Photodetector and optical integrated device

Abstract

To provide a photodetector that achieves both light receiving sensitivity and high-speed operation.SOLUTION: A photodetector includes a light input layer 102 formed of a first semiconductor on a substrate 100 and transparent to the light of the wavelength used, and a light absorption layer 104 formed on the light input layer with a second semiconductor having a bandgap smaller than that of the first semiconductor. The light absorption layer includes a first polar region 105 doped with impurities of the first conductivity type, a second polar region 106 doped with an impurity of a second conductivity type different from the first conductivity type, and an undoped region sandwiched between the first polar region and the second polar region in the direction horizontal to the substrate, and between the first polar region and the second polar region, there is a region with a higher effective refractive index than the rest of the light absorption layer.SELECTED DRAWING: Figure 5

Description

[Technical field]

[0001] The present disclosure relates to photodetectors and integrated optical devices. [Background technology]

[0002] To cope with the increase in communication capacity, research and development in the field known as silicon photonics has been attracting attention. Forming optical circuits on silicon (Si) substrates or silicon-on-insulator (SOI) substrates makes it possible to integrate them with electronic circuits formed on the same substrate. Photonic integrated circuits are expected to improve the conversion efficiency between electrical and optical signals and realize small, high-capacity, low-power optical transceivers.

[0003] In optical integrated circuits, the waveguide parts used for multiplexing, splitting, and modulation of light are required to have the property of not absorbing light in order to suppress excess loss. On the other hand, photodetectors that convert light into electricity must have the property of absorbing light. A promising combination that meets these requirements is one that uses germanium (Ge) for the photodetector and Si for the other waveguide parts, using near-infrared light with wavelengths from 1.2 μm to 1.6 μm. Light in this wavelength range is transparent to Si and is absorbed by Ge.

[0004] When extracting photocarriers from a Ge photodetector connected to a Si waveguide, the configuration shown in Figure 1 can be used. Rectification occurs by doping the Si waveguide (Si-WG) joined to the Ge photodetector (Ge-PD) with p-type impurities and doping the top of the Ge-PD with n-type impurities. When a reverse bias is applied such that the n-type has a higher potential than the p-type, electrons of the photocarriers generated by light absorption in the Ge are extracted to the n-electrode and holes are extracted to the p-electrode. This configuration has a heterointerface between Si and Ge, and a PIN structure is formed in a direction perpendicular to the substrate, so it is called a heterojunction vertical PIN structure.

[0005] A stripe-type Ge photodetector formed on a Si island of an SOI (Silicon-on-Insulator) substrate is known (see, for example, Patent Document 1). This Ge photodetector includes a heterojunction in which a high-concentration p-type impurity region and an n-type impurity region are arranged in a direction parallel to the substrate. [Prior art documents] [Patent documents]

[0006] [Patent Document 1] U.S. Patent No. 9,035,409 Summary of the Invention [Problem to be solved by the invention]

[0007] Regardless of whether the PIN structure is perpendicular or parallel to the substrate, in a configuration that includes a heterojunction in the photocarrier migration path, the photocarriers pass through a heterointerface where the energy bands are discontinuous. In the configuration shown in Figure 1, when holes pass through the Si / Ge heterointerface, some of the holes accumulate at the interface due to a barrier in the valence band. The holes accumulated at the interface block the externally applied voltage and cause a decrease in the electric field strength of the depletion layer. As a result, the response characteristics of the photodetector deteriorate and high-speed operation is hindered.

[0008] An object of the present disclosure is to provide a photodetector that achieves both high light receiving sensitivity and high-speed operation. [Means for solving the problem]

[0009] In one embodiment, the photodetector comprises: an optical input layer formed on a substrate from a first semiconductor and transparent to light of a wavelength to be used; a light absorbing layer formed on the light input layer and made of a second semiconductor having a band gap smaller than that of the first semiconductor; The light absorbing layer has a first polarity region doped with an impurity of a first conductivity type, a second polarity region doped with an impurity of a second conductivity type different from the first conductivity type, and an undoped region sandwiched between the first polarity region and the second polarity region in a direction horizontal to the substrate, and has a region between the first polarity region and the second polarity region whose effective refractive index is higher than that of other parts of the light absorbing layer. Effect of the Invention

[0010] This realizes a photodetector that combines high light receiving sensitivity with high speed operation. [Brief description of the drawings]

