JP3386011B2 - Semiconductor light receiving element - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention
_The present invention relates to a structure of a light-receiving element capable of satisfactorily blocking the influence of transmission light in an optical transmission / reception module in which transmission / reception is performed by one optical fiber using ₁ and λ ₂ . When one optical fiber is used for both transmission and reception, an LD that emits transmission light and a PD that receives reception light are provided on the same container or the same substrate. PD is a highly sensitive element. The LD has a strong power because it carries a signal to a long distance. Although the wavelengths are different, the light receiving element has a sensitivity to the transmitted light, and thus the transmitted light is perceived.

Sensing the transmitted light by the PD is called optical crosstalk. Sensing the transmitted light makes it difficult to detect the received light. Therefore, crosstalk needs to be suppressed as small as possible. There are two types of interactions between the transmitter and the receiver, electrical crosstalk and optical crosstalk due to electromagnetic coupling between electrical circuits. Both are difficult problems to overcome. Here, only optical crosstalk is considered.

Transceivers for transmitting and receiving with one optical fiber have been devised in various forms depending on the method of separating the transmitted light and the received light. The most common method is to use a wavelength demultiplexer to split the transmission light path and the reception light path. The optical crosstalk problem is relatively minor in the case of such spatial separation. As a more special transmission / reception module, there is a transmission / reception module in which PD and LD are arranged in a straight line so that transmission / reception paths are almost the same. The problem of optical crosstalk becomes more serious when the same transmission and reception paths are used.

[0004]

2. Description of the Related Art A typical example of a transceiver module is shown in FIG.
(Wavelength multiplexing two-way communication). LD on the station side₁By
Signal to the subscriber. PD₁By the subscriber side
Receive the signal from. Wavelength demultiplexer 2 sends the upstream and downstream signals
Divided into two different optical fibers 1 and 7 depending on the wavelength difference.
Separated. The wavelength of the downstream signal is λ_Two, The wavelength of the upstream signal
λ ₁And Both are up and down by one optical fiber 3.
Sent in the direction. On the subscriber side, it is transmitted by the wavelength demultiplexer 4.
The reflected light and the received light are separated into different optical fibers 5 and 6. Receiving
Shinmitsu λ_TwoIs PD_TwoTo receive by. Transmit light λ₁Is LD
_TwoCaused by. PD_Two, LD_TwoElectricity before
Although there is a circuit, illustration is omitted here. Send and receive with this description
Belief is what the subscriber sees. The rising light
The transmitted light and the downstream light are the received light. By the wavelength demultiplexer 4
PD to branch_TwoAnd LD_TwoSpatially separated from
It

FIG. 2 shows a case where _two wavelengths of light λ ₁ and λ ₂ are transmitted (wavelength multiplexing one-way communication). On the station side, a multiplexer 8 combines transmission lights having different wavelengths. On the subscriber side, wavelength separation is performed by the wavelength demultiplexer 4. Different P on the receiving side
The lights of λ ₁ and λ ₂ are separated and received by D ₁ and PD ₂ . Also in this case, optical crosstalk between PD ₁ and PD ₂ is a problem.

FIG. 3 shows a conventional PD module that can be used as a receiver in an optical communication system in which optical paths are separated as shown in FIGS. The PD chip 12 is attached to the center of the circular metal stem 10 having the lead pin 9 via the submount 11. On the stem 10, a cylindrical cap 14 having a lens 13 is aligned and welded. Furthermore, a cylindrical sleeve 15 is provided thereon. The ferrule 16 is inserted into the shaft hole of the sleeve 15. The ferrule 16 holds the tip of the optical fiber 17. The tip of the ferrule 16 is obliquely polished. A bend limiter 18 covers the sleeve to protect the optical fiber 17. The transceiver module of FIGS. 1 and 2 additionally includes a transmitter. It is the one in which the PD in FIG. 3 is replaced with an LD, and will not be described.

The present invention can be applied to a system including a wavelength demultiplexer as shown in FIGS. 1 and 2, and FIGS. 1 and 2 include a receiver as shown in FIG. 3 for explanation. The receiver uses a metal case and the optical fiber arrangement is three-dimensional. Although it has high performance, alignment is essential, manufacturing cost is high, and it is difficult to be widely used.

A surface mount type module has been studied as a cheaper light receiving element module. Figure 4 is a back-illuminated P
A conventional example of a surface mount module using D is shown. The present invention can be applied to such a light receiving element and will be described in advance. A vertical V groove 2 is formed in the center of the rectangular Si bench 19.
0 is formed by etching. At the end of the V groove 20, there is an inclined mirror surface 21. This is also produced at the same time by etching. The PD chip 23 is fixed just above the end of the V groove 20. This is a back-illuminated type and has a light receiving portion 24 on the upper surface. The light emitted from the optical fiber 22 is S
The light propagates in the V groove parallel to the i bench surface, is reflected upward by the mirror surface 21, enters the PD 23 from the bottom surface, and reaches the light receiving portion 24. The surface mount type module is easy to manufacture because it has no centering point.

Both the light receiving element modules of FIGS. 3 and 4 can be used to detect the received light separated by the wavelength demultiplexer of FIGS. 1 and 2. The wavelength demultiplexer can be manufactured, for example, by forming a wavelength selective branching waveguide on a Si substrate. However, there are various wavelength demultiplexers, and a prism type wavelength demultiplexer as shown in FIG. 5 can be used. In FIG. 5, transparent triangular prism glass blocks 25, 2
6, the dielectric multilayer film 27 is laminated on the hypotenuse surface, which has wavelength selectivity. The selectivity is such that light of different wavelengths emitted from the optical fiber 28 is reflected on one side and transmitted on the other side. However, here, it has a selectivity for combining the transmitted light and the received light. Due to the reciprocity of light, the same selectivity is only used for different purposes. The received light λ ₂ emitted from the optical fiber 28 is reflected by the multilayer film 27 and guided to the PD 30. The transmission light λ ₁ emitted from the LD 29 passes through the multilayer film 27 and enters the fiber 28. The present invention can be applied to such a transmission / reception module, and thus has been described in a preliminary manner.

However, the best application of the present invention is to a transmitting / receiving module in which the wavelength demultiplexer does not separate the paths. In order to distinguish it from the ones described so far, let's call it non-separable type. It was first proposed by the present inventor and can be said to be a special module. FIG. 6 shows a non-separated type. As the application of the present invention, the one shown in FIG. 6 is a favorite.

Although a Si bench is provided inside the housing 31, it is not shown here. The optical fiber 32 is provided in the vertical direction. LD to face the tip of the optical fiber
33 is attached. WDM in the middle of the optical fiber 32
There is a filter 35 for wavelength separation.
The PD 34 is located directly above the WDM 35. LD of LD33
₁ of the transmitted light is stronger example of 1 mW. This goes out through the optical fiber 32. The received light λ ₂ transmitted from the outside through the optical fiber is reflected by the WDM 35, enters the back surface of the PD 34, and is detected by the light receiving unit 36. The transmitted light λ ₁ is strong light and the received light λ ₂ is weak light. The paths to the WDM in the optical fiber 32 are the same. Although in the opposite direction and separated by WDM, some of the transmitted light may enter the PD. This causes optical crosstalk. Even a small percentage
Since the original transmitted light power is large and the received light is weak, the received light becomes a large noise that cannot be ignored.

