JP2017017112A - Method for manufacturing optical semiconductor element - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a method for manufacturing an optical semiconductor element including an optical waveguide layer which enlarges a spot size of the light input from an active layer and reduces the reflection of the light.SOLUTION: A method for manufacturing an optical semiconductor element comprises a first step, a second step, and a third step. The first step forms a selective growth mask 30 on the laminate laminated with a substrate 11, a clad layer 112 arranged on the substrate, and an active layer 14 arranged on the clad layer. In the first step, the selective growth mask 30 includes an opening 31 extending from an edge of the selective growth mask 30 toward the inner side of the mask. In the opening 31, the width of the inner side is wider than the width of the edge side. The second step removes portions of the active layer 14 located inside the opening 31 using the selective growth mask 30. The third step forms an optical waveguide layer 15 in the opening 31 with the selective growth mask 30 remaining on the active layer 14.SELECTED DRAWING: Figure 11

Description

  The present invention relates to a method for manufacturing an optical semiconductor element.

  Conventionally, between optical elements having different light spot sizes, in order to reduce the optical loss of light propagating between the two optical elements, a spot size converter is used to convert the light spot size. ing.

  Silicon photonics technology that integrates a plurality of optical elements on a silicon substrate at high density is expected as a promising technology for realizing a compact and large-capacity optical transceiver.

  Since silicon is a semiconductor that emits light based on indirect transition of carriers, there are many problems in using it as a material for a light-emitting element such as a semiconductor laser. Therefore, it has been attempted to mount a light emitting element formed of a compound semiconductor that emits light based on direct transition on a silicon substrate.

  Since a light emitting element mounted on a silicon substrate may be disposed in the vicinity of an element that generates a large amount of heat, such as an LSI unit, it is required to operate stably even at high temperatures. Therefore, it has been studied to mount a light emitting element having an active layer including quantum dots that emit light stably over a wide temperature range on a silicon substrate.

  When a light emitting element having an active layer containing quantum dots is mounted on a silicon substrate and optically coupled to a silicon optical waveguide, there may arise a problem that the coupling loss of light increases.

This is because the spot size in the vertical direction of the light emitting element having an active layer containing quantum dots (full width at a position 1 / e ² of the mode field diameter of the near-field image) is narrow (eg, 1.4 μm). This is because the dimensions are different from the spot size of the waveguide (eg, 3 μm).

  Therefore, it has been studied to arrange a spot size conversion section between a light emitting element having an active layer containing quantum dots and a silicon optical waveguide.

  FIG. 1 is a diagram showing a conventional optical semiconductor device.

  The optical semiconductor element 110 includes a light generation unit 110A that generates laser light and a spot size conversion unit 110B. The spot size conversion unit 110B enlarges the spot size of the light output from the light generation unit 110A and outputs it to the outside. The light generation unit 110A and the spot size conversion unit 110B are formed using a compound semiconductor.

  The light generation unit 110A includes a substrate 111, a lower cladding layer 113 disposed on the substrate 111, an active layer 114 disposed on the lower cladding layer 113, and an upper cladding layer 116 disposed on the active layer 114. Have. The active layer 114 has a multiple quantum well (MQW) structure. A first electrode layer 119 a is disposed on the upper cladding layer 116, and a second electrode layer 119 b is disposed under the substrate 111. The substrate 111, the lower cladding layer 113, and the second electrode layer 119b extend from the light generation unit 110A to the spot size conversion unit 110B.

  The spot size conversion unit 110 </ b> B includes an optical waveguide layer 115 disposed on the lower cladding layer 113 and an upper cladding layer 117 disposed on the optical waveguide layer 115. The optical waveguide layer 115 and the active layer 114 are optically coupled by butt joint bonding.

  The optical waveguide layer 115 has the same thickness as that of the active layer 114 at the junction J with the active layer 114, but the thickness rapidly decreases as the optical waveguide layer 115 moves away from the junction J. Light that is output from the active layer 114 and propagates through the optical waveguide layer 115 decreases in thickness, and the amount of light that oozes out from the optical waveguide layer 115 to the upper cladding layer 116 and the lower cladding layer 113 increases. Then the spot size will increase. In this way, the light generated by the light generation unit 110A is enlarged in spot size by the spot size conversion unit 110B and output to the outside of the optical semiconductor element 110 (for example, a silicon optical waveguide).

JP 2001-135877 A JP-A-8-125279 JP-A-9-61652

  In the optical semiconductor device 110 shown in FIG. 1, the thickness of the optical waveguide layer 115 decreases rapidly from the junction J, so that a part of the light output from the active layer 114 is part of the optical waveguide layer 115, the upper cladding layer 117, and the like. Reflects at the interface. Therefore, the optical semiconductor element 110 shown in FIG. 1 has a problem that the loss of light in the spot size conversion unit 110B is large.

  The reason why the thickness of the optical waveguide layer 115 rapidly decreases as the distance from the joint J increases is due to the manufacturing method. A method for manufacturing the optical semiconductor element 110 shown in FIG. 1 will be described below with reference to FIGS.

  First, as shown in FIG. 2, a substrate 111, a lower cladding layer 113 disposed on the substrate 111, an active layer 114 disposed on the lower cladding layer 113, and an upper cladding disposed on the active layer 114. A stacked body 111a including the layer 116 is formed. FIG. 2B is a plan view of the stacked body 111a. 2A is a cross-sectional view taken along line Z1-Z1 of FIG. The substrate 111 is formed using InP, and the lower cladding layer 113 is also formed using InP. The active layer 114 has a multiple quantum well (MQW) structure containing InGaAsP. The upper cladding layer 116 is formed using InP.

  Next, as shown in FIG. 3, a selective growth mask 130 is formed on the stacked body 111a. The selective growth mask 130 has a function of growing a layer in a region not covered with the mask. The selective growth mask 130 has an opening 131. The opening 131 extends from the end in the longitudinal direction of the selective growth mask 130 toward the inside of the mask if the width is reduced. The selective growth mask 130 is formed using SiN. 3B is a plan view of the stacked body 111a, and FIG. 3A is a cross-sectional view taken along the line Z2-Z2 of FIG. 3B.

  Next, as shown in FIG. 4, using the selective growth mask 130, the portions of the upper cladding layer 116 and the active layer 114 located inside the opening 131 are removed by etching, and the lower cladding layer is removed from the opening 131. 113 is exposed. Note that, even if not located inside the opening 131, the portions of the upper cladding layer 116 and the active layer 114 that are not covered by the selective growth mask 130 are removed at the same time. 4B is a plan view, and FIG. 4A is a cross-sectional view taken along line Z3-Z3 in FIG. 4B.