[0011] [Figure 1] FIG. 1 is a schematic diagram of a photodetector with a heterojunction vertical PIN structure. [Diagram 2] FIG. 1 is a diagram showing hole accumulation at a Si / Ge heterointerface. [Diagram 3] FIG. 13 is a diagram showing the evaluation results of response characteristics of an actually fabricated heterojunction photodetector. [Figure 4A] FIG. 1 is a diagram showing technical problems of a homojunction photodetector. [Figure 4B] FIG. 1 is a diagram showing technical problems of a homojunction photodetector. [Figure 4C] FIG. 1 is a diagram showing technical problems of a homojunction photodetector. [Figure 4D] FIG. 1 is a diagram showing technical problems of a homojunction photodetector. [Diagram 5] 1A and 1B are a schematic top view and a schematic cross-sectional view of a photodetector according to a first embodiment. [Figure 6A] FIG. 6 is a diagram illustrating the effect of the configuration in FIG. 5. [Figure 6B] FIG. 6 is a diagram illustrating the effect of the configuration in FIG. 5. [Figure 7A] 6 is a manufacturing process diagram of the photodetector of FIG. 5. [Figure 7B] 6 is a manufacturing process diagram of the photodetector of FIG. 5. [Figure 7C] 6 is a manufacturing process diagram of the photodetector of FIG. 5. [Figure 7D] 6 is a manufacturing process diagram of the photodetector of FIG. 5. [Figure 7E] 6 is a manufacturing process diagram of the photodetector of FIG. 5. [Figure 7F] 6 is a manufacturing process diagram of the photodetector of FIG. 5. [Figure 8] FIG. 11 is a schematic cross-sectional view of a photodetector according to a second embodiment. [Figure 9] FIG. 11 is a schematic cross-sectional view of a photodetector according to a third embodiment. [Figure 10] FIG. 13 is a schematic cross-sectional view of a photodetector according to a fourth embodiment. [Figure 11] 1 is a schematic plan view of an optical integrated device for a transceiver using a photodetector according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0012] Before describing the specific configuration of the photodetector of the embodiment, technical problems that arise in a heterojunction type photodetector will be described in more detail, and then problems that arise in a homojunction type photodetector will be described.

[0013] Figure 2 shows the accumulation of holes at the Si / Ge heterointerface. At the interface between Si and Ge, the energy level of the valence band Ev drops on the Si side and rises on the Ge side. This discontinuity in the energy bands becomes a potential barrier for holes generated by light absorption in Ge.

[0014] Holes can overcome this barrier with a certain probability due to the quantum tunneling effect, but some accumulate at the interface. The accumulated holes block the applied voltage from the outside, reducing the electric field strength of the depletion layer. In a steady state, the transport speed of photocarriers in the depletion layer is roughly proportional to the electric field strength. A reduction in the electric field strength of the depletion layer reduces the carrier transport speed. The more carriers accumulate, the greater the blocking effect and the greater the rate of reduction in the transport speed. Ultimately, a reduction in the carrier transport speed leads to a deterioration in the conversion rate of optical input to an electrical signal, i.e., a deterioration in the response characteristics of the photodetector.

[0015] The degree of degradation of the response speed of a photodetector depends on the amount of photocarriers, so the greater the intensity of the incident light, the greater the degradation of the response speed. In an optical receiver for coherent transmission, local light is introduced into the receiving front-end circuit at the same time as the signal light, and heterodyne or homodyne detection is performed. There is a tendency to increase the intensity of the local light to increase the receiving sensitivity, and bandwidth degradation becomes a particular problem when the light intensity is strong.

[0016] Figure 3 shows the evaluation results of the response characteristics of the heterojunction photodetector shown in Figure 1 that was actually fabricated. The horizontal axis is frequency (GHz) and the vertical axis is relative response strength (dB), and the response characteristics are evaluated at multiple bias voltages. The frequency at which the response strength drops by 3 dB from the response strength to a low-frequency signal (near 0 GHz) is called the response band of the photodetector. In the figure, the 3 dB drop line is shown as a dashed line horizontal to the horizontal axis. Figure 3(a) shows the evaluation results when the optical input strength is weak, and Figure 3(b) shows the evaluation results when the optical input strength is strong.

[0017] It can be seen that when the optical input intensity is strong, the response band tends to narrow and the response speed decreases. It can also be seen that the degradation of the response characteristics is mitigated by increasing the absolute value of the bias voltage. This is because increasing the absolute value of the bias voltage increases the proportion of photocarriers that have the energy to overcome the energy barrier at the Si / Ge interface due to the electric field potential. Although the accumulation of photocarriers can be mitigated by increasing the applied voltage, when a photodetector is actually implemented in an optical integrated circuit and used as a receiver, there is a limit to the bias voltage that can be applied to the photodetector. In reality, it is difficult to apply a voltage high enough to prevent response degradation.

[0018] One way to alleviate these issues is to use a homojunction PIN-PD. With a homojunction, there is no potential barrier in the photocarrier movement path due to the energy band discontinuity, so local accumulation of photocarriers does not occur. There is almost no reduction in the electric field strength of the depletion layer, so bandwidth degradation is unlikely to occur even if the incident light intensity increases. However, there are issues specific to homojunctions.

[0019] Figures 4A to 4D show problems that may occur in homojunction photodetectors. A Ge-PD that absorbs light of the wavelength used is connected to a Si-WG that is transparent to the wavelength used. In a homojunction Ge-PD, high-concentration impurity regions of opposite polarity are provided horizontally or laterally to the substrate inside the light absorption layer. A Ge-PD has a high-concentration n-type impurity region (indicated as "n+" in the figure) and a p-type impurity region (indicated as "p+" in the figure) with an undoped Ge layer sandwiched between them. In this specification, these high-concentration impurity regions in the light absorption layer are called "polar regions." The n-type polar region is connected to an n-electrode (n-ELC), and the p-type polar region is connected to a p-electrode (p-ELC).

