JP2009176908A - Semiconductor optical device and method of manufacturing the same - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a semiconductor optical device that surely suppresses the deterioration in device characteristics. <P>SOLUTION: A semiconductor laser device has a semiconductor layer (an active layer 4) formed with a waveguide area (a conduction area A') extending in one direction. There are arranged below the semiconductor layer (the active layer 4) a plurality of planar-defect generation derivatives 10 apart at intervals in order to form each planar defect 15 in areas excluding the waveguide area (the conduction area A') of the semiconductor layer (the active layer 4). Each planar-defect generation derivative 10 and the waveguide area (the conduction area A') are arranged spaced apart at intervals in a plane view from the surface side of the semiconductor layer (the active layer 4). The plurality of planar-defect generation derivatives 10 are scattered along in the extending direction of the waveguide area (the conduction area A'). <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a semiconductor optical device and a method for manufacturing the same.

  Among III-V nitride semiconductor light-emitting devices based on gallium nitride (GaN), semiconductor lasers emitting blue-violet light have been put into practical use as next-generation high-density optical disk light sources, and the market will expand further in the future. It is expected to go. On the other hand, this material system can emit light in the visible wavelength region from blue to red by controlling the indium composition of the indium gallium nitrogen (InGaN) active layer. In recent years, there has been an increasing demand for the realization of semiconductor lasers that oscillate with visible light.

  FIG. 12 shows a cross-sectional structure of a general ridge stripe type nitride-based semiconductor laser device fabricated on a GaN substrate 201. In this element, an n-type AlGaN cladding layer (lower AlGaN cladding layer) 202, a lower optical waveguide layer 203, an InGaN / GaN quantum well active layer 204, an upper optical waveguide layer 205, a p-type are formed on an n-type GaN substrate 201. An AlGaN cladding layer (upper AlGaN cladding layer) 206 and a p-type GaN contact layer 207 are sequentially stacked. The p-type AlGaN cladding layer 206 is processed into a ridge stripe shape, and a p-type electrode 208 is formed on the ridge. An n-type electrode 212 is formed on the back surface of the GaN substrate 201.

In the device shown in FIG. 12, in order to obtain light emission in a pure blue band having a wavelength of about 450 nm, the indium composition of the quantum well active layer 204 needs to be about 0.2 or more. Therefore, a higher indium composition is required.
However, it is known that there are some difficulties in increasing the indium composition to increase the wavelength in a laser device having an InGaN quantum well as an active layer.

  Since the difference in interatomic distance between InN and GaN is large, phase separation is likely to occur in the InGaN material system, and an indium composition distribution is formed in the plane. This not only widens the emission spectrum and increases the wavelength shift associated with current injection, but also increases the risk of introducing local defects due to increased local strain even if the entire layer is below the strain criticality. As a result, the light emission characteristics deteriorate.

  Furthermore, in the crystal growth of the InGaN layer, since re-evaporation of indium becomes remarkable when the growth temperature becomes high, it must be grown at a low growth temperature of about 800 ° C., and it is difficult to obtain a good quality crystal. It is. Furthermore, it is known that when the cladding layer above the active layer is grown at a higher temperature, the crystals of the active layer grown at a lower temperature deteriorate. In addition, it is considered that all of these effects become conspicuous as the wavelength increases, that is, as the indium composition increases.

The inventors of the present invention made a prototype of a semiconductor laser element that oscillates in the blue band and investigated in detail the cause of the deterioration of the light emission characteristics.
And in the element which deteriorated, it discovered that the surface defect which is an aggregate | assembly of the dislocation extended in the range of several microns or more in the surface of an active layer was formed. As shown in FIG. 13, it has been found that the laser characteristics are greatly deteriorated when such a surface defect 221 and the laser energized region 220 overlap.

  Further, it has been confirmed that such a plane defect does not occur immediately after the growth of the quantum well active layer 204 but occurs by performing high-temperature growth or a process of about 1000 ° C. or higher after crystal growth. It has been found that in the device fabrication method, it is difficult to avoid crystal degradation of the active layer as described above.

In summary, when the indium composition of the quantum well active layer 204 is high, a composition distribution due to phase separation occurs during crystal growth, and in some cases, micro defects such as misfit dislocations in a relatively high region of the indium composition. Will occur. Furthermore, by performing high-temperature processes such as the growth of the upper cladding layer after the active layer growth, strain relaxation and InN dissociation occur from such micro defects as the starting point, and the peripheral crystal is deteriorated over a wide range to cause surface defects. It is speculated that there is a process such as
In addition, although the examination result about the semiconductor light-emitting device which has an InGaN / GaN quantum well active layer was described here, also in other types of active layers, micro defects, such as misfit dislocation, occurred, and it became a plane defect in a wide range. It is thought that this has a great influence on the laser characteristics.
Further, not only the semiconductor light emitting element but also the semiconductor optical element, a surface defect may occur in the waveguide length region, which may greatly affect the element characteristics.

  As a method for preventing the deterioration of device characteristics, particularly a method for preventing the deterioration of the active layer of a semiconductor light emitting device, a method for limiting the growth temperature and growth time of the upper layer, a semiconductor layer containing aluminum with higher stability, and the like. Patent Documents 2 and 3 propose methods for growing on the active layer.

In addition, Patent Document 1 discloses a method of providing a defect introduction region having a higher dislocation density in a lower layer than the active layer in order to alleviate distortion of the active layer and improve reliability.
Furthermore, there are Patent Documents 4 and 5 and Non-Patent Document 1 as background art of the present invention.

JP 2006-24713 A JP 10-335700 A JP-A-10-144612 JP 2001-176809 A WO2005 / 022620 J. W. Matthews and A. E. Blakeslee, Journal of Crystal Growth, 27, 118-125 (1974).

  However, it is considered difficult to reliably suppress the deterioration of device characteristics by these solutions.

  The objective of this invention is providing the semiconductor optical element which can suppress deterioration reliably.

