Optical Circuit
The optical circuit with simulated mesa structures in the window region of AXEL chips addresses the issue of reflected return light and cleavage position variations, enhancing output and quality transmission while increasing manufacturing yield and reducing optical loss.
Patent Information
- Application Number
- JP2023522060
- Authority / Receiving Office
- JP · JP
- Patent Type
- Patents
- Current Assignee / Owner
- Filing Date
- 2021-05-18
- Publication Date
- 2024-11-20
- Estimated Expiration
- 2041-05-18
AI Technical Summary
The integration of a semiconductor optical amplifier (SOA) in EADFB lasers (AXEL) leads to increased reflected return light, affecting operating characteristics and reducing manufacturing yield due to variations in cleavage position during the cleavage process, which is necessary for forming the window structure on the chip end face.
An optical circuit with integrated semiconductor laser and SOA, featuring a window region with simulated mesa structures parallel to the optical axis, thickening the buried InP layer to reduce optical loss and provide a margin for cleavage position variations, using a configuration that includes a plurality of simulated mesas to minimize the area for InP deposition during regrowth.
The solution achieves high output and quality transmission characteristics by stabilizing optical output levels and waveform quality, improving manufacturing yield and reducing optical loss, even with manufacturing errors in cleavage positions.
Smart Images

Figure 0007587189000003 
Figure 0007587189000004 
Figure 0007587189000005
Abstract
Description
[Technical field]
[0001] The present invention relates to optical circuits, and more particularly to optical transmitters. [Background technology]
[0002] In recent years, network traffic has been growing explosively due to the spread of video distribution services and the growing demand for mobile traffic. In the optical transmission lines that support the network, the trend is to reduce the cost of the network by increasing the transmission rate, reducing power consumption, and extending the transmission distance. Semiconductor modulated light sources are also required to achieve high speed and high output while suppressing the increase in power consumption. Distributed feedback (DFB) lasers (hereinafter referred to as EADFB lasers) integrated with electro-absorption (EA) modulators have high extinction characteristics and excellent chirp characteristics compared to directly modulated lasers, and have been used in a wide range of applications.
[0003] 1 is a diagram showing a schematic configuration of a general EADFB laser. An integrated EADFB laser 40 has a structure in which a DFB laser 10 and an EA modulator 20 are integrated on the same chip. The DFB laser 10 has an active layer made of multiple quantum wells (MQWs). The EA modulator 20 has a layer 1 and oscillates at a single wavelength due to a diffraction grating 3 formed in the resonator. The EA modulator 20 has a light absorption layer 2 made of MQW with a different composition from that of the DFB laser, and the amount of light absorption in the light absorption layer 2 is changed by voltage control by a modulation signal source 12. The EA modulator 20 is driven under conditions that allow the output light from the DFB laser 10 to pass or absorb, causing the light to blink, and the electrical signal is converted into a modulated optical signal 4.
[0004] One of the problems with EADFB lasers is that it is difficult to achieve high output power because the EA modulator involves a large optical loss. As a solution to this problem, an EADFB laser (SOA Assisted Extended Reach EADFB Laser: AXEL) has been proposed, in which a semiconductor optical amplifier (SOA) is further integrated at the light output end of the EADFB laser (Non-Patent Document 1).
[0005] FIG. 2 shows the schematic configuration of AXEL, which integrates an SOA in an EADFB laser. In AXEL45, the signal light modulated by the EA modulator 20 is amplified by the integrated SOA region 30 to obtain the signal light 4. In AXEL45, the output power is about twice as high as that of a typical EADFB laser. Due to the highly efficient operation achieved by the SOA integration effect, AXEL can reduce power consumption by 40% when driven under the operating conditions that obtain the same optical output as a typical EADFB laser. In addition, the active layer of the SOA in AXEL uses the same MQW structure as that of a DFB laser. Therefore, there is no need to add a new regrowth process for the integration of the SOA region, and AXEL can be manufactured using the same manufacturing process as that of a conventional EADFB laser.
[0006] One of the issues facing AXEL is the degradation of operating characteristics due to reflected back light, which is caused by the high optical output characteristics associated with SOA integration. In optical transmitters such as general semiconductor lasers, light that is reflected at the semiconductor chip end face and reflected back into the chip adversely affects the operating characteristics of the device. To deal with reflected back light, semiconductor optical transmitters apply an anti-reflection (AR: Anti-Reflection) coating to the chip end face, which generally reduces reflected back light from the chip end face to zero. However, in the case of an EADFB laser (AXEL) integrated with an SOA, even a small amount of reflected light has a significant effect on the operating characteristics due to its high output.
[0007] Consider as a specific example the case where the optical amplification effect (gain) of the AXEL SOA is +3 dB compared to a conventional standalone EADFB laser. In AXEL, the average optical output is increased by 3 dB through the integration of the SOA, which results in an increase of 3 dB in the absolute value of the reflected return light intensity. In addition, the reflected return light at the chip end surface propagates in the reverse direction again within the SOA and is amplified. For this reason, the reflected return light intensity reaching the DFB laser section 10 in AXEL in Figure 2 is increased by 6 dB compared to a standalone EADFB laser. As a countermeasure against the reflected return light caused by the high output of AXEL described above, a structure combining a window structure section and a curved waveguide in addition to the AR coating is adopted.
[0008] FIG. 3 is a diagram for explaining a configuration for reducing the influence of reflected back light in AXEL. FIG. 3 is a top view showing an enlarged view of the vicinity of the output end face 55 of the AXEL semiconductor chip 50. To be precise, FIG. 3 is a cross-sectional view of the inside of the AXEL chip cut by a plane that passes through the cores of the optical waveguides 51 and 52 and is parallel to the substrate surface. It should be noted that the top view in the following description of this specification includes the case of a cross-sectional view. Usually, the output end face of a semiconductor chip is a crystal plane formed by cleavage, and the waveguide in the semiconductor chip is formed at a perpendicular angle to this output end face. Therefore, the light propagating through the optical waveguide enters the output face perpendicularly and is output from the semiconductor chip.
[0009] On the other hand, in the AXEL chip structure shown in FIG. 3, a bent portion 52 is provided in an optical waveguide 51 in the chip, and the bent portion 52 is inclined at an inclination angle 54 (θ wg ) is incident on the waveguide. By providing the inclination angle 54, the light reflected at the exit end face 55 is less likely to be coupled back into the waveguide, so that reflected back light can be suppressed. Generally, the inclination angle θ wg Typically, 4 to 8 degrees is used.
[0010] Furthermore, in the AXEL chip structure shown in FIG. 3, the optical waveguide 51 passes through the bent portion 52 and terminates inside the output end face 55 of the chip 50. The light emitted from the optical waveguide propagates through a bulk semiconductor called the window region 53, reaches the output end face 55, and is output to the outside of the chip. At this time, in the window region 53, the output light propagates while expanding its beam diameter due to the diffraction effect. The expansion of the beam diameter can further reduce the proportion of reflected back light that is reflected at the output end face 55 and recouples with the waveguide inside the chip. The window region 53 is usually fabricated with a length of about 10 μm. In order to obtain high-quality transmission characteristics as well as high-output characteristics in AXEL, sufficient measures against reflected back light must be taken, such as the configuration shown in FIG. 3. [Prior art documents] [Non-patent literature]
[0011] [Non-Patent Document 1] W Kobayashi et al., “Novel approach for chirp and output power compensation applied to a 40-Gbit / s EADFB laser integrated with a short SOA,” Opt. Express, Vol. 23, No. 7, pp. 9533-9542, Apr. 2015 Summary of the Invention [Problem to be solved by the invention]
[0012] However, the above-mentioned structure for reducing the effect of reflected light near the chip end face in AXEL has a problem of reduced chip yield due to manufacturing variations when forming the window structure. The window structure formed on the light-emitting end face is fabricated by cleavage. In the cleavage process of an actual semiconductor chip, a cleavage position error of about ±10 μm generally occurs. In AXEL having a window structure, in order to manufacture an optical transmitter with sufficiently high output and high quality characteristics, it is necessary to control the variation in the cleavage position with sufficient precision. With the current AXEL chip end structure, it is not possible to provide a sufficient margin for the cleavage position variation when forming the emission end face. Due to manufacturing errors in the cleavage position, a certain number of defective products are always produced in AXEL chips, reducing the manufacturing yield. Similar problems can occur in other optical circuits that include cleavage in the manufacturing process and window structures.