When the conventional PD shown in FIG. 7 is used, it is not simply noise. It may happen that the transmitted light noise exceeds the received light itself and the S / N ratio far exceeds 1. The InP-based light receiving element shown in FIG. 7 is manufactured by forming the n-InP buffer layer 38, the n-InGaAs light receiving layer 39, and the n-InP on the n-InP substrate 37.
Starting from an epitaxial wafer on which a window layer 40 is laminated. A p-type region 41 is produced by Zn diffusion, and an annular p-electrode 42 is formed thereon. An antireflection film 43 is placed inside the p-electrode 42, and a passivation film 44 is placed outside.
An n electrode 45 is formed on the back surface. Incident light enters from the upper surface.

FIG. 8 shows the sensitivity characteristic of the light receiving element of FIG.
The P portion where the sensitivity decreases on the short wavelength side corresponds to the band gap of the InP window layer. Light of wavelengths shorter than that is not perceived because it is absorbed by the InP window layer. The R portion where the sensitivity decreases on the long wavelength side corresponds to the band gap of the InGaAs absorption layer. Light with a lower energy (here, the photon energy (hν)) is less than the band gap of the light-receiving layer and cannot be sensed. This light receiving element has sensitivity in a wide range Q from the band gap P of the InP window layer to the band gap R of the InGaAs light receiving layer. It has sufficient sensitivity for both 1.3 μm band and 1.55 μm band.

Therefore, the light receiving element of FIG. 7 is most commonly used for long-wavelength light used for optical communication. In the conventional PD shown in FIG. 7, 1 μm to 1.65 μm as shown in FIG.
There is even sensitivity. The sensitivity in such a wide range is an advantage in that the same light receiving element can be used for 1.3 μm and 1.55 μm. However, on the other hand, when it is used in a transmission / reception module, it has a drawback that it feels transmission light and causes optical crosstalk.

If the energy of the transmitted light is lower than that of the received light, crosstalk can be reduced by devising the light receiving layer of the PD. PD absorbs light having energy higher than the bandgap of the light-receiving layer and converts it into photocurrent, but cannot perceive light having energy lower than the bandgap. If a light-receiving layer having a band gap between the transmitted light energy and the received light energy is selected, a PD that selectively receives only the received light can be formed.

However, on the contrary, the transmitted light λ₁Is the received light λ_TwoYo
High energy (λ₁<Λ_Two) If such a hand
I can't use the steps. The bandgap of the absorption layer is devised
By doing so, you can suppress crosstalk
Absent. The present invention is directed to such cases. That is, send
The light wavelength is shorter than the received light wavelength. For example λ ₁
= 1.3 μm transmission, λ_Two= 1.55μm reception
That is the case. Wavelength with Y-branch as in FIGS. 1 and 2
The performance of WDM is insufficient even when using a duplexer.
Loss talk becomes large, and as shown in Fig. 6, the route is not separated.
In the case of, crosstalk is more remarkable.

In the transmitting / receiving module of FIG.
State why optical crosstalk occurs
U The device of FIG. 6 is actually a Si platform (Si
Intense transmitted light (λ from the LD placed on the bench)₁)
Does not enter the optical fiber
It also illuminates the resin. LD emitted light has a considerable angle
Because it spreads. Transmit light λ₁On the other hand, the Si bench is
It is transparent. This is the origin of the problem. LD
Etc. are covered with transparent resin. Space for Si bench and transparent resin
Transmitted light entering λ₁Is transmitted through Si, reflected and scattered
It Resin distribution, Si platform shape, element distribution
Depending on the position, various scattered light is generated. λ₁Reflection of scattered light
The trajectory is complicated. Seen from PD, Si platform
The whole area is λ ₁It seems that it is shining brightly.

Since it is not visible light, it is not visible to human eyes, but it is visible to PD. From different directions From different heights From λ ₁
Is incident on the PD. P from the back, top, and sides
Enter D. Therefore, the transmitted light that does not pass through the optical fiber enters the PD and causes crosstalk. Such crosstalk due to λ ₁ scattered light that does not pass through WDM cannot be suppressed even if the performance of WDM is improved.

However, since such a thing was not known, it was once thought that the transmitted light enters the PD due to the incompleteness of the WDM filter. If so, WDM to P
It was thought that it would be sufficient to devise to remove only the transmission light wavelength on the route to D.

Therefore, the present inventor uses a PD having an absorption layer having a wavelength selectivity that absorbs only the transmitted light λ _{1 and} transmits the received light λ _2.
He invented a PD with a very clever structure that it is provided in a chip. JP-A-11-83619 (Applicant: Sumitomo Electric Industries, Ltd., inventors: Miki Kohara, Hiromi Nakanishi, Hitoshi Terauchi, 199.
Application dated September 3, 7). FIG. 9 shows the basic structure of a PD having such an absorption layer. An n-type InGaAs light receiving layer 47 is provided on the n-type InP substrate 46. n-type InGa
In the central portion of the As light receiving layer 47, the p-type region 4 is formed by Zn diffusion.
8 is formed. An i layer 49 (depletion layer) is formed in contact with the pn junction. A p-electrode 50 is provided on the p-type region 48. p
A passivation film 51 (eg, SiN) is provided outside the electrodes so as to cover the pn junction. An n-type InGaAsP absorption layer 52 is provided on the back surface of the n-type InP substrate 46. A ring-shaped n-electrode 53 is formed further below. The central portion serves as an entrance window and is covered with the antireflection film 54.

FIG. 10 shows a more specific PD structure.
In practice, the n-type InP substrate 46 is used to improve the crystallinity.
An n-type InP buffer layer 56 is provided between and the InGaAs light receiving layer 47. Furthermore, I is formed on the InGaAs light receiving layer 47.
An nP window layer 55 is provided. The p region 48 is formed from the window layer 55 by Zn diffusion, and the p electrode 50 is formed. Since the InP substrate 46 and the InP buffer layer 56 are provided as described above, if the InGaAsP absorption layer 52 is not provided, the sensitivity lower limit of P occurs in the sensitivity curve as shown in FIG.

This is a back-illuminated type and is used for a module as shown in FIG. The InGaAsP absorption layer 52 is a novel point of this PD. Since it is a quaternary compound, the band gap and lattice constant can be freely selected. About semiconductor,
The insulator can absorb light with energy larger than the band gap and cannot absorb light with energy smaller than the band gap. Therefore, if the band gap of the absorption layer 52 is selected between the energy of the transmitted light λ _{1 and the} energy of the received light λ ₂ , it should be possible to selectively absorb only λ ₁ .

For example, the transmitted light is 1.3 μm and the received light is
Assuming 1.55 μm, the absorption layer bandgap wavelength is
It is about 1.4 μm. This absorption layer is 1 × 10¹⁸cm
^-3The carrier concentration (electron concentration) of
μm. It is a fairly thick absorption layer. Also the electron concentration
Make it quite high. Thick absorption layer is enough for 1.3 μm light
This is because it is absorbed in. This absorption layer is compatible with 1.3 μm light
Damping constant α is α = 10 ^Fourcm^-1Is. Film thickness is 5
If it is μm, the attenuation is exp (-αd) = 0.007 and
The excess rate is 0.7%, which is sufficiently small. Electron concentration is high
One is because the forward resistance of PD does not become too high.
It The other is that the absorption layer absorbs 1.3 μm light and
Although it makes a hole pair, it also means that holes are likely to recombine.
Since it is n-type, there are many electrons from the beginning, which is a problem.
Yes. Absorption layer absorbs transmitted light and selectively passes only received light
To prevent optical crosstalk between transmitted and received light
Is effective.