According to the present invention, in a semiconductor optical device having a semiconductor layer in which a waveguide region extending in one direction is formed, the semiconductor layer has a region below the semiconductor layer other than the waveguide region. A plurality of surface defect generation derivatives for forming a surface defect are arranged apart from each other, and each of the surface defect generation derivatives and the waveguide region are arranged apart from each other in plan view from the semiconductor layer surface side. There is provided a semiconductor optical device in which a plurality of the surface defect generating derivatives are scattered along the extending direction of the waveguide region.
According to the present invention, in the method of manufacturing a semiconductor optical device having a semiconductor layer in which a waveguide region extending in one direction is formed, a step of arranging a plurality of surface defect generating derivatives, and the surface defect generating derivative Forming the semiconductor layer thereon, and forming a surface defect in a region other than the waveguide region of the semiconductor layer by the surface defect generation derivative, A semiconductor optical device arranged so that each of the surface defect-generating derivatives and the waveguide region are separated from each other along the extending direction of the waveguide region in plan view from the semiconductor layer surface side. A manufacturing method can also be provided.

  ADVANTAGE OF THE INVENTION According to this invention, the manufacturing method of the semiconductor optical element which can suppress deterioration of an element reliably, and a semiconductor optical element is provided.

  Hereinafter, preferred embodiments of a semiconductor optical device and a manufacturing method thereof will be described in detail with reference to the drawings. In all the drawings, the same reference numerals are given to the same components, and the description will be omitted as appropriate. Further, the drawings used here are schematic views, and are not based on actual dimensional ratios, and may be displayed exaggerated than actual.

(First embodiment)
(Structure and manufacturing method of semiconductor light emitting device)
FIG. 1 is a cross-sectional view showing a first embodiment of a semiconductor laser device (semiconductor optical device). FIG. 2 is a plan view showing an active layer of the semiconductor laser device of FIG. In FIG. 2, the surface defect generating derivative 10 disposed below the active layer 4 is indicated by a solid line. In the present embodiment, a ridge stripe type semiconductor laser element is exemplified.

First, an outline of the semiconductor laser element will be described.
The semiconductor laser element has a semiconductor layer (active layer 4) in which a waveguide region (energization region A ′) extending in one direction is formed.
Below the semiconductor layer (active layer 4), a plurality of surface defect generating derivatives 10 for forming surface defects in other regions except for the waveguide region (conducting region A ′) of the semiconductor layer (active layer 4) are spaced apart. Are arranged.
In plan view from the surface side of the semiconductor layer (active layer 4), the surface defect generation derivative 10 and the waveguide region (conduction region A ′) are spaced apart and the waveguide region (conduction region A ′) extends. A plurality of surface defect generating derivatives 10 are scattered along the direction.

Next, the semiconductor laser element will be described in detail.
The semiconductor laser element is a nitride semiconductor laser, and is provided on the n-type GaN substrate 1, the surface defect induction region 9 provided in the region B on the substrate 1, and the substrate 1 and the surface defect induction region 9. an n-type AlGaN cladding layer 2 (first cladding layer), a GaN lower optical waveguide layer 3 sequentially provided on the cladding layer 2, an InGaN / GaN quantum well active layer 4, a GaN upper optical waveguide layer 5, A p-type AlGaN cladding layer 6 (second cladding layer) is provided on the GaN upper optical waveguide layer 5 and forms a current confinement portion.
Although the InGaN / GaN quantum well active layer 4 is used as the quantum well active layer, the present invention is not limited thereto, and the barrier layer may be an InGaN layer or an AlInGaN layer. That is, as the barrier layer, an Al _y2 In _x2 Ga _1-x2-y2 N layer (0 <x1 <1, 0 ≦ y2 <1, 0 ≦ x2 <1, 0 ≦ y2 + x2 <1) is preferable.
The emission wavelength of the semiconductor laser element is preferably 440 nm or more and 490 nm or less.

  The cladding layer 6 and the p-type GaN contact layer 7 provided thereon are processed into a ridge stripe shape, and this ridge stripe functions as a current confinement portion. The ridge stripe also functions as a horizontal refractive index waveguide mechanism. The width of the ridge stripe is 1.5 μm, for example. A p-type electrode 8 is provided on the contact layer 7, and an n-type electrode 12 is provided below the GaN substrate 1. 1 is covered with an insulating film 11 except for the portion where the ridge stripe and the p-type electrode 8 are provided.

In the active layer 4, a region having the same width as the ridge stripe located below the ridge stripe becomes the energization region A ′, and the light emitting region has substantially the same width as the ridge stripe. The energization region A ′ has a planar rectangular shape when viewed from the surface side of the active layer 4 and extends in one direction. In this embodiment, since the semiconductor laser element having the ridge stripe structure is used, the energized region A ′ is a region immediately below the ridge stripe. For example, a part of the semiconductor layer containing Al is oxidized. In the case where a current confinement layer is formed, a region immediately below the non-oxidized region of the semiconductor layer containing Al becomes a current-carrying region.
The area A completely includes the energization area A ′ and indicates a wider area than the energization area A ′. In this region A, the surface defect generating derivative 10 is not formed. Region B refers to other regions excluding region A. A surface defect induction region 9 is formed in a part of the region B.

Here, a plurality of spaced-apart surface defect generating derivatives 10 are formed in the surface defect induction region 9 as shown in FIGS. 1 and 2. Further, when the semiconductor laser device is viewed in plan from the substrate surface side, the surface defect 15 exists above the surface defect generating derivative 10 in the active layer 4.
A specific example of the surface defect generating derivative 10 is an AlN thin film having a planar shape of 5 μm square and a thickness of 0.1 μm, for example. The surface defect 15 in the active layer 4 is an aggregate of dislocations generated due to compressive strain, for example, spreading in the horizontal direction.

  With this structure, it is possible to suppress crystal deterioration due to a high-temperature process after crystal growth, which becomes noticeable as the emission wavelength of the InGaN / GaN quantum well active layer 4 becomes longer.