[0013] The present invention has been made in consideration of such problems, and provides an optical circuit that achieves high output characteristics and high-quality transmission characteristics regardless of variations in the cleavage process. [Means for solving the problem]
[0014] One aspect of the present invention is an optical circuit in which a semiconductor laser and a semiconductor optical amplifier (SOA) are integrated on a substrate, the optical circuit comprising: a waveguide optically connected to the SOA and terminated on the inner side relative to an end face of the substrate; and a window region through which light emitted from an end face of the waveguide propagates to the end face, the window region including a plurality of simulated mesas configured from the end face to the end face, parallel to an optical axis of the light and spaced a distance M from the optical axis, and the window region being filled with bulk semiconductor except for the simulated mesas.
[0015] The above-mentioned semiconductor laser may be a distributed feedback laser (DFB laser) integrated with an electroabsorption modulator, the optical circuit may be an optical transmitter, and the waveguide may be a passive waveguide made of a core layer having a bandgap wavelength shorter than the oscillation wavelength of the DFB laser. Effect of the Invention
[0016] To achieve high output characteristics and high quality transmission characteristics of optical circuits, particularly optical transmitters. [Brief description of the drawings]
[0017] [Figure 1] FIG. 1 is a diagram showing a schematic configuration of a general EADFB laser. [Diagram 2] This is a diagram showing the schematic configuration of an AXEL in which an SOA is integrated into an EADFB laser. [Diagram 3] FIG. 13 is a diagram showing a configuration for reducing the influence of reflected back light in AXEL. [Figure 4] FIG. 1 is a diagram illustrating a bar structure for creating a window region at an end face of a device. [Diagram 5] 1A and 1B are diagrams for explaining degradation of quality of a transmitted optical signal due to an error in cleavage position. [Figure 6] 13 is a diagram showing the shape of a light beam near an end face when the window region length L is increased. FIG. [Figure 7] 1A to 1C are diagrams illustrating the manufacturing process of the upper cladding layer in AXEL. [Figure 8] FIG. 1 is a diagram showing a schematic configuration of an AXEL having a window structure according to a conventional technique. [Figure 9] FIG. 13 is a diagram showing the relationship between cleavage position variation and loss in an AXEL having a conventional configuration. [Figure 10] 1A and 1B are diagrams illustrating a schematic configuration of a window structure of an optical transmitter according to the present disclosure. [Figure 11] 1A and 1B are diagrams illustrating the configuration of a window structure portion after cleavage in an optical transmitter according to the present disclosure. [Figure 12] FIG. 13 is a diagram showing the relationship between the simulated mesa position from the optical axis and the amount of InP thickening. [Figure 13] FIG. 13 is a diagram showing the relationship between the propagation distance of light in a window region and loss caused by diffraction. [Figure 14] 2 is a diagram showing a schematic configuration of a window structure of an optical transmitter according to a first embodiment. FIG. [Figure 15] 4 is a diagram showing the vicinity of a window region before embedding and regrowth in the optical transmitter of the first embodiment. FIG. [Figure 16] 4 is a diagram showing the vicinity of a window region after embedding and regrowth in the optical transmitter of the first embodiment. FIG. [Figure 17] FIG. 11 is a diagram showing the relationship between cleavage position variation and loss in Example 1. [Figure 18] 6 is a diagram showing the relationship between cleavage position variation and optical waveform quality in the first embodiment. FIG. [Figure 19] 1A and 1B are diagrams showing a comparison of the arrangements of optical circuit elements of optical transmitters according to first and second embodiments. [Figure 20] 13 is a diagram showing a schematic configuration of a window structure portion of an optical transmitter according to a second embodiment before cleavage. FIG. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0018] The optical circuit of the present disclosure provides a new optical transmitter configuration that achieves both high output and high-quality transmission characteristics regardless of variations in cleavage position. The optical circuit of the present disclosure can be an optical transmitter having an AXEL configuration in which an EADFB laser and a semiconductor optical amplifier (SOA) are integrated. In a window structure portion of the chip emission end face of the AXEL, a partially thickened bulk semiconductor layer is formed by a structure including a simulated mesa configured parallel to the optical axis of the optical waveguide. In the following, first, the effect of variations in cleavage position at the end of the chip on the characteristics of the AXEL is described, and then the configuration and operation of the optical circuit of the present disclosure are described.
[0019] Hereinafter, in the description of this specification, the term "AXEL" refers to an optical transmitter that integrates an EADFB laser and a semiconductor optical amplifier (SOA), and is synonymous with "optical transmitter."
[0020] FIG. 4 is a diagram illustrating a bar structure for producing a window region at a device end face. In an optical semiconductor device such as a semiconductor laser, multiple chips having the same emission end face are cleaved into a bar shape to form the emission end face. By carrying out the cleavage process, a bar is formed, which has a common cleavage plane and multiple chips are lined up along the cleavage plane. The bar is further divided into cleavage planes. The bars are then cut vertically to form individual device chips. Adjacent bars are arranged so that their emission end faces face each other.
[0021] The semiconductor device 60 in Fig. 4 includes two bars 61a, 61b, each of which includes three optical circuits 61-1 to 61-3. The opposing chips have window structures 53a, 53b, respectively, and the ends of the optical waveguides 51a, 51b are located at a position separated by the window region length L from the planned cleavage position 62. The semiconductor device in Fig. 4 shows only the vicinity of the end face of each chip including the final planned cleavage position 62, and each chip shows a part of an AXEL.
[0022] In the cleavage process of a semiconductor chip, there is a limit to the accuracy of cleavage, and even if the planned cleavage position 62 in FIG. 4 is determined, the actual cleavage position will vary to a certain extent. In semiconductor chips having a window structure manufactured by cleavage, the window region length L varies from chip to chip. As described above, in AXEL, the window region length is designed to be about 10 μm. Referring to FIG. 4, if an error of 10 μm or more occurs in the actual cleavage position with respect to the planned cleavage position 62, the window region of one of the two bars 61a, 61b will disappear. If the window region disappears, the effect of suppressing reflected back light will be insufficient.
[0023] Figure 5 is a diagram explaining the degradation of quality of transmitted optical signals due to errors in cleavage position. The graph shows the results of evaluating the optical waveform quality of transmitted optical signals when multiple AXEL chips produced using the same process were mounted on modules. The horizontal axis shows the deviation (μm) of the actual cleavage position of an AXEL chip produced using a manufacturing process that includes a cleavage process, and the vertical axis shows the mask margin (%) of the transmitted optical signal waveform from an optical transmitter module that uses that chip.
[0024] The multiple AXEL chips fabricated here have the same structure and an oscillation wavelength in the 1.3 μm band. The window region length L = 10 μm is the design value, and the only difference between the chips is the window region length L, which varies due to manufacturing errors. Before assembling each chip into a module, the window region length L was measured. After measuring L, each chip was mounted on a typical butterfly-type semiconductor module consisting of a two-lens system. The butterfly-type package is equipped with a high-frequency connector, and each fabricated module was modulated with a 25 Gbit / s NRZ signal to evaluate the optical waveform (EYE waveform) of the transmitted optical signal. The mask margin on the vertical axis of the graph in Figure 5 is a representative index of the quality of the optical waveform, and the larger the margin value (%), the clearer the eye opening and the better the quality of the modulated signal.
[0025] As is clear from the graph in Figure 5, the more the cleavage position is shifted in the negative direction, that is, the smaller the window region length L, the more the mask margin deteriorates. This deterioration in mask margin is thought to occur because as the window region length L becomes smaller, the intensity of light reflected back from the chip end face to the inside increases, destabilizing the AXEL oscillation operation. From the relationship between the window region length (cleavage position deviation) and the transmitted optical signal quality shown in Figure 5, it can be seen that the window region length L needs to be at least 5 μm. However, as will be explained next, there is an upper limit to the window region length L, and if the window region length L is too large, it will have an adverse effect on the AXEL device characteristics.
