JPH09283788A - Optical semiconductor integrated circuit - Google Patents

Abstract

PROBLEM TO BE SOLVED: To prevent optical carriers generated in separation areas from disappearing and the output of photodiodes from deteriorating when the width of the separation areas surrounding the photodiodes is large in an optical semiconductor integrated circuit for forming the photodiodes of two rows and two columns, irradiating the photodiodes with optical spots and detecting dislocation. SOLUTION: When LOCOS oxide films (oxide films having inclinations) 50 are formed on a third separation area 27 in a second epitaxial layer 23, light which is made incident on the separation area diffracts on interfaces having the inclinations of the LOCOS oxide films 50 and light which originally generates the optical carriers in the separation area is generated in the photodiodes. Since impurity concentration is high in the separation area 27, the consumption quantity of the optical carriers is large and the distinction of the optical carriers can be suppressed by diffracting light.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an optical semiconductor integrated circuit, and irradiates adjacent optical elements (which detect the amount of light) with a light beam so that the light beam is at a predetermined position. The present invention relates to an optical semiconductor device used for determining whether or not it is concerned, and particularly to the structure of an isolation region surrounding this optical element.

[0002]

2. Description of the Related Art For example, Japanese Patent Laid-Open No. 07-073505 discloses optical ICs PD1, PD2, PD3 and PD4 in which four optical elements are arranged in a matrix of 2 rows and 2 columns. That is, in FIG. 4, a photodiode is formed in the semiconductor substrate in the photodetection IC. When a predetermined voltage is applied to the photodiode with a reverse bias and light is applied, a photocurrent flows according to the amount of light. In other words, it is possible to know the magnitude and amount of this photocurrent.

That is, the light beam spot 1 is divided into two rows 2
If the outputs of the four photodiodes are located almost in the center of the row and they are the same, it can be confirmed that they are located in the center.
It can be confirmed that the position is shifted. On the other hand, the optical IC
In addition to the photodiode, Tr and the like are incorporated to form a predetermined circuit to achieve the purpose.

On the other hand, a one-stage epitaxial layer optical IC (FIG. 5) such as Japanese Patent Laid-Open No. 02-142181 is known. That is, the photodiode 3 and the Tr 4 are incorporated in the single epitaxial layer 2. Here, in order to increase the light sensitivity of the photodiode 3, it was necessary to change the thickness of the epitaxial layer according to the wavelength of light. However, due to the increase in thickness, a portion that does not deplete occurs in the epitaxial layer 2 of the photodiode 3, and the photocarrier generated in the non-depleted portion has a problem that the transit time becomes longer and the response speed becomes slower. On the Tr side, there is a problem that the collector resistance is increased and the response speed is also slowed down.

In order to solve this, a structure as shown in FIG. 6 has been developed. That is, it is necessary to reduce the junction capacitance by making the epitaxial layer of the photodiode 3 thick with high specific resistance, and to make the portion of Tr4 thin with low specific resistance. In other words, by forming the two-stage epitaxial layers 5 and 6 having a high specific resistance, a sufficient film thickness can be obtained according to the wavelength of light, and at the same time, a portion that is not depleted can be eliminated.
By providing the buried layer 7 between the first epitaxial layer 5 and the second epitaxial layer 6, the collector resistance can be reduced.

[0006]

Since the epitaxial layer of the photodiode has a portion which is not depleted due to the thickening of the epitaxial layer, the photocarrier generated in the portion which is not depleted has a long transit time and a slow response speed. Commented that there was a problem. It is considered that this is because photocarriers move while colliding with impurities in the region where the impurity concentration is high in the non-depleted portion.

On the other hand, there is a P-type isolation region 8 as shown in FIG. 6, which is generally grounded to the ground.
The impurity concentration is set high. This separation area is shown in FIG.
Considering the above, in the region between the four optical elements (the part corresponding to the + solid line in the figure), photo carriers generated by irradiation with light (generated holes move to the substrate side, electrons generate Of the four photodiodes and flows to the anode electrode.) Has a slower mobility as described above. Therefore, if the width of the isolation region is wide, the response speed of the photodiode becomes slower. It is also possible that carriers disappear during this movement. 8 is shown in FIG.
2 shows the output when the beam spot 1 is applied to the photodiodes PD1 and PD2 and is horizontally scanned in the direction of the arrow. That is, since the carriers disappear in the separation region, the output decreases like a wavy line. Therefore, since this output is reduced, there is a problem in that the arithmetic processing circuit malfunctions.