  Next, as shown in FIG. 5, the optical waveguide layer 115 is formed in the opening 131 with the selective growth mask 130 left on the upper cladding layer 116. The optical waveguide layer 115 is formed using InGaAsP by, for example, metal organic chemical vapor deposition or molecular beam epitaxy. 5B is a plan view, and FIG. 5A is a cross-sectional view taken along line Z4-Z4 of FIG. 5B.

  The raw material of the optical waveguide layer 115 is not deposited on the selective growth mask 130 but moves from the selective growth mask 130 onto the lower cladding layer 113 exposed in the opening 131 and is deposited. Therefore, more raw material is deposited on the active layer 114 side of the opening 131 than at the open end of the opening 131. As a result, the optical waveguide layer 115 is formed so as to rapidly decrease in thickness as it moves away from the junction J with the active layer 114.

  The above is an explanation of the reason why the thickness of the optical waveguide layer 115 rapidly decreases as the distance from the joint portion J increases. As described above, when the thickness of the optical waveguide layer 115 is drastically reduced, reflection of light occurs and light coupling loss occurs.

  It is an object of the present specification to provide a method for manufacturing an optical semiconductor element having an optical waveguide layer that increases the spot size of light input from an active layer and reduces light reflection.

  According to an embodiment of the method for manufacturing an optical semiconductor element disclosed in the present specification, a laminate in which a substrate, a cladding layer disposed on the substrate, and an active layer disposed on the cladding layer are stacked. A first step of forming a selective growth mask on the selective growth mask, the selective growth mask having an opening extending inwardly from an edge of the selective growth mask, the opening having a width on the edge side; A first step having a wider inner width than the first step, a second step of removing the portion of the active layer located inside the opening using the selective growth mask, and the selective growth mask And a third step of forming an optical waveguide layer in the opening while remaining on the active layer.

  According to one embodiment of the method for manufacturing an optical semiconductor element disclosed in the present specification described above, an optical semiconductor element having an optical waveguide layer that enlarges the spot size of light input from the active layer and reduces light reflection is obtained. It is done.

  The objects and advantages of the invention will be realized and obtained by means of the elements and combinations particularly pointed out in the appended claims.

  Both the foregoing general description and the following detailed description are exemplary and explanatory and are not restrictive of the invention as claimed.

It is a figure which shows the optical semiconductor element of a prior art example. It is a figure explaining the process (the 1) which manufactures the optical semiconductor element of a prior art example. It is a figure explaining the process (the 2) which manufactures the optical semiconductor element of a prior art example. It is a figure explaining the process (the 3) which manufactures the optical semiconductor element of a prior art example. It is a figure explaining the process (the 4) which manufactures the optical semiconductor element of a prior art example. (A) is sectional drawing which shows 1st Embodiment of the optical semiconductor element disclosed by this specification, (B) is a top view. (A) is X2-X2 sectional view taken on the line which shows 1st Embodiment of the optical semiconductor element disclosed to this specification, (B) is X3-X3 sectional view taken on the line. It is a figure which shows the shape of a different selective growth mask. It is a figure which shows the relationship between the shape of a selective growth mask, and the thickness change of an optical waveguide layer. It is a figure explaining the process (the 1) of 1st Embodiment of the manufacturing method of the optical semiconductor element disclosed to this specification. It is a figure explaining the process (the 2) of 1st Embodiment of the manufacturing method of the optical semiconductor element disclosed to this specification. It is a figure explaining the process (the 3) of 1st Embodiment of the manufacturing method of the optical semiconductor element disclosed to this specification. It is a figure explaining the process (the 4) of 1st Embodiment of the manufacturing method of the optical semiconductor element disclosed to this specification. It is a figure explaining the process (the 5) of 1st Embodiment of the manufacturing method of the optical semiconductor element disclosed to this specification. It is a figure explaining the process (the 6) of 1st Embodiment of the manufacturing method of the optical semiconductor element disclosed to this specification. (A) is sectional drawing which shows 2nd Embodiment of the optical semiconductor element disclosed by this specification, (B) is a top view. (A) is a Y2-Y2 line sectional view showing a 2nd embodiment of an optical semiconductor device indicated to this specification, and is a Y3-Y3 line sectional view. It is a figure explaining the process (the 1) of 2nd Embodiment of the manufacturing method of the optical semiconductor element disclosed to this specification. It is a figure explaining the process (the 2) of 2nd Embodiment of the manufacturing method of the optical semiconductor element disclosed to this specification. It is a figure explaining the process (the 3) of 2nd Embodiment of the manufacturing method of the optical semiconductor element disclosed to this specification. It is a figure explaining the process (the 4) of 2nd Embodiment of the manufacturing method of the optical semiconductor element disclosed to this specification. It is a figure explaining the process (the 5) of 2nd Embodiment of the manufacturing method of the optical semiconductor element disclosed to this specification. It is a figure explaining the process (the 6) of 2nd Embodiment of the manufacturing method of the optical semiconductor element disclosed to this specification. It is a figure which shows the modification 1 of a selective growth mask. It is a figure which shows the modification 2 of a selective growth mask. It is a figure which shows the modification 3 of a selective growth mask. It is a figure which shows the selective growth mask used by the experiment example and the comparative experiment example. It is a figure which shows the thickness change of the optical waveguide layer of an Example and a comparative example.

  Hereinafter, a preferred first embodiment of an optical semiconductor device disclosed in this specification will be described with reference to the drawings. However, the technical scope of the present invention is not limited to these embodiments, but extends to the invention described in the claims and equivalents thereof.

  FIG. 6B is a plan view of the first embodiment of the optical semiconductor element disclosed in this specification, and FIG. 6A is a cross-sectional view taken along line X1-X1 in FIG. 7A is a cross-sectional view taken along line X2-X2 in FIG. 6B, and FIG. 7B is a cross-sectional view taken along line X3-X3 in FIG.

  The optical semiconductor element 10 of the present embodiment includes a light generation unit 10A that generates laser light and a spot size conversion unit 10B. The spot size conversion unit 10B receives light from the light generation unit 10A, and outputs light with an enlarged spot size from the end opposite to the light generation unit 10A. The spot size conversion unit 10B expands the spot size of the light output from the light generation unit 10A in the vertical direction and also in the horizontal direction.