[0026] FIG. 6 is a diagram showing the shape of a light beam near the exit end face when the window region length L is increased. FIG. 6(a) shows a top view of the substrate surface (xy plane) near the exit end face of the AXEL chip 50, and FIG. 6(b) shows a cross-sectional view cut perpendicular to the substrate surface (xy plane) along the line VIb-VIb. In FIG. 6(a), the optical waveguide 51 inside the chip and the curved waveguide section 52 at the tip are seen. The light emitted from the curved waveguide section 52 reaches the cleavage surface at the chip end face as a beam 57a. From the end of the curved waveguide section 52 to the cleavage surface is a window region 53 with a length L. Referring to FIG. 6(b), the light having a profile 58 emitted from the core of the curved waveguide section 52 at height d reaches the cleavage surface as a beam 57b.
[0027] In an optical transmitter for a communication wavelength band (C, L, O band) of 1.55 μm band or 1.3 μm band, when the core material is InGaAsP, the thickness of the core layer is about 200 nm to 300 nm. Usually, in an InP-based optical semiconductor device, the optical confinement is stronger in the vertical direction (z direction) of the waveguide than in the horizontal direction (xy plane) of the optical waveguide. Therefore, the beam spread in the window region has a larger spread angle in the vertical direction of the waveguide. Therefore, when the length L of the window region 53 is long, the upper end of the beam 57b in the vertical direction (z direction) of the waveguide reaches the interface 59 between the cladding layer and the outside of the semiconductor (air or electrode). If a part of the beam reaches the interface 59 of the chip, it causes a defect (shading) in the beam shape or an optical loss. In the beams 57a and 57b shown in FIG. 6, for example, the part attenuated by a certain percentage (for example, 1 / 2) from the peak intensity at the beam center is shown as the beam edge. Therefore, light is distributed even outside the boundary of beam 57b in FIG. 6(b), and light that passes over interface 59 of the upper cladding layer with thickness h is emitted and scattered outside the chip.
[0028] In a typical InP-based optical semiconductor device, the cladding layer above the optical waveguide is formed by regrowth, and the thickness h of the upper cladding layer is about 2 μm. As explained in Figure 6, when providing a window structure at the edge of the chip, the thickness h of the upper cladding layer must be designed taking into account the beam spread in the vertical direction of the waveguide caused by the diffraction effect of the light beam. In AXEL, the thickness of the upper cladding layer in the window region at the edge of the chip is an issue. Therefore, we first examine the manufacturing process of the upper cladding layer in AXEL.
[0029] FIG. 7 is a diagram for explaining a part of the manufacturing process of the upper cladding layer in AXEL. FIG. 7 shows a cross-sectional view taken perpendicular to the light propagation direction of the optical waveguide, with (a) showing the mask formation process, (b) showing the mesa formation process, and (c) showing the burying layer growth process. As also shown in FIG. 6, in the window region 53, the optical waveguide is terminated at the bent portion 52, forming an InP region. The InP in the window region 53 is formed simultaneously during the growth process of the upper cladding layer or the burying growth process. FIG. 7 explains the case where the window region 53 is formed during the burying growth of InP.
[0030] As explained in FIG. 2, the AXEL waveguide has a common buried structure in the laser section 10, the EA modulator section 20, and the SOA section 30. To fabricate this waveguide structure, first, as shown in FIG. 7(a), an insulating film is formed on an initial substrate and patterned to form an insulating film mask 64 in the shape of a waveguide. The initial substrate is formed by successively growing a substrate 60, a lower cladding layer 61, an active layer 62, and an upper cladding layer 63. Thereafter, as shown in FIG. 7(b), a mesa formation process is performed by a dry etching or wet etching process to form a waveguide mesa section 65. Next, as shown in FIG. 7(c), an InP layer 66 that functions as a current blocking layer is regrown to bury the periphery of the waveguide mesa section 65. Finally, the insulating film mask 64 is removed, and a buried hetero (BH) is formed. The structure is complete. The window structure with the window region at the edge of the chip is also formed at the same time as the BH structure is fabricated. The problem with the window structure is that the thickness of the InP in the window region is thinner than the thickness of the InP in the upper cladding of the waveguide section, and it is particularly likely to affect the optical field.
[0031] FIG. 8 is a diagram showing a schematic configuration of an AXEL having a window structure according to a conventional technique. FIG. 8 shows a schematic configuration of the vicinity of the output end surface of an AXEL having a window structure formed by burying and regrowth. FIG. 8(a) shows the window structure of the AXEL as viewed from above the substrate, and is a cross section cut along the xy plane including the center (optical axis) in the height direction of the optical waveguide 51 in FIG. 8(c). FIG. 8(b) is a cross section cut along the yz plane including the line VIIIb-VIIIb in FIG. 8(a) in the waveguide region of the AXEL, and FIG. 8(d) is a cross section cut along the yz plane including the line VIIId-VIIId in FIG. 8(a) in the window region. FIG. 8(c) shows the VIIIc- FIG. 8 is a cross-sectional view taken along the xz plane including line VIIIc.
[0032] In order to simplify the explanation, FIG. 8 illustrates a straight waveguide 51 without a bent waveguide at the end. As shown in the cross-sectional view of the waveguide region in FIG. 8(b), buried regrown InP 66, which is a current blocking layer, is formed on both sides of the waveguide mesa, and the waveguide core is buried with the current blocking layer InP 66. As shown in FIG. 8(c) and FIG. 8(d), the portion that becomes the window region of length L is also etched at the same time as the fabrication of the waveguide mesa, and the same InP layer as the current blocking layer 66 is buried in the burying regrowth process. In a normal burying regrowth process, the thickness of the regrown InP layer is set according to the height of the waveguide mesa so that the mesa structure is completely buried. This is because if the buried regrown InP layer is made too thicker than the waveguide mesa, the InP layer 66 will pile up and be deposited on the insulating film 64 shown in the fabrication process of the upper cladding layer in FIG. 7.
[0033] In contrast to the above-mentioned waveguide region, the window region is a region where the waveguide 51 terminates and no mesa shape exists, as shown in the top view of FIG. 8(a), and therefore the area and volume of the base portion to be filled during the regrowth of the InP layer are larger. If the waveguide region and the window region are filled with InP simultaneously in one regrowth process due to the difference in the size of the buried space, the InP layer 66a in the window region becomes thinner than the InP layer 66 in the waveguide region, as shown in FIG. 8(c). As explained in the light beam shape of FIG. 6, the light field propagates while spreading due to the diffraction phenomenon in the window region, and the spread in the direction perpendicular to the waveguide is larger. If the InP film thickness becomes thin in the window region, the light field is likely to come into contact with the interface between the InP film and air, resulting in light loss.
[0034] Figure 9 shows the relationship between cleavage position variation and loss in the conventional AXEL. As in the mask margin evaluation in Figure 5, we first measured the deviation of the cleavage position from the design value for an AXEL chip fabricated using a normal buried regrowth process. Using the AXEL chips with the measured deviation, we fabricated several AXEL modules and evaluated the optical loss in the modules. The horizontal axis of Figure 9 shows the deviation of the cleavage position (μm), and the vertical axis shows the loss (dB) generated in the optical module. The plot points in Figure 9 correspond to the plot points used in the mask margin evaluation in Figure 5. AXEL chips with the same structure are used in each module, and the design value of the window region length L is 10 μm.
[0035] The loss within the optical module is calculated as follows from the difference between the optical output level of the chip alone and the optical output level when mounted in the module. The optical output level of the chip alone before mounting in the module was measured for each AXEL chip using a large-aperture photodetector. The AXEL chip was then mounted in an optical module, and the optical output intensity coupled to an optical fiber was measured. The difference between the two optical intensities makes it possible to estimate the optical loss caused by mounting in the module. The optical module used in this measurement had a two-lens system, and the AXEL chip was mounted in the module using an active alignment process.
[0036] The relationship between cleavage position variation and loss in Figure 9 shows a tendency for loss inside the module to increase when the window region length L is longer than the design value (deviation amount 0). This increase in loss occurs because, as shown in Figure 6, as the window region length L becomes longer, the upper end of the emitted beam in the direction perpendicular to the waveguide reaches the interface 59 between the cladding layer and the outside of the chip, causing a defect in the beam. Part of the beam that reaches interface 59 becomes light that is emitted and scattered outside the chip, reducing the coupling efficiency from the chip end face to the optical fiber and increasing the loss inside the module.