[0008]

The present invention has been made in view of the above problems. First, in an optical semiconductor integrated circuit that compares the outputs of at least two optical elements, it is formed between two optical elements. A first separation element for refracting the light incident on the third separation area to form two optical elements.
This is solved by providing a means for making the light incident on the semiconductor layer of the second layer or the second layer. The isolation region is highly doped with impurities, and the light incident on the isolation region is bent toward the optical element by the above-mentioned means, so that it originally occurs in the isolation region and disappears or collides with the impurities. While optical IC
Will be generated in the semiconductor layer.

In other words, when the light beam strikes approximately the center of the two optical elements, the above-mentioned means provided in the separation region can form the incident beams into both optical elements in a well-balanced manner. When the output of each optical element is large and the beam is deviated, the output difference of both optical elements is large and the malfunction of calculation can be prevented. Secondly, in an optical semiconductor integrated circuit in which one optical element is formed on an IC substrate,
In the third isolation region formed around the optical element,
Providing a means for refracting the light incident on the third separation region and making it incident on the semiconductor layer of the first layer or the second layer forming the optical element is surrounded by this means.
Photocarriers that originally generate and disappear in the separation region or contribute to the output of the optical IC while colliding with impurities are generated in the semiconductor layer, so that a large output can be obtained.

Thirdly, if the means is an Si oxide film having an inclination in which the interface with the third isolation region gradually decreases in thickness toward the optical element, as shown in FIG. At a certain Si oxide film, the beam is refracted toward the optical element, so that a large output can be obtained. That is, the refractive index n of the Si oxide film is 1.45, and the refractive index n of the Si substrate is 3.42.
It is. Further, if a material having a refractive index substantially the same as that of the Si oxide film (for example, a passivation film of n = 1.5) is formed on the Si oxide film, the oxide film and the semiconductor layer are substantially straightened, and at this interface, Can be bent like.

Fourth, in the second semiconductor layer, the upward diffusion length of the second isolation region is set to exceed 1/2 of the layer thickness of the second semiconductor layer, By overlapping the third separation region at the tip of the second separation region, the lateral diffusion of the third separation region can be shortened. Therefore, the photocarriers generated in the third separation region can move to the semiconductor layer of the optical element in a short time, and the distance of the separation region through which the beam passes becomes short. Will increase and the switching speed will also increase.

Fifth, in the semiconductor layer of the first layer, the upward diffusion length of the first isolation region is set so as to exceed 1/2 of the thickness of the semiconductor layer of the first layer. In the eye semiconductor layer, the upward diffusion length of the second isolation region is set to exceed 1/2 of the layer thickness of the second semiconductor layer. Furthermore, in the first semiconductor layer, the upward first isolation region and the downward second isolation region are overlapped at the tip, and in the second semiconductor layer, the third isolation region is overlapped with the second isolation region. By overlapping at the tip of the separation region, the lateral diffusion of the second separation region can be shortened, so that the third action can be further enhanced.

Therefore, in both the third and fourth embodiments, the width of the isolation region on the surface can be shortened, and the light is refracted, so that even if light is incident, the output of adjacent photodiodes is small with carrier disappearance being small. Will contribute to.

[0014]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, a first embodiment of the present invention will be described in detail with reference to the drawings. In the vicinity of the photodiodes PD1, PD2, PD3 and PD4 of FIG. 4, means for calculating these outputs is incorporated. That is, FIGS. 1 and 2 are sectional views of PD1 and PD2, and FIG. 3 illustrates a sectional view of one element of the arithmetic circuit, that is, an IC in which Tr is incorporated with a photodiode. Since PD1 and PD2 have the same structure, the structure will be described first with reference to FIG.

