JPH09148684A - Semiconductor optical device and optical network using the same - Google Patents

Abstract

PROBLEM TO BE SOLVED: To restrain chirping by adjusting phases of the respective lights to be guided to optical waveguides in a Fabry-Perot resonator region and a distributed feedback type resonator (DFB) region. SOLUTION: The phase of a light which is outputted from a region 1 (distributed feedback type semiconductor laser), reflected by a region 3, and returned to the region 1 is adjusted by changing the refractive index of the region 2. When the light of the distributed feedback type semiconductor laser is intensive, the region 3 constituting a Fabry-Perot laser amplifies the light of DFB mode and consumes the gain, so that the oscillation is restrained. When the light of the DFB mode becomes weak, the consumption of gain in the region 3 decreases, and the gain is given to the Fabry-Perot mode. When pumping is sufficient, oscillation of Fabry-Perot mode is generated. By controlling the carrier injection amount to the respective regions, the threshold gains of both modes can be made equal, and further the switching of stable TM mode and TE mode of multimode is enabled.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a light source used for optical communication and optical information processing and an optical network using the same.

[0002]

2. Description of the Related Art With the increase in capacity of optical communication technology, many problems have been solved. However, the phenomenon called so-called chirping, in which the fluctuation of the refractive index due to the non-uniform carrier distribution in the semiconductor laser during high-speed intensity modulation distorts the oscillation waveform, has not always been solved. The mainstream methods currently used to avoid this problem are:
The semiconductor laser is driven continuously (CW), and the intensity is modulated by an external modulator made of a dielectric or a semiconductor.
With this method, there is a limit to the miniaturization and cost reduction of the device, and when this is used for an optical network, the flexibility of the optical network (such as high ability to simultaneously transmit signals with greatly different modulation speeds) It cannot be said that it is expensive.

On the other hand, as another method, there is a method of switching the polarization plane of the oscillation light of the device according to a signal, that is, a so-called polarization modulation method. This is disclosed in, for example, Japanese Patent Laid-Open No.
-42593, and JP-A-62-144426. The outline is as follows. As shown in FIG. 12, a semiconductor laser having a characteristic that polarization is inverted from TM mode to TE mode at a certain current value is used. Then, as a bias point current value TE and TM modes to oscillate simultaneously, the signal current (modulation current) I _s T
The threshold gains of E and TM are switched, and only the light polarized in a specific direction by the polarizer is sent out to the transmission line.

In the polarization modulation method, since the light density of the semiconductor laser is constant during the modulation (since the driving current is almost constant and is not turned on / off), the carrier fluctuation due to the modulation can be made extremely small. There are advantages. However, in the above conventional example, no specific structure of such a laser is specified.

On the other hand, as a concrete configuration example, there is the following two-electrode configuration. (1) An ordinary DFB laser has a two-electrode configuration, and carriers are non-uniformly injected to simultaneously control the phase and the gain. (2) By making the active layer a quantum well structure, the gain has polarization dependency, and the lack of gain in the TM mode is compensated. (3) Direct polarization modulation is performed by switching the TE mode and TM mode of the Bragg wavelength by phase control.

[0006]

In this method, it is essential to satisfy the oscillation (gain) condition and the phase matching condition of the discrete DFB mode. This reduces the degree of freedom in designing the resonator length, the grating structure, and the layer structure (in particular, the gain spectrum of the active layer), and requires a highly accurate manufacturing technique. As a result, it has been difficult to perform stable polarization modulation.

Therefore, it is an object of the present invention to provide a semiconductor laser which can suppress chirping and has a high degree of freedom in designing and manufacturing, and an optical network using the same.

[0008]

A semiconductor optical device of the present invention which achieves the above object has a composite resonator structure including at least a Fabry-Perot resonator region and a distributed feedback resonator (DFB) region, and has a polarization direction An optical waveguide in each resonator region having a plurality of waveguide modes orthogonal to each other, an optical coupling means for coupling the optical waveguides in each resonator region, a gain medium giving a gain to each resonator region, Phase adjusting means for adjusting the phase of each light guided to the optical waveguide in the resonator region.

