JPH0484128A - Optical demultiplexer - Google Patents

Abstract

PURPOSE:To form the above optical demultiplexer which is smaller in size, is more highly integrated and operates on lower voltages by integrating optical branching circuits, optical gate switches formed by using the electricity absorbing effect of semiconductors and optical amplifiers. CONSTITUTION:This optical demultiplexer is constituted by integrating 1Xn optical branching circuits 14, optical waveguides of a pin structure optically coupled to respective branching destinations, i.e., the optical gate switches 15 formed by using the electricity absorbing effect of the semiconductors, further the optical amplifiers 16 optically coupled to that. The optical gate switches 15 as small as 200mum element length and have excellent characteristics of >=15dB extinction ratio at about 3V voltage and, therefore, the optical demultiplexer which operates on the sufficiently low voltage is obtd. Further, the series connection of the optical gate switches 15 is obviated even if the number of the optical branches is structurally increased. Since the light power attenuated by the optical branching is sufficiently made up by the optical amplifiers 16, the expansion of the number of the branches is extremely easy. The optical demultiplexer which is smaller in size, is more highly integrated and operates on the lower voltages is obtd. in this way.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to an optical demultiplexer used in ultrahigh-speed optical communication systems, optical switching systems, and the like.

[Conventional technology]

With the recent development of optical communication systems and optical switching systems,
It is desired to realize an optical demultiplexer that can process ultra-high-speed optical pulse trains or time-division optical signals at low speed. Such optical demultiplexers are required not only to operate at extremely high speeds but also to operate at low voltages, and to be easily miniaturized and integrated.

As an example of an optical demultiplexer, the 1× prototype produced by Haga et al.
A four-speed optical switch/demultiplexer is described in the IEE Journal of Lightweight Technology, 1985, Volume 3, Page 116. This uses a lithium niobate substrate and integrates three 1x2 switches consisting of a Y branch, an asymmetrical X branch, and an optical waveguide that provides phase modulation between them to form a 1x4 switch. 4G entered the wave path
The operation is realized in which the input optical signal IG b/s is outputted from each of the four optical waveguides as an optical signal of b/s. The total length of the element is as large as approximately 40 mm, and the operating voltage is 6.6 to 133 .mu.m.

[Problem to be solved by the invention]

In the conventional example, the operating principle of the 1x2 switch, which is the basic element, is the change in refractive index due to the electro-optic effect of lithium niobate, which requires a fairly long element length and also requires a relatively high operating voltage. There was a problem.

Therefore, with this element size, IX8. Expansion to lX16 and the like is difficult, and furthermore, application in ultra-high speed regions such as 20 Gb/s and 40 Gb/s is difficult to achieve without using complex structures such as traveling wave electrodes.

The purpose of the present invention is to achieve miniaturization, integration, and low voltage by integrating an optical branch circuit and an optical gate switch and optical amplifier using the electric field absorption effect of semiconductors, and to compensate for losses caused by optical branching. Furthermore, by manufacturing these on a high-resistance semiconductor substrate, the element capacitance of a single optical gate switch can be reduced, and a demultiplexer capable of ultra-high-speed operation can be provided.

[Means to solve the problem]

The optical demultiplexer of the present invention includes, on a high-resistance semiconductor substrate, an optical branching circuit that branches an optical signal input from one input optical waveguide into n beams, and a branching destination of each of the optical branching circuits. a semiconductor optical waveguide having a PIN structure in a positional relationship for optical coupling; means for applying an electric field to the semiconductor optical waveguide having the PIN structure; and each of the PIN structures.
The present invention is characterized in that it includes a semiconductor optical amplifier positioned to be optically coupled to a semiconductor optical waveguide having an il structure.

[Effect]

The present invention comprises an IXn optical branch circuit, an optical waveguide with a PIN structure optically coupled to each branch destination, that is, an optical gate switch using the electric field absorption effect of a semiconductor, and an optical amplifier optically coupled thereto. It has a configuration that integrates. The basic operation is that light is output only to the output boat whose optical gate switch is turned on, so the optical gate switches are turned on in order in synchronization with the optical pulse train or the bit rate of the optical signal. Operation as a demultiplexer can be realized. The optical gate switch used here is small with an element length of 200 μm, and has excellent characteristics such as an extinction ratio of 15 dB or more at a voltage of about 3 cm. Therefore, an optical demultiplexer that operates at a sufficiently low voltage can be obtained. Furthermore, due to the structure, even if the number of optical branches is increased, the optical gate switches are not connected in series, and the optical power attenuated by the optical branches can be sufficiently compensated for by the optical amplifier, so it is very easy to expand the number of branches. It's easy. Furthermore, the overall element length is approximately determined by the length of the optical branch circuit, and by optimizing this, sufficient miniaturization can be achieved. The speed of an electroabsorption type optical switch is determined by the capacitance of the device, and by using a high-resistance semiconductor substrate, the odd capacitance of the device can be significantly reduced.
It is stated in the Proceedings of the 1990 Spring National Conference of the Institute of Electronics, Information and Communication Engineers, pages 4 to 294 that a band of 40 GH2 or more can be obtained. Therefore, since the optical demultiplexer according to the present invention has a configuration in which an optical branch circuit, an optical gate switch using an electric field absorption effect, and an optical amplifier are integrated on a high-resistance semiconductor substrate, the switching speed of the gate switch is very fast. It is possible to process ultra-high-speed time-series optical signals,
Furthermore, it has the feature of low optical power loss.

