4 μm or more can be measured, and can be used for a wider range of material inspection apparatuses.

It is preferable not to have a regrowth interface between the bottom surface of the light receiving layer and the top surface of the semiconductor layer including the light receiving layer and the InP contact layer.
Here, the regrowth interface means that after the first crystal layer is grown by a predetermined growth method, the second crystal layer is exposed to the first crystal layer and exposed to the first crystal layer by another growth method. Refers to the interface between the first crystal layer and the second crystal layer. Usually, oxygen, carbon, and silicon are mixed as impurities at a high concentration. The semiconductor element of the present invention should preferably have no regrowth interface and can maintain good crystallinity up to the surface of the InP contact layer. This can contribute to the reduction of dark current.
Moreover, a semiconductor element can be manufactured efficiently. That is, as will be described later, since the growth is consistently performed by the all organic MOVPE method from the (buffer layer to MQW) to the InP contact layer containing P, the manufacturing is continuously performed in the same growth tank. Can do. For example, even if an InP contact layer containing phosphorus is formed, solid phosphorus is not used as a raw material, so that phosphorus does not adhere to the inner wall of the growth tank. For this reason, there is no fear of ignition at the time of maintenance, and it is excellent in safety.

  The optical sensor device of the present invention is characterized in that any one of the above semiconductor elements is used as a light receiving element. Thus, an optical sensor device having a low dark current and sensitivity on the long wavelength side in the near infrared region can be obtained. This optical sensor device can include a CMOS having a readout electrode from each pixel of a semiconductor element (light receiving element), a spectroscope (diffraction grating), an optical element such as a lens, a control device such as a microcomputer, and the like.

The method for manufacturing a semiconductor device of the present invention manufactures a semiconductor device formed on a III-V semiconductor substrate. This manufacturing method includes a step of forming a type 2 MQW light-receiving layer on a group III-V semiconductor substrate, and MQW includes a layer containing at least Ga, As, and Sb and a layer containing at least In, Ga, and As. In the MQW forming step, the Sb composition x (atm.%) Of the layer containing at least Ga, As, and Sb is set to 44 atm. % Or more, and the Sb composition x (atm.%) And the film thickness z (nm) satisfy z ≧ −0.4x + 24.6.
Further, the III-V semiconductor can be InP.
Furthermore, the layer containing at least Ga, As, and Sb and the layer containing at least In, Ga, and As form a strain compensation structure, and the Sb composition x of the layer containing at least Ga, As, and Sb is 44 atm. %, The In composition y (atm.%) Of the layer containing at least In, Ga, As is changed to 1 atm. It can be grown to decrease at a rate of 0.9 to 1.2 times per% increase.
Further, a step of forming an InP contact layer on the MQW may be provided.

Further, the film thickness z (nm) of the layer containing at least Ga, As, and Sb in MQW is set to 3 nm or more, and (a1) 44 atm. % ≦ x <56.8 atm. %, Z <10 nm, and (a2) 56.8 atm. % ≦ x ≦ 68 atm. It is preferable that z <−0.625x + 45.5 in the range of%.
The point that the dark current of the semiconductor element manufactured by the above method is low and the light receiving sensitivity is expanded in the near-infrared long wavelength region is as described above.

  The film thickness z (nm) of the layer containing at least Ga, As, and Sb in MQW is set to 3 nm or more, and (b1) 44 atm. % ≦ x <54.3 atm. %, Z ≦ 7 nm, and (b2) 54.3 atm. % ≦ x ≦ 68 atm. % ≦ z ≦ −0.27x + 21.7. Thereby, it is possible to pursue an improvement in sensitivity on the long wavelength side in the near infrared region while ensuring good crystallinity.

In the MQW formation process, when the lattice mismatch degree of a layer containing at least In, Ga, As is Δω ₁ and the lattice mismatch degree of a layer containing at least Ga, As, Sb is Δω ₂ , a multiple quantum well structure Total lattice mismatch Δω = {Σ (Δω ₁ × layer thickness including at least In, Ga, As + Δω ₂ × layer thickness including at least Ga, As, Sb)} / {Σ (at least In, Ga, The thickness of the layer containing As + the thickness of the layer containing at least Ga, As, and Sb)} is preferably set to satisfy −0.2% to 0.2%. This makes it possible to maintain the crystal quality of the MQW while changing the band structure of the layer containing at least Ga, As, and Sb and the layer containing at least In, Ga, and As to be advantageous for expansion to the long wavelength side, thereby causing lattice defects. The density can be lowered.
The layer containing at least Ga, As, and Sb can be GaAsSb.
In addition, a layer containing at least In, Ga, and As can be InGaAs.

A semiconductor layer including MQW and InP contact layers should be consistently grown on the InP substrate by a total metalorganic vapor phase epitaxy. Here, the all organic vapor phase growth method refers to a growth method using an organic metal raw material composed of a compound of an organic substance and a metal for all raw materials used for vapor phase growth, and is referred to as an all organic MOVPE method.
According to said method, the above-mentioned semiconductor element can be manufactured efficiently. That is, since the InP contact layer containing P is consistently grown by the all-organic MOVPE method, it can be continuously manufactured in the same growth tank. For example, even if an InP contact layer containing phosphorus is formed, solid phosphorus is not used as a raw material, so that phosphorus does not adhere to the inner wall of the growth tank. For this reason, there is no fear of ignition at the time of maintenance, and it is excellent in safety.
Another advantage of the all organic MOVPE method is that MQWs with steep heterointerfaces between each layer can be obtained. With MQW having a steep hetero interface, high-accuracy spectrum spectroscopy or the like can be performed.

  In the MQW formation step, the MQW is preferably formed at a temperature of 400 ° C. or higher and 560 ° C. or lower. Thereby, MQW excellent in crystallinity can be obtained, and the dark current can be further reduced. The above-mentioned temperature is a substrate surface temperature monitored by a pyrometer including an infrared camera and an infrared spectrometer, and the substrate surface temperature being monitored. Accordingly, although it is the substrate surface temperature, strictly speaking, it is the temperature of the epitaxial layer surface in a state where a film is formed on the substrate. There are various names such as a substrate temperature, a growth temperature, and a film formation temperature, and all refer to the monitored temperatures.

  According to the semiconductor element and the like of the present invention, it is possible to expand the light receiving sensitivity to the long wavelength side of the near infrared while keeping the dark current low. Further, since the organic organic MOVPE method consistently grows from the MQW of the light receiving layer to the InP contact layer, the manufacturing efficiency is high, and there is no adhesion of phosphorus to the inner surface of the growth tank, so that safety is also excellent.

It is a figure which shows the semiconductor element in Embodiment 1 of this invention. It is an enlarged view of MQW of FIG. It is a figure which shows the band structure of type 2 (InGaAs / GaAsSb), and the light absorption (light reception) phenomenon. It is a figure which shows the range which Sb composition x (%) of GaAsSb and film thickness z of GaAsSb satisfy | fill of type 2 InGaAs / GaAsSbMQW. It is a figure which shows the piping system etc. of the film-forming apparatus of all the organic MOVPE method. (A) is a figure which shows the flow of an organometallic molecule | numerator, and the flow of temperature, (b) is a schematic diagram of the organometallic molecule | numerator in the board | substrate surface. 2 is a flowchart of a method for manufacturing the light receiving element 50 of FIG. 1. It is an optical sensor apparatus containing the light receiving element array (semiconductor element) in Embodiment 2 of this invention. In MQW of Example 1, it is a figure which shows the position and PL wavelength of each test body in the relationship between the Sb composition of GaAsSb, and a film thickness. It is a figure which shows the relationship between the film thickness of the quantum well of each test body in Example 2, PL wavelength, and light reception sensitivity. It is a figure which shows the relationship between Sb composition x of GaAsSb of MQW of each test body in Example 3, PL wavelength, and dark current.

(Embodiment 1)
FIG. 1 is a diagram showing a semiconductor element 50 according to the first embodiment of the present invention. The light receiving element 50 has an InP-based semiconductor multilayer structure (epitaxial wafer) having the following configuration on the InP substrate 1. In FIG. 1, light is incident from the InP substrate side, but may be incident from the epitaxial side.
(InP substrate 1 / InP buffer layer 2 / type 2 (InGaAs / GaAsSb) MQW light receiving layer 3 / InGaAs diffusion concentration distribution adjusting layer 4 / InP contact layer 5)
The p-type region 6 positioned so as to reach from the InP contact layer 5 to the MQW light-receiving layer 3 is formed by selectively diffusing Zn of the p-type impurity from the opening of the selective diffusion mask pattern 36 of the SiN film. The The fact that diffusion is introduced into the periphery of the light receiving element 10 in a limited manner in a planar manner can be achieved by diffusion using the selective diffusion mask pattern 36 of the SiN film. A p-side electrode 11 made of AuZn is provided in the p-type region 6, and an n-side electrode 12 made of AuGeNi is provided in ohmic contact with the back surface of the InP substrate 1. In this case, the InP substrate 1 is doped with n-type impurities to ensure a predetermined level of conductivity. An antireflection film 35 made of SiON is provided on the back surface of the InP substrate 1 so that light can enter from the back surface side of the InP substrate. In the type 2 MQW light-receiving layer 3, a pn junction 15 is formed at a position corresponding to the boundary front of the p-type region 6, and a reverse bias voltage is applied between the p-side electrode 11 and the n-side electrode 12. As a result, a depletion layer is formed more widely on the side where the n-type impurity concentration is low (n-type impurity background). The background of the MQW light-receiving layer 3 is about 5E15 cm ⁻³ or less in terms of n-type impurity concentration (carrier concentration). The position 15 of the pn junction is determined by the intersection of the background (n-type carrier concentration) of the light-receiving layer 3 of the multiple quantum well and the concentration profile of the p-type impurity Zn. The diffusion concentration distribution adjustment layer 4 is arranged for adjusting the concentration distribution of the p-type impurity in the MQW constituting the light receiving layer 3, but the diffusion concentration distribution adjustment layer 4 may be omitted. In the light receiving layer 3, the Zn concentration is preferably 5E16 cm ⁻³ or less.

The point of the light receiving element in the present embodiment is as follows.
(P1) The light-receiving layer 3 is composed of type 2 (InGaAs / GaAsSb) MQW, the Sb composition x of GaAsSb is set to 44% or more, the Sb composition x of GaAsSb is increased, and the increased amount is absorbed as a whole. The In composition y of InGaAs is lowered so as to maintain InP lattice matching. That is, GaAsSb and InGaAs form a strain compensation structure. This point (P1) is the same as the factor (F2) for enlarging the light receiving area described later to the near-infrared long wavelength side.
(P2) The film thickness z of the unit quantum well (GaAsSb) is set to 3 nm or more and satisfies z ≧ −0.4x (%) + 24.6. This feature mainly includes the above-mentioned factor (F1) film thickness increase, and also includes the above-mentioned factor (F2) as a secondary factor.
According to the above (P1), the energy difference between the valence band of GaAsSb and the conduction band of InGaAs can be reduced by increasing the Sb composition x and decreasing the In composition y. In addition, by maintaining the InP lattice matching, it is possible to grow an MQW, a contact layer, or the like having a low lattice defect density. As a result, the light receiving area can be expanded to a long wavelength area while keeping the dark current low.
Moreover, according to (P2), the long wavelength range in which light can be received can be 2.4 μm or more.
Furthermore, since it does not have a regrowth interface, high concentration oxygen, carbon, and silicon are not mixed as impurities. As a result, good crystallinity can be maintained up to the surface of the InP contact layer, and dark current can be reduced.

<Expansion of the light receiving area to the long wavelength side of the near infrared>
There are the following two factors for expanding the MQW light receiving area of type 2 (GaAsSb / InGaAs) to the long wavelength range of the near infrared.
(F1) Increase in film thickness of each layer of MQW. In the present embodiment, as shown in FIG. 2, the thickness z of GaAsSb is used as an index. The thickness of InGaAs is in the range of ± 1.0 nm of the thickness z of GaAsSb, and may be considered substantially the same.
(F2) Increase of Sb composition x to 44% or more and decrease of In composition in GaAsSb / InGaAs. The factor (F2) can be said to be the formation of a strain compensation structure from the viewpoint of InP lattice matching. However, in terms of the band structure, the energy difference between the valence band of GaAsSb and the conduction band of InGaAs is reduced. The crying point of this factor (F2) is to ensure InP lattice matching or to reduce lattice defect density.

The above factors (F1) increase in film thickness and (F2) distribution of compositions x and y will be described with reference to FIG. FIG. 3 is a diagram showing a band structure of type 2 (InGaAs / GaAsSb) and light absorption (light reception). When light having a wavelength λ on the long wavelength side is received, the electrons that have occupied the level of the valence band of GaAsSb are excited, and the electrons occupy the level of the conduction band of InGaAs. As a result of light reception, holes are generated in the valence band. In Type 2, electrons are excited from the valence band of GaAsSb to the conduction band of InGaAs, so that the energy difference is smaller than the transition from the valence band to the conduction band in a single phase, and long wavelength light Can be received.
When the wavelength λ of the light on the long wavelength side is received, the energy difference between the generated holes and electrons is h · (c / λ). Here, h is Planck's constant (6.626E-34J · s), and c is the speed of light in the medium. In order to expand the light receiving area to the longer wavelength side, it is necessary to bring both ends of the arrow of h · (c / λ) in FIG. 3 closer. As shown in FIG. 3, the factor (F2) decreases the energy difference ΔEvc between the valence band of GaAsSb and the conduction band of InGaAs in the band structure. That is, the energy difference ΔEvc is reduced by changing the band structure in which the electron level is formed. Further, the factor (F1) is considered to influence as follows. A single layer in the MQW forms one well potential. The energy level of electrons formed in the well potential tends to increase as the well potential width (film thickness) decreases. This corresponds to the fact that when energetic particles (waves) such as electrons are confined in a small space, the energy state becomes higher than when they are spread in a wide space. It may be said that it is a natural property. When the film thickness z is increased, the energy level of electrons (holes) shown in FIG. 3 approaches the valence band and the conduction band. As a result, even if the wavelength λ becomes larger and h · (c / λ) becomes smaller, light can be received.