More specifically, the following modes are also possible. The Bragg wavelength of the distributed feedback resonator region is set near the peak energy of the gain medium. Each of the optical waveguides in each resonator region is at least,
A slab-type optical waveguide comprising a first cladding layer, an active layer, an optical guide layer, and a second cladding layer, wherein the means for coupling the optical waveguides in each of the resonator regions aligns the optical guide layer in the cavity direction. Is to be common. The structure of the active layer serving as the gain medium is composed of a plurality of quantum wells, and the structure of the well layer or barrier layer of the quantum well or both is an asymmetric quantum well different from that of other quantum wells. The gain spectrum generated by the active layer is changed by changing the density of carriers injected into the active layer with respect to a plurality of waveguide modes in which polarizations are orthogonal to each other, which are induced by the optical waveguide including the plurality of gain modes. Control to the threshold oscillation gain of the guided mode. Strain stress is applied perpendicularly to the layer direction on at least one well layer or barrier layer constituting the asymmetric quantum well. The Bragg wavelength of the distributed feedback resonator region is set near the energy of the ground level of the asymmetric quantum well. The device is divided into a plurality of parts in the cavity direction, electrodes independent of each other are formed, and the carrier injection level is changed by injecting carriers into each of the divided regions independently.
For the stop band of the distributed feedback resonator region,
The Fabry-Perot resonator mode is set so that the axial mode spacing is sufficiently small. Two optical waveguides are formed in parallel, and the optical coupling means is an optical coupler means provided at at least one end of the Fabry-Perot resonator region and having a variable branching / merging ratio between the two optical waveguides. .

Further, the optical network of the present invention which achieves the above object, uses the above-mentioned semiconductor optical device, arranges a polarizer on the exit surface of the optical device, and selects only the DFB mode for use in communication. Is characterized by.

The gist of the present invention is as follows. At least in the case where an optical network is constructed using intensity modulation / demodulation, it is not necessary for the polarization-modulated light source to have low chirp for both TM and TE of the oscillation light, as long as either one is used as a signal. Good. Therefore, it is the aim of the semiconductor laser of the present invention to sacrifice the coherence of one mode (single-axis mode property) and instead improve the controllability and reliability of switching.

The basic structure is as follows. 1) A composite resonator including a region having a Fabry-Perot resonator (FP), a region having a DFB, and a phase adjusting region. 2) Controlling the resonator loss spectrum by optimizing or varying the coupling coefficient of the plurality of waveguides forming the composite resonator. 3) Applying a signal current or a signal voltage to the phase adjustment region to change the effective resonator length and selectively oscillate light with different polarization directions.

More specifically, the following is also possible. 4) In the case of an ordinary slab optical waveguide, use that the resonator loss of FP is always smaller in TE than in TM. It can be reversed. 5) By setting the pitch of the grating in the DFB region so that the Bragg wavelength becomes the gain peak wavelength, T
M mode controls gain and resonator loss spectrum so that DFB mode is selected. At the same time, by setting the axial mode interval of the resonator mode of FP to be smaller than that of the stop band of DFB, TE mode is FP,
Controlling gain and resonator loss spectrum so that DFB or both modes are selected simultaneously. 6) Conversely, TE mode is DFB mode, TM mode is F
Controlling the gain and resonator loss spectrum such that P, DFB or both modes are selected simultaneously. 7) Applying an asymmetric (strained) quantum well structure to the active layer to give polarization dependence to the gain spectrum.

The oscillation wavelength of the semiconductor laser is determined by the following oscillation condition formula. _{Γ · g th = Γ · α} in + α M tens of α _sc (1) _{Here, α M = l / 2L eff} · ln (1 / R 1 · 1 /
R ₂ ) exp (i · (2n _eff · L _eff / λ tens φ)) = 0 (2) where Γ: optical confinement coefficient in active layer g _th : threshold gain α _in : internal loss α _M : reflection Loss α _sc : Other losses (scattering loss, coupling loss, etc.) R _i : Effective reflectance in two directions viewed from one point in the resonator n _eff : Effective refractive index of the waveguide L _eff : Effective Resonator length λ: oscillation wavelength φ: phase.

In addition to this, the direct polarization modulation semiconductor laser has, as another condition, polarization modes independent of each other, for example, a TE mode and a TM mode, and their threshold gains are substantially equal, that is, , Γ _TE · g _thTE = Γ _TM · g _thTM (3) is required.