In the figure, 20 is a P-type single crystal silicon semiconductor substrate, and 21 is an I-type (substantially intrinsic) first I-type (substantially intrinsic) film having a thickness of about 4 to 5 μm formed on the substrate 20 by vapor phase epitaxy. The epitaxial layer 23 is an I-type (substantially intrinsic) second epitaxial layer formed on the first epitaxial layer 21 by vapor phase epitaxy and having a thickness of about 3 μm. Here, the term “substantially intrinsic” means that even if the epitaxial layers are intrinsically laminated, the P-type impurities of the substrate are diffused to become a P-type with a very low concentration, or the P-type or N-type is caused depending on the contamination of the chamber. Because it will also be. However, if the concentration is extremely low, the depletion layer of the photodiode expands, so that there is no substantial problem.

The substrate 20 has an impurity concentration of 40 to 60 Ω, which is lower than that of a general bipolar IC (2 to 4 Ω · cm).
・ Use a specific resistance of cm. By stacking the first epitaxial layer 21 non-doped,
000 to 1500 Ω · cm, 200 to 1500 Ω · cm at the time of completion after heat treatment for forming a diffusion region
Specific resistance. Similarly, the second epitaxial layer 23 also has a specific resistance of 200 to 1500 Ω · cm when completed. The specific resistance of the epitaxial layer used in a normal bipolar IC is 0.5 to 2.0 Ω · cm.

First and second epitaxial layers 21, 23
Is electrically separated into a photodiode PD forming portion and an NPNTr forming portion by a P @ + type separating region 24 which completely penetrates both. The isolation region 24 includes a first isolation region 25 that is vertically diffused from the surface of the substrate 20 and a first isolation region 25.
And a second isolation region 26 diffused vertically from the boundary between the second epitaxial layers 21 and 23, and a third isolation region 27 formed from the surface of the second epitaxial layer 23. The first and second epitaxial layers 22 and 23 are separated into island-shaped regions.

On the surface of the second epitaxial layer 23 of the photodiode PD portion, an N + type diffusion region 28 for taking out the cathode of the photodiode PD is formed on almost the entire surface. The surface of the second epitaxial layer 23 is covered with an oxide film, and the cathode electrode is in contact with the N + type diffusion region 28 through a contact hole which is partially opened in the oxide film.
In addition, the isolation region 24 is used as the anode-side low resistance extraction region of the photodiode PD, and the anode electrode is used as the isolation region 2.
The surface of 4 is contacted.

The third isolation region 27 is provided with a Si oxide film 50 having a slope at the interface with the epitaxial layer. This oxide film interface has an inclination in the isolation region that is directed upward toward the photodiode PD. That is, if the interface with the third isolation region is an inclined Si oxide film whose thickness gradually decreases toward the optical element, as shown in FIG. 9, at the inclined Si oxide film,
The beam is refracted towards the optical element.

The present invention is characterized in that a means for bending the incident beam to enter the photodiode is provided. As this means, for example, a LOCOS oxide film, a V-groove, or another method formed is considered, and among them, a Si oxide film having an inclination whose thickness gradually decreases toward the optical element is considered. For example, the first LOCOS oxide film and the third Si oxide film have the same purpose, and the thickness may be gradually reduced toward the optical element. In particular, the LOCOS oxide film can be formed by forming a silicon nitride film, which is an oxidation resistant film, in a region other than this formation region and then oxidizing it.

That is, the refractive index n of the Si oxide film is 1.4.
5, the Si substrate has a refractive index n of 3.42. Also S
If a material having a refractive index substantially the same as that of the Si oxide film (for example, a passivation film with n = 1.5) is formed on the i oxide film, it will go straight to the interface between the oxide film and the semiconductor layer, and at this interface. Can be bent as shown. Further, if the V groove is formed substantially in the third separation region, the interface between the groove and the semiconductor layer has an inclination, and light can be refracted.

As shown in FIG. 9, the light incident from above reaches the interface between the isolation region and the LOCOS oxide film. If a normal line to the inclination is drawn and the angles between the normal line and the light ray are I and I ′, respectively, n (1.45) × Sin I = n
It is refracted by the relationship of (3.42) × Sin I ′. For example PD
When the light enters the surface perpendicularly and becomes I = 45 degrees,
I '= about 17 degrees, and the light entering vertically is bent to the PD side by 28 degrees.