FIG. 4 is a diagram showing a range where the Sb composition x (%) of GaAsSb and the film thickness z of GaAsSb satisfy the type 2 InGaAs / GaAsSbMQW in the light receiving element 50 of the present embodiment. The In composition y of InGaAs is distributed so as to satisfy 100 ≦ x + y ≦ 104. The meaning of each boundary line in FIG. 4 will be described below. The reason for setting the range in FIG. 4 will be described in Example 1 (see FIG. 9).
1. B line: z = −0.4x + 24.6
This B line defines the composition x and the film thickness z that the longest wavelength that can be received is 2.43 μm. In the range where z is above the B line, that is, in the xz range within the B line, the longest wavelength that can be received is 2.4 μm or more. In the present invention, in order to set the longest wavelength to 2.4 μm or more, the film thickness z is set to B line or more.
2. x = 44
The Sb composition x = 44 (%) is a line that can ensure the longest wavelength of 2.3 μm regardless of the film thickness z. When the Sb composition x is 44% or more and the film thickness z is 7 nm or more, the longest wavelength of 2.3 μm or more can be obtained. In the present embodiment, the Sb composition x is 44 (%) or more.
3. z = 3
In the present invention, the film thickness z is set to 3 nm or more in order to ensure the light receiving sensitivity.
4). A line: z = −0.625x + 45.5
The limit at which the dark current becomes large can be determined by the A line. On the outside including the A line (in a range where z is equal to or greater than the A line), the dark current increases and the S / N ratio decreases. On the inner side of this A line (range where z is located below this A line), the closer to the A line, the greater the film thickness z, the Sb composition x, and the light receiving area on the longer wavelength side. Enlarged. In the present embodiment, when low dark current is important, the xz range can be set inside the A line. As described above, unless both the film thickness z and the Sb composition x are close to the A-line, it is impossible to obtain a semiconductor element expanded to the long wavelength side in the near-infrared region where practical use is expected.
5. A2 line: z = −0.27x + 21.7
However, when emphasizing the reduction of dark current, in order to realize good crystallinity, (b1) 44 atm. % ≦ x <54.3 atm. %, Z ≦ 7 nm, and (b2) 54.3 atm. % ≦ x ≦ 68 atm. % Line A2: z ≦ −0.27x + 21.7. The A2 line will be described in detail in the embodiments.
6). z = 10
z = 10 nm defines the sensitivity limit. That is, when the film thickness z is 10 nm, the overlap of wave functions of electrons and holes is reduced, and the sensitivity is reduced. That is, the probability that electrons transition from the valence band of GaAsSb to the conduction band of InGaAs is reduced, and such electron transition (light reception) is less likely to occur. This comes from the demands of quantum mechanics. If the film thickness z is smaller than this, that is, if it is in the range of z <10, the light receiving sensitivity can be ensured.
7. z = 7
However, when importance is attached to the sensitivity, it is preferable that z ≦ 7 (nm).

<How to grow MQW>
Next, a manufacturing method will be described. An InP substrate 1 is prepared, and an InP buffer layer 2 / type 2 (InGaAs / GaAsSb) MQW light-receiving layer 3 / InGaAs diffusion concentration distribution adjusting layer 4 / InP contact layer 5 is grown thereon by an all-organic MOVPE method. To do. Here, the growth method of the light receiving layer 3 of type 2 (InGaAs / GaAsSb) MQW will be described in detail.
FIG. 5 shows a piping system and the like of the all-organic MOVPE film forming apparatus 60. A quartz tube 65 is disposed in the reaction chamber (chamber) 63, and a raw material gas is introduced into the quartz tube 65. A substrate table 66 is disposed in the quartz tube 65 so as to be rotatable and airtight. The substrate table 66 is provided with a heater 66h for heating the substrate. The temperature of the surface of the wafer 50 a during film formation is monitored by the infrared temperature monitor device 61 through a window 69 provided in the ceiling of the reaction chamber 63. This monitored temperature is a temperature at the time of growth or a temperature called a film forming temperature or a substrate temperature. In the production method of the present invention, when MQW is formed at a temperature of 400 ° C. or more and 560 ° C. or less, 400 ° C. or more and 560 ° C. or less are temperatures measured by this temperature monitor. The forced exhaust from the quartz tube 65 is performed by a vacuum pump.

The source gas is supplied by a pipe communicating with the quartz tube 65. The all organic MOVPE method is characterized in that all raw material gases are supplied in the form of an organometallic gas. In FIG. 5, source gases such as impurities are not specified, but impurities are also introduced in the form of an organometallic gas. The raw material of the organometallic gas is put in a thermostat and kept at a constant temperature. Hydrogen (H ₂ ) and nitrogen (N ₂ ) are used as the carrier gas. The organometallic gas is transported by a transport gas, and is sucked by a vacuum pump and introduced into the quartz tube 65. The amount of carrier gas is accurately adjusted by an MFC (Mass Flow Controller). Many flow controllers, solenoid valves, and the like are automatically controlled by a microcomputer.

A method for manufacturing the wafer 50a will be described. First, the n-type InP buffer layer 2 is epitaxially grown on the S-doped n-type InP substrate 1 to a film thickness of 150 nm. TeESi (tetraethylsilane) is preferably used for n-type doping. At this time, TMIn (trimethylindium) and TBP (tertiary butylphosphine) are used as the source gas. The InP buffer layer 2 may be grown using an inorganic raw material PH ₃ (phosphine). In the growth of the InP buffer layer 2, even if the growth temperature is about 600 ° C. or less than about 600 ° C., the crystallinity of the InP substrate located in the lower layer is not deteriorated by heating at about 600 ° C. However, when the InP contact layer is formed, since the MQW containing GaAsSb is formed in the lower layer, it is necessary to strictly maintain the substrate temperature within a range of, for example, a temperature of 400 ° C. or more and 560 ° C. or less. The reason is that when heated to about 600 ° C., the crystallinity of GaAsSb is greatly deteriorated due to heat damage, and when the InP contact layer is formed at a temperature lower than 400 ° C., the decomposition efficiency of the source gas is greatly increased. As a result, the impurity concentration in the InP layer increases and a high-quality InP contact layer cannot be obtained. Next, an n-type doped InGaAs layer is grown on the InP buffer layer 2 to a thickness of 0.15 μm (150 nm). This InGaAs layer is also included in the buffer layer 2 in FIG.

  Next, a type 2 MQW light-receiving layer 3 having InGaAs / GaAsSb as a quantum well pair is formed. As described above, the film thickness z of the GaAsSb in the quantum well is 3 nm or more and less than 10 nm, and the film thickness of the InGaAs 3b is preferably substantially the same as z ± 1.0 nm. In FIG. 1, the MQW light receiving layer 3 is formed by stacking 250 pairs of quantum wells. In the film formation of GaAsSb, triethylgallium (TEGa), tertiary butylarsine (TBAs), and trimethylantimony (TMSb) are used. For InGaAs, TEGa, TMIn, and TBAs can be used. These source gases are all organometallic gases, and the molecular weight of the compound is large. Therefore, it can be completely decomposed at a relatively low temperature of 400 ° C. or higher and 560 ° C. or lower and contribute to crystal growth. The MQW light-receiving layer 3 can be made abrupt in composition change at the interface of the quantum well by using all organic MOVPE. As a result, highly accurate spectrophotometry can be performed.

  As a raw material for Ga (gallium), TEGa (triethylgallium) or TMGa (trimethylgallium) may be used. The raw material for In (indium) may be TMIn (trimethylindium) or TEIn (triethylindium). As a raw material of As (arsenic), TBAs (tertiary butylarsine) or TMAs (trimethylarsenic) may be used. The raw material for Sb (antimony) may be TMSb (trimethylantimony), TESb (triethylantimony), TIPSb (triisopropylantimony), or TDMASb (tridimethylaminoantimony). By using these raw materials, a semiconductor element having a low MQW impurity concentration and excellent crystallinity can be obtained. As a result, for example, when used in a light receiving element, a light receiving element with a small dark current and a high sensitivity can be obtained. Furthermore, using the light receiving element, an optical sensor device capable of capturing a clearer image, such as an imaging device, can be obtained.

Next, the flow state of the source gas when the multiple quantum well structure 3 is formed by the all organic MOVPE method will be described. The source gas is transported through the piping, introduced into the quartz tube 65, and exhausted. Any number of source gases can be supplied to the quartz tube 65 by increasing the number of pipes. For example, even a dozen kinds of source gases are controlled by opening and closing the electromagnetic valve.
The flow rate of the source gas is controlled by a flow rate controller (MFC) shown in FIG. 5, and the flow into the quartz tube 65 is turned on and off by opening and closing the electromagnetic valve. The quartz tube 65 is forcibly exhausted by a vacuum pump. There is no stagnation in the flow of the source gas, and it is performed smoothly and automatically. Therefore, the composition is switched quickly when forming the quantum well pair.
As shown in FIG. 5, since the substrate table 66 rotates, the temperature distribution of the source gas does not have the directivity as on the inflow side or the outlet side of the source gas. Further, since the wafer 50a revolves on the substrate table 66, the flow of the source gas near the surface of the wafer 50a is in a turbulent state, and even the source gas near the surface of the wafer 50a contacts the wafer 50a. Except for the raw material gas, it has a large velocity component in the flow direction from the introduction side to the exhaust side. Therefore, most of the heat flowing from the substrate table 66 to the source gas through the wafer 50a is always exhausted together with the exhaust gas. For this reason, a large temperature gradient or temperature step is generated in the vertical direction from the wafer 50a through the surface to the source gas space.
Furthermore, in the embodiment of the present invention, the substrate temperature is heated to a low temperature range of 400 ° C. or more and 560 ° C. or less. In the case of using the all organic MOVPE method using TBAs or the like as a raw material at the substrate surface temperature in such a low temperature region, since the decomposition efficiency of the raw material is good, the growth of the multiple quantum well structure with the raw material gas flowing in the range very close to the wafer 50a The source gas that contributes to is limited to those efficiently decomposed into the form necessary for growth.

FIG. 6A is a diagram showing the flow of organometallic molecules and the flow of temperature, and FIG. 6B is a schematic diagram of organometallic molecules on the substrate surface. These diagrams are used to explain the importance of setting the surface temperature in order to obtain a steep composition change at the heterointerface of the multiple quantum well structure.
The surface of the wafer 50a is set to a monitored temperature. However, when the material gas space is slightly entered from the wafer surface, the temperature suddenly decreases or a large temperature step is generated as described above. Therefore, in the case of a raw material gas having a decomposition temperature of T1 ° C., the substrate surface temperature is set to (T1 + α), and α is determined in consideration of variations in temperature distribution and the like. In a situation where there is a sudden large temperature drop or temperature step from the surface of the wafer 50a to the source gas space, when large-sized organometallic molecules flow through the wafer surface as shown in FIG. It is considered that the compound molecules that contribute to the growth are limited to those in the range in contact with the surface and the film thickness range of several organometallic molecules from the surface. Therefore, as shown in FIG. 6B, the organic metal molecules in the range in contact with the wafer surface and the molecules located within the film thickness range of several organometallic molecules from the wafer surface are mainly crystal growth. It is considered that the organometallic molecules outside it are discharged out of the quartz tube 65 with almost no decomposition. When the organometallic molecules near the surface of the wafer 50a are decomposed and crystal growth occurs, the organometallic molecules located outside enter the replenishment.
In other words, the range of the organometallic molecules that can participate in crystal growth is limited to a thin source gas layer on the surface of the wafer 50a by making the wafer surface temperature slightly higher than the temperature at which the organometallic molecules decompose. it can.

From the above, when the source gas suitable for the chemical composition of the pair is switched by the electromagnetic valve and forcedly evacuated by the vacuum pump, after the crystal of the previous chemical composition is grown with slight inertia, Thus, it is possible to grow a crystal having a switched chemical composition without being affected by the source gas. As a result, the composition change at the hetero interface can be made steep. This means that the previous source gas does not substantially remain in the quartz tube 65, and the source gas that contributes to the growth of the multiple quantum well structure with the source gas flowing in a range very close to the wafer 50a is: This is because it is limited to those efficiently decomposed into a shape necessary for growth (film formation factor 1). That is, as can be seen from FIG. 5, when one layer of the quantum well is formed, the electromagnetic valve is opened and closed while forcibly evacuating with a vacuum pump, and when the source gas for forming the other layer is introduced, a little Although there are organometallic molecules that participate in crystal growth with inertia, the molecules in one layer that supplements are almost exhausted and gone. The closer the wafer surface temperature is to the decomposition temperature of the organometallic molecule, the smaller the range of organometallic molecules participating in crystal growth (range from the wafer surface).
When forming this multi-quantum well structure, phase growth occurs in the GaAsSb layer of the multi-quantum well structure when grown in a temperature range of about 600 ° C., and the crystal growth surface of the multi-quantum well structure that is clean and excellent in flatness, and A multiple quantum well structure having excellent periodicity and crystallinity cannot be obtained. For this reason, the growth temperature is set to a temperature range of 400 ° C. or more and 560 ° C. or less (deposition factor 2). This film formation method is an all-organic MOVPE method, and all the source gases are organometallic gases with high decomposition efficiency. (Film formation factor 3) strongly depends on film formation factor 1.