[0020] When light is incident on the polar region, a phenomenon called free carrier absorption occurs, so ideally, the light should be absorbed only by the undoped Ge layer. Free carrier absorption is not interband absorption that generates new photocarriers, but light absorption that gives photon energy to already existing carriers. Free carrier absorption does not contribute to an increase in photocurrent, so it is so-called "ineffective absorption."

[0021] The photosensitivity, which is one of the important performance indices of a photodetector, is given by the ratio of the output current to the incident light intensity. One way to increase the photosensitivity is to suppress reactive absorption. In the homojunction type, the relationship between the light intensity distribution and the polar region becomes an issue in order to minimize reactive absorption.

[0022] Immediately after light is incident on the undoped light absorption region of the Ge-PD from the Si-WG, light B1 is distributed between the n-type polar region and the p-type polar region, as shown in Figure 4A. However, as shown in Figure 4B, light B2 propagates a certain distance and spreads throughout the Ge layer, penetrating the polar region. As shown in Figure 4C, when the polar region is formed deep to increase the electric field strength, the overlap between light B3 and the polar region becomes particularly large, making it more likely that ineffective absorption will occur in the polar region. As a result, the light receiving sensitivity decreases.

[0023] In order to avoid the reactive absorption that occurs in Figures 4B and 4C, it is possible to limit the polar regions to shallow regions of the light absorption layer, as in Figure 4D. Although the configuration in Figure 4 suppresses reactive absorption, the intensity distribution of light B4 moves away from the polar regions. The electric field strength (schematically shown by the dashed arrow in the figure) generated by the potential difference between the polarity regions of opposite polarity weakens in the region where light B4 is guided, and the response speed decreases. Since the reduction in electric field strength causes a reduction in the carrier transport speed, it is difficult to achieve both high-speed operation and high light receiving sensitivity in the homojunction structures shown in Figures 4A to 4D.

[0024] First Embodiment Fig. 5 is a schematic diagram of a photodetector 10 according to a first embodiment. Fig. 5(a) is a top view, and Fig. 5(b) is an X-X' cross-sectional view. As coordinate axes, the lamination direction or film thickness direction is the Z direction, and the plane perpendicular to the Z direction is the XY plane. In the embodiment, a photodetector configuration is provided that suppresses reactive absorption in the polar region while maintaining efficient electric field application.

[0025] The photodetector 10 is provided on a dielectric layer 101 on a substrate 100, and has an optical input layer 102 and an optical absorption layer 104. The optical input layer 102 is formed as an extension of an optical waveguide 120, and guides propagating light to the optical absorption layer 104. For optical confinement, the optical waveguide 120 and the photodetector 10 are entirely surrounded by dielectric layers 101 and 116. When the dielectric layers 101 and 116 are made of the same material, they will eventually become an integral layer, but from the viewpoint of the layer structure for the optical input layer 102 and the optical absorption layer 104, the dielectric layer covering the upper surface of the optical absorption layer 104 is referred to as the "dielectric layer 116."

[0026] Assuming that light with a wavelength of 1.2 μm to 1.6 μm (near infrared region) is used in optical communications, the optical waveguide 120 and the optical input layer 102 are formed of a first semiconductor that is transparent to light of the above wavelength. The optical absorption layer 104 is formed of a second semiconductor that absorbs light of the above wavelength. The first semiconductor is, for example, Si that has a band gap that does not absorb light of the above wavelength. The second semiconductor is Ge, SiGe, GeSn, or the like that has a band gap narrower than Si and absorbs light of the above wavelength.

[0027] Ge and Sn are Group IV materials like Si, and have the advantage of being less susceptible to contamination during the manufacturing process compared to the use of III-V mixed crystals, but the material of the light absorption layer 104 is not limited to these materials. Other semiconductor materials with appropriately designed band gaps may be used as long as they can absorb light with wavelengths of 1.2 μm to 1.6 μm.

[0028] The Si optical waveguide 120 and the optical input layer 102 can be fabricated as part of an optical integrated circuit using silicon photonics technology. The dielectric layers 101 and 116 are layers of a material with a large refractive index difference between the first and second semiconductors, such as silicon dioxide (SiO2). When an SOI substrate is used, the substrate 100 may be a Si substrate, and the dielectric layer 101 may be a BOX (Buried Oxide) layer of the SOI substrate. In this case, the Si layer on the BOX layer (so-called SOI layer) is patterned to form the optical waveguide 120 and the optical input layer 102.

[0029] The optical waveguide 120 may include an optical waveguide 121 with a constant width and a tapered waveguide 122 whose width increases continuously toward the optical input layer 102. Light incident on the optical input layer 102 from the optical waveguide 120 is coupled to the optical absorption layer 104 by evanescent optical coupling. Since Ge has a higher refractive index than Si, the light guided through the Si optical waveguide 120 is coupled into Ge by the evanescent field and is absorbed in Ge.