FIG. 11 shows the wavelength dependence of the transmittance of the InGaAsP absorption layer 52. The mixed crystal ratio of InGaAsP is selected so that the bandgap wavelength is λ _g = 1.42 μm. So it absorbs wavelengths shorter than 1.42 μm,
Wavelengths longer than 1.42 μm are transmitted. It is stated that the sensitivity of PD was 1 A / W for 1.55 μm light and 0.01 A / W or less for 1.3 μm light. Extinction ratio is 1
/ 100 or less (20 dB).

Since the back-illuminated PD is used,₁
And λ_TwoIs incident on the absorption layer 52, _TwoThrough,
λ₁Absorbs. In other words, the transmitted light is absorbed here.
Do not go to the light receiving layer. So the LD transmitted light enters PD
No optical crosstalk occurs. This is the optics of FIG.
It is effective in the system, and even in the case of FIG. 1 and FIG.
The 1.3 μm light that could not be separated by the
You can

It is a very clever and excellent invention. n-type In
An n-type InGaAs absorption layer 47 is provided on the upper side of the P substrate 46, and an n-type InGaAsP absorption layer 52 is provided on the lower side. Both are formed by epitaxial growth, but the manufacturing method is rather complicated because both sides are epitaxial. Since the absorption layer and the absorption layer are on the opposite side of the substrate, there is an advantage that electron-hole pairs generated by absorption disappear without affecting the absorption layer. Therefore, when considering the thickness, 1.3
It may be considered as only the photoelectric conversion exp (-αd) of the μm light. I hope this is small enough. Annoying 1.
It was considered that the crosstalk between the transmitted light and the received light can be completely eliminated because the 3 μm light can be removed.

[0027]

However, it has been found that this is not always the case. The invention of (1) is extremely effective for 1.3 μm leaked light that is directly incident from the back surface. This is because it always passes through the absorption layer. However, in the transceiver module, the light from the LD is not such a simple one, but is complicatedly scattered and irregularly reflected. Since the resonator length of the LD is short, the emission angle is wide in the first place. Since no lens is used, the emitted light spreads vertically and horizontally. No significant power of LD light enters the optical fiber. As shown in FIG. 6, in the non-separable type in which the PD and LD are placed on the fiber extension using the Si bench, the light from the LD illuminates the entire Si bench brightly.
This was unexpected. Si bench since it is transparent to λ _1, λ ₁ from the Si bench and a transparent resin
Scattered light is generated. That is, there is much scattered light because the light emitted from the LD does not enter the optical fiber, rather than because the WDM is incomplete. On the other hand, it is meaningless to improve the performance of WDM.

The light emitted from the LD enters the fiber and becomes the transmitted light.
Not λ₁The light is called stray light. Stray light is inside the housing of Figure 6
Can be said to be full. Since it is a back-illuminated type, is it the back side?
Λ ₁Does not always come. InP substrate is 200μm
There is a thickness of. Therefore, there is stray light entering from the side of the board. Up
There is also stray light entering from the surface. InGaAsP absorption below the substrate
Even if there is a collecting layer 52, it is a stray that entered from the PD top surface or PD side surface
Light (λ₁) Cannot be absorbed. Entered from the side
λ₁Stray light reaches the depletion layer 49 and p region 48 in FIG.
Generates electric current. That is, the bottom absorption layer 52 is
There is nothing but helplessness to the stray light from. Shi
Even the stray light entering the PD is from the upper side rather than the lower side.
It has become clear that there are more people.

It is also possible to widen the p electrode against stray light from above and cover the portion other than the p electrode with an opaque material. However, since the PD is manufactured by a wafer process, it is difficult to cut the wafer into a large number of chips, form PD chips, and then perform some processing on the side surface. The thickness of the side surface of the chip is as large as 200 μm, but this remains exposed. It is difficult to cover the sides and prevent stray light from entering from the sides.

The wavelength separation ratio in the WDM filter or WDM coupler is 15 dB to 20 dB. On the other hand, the receiver as a whole needs to have a wavelength separation ratio of at least 30 dB, preferably 40 dB. For example, LD is 1mW (= 0d
In Bm), the minimum reception limit of the received light emitted from the optical fiber is −
If the power ratio of the transmitted light incident on the PD to the received light is -10 dB, the wavelength separation is required to be 40 dB as a whole. If the ratio of transmitted light to received light is -15 dB (about 1/30) and the minimum receiving sensitivity is -35 dBm, a wavelength separation ratio of 50 dB is required.

It is an object of the present invention to provide a structure of a PD which prevents stray light of transmitted light from entering in a device which transmits and receives received light λ ₂ having a long wavelength and transmitting light λ ₁ having a short wavelength.

[0032]

SUMMARY OF THE INVENTION In the present invention, above the substrate
Immediately below the absorption layer at₁To selectively absorb I
An nGaAsP-based absorption layer is provided. On the board unlike
The absorption layer is provided on. The absorption layer is below the substrate ()
Only because it is on the board (invention)
is there. The absorption layer is provided immediately above the substrate and immediately below the light-receiving layer.
It Then, not only from directly below, but also from the side and diagonally below
₁All (1.3 μm light) is also absorbed by this absorption layer. Sucking
Crosstalk is significantly reduced because it is collected and does not reach the light receiving layer
To do. Bandgap wavelength of quaternary mixed crystal InGaAsP
λ_gIs λ₁And λ_TwoIntermediate value of (λ₁<Λ_g<Λ_Two)
To determine the mixed crystal ratio. The thickness d is preferably about 5 μm
Is 3 μm to 10 μm. If it is thin, 1.3 μm light is completed
I can't absorb everything. If it is thick, the epitaxial growth time will increase
The material cost also increases. More about thickness later
Describe. Carrier (electron) concentration must be fairly high
Is. 10¹⁸cm ^-3The degree is desirable. Carrier rich
If the degree is low, the forward resistance of the PD becomes large, which is not good.
It takes time to recombine the holes created by 1.3 μm light
It Unlike that, the disappearance of holes due to recombination is important. This
This will also be described later.

This structure is effective for any PD in which the periphery of the chip is exposed to the outside. It can be applied to any PD such as back-illuminated type, top-illuminated type, side-illuminated type, and waveguide type.

Further, the peripheral portion p is formed so as to surround the central p region.
When a region (referred to as a diffusion shielding layer) is provided, carriers due to light from the immediate side disappear here.

It is more preferable to form an InP window layer on the InGaAs light receiving layer. If an InP window layer is provided, pn
The surface of the junction is stabilized by a passivation film, which is effective in reducing dark current and ensuring long-term reliability.

It is also effective to provide a low-doped InP buffer layer between the absorption layer and the light-receiving layer to improve crystallinity. If the impurity concentration is high, the lattice structure may be disturbed and the crystallinity may deteriorate. Since the absorption layer has a high dopant concentration, the buffer layer restores the crystallinity.