Although a LOCOS oxide film will be described, some ingenuity is required for the method of forming the slope. When comparing two or more photodiodes, the LO in the middle of FIG.
The COS oxide film needs to be terminated at both ends of the isolation region. That is, it is significant that the light is incident on the left PD1 at the left slanted portion and the light is incident on the right PD2 at the right slanted portion. On the other hand, the separation regions on both sides may be terminated as in the middle, or may be inclined only on the inner side and may not be inclined on the outer side. Further, like the LOCOS oxide film on the right side of FIG. 1, if the slope of the LOCOS oxide film is formed from the left end to the right end of the third isolation region, all the light entering the third isolation region is left. Since it can be bent, the efficiency can be further improved.

This can be said also in FIG. 3. When one photodiode and one arithmetic circuit are paired, the LOCOS oxide film on the right side of FIG. 1 is placed on the isolation region in the center of FIG. It is possible to obtain more output by installing in. If the LOCOS oxide film on the Tr side is a LOCOS oxide film (which terminates in the third isolation region) as shown on the left side of FIG. 1, carriers may be generated in Tr, which is not preferable in some cases.

Therefore, between the two photodiodes, as shown in FIG.
It is necessary to use the one whose slope ends at both ends as in the middle of the LOCOS oxidation, but the LOCOS oxide film on both sides has slope that ends on both sides of the isolation region (left in the figure), PD and the isolation region. It does not matter which one has a slope that gradually increases from the interface toward the outside (right in the figure). Further, in the cross-sectional view of Tr in FIG. 3, the thickness L gradually increases from the interface between Tr and the separation region toward the outside of the separation region.
An OCOS oxide film (a LOCOS oxide film on the right side in FIG. 1) may be provided, or a slope may be terminated at both ends of the isolation region. In other words, the slope of this interface should be from the inside to the outside of the separation region,
The separation area may be all or a part.

Next, Tr will be described. NPNTr
At the boundary between the first and second epitaxial layers 21 and 23, a N + type buried layer 29 is buried. On the surface of the second epitaxial layer 23 above the buried layer 29, the specific resistance of the second epitaxial layer 23 is lowered and NP
An N type collector region 30 serving as a collector of NTr is formed so as to be connected to buried layer 29. On the surface of the collector region 30, a P-type base region 3 of NPNTr is formed.
1, an N + type emitter region 32 and an N + type collector contact region 33 are formed. An Al electrode (not shown) is in contact with each diffusion region, and the Al wiring extending on the oxide film connects the respective elements, whereby the photodiode PD serves as an optical signal input portion and the NPNTr serves as another element. Together with this, it forms a signal processing circuit (arithmetic circuit).

The photodiode PD in such a structure is
The cathode electrode is operated in a reverse bias state in which a Vcc potential such as +5 V is applied and an anode electrode is applied with a GND potential.
Since the first and second epitaxial layers 21 and 23 are P-type high resistivity layers, when the reverse bias is applied, the depletion layer becomes the first depletion layer from the junction surface between the N + cathode region 28 and the second epitaxial layer 23. And spreads greatly in the second epitaxial layers 21, 23, and reaches a thickness equal to the sum of the thicknesses of the first and second epitaxial layers 21, 23. This thickness will be described later.

When light having a wavelength of 635 nm is incident on the photodiode 21, the incident light is 10 μm from the silicon surface.
It reaches the above depth (see FIG. 14). The incident light generates photo-generated carriers, and the carriers move into photocurrent. The generation of the photogenerated carriers is roughly classified into carriers generated in the depletion layer generated in the depletion layer and carriers generated outside the depletion layer generated outside the depletion layer. The carriers generated in the depletion layer can be instantaneously moved by being attracted by the electric field, but the carriers generated outside the depletion layer have a slow response because the movement is caused by diffusion. According to the configuration of the present application, since the incident light is received by the thick depletion layer spreading over the first and second epitaxial layers 21 and 23, most of the incident light can be converted into the generated carriers in the depletion layer, which enables the high speed response of the photodiode PD. You can
Since the N + cathode region 28 is formed in a high concentration and shallow (0.3 to 1.0 μ) region due to emitter diffusion, the amount of carriers generated outside the depletion layer in the cathode region 28 is small. Moreover, because of the high concentration, the photo-generated carriers generated in the cathode region 28 are immediately extinguished or can reach the cathode electrode in an extremely short time. Therefore, the delay current due to diffusion movement is extremely small.