[0037] From the relationship between the amount of deviation in the cleavage position and loss shown in Figure 9, in order to keep the optical loss within the module at 3.0 dB or less and to create an optical module with a stable output level, the upper limit of the window region length L of the AXEL chip must be 15 μm or less. In other words, the deviation of the cleavage position from the design value of 10 μm must be +5 μm or less. As already mentioned, from the relationship between the window region length L and the transmitted optical signal quality shown in Figure 5, the window region length L must be at least 5 μm, and the deviation of the cleavage position must be -5 μm or more. This requires an accuracy of ±5 μm for the cleavage position.
[0038] However, in the actual cleavage process of semiconductor chips, an error of about ±10 μm usually occurs in the cleavage position. In AXEL, which has a window structure, it becomes impossible to provide a sufficient margin when forming the output facet in order to manufacture an optical transmitter with sufficient transmission signal quality and optical output level. As a result, the manufacturing error in the cleavage position causes the window region length to deviate from the allowable value, resulting in a certain number of defective chips and reducing the manufacturing yield.
[0039] As mentioned above, the window region length L required to ensure the required characteristics of an AXEL optical transmitter is in the range of 5 μm to 15 μm. This range of the window region length L generally applies in the C-band, L-band, and O-band wavelength bands as long as the core layer at the end of the waveguide is InGaAsP with a band gap wavelength of 1.1 to 1.4 μm.
[0040] The optical transmitter disclosed herein presents a novel configuration in which the buried InP layer is locally thickened in the window region of the AXEL chip. By locally thickening the buried layer in the window region, the optical field can propagate in the direction perpendicular to the waveguide without reaching the chip surface, reducing optical loss in the window region within the chip. Even if the window region length varies from chip to chip due to variations in cleavage position, a stable optical output level and optical waveform quality can be achieved, increasing the yield of AXEL modules.
[0041] The optical transmitter of the present disclosure has multiple "simulated mesa structures" in a window region near the output end face, away from the optical axis. The simulated mesa structures reduce the area of the semiconductor portion that serves as a base for depositing the buried layer in the window region, thereby relatively increasing the growth rate of InP during regrowth, making it possible to make the film thicker than other regions on the wafer, such as the waveguide region. The configuration, fabrication method, and effects of the optical transmitter of the present disclosure will be described below with reference to the drawings.
[0042] FIG. 10 is a diagram for explaining a schematic configuration of a window structure of an optical transmitter of the present disclosure. FIG. 10 is a schematic diagram of the vicinity of two window regions of an AXEL chip 100 in a state in which two adjacent AXEL chips are arranged facing each other before cleavage as shown in FIG. 4. In FIG. 10, in order to simplify the explanation, a window structure in which light propagates from a waveguide without a bent portion at the tip is explained. One window structure portion 108a is in contact with a window structure portion 108b of an adjacent chip, and the two window structures are separated by cleavage to become one AXEL chip. (a) of FIG. 10 is a view of the window structure portion seen from above the substrate surface, and (c) is a cross section cut in an xy plane including the Xa-Xa line. (b) of FIG. 10 is a cross section cut in a yz plane including the Xb-Xb line of (a) in the waveguide region of AXEL, and (d) is a cross section cut in a yz plane including the Xd-Xd line of (a) in the window region. FIG. 10(c) is a cross-sectional view taken along the xz plane including the line Xc-Xc in FIG. 10(a) along the optical waveguides 101a and 101b.
[0043] The characteristic configuration of the optical transmitter of the present disclosure in FIG. 10 becomes clear by comparing it with the corresponding configurations of the prior art in FIG. 8. Referring to FIG. 10(a), the AXEL chip 100 has simulated mesa structures 102-1 and 102-2 in the window region at positions on the extension of the optical axis of the terminated waveguide 101a and at offset distances from the optical axis. Referring to FIG. 10(b) and FIG. 10(d), the simulated mesa structures 102-1 and 102-2 have the same layer configuration as the mesa structure including the waveguide core. The window structure portion having the simulated mesa structure in FIG. 10 is in contrast to the window structure portion in the prior art AXEL shown in FIG. 8, which does not have anything except the InP buried layer up to the cleavage surface of the window region. The simulated mesa structure is just a structure that does not have the function of optical signal processing in AXEL. Therefore, in this specification, the simulated mesa structure can also be called a "dummy mesa". In addition, the simulated mesa structure can also be called a "simulated waveguide" because it has a mesa structure with the same layer configuration as the waveguide mesa including the waveguide core. As can be easily understood from Fig. 10, the simulated mesa structures 102-1 and 102-2 can be fabricated at the same time as the process of fabricating the mesa of the waveguide 101a.
[0044] Therefore, the optical circuit of the present disclosure can be implemented as an optical circuit in which a semiconductor laser and a semiconductor optical amplifier (SOA) are integrated on a substrate 103, the optical circuit comprising: a waveguide 101 optically connected to the SOA and terminated on the inner side of an end face of the substrate; and a window region 108 through which light emitted from an end face of the waveguide propagates to the end face, the window region including a plurality of simulated mesas 102-1, 102-2 configured from the end face to the end face, parallel to the optical axis of the light and spaced a distance M from the optical axis, and the window region being filled with bulk semiconductor 106, 106a except for the simulated mesas.
[0045] By forming the simulated mesa structures 102-1 and 102-2, the area and volume of the region that becomes the base on which the InP layer is deposited in the window region during burying and regrowth are made smaller than in the conventional AXEL technology. As explained in the InP burying and regrowth process in Fig. 7, by performing regrowth with an insulating film mask 64 provided on the top of the waveguide mesa portion 64, InP is deposited on the exposed semiconductor portion other than the mesa portion, and the waveguide mesa is buried.
[0046] Similarly, in the regions 101-1 and 101-2 of the simulated mesa structure in FIG. 10, regrowth is performed with the upper part of the simulated mesa protected by an insulating mask. As a result, in the window region, an InP layer is selectively grown only on the semiconductor exposed portion excluding the simulated mesa structure. By arranging the simulated mesa structure, in the window region, the semiconductor exposed portion that is the base on which the InP layer is deposited, that is, the exposed area of the substrate 103 before the regrowth process (FIG. 7(b)), is reduced. It can be understood that the space of InP to be filled in the InP deposition process is also relatively small. As a result, in the window region, the InP growth rate of the semiconductor exposed portion excluding the simulated mesa structure is relatively increased, and the InP film thickness near the simulated mesa is increased. In particular, in the window region portion between the two simulated mesa structures 102-1 and 102-2, a thicker InP film is formed than around the normal waveguide mesa portion. In (c) and (d) of Figure 10, in the window region on the extension of the optical axis of the waveguide 101a, which is sandwiched between the two simulated mesa structures, the InP film is deposited thicker and has a raised profile 106a compared to the InP layer 106 around the waveguide mesa.
[0047] By making the InP buried layer in the window region thicker, the optical field that is emitted from the end of the waveguide and propagates through the window region while diffracting is less likely to come into contact with the interface between the upper semiconductor and air, and optical loss is reduced. In other words, the window region length L can be designed to be long without increasing optical loss, and a sufficient margin can be provided for cleavage position deviation. In the window region starting from the end position of the waveguide 101a, the simulated mesa structure is located at a position far enough away from the optical axis so that the optical field is not affected. Therefore, the provision of the simulated mesa structure does not have any effect on the emitted beam.
[0048] Fig. 11 is a diagram showing the configuration of the window structure part of the optical transmitter of the present disclosure after cleavage, and shows the AXEL chip on which the output end face is formed after the cleavage process is performed. Fig. 11(a) is a view of the substrate surface, and is a cross-sectional view cut along a plane parallel to the substrate surface passing through the core. Fig. 11(b) is a cross-sectional view cut along a yz plane including the line XIb-XIb in Fig. 11(a) in the waveguide region of the AXEL. After the chip 100 before cleavage shown in Fig. 10 is cleaved, it becomes a single separated AXEL chip.
[0049] In order to thicken the InP layer without the simulated mesa structure affecting the light propagating through the window region, it is desirable to arrange the simulated mesas of the same shape symmetrically along the optical axis in the window region. With reference to FIG. 11, the length of the window region is defined as L, the width of the simulated mesa structures 102-1 and 102-2 as W, and the distance (position) from the optical axis 109 to the end of the simulated mesa structure as M. The simulated mesa structure is formed along the optical axis 1090 from the end of the waveguide to the cleavage plane, and has the same L as the window region length. Here, the change in the InP film thickness generated between the two simulated mesa structures was investigated using the position M and width W of the end of the simulated mesa structure based on the optical axis as parameters. The top of the simulated mesa structures 102-1 and 102-2 was set as the reference height, and the thickening amount Δd of the InP film thickness directly above the optical axis from the reference height was experimentally confirmed.