A method for manufacturing the nitride-based semiconductor laser device shown in FIG. 1 will be described with reference to FIGS.
First, an AlN thin film is deposited on the n-type GaN substrate 1 to a thickness of 0.1 μm at a low temperature of about 400 ° C. using a metal organic chemical vapor deposition method (MOVPE method) or the like.
Next, using a normal photolithography process, for example, a square etching mask of about 5 μm square is formed, and portions other than the mask are removed by wet etching using phosphoric acid. Thus, a 5 μm square AlN thin film functioning as the surface defect generating derivative 10 is formed. The surface defect generation derivatives 10 are arranged in a scattered manner in the surface defect induction region 9 so as to be separated from each other. The planar arrangement of the surface defect generating derivative 10 is as shown in FIG. 2, for example, and is arranged at a pitch of 125 μm in the direction parallel to the extending direction of the ridge stripe in the vicinity of the ridge stripe, that is, in the vicinity of the energization region A ′. They are arranged at a pitch of about 250 μm at a position away from the vicinity of the ridge stripe (step 1, FIG. 4).

Next, the first cladding layer 2, the optical waveguide layer 3, the InGaN / GaN quantum well active layer 4, the optical waveguide layer 5, the second cladding layer 6, and the contact layer 7 are formed on the substrate 1 using the MOVPE method or the like. Are sequentially stacked (step 2, FIG. 5).
Here, the indium composition of the well layer used for the InGaN / GaN quantum well active layer 4 is about 0.22 so that oscillation at a wavelength of 450 nm can be obtained. At this time, for example, the crystal growth temperature is preferably about 800 ° C. Further, when the optical waveguide layer 5 or the cladding layer 6 on the active layer 4 is grown, it is desirable to set a higher growth temperature, for example, 1100 ° C. When the upper optical waveguide layer 5 and the cladding layer 6 are crystal-grown at a high temperature, fine defects are concentrated to some extent in the upper portion of the surface defect generating derivative 10 compared to other regions. Surface defects will occur.

  Next, a stripe-shaped etching mask having a width of about 1.5 μm is formed using a normal photolithography process, and etching is performed halfway through the contact layer 7 and the cladding layer 6 by dry etching using a chlorine-based gas. Thereby, a ridge stripe having a width of about 1.5 μm is formed. The value of the ridge width and the etching depth of the cladding layer 6 affect the current-light output characteristics and current-voltage characteristics, including the horizontal transverse mode characteristics of the semiconductor laser device. Select the optimum value. (Step 3, FIG. 6).

  Next, an insulating film 11 such as a silicon oxide film is formed on the entire element by CVD or the like. Thereafter, the insulating film in the p-type electrode 8 forming portion is removed using a normal photolithography process. Furthermore, titanium and gold are vapor-deposited as the p-type electrode 8, and the p-type electrode 8 is formed by performing an alloy process by heating under appropriate conditions. Further, the n-type electrode 12 is formed by vapor-depositing titanium and gold as the n-type electrode 12 on the back surface of the substrate 1 and performing an alloying process by heating under appropriate conditions. Finally, a laser mirror end face is formed by cleavage (step 4, FIG. 7).

(About the effect of improving the deterioration of the luminous characteristics)
Here, the improvement effect of the deterioration of the light emission characteristics of this embodiment will be described. First, with reference to the example of FIG. 12, a manufacturing process of a layered structure of a blue-band semiconductor laser element having a conventional structure will be described, and a degradation mechanism that is supposed to occur due to this will be described.

  First, an n-type lower AlGaN cladding layer 202, a lower optical waveguide layer (GaN layer) 203, an InGaN / GaN quantum well active layer 204, and an upper optical waveguide layer (GaN layer) 205 are formed on an n-type GaN substrate 201. Then, the layers are sequentially laminated using a metal organic chemical vapor deposition method (MOVPE method) or the like.

Here, the InGaN / GaN quantum well active layer 204 is designed to have an oscillation wavelength of 450 nm, and the average indium composition of the InGaN quantum well layer 204 is 0.22.
The crystal growth temperature is about 850 ° C., and the composition is adjusted to a desired value. The indium composition of the InGaN quantum well layer 204 is about 0.22 on average over the entire device, but phase separation occurs, so that a portion having a high indium composition and a portion having a low indium composition are locally observed. Is present. In a region with a particularly high indium composition, there are micro defects due to misfit dislocations depending on the location. However, no surface defects are generated immediately after crystal growth of the quantum well active layer 204 (step 1 ′).

  Further, the upper AlGaN cladding layer 206 and the p-type GaN contact layer 207 are sequentially stacked using the MOVPE method. In the crystal growth of these indium-free layers, a higher growth temperature is desirable for obtaining good crystals, and the temperature is set to 1100 ° C. Although this crystal growth step is described separately for the purpose of explaining the deterioration mechanism, it may be performed continuously following the crystal growth of the active layer 204 (step 2 ').

  As a result, the thermal energy caused by the high temperature growth causes the growth of defects and further generation of the accompanying defects starting from the minute defects formed by the phase separation generated in the step 1 ′. The crystal is deteriorated over a wide range, and surface defects 221 are generated in the plane of the quantum well active layer 204 as shown in FIG.

  As a result of studying the formation factors of the surface defects in the surface of the quantum well active layer, the present inventor has found that the formation conditions of the surface defects are most dependent on the indium composition of the InGaN quantum well layer itself, but the barrier layer is also The present inventors have found that the structure of the entire active layer including the active layer area is also affected. That is, even when a micro defect occurs due to phase separation or the like of the InGaN quantum well layer, it is possible to suppress the occurrence of surface defects that degrade the laser characteristics by keeping the average strain amount of the entire active layer low. Conceivable.

Further, as a result of further analysis of the deterioration form, the following features were found.
FIG. 8 shows the strain amount and surface of the active layer in samples having InGaN quantum well layers of various indium compositions (structure without the surface defect generation derivative 10, barrier layer is an AlInGaN layer (where the barrier layer has a sufficiently small strain)). This is a plot of the relationship with the occurrence of defects.
As an index of the amount of strain of the active layer, in order to compare samples with different compositions and layer thicknesses on the same basis, the ratio of the layer thickness of the quantum well layer and the calculated value of the critical film thickness corresponding to the indium composition was expressed. (The horizontal axis in FIG. 8).
The occurrence state of the surface defect is indicated by the area coverage of the surface defect in the active layer surface (vertical axis in FIG. 8).
The growth conditions of each active layer used here are almost the same except that the optimum growth temperature corresponding to the indium composition is set. In such a case, the individual surface defects have almost the same size regardless of the indium composition and the strain amount of the quantum well layer, and the formation density with respect to the change of the indium composition and the strain amount as shown in FIG. Were found to be different.
This is because the microdefects themselves due to phase separation of the InGaN layer can occur even when the amount of strain of the entire active layer is small, so the degradation form at each dark spot is similar, but the process of expanding to surface defects Shows that the strain amount of the entire active layer has a great influence.