Further, since the P + isolation region 24 is used as the anode extraction and the isolation region 24 is diffused and formed to the deep portion of the substrate 20, the anode extraction resistance is small. One of the NPNTrs can be set to an impurity concentration suitable for the collector by the collector region 30 formed in the second epitaxial layer 23, so that the transistor characteristics can be satisfied. Moreover, by using the two-step epitaxial method, only the second epitaxial layer 23 needs to be N-type inverted, so that the diffusion heat treatment time does not become extremely long.

Therefore, according to the structure of the present invention, the high speed photodiode PD and the NPNTr can coexist together. Next, FIG. 1 and FIG. 2 will be described. In both figures, the PDs of FIG. 3 are connected, and the detailed description is omitted because it has been described above. However, the structure is such that an inclined oxide film is formed in the third isolation region, and the third isolation region is formed. The feature is that 27 is shallow.

FIGS. 1 and 2 show whether or not the light beams are deviated by observing the output difference of the plurality of photodiodes as shown in FIG. 4, and the + -shaped solid line portion in FIG. The effect is generated by forming the oxide film on the solid line portion of the square shape and the solid line portion of the square shape including both of them. In particular, the first + -shaped portion has the effect of increasing the output of the photodiode of FIG. 8 from the wavy line to the solid line. Returning to FIG. 1, the center LOCOS oxide film corresponds to this, and the center of the beam spot is drawn as the center of the isolation region. That is, without the LOCOS oxide film, the light incident on the high-concentration separation region generates photocarriers, and the mobility of the carriers is slow or the carriers disappear, but if the LOCOS oxide film is present, Since it is bent as shown in FIG. 9, it can be distributed to both photodiodes by this amount, and its value can be increased like the output of the solid line in FIG. In this part, ideally, half of both outputs becomes the maximum output, and it is advantageous that the calculation accuracy can be improved by bringing the output as close to this as possible.

On the other hand, in a device in which one photodiode and an arithmetic element are formed in a pair, that is, in an element that basically determines whether light enters or does not enter, the photodiode PD is replaced by LOCOS as shown in FIG. Surrounding is significant. That is, the light that originally enters the separation region does not contribute much as described above, but by providing LOCOS,
As indicated by the arrow in FIG. 3, the light incident on the separation region can be refracted and incident on the photodiode region, and a large output of the photodiode can be obtained. The same can be said for the solid line portion of the mouth shape on the outer side of FIG.

Next, the reason why the diffusion depth of the third isolation region is made shallow will be described. That is, in the second epitaxial layer 23, if the upper diffusion length of the second isolation region 26 is diffused to a length exceeding half the thickness of the second epitaxial layer 23, the third isolation region becomes Epitaxial layer 2
Since it is less than half the thickness of 3, the lateral diffusion of the third separation region 27 can be shortened accordingly (T3 << L2 / 2). For example, when the beam spot 1 in FIG. 4 is irradiated to the central separation region 24 in FIG. 1 and FIG. 2 and the vicinity thereof, the width of the third separation region 27 is short, so that carriers generated by the light irradiated here are generated. Since what is eliminated is suppressed and flows to the adjacent photodiodes PD1 and PD2, it is possible to suppress the delay of the response speed and simultaneously increase the output thereof as shown by the solid line in FIG.

Although only the lateral direction of the third separation area 27 has been considered above, FIG. 2 also considers the second separation area. That is, the upper diffusion length of the first isolation region is set to a length that exceeds half the thickness L1 of the first epitaxial layer 21, the tip of the second isolation region 26 overlaps with the first isolation region 25, and If the thickness exceeds the half of the thickness L2 of the second epitaxial layer 23, the second isolation region 26
Also, since the lateral diffusion length of the third isolation region 27 can be shortened, both lateral widths can be shortened. Looking at FIG. 7, 90% of the light is absorbed at about 7 μm, and 1 μm from the surface.
m, 1 μm to 2 μm, 2 μm to 3 μm, the deeper the absorption becomes. Therefore, in the isolation region, if the absorption is greater toward the surface and the width thereof is shortened, carriers that contribute to the output of the adjacent photodiode do not disappear and flow to the photodiode.