Since each variable has wavelength dependence and polarization dependence, it is not easy to satisfy all the above conditions (1) to (3). Especially, the expression (2) is an extremely important conditional expression that determines the oscillation mode (wavelength, polarization mode, transverse mode). In the conventional resonator configuration using DFB, the oscillation wavelength is relatively discrete with respect to the gain spectrum between the TE mode and the TM mode, which are oscillation modes, and therefore, (2) and (3) It was difficult to satisfy the formulas at the same time.

In the case of the composite resonator laser including the composite resonator of FP and DFB and the slab waveguide according to the present invention, the reflectance and its loss spectrum at any one point in the resonator are the coupling parameters of the resonator. (Coupling coefficient and phase between each waveguide) can be changed significantly. Furthermore, by utilizing the resonator distribution and polarization dependence of the gain, for example, the threshold gains of the TE mode and the TM mode of the Bragg wavelength can be made almost equal. Further, by providing a phase adjustment region to modulate the phase, feedback from another reflection end (for example, Fabry-Perot resonator) lowers the threshold for the TM-polarized DFB mode, for example, and stabilizes the TM-polarized DFB. Due to the situation that causes the oscillation of the mode and the feedback from the other reflection end (for example, the DFB resonator), the threshold for the DFB mode of the TM polarization is increased, and the DFB mode of the TE polarization and the Fabry-Perot mode (multimode It is possible to switch the situation that causes the oscillation of). As described above, it is the most characteristic part of the present invention that the TE and TM phase matching conditions are easily satisfied as compared with the conventional configuration.

The polarization dependence of the gain will be further described.
The asymmetric quantum well structure has a degree of freedom to change the bandgap and the profile as compared with the symmetric quantum well structure.
Furthermore, by introducing distortion asymmetrically, TE mode and TM
The band gap can be changed selectively with respect to the mode.
This means that the gain generated in the same active layer with the same injected carrier density can be set to the threshold gain of each of the TE mode and the TM mode. A polarization-switched optical output can be obtained by setting the threshold carrier density at the modulation bias point and superimposing a signal on the injection current or performing phase modulation.

[0019]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Embodiment 1 FIG. 1 is a sectional view in the resonator direction of a first embodiment of the present invention. The specific layer configuration is as follows. In FIG. 1, 101 is an n-type InP substrate, 102 is an n-type InP clad layer, 103 is an n-type InGaAsP optical guide layer, 10
4 is an undoped GaInAsP active layer, and 105 is a p-type I
The nP cladding layer 106 is a p-type InGaAsP contact (cap) layer. Further, 107 is a λ formed at the boundary between the n-type cladding layer 102 and the n-type light guide layer 103.
A grating having a / 4 phase shift part,
Reference numeral 8 is a positive electrode, and 109 is a negative electrode.

In FIG. 1, a region 1 is a DFB region, a region 2 is a phase adjustment region, and a region 3 is a Fabry-Perot resonator (F
P) area. By setting the coupling parameter κ · L in the DFB region to about 2 and setting the resonator length in the FP region to 300 μm, the axial mode interval of the Fabry-Perot mode is made sufficiently smaller than that in the stop band. Each region is electrically independent, but optically coupled via a light guide layer 103 aligned in the resonator direction. In the DFB region, the grating 107 is formed in the optical guide layer 103, and the grating pitch is set so that the Bragg wavelength of the TM mode matches the gain peak.

Both end faces in the FP region form Fabry-Perot resonators. In the case of this embodiment, one side is cleaved 11
2 and the other uses the etching surface (etching mirror) 110. In the case of this embodiment, the depth of the etching surface 110 is
Until the active layer 104 is reached. As a result, both functions of the reflecting mirror and the optical coupler can be provided. There is no gain medium in the phase adjustment region, and both end faces are obliquely cut so as not to form a resonator. End face 11 of DFB area
3 and the end face 112 of the FP region may be coated if necessary. In the case of this embodiment, the antireflection (AR) coat 113 is on the DFB side and the 50% reflection coat 1 is on the FP side.
Twelve.