〔Example〕

FIGS. 1(a), (b), and (c) are perspective views showing an embodiment of the optical demultiplexer according to the present invention, and A in the perspective view.
-A', BB' is a cross-sectional view. Here 1×4
In this case, a two-stage Y branch was used as the optical branch circuit. As for the material system, InGaAsP/
Using InP, the optical branch circuit section and optical gate switch section use a double-hetero (DH) optical waveguide structure with different waveguide layer wavelength compositions, and the optical amplifier has an active layer on top of the waveguide layer. The explanation will be based on the case where an evanescent wave coupling structure is used, but the material, composition, and structure are not limited to this, and include InGaAs/InAlAs system,
GaAs/AIGaAs-based materials, multiple quantum well (MQW) structures, etc. may also be used.

First, a manufacturing method of an embodiment of the present invention will be briefly explained using FIG. On a high-resistance InP substrate, the n''-InP cladding layer 2 is 0.5 μm thick, and the iI nGaAsP (band gap wavelength 1.45 μm) light absorption layer 3 is 0.3 μm thick.
, MOVP the p''-I nP cladding layer 4 to 0.8 μm.
The layers are sequentially grown using the E method, and a layered structure that becomes an optical gate switch is first grown. Next, the portion where the optical branch circuit is to be formed is removed by etching until the high-resistance InP substrate 1 appears on the surface, and a 1-InP cladding layer 5 of 0.5 mm is etched only on the etched portion. czm, 1-InGaAsP (band gap wavelength 1.1 μm) guide layer 6 is Q, 3 μm, 1-
An InP cladding layer 7 is selectively grown to a thickness of 08 μm by MOVPE.

Furthermore, the portion where the optical amplifier is formed is made of i-InGaAs.
The P light absorption layer 3 is removed by etching until it appears on the surface, and an i-InGaAsP (band gap wavelength 1.55 μm) active layer 8 is formed on it with a thickness of 0.15 μm and p” −1.
The nP cladding layer 9 is selectively grown again to a thickness of 0.65 μm using the MOVPE method.

As a result, a structure as shown in the cross-sectional view (FIG. 1(C)) approximately along line BB' is formed. 1- of the optical branch circuit 14
InGaAsP guide layer 6 and optical gate switch 15
- InGaAsP light absorption layer 3 and light absorption layer 3 have good optical coupling, and optical gate switch 15 and optical amplifier 1
6 is a common 1-InGaAsP light absorption layer 3 that absorbs light -3
Therefore, there is no problem with optical coupling.

Next, a SiO□ mask pattern is formed by a normal photolithography method in order to form an optical branch circuit, an optical gate switch, and an optical amplifier. The width of the stripe at this time is 1.5 μm. Using this 5i02 mask, n”-
Etching is performed until the InP cladding layer 2 and 1-InP cladding layer 5 are exposed on the surface, first forming an optical waveguide pattern with a high mesa structure, and then using a resist mask, one side of the waveguide of the optical gate switch and optical amplifier is etched. The n''-InP cladding layer 2 is further etched only in the portion where the p-side electrode will be formed, exposing the high-resistance InP substrate 1. After peeling off the resist mask, use any 5i02 mask as it is as a mask for selective growth. Uses high resistance InP
The entire structure is buried with a embedding layer 10. At this point, the optical branch circuit 14 is completed and an optical waveguide with a buried structure is formed.

In order to take out the n-side electrodes of the optical gate switch 15 and the optical amplifier 16, a hole of about 100 μm x 400 μm is made by etching on the opposite side of the part where the p-side electrode of the waveguide is formed, and the n+-InP cladding layer 2 is etched. exposed on the surface. Next, as shown in FIG. 1(c), a groove is provided between the optical gate switch 15 and the optical amplifier 16 in order to electrically isolate them. Finally, the AA' cross-sectional view (Fig. 1 (
As shown in b)), high resistance InP is directly placed on the waveguide of the optical gate switch and optical amplifier and on the high resistance InP substrate 1.
The p-side electrode 11 of the optical gate switch and the p-side electrode 1 of the optical amplifier are placed on the side where the buried layer 10 is located (on the right side of the waveguide in the figure).
2 while maintaining electrical independence, and the p-side electrode 11.
An n-side electrode 13 is formed on the side opposite to 12 where the n+InP cladding layer 2 is exposed, and the device is completed.

The substrate is polished to a thickness of approximately 100 μm, the element length of the optical gate switch portion is 200 μm, and the element length of the optical amplifier portion is 4 μm.
00 μm, and the area of each pad portion of the p-side electrode is 80 μm×80 μm. The interval between each optical gate switch is 250 μm, the branching angle of the Y branch of the optical branch circuit is 5 degrees, and the element size as an optical demultiplexer is 9 am.
It is as small as X 1 am, which is less than a quarter of that of the conventional example.