[0030] The light absorbing layer 104 has therein a first conductive type polar region 105 and a second conductive type polar region 106. The polar region 105 is doped with a high concentration of n-type impurities, for example, and the polar region 106 is doped with a high concentration of p-type impurities. Between the polar regions 105 and 106 of opposite polarities is an undoped light absorbing layer 104. This forms a horizontal PIN structure. If the light travels in the Y direction, the PIN structure is arranged in a direction parallel to the substrate 100 on the XZ plane perpendicular to the Y direction. Since the PIN structure is inside the Ge light absorbing layer 104 and does not have a heterointerface in the current path, it is a homojunction type PIN structure.

[0031] In the embodiment, reactive absorption in the polar regions 105 and 106 is minimized, and photocurrent is efficiently extracted. To achieve this, light incident on the light absorbing layer 104 is efficiently confined between the polar regions 105 and 106, and the spread of light to the polar regions 105 and 106 is suppressed. Specifically, in the region between the polar regions 105 and 106, the film thickness of at least one of the light input layer 102, the light absorbing layer 104, and the dielectric layer 116 is partially changed.

[0032] In the first embodiment, a recess 103 is formed in the light input layer 102 to partially increase the thickness of the light absorbing layer 104. When the upper surface of the light absorbing layer 104 is flat, the thickness of the light absorbing layer 104 increases in the region corresponding to the recess 103 by the amount that the thickness of the light input layer 102 is reduced by the recess 103. Light is trapped in the light absorbing layer 104, which is partially thickened between the polar regions 105 and 106, and ineffective absorption in the polar regions 105 and 106 is suppressed. The light trapped between the polar regions 105 and 106 is subjected to a sufficient electric field strength, and photocarriers generated by light absorption are efficiently extracted. This makes it possible to achieve both light receiving sensitivity and response speed.

[0033] 6A and 6B are diagrams showing the effect of the configuration in Fig. 5. When the light absorbing layer 104 with an increased thickness in the recess 103 is turned upside down, the light absorbing layer 104 becomes a so-called rib-type waveguide, and light is confined in the rib portion with a larger thickness.

[0034] 6A, immediately after light is coupled from the light input layer 102 to the light absorbing layer 104, light 130A exists in the space where the light absorbing layer 104 has become thicker due to the recess 103. Now, consider the effective refractive index of the three layers, the light input layer 102, the light absorbing layer 104, and the dielectric layer 116. The effective refractive index is proportional to the proportion of space occupied by the medium (space occupancy rate).

[0035] In the direction perpendicular to the substrate 100, the refractive index of the Ge light absorption layer 104 is greater than that of the Si light input layer 102 and greater than that of the SiO2 dielectric layer 116, so that the light 130A is confined in the light absorption layer 104. In the horizontal direction, the effective refractive index of the thicker portion of the light absorption layer 104 is greater than that of the regions on both sides of the recess 103 where the thickness is relatively smaller, so that the light 130A is confined in the thicker region. The light 130A is less likely to spread to the polar regions 105 and 106 formed in the regions with a lower effective refractive index.

[0036] In order to efficiently confine the light 130A, it is desirable that the change in the effective refractive index due to the change in film thickness is sufficiently large. For example, it is desirable to change the film thickness at a rate of 1 / 10 or more of the wavelength of the light 130A. When the wavelength of the light 130A is 1500 nm, it is desirable that the change in the film thickness of the light absorption layer 104 that increases in the recess 103 is 150 nm or more. If the change in film thickness is less than 1 / 10 of the wavelength of the light 130A, the difference in the effective refractive index may be insufficient, resulting in insufficient lateral confinement of the light.

[0037] With the above configuration, even after the light 130B propagates a certain distance in the light absorbing layer 104 as shown in Fig. 6B, the light 130B is prevented from spreading throughout the entire light absorbing layer 104. In a plane perpendicular to the light propagation direction, the light 130B has a dominant intensity distribution in the center of the light absorbing layer 104, i.e., in the region between the polar regions 105 and 106. Since the distribution of the light 130B can be separated from the polar regions 105 and 106, it is possible to suppress ineffective absorption due to free carrier absorption and maintain high light receiving sensitivity.

[0038] Returning to Fig. 5, in order to extract photocarriers generated in the light absorbing layer 104 as a photocurrent, an electric field is applied to the light absorbing layer 104 from the electrodes 111 and 112. The electrode 111 is electrically connected to the n-type polar region 105 via a plug electrode 107 formed in the dielectric layer 116. The electrode 112 is connected to the p-type polar region 106 via a plug electrode 108 formed in the dielectric layer 116.

[0039] By applying a positive bias voltage to the electrode 111 and a negative bias voltage to the electrode 112, a potential difference is generated between the polar regions 105 and 106, and the polar region 105 is in a reverse bias state. Electrons generated in the light absorbing layer 104 are drawn out from the electrode 111, and holes are drawn out from the electrode 112. Since reactive absorption in the polar regions 105 and 106 is minimized, a photocurrent according to the amount of incident light is extracted from the electrodes 111 and 112, achieving high light receiving sensitivity. In addition, an electric field of sufficient strength is applied to the photocarriers, maintaining a high response speed.