[0050] Figure 12 shows the relationship between the simulated mesa position from the optical axis and the InP thickening amount. Several samples of simulated mesa structures with different configurations were prepared on the same substrate, and the position M from the optical axis to the simulated mesa and the thickening amount Δd were plotted with the simulated mesa width W as a parameter. Burying and regrowth were performed simultaneously on all samples with the InP growth amount that completely buried the waveguide with a mesa height of 4 μm, the same as in the normal AXEL fabrication. For several samples, the simulated mesa position M from the optical axis was changed from 2 to 10 μm. As a parameter, the width W of the simulated mesa structure was set to three types: 3, 5, and 10 μm, and the plot is made for each width W.
[0051] As is clear from the graph of FIG. 12, the InP film in the window region can be thickened by narrowing the distance M between the simulated mesa structure and the optical axis or widening the width W of the simulated mesa structure. In fact, when the position M of the simulated mesa from the optical axis is 2 μm, the regrown InP climbs up onto the simulated mesa structure regardless of the width W of the simulated mesa structure, and the grown film thickness is not stable. When the position M of the simulated mesa from the optical axis is 7 μm, only a film thickening amount Δd of 200 nm or less is obtained, and no practical effect is obtained. From the results of FIG. 12, it is found that the simulated mesa position from the optical axis that provides a practical and sufficient effect is about M=2 to 6 μm. By using the simulated mesa structure of the optical transmitter of the present disclosure, the InP film in the window structure can be thickened by a maximum of about 1.5 μm. It is found that the greater the width W of the simulated mesa structure, the greater the effect of thickening, and even in the smallest case of W=3 μm, a sufficient effect of thickening is obtained.
[0052] FIG. 13 is a diagram showing the relationship between the light propagation distance in the window region and the optical loss caused by diffraction. For window regions with different buried InP film thicknesses, the amount of vertical beam spread due to diffraction was calculated when light propagates as shown in the schematic diagram on the right side of FIG. 13. The horizontal axis shows the distance Z that the light propagates from the end of the waveguide within the window region 108 of length L, and the vertical axis shows the proportion (%) of optical power that contacts the interface between the InP buried layer and the air above. In FIG. 13, the upper cladding layer thickness Dclad When the thickness of the upper cladding layer is 2 μm, the simulated mesa structure of the present disclosure clad The results are compared when the thickness is increased to 3 μm.
[0053] In the window structure of the conventional technology, when light propagates through the window region for a distance Z of about 15 μm, about 6% of the optical power of the optical field comes into contact with the upper cladding layer interface, resulting in optical loss. In contrast, in the window structure of the InP buried layer thickened by the simulated mesa structure in the optical transmitter disclosed herein, even if light propagates through the window region for 20 μm, the optical power ratio of the optical field that comes into contact with the upper cladding layer interface is about 3%. Compared to the configuration of the conventional technology, the optical power ratio of the light that comes into contact with the interface is less than half, making it possible to sufficiently reduce loss. That is, by using the simulated mesa structure of the optical transmitter disclosed herein, optical loss can be sufficiently suppressed even when the window region length is set to 20 μm, which is the state in which the maximum error occurs on the positive side from the design value of the cleavage position of 10 μm. This allows the design of an optical transmitter that has a sufficient margin for the required performance of high output and high-quality transmission characteristics, even against unavoidable errors in the cleavage position in the manufacturing process. This will reduce the defect rate of AXEL chips, improve yields, and lower manufacturing costs.
[0054] A more specific configuration and effects of the optical transmitter of the present disclosure will be described in detail below with reference to examples. EXAMPLES
[0055] In the optical transmitter of this embodiment, we aimed to increase the optical output power of AXEL during modulation to +9 dBm or more in order to generate a 25 Gbit / s modulated signal and to support optical communication systems with a high loss budget. The typical optical output power of AXEL during modulation currently designed for high output is about +10 dBm.
[0056] FIG. 14 is a diagram showing a schematic configuration of an optical transmitter according to the first embodiment. FIG. 14 is a top view of the chip substrate, showing a chip 200 including two adjacent AXELs 200a and 200b before separation in order to explain the process of forming an output end face by cleavage. One AXEL 200a is a monolithic integrated element including a DFB laser 202a having a length of 300 μm, an EA modulator 203a having a length of 150 μm in front of the AXEL, and an SOA 204a having a length of 200 μm. The three regions are connected by an optical waveguide 201a, and adopt a buried heterostructure using semi-insulating InP that provides high heat dissipation and current confinement effects. In order to obtain a sufficient reflection suppression effect, the AXEL 200a has a bent waveguide portion at the tip of the optical waveguide 201a and a window region 208a up to the planned cleavage surface 212. When the adjacent AXEL 200b is rotated 180 degrees at the intersection of the intended cleavage plane 212 and the optical axis, it overlaps with the AXEL 200a and has the same shape.
[0057] The AXEL chip is formed on the InP substrate (100) surface, and a DFB laser 202a is arranged so as to output light in the substrate orientation
[0011] . The light from the DFB laser 202a passes through an EA modulator 203a and an SOA 204a on the same optical axis, and is rotated θ with respect to the crystal orientation
[0011] . wg In this embodiment, the bending angle θ is set to a value that provides a sufficient reflection suppression effect. wg In FIG. 14, the bending angle θ wg Please note that the drawing is emphasized. The light emitted from the end of the waveguide passes through a window region 208a with a length L, and then reaches the planned cleavage surface (output end surface). The adjacent AXEL chip 200b has a bent waveguide of the waveguide 201b arranged so that the emitted light from the waveguide 201a faces each other on the same optical axis. The optical circuit elements of the two AXEL chips 200a and 200b have the same structure with rotational symmetry at the intersection of the planned cleavage surface and the optical axis line. The two AXEL chips 200a and 200b in the state of FIG. 14 will be separated into two chips by the cleavage process.
[0058] As described above, in the cleaving process, a positional error of about ±10 μm usually occurs. Therefore, the final window region length L is determined at the time when the cleaving process is completed. In order to obtain a sufficient reflection suppression effect, a window region length of 5 μm or more is necessary, and in consideration of the positional error in the cleaving process, the window region length L is set to 15 μm in this embodiment. Therefore, the respective ends of the waveguides 201a, 201b of the two opposing AXEL chips are arranged separated by window regions 208a, 208b having a length of 30 μm before cleaving. In the optical transmitter of this embodiment, two simulated mesa structures 205-1, 205-2 are provided so as to be approximately linearly symmetrical with respect to the optical axis. The positions of the two simulated mesa structures 205-1, 205-2 in the optical axis direction are determined by the bending angle θ wg Note that there is a slight misalignment due to the angle of 5°. When viewed from the substrate surface, both of the two simulated mesa structures 205-1 and 205-2 have a width W and are disposed parallel to the optical axis at a distance M from the optical axis.
[0059] As can be intuitively understood from FIG. 14, the simulated mesa structures 205-1 and 205-2 are located at a position sufficiently far from the optical axis of the extended waveguide, so they do not affect the light emitted from the end of the waveguide. The two simulated mesa structures are located across the planned cleavage surface 212, and have the same effect on the two adjacent AXELs 200a and 200b facing each other. In this embodiment, the width W of the simulated mesa structure is designed to be 5 μm, and the distance M from the optical axis is designed to be 3 μm. As explained in FIG. 12 in relation to the simulated mesa position and the InP thickening amount, the InP buried layer in the window regions 208a and 208b can be sufficiently thickened by setting an appropriate width and position of the simulated mesa structure.
[0060] Next, the procedure for fabricating the optical transmitter of this embodiment will be described. In order to fabricate the element, a lower SCH (Separated Confinement Heterostructure) layer, a multilayered An initial substrate was used on which the active layer (MQW1), which is a child well layer, and the upper SCH layer were grown in sequence. The multiple quantum well layer is made up of six quantum well layers, and has optical gain in the 1.3 μm oscillation wavelength band. The initial growth substrate including this multiple quantum well layer has a structure optimized for highly efficient operation of the DFB laser.
