JP4629687B2 - Optical semiconductor device and manufacturing method thereof - Google Patents

Description

  The present invention relates to an optical semiconductor device, and more particularly to an integrated optical semiconductor device such as a wavelength tunable laser or a semiconductor modulator integrated light source, and a manufacturing method thereof.

  With the explosive growth of the Internet population in recent years, there has been a demand for rapid increase in information transmission and capacity, and optical communication will continue to play an important role in the future. Light sources used in optical communication are required to have higher transmission characteristics. In particular, in long-distance optical communication, a modulator integrated light source in which an optical modulator is integrated in front of the semiconductor laser is used because it cannot be dealt with by directly modulating the semiconductor laser. In addition, in order to cope with wavelength multiplex communication and the like, a light source with high added value such as a wavelength tunable light source capable of instantaneously switching wavelengths is rapidly demanded. In such an integrated optical semiconductor device, regions such as a modulator portion and a wavelength adjusting portion are formed on the same substrate in addition to an active layer (laser portion) that generates light.

  FIG. 1 is an example of a wavelength tunable light source. It is formed of a laser unit that generates light and a wavelength adjustment unit that modulates the phase of light and changes the wavelength. FIG. 2 is an example of a modulator integrated light source. A laser unit that generates light, a modulator unit that modulates the generated light, and a waveguide unit that efficiently propagates light that reaches the modulator unit from the laser unit. In order to propagate light while confining light well in these wavelength adjusting sections and waveguide sections, a semiconductor layer having a thickness of several hundred nm or more is generally used. As a semiconductor material of the integrated optical element, materials such as InGaAsP and InGaAlAs formed on an InP substrate are used. InGaAsP in particular has a characteristic that the material itself is not easily oxidized, and is a very suitable material for manufacturing an integrated optical device using a regrowth process.

  Here, a manufacturing process of the integrated optical device will be described by taking the wavelength tunable laser of FIG. 1 as an example. This is a so-called distributed reflector (DBR) type structure. Process steps mainly relating to crystal growth are shown in FIG. First, as shown in FIG. 3A, an active layer 302 having a quantum well structure is formed on an n-InP substrate 301 by normal crystal growth. Subsequently, after forming the insulator mask 303 as shown in FIG. 3B, unnecessary portions are removed by etching. Further, as shown in FIG. 3C, an InGaAsP waveguide layer 304, an InP spacer layer 305, an InGaAsP diffraction grating layer 306, and the like are formed by regrowth. Finally, as shown in FIG. 3D, after the shape of the diffraction grating 306 is formed, a p-InP cladding layer 307 is formed by regrowth on the whole. Thereafter, the device is completed through a mesa etching process, an electrode deposition process, and the like.

  In manufacturing such an integrated optical device, as described above, an active layer that generates normal light is formed first. This is to avoid the deterioration of the light emission characteristics in the multi-quantum well active layer due to the influence of impurities such as C, O, and Si on the regrowth surface and the increase in strain due to the selective growth effect near the mask. For this reason, the waveguide layer is mostly formed by a regrowth process.

Japanese Journal of Applied Phisics, Volume 21, 1982, p. 797 `` Journal of Crystal Growth 27, 1974, p.118

In an optical element in the 1.55 μm band which is the lowest loss wavelength band of the current silica-based optical fiber, the composition wavelength of the InGaAsP waveguide layer is set to less than 1.55 μm in order to suppress light absorption. Typically, it is set in the range of 1.3 μm to 1.45 μm. The temperature during crystal growth of the InGaAsP layer is usually 500-600 ° C. This is because, if the growth temperature is too high, P atoms having a high vapor pressure escape from the surface, resulting in deterioration of the film quality.
Here, the degree of lattice mismatch ε (%) from the InP substrate of the InGaAsP layer is defined by the following equation.

ε (%) = [{(Lattice constant of InGaAsP)-(Lattice constant of InP substrate)} / (Lattice constant of InP substrate)] x100

In the case of an InGaAsP layer used for a waveguide layer, a thickness of several hundred nm or more is necessary. Typically, it is 200 to 500 nm. Therefore, in order to prevent the occurrence of lattice mismatch dislocations, it is usually formed under conditions that lattice match with the InP substrate. The value of ε at that time is typically set in the range of −0.05% ≦ ε ≦ + 0.05%. Considering the difference in thermal expansion coefficient between the InP substrate and the InGaAsP layer, it is preferable to set it slightly on the negative strain side.

