JP3538731B2 - Submount for receiving photodiode - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to components of a device for exchanging signals using light, and more particularly to components of a communication system for simultaneously transmitting and receiving light of two wavelengths. Not passive components such as light-emitting devices (LD, LED) and light-receiving devices (PD, APD), but passive components that serve as the basis for these devices.
Especially regarding submounts. In the case of a transistor which is a simple element, a metal case serves as a collector. That is, the bottom of the chip is directly attached to the case. In the case of a diode as well, the chip is directly fixed to the metal case.
The case itself is the anode or cathode. So no submount is needed.

An optical element (LD, PD, LED) is also a diode, and has only a cathode and an anode. Although it is a simple bipolar structure, the situation is slightly different. In addition to LDs and PDs, there are monitoring PDs, AMPs, and the like, and the chip side of the LDs and PDs may not be able to be grounded to increase the degree of freedom of the circuit. Therefore, the chip bottom of the optical element cannot be directly bonded to the package (case). In this case, it is necessary to electrically disconnect the chip bottom from the case.
Therefore, the insulator plate inserted between the case and the chip is a submount. A submount is also used for improving heat radiation, adjusting the thermal expansion coefficient, and adjusting the height. In this case, the submount is not necessarily required to be an insulator. In the case of a device housed in a metal case, a submount is a passive component inserted between a package (case) and a chip and having a function of holding the package and the chip. Therefore, metallized surfaces for bonding are provided on both surfaces.

[0003] In recent years, a planar mounting type transmitting / receiving device has been developed. In this case, the structure is such that the chip is fixed on a Si bench (Si platform) and attached to a lead frame and covered with a resin package. In that case, the chip is not attached to the lead frame.
The chip is mounted on a Si bench. It may be fixed directly or may contain a thin member to adjust its height. Such a rectangular member may also be called a submount. In that case, the submount is a passive component interposed between the Si bench and the chip. That's why the concept of a submount has also expanded. Therefore, in the present invention, it will be referred to as a submount or an optical component. The optical component is essentially a submount, which means that it is interposed between the case or bench and the chip.

[0004]

2. Description of the Related Art A submount is directly fixed on a case, and PD and LD chips are mounted thereon. A rectangular opaque insulator is metallized on the front and back surfaces. It must be metallized to bond cases and chips. Since no light enters from the case side, the submount is naturally opaque. A high thermal conductivity is also required to remove the heat of the chip. A dielectric such as ceramic is often used as a submount material. Many of the currently used submounts are made of alumina. There has never been a use where the submount itself must be transparent. So there is no transparent submount. The existing submount is m
/ I / m is a simple rectangular plate having a three-layer structure.

However, there is a submount having an opening. As described above, this is a case where a back (back) incident type PD is used in the surface mount device. We don't often use such things.

For example, 1.55 μm light is received, and 1.3
1.55 μm in a transceiver that transmits μm light
As shown in FIG. 1, an m-light receiving PD has been proposed. A V-groove 2 is formed in the center of the flat Si bench 1 in the vertical direction. The PD chip 3 is fixed near the end of the V groove 2. The PD chip 3 is of a back-illuminated type, and has a ring-shaped n-electrode 6 at the bottom of the chip, in which a multilayer filter 5 is formed. Above the PD chip, a light receiving section 4 composed of a p region is provided. A rectangular annular submount 7 is mounted below the n-electrode 6. There is a metallization pattern on the surface of the Si bench, and the submount 7 is soldered to the metallization. Although the n-electrode on the bottom surface of the PD can be directly bonded to the metallization, a submount having an opening is used here. An optical fiber 8 is embedded in the V groove 2 and fixed by an adhesive.
The end of the V-groove is a mirror surface 9. The light leaving the optical fiber 8 is reflected by the mirror surface 9, passes through the multilayer filter 5, enters the PD 3 from the bottom surface, and reaches the light receiving unit 4.