The thickness d of the absorption layer will be described. I've said this a few times, but here's more. The absorption coefficient is α, the thickness is d, and the transmittance is T. If there is no reflection on the front and back,

T = exp (-αd) (1)

[0039] α changes depending on the crystal composition,
For example, in one example of the present invention, α = 1 × 10 ⁴ cm ⁻¹ when the composition has an absorption edge of 1.42 μm. This relationship is shown in FIG. Since 1.55 μm light is hardly absorbed in the absorption layer (T = 1), T passes through (1.55 μm light and 1.
It can be regarded as a number indicating the effect as a filter (excluding 3 μm light).

Filter effect = −10 log T = 4.343 αd (2) (As α ⁻¹ = 1 μm)

From FIG. 22, a filter effect of 10 dB (T =
10%), it is necessary that d = 2.3 μm or more. 20dB filter effect (T = 1%)
In order to obtain, a thickness of d = 4.6 μm is required. Considering the variation in the thickness of the epitaxial growth layer, 2
When trying to secure the filter effect of 0 dB (1%), d
= 4 to 6 μm, that is, a thickness of about 5 μm is preferable.

In terms of filter effect, it is better to increase the thickness of the absorption layer. However, if the absorption layer is too thick, the crystallinity will decrease. Since the absorption layer has a high dopant concentration and is a quaternary mixed crystal, a thick layer deteriorates the crystallinity. Even if the crystallinity is deteriorated to some extent, the crystallinity is recovered to some extent when an InP buffer layer is put on the crystal layer. Therefore, a thickness of about 10 μm is allowed.

If only the transmitted light is absorbed as described above, the problem of T is sufficient as described in. However, unlike the present invention, since the absorption layer is placed in the vicinity of the light receiving layer,
Another problem is the early recombination of carriers. Unwanted λ ₁ (1.3 μm light) is absorbed by the absorption layer to form electron-hole pairs, but if this flows, it becomes photocurrent and 1.3 μm.
m light becomes noise and is included in the signal. It must be recombined early before it becomes a photocurrent. This means that the recombination thickness is also necessary for the absorption layer. Since the absorption layer is n-type InGaAsP, if electron-hole pairs are formed, the electrons are majority carriers and a large number of conduction electrons already exist, so there is no particular problem.

Holes are minority carriers in the absorption layer. Since the n-type absorption layer has a high carrier concentration, an electric field is not effective (electric field = 0). However, since there is a hole concentration gradient toward the p region, holes diffuse toward the p region due to the concentration gradient. During diffusion, they interact with a large amount of electrons, recombine, and disappear. The length of diffusion until the recombination disappears is called diffusion length L _d . Actually, the diffusion length L _dh is defined as the square root of (D τ) where τ is the hole lifetime and D is the diffusion coefficient. L _dh = (Dτ) ^1/2 . Diffusion is not a movement in one direction, but a movement that goes back and forth, so D
It is itself defined as the square of the displacement x divided by the time t, the limit of x ² / t.

Generally, the electron diffusion length L _de in InGaAsP crystal is long and the hole diffusion length L _dh is short. If the p-type absorption layer is used to absorb λ ₁ , the electrons become minor carriers and the diffusion length is long, so that a photocurrent is easily generated. Since this is a problem, the n-type absorption layer is used to convert the minority carriers into holes. The diffusion length depends on the crystal composition but is not the only one. Even with the same composition, the higher the crystal purity, the less collisions occur, and the longer the diffusion length. Electron diffusion length L _de = 6.0 μ in a high-purity (n = 10 ¹⁵ cm ⁻³ ) InGaAsP crystal
m, hole diffusion length L _dh = 1.6 μm. A crystal with low purity has a shorter diffusion length. This is because the number of majority carriers increases and the recombination cross section increases (τ decreases). Generally, the diffusion length L _dh of holes decreases in proportion to the square root of the carrier concentration n of the n-type region. InGa with a fairly high concentration
Since the AsP absorption layer (n = 10 ¹⁸ cm ⁻³ ) is used, it is reduced to about 1/30 as compared with high-purity InGaAsP. That is, the hole diffusion length in the n-InGaAsP absorption layer is L _dh = 0.05 μm. This means that the hole concentration drops to 1 / e at 0.05 μm.

Since α is large, 1.3 μm light is mostly electron-hole pairs in the first part of the absorption layer. The holes start to diffuse from the first part, but if the thickness of the absorption layer is d = 5 μm, the holes are attenuated to exp (−5 / 0.05) ^{≈10 −44} . The reason why the carrier concentration n of the absorption layer is made considerably high is also to reduce the hole diffusion length L _dh (hereinafter simply referred to as L). Therefore, the lower limit of the carrier concentration of the absorption layer also depends on the thickness d of the absorption layer. If the d is large, the hole diffusion length may be somewhat long, so the concentration n may be slightly small.

Strictly speaking, it is necessary to consider the double phenomenon that unwanted λ ₁ is photoelectrically converted from the bottom of the absorption layer at the z position and holes are reduced by recombination. The rate of photoelectric conversion in z is αexp (−αz) dz. The holes generated here recombine by d and decrease, but the decrease ratio is e
xp {(z−d) / L} (L is the diffusion length). Although the absorption layer has a thickness of z = 0 to d, the ratio S of holes that survive to the end is

S = α∫exp (−αz) exp {(z−d) / L} dz = αL (1-αL) ⁻¹ {exp (−αd) −exp (−d / L)} (3)

It becomes α and 1 / L have the same element. α does not depend on the concentration of the absorption layer, but the hole diffusion length L is the concentration n.
by. When the standard concentration is set to a low doping of 10 ¹⁵ cm ⁻³ , the hole diffusion length is 1.6 μm as described above, so that L is smaller than the general concentration n.

L = 1.6 (10 ¹⁵ / n) ^1/2 (μm) (4)

It can be considered phenomenologically as follows. When this is put into the equation (3), an arbitrary S is obtained by the thickness d and the carrier concentration,

[0052]

[Equation 1]

Is the extinction ratio of transmitted and received light. This is 20
If you need dB (10logS <-20dB)
From here, the carrier concentration n and thickness d of InGaAsP are limited.
Can be calculated. For example, when d = 5 μm, n = 10¹⁸cm
^-3Then S = -35 dB and n = 10¹⁷cm^-3
However, since S = -30 dB, n is 10¹⁷cm^-3
But it is effective. The upper limit of n is to prevent deterioration of crystallinity.
Determined by The upper limit is 10 ¹⁹cm^-3Is.

10 ¹⁷ cm ⁻³ ≦ n ≦ 10 ¹⁹ cm ⁻³ (6)

The difference between the present invention and the above is as follows.
In short, it means that the solid angle Ω desired from the light receiving layer of the PD to the absorption layer is large. The width of the PD chip is W (300 μm to 500 μm). The distance from the light receiving layer to the absorbing layer is g. The solid angle Ω that wants the absorption layer from one point of the absorption layer is in the approximate range.