When we crystallized an InGaAsP waveguide layer with a composition wavelength of 1.4 μm, a film thickness of 300 nm, and ε = −0.05% by the regrowth process as shown in FIG. It was. The surface photograph is shown in FIG. It turns out that a lot of defects are observed. In the fabricated device, a leak current flowed through this defect, and the device characteristics were greatly deteriorated. As a result of subsequent studies, the occurrence of defects in this InGaAsP layer is very small when grown in the first crystal growth process, not in the regrowth process, and even if formed in the same regrowth process. Similarly, when the film thickness was as thin as 100 nm or less, it was found that the generation amount was very small as well. Therefore, the occurrence of this defect is considered to be caused by some physical phenomenon, not simply a shift in crystal growth conditions.
Accordingly, an object of the present invention is to provide an integrated optical semiconductor device that suppresses the occurrence of defects in an InGaAsP waveguide layer formed by regrowth, and thereby has good device characteristics.

  As a result of earnest examination of the cause of this defect, it has been found that it is closely correlated with the following phenomenon. Details are described below. It is known that a semiconductor material called InGaAsP has an immiscible region. InGaAsP in the immiscible region is thermodynamically unstable and easily phase-separated, resulting in defects. Japanese Journal of Applied Physics, Vol. 21, p. 797, 1982 describes the theoretical calculation results of the InGaAsP immiscible region (missibility gap) by Onabe. Fig. 1 is quoted in Fig. 5. The immiscible region is shown inside the elliptical region at each absolute temperature.

  That is, it can be seen that the higher the temperature is, the easier the mixed crystal is mixed and the smaller the elliptical region is, so that the immiscibility is relaxed. The absolute temperature shown here is the growth rate for highly immiscible growth methods such as metal-organic vapor phase epitaxy (MOVPE) and molecular beam epitaxy (MBE). It does not agree well with temperature.

  This figure also shows an equiband gap (composition wavelength) line having a composition wavelength of 1.4 μm and an equi-lattice constant line of the InP substrate. It can be seen that the InGaAsP layer having a composition wavelength of 1.4 μm applied to the above-described device has the most immiscible composition under the composition conditions lattice-matched to the InP substrate. By using the growth method with high non-equilibrium such as MOVPE and MBE described above, such immiscibility is greatly relaxed, and the quality that can be applied to real devices as well as InGaAsP layers with highly immiscible composition. The film has been obtained. However, as the essence of the present material system, it is considered that the factors that easily cause phase separation are always included. Therefore, the occurrence of abnormal defects in the InGaAsP layer with the composition wavelength of 1.4 μm described in the previous section causes partial phase separation in the film, with some impurities adhering to the regrowth semiconductor surface. However, when the film thickness is large, it is considered that it accumulated and appeared as a defect.

  Referring to the equiband gap line (isocomposition wavelength line) in FIG. 5, it can be seen that if both the In composition and the P composition are increased, the composition wavelength can be kept at 1.4 μm and move to the lower left region with low immiscibility. . Therefore, it is considered that the immiscibility can be reduced without affecting the device characteristics and a good quality InGaAsP waveguide layer can be formed. Thus, changing the composition ratio of the InGaAsP layer in the direction of low immiscibility along the isocomposition wavelength line is equivalent to setting ε to the + side as a result. Therefore, in the actual examination described below, description is made using ε.

  4B and 4C show InGaAsP layers (thickness 300 nm, growth temperature 600 ° C.) when ε is changed to the + side by increasing the In composition and the P composition while keeping the composition wavelength at 1.4 μm. ) Surface photograph. At ε = + 0.04% in FIG. 4 (b), the number of defects remained very large, but when ε became + 0.1% or more as shown in FIG. 4 (c), almost no defects were observed. Therefore, it was experimentally confirmed that the occurrence of defects can be suppressed by increasing ε to the + side. FIG. 5 shows the point of ε = + 0.1% at which the effect of the present invention was remarkably obtained, but it appears that the decrease in terms of immiscibility is slight. The reason why such a slight decrease in immiscibility has resulted in a large effect on defect suppression is not clear at this time, but the immiscibility is greater due to the increased strain energy on the + side. It may have declined. In any case, it was found that by setting ε to the + side, it was possible to suppress surface defects that were noticeable in the regrowth process.