This is a device only for receiving light, and there is another module including a light emitting device (LD), which is separated by a WDM filter. Therefore, the received light may be directly guided to the PD, but here, the multilayer filter 5 is provided. Transmit light 1.3 μm, receive light 1.55 μm
Suppose The multilayer filter 5 has 1.3 μm on the light receiving surface of the PD.
The m band is removed and the 1.55 μm band is transmitted. Even though the 1.3 μm light and the 1.55 μm light are separated by the WDM filter, the extinction ratio is low and the light receiving side is 1.3 μm.
m light is slightly mixed. PD is 1.5 in 1.3 μm band
It also has sensitivity in the 5 μm band. If 1.3 μm of the transmission light enters the PD, crosstalk occurs between the transmission side and the reception side. Crosstalk must be minimized. In order to supplement the function of the WDM filter, a multilayer filter is inserted to further remove 1.3 μm light. Why conventional PD is 1.55μ even at 1.3μm
It will be explained whether m also has sensitivity.

Referring to FIG. 2, the internal layer structure of the receiving PD will be described. This is a standard PD of the front-illuminated type. 1 is incident on the back side, but the layer structure is almost the same and only the shape of the electrode is different.

On an n-InP substrate 12, an n-InP buffer layer 13, an n-InGaAs light receiving layer 14, an n-InP
The P window layer 15 is epitaxially grown. A Zn diffusion region 16 is formed in the upper central part. Since this is a front (top) incident type, a ring-shaped p-electrode 17 is provided on the Zn diffusion region 16. A portion surrounded by the p-electrode is an opening through which light enters, and an antireflection film 18 is formed here. This is combined so that light of a desired wavelength is not reflected by the dielectric multilayer film. The portion where the pn junction around the chip is exposed on the surface is protected by the passivation film 19. A uniform n-electrode 20 is provided on the back surface. The n-electrode (cathode) is used with a positive reverse bias and the p-electrode (anode) is negatively biased.

This PD has a wide sensitivity because it uses InGaAs as a light receiving layer. FIG. 3 is a graph showing a sensitivity curve of the InGaAs light receiving layer PD. The rising P around 0.95 μm corresponds to the band gap of the InP window layer. 1.
The fall R around 67 μm corresponds to the band gap of InGaAs. In the area Q in between, the sensitivity is almost uniform. Since the band gaps of InP and InGaAs include 1.3 μm and 1.55 μm, they are sensitive to light having both wavelengths of transmission light and reception light.

Although the WDM filter separates the transmitted light and the received light, it is insufficient.
That is, a multilayer filter 5 that blocks m light is added. 1.3 μm light is transmitted through WDM filter and multilayer filter 5
Is doubly blocked by

In the device shown in FIG. 1, a multilayer filter 5 and a submount 7 having an opening are overlapped, and a PD chip 3 is further attached. In this case, a submount is not required separately, so that the ring-shaped n-electrode of the PD may be directly attached to the filter.

However, the conventional example of FIG. 1 in which the 1.3 μm light blocking filter and the submount are overlapped has two problems. The submount is essentially a rectangular plate, but if light is to be transmitted, the window must be processed. Regardless of the round window or square window, the sub-mount must be machined, which increases the processing cost. Another problem is a multilayer filter. It is said that light of 1.55 μm is transmitted and light of 1.3 μm is blocked, but the multilayer filter has strong incident angle dependence. There is a desired extinction ratio only for normal incidence, but obliquely incident light has no function. Even with 1.3 μm light, a part of the light obliquely incident enters the PD. In addition, multilayer filters essentially eliminate unwanted light by reflecting it. It is not absorbed and extinguished. The multilayer film is formed by alternately laminating thin films having different refractive indices n ₁ and n _{2 so} as to provide transmission characteristics at a certain wavelength and reflection characteristics at a certain wavelength. The refractive index essentially only takes into account the real part.

That is, the multilayer filter merely rejects unnecessary light by reflecting it instead of absorbing it. Unwanted light survives. The light of the wavelength removed by the multilayer filter is simply not transmitted, and is reflected and scattered everywhere after being reflected by the filter. The scattered unnecessary light may enter the PD obliquely.
This causes crosstalk. In addition, even though there is a clear purpose of rejecting 1.3 μm light, combinations of film thicknesses are still diverse. Since the reflection differs depending on the incident angle, it is necessary to design the filter differently when the incident angle differs.