<Method for Manufacturing Semiconductor Device>
In the semiconductor element 50 shown in FIG. 1, the InGaAs diffusion concentration distribution adjustment layer 4 is located on the type 2 MQW light-receiving layer 3, and the InP contact layer 5 is located on the InGaAs diffusion concentration distribution adjustment layer 4. ing. The p-type region 6 is provided by selectively diffusing Zn of the p-type impurity from the opening of the selective diffusion mask pattern 36 provided on the surface of the InP contact layer 5. A pn junction 15 or a pi junction 15 is formed at the tip of the p-type region 6. A depletion layer is formed by applying a reverse bias voltage to the pn junction 15 or the pi junction 15 to capture charges due to photoelectron conversion, and to make the brightness of the pixel correspond to the amount of charges. The p-type region 6 or the pn junction 15 or the pi junction 15 is a main part constituting the pixel. The p-side electrode 11 that is in ohmic contact with the p-type region 6 is a pixel electrode, and reads the above charges for each pixel with the n-side electrode 12 that is set to the ground potential. The selective diffusion mask pattern 36 is left as it is on the surface of the InP contact layer around the p-type region 6. Further, a protective film such as SiON not shown is coated. The selective diffusion mask pattern 36 is left as it is. When the p-type region 6 is formed and then exposed to the atmosphere except for this, a surface level is formed at the boundary with the p-type region on the contact layer surface, and the dark current is formed. This is because of the increase.
One point is to continue the growth in the same film forming chamber or quartz tube 65 by the all organic MOVPE method until the formation of the InP contact layer 5 after the MQW is formed as described above. That is, before the InP contact layer 5 is formed, the wafer 50a is not taken out from the film forming chamber and the InP contact layer 5 is not formed by another film forming method. One point. That is, since the InGaAs diffusion concentration distribution adjusting layer 4 and the InP contact layer 5 are continuously formed in the quartz tube 65, the interfaces 16 and 17 are not regrowth interfaces. For this reason, the concentrations of oxygen, carbon, and silicon are all below a predetermined level, and charge leakage does not occur particularly at the intersection line between the p-type region 6 and the interface 17.

In the present embodiment, a non-doped InGaAs diffusion concentration distribution layer 4 having a film thickness of 1.0 μm, for example, is formed on the MQW light-receiving layer 3. After the InP contact layer 5 is formed, the InGaAs diffusion concentration distribution layer 4 has a high concentration of Zn when the p-type impurity Zn is introduced from the InP contact layer 5 to reach the MQW light receiving layer 3 by a selective diffusion method. When it enters MQW, the crystallinity is damaged, so it is provided for the adjustment. The InGaAs diffusion concentration distribution adjustment layer 4 may be arranged as described above, but may not be provided.
The p-type region 6 is formed by the selective diffusion described above, and the pn junction 15 or the pi junction 15 is formed at the tip thereof. Even when the InGaAs diffusion concentration distribution adjusting layer 4 is inserted, since the InGaAs has a small band gap, the electric resistance of the light receiving element can be lowered even if it is non-doped. By reducing the electrical resistance, it is possible to improve the responsiveness and obtain a moving image with good image quality.
On the InGaAs diffusion concentration distribution adjusting layer 4, the undoped InP contact layer 5 is epitaxially grown to a thickness of, for example, 0.8 μm by the all organic MOVPE method continuously with the wafer 50a placed in the same quartz tube 65. Is good. As described above, trimethylindium (TMIn) and tertiary butylphosphine (TBP) are used for the source gas. By using this source gas, the growth temperature of the InP contact layer 5 can be made 400 ° C. or more and 560 ° C. or less, and further 535 ° C. or less. As a result, the MQW GaAsSb positioned under the InP contact layer 5 is not damaged by heat, and the MQW crystallinity is not impaired. When the InP contact layer 5 is formed, since the MQW containing GaAsSb is formed in the lower layer, the substrate temperature needs to be strictly maintained within a range of, for example, a temperature of 400 ° C. or more and 560 ° C. or less. The reason is that when heated to about 600 ° C., the crystallinity of GaAsSb is greatly deteriorated due to heat damage, and when the InP contact layer is formed at a temperature lower than 400 ° C., the decomposition efficiency of the source gas is greatly increased. Therefore, the impurity concentration in the InP contact layer 5 increases, and a high quality InP contact layer 5 cannot be obtained.

As described above, conventionally, it has been necessary to form the MQW by the MBE method. However, in order to grow an InP contact layer by the MBE method, it is necessary to use a solid raw material as a phosphorus raw material, which is problematic in terms of safety. There was also room for improvement in terms of manufacturing efficiency.
Prior to the present invention, the interface between the InGaAs diffusion concentration distribution adjusting layer and the InP contact layer was a regrowth interface once exposed to the atmosphere. The regrowth interface is specified by satisfying at least one of oxygen concentration of 1E17 cm ⁻³ or more, carbon concentration of 1E17 cm ⁻³ or more, and silicon concentration of 1E17 cm ⁻³ or more by secondary ion mass spectrometry. Can do. The regrowth interface forms a crossing line with the p-type region, and a charge leak occurs at the crossing line, thereby significantly degrading the image quality.
Further, for example, when an InP contact layer is grown by a simple MOVPE method, phosphine (PH ₃ ) is used as a raw material of phosphorus. Therefore, the decomposition temperature is high, and the occurrence of damage due to heat of GaAsSb located in the lower layer is induced to cause the MQW crystal. It will be harmful to sex.

  FIG. 7 is a flowchart of a method for manufacturing the light receiving element 50 of FIG. According to this manufacturing method, only the organometallic gas is used as the source gas (deposition factor 3) until the growth temperature is lowered (deposition factor 2) and the formation of the InP contact layer 5 is completed. Therefore, it is important to have no recrystallization interface (deposition factor 4). As a result, it is possible to efficiently manufacture a large number of photodiodes having a light receiving sensitivity in a wavelength region of 2 μm to 5 μm, which has less charge leakage and excellent crystallinity.

(Embodiment 2)
FIG. 8 shows an optical sensor device 10 including a light receiving element array (semiconductor element) 50 according to the second embodiment of the present invention. Optical parts such as lenses are omitted. Although a protective film 43 made of a SiON film is shown in FIG. 8, it is actually also arranged in FIG. The light receiving element array 50 has the same stacked structure as that of the light receiving elements shown in FIG. 1, and is different in that a plurality of light receiving elements or pixels P are arranged. The film thickness z, the Sb composition s, etc. are the same as those in the semiconductor element of FIG. Also, the interfaces 16 and 17 are not regrowth interfaces, and the concentration of impurities such as oxygen, carbon, and silicon are all low, which is the same as the light receiving element (semiconductor element) in FIG.
In FIG. 8, the light receiving element array 50 and a CMOS 70 constituting a read circuit (Read-Out IC) are connected. The readout electrode (not shown) of the CMOS 70 and the pixel electrode (p-side electrode) 11 of the light receiving element array 50 are bonded with the bonding bump 39 interposed. Further, a common ground electrode (n-side electrode) 12 for each pixel of the light receiving element array 50 and a ground electrode (not shown) of the CMOS 70 are joined via bumps 12b. An imaging device or the like can be obtained by combining the CMOS 70 and the light receiving element array 50 and integrating light reception information for each pixel.
As described above, the light-receiving element array (semiconductor element) 50 of the present invention has sensitivity up to a long wavelength region and has a small dark current (leakage current), and therefore, inspection of living organisms such as animals and plants, environmental monitoring, etc. By using this, high-precision inspection can be performed.

Example 1
The light-receiving element 50 having the same structure as that of the semiconductor element 50 shown in FIG. Was measured. Table 1 shows the Sb composition x of GaAsSb, the In composition y of InGaAs, and the film thickness z (nm) of GaAsSb and InGaAs of each specimen. The film thicknesses of GaAsSb and InGaAs were the same. The composition x and the composition y have a lattice mismatch degree of InGaAs of Δω ₁ and a lattice mismatch degree of GaAsSb of Δω _2, and the lattice mismatch degree of the entire multiple quantum well structure Δω = {Σ (Δω ₁ × ΔGaAs thickness + Δω ₂ × GaAsSb layer thickness)} / {Σ (InGaAs layer thickness + GaAsSb layer thickness)} Δω defined by −0.2% or more and 0.2% or less did.
The light receiving element 50 was manufactured by the all organic MOVPE method as described above. Parts other than the light receiving layer were manufactured by the above method. The manufacturing method of the main epitaxial layer was performed as follows. The growth temperature is 510 ° C. GaAsSb uses TEGa, TBAs and TESb as raw materials, and the Sb composition x has a constant V / III ratio (supply amount of Group V raw materials / supply amount of Group III raw materials), and the supply amount ratio of TBAs and TESb It changed by changing. InGaAs was grown using TEGa, TMIn, and TBAs as raw materials, and the In composition y was grown by changing the supply ratio of TEGa and TMnIn while keeping the V / III ratio constant. In the InP contact layer 5, TMIn and TBP were used as raw materials.

<Evaluation>
Evaluation was performed by measuring the PL wavelength at room temperature and dark current as described below. The dark current was 100 μm in diameter and the reverse bias voltage (Vr) = 1V.
1. PL wavelength at room temperature The wavelength of PL at room temperature corresponds to the bandgap wavelength and corresponds to the longest wavelength that can be received. 9 and 4, in Comparative Examples 2, 3, 8, and 10, the film thickness z is located below the B line: −0.4x + 24.6, The PL wavelengths are 2210 nm, 2290 nm, 2210 nm, and 2300 nm, respectively, and are less than 2400 nm. This is because the film thickness z and the Sb composition x of GaAsSb are not sufficiently high.
On the other hand, in the range where the film thickness z ≧ 3 nm and the film thickness z ≧ −0.4x + 24.6 (B line or more) are satisfied, as shown in Table 1, all the examples of the present invention are PL. The wavelength is 2.4 μm or more. Moreover, the PL wavelength becomes longer as it approaches A line: z = −0.625x + 45.5. In FIG. 9, the arrow and the PL wavelength indicate that the PL wavelength becomes longer as it approaches A line and z = 10.
Further, when approaching the line of film thickness z = 10, PL as in Invention Example 7 (2430 nm) → Invention Example 6 (2620 nm) → Invention Example 5 (2700 nm) → Invention Example 4 (2870 nm) The wavelength is also longer.
2. Dark current The dark current is limited by the A line: z = −0.625x + 45.5. Since the dark current increases as the lattice defect density increases, it tends to be considered that only the Sb composition x and the In composition y. However, it depends on the film thickness as described above, and is influenced not only by the composition x and y but also by the film thickness z. It is thought that lattice defects are accumulated.
Inventive Example 16 and Inventive Example 19 are located on the A line, and the dark currents are 700 nA and 800 nA, respectively. If the dark current is at this level, there are many allowable applications. However, when the dark current is important, the film thickness z is preferably in a range smaller than the A line. In order to further reduce the dark current, it is preferable to set the A2 line: z = −0.27x + 21.7 or less. In this case, Invention Example 16 and Invention Example 19 do not fall within the scope of the present invention and are comparative examples.
The ranges slightly away from the A line, for example, Invention Examples 11, 14, 17, and 18 are 200 nA, 200 nA, 300 nA, and 200 nA, respectively, which are practically satisfactory levels.
3. Photosensitive sensitivity As will be described later in Example 2, when the thickness z of GaAsSb reaches 10 nm, the received light sensitivity decreases. Even if the film thickness is z = 10 nm, there is a certain level of light sensitivity, which is possible depending on the application. However, when the light receiving sensitivity is very problematic, the film thickness z should be smaller than 10 nm. If higher light receiving sensitivity is required, z ≦ 7 (nm) is preferable. In this case, Invention Example 4 is outside the scope of the present invention.
This is a quantum mechanical effect and is considered to hold regardless of the conditions of the examples. Also, when the film thickness z is as thin as 2 nm, the light receiving sensitivity is low. For this reason, the film thickness z is 3 nm or more.

(Example 2)
<Manufacture of specimen>
The epitaxial layer of the buffer layer (InP / InGaAs) / type 2 (InGaAs / GaAsSb) MQW light receiving layer / InGaAs diffusion concentration distribution adjusting layer / InP contact layer is formed on the S-doped InP substrate by the all organic MOVPE method. Formed consistently. An InP contact layer was grown directly on the MQW light-receiving layer 3. TEGa, TMIn, TBAs, TBP, and TMSb were used as raw materials for Ga, In, As, P, and Sb, respectively. TeESi was used for n-type impurity doping.
Specifically, an n-type doped InP buffer layer was grown to 150 nm on an S-doped InP substrate, and an n-type doped InGaAs buffer layer was grown to 0.15 μm thereon. An InGaAs / GaAsSb type 2 MQW light-receiving layer was grown on the two buffer layers. The MQW configuration was repeated 250 pairs with the lower undoped InGaAs layer and the undoped GaAsSb layer as a pair. The composition is a composition that lattice-matches alone, and the Sb composition x of GaAsSb is 49% and the In composition y of InGaAs is 53%. When the lattice mismatch degree of InGaAs is Δω ₁ and the lattice mismatch degree of GaAsSb is Δω ₂ , the lattice mismatch degree of the entire multiple quantum well structure Δω = {Σ (Δω ₁ × InGaAs layer thickness + Δω ₂ × GaAsSb Δω defined by (layer thickness)} / {Σ (InGaAs layer thickness + GaAsSb layer thickness)} is −0.2% or more and 0.2% or less. An InGaAs diffusion concentration distribution adjusting layer was grown by 1.0 μm on the MQW light-receiving layer, and an undoped InP contact layer was grown by 0.8 μm thereon. For the growth of GaAsSb, TEGa, TBAs, and TMSb were used as raw materials. Further, TEBa, TMIn, and TBAs were used for the growth of InGaAs. Further, TMIn and TBP were used for the growth of the InP contact layer or InP buffer layer.

<Evaluation>
1. The PL characteristics are shown in Table 2 and FIG. When the film thickness z of GaAsSb and the film thickness of InGaAs were 5 nm, the PL peak wavelength was 2.5 μm. As both film thicknesses were decreased, the PL wavelength shifted to the short wavelength side. When the film thickness of GaAsSb and InGaAs was 2 nm, the PL wavelength was 1.9 μm.
On the other hand, as the thickness of InGaAs and GaAsSb was increased, the PL wavelength became longer. When both thicknesses were 10 nm, the PL wavelength was 2.9 μm.
2. Light receiving sensitivity The light receiving sensitivity was measured with respect to light having a wavelength of 2000 nm at a reverse bias voltage Vr = −5V. When the film thicknesses of both GaAsSb and InGaAs were 2 nm, the light receiving sensitivity was as low as 0.1 A / W. As both film thicknesses were increased to 3 nm, 5 nm, and 7 m, they improved to 0.6 A / W, 0.6 A / W, and 0.5 A / W. And when both film thicknesses were 10 nm, it deteriorated with 0.2 A / W.