  FIG. 5 shows that the InGaAsP layer having a composition wavelength of 1.3 μm to 1.5 μm has the same high immiscibility under the condition of lattice matching with the InP substrate. Therefore, in the previous experiment, the defect suppression effect when ε was increased to the + side confirmed for the InGaAsP layer with a composition wavelength of 1.4 μm was the same at other composition wavelengths in the range of 1.3 μm to 1.5 μm. Obtainable.

  From the above considerations, in InGaAsP layers formed by regrowth on InP substrates, it is possible to suppress defects that occur during the regrowth process by changing ε to the + side and changing the composition ratio in a direction that reduces immiscibility. I understood. At this time, since the composition wavelength does not change, the influence on the device characteristics is very small. Note that the theoretical calculation shown in FIG. 5 has existed in the past and provided a guide for the immiscibility of InGaAsP. However, it did not quantitatively clarify the range of generation and suppression of defects in an actual InGaAsP layer. This time, we experimentally demonstrated for the first time that defects can be suppressed by setting ε of the InGaAsP layer to an appropriate value on the + side.

  On the other hand, when ε is set to the + side, it is necessary to consider the set film thickness. As is well known, for each ε value, there is a film thickness upper limit (critical film thickness) in terms of the occurrence of lattice mismatch dislocations. The critical film thickness is shown in the theoretical formula of Journal of Crystal Growth 27, 118, 1974 by Matthews et al. The formula is shown below.

  Here, b is the magnitude of the Burgers vector, ν is the Poisson ratio, α is the angle formed by the dislocation line and the Burgers vector, and λ is the angle formed by the plane perpendicular to the line of intersection between the slip surface and the interface. Here, b = 4, ν = 1/3, cosα = 1/2, and cosλ = 1/2 are common values for III-V compound semiconductors, but this value may vary slightly depending on the material. Needless to say.

  FIG. 6 shows the effective range of the present invention in an InGaAsP layer having a composition wavelength of 1.4 μm formed by regrowth on an InP substrate based on the above equation (1). The lower limit of the film thickness of the waveguide layer used in the integrated optical device is typically 150 nm. According to the theoretical formula of Matthews et al., The value of ε corresponding to the critical film thickness of 150 nm is about + 0.4%. In this case, the effective range of the present invention is specifically indicated by the hatched area surrounded by ABC in FIG.

  In the study of FIG. 4, the In composition and the P composition were increased simultaneously to increase ε, but the immiscibility can be lowered by increasing only the In composition or only the P composition. Needless to say. However, in this case, since the composition wavelength changes, it is necessary to consider the influence on the element characteristics in advance.

  The present invention is most effective for the InGaAsP layer formed by regrowth, but the immiscibility as the essence of the material does not change, so even when formed in the first growth step, It is considered that a higher quality InGaAsP film can be obtained by applying the conditions according to the present invention.

  According to the present invention, in an InGaAsP layer having a composition wavelength of 1.4 μm, which is mainly used as a waveguide layer of an optical integrated device in a 1.55 μm band, the degree of lattice mismatch ε with the InP substrate is set to + 0.1% ≦ ε ≦ + By setting it within the range of 0.4%, it is possible to suppress the generation of defects and obtain a good film. As a result, the effect of improving the characteristics of the integrated optical device using this waveguide layer can be obtained.

  Embodiments of the present invention will be described below with reference to FIGS.

  In the first embodiment, the present invention is applied to a wavelength tunable laser. The MOVPE method was used as the growth method. Triethylgallium and trimethylindium were used as the group III element raw material. Arsine and phosphine were used as raw materials for Group V elements. Disilane was used as the n-type dopant, and dimethylzinc was used as the p-type dopant. The growth method is not limited to MOVPE, but includes MBE method, chemical beam epitaxy (CBE) method, metal-organic molecular beam epitaxy (MOMBE) method, etc. It may be used.

