JP7318434B2 - Optical semiconductor device, optical transceiver using the same, and method for manufacturing optical semiconductor device - Google Patents

Description

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

In order to deal with the increase in communication capacity, a technology called silicon photonics has attracted attention, and research and development are proceeding. Forming an optical circuit on a silicon (Si) substrate or SOI (Silicon-on-Insulator) substrate facilitates integration with electronic circuits formed on the same substrate, enabling conversion of optical and electrical signals. Be efficient. Silicon photonics technology is expected to contribute to miniaturization of optical communication modules, large-capacity communication, and reduction of power consumption.

As shown in FIG. 1A, a configuration is known in which a germanium photodetector (Ge-PD) is integrated in a silicon waveguide (Si-WG) (see Patent Documents 1 and 2, for example). FIG. 1B is a cross-sectional view taken along line XX' of FIG. 1A. Since Ge has a higher refractive index than Si, the light propagated through the Si waveguide is incident on the Ge layer by evanescent optical coupling. When incident light is absorbed, photocarriers are generated and a photocurrent is extracted upon application of a bias. For example, when the signal electrode (SIG) is used as the n-side electrode and a reverse bias is applied, electrons are drawn to the signal electrode (SIG) and holes are drawn to the ground electrode (GND), thereby causing a photocurrent to flow.

In a light-receiving element of a compound semiconductor, which is different from silicon photonics, a semi-insulating InP layer is sandwiched between an InGaAs waveguide layer and an InGaAs light absorption layer, and the waveguide layer and the light absorption layer are arranged in the stacking direction or in the plane. A configuration in which they are arranged in parallel in the direction has been proposed (see, for example, Patent Document 3).

JP 2019-16694 A JP 2011-53593 A JP-A-7-79009

In the known configuration shown in FIGS. 1A and 1B, holes pass through the heterointerface between Si and Ge when an external electric field, ie reverse bias, is applied such that the voltage of the n-type semiconductor is higher than that of the p-type semiconductor. At the hetero-interface between Si and Ge, some holes that cannot cross the barrier generated in the valence band accumulate at the interface, degrading the response characteristics of the photodetector.

In addition, since the light absorption coefficient of the semiconductor used as the light absorption layer depends on the wavelength, the light is absorbed in a shorter distance as the light absorption coefficient increases, the carrier density increases, and the deterioration of the response characteristics progresses.

SUMMARY OF THE INVENTION An object of the present invention is to suppress degradation of response characteristics to incident light in an optical semiconductor device and improve photoelectric conversion efficiency.

In one aspect, the optical semiconductor device comprises
a directional coupler having a first waveguide and a second waveguide having a slab region;
a light receiver disposed in the slab region;
has
The first waveguide and the second waveguide are made of a first semiconductor material transparent to the wavelength used,
The photodetector includes a light absorption layer formed of a second semiconductor material having a narrower bandgap than the first semiconductor material and a higher refractive index than the first semiconductor material; A PIN diode having a connected region of a first conductivity type and a region of a second conductivity type.

In the optical semiconductor device, degradation of response characteristics to incident light is suppressed, and photoelectric conversion efficiency is improved.

It is a schematic plan view of a conventional Ge photodetector. 1B is a cross-sectional view taken along line XX' of FIG. 1A; FIG. It is a figure explaining carrier stagnation in a hetero-interface. It is a figure which shows the response characteristic evaluated by the conventional structure. 1 is a schematic diagram of an optical transceiver to which an optical semiconductor device of an embodiment is applied; FIG. It is a plane schematic diagram of the optical semiconductor element of embodiment. FIG. 5B is a sectional view taken along the line II' of FIG. 5A; It is a figure explaining operation|movement of a directional coupler. It is a figure which shows the relationship of the propagation constant of even mode and odd mode. It is a top view of the structure used for calculation. It is a figure which shows the calculation result at the time of d=3 micrometers and a light wavelength of 1.50 micrometers. It is a figure which shows the calculation result at the time of d=3 micrometers and a light wavelength of 1.55 micrometers. It is a figure which shows the calculation result at the time of d=3 micrometers and a light wavelength of 1.60 micrometers. It is a figure which shows the calculation result at the time of d=0 micrometer and light wavelength of 1.50 micrometers. It is a figure which shows the calculation result at the time of d=0 micrometer and a light wavelength of 1.55 micrometers. It is a figure which shows the calculation result at the time of d=0 micrometer and light wavelength of 1.60 micrometers. It is a figure which shows the calculation result at the time of d=5 micrometers and a light wavelength of 1.50 micrometers. It is a figure which shows the calculation result at the time of d=5 micrometers and a light wavelength of 1.55 micrometers. It is a figure which shows the calculation result at the time of d=5 micrometers and a light wavelength of 1.60 micrometers. It is a manufacturing-process figure of the optical semiconductor element of embodiment. It is a manufacturing-process figure of the optical semiconductor element of embodiment. It is a manufacturing-process figure of the optical semiconductor element of embodiment. It is a manufacturing-process figure of the optical semiconductor element of embodiment. It is a manufacturing-process figure of the optical semiconductor element of embodiment. It is a manufacturing-process figure of the optical semiconductor element of embodiment. It is a manufacturing-process figure of the optical semiconductor element of embodiment. It is a manufacturing-process figure of the optical semiconductor element of embodiment. It is a manufacturing-process figure of the optical semiconductor element of embodiment. It is a manufacturing-process figure of the optical semiconductor element of embodiment. It is a manufacturing-process figure of the optical semiconductor element of embodiment. It is a manufacturing-process figure of the optical semiconductor element of embodiment. It is a figure which shows the 1st modification of an optical-semiconductor element. It is a figure which shows the 2nd modification of an optical-semiconductor element. It is a figure which shows the 3rd modification of an optical-semiconductor element. It is a figure which shows the 4th modification of an optical-semiconductor element. It is a figure which shows the 5th modification of an optical-semiconductor element.