It is common practice to eliminate unnecessary light with a multilayer filter as described above, but it must be said that this method still has many disadvantages. WDM with just receiving module
There was a problem of crosstalk due to imperfect filters and the like. However, in the case of a transmitting / receiving module in which a transmitting device and a receiving device are integrated, crosstalk is more serious. LD that generates transmission signal in narrow space, PD that receives reception signal
Is provided adjacent to a narrow space.

Referring to FIGS. 4 and 5, the transmission / reception module
A conventional example will be described. In Fig. 4 by the glass block
A WDM filter 21 is formed. Other styles
A WDM filter may be used. Different transmission / reception routes
I just want to show that this is an example. Isosceles triangle
Multilayer mirror on the hypotenuse surface of columnar glass blocks 22 and 23
24 is formed by vapor deposition or the like. This is for 45 degree incident light
On the other hand, λ₁(Transmitted light) and λ₂(Received light)
It is a multilayer film that emits light. Three sides of WDM filter 21
, An LD 25, a PD 26, and a fiber 27 are provided. LD2
Transmitted light λ generated at 5₁Goes straight through the WDM filter 21
It enters the optical fiber 27 and propagates through it. Light in opposite direction
The light propagating through the fiber 27 is passed through the WDM filter 21.
The light enters the PD 26 after being reflected. Depending on the wavelength of the WDM filter
The ratio of selective reflection and selective transmission is called the extinction ratio. This
It is about 1: 1000 at best. So LD25
Λ ₁Although it is slight, it may be mixed into PD26.
You. This is still a problem because the paths of the received light and transmitted light are different.
Easy to suppress loss talk.

FIG. 5 shows a transmission / reception module in which transmission light passes very close to the PD. The optical fiber 31 is inserted into the housing 30, and the LD 32 is installed at the tip thereof. P
D33 is provided. A WDM filter 34 that diagonally crosses the fiber 31 is provided immediately below the PD 33. LD
The transmission light λ ₁ exiting from 32 enters the optical fiber 31 from the tip. λ ₁ passes through the WDM filter 34 and propagates to the left in the optical fiber. The received light λ ₂ propagates rightward in the optical fiber, is selectively reflected by the WDM filter 34, enters the back surface of the PD 33, is refracted, reaches the upper light receiving unit 35, and generates a photocurrent. At the same time communicate bidirectionally transmit light lambda ₁ of LD part toward the PD for imperfection of the WDM filter. Such leakage light intensity is substantial because the paths are close. Also, light that has exited the LD and did not enter the optical fiber may be scattered and enter the PD. The apparatus of Figure 1 in order to prevent light leakage and stray light such lambda _1,
The multilayer filter is inserted at the bottom of the PD separately from the WDM filter.

In simultaneous two-way communication using light of different wavelengths, P
If D is sensitive to both lights, some of the transmitted light will enter the PD, even with a WDM filter. Then, crosstalk occurs between the transmission unit and the reception unit. In order to compensate for the shortage of the WDM filter, the apparatus shown in FIG. 1 has a multilayer filter that reflects 1.3 μm at the entrance of the PD.

However, as described above, the multilayer filter is designed to completely reflect 1.3 μm of normal incidence. Therefore, 1.3 μm light that is not perpendicularly incident passes through. Since the angle dependence is strong, 1.3 μm light is not eliminated regardless of the direction from which light enters. 1.3 μm light obliquely enters the PD.

Is the dielectric multilayer film a reflection? Is it transparent? There is only one alternative.
Dielectrics are discussed by their complex index of refraction, and the imaginary part represents absorption. However, the imaginary part of the refractive index of the dielectric has little wavelength dependence.