  The light generation unit 10 </ b> A is disposed on the substrate 11, the second lower cladding layer 12 disposed on the substrate 11, and the second lower cladding layer 12, and has a higher refractive index than the second lower cladding layer 12. A lower cladding layer 13 is provided. An active layer 14 having a multilayer quantum dot layer is disposed on the first lower cladding layer 13. The shape of the quantum dot can be appropriately designed so that a gain is obtained at a predetermined wavelength (for example, a wavelength of 1300 nm). A first upper cladding layer 16 having a refractive index lower than that of the active layer 14 is disposed on the active layer 14, and a second refractive index lower than that of the first upper cladding layer 16 is disposed on the first upper cladding layer 16. An upper cladding layer 17 is disposed. A contact layer 18 is disposed on the second upper cladding layer 17.

  A first electrode layer 19 a is disposed on the contact layer 18, and a second electrode layer 19 b is disposed under the substrate 11. The substrate 11, the second lower cladding layer 12, the first lower cladding layer 13, the first upper cladding layer 16, the second upper cladding layer 17, the contact layer 18, and the second electrode layer 19 b It extends from the portion 10A to the spot size conversion portion 10B.

  As shown in FIG. 7A, in the light generation unit 10A, the stacked first upper cladding layer 16, second upper cladding layer 17, and contact layer 18 are narrower than the active layer 14 and have a mesa structure. It is formed.

  The spot size conversion unit 10 </ b> B has an optical waveguide layer 15 disposed on the first lower cladding layer 13. On the optical waveguide layer 15, a first upper clad layer 16, a second upper clad layer 17, and a contact layer 18 are disposed in order. The optical waveguide layer 15 and the active layer 14 are optically coupled by butt joint joining at the joint J.

  As shown in FIG. 7B, in the spot size converter 10B, the first upper clad layer 16, the second upper clad layer 17 and the contact layer 18 are narrower than the optical waveguide layer 15 and form a mesa structure. Is done.

  The optical waveguide layer 15 has the same thickness as that of the active layer 14 at the junction J, but the thickness decreases as the distance from the junction J increases. For example, the thickness of the optical waveguide layer 15 at the junction J can be 400 nm, and the thickness of the end of the optical waveguide layer 15 on the side opposite to the junction J can be 200 nm. That is, the thickness of the optical waveguide layer 15 decreases as the thickness in the direction in which light propagates increases away from the joint J.

  In the process of propagation of light from the active layer 14 to the optical waveguide layer 15, the thickness of the optical waveguide layer 15 is reduced and the effect of being confined in the optical waveguide layer 15 is weakened. . Therefore, the light propagating through the optical waveguide layer 15 is reduced in thickness as the optical waveguide layer 15 is thinned, and oozes out of the optical waveguide layer 15 in the thickness direction of the optical waveguide layer 15 to increase the spot size. I will do it.

  Here, since the refractive index of the second upper cladding layer 17 is smaller than that of the first upper cladding layer 16 and the refractive index of the second lower cladding layer 12 is smaller than that of the first lower cladding layer 13, Expansion beyond the first upper cladding layer 16 and the first lower cladding layer 13 is suppressed. Using such a configuration, the spot size in the thickness direction of the spot size conversion unit 10B is enlarged to a predetermined size.

  As shown in FIG. 6B, the optical waveguide layer 15 extends from the junction J toward the end opposite to the junction J while expanding in width. In the light propagating through the optical waveguide layer 15, the spot size in the in-plane direction of the optical waveguide layer 15 increases as the width of the optical waveguide layer 15 increases.

  In the optical semiconductor element 10, the rate of reduction of the thickness in the direction in which the light of the optical waveguide layer 15 propagates in the spot size conversion unit 10 </ b> B is gentler than that of the above-described conventional optical waveguide layer. Therefore, the amount of light output from the active layer 14 reflected at the interface between the optical waveguide layer 15 and the first upper cladding layer 16 is reduced, so that the light loss in the spot size conversion unit 10B is reduced. In addition, since the rate of reducing the thickness in the propagation direction is moderate, the conversion loss of the mode change of the propagating light expanded by the spot size conversion unit is reduced.

  The inventor of the present application has obtained knowledge for gradual reduction of the thickness of the optical waveguide layer 15 by devising the shape of the selective growth mask used when forming the optical waveguide layer 15. Hereinafter, the relationship between the shape of the selective growth mask and the thickness of the optical waveguide layer 15 will be described with reference to FIGS.

  FIG. 8 is a diagram showing the shapes of different selective growth masks.

  The selective growth mask M1 shown in FIG. 8A has a straight edge E. The selective growth mask M1 is disposed on the active layer 14, and the active layer 14 disposed on the first lower cladding layer 13 is exposed in a portion not covered with the selective growth mask M1.

  The selective growth mask M2 shown in FIG. 8B has an opening N2 extending inward from the edge E of the selective growth mask. The opening N2 has a certain width.

  A selective growth mask M3 shown in FIG. 8C is a mask used to form the optical waveguide layer 15 of the first embodiment described above disclosed in this specification. The selective growth mask M3 has an opening N3 extending inward from the edge E of the selective growth mask M3. The opening N3 has a wider inner width than a width on the edge E side. Specifically, the width of the opening N3 gradually increases from the edge E side of the selective growth mask M3 to the inside. The opening N3 has a first portion 31a extending from the edge E side inward with a predetermined width, and a second portion 31b extending inward from the first portion 31a. The width of the first portion 31a is narrower than the width of the opening N2 shown in FIG. 8B, and the width of the second portion 31b is wider than the width of the opening N2.

  Next, using each of the selective growth masks M1 to M3, the portion of the active layer 14 that is not covered with the mask is removed by etching, and the first lower cladding layer 13 is exposed. Next, an optical waveguide layer was formed by selective growth on the exposed first lower cladding layer 13 with the selective growth masks M1 to M3 left on the active layer. When the selective growth masks M2 and M3 are used, an optical waveguide layer is also formed in the openings N2 and N3 where the first lower cladding layer 13 is exposed.

  A curve C1 shown in FIG. 9 shows a change in the thickness of the optical waveguide layer in the direction along the chain line P1 in FIG. A curve C1 shows the thickness distribution of the optical waveguide layer in the direction away from the edge E of the selective growth mask M1. The origin of the horizontal axis indicates the position of the edge E of the selective growth mask M1.

  A curve C2 shown in FIG. 9 shows a change in the thickness of the optical waveguide layer in the direction along the chain line P2 in FIG. A curve C2 indicates the distribution of the thickness of the optical waveguide layer in the direction away from the edge G of the selective growth mask M2 in the opening N2. The origin of the horizontal axis indicates the position of the edge G in the opening N2.