Further, the thickness L1 of the first epitaxial layer
By making the second epitaxial layer L2 thinner than that, the lateral width of the isolation regions 26 and 27 can be further shortened. For example, FIG.
When the thicknesses L1 and L2 of the first and second epitaxial layers are the same and the upper diffusion length of the first isolation region and the lower diffusion length of the second isolation region are substantially the same, the second isolation region Has a width of about + 2 × 0.8 × L1 / 2 of the diffusion hole. However, as shown in FIG. 2, if L2 <L1 and the upper diffusion length of the first isolation region is set to exceed half the first epitaxial layer thickness L1, the lower diffusion length of the second isolation region becomes L1. Less than 1/2, the second isolation region of less than L1 / 2 diffuses into the second epitaxial layer 23, the thickness of the second epitaxial layer 23 is smaller than that of the first epitaxial layer 23, and When the thickness L2 / 2 of the second epitaxial layer 23 is exceeded, the second isolation region 26 and the third isolation region 27 become shorter than the width + 2 × L1 / 2 of the diffusion hole described above.

For example, if L1 (about 3.5 μm) + L2 (about 3.5 μm) is about 7 μm and the diffusion depth of the first to third isolation regions is about 2 μm, the width of the third isolation region is Is about 4.2 μm. (However, the width of all the diffusion holes is set to 1 μm.) However, L1 (about 4 μm) + L2 (about 3 μm) is set to about 7 μm, and the upper diffusion of the first separation region is 2.6 μm and that of the second separation region. Vertical diffusion of 2 μm, third
If the diffusion depth of the isolation region is about 1.5 μm,
The widths of the second and first isolation regions are about 3.4 and 4.2.
And 5.2 μm (provided that the width of all diffusion holes is 1
μm. ).

Even if the epitaxial layer is 7 μm in total, the remaining 10% of the light reaches the substrate 21 by passing through the epitaxial layer of 7 μm. However, since the light is refracted, a part of the remaining 10% has an apparently thick epitaxial layer in the traveling direction of the light. In particular, in an element whose ON / OFF is controlled by one photodiode as shown in FIG. 1, the light passing through the LOCOS oxide film is refracted and apparently passes through the semiconductor layer for a long time, so that it reaches the substrate. Light that does not work as the generation of photocarriers can be taken into the semiconductor layer.

[0038]

As described above, according to the present invention,
First, in an optical semiconductor integrated circuit that compares the outputs of at least two optical elements, a third semiconductor element formed between two optical elements is used.
Of the first layer or the second layer that forms two optical elements by refracting the light incident on the third isolation region into the second isolation region.
The depletion layer in the semiconductor layer is a photocarrier which originally contributes to the output of the photo IC while being generated in the isolation region and disappearing or colliding with impurities in order to provide a means for making the light incident on the second semiconductor layer. Will occur within.

That is, the outputs of the two optical elements can be made large,
Moreover, if the beams are deviated, the output difference between the two optical elements can be made large, and a malfunction of calculation can be prevented. Secondly, in an optical semiconductor integrated circuit in which one optical element is formed on an IC substrate, the light incident on the third isolation region is supplied to the third isolation region formed around the optical element. When a means for refracting the light and making it incident on the first or second semiconductor layer forming the optical element is provided, the enclosed means originally generate and disappear in the isolation region, and impurities are generated. Optical carriers that contribute to the output of the optical IC while colliding with are generated in the semiconductor layer, so that a large output can be obtained.

Thirdly, if the means is an Si oxide film having an inclination in which the interface with the third isolation region gradually decreases in thickness toward the optical element, as shown in FIG. At a certain Si oxide film, the beam is refracted toward the optical element, so that a large output can be obtained. Fourth, if the second isolation region is set to exceed more than half the thickness of the second epitaxial layer, the third isolation region may diffuse slightly from the surface of the second epitaxial layer. As a result, the width of the third separation region can be shortened. That is, the solid line region of the + character in FIG. 4 is this separation region, and its lateral width can be made extremely short, and considering that the surface has a higher light absorption rate as shown in FIG. The carriers generated by the spots are reduced in disappearance and enter the adjacent photodiodes PD1 to PD4 in a well-balanced manner. Therefore, the output current of the photodiode is increased as compared with the conventional one, and becomes as shown by the solid line in FIG. As a result, the positions of the beams are compared with each other by the adjacent arithmetic circuits, but the accuracy is improved.