[0040] Since the light 130A and 130B (hereinafter sometimes collectively referred to as "light 130") are separated from the polar regions 105 and 106, the polar regions 105 and 106 can be disposed deep in the light absorption layer 104. This allows an electric field to be efficiently applied to the region where the light distribution is dominant, and it is possible to achieve both an improvement in the carrier transport speed and light receiving sensitivity.

[0041] From the viewpoint of applying a sufficient electric field to the light absorbing layer 104, it is desirable that the depth of the polar regions 105 and 106 reach a depth of at least half the average film thickness of the light absorbing layer 104. By confining the light 130 in the region of strong electric field strength, the photocarrier generation region and the region of strong electric field strength coincide with each other. This shortens the carrier transport time and realizes high-speed operation.

[0042] From the viewpoint of separating the polar regions 105 and 106 from the distribution of light 130, it is desirable that the width of the region where the film thickness of the light absorption layer 104 is thick, i.e., the width of the recess 103, is narrower than the width of the undoped region between the polar regions 105 and 106. The width of the recess 103 is the dimension in the (X) direction perpendicular to both the film thickness (Z) direction and the light propagation (Y) direction (see FIG. 5). By controlling the film thickness to an appropriate width and adjusting the effective refractive index distribution, the spread of the light 130 is suppressed, and the deterioration of the light receiving sensitivity caused by free carrier absorption in the polar regions 105 and 106 is suppressed.

[0043] <Photodetector manufacturing process> 7A to 7F are diagrams showing the steps of fabricating a photodetector according to an embodiment. In each of Fig. 7A to 7F, a top view (a) and an X-X' cross-sectional view (b) are shown in parallel. In the following description, the materials, sizes, and forming methods of each component are merely examples and do not limit the present disclosure.

[0044] 7A, an optical input layer 102 and an optical waveguide 120 connected to the optical input layer 102 are formed of Si on a dielectric layer 101 on a substrate 100. When an SOI substrate is used, for example, the thickness of the Si substrate that becomes the substrate 100 is 750 μm, the thickness of the SiO2 layer used as the dielectric layer 101 is 2 μm, and the thickness of the Si layer on the SiO2 layer (so-called SOI layer) is 250 nm. The optical waveguide 120 and the optical input layer 102 are formed by patterning the SOI layer, for example, by electron beam (EB) lithography and inductively coupled plasma (ICP) etching.

[0045] The optical waveguide 120 may include an optical waveguide 121 and a tapered waveguide 122. The optical waveguide 121 may be connected to other optical elements (such as a 90° hybrid optical mixer or a wavelength filter) of the optical integrated circuit. The tapered waveguide 122 has a width that increases continuously from the optical waveguide 121 toward the optical input layer 102. The width of the optical waveguide 121 is, for example, 500 nm. The width of the tapered waveguide 122 may be increased to 1 μm. The size of the optical input layer 102 in the light propagation direction (Y direction) is, for example, 35 μm, and the size in the width direction (X direction) is, for example, 10 μm.

[0046] 7B, a resist mask 124 is formed by photolithography to expose only a partial region of the light input layer 102. An opening 125 in the resist mask 124 has a width of 1 μm. From the opening 125, Si in the light input layer 102 is removed to a depth of 100 nm by dry etching to form a recess 103.

[0047] In FIG. 7C, the resist mask 124 is removed, and a SiO2 film 114 is formed on the entire surface to a thickness of 20 nm. Although the SiO2 film is transparent, in order to facilitate understanding of the manufacturing process, the light input layer 102 and the light waveguide 120 covered with the SiO2 film 114 are shown by dotted lines in the top view (a). A resist mask is formed on the SiO2 film 114 by photolithography, and the SiO2 film 114 is patterned by dry etching using the resist mask to form a mask for Ge growth. A portion of the Si light input layer 102 is exposed through an opening 115 formed in the SiO2 film 114 mask. The resist mask is then removed.

[0048] In Fig. 7D, in the region (within the opening 115) not covered with the SiO2 film 114, a Ge layer is epitaxially grown by low pressure chemical vapor deposition (LP-CVD) to form the light absorbing layer 104. The thickness of the Ge layer is 500 nm. The size of the light absorbing layer 104 is 30 µm in the light propagation direction (Y direction) and 8 µm in the width direction (X direction).

[0049] In FIG. 7E, a resist mask is formed by photolithography to expose a portion of the light absorbing layer 104, and boron (B) is ion-implanted to form an n-type polar region 105. The impurity concentration is, for example, 1E20 [cm-3]. Thereafter, the resist mask is removed, and a resist mask is formed again by photolithography to expose a different region of the light absorbing layer 104. Phosphorus (P) is ion-implanted to form a p-type polar region 106. Thereafter, the resist mask is removed.

[0050] In FIG. 7F, a SiO2 film is formed to a thickness of 1 μm as a dielectric layer 116 on the entire surface by a CVD method, and the implanted impurities are activated by annealing. For example, the impurities in the Ge layer are activated by annealing for about 10 seconds in an atmosphere at 600° C. Then, contact holes reaching the polar regions 105 and 106 are formed in the dielectric layer 116 by photolithography and dry etching. Next, an aluminum (Al) film is formed on the entire surface by sputtering, and the Al film is processed by photolithography and dry etching. As a result, electrodes 111 and 112 connected to the polar regions 105 and 106 are formed. The Al film filling the contact holes becomes the plug electrodes 107 and 108.