[0061] The procedure for fabricating an optical transmitter begins by selectively etching the initial substrate described above, leaving only the areas that will become the DFB laser and SOA regions, and then growing the other active layers. A multiple quantum well layer (MQW2) for the EA modulator is then grown by butt-joint regrowth. In the area that will become the SOA region, the core layer structure formed on the initial growth substrate remains intact, and has the same layer structure as the DFB laser. The only difference between the layer structures of the three regions, the DFB laser, EA modulator, and SOA, is the presence or absence of a diffraction grating. The above procedure makes it possible to manufacture the structure, which integrates multiple regions with different functions, at low cost by limiting the number of regrowths.
[0062] Next, the boundary between the DFB laser and the EA modulator, the boundary between the EA modulator and the SOA, and the region from the end of the SOA to the output facet are selectively etched again, and a butt-joint regrowth is performed to grow a bulk semiconductor that will become the core layer of the passive region. Next, a diffraction grating that operates at an oscillation wavelength of 1.3 μm is formed in the DFB laser region. A diffraction grating is formed in the resonator of the DFB laser so that light is output in the substrate orientation
[0011] direction. After that, a p-InP cladding layer and a contact layer are grown on the entire surface of the element by regrowth again. In this embodiment, the cladding layer thickness is set to 2.0 μm so that the optical field propagating through the waveguide does not reach the interface between the upper cladding layer and the electrode. This cladding layer thickness is a typical value for optical semiconductor devices in normal communication wavelength bands.
[0063] Next, a mesa structure for the portion that will become the waveguide is formed. In this process, the end portion of the waveguide in the window structure portion is also fabricated at the same time, and further, the simulated mesa structure in the window region that is introduced in the optical transmitter of the present disclosure is also fabricated at the same time in this process.
[0064] FIG. 15 is a diagram showing the vicinity of the window region before regrowth of the optical transmitter of this embodiment. As with the optical transmitter shown in FIG. 14, this shows the state before the two adjacent AXELs 200a and 200b are separated by cleavage and before the InP burying layer is fabricated. FIG. 15(a) shows the substrate surface before burying, and shows a cross-sectional view cut along a plane parallel to the substrate surface (xy plane) that passes through the center of the core height of the waveguide. FIG. 15(b) shows a cross-sectional view cut along a cross-sectional view (yz plane) perpendicular to the substrate surface that passes through line XVb-XVb, and FIG. 15(c) shows a cross-sectional view cut along a cross-sectional view (yz plane) perpendicular to the substrate surface that passes through line XVc-XVc. Note that since the tip of the waveguide is a curved waveguide, strictly speaking, the cross-sections of FIG. 15(b) and (c) are slightly tilted (θ) from the yz plane. wg =5°) plane.
[0065] In processing the mesa structure, insulating film mask patterns 207a-207c are formed in the waveguide portion through which light propagates and the simulated mesa portion of the window region, and the mesa structure is formed by etching the semiconductor by dry etching. As shown in the cross-sectional views of (b) and (c) of FIG. 15, in the waveguide portion through which light propagates and the simulated mesa portion, the semiconductor substrate 210 is etched to a depth of about 4 μm so that the core layer 201a formed in the above-mentioned butt joint process is completely exposed from the sidewall. The core layer 201a of the waveguide and the core layers 205-1 and 205-2 of the simulated mesa are made of semiconductors of the same composition grown in the above-mentioned butt joint process of the passive region. When the mesa structures of the waveguide portion and the simulated mesa portion are completed, the mesa structures are then filled with InP by filling and regrowth while leaving the insulating film masks 207a-207c.
[0066] Fig. 16 is a diagram showing the vicinity of the window region after burying and regrowth of the optical transmitter of this embodiment. Fig. 16(a) is a view of the substrate surface after burying the InP layer 211, and shows a cross-sectional view cut along a plane parallel to the substrate surface (xy plane) passing through the center of the core height of the waveguide. Fig. 16(b) is a cross-sectional view cut along a plane (yz plane) perpendicular to the substrate surface passing through line XVIb-XVIb, and Fig. 16(c) is a cross-sectional view cut along a plane (yz plane) perpendicular to the substrate surface passing through line XVIc-XVIc. The difference from Fig. 15 is that the waveguide portion and the simulated mesa portion are buried with the InP burying layer 211.
[0067] The buried InP layer is a semi-insulating InP layer doped with Fe, and functions as a current blocking layer. The amount of InP buried was adjusted so that the mesa structure of the waveguide was completely buried. If the amount of the semi-insulating InP layer is too small, the current blocking effect is not fully exerted, and current is not efficiently injected into the DFB laser or SOA section, resulting in insufficient output and efficient optical amplification characteristics. Conversely, if the amount of the semi-insulating InP layer is too large, abnormal growth occurs in which the regrown InP climbs up onto the insulating film above the waveguide, making it difficult to form electrodes for the DFB laser and SOA sections. To ensure sufficient current confinement effect and flatness above the waveguide, it is necessary to set the amount of InP layer grown during the buried growth.
[0068] In this embodiment, the growth amount of the InP layer was adjusted so that the waveguide mesa is completely buried with InP 211 and the InP 211 has the same height as the waveguide mesa in the waveguide cross section, as shown in Fig. 16(b). On the other hand, in the window region, the exposed portion of the semiconductor 210 is narrowed by providing the simulated mesa structure, as shown in Fig. 16(c), and the grown thickness of the InP film is relatively large compared to the waveguide mesa. This thickening effect can increase the thickness of the buried InP layer between the two simulated mesa structures in the window region.
[0069] In the window region of the AXEL of this embodiment, the thickness of the semiconductor buried layer above the optical axis is 3.1 μm, which is significantly thicker than the 2.0 μm in the window region of the AXEL of the conventional technology. If the growth amount of the InP layer in the window region becomes too large, abnormal growth of InP may occur, where InP climbs up onto the top of the simulated mesa structure. However, the simulated mesa structure itself has no function or role in the operation of the AXEL, and does not cause any problems in the process. The simulated mesa structure functions only as a structure for adjusting the grown film thickness in the window region. In addition, the formation of the simulated mesa structure and the growth of the InP layer in the window region are performed by an etching process and a buried growth process for forming a waveguide. Therefore, with the introduction of the simulated mesa structure in the optical transmitter of the present disclosure, there is no increase or complication in the manufacturing process, and it can be manufactured using the same manufacturing process as the AXEL of the conventional technology.
[0070] Following the above-mentioned embedding and regrowth process of Fig. 16, the insulating film masks 207a to 207c are removed, and then the contact layers between the DFB laser 202a, EA modulator 203a, and SOA 204a regions shown in Fig. 14 are removed by wet etching to electrically isolate each region. Next, a P-side electrode is formed to inject current through the contact layer on each region on the upper surface of the semiconductor substrate. After that, the InP substrate is polished to about 150 μm, and an electrode is formed on the back surface of the substrate to complete the process on the semiconductor wafer.
[0071] Following the process on the semiconductor wafer, a semiconductor bar containing multiple AXEL chips is created by cleaving to form a (011) crystal plane. A typical semiconductor chip cleavage process is used here, with a cleavage position accuracy of ±10μm or less. Using a semiconductor bar with multiple chips connected together created by the cleavage process, an AR coating is applied to the front emission facet and a high reflection coating (HR: High Reflection) is applied to the opposite rear facet.
[0072] In order to confirm the effect of introducing the simulated mesa structure into the window region of the optical transmitter of Example 1 obtained through the series of fabrication steps described above, the transmitter was module-mounted together with multiple AXEL chips fabricated using the same process as the conventional technology and evaluated.
[0073] FIG. 17 is a diagram showing the relationship between the cleavage position variation and the optical loss in the AXEL of Example 1. FIG. 17 shows the results of the AXEL module with the configuration of the conventional technology and the results of the AXEL module of Example 1 at the same time. For both the conventional buried regrowth process and Example 1, the window region length was evaluated as an actual measurement value for all AXEL chips before the module was mounted, and the cleavage position deviation amount on the horizontal axis of FIG. 17 is plotted with the actual measurement value. The optical loss in the module on the vertical axis of FIG. 17 is calculated from the difference between the optical output level evaluated by the large-diameter PD in the AXEL chip before the module was mounted and the fiber-coupled optical output level after the module was mounted. The driving conditions when measuring the optical output level were that the EA modulator was in a non-bias state, and 80 mA and 40 mA CW (Continuous-Wave) currents were applied to the DFB laser and the SOA, respectively. The operating temperature of the AXEL chip was set to 55° C. In FIG. 17, the window region length was evaluated as an actual measurement value for all AXEL chips before the module was mounted, and the cleavage position deviation amount on the horizontal axis of FIG. The results of evaluating a module using the AXEL chip of the present technology under the same conditions are also plotted. The AXEL chip of the prior art is designed with a window region length L of 10 μm. The optical loss results in the AXEL module of the prior art configuration shown by the ○ (white circle) plot in Figure 17 are the same as the plot data already shown in Figure 9.