FIG. 1 shows a cross-sectional structure of the element. The MOVPE method was used as the growth method. Triethylgallium and trimethylindium were used as the group III element raw material. Arsine and phosphine were used as raw materials for Group V elements. Disilane was used as the n-type dopant, and dimethylzinc was used as the p-type dopant.
A 10-cycle 1.55 μm-band multiple quantum well active layer 109 made of InGaAsP was formed on the n-InP substrate 102. Next, an insulator mask was formed in a necessary portion by a photo process. Using this as a mask, unnecessary portions were removed by dry etching and wet etching. Subsequently, an InGaAsP waveguide layer 103 having a thickness of 300 nm and a composition wavelength of 1.4 μm, an InP spacer layer 104, and an InGaAsP diffraction grating layer 105 having a composition wavelength of 1.15 μm were formed. Since the strain amount of the InGaAsP waveguide layer 103 was set to + 0.1%, no defects were generated and the surface state was good. Subsequently, after the shape of the diffraction grating layer 105 was processed and the insulator mask was removed, the whole was buried with the p-InP cladding layer 106, and finally the p ⁺ -InGaAs contact layer 107 was formed. Thereafter, a part of the contact layer was removed to separate the elements, and after mesa etching and substrate polishing, electrodes 101 and 108 were formed. After cleavage, reflection films were formed on both end faces to complete the element. The completed device had a good threshold current of 10 mA and a variable wavelength width of 6 nm.

  In the second embodiment, the present invention is applied to a modulator integrated light source. As a growth method, the MOVPE method is used here, however, it is not limited to this, and other methods may be used as long as the same effect can be obtained. As a raw material, trimethylaluminum was used as an aluminum raw material in addition to Example 1. FIG. 2 shows a cross-sectional structure of the element.

  On the n-InP substrate 202, a four-period multiple quantum well active layer 210 made of InGaAlAs was formed as a laser part. Next, an insulator mask was formed in a necessary portion by a photo process. Using this as a mask, unnecessary portions were removed by dry etching and wet etching. Subsequently, an 8-period multiple quantum well layer 203 made of InGaAlAs was formed as a modulator portion, and an InP spacer layer 204 and an InGaAsP diffraction grating layer 205 were continuously formed. Subsequently, the diffraction grating layer 205 was processed into a convex shape, and after removing the insulator mask, an insulator mask was formed again in a necessary portion by a photo process. Using this as a mask, unnecessary portions were removed by dry etching and wet etching. Subsequently, an InGaAsP waveguide layer 209 having a film thickness of 400 nm and a composition wavelength of 1.3 μm was formed.

Since the strain amount of the InGaAsP waveguide layer 209 was set to + 0.1%, no defects were generated and the surface state was good. Subsequently, after removing the insulator mask, the whole was buried with the p-InP clad layer 206, and finally the p ⁺ -InGaAs contact layer 207 was formed. Thereafter, a part of the contact layer was removed to separate the elements, and after mesa etching and substrate polishing, electrodes 201 and 208 were formed. After cleavage, reflection films were formed on both end faces to complete the element. The completed device had a threshold current of 15 mA, a range of 20 ° C to 85 ° C, and good modulation characteristics of 10 GHz.

  In the third embodiment, the present invention is applied to a 1.55 μm band spot expander integrated laser. As the growth method, the MOVPE method is used here as in the second embodiment, but the method is not limited thereto, and other methods may be used as long as the same effect can be obtained. FIG. 7 shows a cross-sectional structure of the element.

On the n-InP substrate 702, a lower n-InP clad layer 703 was laminated, and subsequently, a 10-cycle multiple quantum well active layer 708 made of InGaAlAs was formed as a laser part. Next, an insulator mask was formed in a necessary portion by a photo process. Using this as a mask, unnecessary portions were removed by dry etching and wet etching. Subsequently, an InGaAsP waveguide layer 704 having a composition wavelength of 1.3 μm was formed in a taber shape as a spot expansion portion so as to have a film thickness of 200 nm as a design value on a flat substrate. Since the strain amount of the InGaAsP waveguide layer 704 was set to + 0.1%, no defects were generated and the surface state was good. Subsequently, after removing the insulator mask, the whole was buried with a p-InP clad layer 705, and finally a p ⁺ -InGaAs contact layer 707 was formed. Thereafter, part of the contact layer was removed to separate the elements, and after mesa etching and substrate polishing, electrodes 701 and 706 were formed. After cleavage, a reflection film was formed on both end faces to complete the element. The completed device had a good threshold current of 5 mA, and FFP was good at 12 degrees in the vertical direction and 11 degrees in the horizontal direction.