  A curve C3 shown in FIG. 9 shows a change in the thickness of the optical waveguide layer in the direction along the chain line P3 in FIG. A curve C3 indicates the distribution of the thickness of the optical waveguide layer in the direction away from the edge G of the selective growth mask M3 in the opening N3. The origin of the horizontal axis indicates the position of the edge G in the opening N3.

  As shown by the curve C1 in FIG. 9, when the optical waveguide layer is formed using the selective growth mask M1, the thickness of the optical waveguide layer increases from the edge E of the selective growth mask M1 and an exponential function. Decrease. In the portion near the edge E of the selective growth mask M1, more raw material for the optical waveguide layer is supplied from the selective growth mask M2 than in the portion far from the edge E, so that the thickness of the optical waveguide layer becomes thicker. .

  As shown by the curve C2 in FIG. 9, when the optical waveguide layer is formed using the selective growth mask M2, the thickness of the optical waveguide layer in the opening N2 is as a whole in the case of FIG. 9A. Than to increase. This is because the material for the optical waveguide layer is supplied into the opening N2 from the portion of the selective growth mask M2 around the opening N2.

  As shown by the curve C3 in FIG. 9, when the optical waveguide layer is formed using the selective growth mask M3, the thickness of the optical waveguide layer in the opening N3 is close to the edge G of the selective growth mask M3. The portion is thinner than the curve C2, while the portion away from the edge G of the selective growth mask M3 is thicker than the curve C2. The reason for this will be described below.

  Since the second portion 31b of the opening N3 of the selective growth mask M3 is wider than the opening N2, the supply amount of the material from the selective growth mask is reduced with respect to the unit area of the second portion 31b. On the other hand, since the width of the first part 31a is narrower than that of the opening N2, the supply amount of the material from the selective growth mask is increased with respect to the unit area of the first part 31a.

  In this way, by using the selective growth mask M3, the growth rate of the optical waveguide layer on the edge G side of the mask is made higher than when the selective growth mask M2 is used. At the same time, the growth rate of the optical waveguide layer at a position away from the edge G of the mask is made lower than when the selective growth mask M2 is used. As a result, by using the selective growth mask M3, the rate of reduction in the thickness of the optical waveguide layer on the edge G side of the mask can be made slower than when the mask M1 or M2 is used.

  The opening N3 of the selective growth mask M3 generally has a pair of side edges F (see FIG. 8C) extending inward from the edge E of the selective growth mask. At least one side edge F of the pair of side edges F has a convex shape toward the inside of the opening N3, so that the thickness of the optical waveguide layer formed in the opening N3 is moderately reduced. It is preferable when forming so. The opening N3 shown in FIG. 8C is an example of a side edge having a stepped side edge F whose width expands inward, and having a convex shape toward the inside of the opening N3. . Further, the side edge F may have a shape that curves convexly toward the inside of the opening (see FIG. 8C).

  Next, a preferred method for manufacturing the above-described optical semiconductor device of the first embodiment will be described below with reference to the drawings.

  First, as shown in FIG. 10, a stacked body 11 a in which a second lower cladding layer 12, a first lower cladding layer 13, and an active layer 14 are stacked is formed on a substrate 11. The active layer 14 has a structure in which multiple quantum dot layers are stacked. 10B is a plan view of the stacked body 11a, and FIG. 10A is a cross-sectional view taken along line X4-X4 of FIG. 10B. In the future, a region where the light generation unit is formed is denoted as R1, and a region where the spot size conversion unit is formed is denoted as R2.

In this embodiment, a p-GaAs substrate having a (100) plane is used as the substrate 11. The second lower cladding layer 12 was formed using p-Al _0.3 Ga _0.7 As. The first lower cladding layer 13 was formed using p-Al _0.27 Ga _0.77 As. The Al composition ratio of the second lower cladding layer 12 was made larger than that of the first lower cladding layer 13, and the refractive index of the second lower cladding layer 12 was made smaller than that of the first lower cladding layer 13. The structure of the active layer 14 is formed by stacking eight quantum dot layers in which quantum dots formed using InAs are arranged in a barrier layer formed using GaAs. The thickness of the barrier layer was 40 nm. Above and below the eight-layered quantum dot layer structure, an SCH (Separate Confinement Heterostructure) layer formed using GaAs is disposed so as to sandwich the quantum dot layer structure. The thickness of each quantum dot layer was 40 nm, the thickness of each SCH layer was 40 nm, and the thickness of the active layer 14 as a whole was 400 nm. Each layer is formed using, for example, a molecular beam epitaxy method (MBE method). The oscillation wavelength of the active layer 14 was 1300 nm.

Next, as shown in FIG. 11, a selective growth mask 30 is formed on the active layer 14 of the stacked body 11a by using a lithography method and an etching method. FIG. 11B is a plan view of the stacked body 11a, and FIG. 11A is a cross-sectional view taken along line X5-X5 of FIG. The selective growth mask 30 is formed using, for example, SiO ₂ or SiN. The selective growth mask 30 has an opening 31. The shape of the opening 31 is the same as the opening shown in FIG. The dimension of the selective growth mask 30 can be set to 800 μm in the longitudinal direction and 400 μm in the width direction. The length L1 of the first part 31a of the opening 31 can be 50 μm, the width L2 can be 10 μm, the length L3 of the second part 31b can be 50 μm, and the width L4 can be 15 μm. The junction J between the active layer 14 and the optical waveguide layer 15 is formed at a position that coincides with the boundary between the region R1 and the region R2, and is located at a position of 100 μm inward from the edge E of the selective growth mask 30.

  Note that the selective growth mask 30 in the region R1 may be formed narrow enough to cover a portion where a mesa structure will be formed in the future, as indicated by a chain line in FIG. Since the portion of the selective growth mask 30 opposite to the region R2 has little influence on the selective growth of the optical waveguide layer in the opening 31, the area may be reduced. Further, by making the selective growth mask 30 small, it is possible to suppress a part of the raw material for forming the optical waveguide layer from being deposited on the mask.

  Next, as shown in FIG. 12, the active layer 14 not covered with the selective growth mask 30 including the portion of the active layer 14 located inside the opening 31 is removed using the selective growth mask 30. The first lower cladding layer 13 is exposed. 12B is a plan view, and FIG. 12A is a cross-sectional view taken along line X6-X6 of FIG.