Before describing the specific configuration of the embodiment, the problem of degradation of response characteristics will be described in more detail. Carrier accumulation at the hetero interface is considered to be the main cause of deterioration of the response characteristics of the photodetector, and the degree of characteristic deterioration is influenced by the wavelength dependence of the light absorption coefficient.

FIG. 2 is a diagram for explaining accumulation of carriers at a heterointerface. In the configuration shown in FIGS. 1A and 1B, an energy barrier occurs in the valence band at the heterointerface between Ge and Si. When Ge is n-type and Si is p-type, holes (schematically represented by white circles in the figure) pass through this hetero-interface to take out an electric signal from the photodetector. Holes can cross the barrier with a certain probability due to the quantum tunnel effect or the like, but some of them accumulate at the interface. The holes accumulated at the interface block the voltage applied from the outside, and the electric field strength applied to the depletion layer decreases.

At steady state, the transport rate of photocarriers in the depletion layer depends on the electric field strength, with decreasing electric field strength decreasing the carrier transport rate. As the amount of accumulated carriers at the hetero interface increases, the electric field shielding effect increases, and the degree of decrease in transport speed also increases. A decrease in the carrier transport rate ultimately causes a deterioration in the conversion rate from input light to an electric signal, that is, the response characteristic of the photodetector. Since the degree of deterioration depends on the amount of photocarriers, the higher the intensity of the incident light, the more pronounced the deterioration of the response speed.

Coherent optical receivers perform heterodyne or homodyne detection using local light, but there is a tendency to increase the intensity of local light to increase reception sensitivity, and deterioration of response characteristics at high intensity is a serious problem.

FIG. 3 shows the response characteristics evaluated for the structures of FIGS. 1A and 1B. FIG. 3A shows the response characteristics when the optical input intensity is weak, and FIG. 3B shows the response characteristics when the optical input intensity is high. The horizontal axis is frequency, and the vertical axis is relative response strength. In both (A) and (B), the applied bias is varied from 0.0 V to −3.0 V to evaluate various response intensities.

The frequency at which the response strength of the low frequency signal drops by 3 dB is called the response band of the receiver. When the optical input intensity is high as shown in FIG. 3B, the response band becomes lower than in FIG. 3A, resulting in a lower response speed. On the other hand, by increasing the bias voltage (absolute value), response deterioration is alleviated. This is because an increase in the bias voltage increases the proportion of carriers exceeding the energy barrier at the Si/Ge interface, thereby alleviating the accumulation of carriers.

However, when a photodetector is actually mounted on an optoelectronic circuit and used as a receiver, there is a limit to the bias level that can be applied, and it is difficult to apply a high voltage sufficient to prevent response deterioration.

Next, the wavelength dependence of the light absorption coefficient will be explained. In the semiconductor used as the light absorption layer, the light absorption coefficient, which indicates the amount of light absorbed per unit propagation distance, depends on the wavelength. At wavelengths at which the light absorption coefficient increases, light is absorbed in a shorter distance and the density of carriers accumulated at the interface increases, leading to faster deterioration. Thus, the response characteristic of the photodetector has wavelength dependence. For example, a Ge light absorption layer has a bandgap near photon energy in the wavelength band of 1.5 μm used in optical communication, and the light absorption coefficient changes sharply near this wavelength. Since the light absorption coefficient is larger on the short wavelength side of the 1.5 μm band than on the long wavelength side, carriers accumulate in the hetero interface near the incident end face, resulting in a greater degree of degradation in response speed.

<Configuration of Embodiment>
In the embodiment, light is gradually absorbed along the propagation direction from the light incident position to the light receiver in order to alleviate the influence of carrier accumulation at the hetero interface and the wavelength dependence of the response characteristics. design. As a result, the accumulation of carriers at the hetero interface is dispersed in the propagation direction, and the influence of the decrease in electric field strength is alleviated.