(Example 3)
The test body was configured as shown in Example 2 and followed the same procedure. However, in this example, the film thickness was fixed at 5 nm for both GaAsSb and InGaAs, and the supply amount of the source gas was adjusted as described above. At this time, when the lattice mismatch degree of InGaAs is Δω ₁ and the lattice mismatch degree of GaAsSb is Δω ₂ , the lattice mismatch degree of the entire multiple quantum well structure Δω = {Σ (Δω ₁ × InGaAs layer thickness + Δω ₂ × GaAsSb layer thickness)} / {Σ (InGaAs layer thickness + GaAsSb layer thickness)} Δω defined by −0.2% to 0.2%.

<Evaluation>
1. The PL characteristics are shown in Table 3 and FIG. When the Sb composition x of GaAsSb was 49% and the In composition y of InGaAs was 53%, the PL wavelength was 2.4 μm. When the Sb composition x of GaAsSb is increased and the InGaAs composition y is decreased, the PL wavelength is shifted to the longer wavelength side. The Sb composition was 62%, the In composition was 38%, and the PL wavelength was 3.0 μm.
2. Dark current Whether the reverse bias voltage Vr is -1V or -5V, when the Sb composition is increased to 44% or more, the dark current gradually increases, but it can be said to be a good level. When the In composition y was lower than 53% and the Sb composition was increased from 62%, the difference in dark current between the reverse bias voltage Vr = −1V and −5V became large. Thus, in order to increase the S / N ratio in a light receiving element having light receiving sensitivity up to a longer wavelength side, it is preferable to set a lower reverse bias voltage (decrease the absolute value).

  Although the embodiments of the present invention have been described above, the embodiments of the present invention disclosed above are merely examples, and the scope of the present invention is not limited to these embodiments. The scope of the present invention is indicated by the description of the scope of claims, and further includes meanings equivalent to the description of the scope of claims and all modifications within the scope.

  According to the semiconductor device of the present invention, the Sb composition is adjusted while adjusting the composition so that the InP lattice matching condition is integrated in both GaAsSb and InGaAs while suppressing the dark current to a level at which there is no problem in practice. By increasing (decreasing the In composition) and increasing the thickness of the quantum well, the light receiving sensitivity can be expanded to the long wavelength side in the near infrared region. As a result, an important application can be covered with a simple device. Further, by consistently growing in the same growth tank by the all organic OMVPE method, impurity contamination can be prevented and high quality crystallinity can be obtained. In addition, the steepness of the composition between the MQW quantum wells of the light receiving layer can be increased, and the analysis of the absorption spectrum and the like can be performed with high accuracy.

1 InP substrate, 2 buffer layer (InP and / or InGaAs), 3 type II MQW light receiving layer, 4 InGaAs layer (diffusion concentration distribution adjusting layer), 5 InP contact layer, 6 p-type region, 10 optical sensor device (detection device) 11 p-side electrode (pixel electrode), 12 ground electrode (n-side electrode), 12b bump, 15 pn junction, 16 interface between MQW and InGaAs layer, 17 interface between InGaAs layer and InP window layer, 20 infrared temperature monitor Equipment, 21 reaction chamber window, 30 reaction chamber, 35 AR (antireflection) film, 36 selective diffusion mask pattern, 39 bonding bump, 43 protective film (SiON film), 50 light receiving element (light receiving element array), 50a wafer ( Intermediate product), 60 all organic MOVPE film forming device, 61 infrared temperature monitoring device, 63 reaction chamber, 65 quartz tube, 69 Reaction chamber window, 66 substrate table, 66h heater, 70 CMOS, P pixel.

  The present invention relates to a group III-V semiconductor device and a method for manufacturing the same, and more specifically, a type 2 multiple quantum well structure having a configuration that has light receiving sensitivity up to a long wavelength region in the near infrared ( The present invention relates to a semiconductor element, an optical sensor device, and a method for manufacturing a semiconductor element, including a multi-quantum well) in a light receiving layer.

A photodiode capable of obtaining a cutoff wavelength of 2 μm or more by forming an InGaAs / GaAsSb type II multiple quantum well structure on an InP substrate of a III-V compound semiconductor is disclosed (Non-patent Document). 1).
Further, an LED and a laser diode having an emission wavelength of 2.14 μm formed by forming InGaAs-GaAsSb type 2 MQW on an InP substrate as an active layer have been disclosed (Non-patent Document 2). This type 2 MQW GaAsSb has an Sb composition of 34 atm. % To 40 atm. %, And a strain compensation structure on the side lower than the lattice matching composition of InP. In the following description, “atm.%” Is simply referred to as “%”.
Further, without forming the type 2 MQW, a GaAsSb single phase with a changed Sb composition is grown on an InP substrate and the wavelength of PL is measured. Then, a light-emitting element using InP / GaAsSb type 2 MQW is considered.
Furthermore, a semiconductor laser element having a GaInNAsSb quantum well structure has been disclosed (Patent Document 1). This GaInNAsSb quantum well structure is a single quantum well structure (that is, the number of pairs = 1).

JP 2005-197395 A

R. Sidhu, et.al. "ALong-Wavelength Photodiode on InP Using Lattice-Matched GaInAs-GaAsSb Type-II Quantum Wells, IEEE Photonics Technology Letters, Vol.17, No.12 (2005), pp.2715-2717 M.Peter, et.al. “Light-emitting diodes and laser diodes based on a Ga1-xInxAs / GaAs1-ySbytype II superlattice on InP substrate” Appl. Phys. Lett., Vol.74, No.14 (5 April 1999 ), pp.1951-1953 M.Peter, et.al. “Band gaps and band offsets in strained GaAs1-ySby on InP grown by metalorganic chemical vapor deposition” Appl. Phys. Lett., Vol. 74, No. 3 (18 January 1999), pp. 410 -412

Since the field of application of the photodiode using the semiconductor element described above is widened, it is desired to increase the light receiving sensitivity to the long wavelength side as much as possible. For this reason, it can be considered that type 2 InGaAs / GaAsSbMQW has a strain compensation structure. In the MQW having a strain compensation structure using InGaAs / GaAsSb, compressive stress is generated in one of the paired InGaAs / GaAsSb, and tensile stress is generated in the other to prevent distortion in the pair. In order to increase the wavelength of light reception sensitivity, it is desirable to generate tensile stress in InGaAs and compressive stress in GaAsSb. In Non-Patent Document 2, the Sb composition of GaAsSb is lowered and pulled by InGaAs. It is assumed that the MQW having such a strain compensation structure first deteriorates in crystallinity. For this reason, there is no example used for the light reception use to which the deterioration of crystallinity is directly linked to the increase in dark current. That is, there is no example used for the light receiving element, only the laser diode in Non-Patent Document 2 described above.
When forming an MQW having a repetition number of 100 to 300, in the MBE (molecular beam epitaxy) method, the molecular beam can be instantaneously switched with a shutter. For this reason, automatic switching of microcomputer-controlled valves is possible, and it has been considered that deposition by the MBE method is almost inevitable for the growth of a steep interface and a high-quality multiple quantum well structure. In particular, when considering only the above-mentioned problem of crystal growth of a GaAsSb layer that easily undergoes phase separation, a crystal growth method with strong non-equilibrium is necessary for epitaxial growth while preventing phase separation. For this reason, the MBE method which is a crystal growth method with strong non-equilibrium is suitable. Actually, the MBE method is used to form a GaAsSb layer (Non-patent Document 1).
All of the above semiconductor devices target near-infrared long-wavelength light. In this case, the cap layer or contact layer located on the surface side of type 2 MQW absorbs light in the target wavelength region. Preferably, it is made of a material that does not. For this reason, InP is often used for the contact layer. When InP is grown by the MBE method, since solid phosphorus is used as a raw material, phosphorus (P) adheres to the inner wall of the growth tank. The adhered phosphorus may ignite when the growth tank is opened during maintenance and is in contact with air. Therefore, a manufacturing method is used in which MQW is grown by the MBE method and the InP contact layer is grown by a growth method other than the MBE method. When switching from the MBE method to the OMVPE method, for example, the intermediate product InP wafer is once exposed to the air. Such exposure to air is susceptible to contamination by impurities. Further, the switching of the growth method as described above inhibits the production efficiency.

  An object of the present invention is to provide a semiconductor element, an optical sensor device, and a method for manufacturing a semiconductor element, in which dark current is low and the light receiving sensitivity is increased on the long wavelength side of the near infrared.

The semiconductor element of the present invention is formed on an InP substrate. The semiconductor device is provided with a light receiving layer of multi-quantum well structure of the type 2 located on the InP substrate, the multiple quantum _{wells, GaAs 1-x} Sb x (hereinafter, GaAsSb) and _layer, In _{y Ga 1-y} As (hereinafter referred to as InGaAs) layer, the range of the Sb composition x (atm.%) Of the GaAsSb layer and the film thickness z of the GaAsSb layer is as follows: It meets all,
(A) The film thickness z is less than 10 nm (44 atm.% <Composition x <54 atm.%) ,
(B) Composition x is 44 atm.% Or more (7 nm ≦ film thickness z <10 nm) ,
(C) Composition x is 54 atm. % (3 nm <film thickness z <10 nm) ,
(D) The film thickness z is a line segment z = −0.4x + 24.6 (B) connecting the point (composition x = 44, film thickness z = 7) and the point (composition x = 54, film thickness z = 3) Line) larger range ,
In addition, a distortion compensation structure is formed, and the longest wavelength at which the light receiving layer has light receiving sensitivity is 2.4 μm or more.
Here, the GaAsSb layer and the InGaAs layer in MQW have substantially the same film thickness within a range of variation of ± 1.0 nm. For this reason, in the following description, when the film thickness z of one layer (for example, GaAsSb layer) is referred to, it is appropriate to interpret it as the film thickness z of the one layer or the other layer (for example, InGaAs layer).

In the above invention, in order to expand the light receiving region to the long wavelength side, the multiple quantum well is composed of a repetitive structure of a GaAsSb layer and an InGaAs layer, and the Sb composition x (atm.%) Of the GaAsSb layer and the The film thickness z of the GaAsSb layer satisfies all of the above (a) to (d) and forms a strain compensation structure. Accordingly, the longest wavelength at which the light receiving layer has light receiving sensitivity is set to 2.4 μm or more.
The above invention applies the strain compensation structure (factor (F2)) in addition to the increase in the thickness of the quantum well (factor (F1)) in order to expand the light receiving region to the longer wavelength side. According to the above configuration, when the Sb composition x of the GaAsSb layer forming the strain compensation structure (factor (F2)) is increased and the In composition y of the InGaAs layer is decreased, lattice matching with InP (hereinafter, “ The energy difference between the valence band of the GaAsSb layer and the conduction band of the InGaAs layer can be reduced while maintaining "InP lattice matching". As a result, it is possible to expand the light receiving range to the long wave side while maintaining a low dark current without increasing the lattice defect density.

Further, (e) the film thickness z is a line segment z = −0.4x + 25.6 connecting the point (composition x = 44, film thickness z = 8) and the point (composition x = 54, film thickness z = 4). A range larger than (a line obtained by translating the B line by +1 in the z direction) can be satisfied.
  As described above, there is a variation of ± 1.0 nm between the GaAsSb layer and the InGaAs layer in MQW. Even if there is such a variation, the lower limit of the film thickness is increased by 1.0 nm while narrowing the entire range in order to reliably obtain an increase in the quantum well film thickness (factor (F1)).

The Sb composition x of the GaAsSb layer and the In composition y (atm.%) Of the InGaAs layer can be distributed so as to satisfy 100 ≦ x + y ≦ 104.
The relation of 100 ≦ x + y ≦ 104 between the Sb composition x of the GaAsSb layer and the In composition y of the InGaAs layer is that the lattice defect density is reduced while maintaining the InP lattice matching while increasing the Sb composition x from 44%. Is necessary to do.

  An InP contact layer may be provided on the MQW. As a result, the contact layer can be formed of a crystal layer that is highly transmissive to light on the long wavelength side in the near infrared region. The InP layer has many achievements as a contact layer and is excellent in surface smoothness. Therefore, the hygroscopic property is small and the durability is excellent.

The film thickness z (nm) of the GaAsSb layer in MQW is 44 atm. % ≦ x <54 atm. % (Ii) z <10 nm, and (d) a line segment connecting a point (composition x = 44, film thickness z = 7) and a point (composition x = 54, film thickness z = 3) z = A range larger than −0.4x + 24.6 (line B) .
As described above, (F1) the increase in the thickness of each MQW layer, and (F2) the increase in the Sb composition in the GaAsSb layer and the decrease in the In composition in the InGaAs layer (strain compensation structure) are effective in expanding the light receiving wavelength range. is there. In particular, regarding the increase in the film thickness of each MQW layer of (F1), when the film thickness z of the GaAsSb layer is increased, the following should be performed in relation to the dark current or the light receiving sensitivity.
44 atm. % ≦ x <54 atm. In the range of%, it is not so difficult to maintain InP lattice matching between the GaAsSb layer and the InGaAs layer. For this reason, even if the film thickness is increased and the deviation from strict lattice matching is enlarged, the deterioration of crystallinity is small. For this reason, by increasing the film thickness, it is possible to promote a longer wavelength while maintaining a low dark current. However, when the film thickness is increased, the overlap of wave functions between the holes confined in the GaAsSb layer and the electrons confined in the conduction band of the InGaAs layer is reduced. As a result, the quantum efficiency may decrease, and the light receiving sensitivity may deteriorate. When the film thickness z of the GaAsSb layer is 10 nm or more, the light receiving sensitivity is greatly reduced. In order to ensure the light receiving sensitivity, the film thickness z is set to less than 10 nm.
As a reference only, 56.8 atm. % ≦ x ≦ 68 atm. In the range of%, the Sb composition x deviates to the higher level from the range where InP lattice matching is obtained with the GaAsSb layer alone. By obtaining a decrease in the In composition y of the InGaAs layer, InP lattice matching can be obtained in MQW. For this reason, roughly, the characteristics of the crystal film are not the best, and the InP lattice matching is barely maintained. When the thickness z of the GaAsSb layer is increased, the deviation from the InP lattice matching of the GaAsSb layer in the MQW becomes clear (the same applies to the InGaAs layer whose thickness is proportionally increased), and the lattice defect density is It increases (deteriorates) as the film thickness z increases. As a result, the dark current increases as the film thickness z increases. Therefore, in such a high Sb composition x range, the dark current depends on the film thickness z as well as the Sb composition x of the GaAsSb layer. The range of the Sb composition x and the film thickness z that brings the dark current within a practically acceptable level is z <−0.625x + 45.5.