  In the etching for removing the active layer 14, it is preferable to use an etchant having an etching selectivity that has a high etching rate for the active layer 14 and a low etching rate for the first lower cladding layer 13. As an etchant having a high etching rate with respect to the active layer 14 mainly composed of GaAs and InAs and a slow etching rate with respect to the first lower cladding layer 13 mainly composed of AlGaAs, specifically, a citric acid / hydrogen peroxide solution system is used. Etchant. In this specification, citric acid means a mixture of pure water and citric acid powder in citric acid: pure water = 1 g: 1 cc. In this embodiment, an etchant of citric acid: hydrogen peroxide = 4: 1 was used.

  Next, as shown in FIG. 13, the optical waveguide layer 15 is formed by selective growth in the opening 31 with the selective growth mask 30 left on the active layer 14. FIG. 13B is a plan view, and FIG. 13A is a cross-sectional view taken along line X7-X7 in FIG. 13B. In the present embodiment, the optical waveguide layer 15 is formed using GaAs by metal organic vapor phase epitaxy (MOVPE). The optical waveguide layer 15 in the opening 31 is formed so that its thickness gradually decreases in the vicinity of the joint J as shown by a curve C3 in FIG. A junction J in which the active layer 14 and the optical waveguide layer 15 are bad-joint is formed at the boundary between the region R1 and the region R2. Then, the selective growth mask 30 is removed from the active layer 14.

Next, as shown in FIG. 14, the first upper cladding layer 16, the second upper cladding layer 17, and the contact are formed so as to cover the active layer 14, the optical waveguide layer 15, and the first lower cladding layer 13. Layers 18 are formed in order. 14B is a plan view, and FIG. 14A is a cross-sectional view taken along line X8-X8 in FIG. 14B. The second upper clad layer 17 was formed using p-Al _0.3 Ga _0.7 As. The first upper cladding layer 16 was formed using p-Al _0.27 Ga _0.77 As. The Al composition ratio of the second upper cladding layer 17 was made larger than that of the first upper cladding layer 16, and the refractive index of the second lower cladding layer 12 was made smaller than that of the first lower cladding layer 13. The contact layer 18 was formed using n-GaAs. Each layer is formed using, for example, a molecular beam epitaxy method (MBE method) or a metal organic chemical vapor deposition method (MOVPE method).

  Next, as shown in FIG. 15, a mask 32 is formed on the contact layer 18. FIG. 15B is a plan view, and FIG. 15A is a cross-sectional view taken along line X9-X9 in FIG. 15B. The mask 32 extends in the longitudinal direction with the same width in the region R1, and extends in the longitudinal direction while expanding in width in the region R2. Then, using the mask 32, the portions from the contact layer 18 to the first upper cladding layer 16 are etched, and the mesa structure of the light generation unit 10A and the spot size conversion unit 10B is formed. The periphery of the mesa structure is surrounded by the atmosphere, and a ridge waveguide structure is formed. Then, the first electrode layer 19a is formed on the contact layer 18 in the region R1, and the second electrode layer 19b is formed under the substrate 11 in the region R1 from the region R1, whereby the optical semiconductor element 10 is obtained. The contact layer 18 in the region R2 may be removed.

  According to the manufacturing method of the optical semiconductor device of the present embodiment described above, the optical semiconductor device 10 having the optical waveguide layer 15 that enlarges the spot size of light input from the active layer 14 and reduces the reflection of light is obtained. Specifically, since the thickness of the optical waveguide layer 15 in a predetermined range from the junction J is gradually reduced, the light input from the active layer 14 is used as the interface between the optical waveguide layer 15 and the first upper cladding layer 16. The amount of reflection from can be reduced.

  In the manufacturing method of the first embodiment described above, the width of the opening 31 of the selective growth mask 30 has been changed in two stages, but the change in the width of the opening 31 may be changed in three or more stages. . In the manufacturing method of the first embodiment described above, the p-GaAs substrate is used, but an n-GaAs substrate may be used. In this case, it is preferable that the upper cladding layer and the lower cladding layer are formed using p-AlGaAs, and the contact layer is formed using p-GaAs. Furthermore, it is preferable to dispose a buffer layer between the substrate 11 and the second lower cladding layer 12 from the viewpoint of improving the crystal quality of the lower cladding layer.

  Next, a second embodiment of the above-described optical semiconductor element will be described below with reference to FIGS. For points that are not particularly described in the other embodiments, the description in detail regarding the first embodiment is applied as appropriate. Moreover, the same code | symbol is attached | subjected to the same component.

  FIG. 16B is a plan view of a second embodiment of the optical semiconductor element disclosed in this specification, and FIG. 16A is a cross-sectional view taken along line Y1-Y1 of FIG. 17A is a cross-sectional view taken along line Y2-Y2 of FIG. 16B, and FIG. 17B is a cross-sectional view taken along line Y3-Y3 of FIG.

  Similarly to the first embodiment described above, the optical semiconductor element 10 of the present embodiment also includes a light generation unit 10A that generates laser light and a spot size conversion unit 10B. The spot size conversion unit 10B receives light from the light generation unit 10A, and outputs light with an enlarged spot size from the end opposite to the light generation unit 10A. The spot size conversion unit 10B expands the spot size of the light output from the light generation unit 10A in the vertical direction and also in the horizontal direction.

  In the optical waveguide layer 15 of the spot size conversion unit 10B of the present embodiment, the rate of reduction of the thickness in the vicinity of the joint J is further gradual than in the first embodiment described above. This reason will be described later in the manufacturing method.

  The light generation unit 10 </ b> A includes a substrate 11 and a first lower clad layer 13 disposed on the substrate 11. An active layer 14 having a multiple quantum well (MQW) structure is disposed on the first lower cladding layer 13. As shown in FIG. 17A, the first lower cladding layer 13 has a central portion in the width direction protruding upward, and the active layer 14 is a protruding portion in the center of the first lower cladding layer 13. Placed on top. The multiple quantum well structure can be appropriately designed so as to obtain a gain at a predetermined wavelength. A first upper cladding layer 16 is disposed on the active layer 14, and a second upper cladding layer 17 is disposed on the first upper cladding layer 16. A contact layer 18 is disposed on the second upper cladding layer 17. In the optical semiconductor device 10 of the present embodiment, the second lower cladding layer is not disposed.

  A first electrode layer 19 a is disposed on the contact layer 18, and a second electrode layer 19 b is disposed under the substrate 11. The substrate 11, the first lower cladding layer 13, the second upper cladding layer 17, the contact layer 18, and the second electrode layer 19b extend from the light generation unit 10A to the spot size conversion unit 10B.

  As shown in FIG. 17A, in the light generation unit 10A, the stacked active layer 14, first upper clad layer 16, second upper clad layer 17, and contact layer 18 are larger than the first lower clad layer 13. The width is narrow and a mesa structure is formed. A buried layer 21 is disposed around the mesa structure.

