JP2625312B2 - Waveguide type matrix optical switch - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a waveguide-type matrix optical switch used in the field of optical communication and the like, and more particularly to a waveguide-type matrix optical switch having a high extinction ratio which is resistant to fabrication errors. .

[0002]

2. Description of the Related Art In order to spread optical fiber communication more widely, in addition to increasing the performance and cost of optical fibers and light receiving / emitting elements, in addition to optical branching couplers, optical multiplexer / demultiplexers, optical switches, and the like. The development of various optical circuit components has reached an essential stage. In particular, optical switches are expected to play an important role in the near future in order to freely switch optical fiber lines according to demand and to secure detours in case of line failure.

Conventionally, as the configuration of the optical switch, 1) bulk type and 2) waveguide type have been proposed. The bulk type is assembled using movable prisms and lenses as components, and has the advantages of low wavelength dependence and relatively low loss, but the assembly adjustment process is complicated, not suitable for mass production, and the disadvantage of high price And has not yet become widely used. The waveguide type is based on an optical waveguide on a flat substrate and uses photolithography and microfabrication technology to mass produce so-called integrated optical switches in a batch, and is expected as a future type of optical switch. Have been. In particular, it is expected that there is no other form in which a relatively large-scale (M × N) matrix optical switch having M input ports and N output ports can be practically constituted by a waveguide type. I have.

FIG. 1 is a conceptual diagram of a (4 × 4) optical switch as an example of an (M × N) matrix optical switch to which the present invention is applied. This (4 × 4) switch apparently has four input waveguides 1a, 1b, 1c, 1d and four output optical waveguides 2a, 2b, 2c, 2d of 4 × 4 = 16.
The optical switch element S has a configuration in which it intersects at 16 points, and the (2 × 2) optical switch element S as a minimum unit of the optical switch
, S33 are arranged. Such a matrix optical switch configuration is called “strictly non-blocking matrix optical switch”, and the input optical waveguides 1 a,
The optical paths of the four-channel signal light input to 1b, 1c and 1d can be distributed to the four output optical waveguides 2a, 2b, 2c and 2d.

For example, when an optical signal incident on the input optical waveguide 1a is output from the output optical waveguide 2b, the optical switch elements S03, S02, S01, S11, S21, S3
1 is formed. At this time, the optical switch element S
01, a bar path is formed in which light entering from the lower left waveguide exits from the lower right waveguide. In other optical switch elements, light entering from the lower left (or upper left) waveguide is shifted to the upper right (or upper left). Or a cross path exiting from the lower right waveguide is formed. To minimize the number of drive switches, the cross path needs to be formed when the optical switch element is off, and the bar path needs to be formed when the switch is on. In this case, only one switch S01 is used. It turns on and the other switches turn off. This is true for any other optical path. For example, an optical path from the input waveguide 1a to the output waveguide 2a is formed by turning on only the optical switch element S00 and turning off the other six optical switch elements S03, S02, S01, S10, S20, and S30. You. Thus, the number of optical switch elements that are turned on to form a bar path is always one, and the number of optical switch elements that form a cross path when turned off varies from 0 to 6. That is, in this (4 × 4) matrix optical switch, the number of optical switch elements through which an optical signal passes is a minimum of one and a maximum of seven.

Conventionally, attempts have been made to construct the above matrix optical switch using optical waveguides of various materials. Among them, thermo-optics utilizing the thermo-optical effect of a quartz optical waveguide on a silicon substrate has been proposed. The matrix optical switch is expected to be the leading candidate for a practical matrix optical switch because it has no inconvenient polarization dependence and excellent connectivity with optical fibers.

FIGS. 2 (A) and 2 (B) show examples of (4) in FIG.
× 4) is a configuration diagram of a conventional thermo-optical (4 × 4) matrix optical switch fabricated on a silicon substrate corresponding to the configuration diagram of the matrix optical switch, FIG. FIG. 2B is an enlarged plan view of the optical switch element. In FIGS. 2A and 2B, eight optical waveguides including four input optical waveguides 1a, 1b, 1c, and 1d constitute an input-side waveguide bundle 4a, and four output optical waveguides 2a.
Eight optical waveguides including a, 2b, 2c, and 2d constitute the output-side waveguide bundle 4b, but the actual arrangement in FIG. 2A is topologically equivalent to the configuration conceptual diagram in FIG. That is easy to understand.

[0008] These waveguide bundles 4a-4b are a series of silica-based single-mode optical waveguides formed on the substrate 3 by a known combination of a flame hydrolysis reaction deposition method and a reactive ion etching technique. Each of the 16 optical switch elements S00,..., S33 has a so-called Mach-Zehnder optical interferometer circuit type (2 × 2) optical switch configuration, as shown in FIG. 2B. . That is, 2
Portions of the optical waveguides 61a-61b and 62a-62b are close to each other at two locations to form directional couplers 63a and 63b. Their optical coupling rates are set to be 50% at the signal light wavelength. The optical path lengths of the two optical waveguides 61a-61b and 62a-62b connecting between the directional couplers 63a and 63b are equal to the length of the two optical waveguides.
The thermo-optical phase shifters 64a and 64b, which are composed of thin film heaters located on the optical waveguide, are set to be identical (symmetric) in a state where the thermo-optical phase shifters 64a and 64b are not operated (off state).

The Mach-Zehnder optical interferometer circuit type (2)
× 2) The switching input / output characteristics of the optical switch are as follows: the power coupling ratio of each directional coupler is k, the power of the input signal light to one optical waveguide is P ₁₀ , and the output signal light of the bar path and the cross path is If the powers are P ₁ and P ₂ , respectively, and the phase difference between the two waveguides connecting the two directional couplers is Δφ, it is expressed by the following equation.

[0010]

P ₁ / P ₁₀ = (1−2k) ² cos ² (Δφ / 2) + sin ² (Δφ / 2) (1)

[0011]

P ₂ / P ₁₀ = 4k (1−k) cos ² (Δφ / 2) (2) When the coupling ratio k = １ ／, in other words, when the directional coupler is a 3 dB coupler The input / output characteristics are as follows.
First, in the off state (the thin film heater is not energized), since the phase difference Δφ = 0, the signal light passes through the optical switch element in a cross path (61a → 62b, 62a → 61b).
On the other hand, the phase shifters 64a, 64a, and 63b are arranged such that an optical path length difference near 1/2 wavelength corresponding to an optical phase of 180 degrees (π radian) is generated in the waveguide between the directional couplers 63a and 63b.
Assuming that at least one of the switches 64b is turned on (power is supplied to the thin-film heater) and is in the ON state, that is, the phase difference is Δφ = π, the signal light passes through the optical switch element through the bar path (61a → 61b, 62a
→ It switches so as to pass at 62b). Thus,
A cross / bar switching operation as a (2 × 2) optical switch element is achieved. However, the conventional waveguide type matrix optical switch having such a (2 × 2) optical switch element as a basic unit has the following major problems in manufacturing.

[0012]

In order for the Mach-Zehnder optical interferometer circuit type optical switch element shown in FIG. 2B to operate ideally, the directional coupler 63, which is a constituent element, is required.
The coupling ratio between a and 63b is exactly 50% at the signal light wavelength.
Is an indispensable condition, but there are some manufacturing errors in the actual optical waveguide manufacturing process, and it is difficult to accurately set the coupling ratio to 50%. This is because the directional coupler is a very structure-sensitive optical element,
The optical waveguide width and the optical waveguide interval, and the core of the optical waveguide
This is because the coupling ratio is likely to fluctuate due to a slight process error such as a relative refractive index difference between the claddings.

When the coupling ratio of the directional coupler deviates from 50%, in the off state, the signal light passes through the cross path (6).
1a → 62b, 62a → 61b) does not pass 100%,
It leaks into the bar path (61a → 61b, 62a → 62b), and the existence of this so-called light crosstalk has been a major problem in the fabrication of the waveguide type matrix optical switch.

For example, when the coupling ratio is from 50% to 5%,
When the optical switch element is off by about 5% or about 5% to the smaller side, the bar path (61a →
61b, 62a → 62b), 1% of the signal light intensity leaked out, and in the (4 × 4) matrix optical switch configuration of FIG. 2A, an extinction ratio of only about 15 dB was finally obtained. This situation becomes more serious as the matrix scale becomes larger. For example, in an (8 × 8) matrix optical switch, the crosstalk characteristic eventually deteriorates to an extinction ratio of about 11 dB.

The setting of the coupling ratio of the directional coupler often involves an error of about 50% ± 5%, and in some cases, about 50% ± 10% in an actual optical waveguide manufacturing process. The above-described coupling rate sensitivity of the matrix optical switch has been the biggest obstacle in manufacturing an optical switch with high yield.

SUMMARY OF THE INVENTION An object of the present invention is to provide a waveguide type matrix optical switch which solves the above-mentioned drawbacks, is resistant to a setting error of a directional coupler, and has an excellent extinction ratio.

[0017]

In order to achieve the above object, the present invention provides a waveguide type matrix optical switch having a plurality of input terminals and a plurality of output terminals for switching and outputting input signal light. , A plurality of input optical waveguides respectively connected to the input end, a plurality of output optical waveguides respectively connected to the output end, and arranged at respective intersections of the input optical waveguide and the output waveguide. A plurality of optical switch elements, wherein each of the optical switch elements makes two optical waveguides close to each other at two locations on a substrate, and the two optical waveguides between the two locations are used for the signal light. Two directional couplers configured to have an effective optical path length difference of a half wavelength and having the same coupling rate, and are disposed on at least one of the two optical waveguides between the two directional couplers. And the effective optical path length difference Characterized by comprising an optical path length difference switching means for switching to an integral multiple of the wavelength of light, and a cross section which crossed the two optical waveguides.

[0018]

According to the present invention, a half-wavelength optical path difference is set between the two directional couplers of the Mach-Zehnder optical interferometer circuit constituting the optical switch element in the off state. Therefore, in the off state, the signal light passes through the Mach-Zehnder optical interferometer circuit as it is input to the optical waveguide (through state) without shifting to another optical waveguide. Further, in the present invention, since the two optical waveguides cross each other at an angle where the crosstalk can be ignored in the optical switch element, the signal light passes through the optical switch element in a cross path. These operations are valid regardless of the values of the two directional couplers, as long as the coupling ratio is the same.

On the other hand, in the ON state, the optical phase shifter provided on the optical waveguide constituting the Mach-Zehnder optical interferometer circuit is driven to cancel the optical path length difference corresponding to the half wavelength. Thereby, the state of the optical switch element including the intersection can be switched to the bar path.

[0020]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The embodiments of the present invention will be described below in detail with reference to the drawings. In the following embodiments, a matrix using a quartz-based single-mode optical waveguide formed on a silicon substrate as an optical waveguide and a thermo-optic Mach-Zehnder optical interferometer circuit type (2 × 2) optical switch as an optical switch element is used. The optical switch will be described. This is because this combination can provide a matrix optical switch having excellent connectivity with a single mode optical fiber and having no polarization dependence. However, the invention is not limited to these combinations.

Embodiment 1 FIGS. 3A to 3C are configuration diagrams of an optical switch element constituting a (4 × 4) matrix optical switch as a first embodiment of a waveguide type matrix optical switch according to the present invention. 3A is a plan view, and FIGS. 3B and 3C are
3A and 3B are enlarged cross-sectional views taken along line segments AA ′ and BB ′ in FIG. The overall plan view of the (4 × 4) matrix optical switch is the same as that of FIG. Silica glass optical waveguide 31 on silicon substrate 11
The two directional couplers 33a and 33a are located close to each other at a distance of several μm at two places.
b (see FIG. 3B). Directional coupler 3
The coupling ratio of 3a and 33b is set at a target of 50% by adjusting the waveguide interval and length of the coupling portion. An optical waveguide 31a- between two directional couplers 33a and 33b.
The effective optical path lengths of the optical waveguides 31b and 32a-32b are
The a-32b side is equivalent to one half of the wavelength of the signal light (1.3 μm in this case) as compared with 31a-31b;
It is set to be about 65 μm longer, and a thin film heater 34a as an optical phase shifter is provided above the clad layer 14 at a portion corresponding to the optical waveguides 31a to 31b (FIG. 3C). Optical waveguides 31a-31b and 3
2a-32b intersect at the intersection 35 at an intersection angle θ at which crosstalk can be ignored.

The region between the directional couplers 33a and 33b constitutes a weakly asymmetric Mach-Zehnder optical interferometer circuit having a half-wavelength optical path difference. After passing through the directional couplers 33a and 33b, the signal light input from the end of the waveguide 31a is output while remaining in the optical waveguides 31a-31b, and the signal light input from the end of the optical waveguide 32a is output from the optical waveguide 32a-31b. 3
It is output while staying at 2b. Since the signal lights that have passed through the optical interferometer circuit area intersect at the intersection 35, a cross path in which the matrix optical switch is turned off can be realized as the entire optical switch element.

Here, the output state when an optical path length difference between zero and λ / 2 is given to the waveguide arm of the Mach-Zehnder interferometer will be visually observed through mode interference.

The transition of the optical power at the directional coupler is as follows.
It can be expressed by interference of two modes, an even mode and an odd mode, that is, superposition. Each mode is represented on the imaginary space to represent a phase state.

FIG. 4 shows the state of the mode in each part of the switch element when the optical path length difference ΔL = 0 (that is, in the case of the conventional example). Here, the incident light is set to E _10, and vector notation the composite electric field on both waveguides center. The state at each of the positions (1) to (4) shown in FIG. 4 is as follows.

(1) The state where light is present on the waveguide I side means that the even mode and the odd mode are in the state shown in the figure, and the waveguide I side is intensified because both modes have the same phase. I
In I, it can be considered that they cancel each other because of the opposite phase.

(2) The odd mode has a slower propagation constant (phase velocity) than the even mode, and therefore rotates toward the imaginary axis as it propagates (assuming the even mode is on the real axis). When the phase difference becomes 90 degrees, the electric field strength of both waveguides becomes the same. This is the state after passing through the 3 dB directional coupler. It should be noted here that the electric field strength is the same, but the phase on the waveguide II side is advanced by 90 degrees.

(3) The state of (2) is maintained because there is no optical path length difference in the waveguide arm.

(4) By passing through the 3 dB directional coupler again, the phase difference between the two modes is further increased by 90 degrees, and this mode is reversed on the waveguide I side, contrary to the state of (1). The phases cancel each other out, and in the waveguide II, they are intensified due to the same phase, so that light exists only in the waveguide II. That is, the light has all been transferred from the waveguide I to the waveguide II.

From the above, when the optical path length difference ΔL = 0, the phase difference between the even mode and the odd mode occurs only in the directional coupler, and if the phase difference is not exactly π, 100%
Does not occur, and some light leaks to the through side.

FIG. 5 shows the state when the optical path length difference ΔL = λ / 2 (that is, in the case of the present invention). The state at each position (1) to (4) shown in FIG. 5 is as follows.

(1) and (2) Same as when optical path length difference ΔL = 0 (3) Only one arm (here, waveguide II) has a phase of π
Because of the delay, the electric field on the waveguide II side in (2) becomes 18
It will be in a state rotated by 0 degrees. This electric field can be considered as a combination of the even mode and the odd mode as shown in the figure.

(4) By passing through the 3-dB directional coupler, the phase difference between the two modes is further increased by 90 degrees, and all the light returns to the waveguide I side.

This is a phenomenon that always occurs regardless of the value of the coupling ratio as long as the coupling ratio of the two directional couplers constituting the interferometer is the same. This is because the odd mode is advanced from the even mode by an angle θ corresponding to the coupling ratio because the optical path length difference of π is given as shown in (3). This is because the relative relationship between the two modes returns to the original state (state of (1)) because of the delay by θ in the container.

This will be represented by a mathematical formula. The parameters of the Mach-Zehnder interferometer are as shown in FIG. That is,
The coupling coefficient of each directional coupler is δ, the coupling length is d, and the wave number is k.
(= 2πn _eff / λ ₀ ), the characteristic matrix of the directional coupler alone is

[0036]

(Equation 3)

Is as follows. δd represents the phase difference between the even and odd modes in the aforementioned mode. Therefore, 3 dB
In a directional coupler

[0038]

## EQU4 ## δd = π / 4.

The phase difference between the waveguide arms of the interferometer is

[0040]

(Equation 5)

## EQU1 ## The matrix notation for the electric field across the interferometer is

[0042]

(Equation 6)

When this is calculated and expressed by power,
The output characteristics are

[0044]

P ₁ = P ₁₀ [cos ² (2δd) cos ² (kΔL / 2) + sin ² (kΔL / 2)] + P ₂₀ [sin ² (2δd) cos ² (kΔL / 2)] (6) P ₂ = P ₁₀ [sin ² (2δd) cos ² (kΔL / 2)] + P ₂₀ [cos ² (2δd) cos ² (kΔL / 2) + sin ² (kΔL / 2)] (7) If only P ₁₀ is input, the optical path length difference [Delta] L = 0,
That is, the case corresponding to the conventional example is as follows.

[0045]

P ₁ = P ₁₀ cos ² (2δd) (8) P ₂ = P ₁₀ sin ² (2δd) (9) Also, the optical path length difference ΔL = λ / 2 (kΔL / 2 = π / 2),
That is, in the case of the present invention,

[0046]

P ₁ = P ₁₀ (10) P ₂ = 0 (11) When the optical path length difference ΔL = 0 as in the related art, δd = π
/ 4 unless (3 dB directional coupler), leakage light occurs from P _1. On the other hand, as in the present invention, the optical path length difference ΔL
= Λ / 2 indicates that leak light does not occur in principle at P ₂ irrespective of the coupling ratio of the directional coupler, which is consistent with the case of the above-described mode superposition.

From the above results, if the state of ΔL = λ / 2 is set to the off state, even if the coupling ratio of the directional coupler deviates from the ideal value of 50% due to a manufacturing error or the like, the two directions are not changed. As long as the coupling ratios of the sex couplers are the same, a high extinction ratio can be realized irrespective of the value, and crosstalk can be significantly reduced even as a matrix switch. However, in the case of configuring a matrix switch, in order to minimize the number of drive switches, it is necessary to configure so that light is output to the cross side in an off state, and therefore, it is necessary to exchange output ports. In the present invention, the two waveguides 31a-31b and 32a are formed within the switch element at an angle where no interference occurs between the two waveguides at the intersection 35.
This is achieved by crossing -32b.

On the other hand, the thin-film heater 34a is energized to slightly increase the refractive index value of the optical waveguides 31a-31b under the thin-film heater 34a by the thermo-optic effect so that the effective optical path difference of the optical interferometer circuit becomes zero. Then, the two directional couplers 33a,
The coupling rate of 33b acts additively as shown in FIG. Therefore, if the ideal coupling rate is set to 50%, the optical interferometer circuit as a whole appears to be 1%.
When the signal light passes through the optical interferometer circuit, the signal light exchanges with the optical waveguide and passes therethrough. Subsequently, by passing through the intersection 35, the optical switch element is switched to a necessary bar path as an ON state.
Ideal coupling ratio of the directional couplers 33a and 33b is 50%
If the value deviates from the value, the signal path cannot be switched to the 100% bar path, and a signal light component that remains in the cross state appears. However, this residual signal light eventually ends up in the output port 2a,
Since the light is emitted from unnecessary output side waveguides other than 2b, 2c, and 2d (see FIG. 2A), loss is increased, but it does not lead to deterioration of the extinction ratio as a matrix optical switch.

FIG. 7 is an explanatory diagram of an excess loss characteristic caused by light crosstalk in a (4 × 4) matrix optical switch, and shows values on a path passing through seven switch elements. Here, the solid curve is the characteristic of the optical switch of this embodiment, and the broken curve is the characteristic of the conventional optical switch. 5% for the higher directional coupler coupling ratio or 5% for the smaller directional coupler coupling ratio
% And 5% from the ideal coupling rate of 50%,
While the loss increase of the conventional optical switch is 0.25 dB, the optical switch according to the present embodiment has an extremely small loss of 0.05 dB, and the coupling ratio of the directional coupler is 35% or 6%.
Even if the coupling rate deviates from the ideal coupling rate by 5% to 5%,
While the loss increase of the conventional optical switch reaches as high as 2.5 dB, the loss increase of the optical switch of this embodiment is extremely small at 0.4 dB.

Here, it should be noted that as the size of the matrix optical switch increases, the tendency of the loss increase in the conventional optical switch becomes remarkable, whereas in the matrix optical switch of the present invention, the value of the loss increase ( 7 (solid curve in FIG. 7) is always the same. In the matrix optical switch configuration, each input signal light is output after passing through several switch elements, but only one of the passed switch elements is in the ON state, and the rest are in the OFF state. It is. In the conventional matrix optical switch, when the coupling ratio of the directional coupler deviates from 50%, light leaks from the switch elements in the off state, so that the size of the matrix increases and the number of switch elements in the off state passes. As the number increases, the excess loss increases. On the other hand, in the matrix optical switch of the present invention, since no leakage light is generated from the switch element in the off state, only the leakage light from one switch element in the on state in each input signal light becomes excessive loss. Thus, no matter how large the size of the matrix, the value of excess loss is always the same.

Next, the wavelength dependence of crosstalk will be described.

FIGS. 8A and 8B show the optical path length difference ΔL between the ON state and the OFF state of the switch element of the present invention and the conventional switch element. It can be seen that the two have the opposite relationship.

FIGS. 9A and 9B are explanatory diagrams of the transmittance-wavelength characteristic of the cross path of the optical switch element. FIG. 9B is a characteristic diagram of the optical switch element used in this embodiment. 2) is a characteristic diagram of the conventional optical switch element shown in FIG. In FIGS. 9A and 9B, the two solid curves show the cross-path light transmission characteristics in the off state and the on state, and the broken curves show the coupling ratio of the directional coupler constituting the optical interferometer circuit. The directional coupler has an ideal coupling ratio of 50% at a wavelength of 1.3 μm. It should be noted here that in FIG. 9A corresponding to the present invention, the wavelength dependence in the off state is small and the wavelength dependence in the on state is large, whereas in FIG. 9B corresponding to the conventional example. In
Conversely, the wavelength dependence of the OFF state is large, and the wavelength dependence of the ON state is small. In an actual (4 × 4) matrix optical switch, signal light passes through a maximum of seven optical switch elements, but six of them are in an off state except for one of them, so that the signal is in an off state. It is understood that the optical switch element of FIG. 9A having a small wavelength dependence is more advantageous in reducing the wavelength dependence of the matrix optical switch as a whole.

Next, the wavelength dependence in the cross path will be described using mathematical expressions.

First, the wavelength dependence of a conventional switch element in a cross path will be examined.

In the conventional switch element, the optical path length difference ΔL = 0 in the off state. Therefore, the output characteristic is obtained from the equation (9).

[0057]

P ₂ = P ₁₀ sin ² (2δd) (12), and has a wavelength characteristic as shown in FIG. 9B according to the wavelength dependence of the coupling rate of the directional coupler.

On the other hand, in the ON state, in addition to the wavelength dependence of the directional coupler, the wavelength dependence also occurs in the optical path length difference between the waveguide arms. The output characteristic is given by equation (7).

[0059]

P ₂ = P ₁₀ [sin ² (2δd) cos ² (kΔL / 2)] (13) Wavelength dependence of the optical path length difference ΔL is within 8% in a wavelength λ = 1.2~1.4μm, cos ²
(KΔL / 2) is a value close to 0. Therefore, the output characteristics are not so affected by the wavelength dependency of the directional coupler, and the output characteristics are close to 0, and the wavelength dependency is small.

Next, as shown in FIGS. 8A and 8B, the cross-path output of the switch element of the present invention has a relationship between ON / OFF and the optical path length difference ΔL in the conventional switch element. On the contrary, the relation of the output ports is also reversed. Therefore, the output characteristics are the characteristics obtained by reversing the notation of ON and OFF in FIG. 9B and symmetrically folded at a line with a transmittance of 50%, that is, FIG. 9A.

As can be seen from these characteristics, the conventional switching element has a very strong wavelength dependence in the off state, whereas the switching element used in the present invention has almost no wavelength dependence in the off state. .

FIG. 10 shows calculation of optical crosstalk (total amount of leakage light from the other three incident ports with respect to main signal light) in a path passing through seven switch elements in a 4 × 4 matrix switch. The results are obtained from the wavelength dependence of the switch element described above. In contrast to the conventional type, which passes through six off-state switching elements having a strong wavelength dependence, in the present invention, only one on-state element has a strong wavelength dependence. ing.

For example, the wavelength range in which a crosstalk of 15 dB or more is obtained is 5
In contrast to 0 nm, in the present embodiment, it is expanded to 170 nm.

The small wavelength dependence of the matrix optical switch of the present invention is closely related to the fact that the matrix optical switch of the present invention is resistant to manufacturing errors.

Next, a more specific manufacturing procedure of this embodiment will be described.

The (4 × 4) matrix optical switch of this embodiment is a silicon substrate 11 having a thickness of 1 mm and a diameter of 6 inches.
Made above.

FIGS. 11A to 11F show an optical switch element having a silica glass single mode optical waveguide having a cross-sectional structure shown in FIGS. 3B and 3C on a silicon substrate. FIG. 3 is a diagram illustrating a process of forming a total of 16 pieces in the layout shown in FIG.

First, silicon trichloride (SiCl ₄ ) as a main component and boron trichloride (BC
l ₃ ), a glass hydrolyzing reaction using a mixed gas containing a small amount of phosphorus trichloride (PCl ₃ ) as a raw material to deposit a glass particle layer 12 a for lower cladding mainly composed of quartz (SiO ₂ ). An appropriate amount of germanium tetrachloride (GeCl
₄₎ is switched to a mixed gas obtained by adding a gas such as SiO ₂ -
A core glass particle layer 13a mainly composed of GeO ₂ is deposited (FIG. 11A). Subsequently, the silicon substrate 11 on which the two glass fine particle layers 12a and 13a are deposited is heated to a high temperature of about 1350 ° C. in an electric furnace to make the glass fine particles transparent and vitrified, thereby forming the lower clad layer 12 and the core layer 13. (FIG. 11B). Subsequently, unnecessary portions of the core layer 13 are removed by a photolithography process and reactive ion etching (RIE) to form core portions 31 and 32 (FIG. 11C). Next, SiCl is again
₄ -BCl ₃ -PCl ₃ gas mixture to deposit the upper cladding glass fine particle layer 14a as by utilizing a flame hydrolysis reaction of raw material embedding the core portion 31 and 32 (FIG. 11 (D)). Subsequently, the substrate 11 is heated again in an electric furnace, and the glass fine particle layer 14a for upper cladding is made transparent and vitrified to form the desired single mode optical waveguide as the upper cladding layer 14 (FIG. 11E )). Further, a thin-film heater 34a as an optical phase shifter is installed at a predetermined portion of each optical switch element on the optical waveguide formed on the silicon substrate in the above process (FIG. 11 (F)).

The core size of the manufactured optical waveguide is 6 μm × 6.
μm, and the relative refractive index difference Δ between the cladding layers 12 and 14 was 0.75%. The (4 × 4) matrix optical switch of the present embodiment uses this silica glass optical waveguide for 250 times.
It is based on a structure in which eight pieces are arranged side by side at a pitch of μm. The optical switch element was basically composed of a straight optical waveguide and a bent optical waveguide having a bending radius of about 4 mm.

FIG. 12 is a diagram showing a design guideline for properly setting the radius of curvature of a curved waveguide, and plots the dependence of the 90-degree bending loss on the radius of curvature of a silica-based glass single-mode optical waveguide (experimental value). It was done. In the optical waveguide of .DELTA. = 0.75% employed in this embodiment, the allowable minimum radius of curvature is about 4 mm, and the entire radius of the matrix optical switch is accommodated on a substrate having a limited area. The allowable minimum radius of curvature of the wave path was set.

The intersection angle θ at the intersection is 30 degrees. The effective optical path length difference between the two directional couplers was accurately set to 0.65 μm, which is half the signal light wavelength of 1.3 μm, by a photolithography process. Considering that the refractive index value of the silica glass is about 1.45, the actual waveguide length difference on the mask pattern is 0.65 μm / 1.45 = 0.45.
It was set to 45 μm. The thin film heater 34a was formed by evaporating a chromium thin film over each optical switch element over a thickness of 0.3 μm, a width of 50 μm, and a length of 4 mm by a vacuum evaporation method using metal chromium as an evaporation source. The total length of the optical switch element including the two directional couplers 33a and 33b, the thin film heater 34a, and the intersection 35 was about 15 mm.

The matrix optical switch manufactured on the silicon wafer through the above-described process was replaced with a 10 mm × 110 m
m by a dicing saw, a heat sink is provided below the silicon substrate, an optical fiber array is connected to the input and output optical waveguides, and a power supply lead is connected to the thin film heater 34a. Matrix optical switch is completed. By appropriately selecting a thin film heater to be supplied with power, a (4 × 4) switch operation was confirmed. The power consumption of each thin film heater 34a required for the switch operation was about 0.5 watt. Since the number of simultaneously operating thin film heaters is up to four, the total power consumption is up to about 2 watts. The loss value of the matrix optical switch is 3 to 4 dB including the connection loss of the optical fiber, and the extinction ratio of the entire matrix optical switch is such that the coupling ratio of the directional coupler is 50% ± 1 due to a manufacturing error.
Even if the deviation was as large as about 0%, it was better than about 20 dB.

Embodiment 2 FIG. 13 is an overall layout diagram of an (8 × 8) matrix optical switch as a matrix optical switch according to a second embodiment of the present invention. 11 is a silicon (Si) substrate, 21a-2
1b is a waveguide bundle composed of 8 + 8 = 16 silica glass optical waveguides. On the way of the waveguide bundles 21a-21b,
Fifteen optical switch groups (# 1, # 2,..., # 15) are arranged. These switch groups are sequentially one,
, 7, 8, 7, ..., 2, 1 optical switch elements. What is characteristic in this arrangement is that between the optical switch groups # 2 and # 3, between # 4 and # 5,
Between # 6 and # 7, between # 7 and # 8, between # 8 and # 9, # 9
Between # 10 and # 10, between # 11 and # 12, and between # 13 and # 14, the waveguide bundles 22a, 22b, 22c, 22d, 22e, 2 having a 90-degree to 180-degree bend, respectively.
2f, 22g, and 22h. In other words, it is characteristic that the fifteen optical switch groups are arranged in a zigzag pattern on a limited substrate size by appropriately using the bent waveguide bundle.

FIG. 14 shows the optical switch group (# 1, # 1) of FIG.
# 2,..., # 15).
It shows how the optical switch elements are arranged on the way of a-21b. These switch groups are sequentially 1, 2,..., 7, 8, 7,.
And one optical switch element (symbolized by an elliptical mark in FIG. 14). The left end of the optical switch group # 1 has eight input optical waveguides 1a, 1b, 1c, 1d, 1e, 1f,
It is connected to the entrance of the waveguide bundle 21a-21b including 1g and 1h. The optical switch groups # 1 and # 2 are directly connected, the optical switch groups # 2 and # 3 are connected by a bent waveguide bundle 22a, and so on. Reference numeral 15 denotes eight output optical waveguides 2a, 2b, 2
Waveguide bundle 21 including c, 2d, 2e, 2f, 2g, 2h
It is led to the outlet of a-21b.

FIG. 15 is a plan view of an optical switch element constituting the matrix optical switch of FIGS. 13 and 14.
Silica-based glass optical waveguides 31a-31b on silicon substrate
And 32a-32b are close to each other at a distance of several μm at two locations to constitute two directional couplers 33a and 33b. In this embodiment, an intersection 35 is provided between the two directional couplers 33a and 33b. The coupling ratio of the directional couplers 33a and 33b is set at a target of 50% by adjusting the waveguide interval and length of the coupling portion. Optical waveguide 31a between two directional couplers 33a, 33b
The effective optical path length difference between −31b and 32a-32b is that the optical waveguide 32a-32b side is longer than the optical waveguide 31a-31b by half the wavelength of the signal light (1.3 μm in this case), that is, about 0.65 μm. The thin-film heaters 34a and 34b as optical phase shifters are provided above the clad layer in portions corresponding to the optical waveguides 31a-31b and 32a-32b. Optical waveguide 31
The a-31b and 32a-32b intersect at an angle θ (here, 30 degrees) at which the crosstalk can be ignored at the intersection 35. The length of the optical switch element shown in FIG. 15 was about 12 mm. In the optical switch element of this embodiment, since the intersection 35 is provided between the directional couplers 33a and 33b, the two optical waveguides 31a-31b and 3
A place where the interval between 2a and 32b is largely apart can be combined into one place. For this reason, the optical switch element length is set to the first
There is an advantage that it can be made slightly shorter than in the case of the embodiment (15 mm).

(8 × 8) fabricated on a silicon wafer
The matrix optical switch is cut out into a 55 mm × 55 mm square by a dicing saw, a heat sink is provided below the silicon substrate, an optical fiber array is connected to the input / output optical waveguides, and power supply leads are connected to the thin film heaters 34 a and 34 b. Are completed, the target matrix optical switch is completed. By appropriately selecting a thin film heater to be supplied with power, an (8 × 8) switch operation was confirmed. The power consumption of each thin-film heater required for the switch operation was about 0.5 watt. Since the number of simultaneously operating thin film heaters is up to eight, the total power consumption is up to about 4 watts. The loss value of the matrix optical switch is 5 to 7 dB including the connection loss of the optical fiber, and the extinction ratio of the entire matrix optical switch is about 50% ± 10% due to a manufacturing error. Even if there was a large deviation, it was as good as about 15 to 20 dB or more.

The optical switch element (FIG. 15) in this embodiment is provided with two thin film heaters 34a and 34b. Normally, only the thin film heater 34a provided on the side having the short optical path length is sufficient. However, if a pair of thin film heaters is provided, even if the optical path length difference is shifted due to a manufacturing accident, any one of the thin film heaters may be used. On the other hand, there is an advantage that compensation can be made by applying a light current.

In the arrangement of the matrix optical switch according to the second embodiment, when the signal light input from the input optical waveguide 1h is extracted from the output optical waveguide 2h, the signal light passes through the optical switch element only once. On the other hand, when the signal light input from the input optical waveguide 1a is taken out from the output optical waveguide 2a, the signal light passes through the optical switch element 15 times, and when each optical switch element generates a constant passage loss, In some cases, the output signal light level fluctuates depending on the path, causing inconvenience. In such a case, the improvement measures of the following embodiment can be taken.

Embodiment 3 FIG. 16 is a configuration diagram of an optical switch group employed in a third embodiment of the present invention. The third embodiment has the same basic structure as the second embodiment described with reference to FIGS. 13 and 14, except that a dummy optical switch element _SD is arranged in addition to the optical switch element S in the optical switch group. The points are different. As the dummy optical switch element _SD , one in which the thin film heaters 34a and 34b are omitted from the optical switch element in FIG. 15 and in which the coupling ratio of the two directional couplers is the same (including 0%) is used. . When such a dummy optical switch element _SD is added, the fluctuation of the optical output level caused by the path difference of the matrix optical switch is reduced, and the loss value of the entire matrix optical switch is 6.5 to 7 dB. The ratio indicates that the coupling ratio of the directional coupler is 50% ± due to a manufacturing error.
Even with a shift of about 10%, 20 dB or more was achieved.

In the above embodiment, the intersection angle θ between the two waveguides is set to 30 degrees. Next, the setting of the intersection angle θ will be considered from the viewpoint of optical crosstalk and cross loss.

FIG. 17 is a view for explaining the amount of signal light leaking into the optical waveguide on the other side where one signal crosses, that is, the so-called cross angle dependence of optical crosstalk.
The optical waveguide used here was the same as the core in the previous embodiment.
Clad relative refractive index difference Δ is 0.75%, core size is 6 μm
× 6 μm. Optical crosstalk decreases as the intersection angle increases. In the present invention, the two optical waveguides in the optical switch element need to intersect at an angle at which the optical crosstalk can be ignored, and the appropriate angle is such that the optical crosstalk due to the intersection between the waveguides is -30 dB or less. It may be regarded as an angle. Although it depends on the relative refractive index difference between the core and the clad of the optical waveguide used, if the crossing angle is about 15 degrees or more, the optical crosstalk between the two optical waveguides is helped by the straightness of light, Practically negligible. At the intersection angle of 30 degrees employed in the above embodiment, the optical crosstalk is -60d
It is as small as B.

FIG. 18 is a diagram for explaining the dependence of the crossing loss on the crossing angle. Again, the optical waveguide used had a core / cladding relative refractive index difference Δ of 0.75% and a core size of 6 μm ×
6 μm. The crossing loss decreases as the crossing angle increases, and the crossing loss per crossing is 0.2 if the crossing angle is about 15 degrees or more where optical crosstalk can be ignored.
dB or less can be realized. Intersection angle 3 adopted in the above embodiment
In the case of 0 degree, it is as small as 0.06 dB, and (8 × 8)
In the entire matrix optical switch, the amount of increase in loss is as small as 1 dB or less.

As can be seen from the above two results, the loss at the intersection and the optical crosstalk generally improve as the intersection angle approaches 90 degrees. Therefore, it is characteristically desirable that the crossing angle be 15 degrees or more and be as close to 90 degrees as possible. However, if the crossing angle is increased, the area occupied by the curved portion increases and the size of the optical switch element tends to increase. Therefore, the matrix switch is determined in consideration of the required performance, the allowable bending radius of the optical waveguide, and the like. .

Next, the shape of the optical waveguide at the intersection will be considered.

FIGS. 19A to 19E are views for explaining the shape of the intersection. In the above embodiment, the width of the optical waveguide is uniform throughout the circuit (FIG. 19A), but it is also effective to change the shape of the intersection in order to reduce the intersection loss at the intersection.

One of them is the normal width optical waveguides 41a, 41b, 42a, and 42b constituting portions other than the intersection before and after the intersection as shown in FIGS. 19B and 19C.
Are connected to the tapered optical waveguides 43a, 43b, 44a, 44b for converting the waveguide width having a tapered shape to convert the optical waveguide width into optical waveguides 45, 46 having widths such that the cross loss is reduced. It is a method of crossing them. In a commonly used optical waveguide having a core / cladding relative refractive index difference Δ = 0.75%, an optical waveguide width and a thickness of 6 μm, the crossing loss is 0.06 dB when the crossing angle of the optical waveguide is 30 degrees. On the other hand, when the width of the optical waveguide at the intersection is 4 μm and 10 μm, the intersection loss is 0.0
It is improved to 5 dB, and it can be seen that the crossing loss can be reduced by optimizing the optical waveguide width. However, the tapered optical waveguides 4 for waveguide width conversion disposed before and after the intersection.
Needless to say, it is necessary to make the tapered shapes of 3a, 43b, 44a, and 44b gentle enough to prevent radiation loss and the like.

Another method is as shown in FIGS. 19D and 19E, where the width of the optical waveguide is reduced or increased up to the intersection, and the tapered optical waveguide 47a-
47b, 48a-48b cross each other.
Similarly to the above example, the core / clad relative refractive index difference Δ = 0.
In the case of 75%, a crossing angle of 30 °, an optical waveguide width and a thickness of 6 μm, the crossing loss is 0.06 dB, for example, the crossing shown in FIG.
m, the crossing loss is 0.03 dB at the intersection between the narrowed optical waveguide widths from 6 μm to 5 μm.
This indicates that this method is effective in reducing the crossing loss.

In the above embodiments, all the optical waveguides are smoothly connected. However, considering the intrinsic electric field distribution of the optical waveguide, the shape and the center position of the electric field distribution of the linear optical waveguide, the curved optical waveguide, and those having different radii of curvature between the curved optical waveguides are different from each other. Therefore, when optical waveguides of different shapes are connected together with their waveguide centers aligned, radiation loss occurs due to the mismatch of the intrinsic electric field distribution of the optical waveguide, and the characteristics of the directional coupler and the Mach-Zehnder optical interferometer are unstable. The light meandering phenomenon occurs. Therefore, at the connection portion such as a straight optical waveguide, a curved optical waveguide, and a curved optical waveguide having different radii of curvature or different bending directions, there is no state where the center axes of the optical waveguides are broken, which is called "offset". Stitched together,
It is also noted that a method of matching the electric field distribution of light unique to two joined optical waveguides as much as possible at the joint portion is effective for reducing loss and suppressing meandering.

FIG. 20 shows the configuration of a directional coupler having an offset structure as an example. As shown here, the connecting portions 53a, 53b, 53c,
53d, 54a, 54b, 54c, 54d and portions 53e, 53f, 54e, in which the bending directions are different between the curves.
At 54f, an offset structure is applied. The amount of the offset varies depending on the relative refractive index difference between the core and the clad, the shape of the optical waveguide, and the wavelength of the signal light, but is generally about submicron.

In the above embodiment, an optical phase shifter (thin film heater) is provided in a waveguide having a wavelength shorter than a half wavelength, and the phase is controlled so as to cancel out a difference in a half optical path length. However, in some cases, conversely, it is also possible to turn on the state by increasing the optical path length difference by one wavelength. However, in this case, there is a disadvantage that the wavelength dependence of the ON state becomes stronger.

In the above embodiment, the signal light wavelength is 1.3.
Although the case of μm was dealt with, the configuration of the present invention can of course be applied to other wavelengths, for example, 1.55 μm wavelength.

As described above, the matrix optical switch of the present invention is strong against the manufacturing error of the directional coupler. As a secondary effect thereof, it is also strong against the polarization dependence of the directional coupler. I also want to point out that.

In the above embodiment, the configuration, operation, and the like of the matrix optical switch based on the silica-based glass optical waveguide on the silicon substrate have been described. However, a Mach-Zehnder optical interferometer circuit type optical switch element can be configured. It should be noted that the present invention can be applied to other materials, for example, a plastic optical waveguide, an ion diffusion type glass optical waveguide, and a lithium niobate optical waveguide in which an optical phase shifter using an electro-optic effect is installed.

[0094]

As described above, the waveguide type matrix optical switch of the present invention employs a Mach-Zehnder optical interferometer circuit structure having an optical path difference corresponding to a half wavelength in an off state as an optical switch element. In addition, it includes the intersection of the optical waveguides in the optical switch element. Therefore, if the coupling ratio of the two directional couplers is equal, even if the value deviates from 50%, no light crosstalk to the bar path occurs in the off state. If it deviates from 50%, all the signal light cannot be switched to the bar path in the ON state, and the signal light slightly remains in the cross path. However, since this residual light is guided to an unnecessary waveguide end in the output side waveguide bundle, it does not lead to deterioration of the extinction ratio as a matrix optical switch. In addition, by providing an intersection on the way of connecting the two directional couplers, it is possible to combine the places where the distance between the two optical waveguides needs to be large into one place, so that the optical switch element is occupied. The length can be minimized. Therefore, the size of a matrix optical switch in which a large number of optical switch elements are integrated on a substrate can be reduced. Further, in the waveguide type matrix optical switch of the present invention, as long as the coupling ratio of the two directional couplers in the switch element is the same, even if the coupling rate deviates from 50%, the signal light in the off state causes the switch element to pass through the switch element. It will pass through a 100% cross route. Therefore, the manufacturing accuracy of the directional coupler is greatly reduced, and a matrix optical switch having an excellent extinction ratio can be easily provided. Further, in the matrix optical switch of the present invention comprising the optical switch element, there is an advantage that the setting accuracy of the signal light wavelength is eased and an inexpensive signal light source with low wavelength accuracy can be introduced. Such a matrix optical switch is expected to make a great contribution to the construction of an optical fiber communication network in which a plurality of signal lights are exchanged.

[Brief description of the drawings]

FIG. 1 is a conceptual diagram showing a configuration of a conventional waveguide type matrix optical switch.

2A is an overall plan view showing a configuration example of a (4 × 4) waveguide matrix optical switch, and FIG. 2B is an enlarged plan view of the optical switch element shown in FIG.

FIG. 3A is a plan view showing the configuration of an optical switch element constituting a (4 × 4) matrix optical switch as a first embodiment of a waveguide type matrix optical switch according to the present invention, and FIG. A) is an enlarged cross-sectional view along AA ′ of FIG.
(C) is an enlarged sectional view along BB 'of (A).

FIG. 4 is a conceptual diagram showing a state of an electric field in each portion of a conventional Mach-Zehnder optical interferometer circuit type optical switch element.

FIG. 5 is a conceptual diagram showing a state of an electric field at each position of a Mach-Zehnder optical interferometer circuit type optical switch element according to the present invention.

FIG. 6 is a conceptual diagram showing each parameter of an optical switch element of the Mach-Zehnder optical interferometer circuit type according to the present invention.

FIG. 7 shows a first embodiment of the present invention and a conventional (4 × 4)
FIG. 4 is an explanatory diagram of an excess loss characteristic caused by light crosstalk with respect to a coupling ratio of a directional coupler of a matrix optical switch.

FIGS. 8A and 8B are conceptual diagrams showing ON / OFF states of a Mach-Zehnder optical interferometer circuit type optical switch element according to the present invention and a conventional Mach-Zehnder optical interferometer circuit type optical switch element. is there.

FIG. 9A is an explanatory diagram of a wavelength characteristic of an optical switch element corresponding to the first embodiment of the present invention, and FIG. 9B is an explanatory diagram of a wavelength characteristic of a conventional optical switch element.

FIG. 10 shows a first embodiment of the present invention and a conventional (4 ×
4) It is an explanatory diagram of a wavelength characteristic of optical crosstalk of the matrix optical switch.

FIGS. 11A to 11F are explanatory views of a process for producing a matrix optical switch constituted by a silica glass waveguide used in the present invention.

FIG. 12 is an explanatory diagram of a bending loss characteristic of the silica-based single mode optical waveguide used in the present invention.

FIG. 13 is an overall layout view of an (8 × 8) matrix optical switch as a second embodiment of the waveguide type matrix optical switch of the present invention.

14 is a diagram showing a configuration of each optical switch group in the matrix optical switch shown in FIG.

FIG. 15 is a plan view showing a configuration of an optical switch element included in the matrix optical switch shown in FIG.

FIG. 16 is a diagram illustrating a configuration of an optical switch group according to a third embodiment of the present invention.

FIG. 17 is an explanatory diagram of the crossing angle dependence of optical crosstalk at one crossing portion of an optical switch element of the Mach-Zehnder optical interferometer circuit type according to the present invention.

FIG. 18 is an explanatory diagram of the crossing angle dependence of crossing loss at one crossing portion of an optical switch element of the Mach-Zehnder optical interferometer circuit type according to the present invention.

FIGS. 19A to 19E are explanatory diagrams of examples of the shape of an intersection portion for the purpose of reducing the intersection loss.

FIG. 20 is a diagram illustrating a configuration of a directional coupler for explaining an offset of an optical switch element of a Mach-Zehnder optical interferometer circuit type according to the present invention.

[Explanation of symbols]

1a, 1b, 1c, 1d, 1e, 1f, 1g, 1h Input optical waveguides 2a, 2b, 2c, 2d, 2e, 2f, 2g, 2h Output optical waveguides 3 Silicon substrate 4a-4b Optical waveguide bundle 11 Silicon substrate 21a- 21b Optical waveguide bundles 22a, 22b, 22c, 22d, 22e, 22f, 2
2g, 22h 90 to 180 degree bent waveguide bundle 31, 31a-31b, 61a-61b Optical waveguide 32, 32a-32b, 62a-62b Optical waveguide 33a, 33b, 63a, 63b Directional coupler 34a, 34b, 64a , 64b Thin film heater (optical phase shifter) 35 Intersection 41, 42, 41a, 41b, 42a, 42b Waveguides 43a, 43b, 44a, 44b of normal width Tapered waveguides 45, 46 before and after the intersection Special-width waveguides 47a, 47b, 48a, 48b Tapered waveguides 51a-51b, 52a-52b forming intersections Optical waveguides 53a, 53b Directional couplers 54a, 54b Thin-film heaters (optical phase shifters) # 1, # 2,..., # 15 Optical switch group S Optical switch element S _D Dummy optical switch element S00 , ..., S33 Optical switch element

Claims (10)

(57) [Claims] 1. A waveguide type matrix optical switch having a plurality of input terminals and a plurality of output terminals for switching and outputting input signal light, wherein a plurality of input optical waveguides respectively connected to the input terminals are provided. A plurality of output optical waveguides respectively connected to the output end, and a plurality of optical switch elements arranged at respective intersections of the input optical waveguide and the output waveguide. A coupling configured such that two optical waveguides are brought close to each other at two locations on a substrate, and the two optical waveguides between the two locations have an effective optical path length difference of a half wavelength of the signal light. Two directional couplers having the same ratio; and at least one of the two optical waveguides between the two directional couplers, wherein the effective optical path length difference is an integral multiple of the wavelength of the signal light. Path length difference switching means for switching to And a crossing portion where the two optical waveguides cross each other. 2. The waveguide type matrix optical switch according to claim 1, wherein said optical path length difference switching means is an optical phase shifter. 3. The waveguide type matrix optical switch according to claim 2, wherein said optical phase shifter is a thermo-optic effect phase shifter comprising a thin film heater. 4. The waveguide matrix according to claim 1, wherein said optical path length difference switching means is provided in each of said two waveguides between said two directional couplers. Light switch. 5. The waveguide type matrix optical switch according to claim 1, wherein said crossing portion is disposed between said two directional couplers. 6. The waveguide matrix optical switch according to claim 1, wherein said two optical waveguides intersect at an intersection angle of 15 ° or more at said intersection. 7. The waveguide-type matrix optical switch according to claim 1, wherein said optical switch elements are connected by a waveguide bundle which is bent in a zigzag manner. 8. The waveguide type matrix optical switch according to claim 1, wherein a dummy switch is provided at a position corresponding to the optical switch element on a waveguide connecting the optical switch element. 9. The dummy switch, wherein two optical waveguides are brought close to each other at two locations on a substrate, and the two optical waveguides between the two locations are arranged such that the two optical waveguides have an effective optical path length corresponding to a half wavelength of the signal light. 9. The waveguide type according to claim 8, further comprising: two directional couplers configured to have a difference in coupling ratio and having an intersection portion where the two optical waveguides intersect. Matrix light switch. 10. A connection in which the waveguide center is offset at a portion where the waveguide transitions from a straight line to a curve, at a portion where the radius of curvature changes at the curve portion, or at a portion where the direction of the curve portion changes. The waveguide type matrix optical switch according to claim 7, wherein