Next, the operation principle of this embodiment will be described. FIG. 3 shows the gain spectrum and the resonance loss of each polarized light by focusing on the individual polarized lights (TE and TM) (the horizontal axis represents wavelength and the vertical axis represents gain or loss). In an actual device, (a) and (b) of FIG.
Is a state in which the graphs are overlapped, and the mode of the oscillating light is such that the resonator loss is lower than the gain. For simplicity, if the gain has the same spectral distribution regardless of the polarization state (for example, in the case of a bulk active layer), whether or not to oscillate depends only on the resonator loss. In the case of the present embodiment, it is important to adjust the phase of the light emitted from the region 1 (distributed feedback semiconductor laser) and reflected by the region 3 and returning to the region 1 by changing the refractive index of the region 2. .

In general, the waveguides have different equivalent refractive indices for TE-polarized light and TM-polarized light. In the case of the present embodiment, the equivalent refractive index for TE polarized light is about 0.2% larger than the equivalent refractive index for TM polarized light. Therefore, even if the physical distances to the reflecting surfaces are the same and the phases of the TE and TM polarized light are the same at the interface between the region 1 and the region 2, due to the difference in equivalent refractive index The distance (optical length) felt by is different, and the phase of the light returned by reflection is different. Moreover, the amount of change in the refractive index for each polarized light when the refractive index of the region 2 is changed by the plasma effect is also different. In the case of a semiconductor laser, the threshold value changes depending on the phase of reflected light. For example, if the phase of the reflected light differs from that of the emitted light by a half wavelength, they will cancel each other out if they have the same intensity, which has the effect of raising the threshold value. On the other hand, if they are in phase, resonance will be assisted and the threshold value will be reduced. In the Fabry-Perot laser, oscillation from the other reflection end (for example, DFB resonator) causes oscillation at a wavelength that is a phase condition that helps resonance.

In the case of a distributed feedback semiconductor laser, since the wavelength having a low resonator loss is determined by the diffraction grating, the return light from another reflection end (for example, Fabry-Perot resonator) changes the DFB mode threshold value. Can be made. On the other hand, the region 3 that constitutes the Fabry-Perot laser
When the light of the distributed feedback semiconductor laser is strong, the DFB mode light is amplified and the gain is consumed, so that the oscillation is suppressed. At this time, when the DFB mode light becomes weak, the consumption amount of the gain in the region 3 is reduced, the gain to the Fabry-Perot mode is given, and if sufficiently excited, the Fabry-Perot mode oscillation is generated. Become. In the present invention, when the DFB mode is switched to the Fabry-Perot mode, the polarization mode in which the distributed feedback semiconductor laser is oscillating is switched (for example, TM polarization is switched to TE polarization). By doing so, the TM-polarized DFB mode to the TE-polarized Fabry-Perot mode and the DF
It is possible to switch between the B mode and the mixed mode.

Therefore, by controlling the carrier injection amount in each region, not only the threshold gains of both modes can be made equal, but also stable TE mode and TE of multimode can be obtained.
Modes can be switched. Whether the LD as a whole oscillates in TE mode or TM mode,
Determined by the phase condition. In the phase adjustment region, the carrier is injected to change the effective refractive index by the plasma effect.

FIG. 4 shows current (I) vs. optical output (L) characteristics when a minute signal current is superposed with the above carrier injection amount as a modulation bias. A major feature of the conventional example is that the TE and TM modes can be switched with a slight change in current. As a result, the effect of chirping is hardly seen in combination with the fact that the total optical output does not change during modulation. Statically, both TE mode and TM mode are single axis modes. In the high speed modulation, the TE mode has multiple modes, but in the TM mode, not only the single axis mode but also the low chirp is maintained. By disposing a polarizer in front of the emitting end face and selecting only TM light, a low chirp optical signal can be extracted.

In this embodiment, the DFB mode is used as the TM mode and the FP / DFB multi mode is used as the TE mode, but the reverse is also possible. For example, by using the polarization dependence of the reflectance of the end face coating, it is possible to create a region in which the reflectance of TM is higher than that of TE with respect to the slab optical waveguide. In this case, the grating pitch is set so that the Bragg wavelength in the TE mode matches the gain peak.

Embodiment 2 FIG. 2 shows that the same operation and effect as those of the first embodiment can be achieved with a different structure. In FIG. 2, area 1 is a gain area, area 2 is a DFB area, and area 3 is a phase adjustment area. The end face 212 and the end face 213 form a Fabry-Perot resonator, and the grating 207 forms a second resonator. 201 is an n-type InP substrate, 20
2 is an n-type InP clad layer, 203 is an n-type InGaAs
P light guide layer, 204 undoped GaInAsP active layer, 205 p-type InP clad layer, 206 p-type In
It is a GaAsP contact layer. Reference numeral 207 is a grating having a λ / 4 phase shift portion formed at the boundary between the n-type cladding layer 202 and the n-type optical guide layer 203, 208 is a positive electrode, and 209 is a negative electrode.