That is, in the process of surface defect formation, when the compressive strain of the active layer exceeds a certain critical value, dislocation growth occurs while relaxing the strain starting from some of the most fragile micro defects. When the surface defect expands to the above size, the average strain amount in the surface having no surface defect at that time becomes less than the critical value, and it is considered that no further growth occurs. After such growth stop, it is considered that the compressive strain is divided by the surface defect in the peripheral portion of the surface defect, the strain stress is reduced, and a good crystal is retained.
Therefore, by controlling the position so that such a surface defect does not overlap with the energized region, the surface defect does not occur in the energized region and can be held well.

  Therefore, by providing a structure that induces the generation of microdefects that are more likely to fail in the active layer in the outer region excluding the energized region, that is, the surface defect generation derivative 10, A method to suppress the occurrence of surface defects in the region was devised.

In the process of FIG. 5, the first cladding layer 2, the optical waveguide layer 3, and the InGaN / GaN quantum well active layer 4 are grown using the MOVPE method or the like. At this stage, the portion grown on the thin film AlN, which is the surface defect generating derivative 10, has a low quality crystallinity of the thin film AlN grown at a low temperature and a large lattice mismatch with the thin film AlN. Many dislocations that propagate vertically to the lower optical waveguide layer 3 are generated.
Therefore, when the quantum well active layer 4 is crystal-grown, the crystallinity of the base is deteriorated in the portion corresponding to the upper portion of the surface defect generating derivative 10 as compared with the portion where the surface defect generating derivative 10 is not present in the lower portion. Therefore, minute defects such as misfit dislocations due to the compressive strain of the InGaN layer are easily formed.
Further, in the case where dislocations propagating perpendicularly to the underlayer are present, it is considered that indium composition concentration is likely to occur, and there is a high possibility that minute defects are concentrated to some extent.
On the other hand, even in a portion where the surface defect generating derivative 10 is not present in the lower portion, a composition distribution due to a high indium composition is also spontaneously generated. A micro defect is formed.

  In such a situation, when the upper optical waveguide layer 5 and the cladding layer 6 are further crystal-grown at a high temperature, when the amount of strain of the active layer 4 is high, a surface defect occurs. In the upper part of the surface defect generating derivative 10, minute defects are concentrated to some extent as compared to other regions, and therefore, the surface defect is preferentially generated in the upper part of the surface defect generating derivative 10. At this time, if the formation density of the surface defect generation derivative 10 is appropriately set with respect to the strain amount of the active layer 4, after the surface defect occurs in a part or all of the surface defect generation derivative 10, Since the compressive strain in the other region (the region including the energization region A ′) is reduced, it is maintained well. Through the above process, it is possible to control the occurrence position of the surface defect outside the energized region.

  On the other hand, Patent Document 1 discloses a method of providing a defect introduction region having a higher dislocation density below the active layer in order to relax the strain of the InGaN quantum well active layer.

In Patent Document 1, a V-shaped groove having a wide area and extending along a ridge stripe is formed in a contact layer on a GaN substrate. A high threading dislocation region is formed in a wide range along the V-shaped groove.
For this reason, the distortion is most relaxed at the periphery of the V-shaped groove of the active layer due to the influence of the V-shaped groove and the high threading dislocation region. Therefore, it is difficult to intentionally generate a surface defect at this position when a high temperature process is performed after crystal growth.
On the other hand, in the region away from the periphery of the V-shaped groove, the distortion is not eliminated, and surface defects are likely to occur when a high temperature process is performed after crystal growth.
Therefore, when the distance between the V-shaped groove and the energization region is wide, distortion is not eliminated in the energization region of the active layer, and a surface defect occurs in the energization region. In particular, when the strain of the active layer is large, surface defects are likely to occur in the energized region.
Here, in order to prevent the occurrence of surface defects in the energized region of the active layer, it is conceivable to bring the energized region close to the position of the V-shaped groove. There is a possibility that threading dislocations may occur in the current-carrying region by overlapping the current-carrying region.

  On the other hand, in this embodiment, a plurality of surface defect generating derivatives 10 are scattered along the longitudinal direction of the energization region A ′. By interspersing the plurality of surface defect generating derivatives 10, threading dislocations can be collected at a portion corresponding to the upper portion of the surface defect generating derivative 10 of the active layer 4, but excessive strain relaxation can be prevented. Therefore, when the high temperature process is performed after the active layer 4 is grown, the surface defect can be generated starting from the portion corresponding to the upper portion of the surface defect generating derivative 10. By adjusting the arrangement position of the surface defect generating derivative 10, it is possible to prevent the surface defect from reaching the energized region A 'of the active layer 4 and to reliably suppress the deterioration of the semiconductor laser element.

(Regarding the arrangement and size of surface defect-generating derivatives)
In order to control the occurrence position of the surface defect, the surface defect generating derivative 10 is used in the active layer 4 without excessively relaxing the distortion of the peripheral portion of the active layer 4 corresponding to the upper portion of the surface defect generating derivative 10. In order to trigger the occurrence of surface defects, the size of each structure is comparatively small, and it is necessary to appropriately select the structure and arrangement so that the structures are separated from each other with a certain distance.

Here, the setting policy of the formation density of the surface defect generating derivative 10 and its application range will be described.
FIG. 9 is a plot of the average surface defect spacing obtained from the surface defect formation density of the sample of FIG. 8 against the strain amount of the active layer. Therefore, if the surface defect generating derivative 10 is formed at a density that is about this average interval with respect to the strain amount of a certain active layer, surface defect formation at portions other than the upper portion of the surface defect generating derivative 10 is suppressed. It is considered possible. In this case, the size of the surface defect does not greatly depend on the strain amount, and is about 50 μm in diameter.