The film thickness z (nm) of the GaAsSb layer in MQW is 44 atm. % ≦ x <54 atm. % (B) less than 10 nm and (d) larger than the B line. More preferably, (e) a line obtained by translating the B line by +1 in the z-axis direction is larger . As a result, it is possible to pursue improvement in sensitivity on the long wavelength side in the near infrared region while placing importance on good crystallinity and on reducing dark current.

When the lattice mismatch degree of the InGaAs layer in MQW is Δω ₁ and the lattice mismatch degree of the GaAsSb layer is Δω ₂ , the lattice mismatch degree of the entire multiple quantum well structure Δω = {Σ (Δω ₁ × InGaAs layer thickness It is preferable that Δω defined by + Δω ₂ × GaAsSb layer thickness} / {Σ (InGaAs layer thickness + GaAsSb layer thickness)} satisfy −0.2% or more and 0.2% or less.
Here, “the lattice constant of the substrate is a and the lattice constant of the InGaAs layer is a ₁ , the degree of lattice mismatch Δω ₁ of the GaAsSb layer is Δω ₁ = {(a−a ₁ ) / a} × 100%” Is defined. For even [Delta] [omega _2, is the same. In the following description, the degree of lattice mismatch is the same.
With the above configuration, the GaAsSb layer alone or the InGaAs layer alone cannot satisfy the InP lattice matching, but both can deviate from the lattice constant in the opposite direction to obtain the InP lattice matching when integrated. Can do. As a result, the lattice defect density can be lowered and the low dark current can be maintained while changing the band structure so as to be advantageous for expansion to the longer wavelength side.

  A GaAsSb layer is used. This makes it easy to obtain lattice matching with the InP substrate and to obtain a crystal layer having a relatively high level of conduction band and valence band, which is a favorable condition for constituting one layer of type 2 MQW. Can do.

  An InGaAs layer is used. This makes it easy to obtain lattice matching with the InP substrate and to obtain a crystal layer having a conduction band and a valence band at a relatively low level, which is a favorable condition for forming the other layer of the type 2 MQW. Can do.

  The longest wavelength with which the light receiving layer has light receiving sensitivity is set to 2.4 μm or more. By satisfying the above xz relationship, the energy difference between the valence band of GaAsSb and the conduction band of InGaAs is reduced to a wavelength equivalent to 2.4 μm or less. As a result, a substance having an absorption band of 2.4 μm or more can be measured, and can be used for a wider range of material inspection apparatuses.

It is preferable not to have a regrowth interface between the bottom surface of the light receiving layer and the top surface of the semiconductor layer including the light receiving layer and the InP contact layer.
Here, the regrowth interface means that after the first crystal layer is grown by a predetermined growth method, the second crystal layer is exposed to the first crystal layer and exposed to the first crystal layer by another growth method. Refers to the interface between the first crystal layer and the second crystal layer. Usually, oxygen and carbon are mixed in as a high concentration as impurities. The semiconductor element of the present invention should preferably have no regrowth interface and can maintain good crystallinity up to the surface of the InP contact layer. This can contribute to the reduction of dark current.
Moreover, a semiconductor element can be manufactured efficiently. That is, as will be described later, since the growth is consistently performed by the all organic MOVPE method from the (buffer layer to MQW) to the InP contact layer containing P, the manufacturing is continuously performed in the same growth tank. Can do. For example, even if an InP contact layer containing phosphorus is formed, solid phosphorus is not used as a raw material, so that phosphorus does not adhere to the inner wall of the growth tank. For this reason, there is no fear of ignition at the time of maintenance, and it is excellent in safety.

  The optical sensor device of the present invention is characterized in that any one of the above semiconductor elements is used as a light receiving element. Thus, an optical sensor device having a low dark current and sensitivity on the long wavelength side in the near infrared region can be obtained. This optical sensor device can include a CMOS having a readout electrode from each pixel of a semiconductor element (light receiving element), a spectroscope (diffraction grating), an optical element such as a lens, a control device such as a microcomputer, and the like.

The method for manufacturing a semiconductor device of the present invention manufactures a semiconductor device formed on an InP substrate. This manufacturing method includes a step of forming a type 2 multiple quantum well structure light-receiving layer on an InP substrate. The multiple quantum well includes a GaAs _1-x Sb _x (hereinafter referred to as GaAsSb) layer and an In _y Ga ₁ layer. _-Y As (hereinafter referred to as InGaAs) layer. In the step of forming a multiple quantum well structure, the Sb composition x (atm.%) Of the GaAsSb layer and the film thickness z of the GaAsSb layer are expressed as follows: It shall satisfy all of (i) to (d)
(A) The film thickness z is less than 10 nm (44 atm.% <Composition x <54 atm.%) ,
(B) Composition x is 44 atm.% Or more (7 nm ≦ film thickness z <10 nm) ,
(C) Composition x is 54 atm. % (3 nm <film thickness z <10 nm) ,
(D) The film thickness z is a line segment z = −0.4x + 24.6 (B) connecting the point (composition x = 44, film thickness z = 7) and the point (composition x = 54, film thickness z = 3) Line) larger range ,
In addition, a distortion compensation structure is formed, and the longest wavelength at which the light receiving layer has light receiving sensitivity is 2.4 μm or more.
Further, for the Sb composition x of the GaAsSb layer and the In composition y of the InGaAs layer, 1 atm. The In composition y can be grown at a rate of 0.9 to 1.2 times per% increase.
Further, a step of forming an InP contact layer on the MQW may be provided.

The film thickness z (nm) of the GaAsSb layer in MQW is 44 atm. % ≦ x <54 atm. % (Ii) z <10 nm, and (d) a line segment connecting a point (composition x = 44, film thickness z = 7) and a point (composition x = 54, film thickness z = 3) z = A range larger than −0.4x + 24.6 (line B) .
The point that the dark current of the semiconductor element manufactured by the above method is low and the light receiving sensitivity is expanded in the near-infrared long wavelength region is as described above.

More preferably, (e) a line obtained by translating the B line by +1 in the z-axis direction is larger.
As described above, there is a variation of ± 1.0 nm between the GaAsSb layer and the InGaAs layer in MQW. Even if there is such a variation, the lower limit of the film thickness is increased by 1.0 nm while narrowing the entire range in order to reliably obtain an increase in the quantum well film thickness (factor (F1)). Thereby, it is possible to pursue an improvement in sensitivity on the long wavelength side in the near infrared region while ensuring good crystallinity .

In the MQW formation step, when the lattice mismatch degree of the InGaAs layer is Δω ₁ and the lattice mismatch degree of the GaAsSb layer is Δω ₂ , the lattice mismatch degree of the entire multiple quantum well structure Δω = {Σ (Δω ₁ × The thickness of the InGaAs layer + Δω ₂ × the thickness of the GaAsSb layer)} / {Σ (the thickness of the InGaAs layer + the thickness of the GaAsSb layer)} is satisfied so that Δω satisfies −0.2% to 0.2%. It is good to form. This makes it possible to reduce the lattice defect density while maintaining the crystal quality of the MQW while changing the band structures of the GaAsSb layer and the InGaAs layer so as to be advantageous for expansion to the longer wavelength side.

A semiconductor layer including MQW and InP contact layers should be consistently grown on the InP substrate by a total metalorganic vapor phase epitaxy. Here, the all organic vapor phase growth method refers to a growth method using an organic metal raw material composed of a compound of an organic substance and a metal for all raw materials used for vapor phase growth, and is referred to as an all organic MOVPE method.
According to said method, the above-mentioned semiconductor element can be manufactured efficiently. That is, since the InP contact layer containing P is consistently grown by the all-organic MOVPE method, it can be continuously manufactured in the same growth tank. For example, even if an InP contact layer containing phosphorus is formed, solid phosphorus is not used as a raw material, so that phosphorus does not adhere to the inner wall of the growth tank. For this reason, there is no fear of ignition at the time of maintenance, and it is excellent in safety.
Another advantage of the all organic MOVPE method is that MQWs with steep heterointerfaces between each layer can be obtained. With MQW having a steep hetero interface, high-accuracy spectrum spectroscopy or the like can be performed.

  In the MQW formation step, the MQW is preferably formed at a temperature of 400 ° C. or higher and 560 ° C. or lower. Thereby, MQW excellent in crystallinity can be obtained, and the dark current can be further reduced. The above-mentioned temperature is a substrate surface temperature monitored by a pyrometer including an infrared camera and an infrared spectrometer, and the substrate surface temperature being monitored. Accordingly, although it is the substrate surface temperature, strictly speaking, it is the temperature of the epitaxial layer surface in a state where a film is formed on the substrate. There are various names such as a substrate temperature, a growth temperature, and a film formation temperature, and all refer to the monitored temperatures.

  According to the semiconductor element and the like of the present invention, it is possible to expand the light receiving sensitivity to the long wavelength side of the near infrared while keeping the dark current low. Further, since the organic organic MOVPE method consistently grows from the MQW of the light receiving layer to the InP contact layer, the manufacturing efficiency is high, and there is no adhesion of phosphorus to the inner surface of the growth tank, so that safety is also excellent.

It is a figure which shows the semiconductor element in Embodiment 1 of this invention. It is an enlarged view of MQW of FIG. It is a figure which shows the band structure of type 2 (InGaAs / GaAsSb), and the light absorption (light reception) phenomenon. It is a figure which shows the range which Sb composition x (%) of GaAsSb and film thickness z of GaAsSb satisfy | fill of type 2 InGaAs / GaAsSbMQW. It is a figure which shows the piping system etc. of the film-forming apparatus of all the organic MOVPE method. (A) is a figure which shows the flow of an organometallic molecule | numerator, and the flow of temperature, (b) is a schematic diagram of the organometallic molecule | numerator in the board | substrate surface. 2 is a flowchart of a method for manufacturing the light receiving element 50 of FIG. 1. It is an optical sensor apparatus containing the light receiving element array (semiconductor element) in Embodiment 2 of this invention. In MQW of Example 1, it is a figure which shows the position and PL wavelength of each test body in the relationship between the Sb composition of GaAsSb, and a film thickness. It is a figure which shows the relationship between the film thickness of the quantum well of each test body in Example 2, PL wavelength, and light reception sensitivity. It is a figure which shows the relationship between Sb composition x of GaAsSb of MQW of each test body in Example 3, PL wavelength, and dark current.

(Embodiment 1)
FIG. 1 is a diagram showing a semiconductor element 50 according to the first embodiment of the present invention. The light receiving element 50 has an InP-based semiconductor multilayer structure (epitaxial wafer) having the following configuration on the InP substrate 1. In FIG. 1, light is incident from the InP substrate side, but may be incident from the epitaxial side.
(InP substrate 1 / InP buffer layer 2 / type 2 (InGaAs / GaAsSb) MQW light receiving layer 3 / InGaAs diffusion concentration distribution adjusting layer 4 / InP contact layer 5)
The p-type region 6 positioned so as to reach from the InP contact layer 5 to the MQW light-receiving layer 3 is formed by selectively diffusing Zn of the p-type impurity from the opening of the selective diffusion mask pattern 36 of the SiN film. The The fact that diffusion is introduced into the periphery of the light receiving element 10 in a limited manner in a planar manner can be achieved by diffusion using the selective diffusion mask pattern 36 of the SiN film. A p-side electrode 11 made of AuZn is provided in the p-type region 6, and an n-side electrode 12 made of AuGeNi is provided in ohmic contact with the back surface of the InP substrate 1. In this case, the InP substrate 1 is doped with n-type impurities to ensure a predetermined level of conductivity. An antireflection film 35 made of SiON is provided on the back surface of the InP substrate 1 so that light can enter from the back surface side of the InP substrate. In the type 2 MQW light-receiving layer 3, a pn junction 15 is formed at a position corresponding to the boundary front of the p-type region 6, and a reverse bias voltage is applied between the p-side electrode 11 and the n-side electrode 12. As a result, a depletion layer is formed more widely on the side where the n-type impurity concentration is low (n-type impurity background). The background of the MQW light-receiving layer 3 is about 5E15 cm ⁻³ or less in terms of n-type impurity concentration (carrier concentration). The position 15 of the pn junction is determined by the intersection of the background (n-type carrier concentration) of the light-receiving layer 3 of the multiple quantum well and the concentration profile of the p-type impurity Zn. The diffusion concentration distribution adjustment layer 4 is arranged for adjusting the concentration distribution of the p-type impurity in the MQW constituting the light receiving layer 3, but the diffusion concentration distribution adjustment layer 4 may be omitted. In the light receiving layer 3, the Zn concentration is preferably 5E16 cm ⁻³ or less.