[0051] The above process completes the photodetector 10A, in which the thickness of the light absorption layer 104 is partially thickened. The photodetector 10A is fabricated on an SOI substrate simultaneously with other optical waveguides and optical devices of an optical integrated circuit, but the steps for forming the other optical devices are omitted here.

[0052] <Second embodiment> 8 is a schematic cross-sectional view of a photodetector 20 according to the second embodiment. In the second embodiment, no recess is provided in the light input layer, and the thickness of the light absorption layer is partially increased on the upper surface.

[0053] The photodetector 20 is provided on a dielectric layer 201 on a substrate 200, and has a light absorbing layer 204 and a light input layer 202 that guides light to the light absorbing layer 204. The entire photodetector 20 is surrounded by dielectric layers 101 and 216. The light input layer 202 is formed of a first semiconductor that is transparent to light of the wavelength used, and the light absorbing layer 204 is formed of a second semiconductor that absorbs light of the wavelength used. As an example, the first semiconductor is Si and the second semiconductor is Ge.

[0054] The light absorbing layer 204 has therein a first conductivity type polar region 205 and a second conductivity type polar region 206. The polar region 205 is doped with a high concentration of n-type impurities, for example, and the polar region 206 is doped with a high concentration of p-type impurities. Between the polar regions 205 and 206 of opposite polarities is an undoped light absorbing layer 204. The photodetector 20 has a homojunction PIN structure that does not have a heterointerface in the current path.

[0055] The light absorbing layer has a protrusion 215 between the polar regions 205 and 206. The protrusion 215 partially thickens the thickness of the light absorbing layer 204, forming a rib-type waveguide. The effective refractive index in the region where the thickness is increased by the protrusion 215 is higher than the effective refractive index in the regions where the polar regions 105 and 106 are formed on both sides. Light coupled from the light input layer 202 to the light absorbing layer 204 is efficiently trapped between the polar regions 205 and 206. The spread of light to the polar regions 205 and 206 is suppressed, and ineffective absorption is suppressed.

[0056] A positive bias is applied to polar region 205 via electrode 211 and plug electrode 207, and a negative bias is applied to polar region 206 via electrode 212 and plug electrode 208. Photocarriers generated in light absorbing layer 204 receive a sufficient electric field strength between polar regions 205 and 206 and are drawn to the corresponding electrodes 211, 212.

[0057] The protrusions 215 of the light absorption layer 204 are formed by dry etching the epitaxially grown Ge layer in FIG. 7D without forming the recesses 103 of the first embodiment. As an example, the regions on both sides of the epitaxially grown Ge layer in the Y direction are etched by 100 nm to form protrusions 215 with a height of 100 nm on the upper surface of the Ge layer. The configuration of the second embodiment also makes it possible to achieve both high light receiving sensitivity and high response speed.

[0058] <Third embodiment> 9 is a schematic cross-sectional view of a photodetector 30 according to the third embodiment. In the third embodiment, the thickness of a dielectric layer covering the photodetector is partially changed.

[0059] The photodetector 30 is provided on a dielectric layer 301 on a substrate 300, and has a light absorbing layer 304 and a light input layer 302 that guides light to the light absorbing layer 304. The entire photodetector 30 is surrounded by dielectric layers 301 and 316. The light input layer 302 is formed of a first semiconductor that is transparent to light of the wavelength used, and the light absorbing layer 304 is formed of a second semiconductor that absorbs light of the wavelength used. As an example, the first semiconductor is Si and the second semiconductor is Ge.

[0060] The light absorbing layer 304 has therein a first conductivity type polar region 305 and a second conductivity type polar region 306. The polar region 305 is doped with a high concentration of n-type impurities, for example, and the polar region 306 is doped with a high concentration of p-type impurities. Between the polar regions 305 and 306 of opposite polarities is an undoped light absorbing layer 304. The photodetector 30 has a homojunction PIN structure that does not have a heterointerface in the current path.

[0061] The dielectric layer 316 covering the light absorbing layer 304 has a protrusion 315 in a portion corresponding to the undoped region between the polar regions 305 and 306 of the light absorbing layer 304. Here, a four-layer slab is considered, which is the light input layer 302, the light absorbing layer 304, the dielectric layer 316, and an air layer. The total effective refractive index of the region where the dielectric layer 316 is thicker than the air layer is higher than the total effective refractive index of the region where the dielectric layer 316 is thinner. Thus, the light coupled from the light input layer 302 to the light absorbing layer 304 is confined in the region between the polar regions 305 and 306 where the effective refractive index is high. The spread of light to the polar regions 305 and 306 is suppressed, and the ineffective absorption is suppressed.

[0062] A positive bias is applied to polar region 305 via electrode 311 and plug electrode 307, and a negative bias is applied to polar region 306 via electrode 312 and plug electrode 308. Photocarriers generated in light absorbing layer 304 receive a sufficient electric field strength between polar regions 305 and 306 and are drawn to the corresponding electrodes 311, 312.