Ω = 2π [1-g / {(W / 2) ² + g ² } ^1/2 ] (7)

It is In the case of the present invention, the distance g between the absorption layer and the light receiving layer is about 2 μm to 10 μm, which is extremely short. But W / 2
Is as large as about 200 μm. Therefore, in the present invention, approximately Ω ≈
2π. However, in this case, since g includes the substrate thickness (200 μm to 300 μm), since this has a size similar to W / 2, Ω≈π. This simple geometrical difference clearly distinguishes the present invention from the present invention. Another difference is in manufacturing technology. Needs to epitaxially grow the absorption layer and the absorption layer on both sides of the InP substrate. It is necessary to epitaxially grow on one surface of the substrate, turn it over, and perform epitaxial growth again. In other words, it becomes double-sided epitaxial. Therefore, the process of manufacturing the epitaxial wafer is duplicated and the cost is high. In the present invention, the absorption layer and the light receiving layer are provided on the same surface of the substrate, so that the single-sided epitaxial process is simpler.

[0058]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS [Embodiment 1 (back-thinned type; basic
The present invention is applied to a back illuminated PD according to FIG.
PD will be described. n⁺N on the InP substrate 57⁺Type
InGaAsP light absorption layer 58, n-type InGaAs light receiving layer
Layer 59 is grown epitaxially. Board like this
The characteristic feature of the present invention is that the absorption layer and the light receiving layer are used.
n⁺-The InP substrate is, for example, a 200 μm thick carrier
(Electron) concentration is n = 3 × 10¹⁸cm^-3Is. n⁺
-The InGaAsP absorption layer has a thickness of, for example, 5 μm
Concentration is n = 10¹⁸cm^-3Is. n-InGaAs
The light-receiving layer 59 has, for example, a thickness of 3 to 4 μm and a carrier concentration of
Is n = 10 ¹⁵cm^-3Is.

Z is formed in the center of the n-type InGaAs absorption layer 59.
A p ⁺ region 60 is formed by n diffusion. The carrier concentration p is not constant because it is produced by diffusion, but n-type I
A pn junction 61 is formed where the carrier concentration becomes equal to that of the nGaAs light receiving layer 59 (p = n). Since the light-receiving layer 59 has a low carrier concentration n, the thick depletion layer (i layer) 62 is p
It occurs continuously under the n-junction. A wide p electrode 63 is formed on the p region 60. Since it is a back-illuminated type, there is no problem even if the p-electrode substantially covers the p-region. A passivation film 64 is formed on the peripheral portion. For example, it is a SiN film and protects the end of the pn junction 61. An annular n electrode 65 is formed on the back surface of the n-InP substrate 57. Since light enters from the back surface, the n-electrode 65 has a ring shape. The antireflection film 66 covers the center of the back surface of the n-InP substrate 57.

Incident light A of λ ₂ entered from the back surface is depleted layer 6
It is absorbed at 2 and forms an electron-hole pair. This is the light sensitive area. Due to reverse bias, electrons travel to the n-type region and holes travel to the p-type region, causing photocurrent to flow. This is normal P
This is a photoelectric conversion operation of D.

Similarly, the incident light B having a wavelength of λ ₁ (transmitted light wavelength) entered from the back surface is absorbed by the n ⁺ -InGaAsP absorption layer 58. The light B of λ ₁ does not reach the light receiving layer 59. Therefore, this PD has no sensitivity to the transmission light wavelength λ ₁ . That is not all. Even if the incident light C of λ ₁ enters the PD from diagonally below, it is absorbed by the absorption layer 58.
The obliquely incident light C does not reach the light receiving layer 59. The PD does not detect the obliquely incident light C. Even for obliquely incident light C, P of the present invention
D is insensitive. Insensitive to the λ ₁ lights B and C. In particular, it is important not to feel the obliquely incident light C.

FIG. 12 shows a basic form, and the structure as it is can operate. On the n-InGaAs absorption layer 59
It is even better to insert an nP window layer. InP substrate 57
It is effective to insert a buffer layer between the InGaAsP absorption layer 58 and the InGaAsP absorption layer 58 in order to improve the crystallinity.

[Second Embodiment (Back-thinned Type; Diffusion Shielding Layer)] FIG. 13 shows an example of application of the present invention to the back-illuminated type, in which a diffusion shielding layer is provided around the p-electrode. . InGa of ^{n +} type on top of the n ⁺ -type InP substrate 57
AsP light absorbing layer 58, n-type InGaAs light receiving layer 59, n
-Epitaxially grow the InP window layer 69. Board
The feature of the present invention is that it serves as an absorption layer and a light receiving layer, but an InP window layer 69 is further added here. By using the InP window layer having a wide band gap, it is possible to stabilize the passivation of the pn junction and to reduce the dark current. It also has an effect of giving the absorption edge P of FIG. 8 to the light from the outside.

N⁺-InP substrate is, for example, 200 μm thick
And carrier concentration is n = 3 × 10¹⁸cm^-3Is. n
⁺-The InGaAsP absorption layer is, for example, 5 μm thick and is a carrier.
Ya concentration is n = 10¹⁸cm^-3Is. n-InGaA
The light receiving layer 59 has a thickness of, for example, 3 to 4 μm, and has a carrier concentration of
Degree n = 10¹⁵cm^-3Is. n-type InP window layer 69
Has a thickness of 2 μm and a carrier concentration of n = 2 × 10
¹⁵cm^-3Is. n-InP window layer 69, n-type InG
In the central part of the aAs light receiving layer 59, p is formed by Zn diffusion. ⁺region
Form 60. A pn junction 61 is created. Thick depletion
A layer (i layer) 62 is continuously formed below the pn junction 61.
It A wide p electrode 63 is formed on the p region 60.

Not only that, but Zn is simultaneously applied to the peripheral portion of the upper surface.
Diffusion is performed to form the p ⁺ region 70. This p ⁺ region 7
The present inventor calls 0 as a “diffusion shielding layer”. It does not mean that the diffusion of electrons is prevented, but it means that a p region is formed by Zn diffusion and the influence of incident light on the peripheral portion is shielded. Although not necessarily a proper term, the present inventors have long used such a term to describe the peripheral p region. Peripheral p
A pn junction 71 is formed immediately below the region 70, and a depletion layer 72 is generated below it. The SiN passivation film 64 protects the exposed portions of the pn junctions 61 and 71. A ring-shaped n electrode 65 and an antireflection film 66 are formed on the back surface of the n-InP substrate 57.

The back-illuminated type λ ₂ incident light A from directly below reaches the central depletion layer 62 to form an electron-hole pair, which becomes a photocurrent. The transmission light (λ ₁ ) B incident from directly below and the λ ₁ light C from obliquely below are absorbed by the n ⁺ type InGaAsP light absorption layer 58. In addition to that, λ ₁ coming from the side
Incident light D is also not perceived. The reason is not due to the conventional absorption layer. The side incident light D generates electron-hole pairs in the depletion layer 72 of the diffusion blocking layer 70, but the holes are generated in the peripheral portion p.
Since it enters the region 70 and disappears, the current does not flow to the central p-electrode. The influence of the side incident light is canceled by the diffusion shield layer 70.