[0021]

SUMMARY OF THE INVENTION It is an object of the present invention to provide an optical component which can be provided at the entrance of a PD so as to reject unnecessary light. It is a second object of the present invention to provide an optical component that can prevent leakage or stray light of transmission light from entering a PD in a transmission / reception module. It is a third object of the present invention to provide an optical component which does not have an angle dependency like a dielectric multilayer filter and can eliminate unnecessary light at any incident angle. A fourth object is to provide an optical component that is resistant to heat and easy to mount.

[0022]

The optical component of the present invention comprises a wavelength selective layer (epitaxial layer) that absorbs light of a wavelength to be removed on one or both sides of a substrate (semiconductor or insulator) that transmits light of a required wavelength. Layer). For example to form a wavelength selective layer which absorbs lambda ₁ on the transparent substrate through the lambda ₂ (epi layer). This can be used as a sub-mount for PD mounting of the transceiver module using a transmission light lambda ₁ and the receiving light lambda _2. Conventional submounts were opaque and it was not common to pass light. Although the opaque ceramic was partially cut away to allow the necessary light to pass, the submount body was opaque. However, the present invention provides a transparent submount. Although it is transparent, it is transparent for desired light but opaque for unnecessary light.

Since the submount is an indirect pedestal for mounting the PD chip to the package and must be soldered to the package or chip, a metallized surface is provided on both surfaces or at least one surface.
The metallized surface only needs to have an area necessary for soldering.
The central part needs to pass light. Therefore, the metallized surface is provided only on the peripheral portion of the front surface and the back surface. The central portion through which light passes may be exposed, but it is more preferable to provide an antireflection film.

A semiconductor crystal is optimal for the light-transmitting substrate.
However, the invention is not limited thereto, and a plate of glass, ceramic, dielectric, or the like can be used.

The semiconductor layer is also optimal for the wavelength selection layer (epi layer). The band gap of the wavelength selection layer is determined by the wavelength of light to be cut off. Once the bandgap is determined, the material is limited. In addition to the semiconductor thin film, a dielectric thin film or an amorphous thin film can be used as the wavelength selection layer. Although amorphous is not crystalline, it still has a band gap and can be defined without any doubt. Dielectric is sometimes used in the narrow sense of ferroelectric. Here, it means broadly an insulator excluding semiconductors and metals. That is, it is a material whose imaginary part of the dielectric constant is not so large.

It has been described that the wavelength selection layer may be amorphous as long as the band gap can be defined. In the same sense, polycrystal or single crystal may be used. If the material is the same, the band gap does not change whether it is polycrystal or single crystal. However, it is best that a semiconductor wavelength selective layer is epitaxially grown on a semiconductor single crystal substrate.

The light removing mechanism of the present invention is similar to the principle of a photodiode. However, it cannot be reverse biased like a photodiode. Because it is a passive component, it cannot actively apply voltage. Creating a depletion layer by reverse bias in the photodiode,
Here, light is photoelectrically converted into an electron-hole pair, and electrons are drawn to the n-region and holes are drawn to the p-region by a reverse bias voltage to be a photocurrent.

In the case of the present invention, no reverse bias exists because no voltage is applied. But that is no problem. The reverse bias in the PD is to allow a photocurrent to flow through the electrode. The photoelectric conversion itself that converts light into electron-hole pairs occurs irrespective of reverse bias. It happens whenever the light energy is greater than the band gap. The probability does not depend on the reverse bias.

What is different? The only thing is that the photocurrent does not flow without the reverse bias. Electron-hole pairs generated by photoelectric conversion recombine and disappear in the same layer. No photocurrent is drawn out to the outside. Since there is no bias, both electrons and holes do not move at high speed due to the electric field in the InGaAsP absorption layer. It only moves by diffusion, but only heat moves because there is no concentration difference. The average speed is almost 0 because of thermal motion. That means staying in the same place. Eventually, the electrons collide with holes and recombine.
Eventually, the light of λ ₁ has disappeared.

Then, what thickness is necessary for the absorption layer? That is a problem. In the absorption layer, an absorption coefficient α for light of an unnecessary wavelength can be defined. This is the energy decay rate when the unit length has advanced. It is 0 for light having an energy smaller than the band gap, but has a finite value for light having a wavelength larger than the band gap. When light travels the distance x, exp (-αx)
Attenuate. Assuming that the thickness is d, this thin film reduces unnecessary light to exp (−αd). Required attenuation rate is 1/1
When 00 or 1/1000 is determined, the thickness d is determined.