Fifth, the upper diffusion of the first isolation region is made to exceed more than half the thickness of the first epitaxial layer, and the lower diffusion tip of the second isolation region overlaps with the tip of the first isolation region. By doing so, the lateral width of the second isolation region can be shortened, and further, if the upper diffusion of the second isolation region is set to exceed half of the second epitaxial layer thickness, the third isolation region becomes , Can be made shorter.

Particularly, if the thickness of the second epitaxial layer is smaller than that of the first epitaxial layer, the diffusion depth of the second isolation region may exceed half of that of the thin second epitaxial layer. Therefore, the diffusion depth of the third isolation region can be shallower. Therefore, the width can be narrowed to the second isolation region, the disappearance of carriers generated here can be suppressed, the carriers can be made to flow to the adjacent photodiode, and the output can be made higher.

[Brief description of drawings]

FIG. 1 is a cross-sectional view illustrating an optical semiconductor integrated circuit according to a first embodiment of the present invention.

FIG. 2 is a sectional view illustrating an optical semiconductor integrated circuit according to a second embodiment of the present invention.

FIG. 3 is a sectional view illustrating an optical semiconductor integrated circuit according to a third embodiment of the present invention.

FIG. 4 is a plan view showing a conventional optical semiconductor integrated circuit.

FIG. 5 is a cross-sectional view of a conventional optical semiconductor integrated circuit.

FIG. 6 is a sectional view illustrating a conventional optical semiconductor integrated circuit.

FIG. 7 is a diagram showing the absorptance of light from the Si surface in the depth direction.

8 is a diagram showing an output when a beam is scanned in a direction of an arrow in FIG.

FIG. 9 is a diagram illustrating refraction of light in a LOCOS oxide film.

FIG. 10 is a diagram illustrating refraction of light in a V groove.

Claims (5)

[Claims] 1. A two-layer semiconductor layer is laminated on a semiconductor substrate, at least two optical elements are incorporated in the semiconductor layer, and a second layer from the semiconductor substrate is provided between the two optical elements. The isolation region is surrounded by an isolation region that reaches the semiconductor layer, and the isolation region extends from a boundary between the semiconductor substrate and the first semiconductor layer, the first semiconductor layer, and the first isolation region. An optical semiconductor integrated circuit having a second isolation region extending from a boundary of a second semiconductor layer and a third isolation region extending from a surface of the second semiconductor layer to a lower layer, In the third separation region formed between the optical elements, the first layer or the second layer that forms the two optical elements by refracting the light incident on the third separation region Optical semiconducting device characterized by providing means for making the light incident on the semiconductor layer of Integrated circuit. 2. A two-layer semiconductor layer is laminated on a semiconductor substrate, an optical element is incorporated in the semiconductor layer, and the periphery of the optical element reaches from the semiconductor substrate to a second semiconductor layer. The semiconductor device is surrounded by an isolation region, the isolation region extending from a boundary between the semiconductor substrate and the first semiconductor layer, the first semiconductor layer and the second semiconductor layer. An optical semiconductor integrated circuit having a second isolation region extending from a boundary of a semiconductor layer and a third isolation region extending from a surface of the second semiconductor layer to a lower layer is formed around the optical element. In the third separation region, the light incident on the third separation region is refracted,
An optical semiconductor integrated circuit, characterized in that means is provided for making the light incident on the first or second semiconductor layer forming the optical element. 3. The method according to claim 1, wherein the means is a Si oxide film having an inclination in which the interface with the third isolation region gradually decreases in thickness toward the optical element. The optical semiconductor integrated circuit described. 4. In the second semiconductor layer, the upward diffusion length of the second isolation region is set to exceed 1/2 of the layer thickness of the second semiconductor layer, 4. The optical semiconductor integrated circuit according to claim 1, wherein the third isolation region is overlapped at the tip of the second isolation region. 5. In the first semiconductor layer, the first semiconductor layer is formed.
The upward diffusion length of the isolation region is set to exceed 1/2 of the thickness of the first semiconductor layer. In the second semiconductor layer, the upward diffusion length of the second isolation region is Second
It is set to exceed 1/2 of the layer thickness of the first semiconductor layer, and in the first semiconductor layer, the upward first isolation region and the downward second isolation region overlap at the tip, 2. The semiconductor layer of the second layer, wherein the third isolation region is overlapped at the tip of the second isolation region.
An optical semiconductor integrated circuit according to claim 2 or 3.