For example, in this embodiment, the indium composition of the active layer 4 is 0.22, and the critical thickness ratio of the well layer thickness is about 117%. Therefore, the average plane defect interval is about 130 μm from FIG. Since the surface defect size is about 50 μm in diameter, a simple surface defect non-formation region having a width of about 80 μm can be obtained around the surface defect.
Therefore, as shown in FIG. 2, if the surface defect generating derivative 10 is arranged so that the energized region A ′ becomes a surface defect non-formed region, the occurrence of surface defects in the energized region A ′ can be suppressed. Can do.

  Here, a patterned AlN thin film is used as the surface defect generation derivative 10, but the present invention is not limited to this as long as it similarly promotes the formation of micro defects in the active layer 4 and the concentration of indium composition. The advantages of using an AlN thin film grown by the MOVPE method are as follows. The AlN thin film deposited by low temperature growth maintains a crystal structure although it contains many defects due to a large lattice mismatch. Therefore, even in a crystal regrown on the upper part, many dislocations propagating vertically are generated, but there are few factors that cause growth disturbance to other parts. Therefore, it is considered that there is little possibility that the crystal breaks down immediately after the growth of the active layer and a wide range of crystals are deteriorated. Of course, an AlGaInN film containing Ga or In may be used instead of the AlN thin film, and the degree of lattice mismatch and dislocation propagation characteristics can be adjusted by the composition ratio.

  Examples of other surface defect generating derivatives may be silicon oxide films, nitride films, alumina films, and the like. In these cases, few crystals are directly grown in the surface defect generation derivative portion, and crystals grow laterally from the edge of the structure, and finally a crystal is also formed on the surface defect generation derivative portion. However, when the crystals grow and meet from the lateral direction in this way, dislocations that propagate vertically occur in the meeting portion. Therefore, the same effect as in the case of the AlN thin film can be obtained. Any material other than this may be used as long as the same effect can be obtained. In addition, as for the shape of the surface defect generating derivative, one example in which a minute defect is effectively obtained is shown here, but it is only an example, and any shape may be used.

  Here, the pattern of the AlN thin film, which is a surface defect-generating derivative, is 5 μm square, but it may be small enough not to make the process difficult. In a general process that does not use special technology, even if the formation size is reduced, the defect is larger than a naturally occurring micro defect, the selectivity of the surface defect occurrence position is sufficiently maintained, and the distortion in the peripheral part is excessive. Therefore, surface defects can be generated effectively. Specifically, the length along the substrate surface direction of the surface defect generating derivative 10 is preferably 0.1 μm or more and 10 μm or less. By setting the thickness to 0.1 μm or more, the surface defect generating derivative 10 can be easily formed. On the other hand, by setting the thickness to 10 μm or less, there is an effect that it is possible to prevent the strain from being excessively relaxed in the upper region of the surface defect generating derivative 10 when the active layer 4 is grown.

  Next, the arrangement of the surface defect generating derivative 10 will be described with reference to FIGS. FIG. 3 shows only the surface defect generating derivative 10 shown in FIG. 2 formed near the ridge stripe. Further, a surface defect 15 having a diameter of about 50 μm generated by the surface defect generating derivative 10 and a portion where distortion can be reduced and a surface defect non-formation region 16 can be formed by the surface defect generation are displayed. Here, these surface defect generation derivatives 10 are formed at a pitch of about 125 μm, for example, in a direction parallel to the ridge stripe. As shown in FIG. 3, when a surface defect occurs in all of these surface defect generating derivatives 10, the energized region A ′ is covered with a portion showing the surface defect non-formation region 16 as shown in FIG. In the region A ′, it is considered that the influence of the compressive strain in the crystal is reduced and the possibility of being kept good is increased.

In this case, since the space between the surface defects located on both sides of the ridge stripe is about 80 μm, the surface defect generating derivative 10 may be formed at a position about 40 μm away from the center of the stripe or inward thereof. Since the width of the light emitting region has an effect such as current spreading, the width of the ridge stripe (current constriction portion) of the laser is about 1 to 2 μm wider than one side. Therefore, the region A has a margin of about 5 μm. In consideration of this, the value may be equal to or more than the value obtained by adding about 12 to 14 μm to the ridge stripe width. The arrangement of the surface defect generating derivative 10 at the position closest to the ridge stripe may be determined from the required amount of the width of the region A and the value of the width of the surface defect non-forming region. In the example of FIG. 1, the arrangement is such that the width of the area A can be widened, but if the width of the area A is narrower, the pitch in the direction parallel to the stripe may be wider.
Considering that the diameter of the surface defect in the active layer is 50 μm, it is preferable that the distance between the energized region A ′ and the surface defect generating derivative 10 in a plan view is at least 25 μm. Further, the arrangement pitch of the surface defect generating derivatives 10 is preferably 50 μm or more.

On the other hand, in FIG. 2, the disposition of the surface defect generation derivative 10 is set so as to be sparser than the vicinity of the ridge stripe at a position away from the vicinity of the ridge stripe. Specifically, the arrangement pitch of the array closest to the energization region A ′ among the plurality of arrangements along the energization region A ′ of the surface defect-generating derivative 10 in a state viewed in plan from the substrate surface side is reduced. The arrangement pitch of the arrangement is made larger than the arrangement pitch of the arrangement closest to the energization region A ′.
This is because in order for the current-carrying region A ′ to be a surface defect non-formation region, it is desirable that a large number of surface defects exist in the vicinity of the current-carrying region A ′. This is because a small value is desirable in considering the influence of local distortion. Therefore, in this region, a surface defect generating derivative may be formed in which a micro defect that hardly breaks down compared to the surface defect generating derivative in the vicinity of the energized region A ′ may be formed. It does not have to be formed. In the latter case, it is considered that a surface defect is generated starting from a minute defect formed spontaneously.