In order to slow down the propagation of light from the optical waveguide to the optical absorption layer of the receiver, the optical waveguide and the optical absorption layer are embedded with a material having a sufficiently lower refractive index than these materials, so that the optical waveguide and the optical absorption Strengthen the optical confinement in the layer. In the configuration of Patent Document 3 described above, since epitaxial growth is used, the intermediate layer inserted between the waveguide layer and the light absorption layer is also formed of a semiconductor, and a large refractive index difference cannot be obtained. Therefore, light propagation from the waveguide layer to the light absorption layer progresses relatively quickly in the vicinity of the incident end face, and it is difficult to suppress uneven distribution of carriers as expected.

FIG. 4 is a schematic diagram of an optical transceiver 1 to which the optical semiconductor device 10 of the embodiment is applied. The optical transceiver 1 has, for example, a coherent optical receiver RX and an optical transmitter TX. The optical semiconductor element 10 of the embodiment including the photodetector 100 is applied to the photodetector (PD) of the optical receiver RX.

The optical transceiver 1 operates, for example, in a polarization multiplexing modulation scheme. In the optical transmitter TX, input light is modulated with transmission data in an X-side IQ modulator (X-IQmod) and a Y-side IQ modulator (Y-IQmod). After the plane of polarization of one modulated light is rotated by 90° by a polarization rotator (PR), the two polarized waves are combined by a polarization beam combiner (PBC) and output.

In the optical receiver RX, an input optical signal is polarization-split by a polarization beam splitter (PBS), and one polarization component undergoes polarization rotation by a polarization rotator (PR). In the 90° hybrid optical mixer (denoted as “90° Hybrid” in the figure), the I component and the Q component are extracted by interference between each optical component and local light. Output light from the 90° hybrid optical mixer is detected by each photodetector 100 of the photodetector (PD). In the example of FIG. 4, the dual balanced photodiodes are used for differential output, but the present invention is not limited to this example.

Of the optical components forming the optical transceiver 1, the multiplexer/demultiplexer (PBC, PBS) and optical modulators (X-IQmod, Y-IQmod) are required to have the property of not absorbing light in order to avoid excess loss. On the other hand, a light-receiving part (PD) that converts light into electricity must have the property of absorbing light. As a powerful combination that satisfies this requirement, for example, a Ge-based material is used for the light-receiving part PD, and a Si material is used for the other parts to transmit and receive optical signals in the wavelength band of 1.2 to 1.6 μm. Light in this wavelength band is transparent to Si and is absorbed by Ge-based materials having a bandgap smaller than that of Si. There are other semiconductor materials that absorb light in this wavelength band, but Ge is a Group IV material like Si, and compared to III-V compound semiconductors, it is less affected by contamination during the manufacturing process. few. Embodiments use Ge or a Ge compound for the light absorbing layer.

A coherent optical receiver tends to increase the intensity of local light in order to increase reception sensitivity. As will be described later, the optical semiconductor device 10 of the embodiment adopts a configuration that suppresses degradation of response characteristics, and mitigates band degradation due to high-intensity incident light to improve photoelectric conversion efficiency.

5A is a schematic plan view of the optical semiconductor device 10 of the embodiment, and FIG. 5B is a cross-sectional view taken along line II' of FIG. 5A. The optical semiconductor device 10 is a waveguide type photodetector having a directional coupler 13 formed by a first waveguide 11 and a second waveguide 12 and a photodetector 100 .

The second waveguide 12 has a slab region 120 in its part, and the light receiver 100 is formed in the slab region 120 . For example, light propagated through first waveguide 11 is optically coupled to second waveguide 12 by directional coupler 13 and detected by photodetector 100 .

The first waveguide 11 and the second waveguide 12 (including the slab region 120) are made of Si, which is transparent to the working wavelength. The photodetector 100 has a light absorption layer 15 made of a semiconductor material having a smaller bandgap and a higher refractive index than Si. Ge, SiGe, GeSn, or the like can be used as the material of the light absorption layer 15 . FIGS. 5A and 5B illustrate an example using a Ge light absorption layer 15. FIG.

Referring to FIG. 5B, the first waveguide 11 and the second waveguide 12 are embedded with a clad having a sufficiently lower refractive index than Si and Ge at the working wavelength. The clad preferably has a refractive index difference of 2 or more from Si, and an insulating layer such as SiO ₂ is used. A first waveguide 11 and a second waveguide 12 are formed on an insulating layer 32 that serves as a lower clad. The first waveguide 11, the second waveguide 12, and the photodetector 100 stacked on the second waveguide 12 are embedded with an insulating layer 37 such as _SiO2 .