[0063] The protrusions 315 of the dielectric layer 316 are formed by forming a dielectric layer 316 covering the entire surface after FIG. 7E without forming the recess 103 of the first embodiment, and dry etching the dielectric layer 316. For example, a SiO2 layer may be formed on the entire surface as the dielectric layer 316, and the SiO2 layer may be dry etched to 500 nm to form protrusions with a height of 500 nm. The configuration of the third embodiment can also achieve both high light sensitivity and high response speed.

[0064] <Fourth embodiment> 10 is a schematic cross-sectional view of a photodetector 40 according to the fourth embodiment. In the fourth embodiment, the thickness of the light absorption layer is partially increased by combining the first and second embodiments.

[0065] The photodetector 40 is provided on a dielectric layer 401 on a substrate 400, and has a light absorbing layer 404 and a light input layer 402 that guides light to the light absorbing layer 404. The entire photodetector 40 is surrounded by dielectric layers 401 and 416. The light input layer 402 is formed of a first semiconductor that is transparent to light of the wavelength used, and the light absorbing layer 404 is formed of a second semiconductor that absorbs light of the wavelength used. As an example, the first semiconductor is Si and the second semiconductor is Ge.

[0066] The light absorbing layer 404 has therein a first conductivity type polar region 405 and a second conductivity type polar region 406. The polar region 405 is doped with a high concentration of n-type impurities, for example, and the polar region 406 is doped with a high concentration of p-type impurities. Between the polar regions 405 and 406 of opposite polarities is an undoped light absorbing layer 404. The photodetector 40 has a homojunction PIN structure that does not have a heterointerface in the current path.

[0067] The light input layer 402 has a recess 403, and the light absorbing layer 404 has a protrusion 415 in the region between the polar regions 405 and 406. In the light absorbing layer 404, the undoped regions of the polar regions 405 and 406 have a larger film thickness, so that the effective refractive index of this portion is larger than the effective refractive index on both sides. Light coupled from the light input layer 402 to the light absorbing layer 404 is confined in the region between the polar regions 405 and 406, where the effective refractive index is high. The spread of light to the polar regions 405 and 406 is suppressed, and ineffective absorption is suppressed.

[0068] A positive bias is applied to the polar region 405 via the electrode 411 and the plug electrode 407, and a negative bias is applied to the polar region 406 via the electrode 412 and the plug electrode 408. Photocarriers generated in the light absorption layer 404 receive a sufficient electric field strength between the polar regions 405 and 406, and are drawn out to the corresponding electrodes 411, 412. The configuration of the fourth embodiment also makes it possible to achieve both high light receiving sensitivity and high response speed.

[0069] 11 is a schematic diagram of an optical integrated device 50 for a transceiver using the photodetector of the embodiment. The optical integrated device 50 includes an optical transmitter TX and an optical receiver RX. The optical integrated device 50 is used as a transceiver for coherent transmission, and local light input from a port P2 is supplied to the optical transmitter TX and the optical receiver RX.

[0070] In the optical receiver RX, the signal light received at port P1 is split into two mutually orthogonal polarized waves by the polarizing beam splitter PBS. One polarized wave is input to the 90° hybrid optical mixer 51a. The polarization axis of the other polarized wave is rotated 90 degrees by the polarization rotator PR and input to the 90° hybrid optical mixer 51b.

[0071] The 90° hybrid optical mixer 51a detects the input signal light (polarized wave) with the local light and outputs four lights with a phase shift of 90° each. The four lights are detected by photodetectors 10-1 to 10-4. The 90° hybrid optical mixer 51b detects the input signal light (polarized wave) with the local light and outputs four lights with a phase shift of 90° each. The four lights are detected by photodetectors 10-5 to 10-8.

[0072] Photodetectors 10-1 and 10-2, and photodetectors 10-3 and 10-4 may form balanced photodiodes, respectively. In this case, the difference between the outputs of photodetectors 10-1 and 10-2 is extracted as an in-phase (I) component of the first polarized wave, and the difference between the outputs of photodetectors 10-3 and 10-4 is extracted as a quadrature-phase (Q) component of the first polarized wave.

[0073] Similarly, photodetectors 10-5 and 10-6, and photodetectors 10-7 and 10-8 may form balanced photodiodes, respectively. In this case, the difference between the outputs of photodetectors 10-5 and 10-6 is extracted as an in-phase (I) component of the second polarized wave, and the difference between the outputs of photodetectors 10-7 and 10-8 is extracted as a quadrature-phase (Q) component of the second polarized wave.

[0074] The polarizing beam splitter PBS, the polarization rotator PR, and the 90° hybrid optical mixers 51a and 51b are included in an optical circuit connected to the photodetectors 10-1 to 10-8. Each of the photodetectors 10-1 to 10-8 has suppressed reactive absorption and improved electric field application efficiency, and thus has high light receiving sensitivity and can operate at high speed. The photocurrent output from the photodetectors 10-1 to 10-8 is amplified by an electric circuit including a transimpedance amplifier and input to a digital signal processor (DSP) at the downstream stage.