  Thus, when many surface defect generation | occurrence | production derivatives 10 are formed, a surface defect does not necessarily generate | occur | produce on all the structures. Here, the arrangement is determined from the average surface defect density, but in reality, the strain distribution as assumed including phase separation or the like is not always formed. However, when surface defects are suppressed in a certain part, it indicates that the strain in the region other than the surface defects is stabilized at a critical value or less in the peripheral part, and in this case, the plane defect interval is stabilized. In some cases, the surface defect non-formation region may be obtained with a width wider than the value obtained from the above.

The arrangement of the surface defect generating derivative 10 is not necessarily regular, and may be provided at an average interval or more, and the arrangement example shown here is only an example. However, in order to effectively suppress the occurrence of surface defects in the portion other than the upper portion of the surface defect generating derivative 10 (portion other than the vicinity of the current-carrying region of the active layer) It is preferable to keep the arrangement of the structures to the minimum necessary that can be estimated from the conditions described later.
Even when the present invention is applied to a light emitting element other than a laser, the arrangement may be determined in accordance with the energization region.

(Examination of surface defect formation density and strain)
Next, the surface defect formation density and the amount of strain will be examined in more detail. Thereby, by clarifying the applicable range of the InGaN quantum well layer structure, it is possible to obtain the maximum effect of improving the light emission characteristics.

When the present inventor examined a semiconductor light emitting device having a conventional InGaN quantum well layer structure, it was found that the average size of surface defects may be a certain value, for example, a circular or elliptical shape with a diameter of 50 μm. Therefore, when the average amount of surface defect is 50 μm or less and the strain amount is more than the critical thickness ratio of the well layer thickness is about 140% or more, surface defects may be formed in many regions in the active layer surface. Even when the surface defect generating derivative 10 is provided, it is difficult to concentrate the generation of surface defects on the surface defect generating derivative 10 and the effectiveness of the present invention may be reduced. In the quantum well layer and the GaN barrier layer having an indium composition of 0.22, the wavelength in this case corresponds to 490 nm.
On the other hand, as shown in FIG. 8, when the critical thickness ratio of the well layer thickness is 105% or less, the occurrence of surface defects may be difficult to occur, and the effectiveness of the present invention is reduced. In this case, the wavelength is 440 nm in the case of a quantum well layer having an indium composition of 0.22 and a GaN barrier layer.

  Curve A in FIG. 10 is based on the indium composition and critical film thickness of the InGaN quantum well layer on the GaN substrate c-plane calculated based on Mathews and Blakeslee theory (see Non-Patent Document 1). The upper and lower limits of the range are shown. That is, the curve A in FIG. 10 shows 105% of the critical film thickness, and the curve B shows 140% of the critical film thickness. It is considered that the present invention is particularly effective in the range between the curves.

  On the other hand, in the InGaN quantum well layer on c-plane GaN, the spatial overlap of the wave function of electrons and holes in the quantum well is reduced due to the influence of the internal electric field due to spontaneous polarization and piezo polarization, thereby reducing the light emission probability. It has been known. This effect becomes more prominent when the indium composition becomes higher. To avoid this effect, it is desirable to make the quantum well layer thickness within a relatively thin range of about 1.0 to 3.0 nm.

  In consideration of these, as shown in FIG. 10, when the quantum well layer thickness is in the range of 1.0 to 3.0 nm, the indium composition of the quantum well layer that exceeds the critical value, that is, the present invention is particularly effective is 0.17. ˜0.39. Therefore, the effect of improving the light emission characteristics according to the present invention is 0.17 to 0.39, assuming the relationship between the strain amount and the surface defect average interval (surface defect density) in FIG. 9 and the composition dependency in FIG. Particularly effective in the indium composition range.

However, even if the amount of strain is lower than the curve A, there is no possibility that surface defects will occur, and it is considered that the present invention can avoid the deterioration of the semiconductor laser device. Even if the amount of strain is higher than the curve B, the present invention is effective because the size of the surface defect may change by devising the arrangement and shape of the surface defect-generating derivative.
Further, as described above, it has been found that the surface defect condition is not determined only by the indium composition of the quantum well layer but is influenced by the strain amount of the entire active layer such as the barrier layer structure and the number of quantum wells. In particular, the use of Al-containing materials for the barrier layer and the layer adjacent to the active layer reduces the strain of the active layer as a whole, and suppresses deterioration even in quantum well structures that exceed the composition and wavelength values shown here. It is considered possible.
Also, the calculated value of the critical film thickness used here is a value based on the theory when a certain condition is satisfied, and does not necessarily match the composition condition where actual deterioration occurs, and the numerical value varies depending on the condition. It can be.

Therefore, the present inventors have used a structure according to the background art as shown in FIG. 12 (the active layer is an In _x1 Ga _1-x1 N well layer / Al _y2 In _x2 Ga _1-x2-y2 N barrier layer). (0 <x1 <1, 0 ≦ y2 <1, 0 ≦ x2 <1, 0 ≦ y2 + x2 <1) (however, the barrier strain is sufficiently small) was experimentally studied. In practice, since it is difficult to accurately identify the indium composition, the composition conditions causing deterioration are estimated using the emission wavelength that varies depending on the indium composition of the quantum well layer as a guide. When the emission wavelength is 440 nm or more, the layer structure In some cases, deterioration may occur, and surface defects were observed in a wide range at 470 nm or more.
On the other hand, when the structure of the present invention is applied, it has been confirmed that surface defects in the stripe region may be suppressed at least until the emission wavelength is 490 nm.