1 is a structural diagram of a wavelength tunable laser according to the present invention. FIG. 3 is a structural diagram of a modulator integrated light source according to the present invention. (a)-(d) is the crystal growth process process drawing of the element shown in FIG. (a)-(c) are surface photographs of InGaAsP layers with different strains. Theoretical calculation diagram of immiscible region of InGaAsP layer. The figure which shows the relationship between the absolute value of the lattice mismatching degree of this invention, and a film thickness (the shaded area enclosed with ABC is this invention range). 1 is a structural diagram of a spot expander integrated laser according to the present invention. FIG.

Explanation of symbols

101 ... n-side electrode,
102… n-InP substrate,
103 ... InGaAsP waveguide layer,
104 ... InP spacer layer,
105 ... InGaAsP diffraction grating layer,
106… p-InP cladding layer,
107… p + -InGaAs contact layer,
108 ... p-side electrode,
109… InGaAsP multiple quantum well active layer,
201 ... n-side electrode,
202… n-InP substrate,
203… InGaAlAs multiple quantum well active layer,
204… InP spacer layer,
205 ... InGaAsP diffraction grating layer,
206… p-InP cladding layer,
207 ... p + -InGaAs contact layer,
208… p-side electrode,
209 ... InGaAsP waveguide layer,
210… InGaAlAs quantum well layer,
301… n-InP substrate,
302 ... InGaAsP multiple quantum well active layer,
303… Insulator mask,
304 ... InGaAsP waveguide layer,
305… p-InP spacer layer,
306 ... InGaAsP diffraction grating layer,
307 ... p-InP cladding layer,
308… p + -InGaAs contact layer,
701 ... n-side electrode,
702… n-InP substrate,
703 ... n-InP cladding layer,
704 ... InGaAsP waveguide layer,
705… p-InP cladding layer,
706 ... p-side electrode,
707… p + -InGaAs contact layer,
708 ... InGaAlAs multiple quantum well active layer.

Claims (4)

An InP substrate having a first region and a second region;
The first region on the InP substrate has an active layer that generates light ;
The second region on the InP substrate has an optical waveguide layer that propagates light from the active layer ;
The waveguide layer has an InGaAsP layer provided on the InP substrate with the optical axis aligned with the active layer ,
The InGaAsP layer has a composition wavelength in the range of 1.3 μm to 1.5 μm,
In the InGaAsP layer, each elemental composition of InGaAsP is adjusted so that the lattice mismatch (ε) with respect to the InP substrate is + 0.1% or more ,
The upper limit of the film thickness before Symbol InGaAsP layer, Ri critical thickness range near the lattice misfit dislocations are not generated, which is calculated from the lattice mismatch to be less than 0.1% (epsilon),
An optical semiconductor element , wherein the InGaAsP layer has a thickness of 150 nm or more . The composition wavelength of the InGaAsP layer is 1.4 μm ,
Before Symbol lattice mismatch with respect to the InP substrate InGaAsP layer (epsilon), the optical semiconductor device according to claim 1, wherein a is in the range of + 0.4% or less at + 0.1% or more. Forming a first optical waveguide layer on the InP substrate;
Selectively forming a first mask on a portion of the first optical waveguide layer;
Removing a region of the first optical waveguide layer where the first mask is not formed by etching to expose the surface of the InP substrate;
Forming a second optical waveguide layer by crystal growth of InGaAsP having a composition wavelength in the range of 1.3 μm to 1.5 μm on the InP substrate exposed by the etching ;
The upper limit value of the film thickness of the second waveguide layer is that the lattice mismatch degree (ε) when the lattice mismatch degree (ε) of the second waveguide layer with respect to the InP substrate is set to + 0.1% or more. Ri thickness range der lattice misfit dislocations are not generated, which is calculated from the epsilon),
The method of manufacturing an optical semiconductor element, wherein the InGaAsP layer has a thickness of 150 nm or more . The composition wavelength of the second optical waveguide layer is 1.4 μm ,
Lattice mismatch (epsilon) is for the previous SL InP substrate, a manufacturing method of the optical semiconductor element according to claim 3, wherein a is in the range of + 0.1% to + 0.4% or less.

Images