[0075] The optical transmitter TX has optical modulators 53a and 53b. In this example, the optical modulators 53a and 53b are Mach-Zehnder interferometer type IQ modulators. A local light beam serving as a carrier wave is modulated by a data signal input from the DSP to each of the optical modulators 53a and 53b. The output light of the optical modulator 53b is rotated by the polarization rotator PR, and then multiplexed with the output of the optical modulator 53a by the polarization beam combiner PBC. The multiplexed light is output from port P3 as a polarization multiplexed signal light.

[0076] The optical integrated device 50 is not limited to use in a polarization multiplexing transceiver, and may be used in an optical transceiver of a QPSK or multilevel QAM system. The multiple photodetectors 10 may be arranged not only in a one-dimensional array, but also in a two-dimensional array. In either case, high light receiving sensitivity and operating speed enable high-speed, high-sensitivity light detection operation.

[0077] It goes without saying that the same effect can be obtained by using the photodetectors 20, 30, or 40 instead of the photodetector 10. Also, when a SiGe photodetector or a GeSn photodetector is used instead of the Ge photodetector, it is possible to suppress reactive absorption and apply an electric field efficiently by using a similar light confinement configuration. [Explanation of symbols]

[0078] 10, 10A, 20, 30, 40 Photodetector 50 Optical integrated device 100, 200, 300, 400 boards 101, 116, 201, 216, 301, 316, 401, 416 Dielectric layer 102, 202, 302, 402 Optical input layer 103, 403 Recess 104, 204, 304, 404 Light absorbing layer 105, 106, 205, 206, 305, 306, 405, 406 polar region 120 Optical waveguide 130, 130A, 130B light 215, 315, 415 protrusion

Claims

1. an optical input layer formed on a substrate from a first semiconductor and transparent to light of a wavelength to be used; a light absorbing layer formed on the light input layer and made of a second semiconductor having a band gap smaller than that of the first semiconductor; having the light absorbing layer has a first polarity region doped with an impurity of a first conductivity type, a second polarity region doped with an impurity of a second conductivity type different from the first conductivity type, and an undoped region sandwiched between the first polarity region and the second polarity region in a direction parallel to the substrate, and has a region between the first polarity region and the second polarity region, the effective refractive index of which is higher than that of other portions of the light absorbing layer; the light absorbing layer has a region with a partially thicker film thickness between the first polarity region and the second polarity region, the width of the thicker film thickness region is narrower than the width of the undoped region, and a difference between a thickness of the thicker film thickness region and a maximum thickness of the first polarity region and the second polarity region is 1 / 10 or more of the wavelength of use.

2. The photodetector according to claim 1 , wherein the light input layer has a recess at a position corresponding to the thick region of the light absorption layer.

3. the light absorbing layer has a protrusion in the region having a large thickness; 3. The photodetector according to claim 1 or 2.

4. An optical input layer formed on a substrate using a first semiconductor and transparent to light having a wavelength to be used; a light absorbing layer formed on the light input layer and made of a second semiconductor having a band gap smaller than that of the first semiconductor; having the light absorbing layer has a first polarity region doped with an impurity of a first conductivity type, a second polarity region doped with an impurity of a second conductivity type different from the first conductivity type, and an undoped region sandwiched between the first polarity region and the second polarity region in a direction parallel to the substrate, and has a region between the first polarity region and the second polarity region, the effective refractive index of which is higher than that of other portions of the light absorbing layer; the light absorbing layer has a region having a partially large film thickness between the first polarity region and the second polarity region, the light input layer has a recess at a position corresponding to the thick region of the light absorption layer.

5. An optical input layer formed on a substrate using a first semiconductor and transparent to light having a wavelength to be used; a light absorbing layer formed on the light input layer and made of a second semiconductor having a band gap smaller than that of the first semiconductor; a dielectric layer covering the light absorbing layer; having the light absorbing layer has a first polarity region doped with an impurity of a first conductivity type, a second polarity region doped with an impurity of a second conductivity type different from the first conductivity type, and an undoped region sandwiched between the first polarity region and the second polarity region in a direction parallel to the substrate, and has a region between the first polarity region and the second polarity region, the effective refractive index of which is higher than that of other portions of the light absorbing layer; the dielectric layer has a thickness that is partially increased in a region that covers the area between the first polarity region and the second polarity region of the light absorbing layer, and the width of the region where the thickness is increased is narrower than the width of the undoped region; Photodetector.

6. The first polar region and the second polar region reach a depth from the surface of the light absorbing layer to a depth of at least half of the average thickness of the light absorbing layer.

6. A photodetector according to claim 1.

7. The first semiconductor is Si, and the second semiconductor is Ge, SiGe, or GeSn; 7. A photodetector according to any one of claims 1 to 6.

8. a first electrode electrically connected to the first polarity region; and a second electrode electrically connected to the second polarity region; a reverse bias is applied to the first polarity region and the second polarity region; 8. A photodetector according to any one of claims 1 to 7.

9. A photodetector according to any one of claims 1 to 8; an optical circuit optically connected to the photodetector; An optical integrated device comprising:

Images