  The spot size conversion unit 10 </ b> B has an optical waveguide layer 15 disposed on the first lower cladding layer 13. As shown in FIG. 17B, the first lower cladding layer 13 has a central portion in the width direction protruding upward, and the optical waveguide layer 15 protrudes at the center of the first lower cladding layer 13. Placed on the part. On the optical waveguide layer 15, a third upper cladding layer 20, a second upper cladding layer 17, and a contact layer 18 are disposed in order. The optical waveguide layer 15 and the active layer 14 are optically coupled by butt joint joining at the joint J.

  As shown in FIG. 17B, in the spot size conversion unit 10B, the optical waveguide layer 15, the third upper cladding layer 20, the second upper cladding layer 17, and the contact layer 18 are wider than the first lower cladding layer 13. Is narrow and a mesa structure is formed. A buried layer 21 is disposed around the mesa structure. The buried layer 21 extends from the light generation unit 10A to the spot size conversion unit 10B.

  The optical waveguide layer 15 has the same thickness as that of the active layer 14 at the junction J with the active layer 14, but the thickness decreases as the distance from the junction J increases. For example, the thickness of the optical waveguide layer 15 at the junction J can be set to 400 nm, and the thickness of the end of the optical waveguide layer 15 on the side opposite to the junction J can be set to 100 nm. That is, the thickness of the optical waveguide layer 15 decreases as the thickness in the direction in which light propagates increases away from the joint J.

  In the process of propagation of light from the active layer 14 to the optical waveguide layer 15, the thickness of the optical waveguide layer 15 is reduced and the effect of being confined in the optical waveguide layer 15 is weakened. . Therefore, the light propagating through the optical waveguide layer 15 is reduced in thickness as the optical waveguide layer 15 is thinned, and oozes out of the optical waveguide layer 15 in the thickness direction of the optical waveguide layer 15 to increase the spot size. I will do it.

  In addition, as shown in FIG. 16B, the optical waveguide layer 15 extends from the joint J toward the end opposite to the joint J while being reduced in width. The light propagating through the optical waveguide layer 15 has an effect that the spot size in the layer direction of the optical waveguide layer 15 is confined in the optical waveguide layer 15 as the width of the optical waveguide layer 15 decreases. For this reason, the light propagating through the optical waveguide layer 15 is reduced in the width of the optical waveguide layer 15 and oozes out to the buried layer 21 outside the optical waveguide layer 15 in the layer direction of the optical waveguide layer 15, resulting in a spot size. Will expand.

  Next, a preferred method for manufacturing the optical semiconductor element 10 of the second embodiment will be described below with reference to the drawings.

  First, as illustrated in FIG. 18, a stacked body 11 a in which a first lower cladding layer 13, an active layer 14, and a first upper cladding layer 16 are stacked is formed on a substrate 11. The active layer 14 has a multiple quantum well structure. 18B is a plan view of the stacked body 11a, and FIG. 18A is a cross-sectional view taken along line Y4-Y4 of FIG. 18B. In the future, a region where the light generation unit is formed is denoted as R1, and a region where the spot size conversion unit is formed is denoted as R2.

  In this embodiment, an n-InP substrate having a (100) plane was used as the substrate 11. The first lower cladding layer 13 was formed using n-InP. The multi-quantum well structure of the active layer 14 includes a quantum well layer having a barrier layer formed using i-InGaAsP having a thickness of 10 nm and a well layer formed using i-InGaAsP having a thickness of 5 nm. 6 layers are laminated. Above and below the multiple quantum well structure, an SCH (Separate Confinement Heterostructure) layer formed using i-InGaAsP is disposed so as to sandwich the multiple quantum well structure. The thickness of each SCH layer was 55 nm, and the thickness of the active layer 14 as a whole was 400 nm. Each layer is formed using, for example, the MOVPE method.

Next, as shown in FIG. 19, a selective growth mask 30 is formed on the first upper cladding layer 16 on the active layer 14 of the stacked body 11a by using a lithography method and an etching method. FIG. 19B is a plan view of the stacked body 11a, and FIG. 19A is a cross-sectional view taken along line Y5-Y5 of FIG. 19B. The selective growth mask 30 is formed using, for example, SiO ₂ or SiN. The selective growth mask 30 has an opening 31. The opening 31 includes a first portion 31a extending from the edge E side inward with a predetermined width, and a second portion 31c extending further inward while the width continuously extends linearly from the first portion 31a. Have.

  By adjusting the rate of change in the width of the second portion 31 b of the opening 31, the rate of change in the thickness of the optical waveguide layer 15 in the vicinity of the joint J can be adjusted. For example, the thickness of the optical waveguide layer 15 in a predetermined region near the joint portion J can be made substantially constant by adjusting the rate of change in the width of the second portion 31b.

  The dimension of the selective growth mask 30 can be 600 μm in the length in the longitudinal direction and 400 μm in the length in the width direction. The length L1 of the first portion 31a of the opening 31 can be 50 μm, the width L2 can be 10 μm, the length L3 of the second portion 31b can be 50 μm, and the width L4 can be 20 μm. The joint J between the active layer 14 and the optical waveguide layer 15 is formed at the boundary between the region R1 and the region R2, and is located at a position of 100 μm inward from the edge E of the selective growth mask 30.

  Next, as illustrated in FIG. 20, the selective growth mask 30 is used to cover the first upper cladding layer 16 and the active layer 14 located inside the opening 31, and cover the selective growth mask 30. The first upper cladding layer 16 and the active layer 14 that are not present are removed, and the first lower cladding layer 13 is exposed. 20B is a plan view, and FIG. 20A is a cross-sectional view taken along line Y6-Y6 of FIG.

  In the present embodiment, first, the first upper cladding layer 16 is etched using an etchant having a high etching selectivity with respect to the first upper cladding layer 16, and then the etching selectivity with respect to the active layer 14 is high. The active layer 14 was etched using an etchant. Specifically, hydrochloric acid was used as an etchant having high etching selectivity with respect to the first upper cladding layer 16 containing InP as a main component. In addition, as an etchant having high etching selectivity with respect to the active layer 14 containing InGaAsP as a main component, a sulfuric acid: hydrogen peroxide aqueous etchant was used.