[Embodiment 3 (back-illuminated type; diffusion shielding layer +
(Reflecting film)] FIG. 14 shows a back-illuminated type in which a reflecting film is further provided in addition to the diffusion shielding layer in order to prevent the transmission light λ ₁ from entering from the upper surface. On the n ⁺ type InP substrate 57, an n ⁺ type InGaAsP light absorption layer 58, an n type InG
The aAs light receiving layer 59 and the n-InP window layer 69 are epitaxially grown. n-InP window layer 69, n-type InGaAs
A p ⁺ region 60 is formed in the center of the light receiving layer 59 by Zn diffusion. A pn junction 61 is created. Thick depletion layer (i
A layer 62 occurs below the pn junction 61. A wide p electrode 63 is formed on the p region 60.

At the same time, Zn is diffused to the peripheral portion of the upper surface, and p
^{A +} region (diffusion shield layer) 70 is produced. Peripheral p region 70
A pn junction 71 is formed immediately below the depletion layer, and a depletion layer 72 is formed thereunder. The portion where the pn junctions 61 and 71 are exposed on the surface is SiN.
It is protected by the passivation film 64. The center of the upper surface of the PD is covered with the metal p-electrode 63, and the outside thereof is covered with the reflective film 73. When the PD is viewed from above, it is entirely covered with the p-electrode 63 and the reflective film 73. 1.
The transmitted light of 3 μm (λ ₁ ) is downward incident B, oblique downward incident C,
There may be side incidence D, upward incidence E, etc. B to D can be prevented also in the second embodiment. In addition to this, in this embodiment, the upward incident E can be rejected by the reflective film. This is all right. No matter where stray light (λ ₁ ) comes from, it does not reach the central depletion layer 62 of the PD and is not sensed.

[Embodiment 4 (back-illuminated type; diffusion shielding layer +
Reflective film; buffer layer)] If a high-concentration InGaAsP absorption layer 58 is provided, the crystallinity may be disturbed. In that case, the crystallinity can be improved by adding a buffer layer having a low impurity concentration. The buffer layer can prevent the crystal defects and impurities of the absorption layer 58 from affecting the light receiving layer. Figure 15
Is an n-InP buffer layer 74 provided between the absorption layer 58 and the light receiving layer 59.

On the n ⁺ -type InP substrate 57, n ⁺ -type In
GaAsP light absorption layer 58, n-InP buffer layer 74,
The n-type InGaAs light receiving layer 59 and the n-InP window layer 69 are epitaxially grown. n-InP window layer 69, n-type I
In the central part of the nGaAs light receiving layer 59, p ⁺
A region 60 is formed. A pn junction 61 is formed, and a depletion layer 62 is formed below it. Zn is simultaneously applied to the peripheral area of the upper surface.
Diffusion is performed to form the p ⁺ region (diffusion shield layer) 70.
A pn junction 71 is formed just below the peripheral p region 70, and a depletion layer 72 is formed below it.

A p electrode 63 is attached to the central p region 60.
The SiN passivation film 64 allows the pn junctions 61 and 7 to be formed.
1 is protected. The center of the upper surface of the PD is the metal p-electrode 63.
Covered with. The outer side is entirely covered with a reflective film 73. The buffer layer 74 has a thickness of, for example, 2 μm to 4 μm, and the carrier concentration is about n = 10 ¹⁵ cm ⁻³ .
This is only because the buffer layer 74 is inserted between the absorption layer 58 and the light receiving layer 59, and can be applied to the configurations shown in FIGS. 12 and 13. The property that the incident lights B, C, D, and E of λ ₁ are not sensed is the same as in the previous embodiment.

With this structure, 500 μm square and 200 μm thick
The PD was prepared and the sensitivity was measured. It was confirmed that the high sensitivity was 1.0 A / W for 1.55 μm light and the low sensitivity was 0.01 A / W or less for 1.3 μm light. In addition, when light of 1.3 μm was emitted from the surroundings, the sensitivity to these lights was as low as 0.01 A / W or less. Next, when this PD was integrated with the 1.3 μm-LD in the configuration of FIG. 6, even when the transmission power was 1 mW, it was 1.5
A reception sensitivity of 5 μm light of −35 dBm (combined with an amplifier) was realized without crosstalk.

[Embodiment 5 (Top incident type; Diffusion shielding layer + Reflective film; Buffer layer)] The present invention can be applied to a top incident type PD. It is easy to rewrite the back-illuminated type shown in FIGS. 12, 13 and 14 into the top-illuminated type. However, here, a top-illuminated PD corresponding to FIG. 15 will be described. FIG. 16 shows a top-illuminated type P
3 shows an embodiment of D. ⁿ on the n ⁺ type InP substrate 57 ⁺
Type InGaAsP light absorption layer 58, n-InP buffer layer 74, n type InGaAs light receiving layer 59, n-InGaA
The sP window layer 86 is epitaxially grown. n-InG
A p ⁺ region 60 is formed by Zn diffusion in the central portion of the aAsP window layer 86 and the n-type InGaAs light receiving layer 59. A pn junction 61 and a depletion layer 62 thereunder are generated. At the same time, Zn is diffused to the peripheral area of the upper surface to form ap ⁺ region (diffusion shield layer)
70 is produced. A pn junction 71 is formed immediately below the peripheral p region 70.
However, a depletion layer 72 is generated thereunder.

A ring-shaped p electrode 75 is formed in the central p region 60.
Attach. Since the light is incident on the top surface, the p-electrode 75 has a ring shape for transmitting incident light. A central portion surrounded by the p-electrode 75 is a portion where light enters, and an antireflection film 76 for λ ₂ is formed. The pn junction 6 is formed by the SiN passivation film 64.
Protects 1, 71. The outside of the annular p-electrode 75 including the passivation film 64 is entirely covered with the reflective film 73. The back surface of the n ⁺ -InP substrate 57 is widely covered with the n-electrode 77. Since the window layer is made of InGaAsP instead of InP, the transmitted light (λ ₁ ) having high energy is absorbed and removed by the window layer. The window layer is also skillfully used.

Stray light (λ ₁ ) B from below can be blocked by the n-electrode 77. The stray light C and D from the side can be blocked by the InGaAsP absorption layer 58 and the diffusion blocking layer 70. The stray light E traveling from above to the peripheral portion is reflected by the reflective film 73.
The λ ₁ leaked light from the upper center is InGaAs as described above.
Excluded with P window layer. The stray light is further attenuated by the λ ₂ antireflection film 76. An antireflection film is a film in which two types of dielectrics having different refractive indexes are alternately stacked and transmitted when light of a specific wavelength is incident at a right angle without reflection. Light of other wavelengths is partially reflected. Also, there is reflection of light of a specific wavelength other than normal incidence. Refractive index,
Since there is a degree of freedom in thickness, λ ₁ at normal incidence can be reflected and λ _{2 at} normal incidence can be transmitted. In the case of top incidence, a WDM is always provided as shown in FIGS. WDM
Stray light from above is suppressed by the action of the InGaAsP window layer and the λ ₂ antireflection film.