For example, the photodetector 100 includes p-type regions 121 and 122 doped with p-type impurities, an undoped Ge light absorption layer 15, and an n-type conductive region 17 doped with n-type impurities. It is a PIN type photodetector with The n-type conductive region 17 is in ohmic contact with the electrode 21 , and the p-type region 122 heavily doped with impurities is in ohmic contact with the electrode 22 .

By using the electrode 21 as a signal electrode and the electrode 22 as a ground electrode and applying a reverse bias to the electrode 21, photocarriers are extracted and a photocurrent corresponding to incident light is obtained.

Returning to FIG. 5A, since the refractive index difference between Si, which is the material of the first waveguide 11 and the second waveguide 12, and SiO ₂ of the clad is large, light confinement in the waveguide is strong, and directional coupling In the device 13, the light gradually propagates from the first waveguide 11 to the second waveguide 12 along the direction of propagation.

The directional coupler 13 has wavelength dependence such that the light on the longer wavelength side with weaker optical confinement propagates faster to the second waveguide 12 . The longer wavelength light (λ _lw ) is first coupled into the second waveguide 12 , and the shorter wavelength light (λ _sw ) is also optically coupled into the second waveguide 12 as it propagates.

In the slab region 120 of the second waveguide 12, since Ge used for the light absorption layer 15 has a higher refractive index than Si, the light in the second waveguide 12 is guided to the light absorption layer 15 by evanescent optical coupling. They are absorbed and photocarriers are generated. The side of the light absorption layer 15 facing the first waveguide 11 serves as a light introducing portion.

In this configuration, at the input side of the directional coupler 13, the light is not coupled from the first waveguide 11 to the second waveguide 12 all at once, but is gradually coupled to the second waveguide 12 along the direction of propagation. . As a result, local concentration of photocarriers (especially at the incident start position) can be suppressed, and the light intensity can be dispersed in the propagation direction.

FIG. 6 is a diagram explaining the principle of operation of the directional coupler. In a directional coupler, a first waveguide 11 and a second waveguide 12 are positioned close to each other and parallel, and light travels between the waveguides. This principle can be explained as follows. Consider the case where each waveguide is a single mode waveguide.

A system in which two waveguides are combined becomes a two-mode waveguide of an even mode and an odd mode generated by coupling the modes of the two waveguides. The even mode is the mode symmetrical with respect to the central axis between the two waveguides as indicated by the solid line, and the odd mode is the mode antisymmetrical with respect to the central axis as indicated by the dotted line. The intensity of the propagating light in the two-mode waveguide can be represented by superposition of the even mode and the odd mode, as indicated by the thick line.

The even mode (propagation constant β ₁ ) and the odd mode (propagation constant β ₂ ) differ in the amount of phase change with respect to the propagation distance. , traverses between the two waveguides as shown in FIG.

From the phase relationship between the odd mode and the even mode, the propagating light intensity exists only in the first waveguide 11 at the position x0. The phase relationship between the two modes changes along with the propagation, resulting in uniform light intensity distributions in the first waveguide 11 and the second waveguide 12 at the position x1. At the position x2, light intensity is distributed only in the second waveguide 12 . At the position x3, the first waveguide 11 and the second waveguide 12 have a uniform light intensity distribution again. The distance over which the propagating light intensity is completely transferred from one waveguide to the other ((x2-x0) in FIG. 6) is called the complete coupling length L, and L=π/(β ₁ -β ₂ ). expressed.

FIG. 7 is a diagram showing the relationship between the propagation constants of the even mode and the odd mode. The horizontal axis is the waveguide spacing normalized by wavelength, and the vertical axis is the propagation coefficient β. The even-mode propagation constant and the odd-mode propagation constant coincide with the propagation constants peculiar to each waveguide when the two waveguides are separated from each other and do not communicate. have different values. This propagation constant difference becomes larger as the waveguide spacing with respect to the optical wavelength becomes narrower. Since the waveguide spacing on the horizontal axis is normalized by wavelength, in a directional coupler having a certain waveguide spacing, the longer the wavelength, the greater the propagation constant difference. As is clear from L=π/(β ₁ -β ₂ ), the longer the wavelength, the smaller the value of L, and the light intensity distribution quickly moves to the waveguide on the opposite side.

By using the wavelength dependence of the directional coupler 13 to cancel out the wavelength dependence of the light absorption coefficient of the semiconductor of the light absorption layer 15, the wavelength dependence of the response characteristics of the photodetector 100 is suppressed. can do.

For example, Ge used as the light absorption layer 15 in the embodiment has a smaller light absorption coefficient with respect to light in the wavelength band of 1.2 μm to 1.6 μm as the wavelength becomes longer. In this case, as shown in FIG. 5A, by laminating the light absorption layer 15 on the second waveguide 12 on the opposite side of the first waveguide 11, which is the optical input waveguide of the directional coupler 13, the length of the light absorption is weak. Light components on the wavelength side migrate quickly to the light absorption layer 15 . Light components on the short wavelength side, which are strongly absorbed, gradually migrate to the light absorption layer 15 . With this configuration, it is possible to cancel the wavelength dependency caused by the fact that the shorter the wavelength of the light, the faster the light is absorbed by the light absorption layer 15 .