For example, InGaAs of λ _g = 1.42 μm
For sP, α for 1.3 μm is α = 10 ⁴ cm ⁻¹ . If d = 5 μm, this becomes 1/100 or less.

[0032]

DESCRIPTION OF THE PREFERRED EMBODIMENTS The submount (optical component) of the present invention comprises a light-transmitting substrate and a wavelength-selective layer. The translucent substrate itself transmits both necessary light and unnecessary light. The wavelength selection layer (epi layer) transmits only necessary light and does not transmit unnecessary light. It is absorbed not only simply not passing. In the present invention, unnecessary light is absorbed by the epi layer, and the unnecessary light disappears there.
Any incident light can be absorbed. This is completely different from a dielectric multilayer film.

The multilayer film is designed to reflect 100% of unnecessary light using only the real part of the refractive index. However, the multilayer film can only reflect light incident at a certain angle. It is useless for objects with random angles of incidence, such as stray light. Moreover, the reflected unnecessary light is further reflected and returns to the PD. Multilayer filters have these inherent disadvantages.

The present invention is far superior to a multilayer filter. Moreover, the present invention is not used as a filter but as a submount, that is, used as a base of a PD.

The principle of the present invention will be described. Semiconductors and insulators have a forbidden band (band gap) between the conduction band and the valence band, and the Fermi order is within the band gap. There are no free electrons because the Fermi order is not in the conduction band. Therefore, it is not a metal but a semiconductor or an insulator. The difference between a semiconductor and an insulator is relative to each other due to the width of the band gap and the amount of the dopant. Here, no distinction is made between insulators and semiconductors. If the impurity order does not exist in the band gap, light having an energy lower than the energy of the band gap is not absorbed. Therefore, it penetrates semiconductors and insulators.

The energy of the band gap is defined as Eg.
And the corresponding photon wavelength is λ_g= Hc / Eg (h
Is the Planck constant, and c is the speed of light). This is a van
It is called the gap wavelength. Characterize the semiconductor and insulator
Physical quantity. When light is absorbed by semiconductors and insulators
This means that electrons have transitioned across the band gap.
And So even energy above the band gap
Only light is absorbed by the semiconductor / insulator. I mean
In the case of a semiconductor / insulator having no impurity order, λ <λ _gLight of
Is absorbed and λ> λ_gLight is transmitted.

It is assumed that unnecessary light is λ ₁ and required light is λ ₂ . In the case of bidirectional communication, it corresponds to the transmission light λ ₁ and the reception light λ ₂ .
The bandgap wavelength of the substrate is _λg2 . Since the substrate transmits both lights, λ _g2 is shorter than either wavelength. The wavelength of the epi layer is λ _g1 . Since the epi layer passes through λ ₂ and absorbs λ ₁ , λ _g1 is shorter than λ _{2 and} longer than λ ₁ .

Λ ₁ <λ _g1 <λ ₂ (1) Substrate; λ _g2 <λ ₁ , λ ₂ (2) Conditions (1) and (2) are asymmetric. The condition (1) is essential, but the condition (2) for the substrate is optional.

As for the substrate, the substrate may _{satisfy the following condition} : λ ₁ <λ _g2 <λ ₂ (3). In this case, both the substrate and the epi layer satisfy the inequalities (1) and (3) of the same shape. In that case, the epi layer itself is unnecessary, and the predetermined object can be achieved only with the substrate. But even so, there's nothing wrong with the epi layer. About substrate (2)
The condition of + (3) may be used.

That is, in a broader sense, the present invention relates to an epilayer; λ ₁ <λ _g1 <λ ₂ (1) substrate; λ _g2 <λ ₂ (4)

Is required. That is, the substrate only needs to transmit necessary light (λ ₂ ), and unnecessary light may be transmitted or absorbed. Therefore, not only a semiconductor with a narrow band gap but also an insulator with a wide band gap may be used. The epi layer must transmit the required light (λ ₂ ) and absorb the unwanted light (λ ₁ ). This is limited in the band gap by the unnecessary light wavelength. The epi layer is often a semiconductor with a narrow band gap.