(About the effect of this embodiment)
As described above, in the semiconductor laser device of FIG. 1, deterioration due to generation of surface defects in the active layer can be suppressed in the energization region A ′. That is, in this embodiment, a plurality of surface defect generating derivatives 10 for intentionally generating surface defects in the active layer 4 are provided. And the plane defect generation | occurrence | production derivative | guide_body 10 is spaced apart and arrange | positioned from the electricity supply area | region A 'of the active layer 4 in planar view from the substrate surface side, and it is active by adjusting the arrangement position of the plane defect generation | occurrence | production derivative | guide_body 10. It is possible to prevent the surface defect from reaching the energization region A ′ of the layer 4.
In particular, it is possible to suppress deterioration in the energized region A ′ due to generation of surface defects due to a high-temperature process after crystal growth, which becomes remarkable as the emission wavelength of the quantum well active layer 4 becomes longer. Specifically, the indium composition (x1) is in the range of 0.17 to 0.39, and the emission wavelength is In _x1 Ga _1-x1 N (well layer) / corresponding to the range from 440 nm to 490 nm. Al _y2 In _x2 Ga _1--x2-y2 N (barrier layer) (0 <x1 <1, 0 ≦ y2 <1, 0 ≦ x2 <1, 0 ≦ y2 + x2 <1) Particularly effective in the quantum well structure An effect is obtained.

  Furthermore, in this embodiment, an AlN thin film is used as the surface defect generating derivative 10. This AlN thin film maintains a crystal structure although it contains many defects due to a large lattice mismatch. Therefore, even in a crystal regrown on the upper part, many dislocations propagating vertically are generated, but there are few factors that cause growth disturbance to other parts. Therefore, it is considered that there is little possibility that the crystal breaks down immediately after the growth of the active layer and a wide range of crystals are deteriorated.

  In addition, the size of the surface defect 15 generated in the active layer 4 by the surface defect generating derivative 10 is 50 μm in this embodiment. Therefore, it is possible to prevent the surface defects 15 generated in the active layer 4 from overlapping each other by setting the arrangement pitch of the surface defect generating derivatives 10 to 50 μm or more.

  Further, in the present embodiment, in the plan view from the substrate surface side, the arrangement density of the surface defect generation derivative 10 in the vicinity of the energization region A ′ of the active layer 4 is the other region excluding the vicinity of the energization region A ′ of the active layer 4. It is higher than the arrangement density of the surface defect generating derivative 10 in FIG. In order for the inside of the energization region A 'to be the surface defect non-formation region 16, it is desirable that many surface defects 15 exist in the vicinity of the energization region A'. For this purpose, it is desirable that the surface defect density is small in the outer region in the vicinity of the energized region A ′ in view of the influence of local strain.

  In actual application, it is desirable to investigate the relation between the formation density of surface defects and the amount of strain in the structure and conditions to be used, and to select conditions appropriately. In addition, the arrangement of the structures in the surface defect induction region can also be set appropriately according to the situation.

  In the process as described above, the element structure is almost the same as the conventional structure except for the presence of the surface defect generating derivative, and the process exactly the same as the conventional structure can be applied after forming the surface defect generating derivative.

(Second Embodiment)
Next, a second embodiment of the semiconductor laser device according to the present invention will be described with reference to FIG.
FIG. 11 is a sectional view of the semiconductor laser device according to the present embodiment. In this embodiment, substantially the same manufacturing method as that of the first embodiment shown in FIG. 1 is adopted, but the formation position and method of the surface defect generating derivative in step 1 are different.

In the embodiment of FIG. 11, the surface defect generating derivative 10 ′ is a recess formed on the surface of the GaN substrate 1. The other points are the same as in the above embodiment.
In the step of forming the surface defect generating derivative 10 ′, for example, a mask opening of about 2 μm square is formed on the GaN substrate 1 using ordinary photolithography, and then about 3 μm using a chlorine-based dry etching or wet etching method. By forming a recess having a depth of 10 mm, the surface defect generating derivative 10 ′ is formed.
After that, it is manufactured by the same method as in the above embodiment.
When each layer is crystal-grown using the MOVPE method, the supply of the raw material is smaller in the portion grown on the concave portion that becomes the surface defect generating derivative 10 'compared to the flat portion. Therefore, the first cladding layer 2 and the optical waveguide layer 3 are formed while maintaining the step of the recess to some extent. When the InGaN / GaN quantum well active layer 4 is grown thereover, the composition of indium is likely to occur because the underlying layer is concave above the surface defect generating derivative 10 '.
Therefore, the same effect as the first embodiment can be obtained.
Further, depending on the etching conditions at the time of forming the recess, defects and impurities due to etching damage are generated in the recess, and the effect of forming dislocations propagating vertically can be obtained at the same time.
The arrangement pitch, the arrangement density, and the like of the recesses are the same as those of the surface defect generating derivative 10 of the above embodiment.

(Third embodiment)
Furthermore, a third embodiment of the semiconductor laser device according to the present invention will be described. The structure is the same as that of the second embodiment of FIG. 11 and employs substantially the same manufacturing method as that of the first and second embodiments, but the method for forming the surface defect generating derivative in step 1 is different.

In the present embodiment, the surface defect generating derivative is a recess formed on the surface of the GaN substrate.
The other points are the same as in the first embodiment.
In the formation process of the surface defect generating derivative, a mask opening of about 5 μm square, for example, is formed on the GaN substrate position by using normal photolithography.
Thereafter, the surface defect generating derivative is formed by raising the temperature to, for example, 800 ° C. in a nitrogen atmosphere. When such a process is performed, nitrogen is released from the portion of the crystal of the GaN substrate that is not covered with the mask, and as a result, many vacancy defects (concave portions) are generated on the surface of the GaN substrate. This part functions as a surface defect generating derivative.
After removing the mask, it is manufactured by the same method as in the first embodiment.
When each layer is crystal-grown using the MOVPE method, dislocations that propagate perpendicularly starting from vacancy defects are generated in the portion to be a surface defect generating derivative, and the same effect as in the first embodiment can be obtained.

Further, as a different manufacturing method, an ion implantation process can be used instead of the temperature raising process. After forming a mask opening of about 5 μm square as described above, for example, argon ions are accelerated to about 100 keV to perform ion implantation. In this case, the mask material and thickness are appropriately selected so that ions do not enter the mask covering region. By this process, in the part not covered with the mask, ions enter the GaN crystal and a large number of nitrogen vacancy defects are generated on the surface of the GaN substrate. This part similarly functions as a surface defect generating derivative.
The arrangement pitch, the arrangement density, and the like of the recesses are the same as those of the surface defect generating derivative 10 of the above embodiment.