[Embodiment 6 (waveguide type; buffer layer)]
So far, the application of the present invention to the back illuminated PD and the top illuminated PD has been described. Other than that, various types of PDs have been proposed. The present invention can also be applied to a waveguide type PD. A case where the waveguide type PD is required will be described with reference to FIG. A demultiplexing waveguide 79 is provided on the substrate 78. This includes a common path 80 and branched paths 81 and 83. In the coupling section 82, the spacing and distance are strictly defined to give wavelength selectivity. The optical fiber is coupled to the common path 80.
A waveguide type PD 84 is provided at the tip of the branch path 81, and an LD 85 is provided at the tip of the branch path 83. The LD 85 generates transmission light of λ ₁ and sends it out to the optical fiber from the branch path 83, the coupling section 82, and the common path 80. Received light λ ₂ from the optical fiber propagates from the waveguides 80 and 81 to the PD 84. The waveguide type PD is such that the received light can be directly received from the side on such a waveguide surface.

FIG. 18 shows an embodiment applied to a waveguide type.
You Since light enters from the side, Zn diffusion is the same as before.
Therefore, the p-type layer is grown on the n-type absorption layer without forming the p-region.
Let n⁺N on the InP substrate 57⁺Type InGaA
sP light absorption layer 58, n-InP buffer layer 74, n-type I
The nGaAs light receiving layer 59 and the p-InP window layer 87 are epitaxially formed.
Shall grow. Since the InP window layer 87 is p-type,
The interface between the window layer 87 and the light receiving layer 59 becomes a pn junction. To spread
Therefore, the pn junction end is not exposed on the upper surface. Light is the top
It does not enter from the bottom or the back. The entire upper surface is made of metal
Covered by pole 88. n⁺-InP substrate 57 is an n-electrode 8
It can be covered by 9. Received light λ _TwoK is the side
The light enters the n-type InGaAs absorption layer 59 as the direction light. This place
In the case of n-type InGaAs light receiving layer, on the InP substrate
⁺A type InGaAsP absorption layer 58 is provided. Electrodes 88, 8
As it is covered with 9, stray light entering from directly below, stray light entering from directly above
There is no such thing. Against stray light from diagonally below
The InGaAsP absorption layer 58 works effectively. This
As shown in Fig. 17, the waveguide type PD has an LD and a path in the first place.
Apart from that, it has a structure that makes it difficult for stray light to enter.
It is suitable for reducing optical crosstalk.

[Embodiment 7 (side incidence type; inclined side surface; diffusion shielding layer + reflection film; buffer layer)] The present invention can also be applied to a side incidence type PD. The waveguide type of the sixth embodiment is also a side-incident type, but in reality, the light receiving layer is thin and it is difficult to couple with the waveguide. The side-incident type referred to here is one in which a part of the bottom portion of the PD is obliquely cut out to refract side light obliquely upward. FIG. 19 shows that the present invention is applied to side incidence.

On the n ⁺ -type InP substrate 57, n ⁺ -type In
GaAsP light absorption layer 58, n-InP buffer layer 74,
The n-type InGaAs light receiving layer 59 and the n-InP window layer 69 are epitaxially grown. n-InP window layer 69, n-type I
In the central part of the nGaAs light receiving layer 59, p ⁺
A region 60 is formed. Below that, a pn junction 61 and a depletion layer 6 are formed.
2 is generated. At the same time, Zn diffusion is performed on the peripheral portion of the upper surface to form the p ⁺ region (diffusion shield layer) 70. A pn junction 71 is formed just below the peripheral p region 70, and a depletion layer 72 is formed below it.

A uniform p electrode 63 is attached to the central p region 60.
Kick Since it is incident on the side, no opening is required in the p electrode 63
It The pn junction 6 is formed by the SiN passivation film 64.
Protects 1, 71. Including passivation film 64
The outside of the p-electrode 63 is entirely covered with the reflective film 73. n
⁺The back surface of the -InP substrate 57 is widely covered with the n-electrode 91. Chi
And then cut the bottom corner of the chip diagonally
The inclined surface 90 is _Two) It becomes the part where L enters. Within
Since the outside refractive index is significantly different, λ_TwoIs upwards at slope 90
It is refracted and reaches the depletion layer in the center to become a photocurrent.

The stray light of the transmitted light (λ ₁ ) cannot enter the PD from below or above. Stray light from diagonally below is absorbed by the InGaAsP absorption layer 58.

[Embodiment 8 (side incident type; V-shaped notch; diffusion shield layer + reflection film; buffer layer)] The side incident type is a waveguide type, an inclined surface type, or a V groove 93 is dug in the bottom surface. There is also PD. FIG. 20 shows an application of the present invention thereto. M which is the received light λ ₂ from the side is the V groove surface 93 on the bottom surface.
It is reflected by and goes to the central depletion layer to generate a photocurrent. The other structure is the same as that described so far, and the description thereof is omitted.

[Embodiment 9 (side incidence type; mesa type; diffusion shielding layer + reflection film; buffer layer)] The side incidence type is a waveguide type, an inclined surface type, a bottom V-groove type, and a mesa type on the upper side. Some PDs have been etched. The received light is refracted by the mesa. FIG. 21 shows a mesa PD to which the present invention is applied. The received light λ ₂ from the side is reflected by the upper mesa surface 95 and travels to the central depletion layer to generate a photocurrent. The waveguide type, the inclined surface type, and the bottom surface V-groove type all have the epitaxial layer on top (referred to as epi-up), but they are mounted upside down (epi-down) on the substrate. The other structure is the same as that described so far, and the description thereof is omitted.

[0084]

As described above, the present invention provides a transceiver module using a short wavelength λ ₁ for transmission and a long wavelength λ ₂ for reception.
An absorption layer that absorbs ₁ is provided between the substrate of the light receiving element and the light receiving layer. A strong λ ₁ is emitted from the LD, which is scattered and becomes stray light, which fills the inside of the module. Since the light receiving element of the present invention has the absorption layer of λ ₁ in close proximity to the light receiving layer, it does not reach the light receiving layer of the light receiving element. As a result, the transmission signal can be eliminated from the light receiving element. Compared to the case where the absorption layer is provided under the substrate, a large amount of λ ₁ stray light obliquely from below can be prevented. It is possible to severely prevent the transmitted light from entering the light receiving layer and significantly reduce the optical crosstalk. In particular, it is extremely effective in the optical path non-separation type transceiver module as shown in FIG. Of course, the present invention can also be applied to the optical path separation type transceiver module as shown in FIGS.

[Brief description of drawings]

FIG. 1 is a schematic view of wavelength division multiplexing bidirectional communication as one application candidate of a light receiving element of the present invention.

FIG. 2 is a schematic diagram of wavelength division multiplexing one-way communication as one application target candidate of the light receiving element of the present invention.

FIG. 3 is a vertical perspective view showing an example of a light receiving element according to a conventional example used for optical communication.

FIG. 4 is a surface-mounted light-receiving module using a back-illuminated PD.

FIG. 5 is a schematic diagram of a transmission / reception module using a dielectric multilayer film WDM.

FIG. 6 is a schematic configuration diagram of a path non-separation type optical transceiver module in which PD and LD are provided on the fiber extension line.

FIG. 7 is a sectional view of an InGaAs photodiode according to a conventional example.

FIG. 8 is a sensitivity characteristic diagram of an InGaAs photodiode according to a conventional example.

FIG. 9 is a schematic cross-sectional view of a light receiving element previously proposed by the present inventor that senses only 1.35 μm light and senses only 1.55 μm light.

FIG. 10 is a more specific cross-sectional view of a light receiving element previously proposed by the present inventor, which receives 1.3 μm light and senses only 1.55 μm light.