The conditions imposed on the present invention are broadly (1) +
As for (4), the case of (1) + (2) is most important in practice. The reason will be described below.

Since the epi layer can produce crystals having an arbitrary mixed crystal ratio, λ _g1 can be freely determined. Therefore, λ _g1 can be given as a value between λ ₁ and λ ₂ . However, a substrate needs to be mass-produced at low cost by a mature crystal growth technique such as the Czochralski method or the Bridgman method. In that case, the crystal structure cannot be freely selected. A large number of single crystal substrates which can be easily obtained are Si, GaAs, InP and GaP. Therefore, λ _g2 has little choice.
Next, combinations of substrate materials and epilayer materials will be described.
The epilayer is written on the left, and the substrate material is written on the right, such as (epilayer) / (substrate).

1. InGaAsP / InP 2. GaInNAs / GaAs 3. GaAsSb / InP 4. AlGaAsSb / InAs 5. TlInGaP / InP 6. TlInGaP / GaAs 7. TlInGaAs / InP 8. Si / Ge directly bonded 9. Ge / GaAs directly bonded 10. Ge / InP directly bonded 11. GaInAsBi / InP 12. BGaInAs / GaAs

Each of these combinations is applicable when the unnecessary light is 1.3 μm and the required light is 1.55 μm. The band gap wavelength λ _g1 of the epitaxial thin film is 1.4.
It can be set to about μm.

In the combinations of 8 to 10, the band gap wavelength of Si is 1.1 μm, the band gap wavelength of GaAs is 0.87 μm, and the band gap wavelength of InP is 0.92 μm. Ge has a band gap wavelength of 1.55 μm, but when bonded to Si, GaAs, and InP, the band gap wavelength λ _g1 can be reduced to about 1.4 μm.

This is because 1.3 μm light and 1.55 μm light are used for the transmission / reception module, and the boundary wavelength is 1.4 μm. The present invention can of course be applied to other combinations of wavelengths. In other combinations of wavelengths, since the boundary wavelength is not 1.4 μm, the combination of (epi layer) / (substrate) is also different.

[0048]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The present inventor has proposed a filter (for example, an InGaAsP epitaxial layer) having a constant removal ability for unnecessary light (for example, 1.3 .mu.m) regardless of the incident direction of light.
(For example, InP) is proposed on an optical component (submount) provided on a substrate. This is a very novel idea of using a semiconductor thin film to eliminate unnecessary light. The multilayer film has a strong direction dependency and is ineffective for eliminating stray light. In the present invention, unnecessary light is absorbed by using absorption of a semiconductor. In the first place, the only way to eliminate the directionality of the incident light that can be removed is to use an absorber.

Here, a bidirectional optical transmitting / receiving module in which the transmission light λ ₁ is in the 1.3 μm band and the reception light λ ₂ is in the 1.55 μm band is described as an embodiment. Unnecessary light for PD is 1.3 μm,
The required light is 1.55 μm and the boundary wavelength is about 1.4 μm. When the boundary wavelength is 1.4 μm, a high absorption rate can be obtained with a thin absorption layer, and a material having wavelength selectivity is I.
A quaternary epitaxial layer of nGaAsP is preferred.

FIG. 6 is a sectional view of an optical component (submount) according to the embodiment. An InGaAsP absorption layer 41 is formed on a semi-insulating InP substrate 40 by epitaxial growth using a chloride method. On the lower surface of the InP substrate 40, a metallized surface 42 is formed only at the peripheral portion. On the upper surface of the absorption layer 41, only the metallized surface 4
3 is formed. The metallized surface is necessary for soldering when fixing the chip to the package as a submount. The entire surface cannot be metallized because light must pass through. An antireflection film 45 made of a dielectric multilayer film is provided at the center of the lower surface of the substrate. The same antireflection film 46 is provided at the center of the upper surface of the substrate. This is a multilayer film that does not reflect light of a required wavelength. That is, here, it is an antireflection film for 1.55 μm.
However, the antireflection films 45 and 46 may not be provided.