The point of the light receiving element in the present embodiment is as follows.
(P1) The light-receiving layer 3 is composed of type 2 (InGaAs / GaAsSb) MQW, the Sb composition x of GaAsSb is set to 44% or more, the Sb composition x of GaAsSb is increased, and the increased amount is absorbed as a whole. The In composition y of InGaAs is lowered so as to maintain InP lattice matching. That is, GaAsSb and InGaAs form a strain compensation structure. This point (P1) is the same as the factor (F2) for enlarging the light receiving area described later to the near-infrared long wavelength side.
(P2) The thickness z of the unit quantum well (GaAsSb) is set to 3 nm or more, and z> −0.4x (%) + 24.6 is satisfied. This feature mainly includes the above-mentioned factor (F1) film thickness increase, and also includes the above-mentioned factor (F2) as a secondary factor.
According to the above (P1), the energy difference between the valence band of GaAsSb and the conduction band of InGaAs can be reduced by increasing the Sb composition x and decreasing the In composition y. In addition, by maintaining the InP lattice matching, it is possible to grow an MQW, a contact layer, or the like having a low lattice defect density. As a result, the light receiving area can be expanded to a long wavelength area while keeping the dark current low.
Moreover, according to (P2), the long wavelength range in which light can be received can be 2.4 μm or more.
Furthermore, since it does not have a regrowth interface, high-concentration oxygen and carbon are not mixed as impurities. As a result, good crystallinity can be maintained up to the surface of the InP contact layer, and dark current can be reduced.

<Expansion of the light receiving area to the long wavelength side of the near infrared>
There are the following two factors for expanding the MQW light receiving area of type 2 (GaAsSb / InGaAs) to the long wavelength range of the near infrared.
(F1) Increase in film thickness of each layer of MQW. In the present embodiment, as shown in FIG. 2, the thickness z of GaAsSb is used as an index. The thickness of InGaAs is in the range of ± 1.0 nm of the thickness z of GaAsSb, and may be considered substantially the same.
(F2) Increase of Sb composition x to 44% or more and decrease of In composition in GaAsSb / InGaAs. The factor (F2) can be said to be the formation of a strain compensation structure from the viewpoint of InP lattice matching. However, in terms of the band structure, the energy difference between the valence band of GaAsSb and the conduction band of InGaAs is reduced. The crying point of this factor (F2) is to ensure InP lattice matching or to reduce lattice defect density.

The above factors (F1) increase in film thickness and (F2) distribution of compositions x and y will be described with reference to FIG. FIG. 3 is a diagram showing a band structure of type 2 (InGaAs / GaAsSb) and light absorption (light reception). When light having a wavelength λ on the long wavelength side is received, the electrons that have occupied the level of the valence band of GaAsSb are excited, and the electrons occupy the level of the conduction band of InGaAs. As a result of light reception, holes are generated in the valence band. In Type 2, electrons are excited from the valence band of GaAsSb to the conduction band of InGaAs, so that the energy difference is smaller than the transition from the valence band to the conduction band in a single phase, and long wavelength light Can be received.
When the wavelength λ of the light on the long wavelength side is received, the energy difference between the generated holes and electrons is h · (c / λ). Here, h is Planck's constant (6.626E-34J · s), and c is the speed of light in the medium. In order to expand the light receiving area to the longer wavelength side, it is necessary to bring both ends of the arrow of h · (c / λ) in FIG. 3 closer. As shown in FIG. 3, the factor (F2) decreases the energy difference ΔEvc between the valence band of GaAsSb and the conduction band of InGaAs in the band structure. That is, the energy difference ΔEvc is reduced by changing the band structure in which the electron level is formed. Further, the factor (F1) is considered to influence as follows. A single layer in the MQW forms one well potential. The energy level of electrons formed in the well potential tends to increase as the well potential width (film thickness) decreases. This corresponds to the fact that when energetic particles (waves) such as electrons are confined in a small space, the energy state becomes higher than when they are spread in a wide space. It may be said that it is a natural property. When the film thickness z is increased, the energy level of electrons (holes) shown in FIG. 3 approaches the valence band and the conduction band. As a result, even if the wavelength λ becomes larger and h · (c / λ) becomes smaller, light can be received.

FIG. 4 is a diagram showing a range where the Sb composition x (%) of GaAsSb and the film thickness z of GaAsSb satisfy the type 2 InGaAs / GaAsSbMQW in the light receiving element 50 of the present embodiment. The In composition y of InGaAs is distributed so as to satisfy 100 ≦ x + y ≦ 104. The meaning of each boundary line in FIG. 4 will be described below. The reason for setting the range in FIG. 4 will be described in Example 1 (see FIG. 9).
1. B line: z = −0.4x + 24.6
This B line defines the composition x and the film thickness z that the longest wavelength that can be received is 2.43 μm. In the range where z is above the B line, that is, in the xz range within the B line, the longest wavelength that can be received is 2.4 μm or more. In the present invention, the film thickness z is made larger than the B line in order to set the longest wavelength to 2.4 μm or more.
2. x = 44
The Sb composition x = 44 (%) is a line that can ensure the longest wavelength of 2.3 μm regardless of the film thickness z. When the Sb composition x is 44% or more and the film thickness z is 7 nm or more, the longest wavelength of 2.3 μm or more can be obtained. In the present embodiment, the Sb composition x is 44 (%) or more.
3. z = 3
In the present invention, the film thickness z is set to 3 nm or more in order to ensure the light receiving sensitivity.
4). A line: z = −0.625x + 45.5
The limit at which the dark current becomes large can be determined by the A line. On the outside including the A line (in a range where z is equal to or greater than the A line), the dark current increases and the S / N ratio decreases. On the inner side of this A line (range where z is located below this A line), the closer to the A line, the greater the film thickness z, the Sb composition x, and the light receiving area on the longer wavelength side. Enlarged. In the present embodiment, when low dark current is important, the xz range can be set inside the A line. As described above, unless both the film thickness z and the Sb composition x are close to the A-line, it is impossible to obtain a semiconductor element expanded to the long wavelength side in the near-infrared region where practical use is expected.
5. A2 line: z = −0.27x + 21.7
However, when emphasizing the reduction of dark current, in order to realize good crystallinity, (b1) 44 atm. % ≦ x <5 4a tm. In% of range, and z <10 nm, and, cited as a reference (b2) 5 4a tm. % ≦ x ≦ 68 atm. % Line A2: z ≦ −0.27x + 21.7. The A2 line will be described in detail in the embodiments.
6). z = 10
z = 10 nm defines the sensitivity limit. That is, when the film thickness z is 10 nm, the overlap of wave functions of electrons and holes is reduced, and the sensitivity is reduced. That is, the probability that electrons transition from the valence band of GaAsSb to the conduction band of InGaAs is reduced, and such electron transition (light reception) is less likely to occur. This comes from the demands of quantum mechanics. If the film thickness z is smaller than this, that is, if it is in the range of z <10, the light receiving sensitivity can be ensured.
7. z = 7
However, when importance is attached to the sensitivity, it is preferable that z ≦ 7 (nm).

<How to grow MQW>
Next, a manufacturing method will be described. An InP substrate 1 is prepared, and an InP buffer layer 2 / type 2 (InGaAs / GaAsSb) MQW light-receiving layer 3 / InGaAs diffusion concentration distribution adjusting layer 4 / InP contact layer 5 is grown thereon by an all-organic MOVPE method. To do. Here, the growth method of the light receiving layer 3 of type 2 (InGaAs / GaAsSb) MQW will be described in detail.
FIG. 5 shows a piping system and the like of the all-organic MOVPE film forming apparatus 60. A quartz tube 65 is disposed in the reaction chamber (chamber) 63, and a raw material gas is introduced into the quartz tube 65. A substrate table 66 is disposed in the quartz tube 65 so as to be rotatable and airtight. The substrate table 66 is provided with a heater 66h for heating the substrate. The temperature of the surface of the wafer 50 a during film formation is monitored by the infrared temperature monitor device 61 through a window 69 provided in the ceiling of the reaction chamber 63. This monitored temperature is a temperature at the time of growth or a temperature called a film forming temperature or a substrate temperature. In the production method of the present invention, when MQW is formed at a temperature of 400 ° C. or more and 560 ° C. or less, 400 ° C. or more and 560 ° C. or less are temperatures measured by this temperature monitor. The forced exhaust from the quartz tube 65 is performed by a vacuum pump.

The source gas is supplied by a pipe communicating with the quartz tube 65. The all organic MOVPE method is characterized in that all raw material gases are supplied in the form of an organometallic gas. In FIG. 5, source gases such as impurities are not specified, but impurities are also introduced in the form of an organometallic gas. The raw material of the organometallic gas is put in a thermostat and kept at a constant temperature. Hydrogen (H ₂ ) and nitrogen (N ₂ ) are used as the carrier gas. The organometallic gas is transported by a transport gas, and is sucked by a vacuum pump and introduced into the quartz tube 65. The amount of carrier gas is accurately adjusted by an MFC (Mass Flow Controller). Many flow controllers, solenoid valves, and the like are automatically controlled by a microcomputer.

A method for manufacturing the wafer 50a will be described. First, the n-type InP buffer layer 2 is epitaxially grown on the S-doped n-type InP substrate 1 to a film thickness of 150 nm. TeESi (tetraethylsilane) is preferably used for n-type doping. At this time, TMIn (trimethylindium) and TBP (tertiary butylphosphine) are used as the source gas. The InP buffer layer 2 may be grown using an inorganic raw material PH ₃ (phosphine). In the growth of the InP buffer layer 2, even if the growth temperature is about 600 ° C. or less than about 600 ° C., the crystallinity of the InP substrate located in the lower layer is not deteriorated by heating at about 600 ° C. However, when the InP contact layer is formed, since the MQW containing GaAsSb is formed in the lower layer, it is necessary to strictly maintain the substrate temperature within a range of, for example, a temperature of 400 ° C. or more and 560 ° C. or less. The reason is that when heated to about 600 ° C., the crystallinity of GaAsSb is greatly deteriorated due to heat damage, and when the InP contact layer is formed at a temperature lower than 400 ° C., the decomposition efficiency of the source gas is greatly increased. As a result, the impurity concentration in the InP layer increases and a high-quality InP contact layer cannot be obtained. Next, an n-type doped InGaAs layer is grown on the InP buffer layer 2 to a thickness of 0.15 μm (150 nm). This InGaAs layer is also included in the buffer layer 2 in FIG.

  Next, a type 2 MQW light-receiving layer 3 having InGaAs / GaAsSb as a quantum well pair is formed. As described above, the film thickness z of the GaAsSb in the quantum well is 3 nm or more and less than 10 nm, and the film thickness of the InGaAs 3b is preferably substantially the same as z ± 1.0 nm. In FIG. 1, the MQW light receiving layer 3 is formed by stacking 250 pairs of quantum wells. In the film formation of GaAsSb, triethylgallium (TEGa), tertiary butylarsine (TBAs), and trimethylantimony (TMSb) are used. For InGaAs, TEGa, TMIn, and TBAs can be used. These source gases are all organometallic gases, and the molecular weight of the compound is large. Therefore, it can be completely decomposed at a relatively low temperature of 400 ° C. or higher and 560 ° C. or lower and contribute to crystal growth. The MQW light-receiving layer 3 can be made abrupt in composition change at the interface of the quantum well by using all organic MOVPE. As a result, highly accurate spectrophotometry can be performed.

  As a raw material for Ga (gallium), TEGa (triethylgallium) or TMGa (trimethylgallium) may be used. The raw material for In (indium) may be TMIn (trimethylindium) or TEIn (triethylindium). As a raw material of As (arsenic), TBAs (tertiary butylarsine) or TMAs (trimethylarsenic) may be used. The raw material for Sb (antimony) may be TMSb (trimethylantimony), TESb (triethylantimony), TIPSb (triisopropylantimony), or TDMASb (tridimethylaminoantimony). By using these raw materials, a semiconductor element having a low MQW impurity concentration and excellent crystallinity can be obtained. As a result, for example, when used in a light receiving element, a light receiving element with a small dark current and a high sensitivity can be obtained. Furthermore, using the light receiving element, an optical sensor device capable of capturing a clearer image, such as an imaging device, can be obtained.

Next, the flow state of the source gas when the multiple quantum well structure 3 is formed by the all organic MOVPE method will be described. The source gas is transported through the piping, introduced into the quartz tube 65, and exhausted. Any number of source gases can be supplied to the quartz tube 65 by increasing the number of pipes. For example, even a dozen kinds of source gases are controlled by opening and closing the electromagnetic valve.
The flow rate of the source gas is controlled by a flow rate controller (MFC) shown in FIG. 5, and the flow into the quartz tube 65 is turned on and off by opening and closing the electromagnetic valve. The quartz tube 65 is forcibly exhausted by a vacuum pump. There is no stagnation in the flow of the source gas, and it is performed smoothly and automatically. Therefore, the composition is switched quickly when forming the quantum well pair.
As shown in FIG. 5, since the substrate table 66 rotates, the temperature distribution of the source gas does not have the directivity as on the inflow side or the outlet side of the source gas. Further, since the wafer 50a revolves on the substrate table 66, the flow of the source gas near the surface of the wafer 50a is in a turbulent state, and even the source gas near the surface of the wafer 50a contacts the wafer 50a. Except for the raw material gas, it has a large velocity component in the flow direction from the introduction side to the exhaust side. Therefore, most of the heat flowing from the substrate table 66 to the source gas through the wafer 50a is always exhausted together with the exhaust gas. For this reason, a large temperature gradient or temperature step is generated in the vertical direction from the wafer 50a through the surface to the source gas space.
Furthermore, in the embodiment of the present invention, the substrate temperature is heated to a low temperature range of 400 ° C. or more and 560 ° C. or less. In the case of using the all organic MOVPE method using TBAs or the like as a raw material at the substrate surface temperature in such a low temperature region, since the decomposition efficiency of the raw material is good, the growth of the multiple quantum well structure with the raw material gas flowing in the range very close to the wafer 50a The source gas that contributes to is limited to those efficiently decomposed into the form necessary for growth.