FIG. 11 is a graph showing wavelength dependence of light transmittance of an InGaAsP absorption layer epitaxially grown on the back surface of the substrate as a 1.3 μm light absorption layer.

FIG. 12 is a sectional view showing a back-illuminated type light receiving element of the present invention.

FIG. 13 is a back-thinned type in which Zn is formed around the central p region.
Sectional drawing of the light receiving element of this invention which provided the diffusion shielding layer by diffusion.

FIG. 14 is a cross-sectional view of a light-receiving element of the present invention, which is a back-illuminated type, in which a diffusion shield layer is provided around the central p region, and the periphery of the p electrode is covered with a reflective film.

FIG. 15 is a cross-sectional view of a back-illuminated type light receiving element of the present invention in which a diffusion shield layer is provided around the central p region, the periphery of the p electrode is covered with a reflective film, and a buffer layer is further provided.

FIG. 16 is a cross-sectional view of a light-receiving element of the present invention which is a top-illuminated type, in which a diffusion shield layer is provided around the central p region, and the periphery of the p electrode is covered with a reflective film.

FIG. 17 is a plan view of an optical receiver module in which an LD and a waveguide PD are provided at two ends of a waveguide WDM.

FIG. 18 is a cross-sectional view showing a waveguide type light receiving element of the present invention.

FIG. 19 is a sectional view showing a side-incident light-receiving element of the present invention in which a bottom corner is obliquely cut away.

FIG. 20 is a cross-sectional view showing a side-incident light-receiving element of the present invention in which the central portion of the bottom surface is cut out in a wedge shape.

FIG. 21 is a cross-sectional view showing a side-incident light-receiving element of the present invention in which an upper surface side portion is cut out in a mesa shape.

FIG. 22 is a graph showing the relationship between the thickness d of the InGaAsP absorption layer and the transmittance T for 1.3 μm light.

[Explanation of symbols]

1 optical fiber 2 wavelength demultiplexer 3 optical fiber 4 wavelength demultiplexer 5 optical fiber 6 optical fiber 7 optical fiber 8 wavelength multiplexer 9 lead pin 10 stem 11 submount 12 PD chip 13 lens 14 cap 15 sleeve 16 ferrule 17 optical fiber 18 bend limiter 19 Si bench 20 V groove 21 mirror surface 22 optical fiber 23 PD chip 24 light receiving portion 25 glass block 26 glass block 27 dielectric multilayer film 28 optical fiber 29LD 30PD 31 housing 32 optical fiber 33LD 34PD 35WDM 36 light receiving portion 37n-InP substrate 38n -InP buffer layer 39n-InGaAs light receiving layer 40n-InP window layer 41p type region 42p electrode 43 antireflection film 44 passivation film 45n electrode 46n type InP substrate 47n type InGaAs light receiving layer 48p type region 9 depletion layer 50p electrode 51 passivation film 52InGaAsP absorbing layer 53n electrode 54 antireflection film 55InP window layer 56n-type InP buffer layer 57n ⁺ -type InP substrate 58InGaAsP light absorbing layer 59n-type InGaAs light receiving layer 60p ⁺ region 61pn junction 62 depletion layer 63p electrode 64 Passivation film 65n electrode 66 antireflection film 69n-InP window layer 70p ⁺ region 71pn junction 72 depletion layer 73 reflection film 74n-InP buffer layer 75p electrode 76 antireflection film 77n electrode 78 substrate 79 demultiplexing waveguide 80 common path 81 branch path 82 Coupling portion 83 Branch path 84PD 85LD 86n-InGaAsP window layer 87p-InP window layer 88p electrode 89n electrode 90 inclined surface 91n electrode 92n electrode 93V groove 94n electrode 95 mesa surface

─────────────────────────────────────────────────── ─── Continuation of the front page (56) Reference JP-A-11-83619 (JP, A) JP-A-63-14470 (JP, A) JP-A-3-4571 (JP, A) JP-A-3- 206671 (JP, A) JP 62-142375 (JP, A) JP 2000-77702 (JP, A) (58) Fields investigated (Int.Cl. ⁷ , DB name) H01L 31/0232 H01L 31 / Ten

Claims (11)

(57) [Claims] 1. An InG substrate including an InP substrate and an InGaAs light receiving layer provided on the InP substrate and having a pn junction.
An InGaAsP absorption layer having a shorter absorption edge wavelength than that of the aAs absorption layer, the same conductivity type as the InP substrate, no pn junction, a thickness of 3 μm to 10 μm, and a carrier concentration of 10 ¹⁷ cm ⁻³ to 10 ¹⁹ cm ⁻³ Substrate and InGaAs
A semiconductor light receiving element, characterized in that at least one is provided between the light receiving layers. 2. The semiconductor light receiving element according to claim 1, wherein an InP buffer layer is grown between the InGaAsP absorption layer and the InGaAs light absorption layer. 3. The semiconductor light receiving element according to claim 1, wherein a diffusion blocking layer made of a p region is formed around the pn junction of the InGaAs light receiving layer. 4. The semiconductor light receiving element according to claim 1, wherein an InP window layer is provided on the InGaAs light receiving layer, and a pn junction is provided in the InGaAs light receiving layer and the InP window layer. 5. InGaAs on the InGaAs light receiving layer
A P window layer (absorption edge wavelength of about 1.4 μm) is provided, and InGaA
4. The semiconductor light receiving element according to claim 1, wherein a pn junction is provided in the s light receiving layer and the InGaAsP window layer. 6. A back-illuminated type having a ring-shaped electrode on the substrate side, an opaque metal electrode above the light-receiving layer, and most of the light-receiving layer covered with the opaque electrode. The semiconductor light receiving element according to any one of 1 to 5. 7. The semiconductor light receiving element according to claim 3, wherein the diffusion shielding layer and the opaque electrode are entirely covered with a dielectric multilayer film for a wavelength λ ₁ to be blocked. 8. A waveguide type having an electrode covering the entire bottom surface on the bottom surface side of the substrate and covering the entire upper part of the light receiving layer with the electrode so that light is introduced into the light receiving layer from the side. The semiconductor light receiving element according to claim 1. 9. A substrate bottom surface has an electrode covering the entire bottom surface, and an electrode covering the entire surface is provided on the upper part of the light-receiving layer. The side surface of the bottom surface is obliquely cut out, and the side surface notch is formed. 5. The semiconductor light receiving element according to claim 1, wherein the semiconductor light receiving element is a side-incident type in which light is incident from and refracted to reach a light receiving portion. 10. An electrode that covers the entire bottom surface is provided on the bottom surface side of the substrate, and an electrode that covers the entire bottom surface is provided on the upper surface of the light-receiving layer. 5. The semiconductor light receiving element according to claim 1, wherein the semiconductor light receiving element is a side-incident type in which light is reflected by the cutout portion and enters the light receiving layer. 11. An electrode that covers the entire lower surface of the substrate, an electrode that covers the upper portion of the light-receiving layer is provided, both sides of the upper surface are notched in a mesa shape, and they are attached to the package by epi-down and incident from the side surface. 5. The semiconductor light receiving element according to claim 1, wherein the semiconductor light receiving element is a side-incident type in which light is refracted at the mesa surface and enters the light receiving layer.