Here, the InP substrate 40 has a thickness of 200 μm. The InGaAsP absorption layer 41 has a thickness of 5 μm. The absorption edge wavelength of InP is the rising P of FIG.
9 μm). Therefore, 1.3μm light is also 1.
55 μm light also passes through the InP substrate. The InGaAsP thin film 41 has selectivity.

InGaAsP has two mixed crystal parameters. Here, the notation of the parameter is omitted. One parameter decreases under the lattice matching condition with InP, but one degree of freedom still remains, so that the absorption edge wavelength can be freely determined. Here, the absorption edge wavelength λ _g1 = 1.42 μm.

FIG. 7 is a graph showing the wavelength dependence of the transmittance of the obtained filter. The horizontal axis represents the wavelength (μm), and the vertical axis represents the transmittance (%). The transmittance is almost 0% for a wavelength of 1.36 μm or less. Light having a wavelength less than this is all absorbed. Corresponding to the band gap of the InGaAsP absorption layer being 1.42 μm, the transmittance curve rises from 1.41 μm and shows 50% transmittance at 1.42 μm and 80% transmittance at 1.45 μm. 1.52
It becomes 100% in μm and keeps 100% transmittance thereafter. The ratio of the transmittance between 1.3 μm and 1.55 μm is 1: 1.
00 or more. This is the absorption coefficient 1 of InGaAsP.
× 10 ⁴ cm ⁻¹ , as estimated from the thickness of 5 μm.

Using this submount with a filter,
A PD was mounted thereon and the sensitivity of 1.3 μm light and 1.55 μm light was evaluated. FIG. 8 shows a state where the PD is placed on the submount. The submount of the present invention is in the lower half. An InGaAsP absorption layer 41 is epitaxially grown on the InP substrate 40, and a metallized surface 42 is provided on a peripheral portion of the back surface of the InP substrate 40, and a metallized surface 43 is provided on a peripheral portion of the front surface of the InGaAsP absorption layer 41. Antireflection films 45 and 46 for 1.55 μm light are deposited at the center. The back surface electrode (n-electrode) of the PD 47 is bonded to the metallized surface 43 on the upper surface.

This is a back-illuminated type PD as shown in FIGS. The back surface has a ring-shaped n-electrode and an opening, and the opening is covered with an antireflection film. The PD 47 has the same layer structure as that shown in FIG. An n-InP buffer layer, an InGaAs light-receiving layer, and an InP window layer 2 are epitaxially grown on an n-InP substrate, and a p region is formed in the center by zinc diffusion. The p electrode is a blind plate that covers the entire p region, unlike FIG. The portion 48 of the p region becomes a light receiving portion. The submount itself is fixed on a Si bench as shown in FIG. 8 and 5, the Si bench is omitted, but an Si bench or the like is actually used. FIG. 1 shows elements for receiving only, and FIG. 5 shows elements for transmitting and receiving. The present invention can be applied to any of them. When measuring the sensitivity ratio between 1.3 μm light and 1.55 μm light, in the case of FIG. 1,
The light is guided to the PD through the 1.3 μm light and the 1.55 μm light from the optical fiber 8 and the sensitivity is measured and compared. In such a module, 1:10 which is the same as the evaluation of the submount itself is used.
A sensitivity ratio of 0 was obtained.

[0056]

According to the present invention, a submount conventionally made of opaque ceramic is formed into a transparent plate. Light can be introduced from the submount to the PD. Since the epi layer that absorbs unnecessary light is provided on the substrate, unnecessary light can be prevented from entering the PD. In a transmission / reception type module, light of two wavelengths is separated by a WDM filter.
Still not enough. The optical component of the present invention can compensate for imperfect wavelength selectivity of the WDM filter. It is possible to prevent crosstalk in which transmitted light is mixed with received light. Although it is a simple part, the effect is great.