As another feature of the embodiment, the starting point of the light absorbing layer 15 along the direction of propagation is located a predetermined distance behind the starting point of the directional coupler 13 in the direction of propagation. This is because, in the starting region of the directional coupler 13, the symmetry of the waveguides is used to efficiently couple the light from the first waveguide 11 to the second waveguide until the end of the second waveguide. This is for minimizing the proportion of light that does not move to the polarizer and ensuring the reception sensitivity.

FIG. 8 is a top view of a model used for calculation of light propagation effects in a directional coupler. A directional coupler is formed by the first waveguide 11 and the second waveguide 12 . In the slab region 120 of the second waveguide 12, a stack (Ge-mesa) including a Ge light absorption layer is arranged. Let 'd' be the offset distance from the starting point P1 of the directional coupler to the starting point P2 of the Ge light absorbing layer in the direction of light propagation. The light propagation state of the structure of the embodiment is calculated by BPM (Beam Propagation Method) while varying the distance d and the wavelength λ of the propagating light.

In the model of FIG. 8, the first waveguide 11 and the second waveguide are Si waveguides arranged on SiO ₂ , the thickness of the waveguide layer is 200 nm, and the width of the waveguide is 400 nm. Assume that the distance between the first waveguide 11 and the second waveguide 12 in the directional coupler is 100 nm, and the thickness of the Ge light absorption layer is 500 nm. Here, for simplicity, the optical absorption coefficient of Ge is set to a constant value (1000 cm ^-1 ) regardless of the light wavelength.

9A to 9C are calculation results when the offset distance d=3 μm. The light wavelength is changed to 1.50 μm, 1.55 μm and 1.60 μm, respectively. In each figure, the position (Z) in the propagation direction of the two-dimensional map of light intensity corresponds to the position in the propagation direction of the intensity profile on the right.

Common to FIGS. 9A to 9C, near Z=14 μm, the monitored intensity of the light W ₁₁ propagating through the first waveguide 11 drops sharply, and the light propagates to the second waveguide 12 . By providing a region of only the directional coupler with a predetermined offset before the position where the light absorption layer 15 is arranged, the intensity of the propagating light is efficiently coupled to the second waveguide 12 .

Focusing on the intensity of the light W ₁₂ at the starting position of the Ge light absorption layer indicated by the arrow in the intensity profile on the right side, the longer the wavelength, the steeper the rising edge of the light W ₁₂ propagating through the second waveguide 12 and the higher the light intensity. Fast transition. On the other hand, at short wavelengths, the transition of light intensity is slow, the light intensity is relatively low at the starting point of the Ge light absorption layer, and the light intensity shifts gradually along the propagation direction. That is, the incidence of light on the Ge mesa is moderate.

In the calculation, the optical absorption coefficient of Ge is constant, but in reality, the optical absorption coefficient of Ge depends on the wavelength, and in the wavelength band of 1.50 μm to 1.60 μm, the optical absorption coefficient for short waves increases. This wavelength dependence of the light absorption coefficient cancels out the wavelength dependence of the light intensity shift shown in FIGS. can be Thereby, the wavelength dependence of the response characteristics of the photodetector 100 is relaxed.

10A to 10C are calculation results when the offset distance d=0 μm. The light wavelengths are changed to 1.50 μm, 1.55 μm and 1.60 μm, respectively. The fact that the offset distance d is zero means that the starting point P1 of the directional coupler and the starting point P2 of the Ge light absorbing layer are at the same position in the propagation direction.

Common to FIGS. 10A to 10C, as shown in the profile of W ₁₁ , the transition of light intensity from the first waveguide 11 to the second waveguide 12 is slow, and the starting point position of the Ge light absorption layer Even if the propagation distance increases from (Z=14 μm), the light intensity of the light W ₁₂ propagating through the second waveguide 12 is low. In FIGS. 10A and 10C, the light intensity barely moves from the first waveguide 11 to the second waveguide 12, but in FIG. 10B, the light does not completely move to the second waveguide 12 and the light receiving sensitivity is lowered.

This is because the directional coupler has the property that when the two waveguides are symmetrical with respect to the central axis, the light intensity is completely transferred from one waveguide to the other. If the offset distance d is zero and there is no symmetrical region, light cannot be efficiently incident on the Ge light absorption layer.

11A to 11C are calculation results when the offset distance d=5 μm. The light wavelengths are changed to 1.50 μm, 1.55 μm and 1.60 μm, respectively. Common to FIGS. 11A-11C, along the direction of propagation, a significant amount of light is transferred to the second waveguide 12 before the beginning of the Ge light absorbing layer. In this case, the light intensity is concentrated at the starting position of the Ge light absorption layer and the photosensitivity is increased. becomes difficult.