In the case of this embodiment, the manufacture is easy, but
Since the coupling coefficient of the waveguide is almost 100%, the load on other parameters (coupling parameter κ · L, etc. in the DFB region) may become large. In this embodiment, the means for coupling a plurality of optical waveguides is to use a common light guide layer. The operation principle is substantially the same as that described in the first embodiment with reference to FIGS.

Embodiment 3 In conditional expressions (2) and (3), the characteristics can be further improved by controlling the profile of gain g. An embodiment in which the polarization dependence of gain is positively used will be described.

In general, when strain stress is applied to a quantum well, degeneration of the valence band is released, and a property considerably different from that of a strainless system appears. For example, in a normal quantum well, the gain of the TE mode is slightly larger than that of the TM mode, but this tendency is emphasized by giving compressive strain, and conversely, the gain of TM mode can be increased by giving tensile strain. it can. That is, the polarization dependence can be controlled. Further, since the differential gain (the ratio of the amount of change in carrier to the amount of change in gain) can be increased at the same time, it is known that the threshold value can be lowered and the modulation limit frequency can be improved.

FIG. 5 shows a schematic diagram of the band structure of the active layer. The photoluminescence peak wavelengths of the quantum well 1 and the quantum well 2 (which are quantum well layers on the low energy side and contribute to the gain of the TE mode) are 1.550 μm, respectively.
m and 1.565 μm. Strain (tensile strain) of 2.0% is introduced only in the quantum well 1 (the quantum well layer on the high energy side, which contributes to the gain of the TM mode). In this embodiment, the active layers 104 and 204 of the first or second embodiment are replaced with active layers having this band structure. The grating pitch in the DFB area is
At least one of the Bragg wavelengths of the two guided modes is
It is set so as to come close to the peak wavelength of the gain spectrum of the asymmetric quantum well.

As a result, although the gain is increased by the carrier injection, unlike the usual gain of the bulk crystal or the quantum well layer, the gain profile (gain spectrum) has a large carrier density dependency. The gain spectrum of Γ · g at a certain carrier density (injection current) is shown in FIG. Compared with the first and second embodiments, the threshold carrier density (threshold gain) is lower (due to the shape change of the energy band structure in the k space) and the modulation efficiency is higher (smaller current or voltage). Polarization can be modulated by changes, etc. (This is because the effective mass becomes smaller and the carriers move easily.), The polarization dependence of the gain spectrum can be increased, and the gain peak wavelengths of the TE mode and TM mode can be changed. Is the difference.

FIG. 7 shows the relationship between Γ · g (gain × confinement coefficient) and resonator loss at a given point in the state where carriers are injected in the vicinity of the threshold value of the present embodiment. It is shown separately for mode and. In the case of this embodiment, there is a merit that the gain profile can be optimized for each of TM and TE. The operation principle is substantially the same as that described in the first embodiment with reference to FIGS.

In this embodiment, tensile strain was introduced only into the quantum wells on the high energy side, but compressive strain was introduced into the quantum wells on the low energy side, or tensile strain and compressive strain were introduced into the quantum wells with high and low energy. It is also possible to introduce it. In a two-layer quantum well, saturation of gain may be a problem, so that the active layer may be an asymmetric multi-strained quantum well layer that is stacked in multiple stages.

Embodiment 4 In order to control the resonator loss spectrum, it is effective to introduce a variable branching ratio optical coupler into the composite resonator. As a branching ratio variable mechanism suitable for the present invention, Japanese Patent Application Nos. 5-80170 and 5-3527 filed by the present applicant.
There is one indicated by 09. In this embodiment, this branching ratio variable mechanism is applied to the second embodiment. FIG. 8 is a schematic sectional view in the resonator direction, and FIG. 9 is a sectional view perpendicular to the resonator direction. Since the branching / merging ratio can be controlled by the couplers in the regions 1 and 3, the control range of the resonator loss spectrum is expanded. That is, it is possible to widen the range in which polarization modulation is possible (the current, applied voltage, etc. in each region need to be adjusted as a pre-stage of polarization modulation, but the adjustment range becomes wider).