  It should be noted that the present invention is not limited to the above-described embodiments, and modifications, improvements, and the like within the scope that can achieve the object of the present invention are included in the present invention.

In each of the above embodiments, the active layer is an InGaN / GaN layer, but is not limited thereto.
For example, the substrate may be a GaN substrate, and the active layer may be a layer composed of a layer containing at least In, Ga, and N.
In each of the embodiments, the semiconductor laser element on the n-type GaN substrate is taken as an example. However, a semiconductor laser element on a substrate other than the GaN substrate such as a sapphire substrate or a silicon substrate may be used.

In the above-described embodiment, the ridge stripe type semiconductor laser structure has been described. However, the present invention is not particularly limited to such a structure. The present invention can be applied to an inner stripe type semiconductor laser element, a mesa type light emitting diode element, a surface emitting laser element, and the like.
Furthermore, the present invention may be applied not only to light emitting elements but also to light receiving elements and optical waveguide elements.
Moreover, in the said embodiment, although the III-V nitride type semiconductor light-emitting device was illustrated as a semiconductor light-emitting device, it is not restricted to this.
Furthermore, the semiconductor laser device according to each of the embodiments can be used for a light source for video equipment such as a next-generation laser display.

It is sectional drawing which shows the semiconductor laser element of the 1st Embodiment of this invention. It is a top view of the active layer of a semiconductor laser element. It is a top view which shows the principal part of a semiconductor laser element. It is sectional drawing which shows the manufacturing process of a semiconductor laser element. It is sectional drawing which shows the manufacturing process of a semiconductor laser element. It is sectional drawing which shows the manufacturing process of a semiconductor laser element. It is sectional drawing which shows the manufacturing process of a semiconductor laser element. It is a figure which shows the relationship between the distortion amount of an active layer, and the generation | occurrence | production state of a surface defect. It is the figure which plotted the space | interval of the average surface defect calculated | required from the sample of FIG. 8 with respect to the distortion amount of the active layer. It is the figure which showed the relationship between the indium composition of an InGaN quantum well layer, and a film thickness. It is sectional drawing of the semiconductor laser element which is the 2nd Embodiment of this invention. It is sectional drawing of a nitride-type semiconductor laser element. It is a top view of the active layer of a nitride-type semiconductor laser element.

Explanation of symbols

1 Substrate 2 Clad layer 3 Lower optical waveguide layer 4 Active layer (quantum well active layer)
5 Upper optical waveguide layer 6 Clad layer 7 Contact layer 8 P-type electrode 9 Planar defect induction region 10 Planar defect generation derivative 11 Insulating film 12 N-type electrode 15 Planar defect 16 Surface defect non-formation region 201 Substrate 202 Clad layer 203 Lower optical Waveguide layer 204 Quantum well active layer 205 Upper optical waveguide layer 206 Cladding layer 207 Contact layer 208 p-type electrode 212 n-type electrode 220 conducting region 221 surface defect A ′ conducting region A region B region

Claims (12)

In a semiconductor optical device having a semiconductor layer in which a waveguide region extending in one direction is formed,
Below the semiconductor layer, a plurality of surface defect generating derivatives for forming surface defects in other regions other than the waveguide region of the semiconductor layer are arranged apart from each other,
The planar defect generating derivatives and the waveguide region are spaced apart from each other in plan view from the semiconductor layer surface side, and a plurality of the surface defect generating derivatives are pointed along the extending direction of the waveguide region. Existing semiconductor optical device. The semiconductor optical device according to claim 1,
In a plan view from the surface side of the semiconductor layer, the arrangement density of the surface defect generating derivative in the vicinity of the waveguide region in the other region is the surface defect in a region except the vicinity of the waveguide region in the other region. A semiconductor optical device having a higher arrangement density of the generated derivative. The semiconductor optical device according to claim 1 or 2,
A semiconductor optical device having an arrangement pitch of the surface defect generating derivatives of 50 μm or more. The semiconductor optical device according to any one of claims 1 to 3,
The surface defect generating derivative is a semiconductor optical device that is arranged at a distance of 25 μm or more from the waveguide region in a plan view from the semiconductor layer surface side. The semiconductor optical device according to claim 1,
A semiconductor optical device, wherein a length along the surface of the semiconductor layer of the surface defect generating derivative is 0.1 μm or more and 10 μm or less. The semiconductor optical device according to claim 1,
The semiconductor optical device is a semiconductor light emitting device,
The semiconductor optical device, wherein the semiconductor layer is an active layer provided on a substrate, and the waveguide region is a current-carrying region. The semiconductor optical device according to claim 6,
The semiconductor optical device, wherein the surface defect generating derivative is formed of a thin film containing at least Al and N. The semiconductor optical device according to claim 6,
The surface defect generating derivative is a semiconductor optical device which is a recess formed in the substrate surface. The semiconductor optical device according to claim 6,
The substrate is a GaN substrate;
A semiconductor optical device, wherein the active layer comprises a layer containing at least In, Ga and N. The semiconductor optical device according to claim 9, wherein
The active layer is an In _x1 Ga _1-x1 N well layer / Al _y2 In _x2 Ga _1-x2-y2 N barrier layer (0 <x1 <1, 0 ≦ y2 <1, 0 ≦ x2 <1, 0 ≦ y2 + x2 < 1) quantum well structure,
A semiconductor optical device in which x1 is 0.17 or more and 0.39 or less. The semiconductor optical device according to any one of claims 6 to 10,
A semiconductor optical device in which the emission wavelength of the active layer is not less than 440 nm and not more than 490 nm. In a method for manufacturing a semiconductor optical device having a semiconductor layer in which a waveguide region extending in one direction is formed,
Arranging a plurality of surface defect generating derivatives;
Forming the semiconductor layer on the surface defect generating derivative;
Forming a surface defect in a region other than the waveguide region of the semiconductor layer by the surface defect generation derivative,
In the planar view from the semiconductor layer surface side, the surface defect generation derivatives are spaced apart from the waveguide regions and are scattered along the extending direction of the waveguide regions. A method for manufacturing a semiconductor optical device disposed on a substrate.

Images