  Next, as shown in FIG. 21, with the selective growth mask 30 left on the first upper cladding layer 16 positioned on the active layer 14, the optical waveguide layer 15 is formed in the opening 31 by selective growth. It is formed. FIG. 21B is a plan view, and FIG. 21A is a cross-sectional view taken along line Y7-Y7 in FIG. In the present embodiment, the optical waveguide layer 15 is formed using InGaAsP by metal organic vapor phase epitaxy (MOVPE). Since the width of the second portion 31b of the selective growth mask 30 of this embodiment is continuously widened, the rate of reduction in the thickness of the optical waveguide layer 15 in the vicinity of the joint portion J is the above-described first embodiment. It becomes more gentle than the optical waveguide layer. A junction J in which the active layer 14 and the optical waveguide layer 15 are bad-joint is formed at the boundary between the region R1 and the region R2. Then, the selective growth mask 30 is removed from the active layer 14.

  Here, due to the difference in diffusion length of the raw materials (for example, In and Ga) forming the optical waveguide layer 15, the composition and the number of lattices of the optical waveguide layer 15 may not be uniform. Therefore, from the viewpoint of improving the crystallinity of the optical waveguide layer 15, the film formation conditions for the optical waveguide layer 15 are selected so that the optical waveguide layer 15 is lattice-matched with the first lower cladding layer 13 in the vicinity of the junction J. It is preferable to do. Since the critical film thickness is determined by the product of strain and film thickness, the optical waveguide layer 15 is thick in the vicinity of the junction J, so that the optical waveguide layer 15 and the first lower cladding layer 13 are lattice-matched. It is preferable to make it. On the other hand, in the region away from the junction J, the thickness of the optical waveguide layer 15 is thin, so that the optical waveguide layer 15 and the first lower cladding layer 13 do not need to be lattice-matched.

  Next, the third upper cladding layer 20 is formed by selective growth on the optical waveguide layer 15 with the selective growth mask 30 left on the first upper cladding layer 16 positioned on the active layer 14. . In the present embodiment, the third upper cladding layer 20 is formed using i-InP using a metal organic chemical vapor deposition (MOVPE) method.

  Then, as shown in FIG. 22, after the selective growth mask 30 is removed from the active layer 14, the second upper cladding layer 17 and the contact layer are formed on the first upper cladding layer 16 and the third upper cladding layer 20. 18 are formed in order. 22B is a plan view, and FIG. 22A is a cross-sectional view taken along line Y8-Y8 in FIG. 22B. The second upper cladding layer 17 was formed using p-InP. The contact layer 18 was formed using p-InGaAs. Each layer is formed using, for example, a metal organic chemical vapor deposition (MOVPE) method.

  Next, as shown in FIG. 23, a mask 32 is formed on the contact layer 18. FIG. 23B is a plan view, and FIG. 23A is a cross-sectional view taken along line Y9-Y9 in FIG. The mask 32 extends with the same width in the longitudinal direction in the region R1, and extends while narrowing in the longitudinal direction in the region R2. Then, using the mask 32, the contact layer 18 to the surface portion of the first lower cladding layer 13 are etched to form the mesa structure of the light generation unit 10A and the spot size conversion unit 10B. Note that as the shape of the mask 32 illustrated in FIG. 23B, a shape extending while expanding in the longitudinal direction may be used as illustrated in FIG. 15B.

  Then, the embedded layer 21 is formed so as to embed the mesa structure of the light generation unit 10A and the spot size conversion unit 10B. In the present embodiment, the buried layer 21 is formed using InP to which Fe is added by the MOVPE method. Fe in the InP layer forms a deep acceptor level in the band gap and traps electrons, so that it becomes a semi-insulating semiconductor layer, and the InP layer to which Fe is added has high resistance. In this way, a buried waveguide structure having a high resistance is formed.

  Then, the first electrode layer 19a is formed on the contact layer 18 in the region R1, and the second electrode layer 19b is formed under the substrate 11 in the region R1 from the region R1, whereby the optical semiconductor element 10 is obtained.

  According to the method for manufacturing an optical semiconductor element of the present embodiment described above, the optical semiconductor element 10 having the optical waveguide layer 15 that further reduces reflection of light input from the active layer 14 is obtained. In the optical semiconductor device of the present embodiment, the rate of reduction of the thickness of the optical waveguide layer 15 in the vicinity of the joint J is further gradual than the optical waveguide layer of the first embodiment described above. Therefore, the amount of light input from the active layer 14 reflected from the interface between the optical waveguide layer 15 and the third upper cladding layer 20 can be further reduced.

  Next, modifications of the selective growth mask used in the above-described method for manufacturing an optical semiconductor element disclosed in this specification will be described below with reference to the drawings.

  FIG. 24 is a diagram showing a first modification of the selective growth mask.

  The selective growth mask 30 of Modification 1 has a narrow portion 30a whose width is narrowed. It is preferable that the narrow portion 30 a is arranged with respect to the opening 31 in a direction intersecting the direction in which the opening 31 extends inwardly from the edge E of the selective growth mask 30. By providing the narrow portion 30a, the area of the selective growth mask 30 is partially reduced, and the amount of the raw material of the optical waveguide layer supplied into the opening 31 located near the narrow portion 30a is reduced. Thus, the change in the thickness of the optical waveguide layer in the vicinity of the junction can be moderated.

  In FIG. 24, the narrow portion 30b of another form is indicated by a chain line. The narrow portion 30b is formed so that the width of the selective growth mask 30 is continuously narrowed.

  Next, modified example 2 of the selective growth mask will be described below.

  FIG. 25 is a diagram showing a second modification of the selective growth mask.

  In the description of the manufacturing method of the first embodiment and the second embodiment described above, it has been described that individual optical semiconductor elements are manufactured. In practice, a plurality of optical semiconductor elements may be simultaneously formed on a single large substrate. FIG. 25 shows a state in which a plurality of selective growth masks 30 used in such a case are arranged side by side. A portion shown in a frame T in FIG. 25 is a selective growth mask for forming one optical semiconductor element.

  In the example shown in FIG. 25, a plurality of selective growth masks 30 are combined and arranged so that the regions R1 face each other. A plurality of selective growth masks 30 may be combined and arranged so that the regions R2 face each other.

  Next, Modification 3 of the selective growth mask will be described below.

  FIG. 26 is a diagram showing a third modification of the selective growth mask.

  This modification is different from Modification 2 described above in that the regions R2 of the selective growth masks 30 adjacent in the vertical direction are combined. Other structures are the same as in the second modification.

  Hereinafter, the manufacturing method of the optical semiconductor element disclosed in this specification will be further described with reference to examples. However, the scope of the present invention is not limited to such examples.