In FIG. 8, 5 is set with respect to the traveling direction of light.
It has a structure in which individual regions are connected in series. Area 1
And 3 are optical couplers, regions 4 and 5 are optical amplifiers, and region 2 is a phase adjuster. For example, in FIG.
00 is an n-type GaAs substrate, 301 is an n-type AlGaAs first cladding layer, 302 is an n-type AlGaAs first core layer,
303 is a p-type AlGaAs second cladding layer, 304 is an undoped AlGaAs second core layer, and 305 is an n-type AlG.
aAs third clad layer, 306 is a contact layer (GaA
s). 307a is a positive electrode, 307b and 308
Is a negative electrode, and 309 is a periodic current confinement layer for modulating the refractive index. In addition, the reflective coat 31
1 is given.

The manufacturing method of this embodiment will be briefly described. For example, ordinary metal organic chemical vapor deposition (MOCVD)
Method) or molecular beam epitaxial growth method (MBE method), first, the first clad layer 301 to the lower layer of the second clad layer 303 are grown on the n-type GaAs substrate 300. After growing the lower layer of the second cladding layer 303, n
After forming the type AlGaAs layer, a grating pattern with an appropriate pitch is etched so that the depth thereof reaches the p-type AlGaAs layer (the lower layer of the second cladding layer 303). After that, p-type AlG is continuously grown by MOCVD.
The upper layer of the aAs layer 303 is grown. As a result, the periodic current constriction layer 309 is produced. After this, undoped A
The lower layers of the 1GaAs second core layer 304 and the third clad layer 305 are grown to form the grating 310 in the region 4. Then, the upper layer of the third cladding layer 305 and the contact layer 306 are grown. In order to control the transverse mode, an embedded structure or the like is formed, and electrodes 307a, 307b,
This embodiment is completed by forming 308.

In the schematic cross-sectional view of FIG. 9, 320 is a p-type AlGaAs buried layer, 321 is an n-type buried layer, and 330 is a current path between the positive electrode 307a and the negative electrode 307b.

Next, the operating principle of this embodiment will be described. For example, when carriers are injected into the regions 1 and 3 (between the positive electrode 307a and the negative electrode 308), the carrier distribution is modulated by the periodic current constriction layer 309, and the carrier distribution is generated in the first core layer 302. , An equivalent refractive index distribution corresponding to this is formed. Therefore, the light guided in the two waveguides centering on the first and second core layers 302 and 304 is coupled by the periodic current confinement layer 309. The degree of coupling can be controlled by controlling the state of carrier injection into regions 1 and 3.
This corresponds to controlling the coupling coefficient of the waveguide in the DFB resonator region and the waveguide in the Fabry-Perot resonator region.
Therefore, current injection into regions 2, 4 and 5 (positive electrode 30
It is possible to oscillate in the TE mode or the TM mode based on the same operation principle as the above embodiment, in combination with the control of the current path 330 between the 7a and the negative electrode 307b. The optical coupler section may be provided on only one side.

Embodiment 5 Next, an example in which the above device is applied to an optical network will be described. FIG. 10 and FIG. 11 are application examples to a bus type optical network and a ring type optical network, and the above devices are mounted on optical nodes 401 to 406. By disposing a polarizer on the emission surface of the semiconductor laser described in the first to fourth embodiments, only specific polarized light (for example, TE light) can be extracted and sent out to the transmission line. Reference numeral 400 is an optical bus line, and 411 to 416 are terminal devices. The semiconductor laser of the present invention has a narrow line width and is stable even at high speed modulation.

[0042]

The effects of the present invention are as follows. 1) Since the total optical output hardly changes before and after polarization switching, there is almost no carrier fluctuation in the laser resonator, and chirping can be suppressed. 2) A light source capable of direct polarization modulation can be manufactured with high yield. 3) The degree of freedom in designing a device capable of polarization modulation is high.

[Brief description of the drawings]

FIG. 1 is a sectional view of a first embodiment in a resonator direction.

FIG. 2 is a sectional view of a second embodiment in the resonator direction.

FIG. 3 is a schematic diagram illustrating the relationship between the gain spectrum and the resonator loss spectrum of the first embodiment for each mode.

FIG. 4 is a graph showing a modulation characteristic of the example of the present invention.

FIG. 5 is a band diagram of an asymmetric strain quantum well.

FIG. 6 is a schematic diagram illustrating a gain spectrum of an asymmetric strain quantum well for each mode.

FIG. 7 is a schematic diagram illustrating a relationship between a resonator loss spectrum and a gain spectrum in each mode when an asymmetric strain quantum well is used.

FIG. 8 is a sectional view of a fourth embodiment in the resonator direction.

FIG. 9 is a lateral cross-sectional view of the fourth embodiment.

FIG. 10 is a diagram showing an example in which the semiconductor optical device of the present invention is applied to an optical network.

FIG. 11 is a diagram showing an example in which the semiconductor optical device of the present invention is applied to an optical network.

FIG. 12 is a diagram illustrating a conventional example.

[Explanation of symbols]

101, 201, 300 Substrate 102, 105, 202, 205, 301, 305
Cladding layer 103, 203 Optical guide layer 104, 204 Active layer 106, 206, 306 Cap layer 107, 207, 310 Phase shift grating 108, 208, 307a Positive electrode 109, 209, 307b, 308 Negative electrode 110 Etching mirror 112, 212 213 coating surface 113 AR coating 302 first core layer 304 second core layer 309 periodic current constriction layer 320 p-type buried layer 321 n-type buried layer 330 current path 400 optical bus line 401-406 optical node 411-416 terminal device

Claims (10)

[Claims] 1. An optical waveguide in each resonator region having a composite resonator structure comprising at least a Fabry-Perot resonator region and a distributed feedback resonator (DFB) region and having a plurality of waveguide modes whose polarization directions are orthogonal to each other. An optical coupling means for coupling the optical waveguides of the respective resonator regions, a gain medium that gives a gain to the respective resonator regions, and a phase of each light guided in the optical waveguides of the respective resonator regions. A semiconductor optical device comprising a phase adjusting means for adjusting. 2. The semiconductor optical device according to claim 1, wherein the Bragg wavelength of the distributed feedback resonator region is set near the peak energy of the gain medium. 3. Each of the optical waveguides in each of the resonator regions is a slab type optical waveguide including at least a first cladding layer, an active layer made of a gain medium, an optical guide layer and a second cladding layer. 3. The semiconductor optical device according to claim 1, wherein the means for coupling the optical waveguides in the resonator region is to align or share the optical guide layer in the resonator direction. 4. Each of the optical waveguides in each resonator region is a slab type optical waveguide including at least an active layer formed of a gain medium, and the structure of the active layer formed of the gain medium includes a plurality of quantum wells. , And the structure of the well layer and / or the barrier layer of the quantum well is an asymmetric quantum well different from that of the other quantum wells, and a plurality of polarizations orthogonal to each other induced in the optical waveguide including the active layer are provided. The gain spectrum generated by the active layer is controlled to the threshold oscillation gain of the plurality of waveguide modes by changing the density of carriers injected into the active layer with respect to the waveguide mode. The semiconductor optical device according to claim 1, 2 or 3. 5. The semiconductor optical device according to claim 4, wherein at least one well layer or barrier layer constituting the asymmetrical quantum well is subjected to strain stress perpendicular to the layer direction. 6. The semiconductor optical device according to claim 4, wherein the Bragg wavelength of the distributed feedback resonator region is set near the energy of the ground level of the asymmetric quantum well. 7. The carrier injection level is changed by dividing the device into a plurality of parts in the cavity direction, forming electrodes independent of each other, and injecting carriers into each of the divided regions independently. The semiconductor optical device according to any one of claims 1 to 6. 8. The axial mode interval of the Fabry-Perot resonator mode is set to be sufficiently small with respect to the stop band of the distributed feedback resonator region, according to any one of claims 1 to 7. The semiconductor optical device according to 1. 9. Two optical waveguides are formed in parallel, and the optical coupling means has a variable branching / merging ratio between the two optical waveguides provided at at least one end of the Fabry-Perot resonator region. 9. The semiconductor optical device according to claim 1, which is an optical coupler means. 10. The semiconductor optical device according to claim 1, wherein a polarizer is arranged on the emission surface of the optical device, and only the DFB mode is selected for use in communication. Optical network to do.