FIG. 6A is a diagram showing the flow of organometallic molecules and the flow of temperature, and FIG. 6B is a schematic diagram of organometallic molecules on the substrate surface. These diagrams are used to explain the importance of setting the surface temperature in order to obtain a steep composition change at the heterointerface of the multiple quantum well structure.
The surface of the wafer 50a is set to a monitored temperature. However, when the material gas space is slightly entered from the wafer surface, the temperature suddenly decreases or a large temperature step is generated as described above. Therefore, in the case of a raw material gas having a decomposition temperature of T1 ° C., the substrate surface temperature is set to (T1 + α), and α is determined in consideration of variations in temperature distribution and the like. In a situation where there is a sudden large temperature drop or temperature step from the surface of the wafer 50a to the source gas space, when large-sized organometallic molecules flow through the wafer surface as shown in FIG. It is considered that the compound molecules that contribute to the growth are limited to those in the range in contact with the surface and the film thickness range of several organometallic molecules from the surface. Therefore, as shown in FIG. 6B, the organic metal molecules in the range in contact with the wafer surface and the molecules located within the film thickness range of several organometallic molecules from the wafer surface are mainly crystal growth. It is considered that the organometallic molecules outside it are discharged out of the quartz tube 65 with almost no decomposition. When the organometallic molecules near the surface of the wafer 50a are decomposed and crystal growth occurs, the organometallic molecules located outside enter the replenishment.
In other words, the range of the organometallic molecules that can participate in crystal growth is limited to a thin source gas layer on the surface of the wafer 50a by making the wafer surface temperature slightly higher than the temperature at which the organometallic molecules decompose. it can.

From the above, when the source gas suitable for the chemical composition of the pair is switched by the electromagnetic valve and forcedly evacuated by the vacuum pump, after the crystal of the previous chemical composition is grown with slight inertia, Thus, it is possible to grow a crystal having a switched chemical composition without being affected by the source gas. As a result, the composition change at the hetero interface can be made steep. This means that the previous source gas does not substantially remain in the quartz tube 65, and the source gas that contributes to the growth of the multiple quantum well structure with the source gas flowing in a range very close to the wafer 50a is: This is because it is limited to those efficiently decomposed into a shape necessary for growth (film formation factor 1). That is, as can be seen from FIG. 5, when one layer of the quantum well is formed, the electromagnetic valve is opened and closed while forcibly evacuating with a vacuum pump, and when the source gas for forming the other layer is introduced, a little Although there are organometallic molecules that participate in crystal growth with inertia, the molecules in one layer that supplements are almost exhausted and gone. The closer the wafer surface temperature is to the decomposition temperature of the organometallic molecule, the smaller the range of organometallic molecules participating in crystal growth (range from the wafer surface).
When forming this multi-quantum well structure, phase growth occurs in the GaAsSb layer of the multi-quantum well structure when grown in a temperature range of about 600 ° C., and the crystal growth surface of the multi-quantum well structure that is clean and excellent in flatness, and A multiple quantum well structure having excellent periodicity and crystallinity cannot be obtained. For this reason, the growth temperature is set to a temperature range of 400 ° C. or more and 560 ° C. or less (deposition factor 2). This film formation method is an all-organic MOVPE method, and all the source gases are organometallic gases with high decomposition efficiency. (Film formation factor 3) strongly depends on film formation factor 1.

<Method for Manufacturing Semiconductor Device>
In the semiconductor element 50 shown in FIG. 1, the InGaAs diffusion concentration distribution adjustment layer 4 is located on the type 2 MQW light-receiving layer 3, and the InP contact layer 5 is located on the InGaAs diffusion concentration distribution adjustment layer 4. ing. The p-type region 6 is provided by selectively diffusing Zn of the p-type impurity from the opening of the selective diffusion mask pattern 36 provided on the surface of the InP contact layer 5. A pn junction 15 or a pi junction 15 is formed at the tip of the p-type region 6. A depletion layer is formed by applying a reverse bias voltage to the pn junction 15 or the pi junction 15 to capture charges due to photoelectron conversion, and to make the brightness of the pixel correspond to the amount of charges. The p-type region 6 or the pn junction 15 or the pi junction 15 is a main part constituting the pixel. The p-side electrode 11 that is in ohmic contact with the p-type region 6 is a pixel electrode, and reads the above charges for each pixel with the n-side electrode 12 that is set to the ground potential. The selective diffusion mask pattern 36 is left as it is on the surface of the InP contact layer around the p-type region 6. Further, a protective film such as SiON not shown is coated. The selective diffusion mask pattern 36 is left as it is. When the p-type region 6 is formed and then exposed to the atmosphere except for this, a surface level is formed at the boundary with the p-type region on the contact layer surface, and the dark current is formed. This is because of the increase.
One point is to continue the growth in the same film forming chamber or quartz tube 65 by the all organic MOVPE method until the formation of the InP contact layer 5 after the MQW is formed as described above. That is, before the InP contact layer 5 is formed, the wafer 50a is not taken out from the film forming chamber and the InP contact layer 5 is not formed by another film forming method. One point. That is, since the InGaAs diffusion concentration distribution adjusting layer 4 and the InP contact layer 5 are continuously formed in the quartz tube 65, the interfaces 16 and 17 are not regrowth interfaces. For this reason, the concentrations of oxygen, carbon, and silicon are all below a predetermined level, and charge leakage does not occur particularly at the intersection line between the p-type region 6 and the interface 17.

In the present embodiment, a non-doped InGaAs diffusion concentration distribution layer 4 having a film thickness of 1.0 μm, for example, is formed on the MQW light-receiving layer 3. After the InP contact layer 5 is formed, the InGaAs diffusion concentration distribution layer 4 has a high concentration of Zn when the p-type impurity Zn is introduced from the InP contact layer 5 to reach the MQW light receiving layer 3 by a selective diffusion method. When it enters MQW, the crystallinity is damaged, so it is provided for the adjustment. The InGaAs diffusion concentration distribution adjustment layer 4 may be arranged as described above, but may not be provided.
The p-type region 6 is formed by the selective diffusion described above, and the pn junction 15 or the pi junction 15 is formed at the tip thereof. Even when the InGaAs diffusion concentration distribution adjusting layer 4 is inserted, since the InGaAs has a small band gap, the electric resistance of the light receiving element can be lowered even if it is non-doped. By reducing the electrical resistance, it is possible to improve the responsiveness and obtain a moving image with good image quality.
On the InGaAs diffusion concentration distribution adjusting layer 4, the undoped InP contact layer 5 is epitaxially grown to a thickness of, for example, 0.8 μm by the all organic MOVPE method continuously with the wafer 50a placed in the same quartz tube 65. Is good. As described above, trimethylindium (TMIn) and tertiary butylphosphine (TBP) are used for the source gas. By using this source gas, the growth temperature of the InP contact layer 5 can be made 400 ° C. or more and 560 ° C. or less, and further 535 ° C. or less. As a result, the MQW GaAsSb positioned under the InP contact layer 5 is not damaged by heat, and the MQW crystallinity is not impaired. When the InP contact layer 5 is formed, since the MQW containing GaAsSb is formed in the lower layer, the substrate temperature needs to be strictly maintained within a range of, for example, a temperature of 400 ° C. or more and 560 ° C. or less. The reason is that when heated to about 600 ° C., the crystallinity of GaAsSb is greatly deteriorated due to heat damage, and when the InP contact layer is formed at a temperature lower than 400 ° C., the decomposition efficiency of the source gas is greatly increased. Therefore, the impurity concentration in the InP contact layer 5 increases, and a high quality InP contact layer 5 cannot be obtained.

As described above, conventionally, it has been necessary to form the MQW by the MBE method. However, in order to grow an InP contact layer by the MBE method, it is necessary to use a solid raw material as a phosphorus raw material, which is problematic in terms of safety. There was also room for improvement in terms of manufacturing efficiency.
Prior to the present invention, the interface between the InGaAs diffusion concentration distribution adjusting layer and the InP contact layer was a regrowth interface once exposed to the atmosphere. The regrowth interface is specified by satisfying at least one of oxygen concentration of 1E17 cm ⁻³ or more, carbon concentration of 1E17 cm ⁻³ or more, and silicon concentration of 1E17 cm ⁻³ or more by secondary ion mass spectrometry. Can do. The regrowth interface forms a crossing line with the p-type region, and a charge leak occurs at the crossing line, thereby significantly degrading the image quality.
Further, for example, when an InP contact layer is grown by a simple MOVPE method, phosphine (PH ₃ ) is used as a raw material of phosphorus. Therefore, the decomposition temperature is high, and the occurrence of damage due to heat of GaAsSb located in the lower layer is induced to cause the MQW crystal. It will be harmful to sex.

  FIG. 7 is a flowchart of a method for manufacturing the light receiving element 50 of FIG. According to this manufacturing method, only the organometallic gas is used as the source gas (deposition factor 3) until the growth temperature is lowered (deposition factor 2) and the formation of the InP contact layer 5 is completed. Therefore, it is important to have no recrystallization interface (deposition factor 4). As a result, it is possible to efficiently manufacture a large number of photodiodes having a light receiving sensitivity in a wavelength region of 2 μm to 5 μm, which has less charge leakage and excellent crystallinity.

(Embodiment 2)
FIG. 8 shows an optical sensor device 10 including a light receiving element array (semiconductor element) 50 according to the second embodiment of the present invention. Optical parts such as lenses are omitted. Although a protective film 43 made of a SiON film is shown in FIG. 8, it is actually also arranged in FIG. The light receiving element array 50 has the same stacked structure as that of the light receiving elements shown in FIG. 1, and is different in that a plurality of light receiving elements or pixels P are arranged. The film thickness z, the Sb composition s, etc. are the same as those in the semiconductor element of FIG. Also, the interfaces 16 and 17 are not regrowth interfaces, and the concentration of impurities such as oxygen, carbon, and silicon are all low, which is the same as the light receiving element (semiconductor element) in FIG.
In FIG. 8, the light receiving element array 50 and a CMOS 70 constituting a read circuit (Read-Out IC) are connected. The readout electrode (not shown) of the CMOS 70 and the pixel electrode (p-side electrode) 11 of the light receiving element array 50 are bonded with the bonding bump 39 interposed. Further, a common ground electrode (n-side electrode) 12 for each pixel of the light receiving element array 50 and a ground electrode (not shown) of the CMOS 70 are joined via bumps 12b. An imaging device or the like can be obtained by combining the CMOS 70 and the light receiving element array 50 and integrating light reception information for each pixel.
As described above, the light-receiving element array (semiconductor element) 50 of the present invention has sensitivity up to a long wavelength region and has a small dark current (leakage current), and therefore, inspection of living organisms such as animals and plants, environmental monitoring, etc. By using this, high-precision inspection can be performed.

Example 1
The light-receiving element 50 having the same structure as that of the semiconductor element 50 shown in FIG. Was measured. Table 1 shows the Sb composition x of GaAsSb, the In composition y of InGaAs, and the film thickness z (nm) of GaAsSb and InGaAs of each specimen. The film thicknesses of GaAsSb and InGaAs were the same. The composition x and the composition y have a lattice mismatch degree of InGaAs of Δω ₁ and a lattice mismatch degree of GaAsSb of Δω _2, and the lattice mismatch degree of the entire multiple quantum well structure Δω = {Σ (Δω ₁ × ΔGaAs thickness + Δω ₂ × GaAsSb layer thickness)} / {Σ (InGaAs layer thickness + GaAsSb layer thickness)} Δω defined by −0.2% or more and 0.2% or less did.
The light receiving element 50 was manufactured by the all organic MOVPE method as described above. Parts other than the light receiving layer were manufactured by the above method. The manufacturing method of the main epitaxial layer was performed as follows. The growth temperature is 510 ° C. GaAsSb uses TEGa, TBAs and TESb as raw materials, and the Sb composition x has a constant V / III ratio (supply amount of Group V raw materials / supply amount of Group III raw materials), and the supply amount ratio of TBAs and TESb It changed by changing. InGaAs was grown using TEGa, TMIn, and TBAs as raw materials, and the In composition y was grown by changing the supply ratio of TEGa and TMnIn while keeping the V / III ratio constant. In the InP contact layer 5, TMIn and TBP were used as raw materials.

<Evaluation>
Evaluation was performed by measuring the PL wavelength at room temperature and dark current as described below. The dark current was 100 μm in diameter and the reverse bias voltage (Vr) = 1V.
1. PL wavelength at room temperature The wavelength of PL at room temperature corresponds to the bandgap wavelength and corresponds to the longest wavelength that can be received. 9 and 4, in Comparative Examples 2, 3, 8, and 10, the film thickness z is located below the B line: −0.4x + 24.6, The PL wavelengths are 2210 nm, 2290 nm, 2210 nm, and 2300 nm, respectively, and are less than 2400 nm. This is because the film thickness z and the Sb composition x of GaAsSb are not sufficiently high.
On the other hand, in the range satisfying the film thickness z ≧ 3 nm and the film thickness z> −0.4x + 24.6 ( larger than the B line), as shown in Table 1, any of the present invention examples The PL wavelength is 2.4 μm or more. Moreover, the PL wavelength becomes longer as it approaches A line: z = −0.625x + 45.5. In FIG. 9, the arrow and the PL wavelength indicate that the PL wavelength becomes longer as it approaches A line and z = 10.
Further, when approaching the line of film thickness z = 10, the PL wavelength is as in Reference Example 7 (2430 nm) → Example 6 (2620 nm) → Example 5 (2700 nm) → Reference Example 4 (2870 nm). After all it becomes long.
2. Dark current The dark current is limited by the A line: z = −0.625x + 45.5. Since the dark current increases as the lattice defect density increases, it tends to be considered that only the Sb composition x and the In composition y. However, it depends on the film thickness as described above, and is influenced not only by the composition x and y but also by the film thickness z. It is thought that lattice defects are accumulated.
Reference Example 16 and Reference Example 19 are located on the A line, and the dark currents are 700 nA and 800 nA, respectively. If the dark current is at this level, there are many allowable applications. However, when the dark current is important, the film thickness z is preferably in a range smaller than the A line. In order to further reduce the dark current, it is preferable to set the A2 line: z = −0.27x + 21.7 or less. In this case, Invention Example 16 and Invention Example 19 do not fall within the scope of the present invention and are comparative examples.
The ranges slightly apart from the A line, for example, Reference Examples 11, 14, 17, and 18, are 200 nA, 200 nA, 300 nA, and 200 nA, respectively, which are practically satisfactory levels.
3. Photosensitive sensitivity As will be described later in Example 2, when the thickness z of GaAsSb reaches 10 nm, the received light sensitivity decreases. Even if the film thickness is z = 10 nm, there is a certain level of light sensitivity, which is possible depending on the application. However, when the light receiving sensitivity is very problematic, the film thickness z should be smaller than 10 nm. If higher light receiving sensitivity is required, z ≦ 7 (nm) is preferable. In this case, Reference Example 4 is outside the scope of the present invention.
This is a quantum mechanical effect and is considered to hold regardless of the conditions of the examples. Also, when the film thickness z is as thin as 2 nm, the light receiving sensitivity is low. For this reason, the film thickness z is 3 nm or more.