[0074] As mentioned above, there is a limit to the positional accuracy of the cleavage, which results in variation in the window region length L. In the conventional AXEL technology, the design value for the window region length is 10 μm, but due to a positional deviation error of about ±10 μm that occurs during the cleavage process, the actual measured value of the window region length L varies from 0 μm to 20 μm. The variation in the window region length L also causes the optical loss within the module to fluctuate, and the optical loss increases rapidly especially when the window region length is 10 μm or more, which is larger than the design value. The net optical output of the fabricated AXEL chip is the modulated output P avgIt is estimated that the average output power of the optical transmitter is about +12 dBm. Therefore, if the optical loss increases by 3 dB or more during module installation, the target optical output power standard of +9 dBm for the optical transmitter cannot be achieved. This was the main cause of the decrease in the manufacturing yield of conventional AXEL modules.
[0075] In contrast, in the AXEL chip group of this embodiment, which has the simulated mesa structure shown by the black circle plot in FIG. 17, the design value of the window region length L is 15 μm, and the actual measured values are distributed in the range of 5 to 25 μm. It can be confirmed that the optical loss in the module is relatively small even when the window region length L exceeds 10 μm, where the loss increases sharply in the conventional technology. Although a slight increase in optical loss is observed near the window region length L of 25 μm, where the error is maximum on the positive side, the optical loss is suppressed to 3 dB or less in all modules of this embodiment. As can be seen from the optical loss results shown in FIG. 17, the introduction of the simulated mesa structure has made it possible to manufacture AXEL modules, which had unavoidable variations in optical loss, with a high yield. For a target value of optical output during modulation of +9 dBm or more, the yield of modules using chips with window structures of the conventional technology was about 40%. On the other hand, in the module using the chip having the simulated mesa structure of this embodiment, the manufacturing yield was about 75%, which was a dramatic improvement of nearly double the manufacturing yield of the conventional technology.
[0076] Finally, the AXEL module was used to evaluate the modulation characteristics of 25 Gbit / s and confirm the operation quality of the optical transmission signal. The modulation signal was NRZ, Pseudo-Random Binary Sequence (PRBS2). 31 -1 was used for all AXEL. The comparison was performed with the laser current set to 80 mA and the voltage applied to the EA modulator set to -1.5 V. The drive current of the SOA was set to 70 mA. With the optical transmitter of this example, a dynamic extinction ratio of 9.1 dB was obtained from evaluation of the EYE waveform at a data rate of 25 Gbit / s.
[0077] FIG. 18 is a diagram showing the relationship between the cleavage position variation and the optical waveform quality in the AXEL of the first embodiment. The horizontal axis shows the actual cleavage position deviation (μm) of the AXEL chip, and the vertical axis shows the mask margin (%) of the optical transmission signal waveform from the optical transmitter module using that chip. This is exactly the same as the relationship between the cleavage position deviation of the conventional technology shown in FIG. 5 and the mask margin (%) of the EYE waveform. The ○ (white circle) plots for the AXEL of the conventional technology configuration shown in FIG. 18 are the same as the data plots of the graph shown in FIG. 5. The ● (black circle) plots for the AXEL of the present embodiment show the mask margin of the EYE waveform for the same sample as the sample showing the optical loss in FIG. 17.
[0078] As is clear from the graph in Figure 18, the AXEL of this embodiment mounted on a module provides a stable mask margin of 30% or more, regardless of the amount of cleavage position misalignment. This mask margin value is at the same level as samples that provided good quality using the AXEL of the prior art mounted on a module. In the optical transmitter incorporating the simulated mesa structure of this embodiment, no reduction in mask margin was confirmed, and no deterioration in the waveform quality of the optical signal occurred, making it clear that the manufacturing yield can be significantly improved.
[0079] When the transmission characteristics of a 40 km single mode fiber were evaluated using the operating conditions of the AXEL module described above, a bit error rate of 10 -12 From the above results, it was confirmed that the optical transmitter with the simulated mesa structure in the window region dramatically improved the manufacturing yield of the optical transmission module without degrading the module performance due to the variation in the cleavage position, as in the conventional technology configuration. EXAMPLES
[0080] In the above-mentioned embodiment 1, a basic configuration example of an optical transmitter (AXEL) in which two simulated mesa structures are introduced in the window region is shown, and the effects of the compatibility of the tolerance to misalignment of the cleavage plane, the optical output level and the quality of the transmission waveform, and the improvement of the yield of the manufacturing process are explained. The simulated mesa structure in the window region is not limited to the one having two simulated mesas on both sides of the optical axis of the output waveguide as shown in embodiment 1, and various variations are possible. In this embodiment, an example of introducing a simulated mesa structure into a chip design that can increase the number of products produced from one wafer is shown. In the embodiment of the optical transmitter described below, three simulated mesa structures are introduced into the window region in a 1.5 μm band AXEL chip.
[0081] FIG. 19 is a diagram showing an example of the layout of optical circuit elements of the AXEL chip of the second embodiment. FIG. 19 shows two AXEL chips including window regions sandwiching a planned cleavage surface, where (a) shows a schematic layout of the optical circuit elements of chip 310 of the first embodiment, and (b) shows a schematic layout of the optical circuit elements of chip 300 of the present embodiment. Each chip includes a DFB laser, an EA modulator, and an SOA, and is provided with a main waveguide 301a passing through these optical circuit elements. The optical axis of a bent waveguide 305a at the tip of main waveguide 301a is inclined at an inclination angle (θ wg ) from the perpendicular direction to the cleavage plane 303. wg ) and emerges from the end face of the chip. Note that for the sake of explanation, the simulated mesa structure is not drawn in FIG.
[0082] In the optical transmitter of the first embodiment, as shown in FIG. 19(a), in two adjacent AXELs of a chip 310 before cleavage, the optical circuit elements are arranged so that the optical axes 306 of two curved waveguides 305a, 305b facing each other on the planned cleavage surface 303 are aligned. In the arrangement of the optical circuit elements in FIG. 19(a), a large offset occurs in the arrangement of DFB lasers and the like having waveguides in the substrate crystal orientation (011) in the direction along the cleavage surface 303 that becomes the output end surface in the (100) substrate on which the optical circuit is formed. From the viewpoint of the margin of the cleavage position of the sidewall 304 parallel to the main waveguide 301a of the chip 310 and the securing of a sufficient area of electrodes, it is desirable to arrange the regions of the DFB lasers, EA modulators, and the like in the center of the chip as far away from the sidewall 304 as possible. Therefore, in the arrangement of the optical circuit elements as shown in FIG. 19(a), the chip width (vertical direction in the figure) must be designed to be relatively large. When such optical circuit elements are arranged on a typical AXEL chip, a chip width of at least about 300 μm is required.
[0083] In contrast, in the optical transmitter of this embodiment shown in Fig. 19(b), the optical circuit elements are arranged so that the optical axes 306a, 306b of the emitted light from two adjacent AXEL chips do not coincide. This makes it possible to arrange the waveguide region of the DFB laser or the like in the substrate crystal orientation (011) direction near the center of the chip, thereby making it possible to narrow the chip width.
[0084] When adopting a structure in which the chip width is reduced by arranging optical circuit elements as shown in FIG. 19(b), the simulated mesa structure of the first embodiment causes a problem. In the optical transmitter of the first embodiment, the two simulated mesa structures are arranged on both sides of the optical axis on the extension of the curved waveguide, straddling the planned cleavage position. When the optical circuit elements are arranged such that the two optical axes 306a, 306b of adjacent AXELs do not coincide as shown in FIG. 19(b), two simulated mesa structures cannot be arranged in this manner. In this embodiment, a structure is adopted in which three simulated mesa structures are arranged at equal intervals for two AXELs arranged adjacent to each other on the planned cleavage surface.
[0085] Fig. 20 is a diagram showing a schematic configuration of a window structure part before cleavage of an optical transmitter of Example 2. Fig. 20 is a diagram showing the substrate surface of two adjacent AXEL chips 300 before cleavage, and more precisely, a cross-sectional view cut parallel to the substrate surface passing through the waveguide core in the thickness direction of the substrate. The optical axes 306a and 306b of the curved waveguides of the waveguides of each AXEL are misaligned as shown in Fig. 19(b).
[0086] In the optical transmitter of this embodiment, the three simulated mesa structures 302-1 to 302-3 are all parallel to the optical axes of the respective curved waveguides 305a, 305b so as to straddle the cleavage plane 303, and are disposed at a distance M from the optical axes. Of the three simulated mesa structures, the middle simulated mesa structure 302-2 is disposed at a position at a distance M from both optical axes 306a, 306b of the two adjacent AXEL chips. Of the three simulated mesa structures, the two simulated mesa structures at both ends are also disposed at a distance M from the optical axes of the respective AXEL chips. All the simulated mesa structures have the same width W.
[0087] Therefore, in the optical circuit of the present disclosure, the waveguide 301a is bent toward the terminal end at an angle of θ wg and an optical axis 306a is an extension of the optical axis of the curved waveguide portion, and the above-mentioned multiple simulated mesas can be implemented as including two simulated waveguides 302-1 and 302-2 arranged on both sides of the optical axis 306a, and a third simulated waveguide 302-3 separated by a distance of 2M from one of the two simulated waveguides, the simulated waveguide 302-2, which is on the opposite side to the bending of the curved waveguide portion 305a, and parallel to the two simulated waveguides.
[0088] By arranging the simulated mesa structure as shown in Fig. 20, it is possible to obtain the same effect of thickening the InP window region as in Example 1. At the same time, as explained in Fig. 19, it is possible to reduce the size of the AXEL chip and increase the yield of chip manufacturing within the wafer.
[0089] The manufacturing process of the AXEL chip of this embodiment is almost the same as that of the AXEL chip of the above-mentioned embodiment 1. The lengths of the DFB laser, EA modulator, and SOA in the AXEL chip are 350 μm, 200 μm, and 200 μm, respectively. The diffraction grating formed in the DFB laser has a period designed to make the oscillation wavelength 1.55 μm. As shown in FIG. 20, three simulated mesa structures are arranged in the window region near the emission end, and the InP buried layer in the window region is thickened. The distance M from the optical axis to the end of the simulated mesa structure is 4 μm, and the width W of the simulated mesa is 7 μm.
[0090] As is clear from FIG. 20, the distance M from the optical axis to the simulated mesa structure is uniquely determined when the waveguide arrangement of the two adjacent AXEL chips is determined. On the other hand, since the width W of the simulated mesa structure can be designed to any value, the InP film in the window region can be adjusted to a desired thickness by appropriately setting the width W. Compared to the optical transmitter of Example 1, the distance M from the optical axis to the simulated mesa structure is increased in this example. Due to the simulated mesa structure, the degree of reduction in the exposed semiconductor portion that serves as the base for the deposition of the InP layer is smaller than in Example 1, and it is expected that the effect of thickening the InP film in the window region will be reduced.
[0091] In this embodiment, by increasing the width W of the simulated mesa structure, it is possible to successfully obtain the same InP thickening effect as in the first embodiment. In the actually fabricated AXEL chip, the cladding layer thickness on the optical axis in the window region was confirmed to be 3.0 μm, and even if the window region length L is designed to be 15 μm, the effect of improving the yield can be obtained without increasing the optical loss in the window region. Furthermore, in the optical transmitter of this embodiment, the chip width could be narrowed from 300 μm in the conventional technology to 200 μm. As a result, it was possible to successfully increase the number of AXEL chips that can be fabricated on the same wafer by 50%.
[0092] Next, the basic characteristics of the AXEL chip of Example 2 were evaluated. At an operating temperature of 55° C., currents of 80 mA and 100 mA were applied to the DFB laser and SOA, respectively, and a voltage of −1.5 V was applied to the EA modulator. ppWhen modulation was performed with a 10 Gbit / s NRZ signal at VDD = 1.5 V, a high optical output level of up to +11 dBm was confirmed as the optical output during modulation. When the yield rate was evaluated when modularizing an AXEL chip having the simulated mesa structure of this embodiment, very good results were confirmed, with the manufacturing yield rate of the module being approximately 60%, when the target value of the optical output during modulation was set to +10 dBm or more.
[0093] In the above-mentioned examples, all the simulated mesa structures were rectangular waveguide mesas when viewed from the substrate surface. However, in order to thicken the buried InP layer between the two simulated mesa structures, it is sufficient to reduce the area and space to the base for the deposition of the InP layer in the window region. Therefore, the shape of the simulated mesa structure may be tapered from the waveguide end toward the emission end face, for example. Even in this case, it is preferable that the distance M between the optical axis and the simulated mesa structure is constant throughout the window region so that the buried InP layer between the two simulated mesa structures in the window region is thickened uniformly. It is also preferable that the two simulated mesa structures are symmetrical with respect to the optical axis.
[0094] In the above explanation, an example was shown in which the waveguide of the DFB laser was configured to output light in the direction of
[0011] , but the substrate orientation is It could also be TIFF0007587189000001.tif621.
[0095] As explained in detail above, the optical transmitter disclosed here, which introduces a simulated mesa structure in the window region near the cleavage plane of the chip, has made it possible to achieve both the high output and high-quality transmission characteristics of AXEL. It has become possible to provide a sufficient margin in the window region length L for misalignment during cleavage, which has significantly improved the yield of optical transmitter modules and also reduced costs. [Industrial Applicability]
[0096] The present invention can be used in optical communications.
Claims
1. In an optical circuit in which a semiconductor laser and a semiconductor optical amplifier (SOA) are integrated on a substrate, a waveguide optically connected to the SOA and terminated on an inner side relative to an end face of the substrate, the waveguide including a bent waveguide portion having a bend angle θ wg toward the termination; a window region through which light emitted from the end of the curved waveguide portion propagates to the end face, the window region including a plurality of simulated mesas configured from the end to the end face, parallel to an optical axis of the light and spaced a distance M from the optical axis, and the window region being filled with a bulk semiconductor except for the simulated mesas; Equipped with the end face is a cleavage face, The plurality of simulated mesas include Two simulated waveguides arranged on either side of the optical axis; and an optical circuit comprising: a third simulated waveguide that is spaced a distance of 2M from one of the two simulated waveguides on an opposite side to the bend of the bent waveguide portion and that is parallel to the two simulated waveguides.
2. 2. The optical circuit according to claim 1, wherein the plurality of simulated mesas have the same layer structure as the waveguide.
3. 3. The optical circuit according to claim 1, wherein the plurality of simulated mesas have a mesa structure with a width W.
4. The light is bent at a bend angle θ wg 4. The optical circuit according to claim 1, wherein the light is incident on said end face at an inclination of only 1.
5. the semiconductor laser is a distributed feedback laser (DFB laser) integrated with an electroabsorption modulator, The optical circuit is an optical transmitter, The waveguide is a passive waveguide made of a core layer having a band gap wavelength shorter than the oscillation wavelength of the DFB laser.
5. The optical circuit according to claim 1, wherein:
6. The DFB laser, the electroabsorption modulator, and the SOA are formed on an InP substrate (100); The optical axis of the DFB laser is aligned along the substrate crystal orientation.
6. The optical circuit according to claim 5, wherein the direction is parallel to the direction of the optical axis.
7. the distance M is in the range of 2 μm<M<6 μm, a width W of each of the plurality of simulated mesas is in the range of 1 μm<W; The distance L between the end and the end face is in the range of 5 μm<L<25 μm.
7. The optical circuit according to claim 1, wherein:
Citation Information
Patent Citations
Semiconductor light element and manufacturing method thereof
JP2003069149A
Optical semiconductor device and manufacturing method therefor
JP2008066647A
Semiconductor device and manufacturing method therefor
JP2008177405A
Semiconductor optical element, and optical module
JP2009246241A
Semiconductor monolithic integrated optical transmitter
US20050006654A1