[Example]
First, a laminated body in which a lower clad layer and an active layer were laminated was formed on a substrate. As the substrate 11, a p-GaAs substrate having a (100) plane was used. The lower cladding layer was formed using p-Al _0.27 Ga _0.73 As, and the thickness was 2.5 μm. The structure having the multilayer quantum dot layer of the active layer is formed by stacking eight quantum dot layers in which quantum dots formed using InAs are arranged in a barrier layer formed using GaAs. . The thickness of the barrier layer was 40 nm. Above and below the multiple quantum well structure, an SCH (Separate Confinement Heterostructure) layer formed using GaAs is disposed so as to sandwich a structure having multiple quantum dot layers. The thickness of each quantum dot layer was 40 nm, the thickness of each SCH layer was 40 nm, and the thickness of the active layer as a whole was 400 nm.

Next, a selective growth mask was formed on the active layer of the stacked body using a lithography method and an etching method. The selective growth mask was formed using, for example, SiO ₂ . The shape of the selective growth mask is shown in FIG. The shape of the opening 31 shown in FIG. 27A is the same as the opening shown in FIG. The dimensions of the selective growth mask 30 were 400 μm in the longitudinal direction and 400 μm in the width direction. The length L1 of the first part 31a of the opening 31 was 50 μm, the width L2 was 10 μm, the length L3 of the second part 31b was 50 μm, and the width L4 was 15 μm.

  Next, using the selective growth mask 30, the active layer not covered by the selective growth mask 30 including the portion of the active layer located inside the opening 31 was removed, and the lower cladding layer was exposed. . The active layer was removed by etching using an etchant of citric acid: hydrogen peroxide solution = 4: 1.

  Next, an optical waveguide layer was formed in the opening 31 with the selective growth mask 30 left on the active layer.

[Comparative example]
A comparative example was obtained in the same manner as the above-described example except that the selective growth mask 230 shown in FIG. 27B was used as the selective growth mask. The selective growth mask 230 has an opening 231 extending inward from the edge of the selective growth mask. The opening 231 has a certain width. The dimensions of the selective growth mask 230 were 400 μm in the longitudinal direction and 400 μm in the width direction. The length L5 of the opening 231 was 100 μm and the width was 10 μm.

  FIG. 28 is a diagram showing a change in the thickness of the optical waveguide layer of the example and the comparative example. A curve D1 shows a change in the thickness of the optical waveguide layer in the direction along the chain line Q1 in FIG. A curve D2 shows a change in the thickness of the optical waveguide layer in the direction along the chain line Q1 in FIG. The horizontal axis in FIG. 28 indicates the distance from the joint.

  According to the curve D1 of the example, the thickness of the optical waveguide layer is 400 nm which is the same as that of the active layer at the junction with the active layer, and is approximately 400 nm when the distance from the junction is about 50 μm. The thickness is maintained, and then the thickness gradually decreases.

  According to the curve D2 of the comparative example, the thickness of the optical waveguide layer decreases almost exponentially as the distance from the joint portion increases.

  In the present invention, the method of manufacturing the optical semiconductor element of the above-described embodiment can be appropriately changed without departing from the gist of the present invention. In addition, the configuration requirements of one embodiment can be applied as appropriate to other embodiments or modifications.

  For example, in each of the embodiments described above, the active layer has a multilayer quantum dot layer structure or a multiple quantum well structure, but the active layer has a multilayer quantum dot layer structure or a multiple quantum well structure. Other light emitting structures may be used.

  In each of the above-described embodiments, the light generation unit outputs laser light. However, the light generation unit may be an optical amplification unit that inputs light from the spot size conversion unit and amplifies the light. Good. Further, the light generation unit may be a light detection unit that receives light from the spot size conversion unit and detects light. In these cases, the spot size conversion unit reduces the spot size of light input from the outside, and inputs the reduced light spot size to the light amplification unit or the light detection unit.

  The material for forming the active layer or the optical waveguide layer described above is not particularly limited. For example, AlGaInAs, GaInNAs, or InGaAs may be used as a material for forming the active layer or the optical waveguide layer.

  All examples and conditional words mentioned herein are intended for educational purposes to help the reader deepen and understand the inventions and concepts contributed by the inventor. All examples and conditional words mentioned herein are to be construed without limitation to such specifically stated examples and conditions. Also, such exemplary mechanisms in the specification are not related to showing the superiority and inferiority of the present invention. While embodiments of the present invention have been described in detail, it should be understood that various changes, substitutions or modifications can be made without departing from the spirit and scope of the invention.

DESCRIPTION OF SYMBOLS 10 Optical semiconductor element 10A Light generation part 10B Spot size conversion part 11 Board | substrate 12 2nd lower clad layer 13 1st lower clad layer 14 Active layer 15 Optical waveguide layer 16 1st upper clad layer 17 2nd upper clad layer 18 Contact layer 19a First electrode layer 19b Second electrode layer 20 Third upper cladding layer 21 Buried layer M1, M2, M3 selective growth mask N2, N3 opening C1, C2, C3 thickness curve R1 first region R2 second region 30 selection Growth mask 30a Narrow part 31 Opening part 31a First part 31b, 31c Second part 33 Mask T Element region

Claims (7)

The selective growth mask is a first step of forming a selective growth mask on a laminated body in which a substrate, a clad layer disposed on the substrate, and an active layer disposed on the clad layer are laminated. Has an opening extending inwardly from an edge of the selective growth mask, and the opening has a first step in which the inner width is wider than the width on the edge side;
A second step of removing a portion of the active layer located inside the opening by using the selective growth mask;
A third step of forming an optical waveguide layer in the opening while leaving the selective growth mask on the active layer;
A method of manufacturing an optical semiconductor device comprising:   2. The method of manufacturing an optical semiconductor device according to claim 1, wherein a width of the opening is gradually increased from the edge side of the selective growth mask toward the inside.   2. The method of manufacturing an optical semiconductor device according to claim 1, wherein the width of the opening continuously increases from the edge side of the selective growth mask toward the inside.   4. The method of manufacturing an optical semiconductor element according to claim 1, wherein the opening has a constant width at a predetermined distance from the edge of the selective growth mask toward the inside. 5. The opening has a pair of side edges extending inwardly from the edge of the selective growth mask;
5. The method of manufacturing an optical semiconductor element according to claim 1, wherein at least one of the side edges has a shape protruding toward the inside of the opening.   The method for manufacturing an optical semiconductor element according to claim 1, wherein the selective growth mask has a narrow portion with a narrow width.   The optical semiconductor element according to claim 6, wherein the narrow portion is disposed in a direction intersecting a direction in which the opening extends from the edge of the selective growth mask toward the inside with respect to the opening. Manufacturing method.

Images