There may be a sense of resistance in using a semiconductor film for the submount. Although semiconductor elements seem to be expensive, the wafer size is now increasing and the wafer cost per chip is decreasing. With such a simple single-layer structure, epitaxial growth can be performed at very low cost. On the other hand, the dielectric multilayer film must be coated on the resin film (dielectric multilayer film / resin film), so that the layer structure is complicated and rather expensive.

The resin filter having the (dielectric multilayer film / resin film) structure is weak against heat. Therefore, a resin filter must be inserted after the LD or PD is die-bonded. There is also a problem in assembly. In this regard, the submount of the present invention can be manufactured by the same manufacturing method as that for LD and PD. Handling is the same as that of a semiconductor element, and can be handled in the same manner as a semiconductor chip such as a PD or LD in all aspects.

[Brief description of the drawings]

FIG. 1 is a sectional view showing an example of a light receiving module according to a conventional example in which a Si bench having a V groove and a back-illuminated PD are combined.

FIG. 2 is a vertical sectional view of a top-incidence type PD according to a conventional example having an InGaAs light receiving layer.

FIG. 3 shows a conventional PD having an InGaAs light receiving layer.
5 is a graph showing the sensitivity characteristics of FIG.

FIG. 4 is a diagram illustrating a WDM in which a PD and an LD are housed in the same housing.
FIG. 9 is a schematic configuration diagram of a transmission / reception module according to a conventional example in which a path of transmission light and a path of reception light are separated by a filter.

FIG. 5 is a schematic cross-sectional view of a transmission / reception module according to a conventional example in which a PD and an LD are provided on the same straight line of a Si bench, and transmission light and reception light are separated by a WDM filter provided in the middle of an optical fiber.

FIG. 6 is a sectional view of an optical component according to an embodiment of the present invention.

FIG. 7 is a graph showing the wavelength dependence of the transmittance of the optical component according to the example of the present invention.

FIG. 8 is a cross-sectional view showing a state in which a back-illuminated type PD is placed on a submount according to an embodiment of the present invention.

[Explanation of symbols]

1 Si bench 2 V groove 3 PD 4 Receiver 5 Multilayer filter 6 n electrode 7 Submount 8 Optical fiber 9 Mirror surface 12 n-InP substrate 13 n-InP buffer layer 14 n-InGaAs light receiving layer 15 n-InP window layer 16 Zn diffusion region 17 p electrode 18 Anti-reflective coating 19 Passivation film 20 n electrodes 21 Multi-layer filter WDM 22 Triangular prism glass 23 Triangular prism glass 24 Multilayer mirror 25 LD 26 PD 27 Optical fiber 30 Housing 31 Optical fiber 32 LD 33 PD 34 WDM filter 35 Receiver 40 InP substrate 41 InGaAsP absorption layer 42 Metallized surface 43 Metallized surface 45 Anti-reflective coating 46 Anti-reflective coating 47 PD 48 Receiver

Continuation of front page       (56) References JP-A-8-160259 (JP, A)                 JP-A-9-139512 (JP, A)

Claims (4)

(57) [Claims] 1. An independent member interposed between a package and a receiving photodiode chip and fixed to both the package and the photodiode chip. The member is a light-transmitting semiconductor substrate and a wavelength λ of light to be transmitted. more the wavelength selective layer of the semiconductor single crystal thin film that absorbs a wavelength lambda ₁ of light grown epitaxially formed on one surface or both surfaces of the substrate have a longer bandgap wavelength lambda _g than the wavelength lambda ₁ of the light to be rejected less than ₂ A submount for a receiving photodiode characterized by having no pn junction. 2. The sub-mount for a receiving photodiode according to claim 1, wherein metallized surfaces are formed in peripheral portions of upper and lower surfaces. 3. The sub-mount for a receiving photodiode according to claim 1, wherein the substrate is an InP single crystal, and the wavelength selection layer is an InGaAsP single crystal thin film. 4. The submount for a receiving photodiode according to claim 3, wherein the band gap wavelength of InGaAsP is in a 1.4 μm band.