(Example 2)
<Manufacture of specimen>
The epitaxial layer of the buffer layer (InP / InGaAs) / type 2 (InGaAs / GaAsSb) MQW light receiving layer / InGaAs diffusion concentration distribution adjusting layer / InP contact layer is formed on the S-doped InP substrate by the all organic MOVPE method. Formed consistently. An InP contact layer was grown directly on the MQW light-receiving layer 3. TEGa, TMIn, TBAs, TBP, and TMSb were used as raw materials for Ga, In, As, P, and Sb, respectively. TeESi was used for n-type impurity doping.
Specifically, an n-type doped InP buffer layer was grown to 150 nm on an S-doped InP substrate, and an n-type doped InGaAs buffer layer was grown to 0.15 μm thereon. An InGaAs / GaAsSb type 2 MQW light-receiving layer was grown on the two buffer layers. The MQW configuration was repeated 250 pairs with the lower undoped InGaAs layer and the undoped GaAsSb layer as a pair. The composition is a composition that lattice-matches alone, and the Sb composition x of GaAsSb is 49% and the In composition y of InGaAs is 53%. When the lattice mismatch degree of InGaAs is Δω ₁ and the lattice mismatch degree of GaAsSb is Δω ₂ , the lattice mismatch degree of the entire multiple quantum well structure Δω = {Σ (Δω ₁ × InGaAs layer thickness + Δω ₂ × GaAsSb Δω defined by (layer thickness)} / {Σ (InGaAs layer thickness + GaAsSb layer thickness)} is −0.2% or more and 0.2% or less. An InGaAs diffusion concentration distribution adjusting layer was grown by 1.0 μm on the MQW light-receiving layer, and an undoped InP contact layer was grown by 0.8 μm thereon. For the growth of GaAsSb, TEGa, TBAs, and TMSb were used as raw materials. Further, TEBa, TMIn, and TBAs were used for the growth of InGaAs. Further, TMIn and TBP were used for the growth of the InP contact layer or InP buffer layer.

<Evaluation>
1. The PL characteristics are shown in Table 2 and FIG. When the film thickness z of GaAsSb and the film thickness of InGaAs were 5 nm, the PL peak wavelength was 2.5 μm. As both film thicknesses were decreased, the PL wavelength shifted to the short wavelength side. When the film thickness of GaAsSb and InGaAs was 2 nm, the PL wavelength was 1.9 μm.
On the other hand, as the thickness of InGaAs and GaAsSb was increased, the PL wavelength became longer. When both thicknesses were 10 nm, the PL wavelength was 2.9 μm.
2. Light receiving sensitivity The light receiving sensitivity was measured with respect to light having a wavelength of 2000 nm at a reverse bias voltage Vr = −5V. When the film thicknesses of both GaAsSb and InGaAs were 2 nm, the light receiving sensitivity was as low as 0.1 A / W. As both film thicknesses were increased to 3 nm, 5 nm, and 7 m, they improved to 0.6 A / W, 0.6 A / W, and 0.5 A / W. And when both film thicknesses were 10 nm, it deteriorated with 0.2 A / W.

(Example 3)
The test body was configured as shown in Example 2 and followed the same procedure. However, in this example, the film thickness was fixed at 5 nm for both GaAsSb and InGaAs, and the supply amount of the source gas was adjusted as described above. At this time, when the lattice mismatch degree of InGaAs is Δω ₁ and the lattice mismatch degree of GaAsSb is Δω ₂ , the lattice mismatch degree of the entire multiple quantum well structure Δω = {Σ (Δω ₁ × InGaAs layer thickness + Δω ₂ × GaAsSb layer thickness)} / {Σ (InGaAs layer thickness + GaAsSb layer thickness)} Δω defined by −0.2% to 0.2%.

<Evaluation>
1. The PL characteristics are shown in Table 3 and FIG. When the Sb composition x of GaAsSb was 49% and the In composition y of InGaAs was 53%, the PL wavelength was 2.4 μm. When the Sb composition x of GaAsSb is increased and the InGaAs composition y is decreased, the PL wavelength is shifted to the longer wavelength side. The Sb composition was 62%, the In composition was 38%, and the PL wavelength was 3.0 μm.
2. Dark current Whether the reverse bias voltage Vr is -1V or -5V, when the Sb composition is increased to 44% or more, the dark current gradually increases, but it can be said to be a good level. When the In composition y was lower than 53% and the Sb composition was increased from 62%, the difference in dark current between the reverse bias voltage Vr = −1V and −5V became large. Thus, in order to increase the S / N ratio in a light receiving element having light receiving sensitivity up to a longer wavelength side, it is preferable to set a lower reverse bias voltage (decrease the absolute value).

  Although the embodiments of the present invention have been described above, the embodiments of the present invention disclosed above are merely examples, and the scope of the present invention is not limited to these embodiments. The scope of the present invention is indicated by the description of the scope of claims, and further includes meanings equivalent to the description of the scope of claims and all modifications within the scope.

  According to the semiconductor device of the present invention, the Sb composition is adjusted while adjusting the composition so that the InP lattice matching condition is integrated in both GaAsSb and InGaAs while suppressing the dark current to a level at which there is no problem in practice. By increasing (decreasing the In composition) and increasing the thickness of the quantum well, the light receiving sensitivity can be expanded to the long wavelength side in the near infrared region. As a result, an important application can be covered with a simple device. Further, by consistently growing in the same growth tank by the all organic OMVPE method, impurity contamination can be prevented and high quality crystallinity can be obtained. In addition, the steepness of the composition between the MQW quantum wells of the light receiving layer can be increased, and the analysis of the absorption spectrum and the like can be performed with high accuracy.

  1 InP substrate, 2 buffer layer (InP and / or InGaAs), 3 type II MQW light receiving layer, 4 InGaAs layer (diffusion concentration distribution adjusting layer), 5 InP contact layer, 6 p-type region, 10 optical sensor device (detection device) 11 p-side electrode (pixel electrode), 12 ground electrode (n-side electrode), 12b bump, 15 pn junction, 16 interface between MQW and InGaAs layer, 17 interface between InGaAs layer and InP window layer, 20 infrared temperature monitor Equipment, 21 reaction chamber window, 30 reaction chamber, 35 AR (antireflection) film, 36 selective diffusion mask pattern, 39 bonding bump, 43 protective film (SiON film), 50 light receiving element (light receiving element array), 50a wafer ( Intermediate products), 60 all-organic MOVPE film deposition equipment, 61 infrared temperature monitoring equipment, 63 reaction chamber, 5 quartz tube, 69 a window of the reaction chamber, 66 a substrate table, 66h heater, 70 CMOS, P pixels.

Claims (23)

A semiconductor element formed on a group III-V semiconductor substrate,
A light-receiving layer of type 2 multiple quantum well structure located on the III-V semiconductor substrate;
The multiple quantum well is composed of a repeating structure of a layer containing at least Ga, As, and Sb and a layer containing at least In, Ga, and As.
The layer containing at least Ga, As, and Sb has an Sb composition x (atm.%) Of 44 atm. %, And the film thickness z (nm) is 3 nm or more and satisfies z (nm) ≧ −0.4x (atm.%) + 24.6.   The semiconductor device according to claim 1, wherein the group III-V semiconductor is InP.   The multiple quantum well structure has a strain compensation structure, and the Sb composition x of the layer containing at least Ga, As and Sb, and the In composition y (atm.%) Of the layer containing at least In, Ga and As, and Is a distribution satisfying 100 ≦ x + y ≦ 104. 3. The semiconductor device according to claim 1, wherein the distribution satisfies 100 ≦ x + y ≦ 104.   The semiconductor device according to claim 1, further comprising an InP contact layer on the multiple quantum well structure.   The film thickness z (nm) of the layer containing at least Ga, As, and Sb in the multiple quantum well structure depends on (a1) 44 atm. % ≦ x <56.8 atm. %, Z <10 nm, (a2) 56.8 atm. % ≦ x ≦ 68 atm. The semiconductor device according to claim 1, wherein z <−0.625x + 45.5 in a range of%.   The film thickness z (nm) of the layer containing at least Ga, As, and Sb in the multiple quantum well structure depends on (b1) 44 atm. % ≦ x <54.3 atm. %, Z ≦ 7 nm, and (b2) 54.3 atm. % ≦ x ≦ 68 atm. % In the range of z ≦ −0.27x + 21.7. 6. The semiconductor device according to claim 1, wherein z ≦ −0.27x + 21.7. When the lattice mismatch degree of the layer containing at least In, Ga and As in the multiple quantum well structure is Δω ₁ and the lattice mismatch degree of the layer containing at least Ga, As and Sb is Δω ₂ , the multiple quantum well structure Total lattice mismatch Δω = {Σ (Δω ₁ × layer thickness including at least In, Ga, As + Δω ₂ × layer thickness including at least Ga, As, Sb} / {Σ (at least In, Ga, As The thickness of the layer containing N + the thickness of the layer containing at least Ga, As, and Sb)} satisfies −0.2% or more and 0.2% or less. The semiconductor element according to any one of the above. The semiconductor element according to claim 1, wherein the layer containing at least Ga, As, and Sb is GaAs _1-x Sb _x (hereinafter, GaAsSb). The semiconductor element according to claim 1, wherein the layer containing at least In, Ga, As is In _y Ga _1-y As (hereinafter, InGaAs).   10. The semiconductor device according to claim 1, wherein a longest wavelength at which the light receiving layer has light receiving sensitivity is 2.4 μm or more. 11.   11. The regrowth interface is not provided between the bottom surface of the light receiving layer and the top surface of the semiconductor layer including the light receiving layer and the InP contact layer. 11. Semiconductor element.   An optical sensor device using the semiconductor element according to claim 1 as a light receiving element. A method of manufacturing a semiconductor device formed on a group III-V semiconductor substrate,
Forming a type 2 multiple quantum well structure light-receiving layer on the III-V semiconductor substrate;
The multiple quantum well is composed of a layer containing at least Ga, As, and Sb and a layer containing at least In, Ga, and As.
In the step of forming the multiple quantum well structure, the Sb composition x (atm.%) Of the layer containing at least Ga, As, and Sb is set to 44 atm. %, And the Sb composition x (atm.%) And the film thickness z (nm) satisfy z ≧ −0.4x + 24.6. .   The method of manufacturing a semiconductor device according to claim 13, wherein the III-V semiconductor is InP.   The layer containing at least Ga, As and Sb and the layer containing at least In, Ga and As form a strain compensation structure, and the Sb composition x of the layer containing at least Ga, As and Sb is 44 atm. %, The In composition y (atm.%) Of the layer containing at least In, Ga, As is 1 atm. Of the Sb composition x. The method for manufacturing a semiconductor device according to claim 13, wherein the growth is performed such that the growth is reduced at a rate of 0.9 to 1.2 times per% increase.   16. The method for manufacturing a semiconductor device according to claim 13, further comprising a step of forming an InP contact layer on the multiple quantum well structure.   The film thickness z (nm) of the layer containing at least Ga, As, and Sb in the multiple quantum well structure is set to 3 nm or more, and (a1) 44 atm. % ≦ x <56.8 atm. %, Z <10 nm, and (a2) 56.8 atm. % ≦ x ≦ 68 atm. The method of manufacturing a semiconductor element according to claim 13, wherein z <−0.625x + 45.5 in a range of%.   The film thickness z (nm) of the layer containing at least Ga, As, and Sb in the multiple quantum well structure is set to 3 nm or more, and (b1) 44 atm. % ≦ x <54.3 atm. %, Z ≦ 7 nm, and (b2) 54.3 atm. % ≦ x ≦ 68 atm. The method of manufacturing a semiconductor device according to claim 13, wherein z ≦ −0.27x + 21.7 in a range of%. In the step of forming the multiple quantum well structure, when at least In, Ga, lattice mismatch of the layer containing As and [Delta] [omega _1, at least Ga, As, lattice mismatch of the layer containing Sb and [Delta] [omega _2, multiple quantum well structure entire lattice mismatch Δω = {Σ (Δω ₁ × least in, Ga, thickness + [Delta] [omega ₂ × least Ga layer comprising As, As, thickness of the layer containing Sb)} / {Σ (at least The thickness of the layer containing In, Ga, As and the thickness of the layer containing at least Ga, As, Sb)} is defined such that Δω satisfies −0.2% to 0.2%. The method for manufacturing a semiconductor element according to claim 13, wherein: 20. The method of manufacturing a semiconductor device according to claim 13, wherein the layer containing at least Ga, As, and Sb is GaAs _1-x Sb _x (hereinafter, GaAsSb). Wherein at least In, Ga, a layer containing As is _In y _{Ga 1-y} As (hereinafter, InGaAs) characterized in that it is a method of manufacturing a semiconductor device according to any one of claims 13 to 20.   The semiconductor layer including the multiple quantum well structure and the InP contact layer is continuously grown on the InP substrate by a total metal organic chemical vapor deposition method. 2. A method for producing a semiconductor element according to item 1. The semiconductor device according to any one of claims 13 to 22, wherein in the step of forming the multiple quantum well structure, the multiple quantum well structure is formed at a temperature of 400 ° C or higher and 560 ° C or lower. Production method.

Images