Next, the operation of the optical demultiplexer of this embodiment will be explained using FIGS. 1, 2, and 3. first first
The basic characteristics will be described using diagrams.

The wavelength of the incident light is 1.55 μm. The light incident on the 14 optical branch circuits passes through two Y-branches, is divided into four parts, and is guided to each optical gate switch. The branching angle of the Y branch is very small at 5 degrees, and the wavelength composition of the guide layer 6 at the optical branching circuit section is 1.1 μm, so excessive loss at the branching section,
There is almost no absorption loss, and the attenuation of the power of the incident light is almost solely due to branching. The light incident on the light absorption layer 3 of the optical gate switch is transmitted when the reverse bias voltage of the switch is 0.
In case (2), the optical gate switch is in the ON state, so the light passes through this as it is and is guided to the next stage optical amplifier. A current is constantly injected into the optical amplifier, and the light with a wavelength of 1.55 μm is sufficiently amplified here, and the loss caused by splitting the incident light into four is compensated for and output. When a reverse bias voltage is applied to the light absorption layer 3, light is absorbed by the electric field absorption effect and the optical gate switch is turned off. The extinction ratio (ON-OFF ratio) at this time is determined by the wavelength used, the wavelength composition of the light absorption layer, and the length of the absorption layer, but in the case of this example, an extinction ratio of 15 dB was obtained at a voltage of 3 Sufficient characteristics have been obtained as a gate switch.

Next, we will discuss switching speed. As mentioned in the section on operation, the switching speed or modulation band Δf of a switch using a field effect is approximately determined by the capacitance C of the element, and is expressed as Δf=1/(πCR). In the present invention, a high-resistance semiconductor substrate is used, and since the entire area under the pad of the p-side electrode is a high-resistance semiconductor layer, the parasitic capacitance of the element can be almost ignored, resulting in a structure in which the element capacitance is extremely small. In the case of the example, the capacitance of the entire device is 0.12 pF. Therefore, the modulation frequency band of the optical gate switch used in the present invention is 50 GHz or more, and ultra high-speed switching is possible.

Next, the operation of the optical demultiplexer will be explained using FIGS. 2 and 3. Here, a case is shown in which a 40 Gb/s optical pulse train of incident light is divided into four and converted into four 10 Gb/s optical pulse trains.

The incident optical pulse train passes through the optical branching circuit 21 and is divided into four parts, each of which reaches the optical gate switches 2223, 24, and 25 at the same time. Figure 3 shows the temporal relationship between the incident optical pulse train and the reverse bias voltages vl to V4 applied to each optical gate switch. The time required for this is 25 ps, which is the time equivalent to one bit, and an electrical signal is applied to each optical gate switch with a delay of 2 Sps compared to the previous stage. As a result, as shown in Figure 2, 40 Gb/
Each of the s optical pulse trains is converted into four 10 Gb/S optical outputs and output. As explained earlier, this optical gate switch is capable of ultra-high-speed operation, so it
It is easy to extract an optical signal of approximately b/s.

Here, 7 branches are connected in cascade as an optical branch circuit, 1×
4 optical demultiplexer has been described, but expansion to 1.times.8.1.times.16 with a similar configuration is also fully possible since losses due to branching can be compensated for by optical amplifiers. In this case, the optical branching circuit section becomes long, but if an optical branching circuit that vertically folds the light using a 90-degree mirror or the like is used, the device can be realized within a few steps.

〔Effect of the invention〕

As described in detail above, according to the present invention, an optical demultiplexer capable of ultra-high-speed operation can be obtained, which will greatly contribute to the realization of future ultra-high-speed optical communication systems and optical switching systems.

[Brief explanation of drawings]

FIGS. 1(a), (b), and (c) are a perspective view and a sectional view of each part showing an embodiment of the optical multiplexer of the present invention,
FIGS. 2 and 3 are diagrams for explaining the operation of the optical demultiplexer of the present invention. In the figure, 1 is a high-resistance InP substrate, 2 is an n+InP cladding layer, 3 is a 1-InGaAsP light absorption layer, 4.9 is a p''-InP cladding layer, 5 is a 1-InP cladding layer,
6 is i-InGaAs P guide layer, 7 is 1-In
P cladding layer, 8 i-I nGaAsP active layer, 1o
is a high resistance InP buried layer, 11.12 is a p-side electrode, 1
3 is an nl electrode, 14.21 is an optical branch circuit, 15, 22.
23.24.25 is an optical gate switch, and 16 is an optical amplifier.

Claims (1)

[Claims] On a high-resistance semiconductor substrate, there is an optical branching circuit that branches an optical signal input from one input optical waveguide into n beams, and a positional relationship that optically couples each branch destination of the optical branching circuit. A semiconductor optical waveguide having a certain PIN structure, means for applying an electric field to the semiconductor optical waveguide having the PIN structure, and a semiconductor optical amplifier positioned to be optically coupled to each of the semiconductor optical waveguides having the PIN structure. An optical demultiplexer comprising: