JPS60121404A - Optical demultiplexer - Google Patents

Abstract

PURPOSE:To improve the demultiplexing efficiency with a simple constitution and to reduce influences upon the other wavelength components by prescribing widths of single mode optical wavegudes of both channels coupled to each other in the position where they are parallel with and close to each other and providing a grating only one optical waveguide. CONSTITUTION:When single mode optical waveguides 1 and 2 of the first and the second channels which have core widths a1 and a2 different from each other are coupoed in a coupling part 3 where they are parallel with and close to each other, the light having a prescribed wavelength which is made incident on the waveguide 1 is branched through the coupling part 3 and goes backward on the waveguide 2 and exits. The efficiency of taking-out of the branched light is maximized if widths a1 and a2 are set to satisfy 1.2a2<=a1<=3a2, and harmful coupling is suppressed to eliminate crosstalk if a grating 6 having a prescribed frequency in the light propagation direction is provided only in one waveguide 2, and thus, the demultiplexing efficiency is improved with a simple constitution, and the light is demultiplexed with less influences upon the other wavelength components.

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to an optical demultiplexer that is used, for example, in optical communications, and selects light of a specific wavelength from one optical waveguide and guides it to another optical waveguide.

<Prior art> Optical demultiplexers are essential components for wavelength division multiplexing communications. In wavelength division multiplexing communications, the density of wavelength multiplexing is determined by the interval of optical wavelengths that can be separated by an optical demultiplexer, so it is desired to realize an optical demultiplexer with high resolution and low insertion loss.

Currently, optical demultiplexers based on the wavelength dispersion properties of prisms, diffraction gratings, and dielectric multilayer films are being developed. These are low-resolution optical demultiplexers developed for multimode optical fiber transmission systems.

Since there is only one propagation constant β in a single mode optical fiber transmission system, a narrowband optical demultiplexer that takes advantage of the strong β selectivity of the reflection characteristics of an optical waveguide provided with a grating is promising. As such an optical demultiplexer, the inventor previously proposed an optical demultiplexer that extracts light with a selected wavelength λ0 from an output optical waveguide that is different from the input optical waveguide (Patent Application No. 55).
-148200). However, in this optical demultiplexer, the optical properties are changed periodically, that is, the grating is provided by periodically changing the position of the twisted core itself while keeping the width of the core constant. This corresponds to attaching gratings with opposite phases on both sides of the core, and in this case, the effects of both gratings weaken each other, resulting in a low extraction efficiency of demultiplexed light.

That is, with this structure, sufficient demultiplexing characteristics cannot be obtained. Furthermore, formation of such a grating is difficult to manufacture. Furthermore, the formula for determining the period A of the grating is
Using the propagation constant β1 of the input optical waveguide and the propagation constant β2 of the output optical waveguide, A=2π/(β1+β2) (1). This expression uses the propagation constant when the input optical waveguide and the output optical waveguide are independent. In reality, both are in a state where they interfere with each other, which is called a coupled state, and when they are independent, β1. Since it cannot be expressed as β2, equation (1) is only an approximation that gives an inaccurate period Δ, and has the disadvantage that the center wavelength deviates from the desired wavelength.

Furthermore, in the past, attention has been paid only to the optical demultiplexing of the wavelength λ0, and it has not been discussed that in a configuration that obtains this demultiplexing characteristic, other wavelength components may have an adverse effect.

<Summary of the Invention> An object of the present invention is to provide an optical demultiplexer with a simple structure, high demultiplexing efficiency, and less influence on other wavelength components.

According to this invention, the first and second channel type single mode
Parts of the optical waveguides are brought close to each other in parallel and coupled. The widths of the cores of the second optical waveguide are different, and in particular, if the core widths are a1 and a2, the distance between them is 1.2a1≦
a2≦3a1. Alternatively, the relationship of 1.2a2≦a1≦3a2 is selected, and furthermore, a grating provided at the coupling portion is provided only for the output optical waveguide.

Furthermore, it is preferable that the crating has irregularities with a period A in the light propagation direction on the upper or lower surface of the core. Further, the period A is preferably selected to be 2π/(βθ+β0). The coefficients βe and β0 are propagation constants at the demultiplexed wavelength λ0 of even mode and odd mode, which are propagation modes generated in the coupling portion, respectively.

<Embodiment> FIG. 1 is a plan view of an embodiment of the present invention, in which the main optical waveguide 1 and the output optical waveguide 2 are both channel type single mode optical waveguides, and some of them are parallel to each other and close to each other. The connecting portion 3 is defined as the connecting portion 3. Further, the main optical waveguide 1 and the output optical waveguide 2 each have a core width of al.

a2, the thickness in the depth direction -6'=b, and the distance between the cores is d. If we let the electric field distributions of the main optical waveguide l and the output optical waveguide 2 be El and E2, respectively, and let the forward waves and retrograde waves be expressed by suffixes of 10 and -, respectively, then the incident light 4
is represented by E; On the other hand, in the coupling part 3, if the two cores of the optical waveguides 1 and 2 are considered as one waveguide, there exists a field effect cloth of an even mode Ee and an odd mode Eo corresponding to the fundamental mode and the -order mode. Therefore, the electric field distribution E+ of the incident light 4 is divided into a traveling wave mode E+ and an odd mode E+ at the Z=00 point (the point of incidence of the coupling portion 30). now,
e O If the optical waveguide 2 is made thicker than 1, the electric field distribution of Eo will be equal to the electric field distribution of 'fEx, so most of it will be E+
, and a portion of it is transmitted slightly as E;. These electric field distributions show a normalized state. E+ and Eki are the phases exp[i(2→−βez) characterized by respective propagation constants βe and 2πC β0
]2πC and e These are the coordinates.The time dependence is common, so if omitted, it becomes exp(-iβez) and exp(-iβoz), respectively.However, here the period A
Since there is a grating, coupling occurs between the forward wave and the retrograde wave through the wave number part. Now, let A be A-4π/(
βe + βo ) (21 If given as 2π, then βe = (-βo ) +J-1 and βo = (-β'') 1; Therefore, βe, which is an even mod traveling wave, is -β0, which is an odd mode. βO, or odd mode traveling wave, is converted into -βe, or even mode retrograde wave. Each retrograde wave returns to the point z = Q, where they are superimposed in the same phase, and then It is divided into the E'' component and the E- component. El is the demultiplexed light 5 taken out from the output optical waveguide 2, and its intensity should be 1 based on the intensity of the first incident light 4. ET is the reflected light returning to the incident optical waveguide 1, and 0
It is desired that To summarize the above, the transition from the incident light to the output light at the wavelength λ0 of the demultiplexed light is shown in FIG.

Therefore, by following the paths indicated by the arrows in FIG. 2, the efficiency of conversion from incident light to demultiplexed light and the efficiency of conversion to reflected light can be determined as follows.

η=(elo e06 eez+ et6 eeQ e
oz) (31R== (elo eoe eel +
ele eeo eot) (4 again, η: intensity conversion efficiency from incident light 4 to demultiplexed light 5, R: intensity conversion efficiency from incident light 4 to reflected light, elo: amplitude conversion efficiency from El to E; ,”oe*E // E.

〃 eoz: Amplitude conversion efficiency from E- to E;, - "te, @E1//E.

eeo2E+〃E-〃 e　O e　62　:　E//　E　; /1 eel: E-tt E.

〃 eol ”, E // E-, /1.Here, eio + eoz + ele +
Co2 + eel.

'1401 has nothing to do with the grating and can be expressed only by the superposition integral of the electric field distribution as follows.

Regarding eeo and eoe, the coupling theory (A, Y
ar iV 1"' Coupled -mode t
theory for guided wave opt
ics+” IEEE J, Quantum Elec
tron, vol, QE-9.

pp. 919-933.1973) as a function of the depth, period, position, etc. of the grating. Using eol, etc., (31, (41 formula is η = (eio e2e + e1e e2o)・e
o e (61R=(2e1oe1e)・eeQ2(7
) can be written. (61, (In Equation 71, the first term is a term that depends only on the structure of the optical waveguide 1.2, and the second term is a term that depends on the grating. Both terms are 0 and 1.
Since the value is between, in order to set η = 1, (6)
The first and second terms on the right side of the equation must each be set to 1. As mentioned above, if the optical waveguide 2 is made thicker than the optical waveguide 1, the electric field distributions of El and Eo and the electric field distributions of E2 and Ee can be made equal, so elo ``= e2e '';
1 and ete″!re2o″i:Q. Therefore, the first term can be set to so1. At this time, the first term of equation (7) automatically becomes yo.

In other words, in this invention, the widths a1 and a of the optical waveguides 1 and 2 are
2 to be different from each other.

The above is a phenomenon at the wavelength λ0, but there is also harmful coupling due to the grating, and one of the objects of the present invention is to eliminate this. This will be explained in detail below. Now, for the period A given by equation (2), A−π/β
A wavelength λ1 where e(λ1) and a wavelength λ2 where A=π/βO(λ2) exist near the demultiplexing wavelength λ0. For example, in the example described later, the difference between λ1 and λ2 and λ0 is about 1 nm. Such a wavelength λ1. In λ2, the bond between te and -βe, (in the case of (al)a2) or the bond between β0 and -βO (in the case of (al)a2)
a 1 < a 2), which causes reflected light returning to the main optical waveguide 1 or causes crosstalk in which light that should originally be transmitted leaks to the output optical waveguide 2. When demultiplexing the wavelength λ0, the wavelength λ1. If λ2 is reflected without being transmitted, it may leak into the output optical waveguide.

Figure 3 shows the results of changes in incident light, transmittance, reflectance, and crosstalk based on the same concept as in the case of wavelength λ0, including the case of λ0. The following can be understood from each hole. If elO→1 and ele−+O are set in order to set η=1, the transmittance T1, reflectance R1, crosstalk C1, and transmittance T automatically approach the target values. On the other hand, the transmittance T
If eoo2 is set to O in order to set 2 to 1, the reflectance R2,
Crosstalk C2 automatically becomes 0. Therefore η and T
By designing so that 2 becomes 1, all the characteristics match the target values, the influence of harmful coupling can be removed, and the extraction efficiency 7 can be optimized. According to coupling theory, e
oou is a monotonically increasing function of the electric field distribution of the odd mode I''JO at the location where the crating is present.

Therefore e. In order to reduce 0, a grating may be provided at a location where the odd mode Eo is small. Therefore, in this invention, the grating 6 is provided only in the output optical waveguide 2 in which the odd mode EO is small. At this time, e. It would be meaningless if e also became smaller at the same time. Conventional one (patent application 1986-14)
8200), the position of the core itself was periodically varied without changing the core width of the optical waveguide. This corresponds to attaching gratings with opposite phases on opposite sides of the optical waveguide, so both effects work in a direction that cancels each other out, resulting in a significantly small value of eve. In the present invention, it is preferable to provide the crating on either the upper or lower surface of the core rather than the side surfaces.

In addition, this kind of grazing creates a crating in places where the electric field is strong, making the effect even greater. As a crating, it is limited to a change in shape and may also be a change in refractive index.

A numerical design example in which a grating with a depth of 1 μm is provided only in the optical waveguide 2 will be described below. Optical waveguide 1
, 2 is a buried quartz single-mode waveguide. This has a rectangular core with dimensions of about 3 to 10 μm, and the refractive index of the cladding is 1.4-5 that of quartz, and the refractive index of the core is 01 to 1.0 times higher than that of the cladding. First, in order to maximize the first term of equation (6), FIG. 4 shows a graph obtained as a function of the core width a1 of the main optical waveguide 1. Here, the core width a2 of the output optical waveguide 2 was 5 μm, the relative refractive index difference between the core and the cladding was 0.5%, and the wavelength was 1.55 μm. The case where the core spacing d is 3 μm to 7 μm as a parameter is shown. As will be described later, it is preferable that d=6 μm or more, and considering this point, from FIG. 4, 1.2a1≦a2≦3ax (a1=
It can be seen that there is a maximum value in the case of 1.6 to 4 μm). The results show that the larger d is, the better, but this is because the effect of grating 6 is not considered at this time, and if the length of the grating is finite, an optimal value for d will come out. . - To see this, Figure 5 shows η including the effect of the grating as a function of d. In Fig. 5, if the length of grating 6 is 30 circles or more, d''?
It can be seen that η≧90q6 when the thickness is 6 μm.

As described above with reference to FIG. 3, it is sufficient to set η and T2 to 1, so next it is necessary to carefully examine the transmittance T2 of the optical demultiplexer manufactured in this way. If we take d on the horizontal axis as in the younger brother 5 diagram,? When T2 is calculated using L as a parameter, the result is as shown in FIG.

It can be seen from FIG. 6 that when L=30 mm and d=6 tt m, T2=90%. Therefore, both η and T2 can be set to 90q6 or more.

FIG. 7 summarizes the structural parameters and structural characteristics used in this example. In particular, the demultiplexed light extraction efficiency η and reflectance Ro1 at wavelength λ0, transmittance Tl, T at wavelengths λ1 and λ2
Target values are also written for I2, reflectance Rt, R2, crosstalk CI, C2, and transmittance T for other wavelengths. From this, it can be seen that in this example, the values of each characteristic are close to the target values. Furthermore, in order to show how good characteristics can be obtained with this invention, two examples of the structure and characteristics of an optical demultiplexer that differs from the configuration of this invention are shown in FIG. The first example ignores the asymmetry (a]\a2) of the coupling portion, which is the first condition of this invention, and the width a1. This is the case where a2 is equally set to 4 μm. As shown in FIG. 7, as a result, the demultiplexed light extraction efficiency η is 0, the reflectance Ro is 099, and it can be seen that at λ0, all the light returns to the main optical waveguide 1. Other transmittance, reflectance, and crosstalk are also far from the target values.

The second example is a case where gratings are provided on both optical waveguides 1 and 2, ignoring the second condition of the present invention, which is the requirement to provide a grating only on the output optical waveguide. As will be described later, gratings are generally produced by interference exposure, but in this case it is easier to produce gratings on both optical waveguides at once. As shown in FIG. 7, the branched light extraction efficiency η is 055. wavelength λ
2, the transmittance and crosstort are T2~0, respectively.
C2-1, which is the opposite of the target value. This is the wavelength λ
This is due to the strong harmful binding in 2.

As can be seen from the above IJ on the 21st, only when the two conditions of the present invention, that is, the grating is provided only in the al\a2 output optical waveguide, can the demultiplexed light extraction efficiency, transmittance, reflectance, cross ) It can be seen that - can be brought closer to the ideal value.

Considering the above, in the above embodiment, η is optimized by setting the relationship between the two core widths as 1.2al≦a2≦3at, and by providing the grating only in the optical waveguide 2, while maintaining the optimum value of η. T2 was also optimized by removing deleterious binding. In addition, the propagation constants of even mode and odd mode were used as a table for the period A of the grating (2)
When formula (2) is adopted and the period A determined by formula (2) is the period of the grating, a correct state is represented, and a desired central wavelength λ0 can be obtained.

Note that the relationship between the two core widths is swapped, ], 2a2≦a
It is also possible to optimize η and T2 in a structure in which C2 is thicker than al, such as 1≦3a2. Even when the optical demultiplexer is constructed of a semiconductor material different from quartz, the relationship between al and C2 can be maintained by appropriately selecting d.
It was confirmed through experiments that the same results as the above two relationships were obtained.

The optical demultiplexer of the present invention can be manufactured by a known method (for example, see Imoto et al., Japanese Patent Application No. 109795/1982). That is, for example, T1 of an optical waveguide pattern is formed on a quartz substrate.
Masks can be created using vapor deposition and photolithography techniques,
A trench with a depth of about 8 μm is dug by reactive sputter etching. Next, a quartz material serving as a core is deposited by bias sputtering, and a grating portion is attached by interference exposure and reactive sputter etching.

Finally, a quartz material that will become the cladding is sputtered to form a buried mold.

Effect> As explained above, the optical demultiplexer according to the invention maximizes the demultiplexed light extraction efficiency by making the widths of the two cores of the coupling part asymmetrical (a1\a2), and connects the grating to the output optical waveguide. By providing the optical demultiplexer only in the center, harmful coupling can be suppressed and reflections and crosstalk can be eliminated.Furthermore, it can be fabricated using conventional methods, making it easy to fabricate a compact and high-performance waveguide type optical demultiplexer. It has advantages such as being able to

If the period of the grating is determined using equation (2), which is an accurate value of the propagation constant in the coupling part, ideal demultiplexing characteristics can be obtained. Further, when the grating is provided on the upper surface or lower surface of the output optical waveguide, the effect of the grating is strong, and the eoe can be increased, that is, the efficiency η can be increased.

[Brief explanation of the drawing]

FIG. 1 is a plan view showing an embodiment of the present invention, FIG. 2 is a diagram showing the state of transition from incident light to output light, FIG. 3 is a diagram showing the transition of incident light and each characteristic, and FIG. The figure shows the dependence of the demultiplexed light extraction efficiency η on the optical waveguide width a1 to optimize the optical waveguide structure of the coupling part in this example. Figure 6 shows the dependence on length and core spacing d.
FIG. 7, which is a diagram showing the dependence, is a diagram showing the structural parameters and characteristics of this example and a comparative example. ■: Main optical waveguide, 2: Output optical waveguide, 3: Coupling section, 4:
Incident light, 5: demultiplexed light, 6: grating. Patent applicant: Takashi Kusano, agent of Nippon Telegraph and Telephone Public Corporation

Claims (1)

[Claims] (1) A first channel type single-mode optical waveguide and a second channel type single-mode optical waveguide are placed close to each other in parallel in a section to form a coupling part, and a grating with a period A in the light propagation direction is installed in the coupling part. By providing a specific wavelength λ0 in the light incident on the first channel type single mode optical waveguide.
In an optical demultiplexer configured to convert the light into a retrograde wave of the second channel type single-mode optical waveguide and extract it,
Above 1. When the core widths of the second channel type single-mode optical waveguide are a1 and a2, the relationship between these core widths a1 and a2 is 1.2a1≦a2≦3ax or 1.2a2≦a1≦3
a2, and the grating is provided only in the second channel type single-mode optical waveguide.