From these calculation results, the smaller the offset distance d, the more the light intensity at the starting position of the Ge light absorption layer is relaxed, and the light intensity distribution is dispersed along the propagation direction. On the other hand, the larger the offset distance d, the more efficient the transfer of the light intensity to the Ge light absorption layer and the higher the photosensitivity, but the more difficult it is to make the light intensity uniform along the propagation direction. The homogenization of the light intensity distribution and the light receiving sensitivity are in a trade-off relationship.

By appropriately setting the offset distance d, the local concentration of light intensity in the light absorption layer can be alleviated and the light receiving sensitivity can be ensured. For the structure used in the model of FIG. 8, the offset distance d is preferably greater than 0 and less than 5 μm (0<d<5 μm), more preferably greater than or equal to 1 μm and less than or equal to 4 μm.

12A to 12L are manufacturing process diagrams of the optical semiconductor device 10 of the embodiment. Here, the optical semiconductor device 10 of the embodiment is manufactured as part of an optical integrated circuit used for a coherent optical receiver. In each process drawing, a cross-sectional view taken along the line II' is shown together with a top view.

In FIG. 12A, for example, a Si waveguide pattern is formed on the SiO ₂ layer 32 by using an SOI substrate in which a SiO ₂ layer 32 and a thin film of Si are laminated on a Si substrate 31 . The thickness of the Si substrate 31 is, for example, 750 μm, the thickness of the SiO ₂ layer 32 is, for example, 2 μm, and the thickness of the Si thin film is, for example, 250 nm. A desired waveguide pattern is obtained by processing the Si thin film by electron beam (EB) lithography and ICP (Inductively Coupled Plasma) dry etching.

The waveguide pattern includes first waveguides 11 and second waveguides 12 . A directional coupler 13 is formed at portions of the first waveguide 11 and the second waveguide adjacent to each other at a predetermined interval. The second waveguide 12 has a slab region 120 in the region where the directional coupler 13 is formed. The first waveguide 11 or the second waveguide 12 connected to the slab region 120 is used as an optical input waveguide. The width of the first waveguide 11 and the second waveguide 12 is, for example, 400 nm, and the distance between the waveguides in the directional coupler 13 is, for example, 100 nm.

In FIG. 12B, a resist mask 41 having a predetermined opening 42 is formed by photolithography, and ion implantation is performed to form a p-type conductive region 121 in part of the slab region 120 (see FIG. 12A). As an example, B (boron) ions are implanted so that the impurity concentration is 2×10 ¹⁸ cm ⁻³ .

In FIG. 12C, the resist mask 41 is removed, a resist mask 43 having an opening 44 is formed again by photolithography, and a p + -type conductive region 122 is formed in another portion of the slab region 120 . For example, B ions are implanted so that the impurity concentration is 1×10 ¹⁹ cm ⁻³ . In FIG. 12D, resist mask 43 is removed.

In FIG. 12E, an SiO2 film 34 having a thickness of 20 nm is formed on the entire surface by, eg, CVD, and annealing is performed to activate the implanted impurities. For annealing, the dopant in the Si layer is activated by, for example, placing it in an atmosphere of 1000° C. for about 1 minute.

In FIG. 12F, the SiO ₂ film 34 is patterned by photolithography and dry etching to form an opening 35 to expose the p-type conductive region 121 of the slab region 120 . The SiO ₂ film 34 with the opening 35 formed therein is used as a mask for selective growth of Ge.

In FIG. 12G, a mesa-shaped Ge light absorbing layer 15 is grown on the exposed p-type conductive region 121 by low pressure CVD (LP-CVD). The thickness of light absorption layer 15 is, for example, 500 nm. The dimensions of the Ge mesa are, for example, 30 μm long and 5 μm wide in the propagation direction.

In FIG. 12H, a resist mask is formed by photolithography to expose only the upper surface of the light absorbing layer 15, and for example P is implanted at a concentration of 1×10 ¹⁹ cm ⁻³ to form an n-type conductive region 17 . do. After that, the resist mask is removed. As a result, the PIN diode photodetector 100 having the p + -type conductive region 122, the undoped Ge light absorbing layer 15, and the n-type conductive region 17 is formed.

In FIG. 12I, an insulating layer 37 such as SiO ₂ is formed on the entire surface by CVD, and a resist mask 45 having a predetermined opening 46 is formed by photolithography.

12J, dry etching is performed using a resist mask 45 to form a contact hole 38 reaching the n-type conductive region 17 and a contact hole 39 reaching the p+-type conductive region 122. In FIG. After that, the resist mask 45 is removed and a metal film 49 such as Al is sputtered on the entire surface.

In FIG. 12K, photolithography is used to form a resist mask 47 only in areas where the metal film 49 is to remain.

In FIG. 12L, the metal film 49 is shaped by dry etching to form the electrode 21 in ohmic contact with the n-type conductive region 17 and the electrode 22 in ohmic contact with the p + -type conductive region 122 . Thus, the optical semiconductor device 10 having the directional coupler 13 formed by the first waveguide 11 and the second waveguide 12 and the photodetector 100 arranged in the slab region 120 of the second waveguide 12 is manufactured. be.

FIG. 13 is a schematic plan view of an optical semiconductor device 10A as a first modified example. Unlike the configuration example of FIG. 5A, in FIG. 13, the second waveguide 12 is the optical input waveguide, and the light on the longer wavelength side that has once moved to the first waveguide 11 gradually returns to the second waveguide 12. is introduced into the light absorbing layer 15 by By using the structure of this modified example, favorable characteristics can be obtained when the light absorption layer 15A is formed of a semiconductor material having a larger light absorption coefficient as the wavelength becomes longer.

As for the input light, when the light propagating through the second waveguide 12 enters the directional coupler 13A, the light (λsw) on the short wavelength side with strong optical confinement propagates through the second waveguide 12 as it is, and is absorbed by the directional coupler 13A. It is led to layer 15A.

The light (λlw) on the long wavelength side with weak optical confinement is once coupled to the first waveguide 11 at the incident end of the directional coupler 13A. After that, it gradually shifts to the second waveguide 12 on which the light absorption layer 15A having a large light absorption coefficient for long waves is mounted.

With this waveguide configuration, the wavelength dependence of the light absorption coefficient can be canceled and the light absorption can be made uniform over the propagation direction.

FIG. 14 is a schematic plan view of an optical semiconductor device 10B as a second modification. In the optical semiconductor device 10B, the distance between the first waveguide 11B and the second waveguide 12 changes in the propagation direction in the directional coupler 13B. Also in this example, the light absorption layer 15B having a large light absorption coefficient for long waves is used.

The first waveguide 11B is an input waveguide, and the light receiver 100 is mounted in the slab region 120 of the second waveguide 12. FIG. The interval between the waveguides is gradually narrowed along the propagation direction from the incident side of the directional coupler 13B.

The light propagated through the first waveguide 11B is coupled to the second waveguide 12 from the light on the long wavelength side (λ _lw ) with weak optical confinement in the directional coupler 13B. Since the distance between the first waveguide 11B and the second waveguide 12 decreases along the direction of propagation, the amount of the short-wavelength light component (λ _sw ) coupled to the second waveguide 12 increases as it propagates. Since the light absorption layer 15B has a small light absorption coefficient for short waves, as a result, the distribution of light intensity absorbed by the light absorption layer 15B is made uniform in the propagation direction.

With this configuration, the wavelength dependence of optical coupling in the directional coupler 13B and the wavelength dependence of the optical absorption coefficient can be canceled, and the amount of photocarriers generated in the optical absorption layer 15 can be dispersed in the propagation direction.

FIG. 15 is a schematic plan view of an optical semiconductor element 10C as a third modification. In the optical semiconductor device 10 , the first waveguide 11</b>C and the second waveguide 12</b>C join together at the directional coupler 13 . Also in this example, the light absorption layer 15C having a large light absorption coefficient for long waves is used.

The light propagated through the first waveguide 11C is coupled in the directional coupler 13C from the light on the long wavelength side (λ _lw ) with weak optical confinement to the second waveguide 12C. Since the distance between the first waveguide 11B and the second waveguide 12C decreases along the propagation direction, the amount of the short-wavelength light component (λ _sw ) coupled to the second waveguide 12 increases as the light propagates.

Furthermore, since the first waveguide 11C merges with the second waveguide 12C, all highly confined short-wavelength light is finally absorbed by the light absorption layer 15C.

The configuration of FIG. 15 can reliably introduce almost all of the input light into the light absorption layer 15C, so that the effect of improving the photoelectric conversion efficiency is high.

FIG. 16 is a schematic cross-sectional view of an optical semiconductor device 10D that is a fourth modification. In the optical semiconductor device 10 of FIG. 5B, the p + -type conductive region 122 is provided in Si forming the optical waveguide, and the n-type conductive region 17 is provided in Ge forming the light absorption layer 15 to realize a PIN diode. Was. In this structure, conductive polar regions were formed in each of first and second semiconductors with different bandgaps.

In the optical semiconductor device 10D of FIG. 16, a p + -type conductive region 122 is provided in Si, and a Si conductive region 171 doped with n-type impurities at a high concentration is arranged on the top surface of the Ge light absorption layer 15 in the stacking direction. Then, a photodetector 100D having a PIN structure is formed. As a result, a structure is obtained in which a conductive polar region is formed only in the first semiconductor having a large bandgap.

As a manufacturing process, selective growth of Si is performed after selective growth of the light absorption layer 15 of Ge (see FIG. 12G). By using a resist mask that exposes only the Si grown on the upper surface of the light absorbing layer 15 and implanting an n-type impurity such as P at a high concentration, an n + -type Si conductive region 171 is formed. The side walls of the Ge mesa forming the light absorbing layer 15 are covered with an undoped Si layer 172 .

FIG. 17 is a schematic cross-sectional view of an optical semiconductor element 10E that is a fifth modification. In FIG. 5B, the PIN diodes are formed in the stacking direction, but in FIG.

In the optical semiconductor device 10E, the light receiver 100E formed in the slab region 120 includes an undoped Ge light absorption layer 151 and a p+ type Ge conductive layer disposed on one side of the light absorption layer 151 in the in-plane direction. and an n+ type Ge conductive region 174 arranged on the opposite side of the p+ type conductive region 183 with the light absorbing layer 151 interposed therebetween. As a result, a structure is obtained in which a conductive polar region is formed only in the second semiconductor with a small bandgap.

As a manufacturing process, after selective growth of the Ge light absorption layer 15 (see FIG. 12G), a resist mask that exposes only the region into which the p-type impurity is to be implanted is used to add a p-type impurity such as B to a high concentration. inject. This resist mask is removed, and another resist mask is used to expose only the region into which the n-type impurity is to be implanted, and an n-type impurity such as P is implanted at a high concentration.

Also in this configuration, the light preferentially coupled to the second waveguide 12 from the long wavelength side light in the directional coupler 13 enters the light absorption layer 151 by evanescent coupling, and the photocarrier distribution is uniform in the propagation direction. can be

Although the present invention has been described above based on specific configuration examples, the present invention is not limited to the above-described configuration examples, and combinations of the embodiments and modifications, and modifications with each other are also possible. Both receiver 100D of FIG. 16 and receiver 100E of FIG. 17 are applicable to the waveguide configurations of FIGS. 13-15.

In either case, the amount of light absorbed by the light absorption layer can be dispersed in the propagation direction to suppress the deterioration of response characteristics to incident light and improve the photoelectric conversion efficiency.

10, 10A to 10E Optical semiconductor element 11 First waveguide 12 Second waveguide 120 Slab regions 13, 13A to 13C Directional couplers 15, 15A to 15C, 151 Light absorption layers 17, 171, 174 n-type conductivity Region (region of second conductivity type)
100, 100D, 100E photodetectors 122, 173 p-type conductive regions (first conductivity type regions)

Claims (11)

a directional coupler having a first waveguide and a second waveguide having a slab region;
a light receiver disposed in the slab region;
has
the slab region is adjacent to the directional coupler;
The first waveguide and the second waveguide are made of a first semiconductor material transparent to the wavelength used,
The photodetector includes a light absorption layer formed of a second semiconductor material having a narrower bandgap than the first semiconductor material and a higher refractive index than the first semiconductor material; An optical semiconductor device, which is a PIN diode having a connected region of a first conductivity type and a region of a second conductivity type. The slab region is not adjacent to the first waveguide except for the side adjacent to the directional coupler,
The optical semiconductor device according to claim 1. 3. The optical semiconductor device according to claim 1, wherein the start point of the light absorption layer is located behind the start point of the directional coupler along the light propagation direction. 4. The method according to any one of claims 1 to 3 , wherein the first waveguide is an optical input waveguide, and the light absorption layer has a light introducing portion on a side facing the first waveguide. opto-semiconductor device. 5. The optical semiconductor device according to claim 4 , wherein the second semiconductor material is a material whose light absorption coefficient decreases with increasing wavelength. The second waveguide is an optical input waveguide, and the light absorbing layer is configured to introduce a portion of light from the second waveguide and another portion of the light from the first waveguide. 4. The optical semiconductor device according to any one of claims 1 to 3 . 7. The optical semiconductor device according to claim 6 , wherein the second semiconductor material is a material whose light absorption coefficient increases with increasing wavelength. 8. The directional coupler according to any one of claims 1 to 7 , wherein the distance between the first waveguide and the second waveguide varies along the light propagation direction. opto-semiconductor device. 9. The space between the first waveguide and the second waveguide is filled with a material having a refractive index 2 or more lower than that of the first semiconductor material. The optical semiconductor device according to 1. An optical transceiver comprising the optical semiconductor device according to any one of claims 1 to 9 . forming a directional coupler on a substrate in a first semiconductor material with a first waveguide and a second waveguide having a slab region, the slab region adjacent to the directional coupler;
forming a thin film in the slab region with a second semiconductor material having a bandgap smaller than that of the first semiconductor material and a refractive index higher than that of the first semiconductor material;
providing a region of a first conductivity type and a region of a second conductivity type connected to the thin film to form a PIN diode;
A method for manufacturing an optical semiconductor device, characterized by:

Images