JP2009145695A - Optical modulator - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an optical modulator of high productivity and high mechanical strength. <P>SOLUTION: The optical modulator provided with a first input feed-through section for transmitting a high frequency signal to an interaction part for modulating light, has a separate substrate having a second input feed-through section. The resonance frequency of dielectric resonance determined from the thickness of a substrate and that of the separate substrate is higher than the frequency of the high frequency signal. The separate substrate transmits a high frequency signal from an external circuit to the first input feed-through section, and has an impedance converter for reducing impedance mismatching between the impedance of the interaction part and the characteristic impedance of the second input feed-through section, the characteristic impedance of a connector electrically connected to the second input feed-through section, or the characteristic impedance of the external circuit. The high frequency signal applied to the second input feed-through section from the external circuit is transmitted to the interaction part while causing residual reflection with less electric reflection compared with the case of having no impedance converter on the separate substrate. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention belongs to the field of optical modulators having a low driving voltage and capable of high-speed modulation.

An optical waveguide and a traveling wave electrode are formed on a substrate having a so-called electro-optic effect (hereinafter, the lithium niobate substrate is abbreviated as an LN substrate) such as lithium niobate (LiNbO ₃ ) whose refractive index changes by applying an electric field. The formed traveling-wave electrode type lithium niobate optical modulator (hereinafter abbreviated as LN optical modulator) is applied to a 2.5 Gbit / s, 10 Gbit / s large capacity optical transmission system because of its excellent transmission characteristics. . Recently, application to an ultra large capacity optical transmission system of 40 Gbit / s is also being studied, and it is expected as a key device.

[First prior art]
This LN optical modulator includes a type using a z-cut LN substrate and a type using an x-cut LN substrate (or a y-cut LN substrate). Here, as a first conventional technique, a z-cut LN substrate LN optical modulation using a coplanar waveguide (CPW) traveling wave electrode having a z-cut LN substrate and two ground conductors, which is advantageous for propagation in the fundamental mode. FIG. 17 shows a perspective view of the container. 18 is a cross-sectional view taken along line AA ′ of FIG. Note that the following discussion also holds true for x-cut LN substrates and y-cut LN substrates.

In the figure, 1 is a z-cut LN substrate, 2 is a transparent SiO ₂ buffer layer having a thickness of about 200 nm to 1 μm in a wavelength region used in optical communication such as 1.3 μm or 1.55 μm, and 3 is a z-cut LN. This is an optical waveguide formed by thermally diffusing Ti at 1050 ° C. for about 10 hours after depositing Ti on the substrate 1, and constitutes a Mach-Zehnder interference system (or Mach-Zehnder optical waveguide). Reference numerals 3a and 3b denote optical waveguides (or interactive optical waveguides) in a portion where an electrical signal and light interact (referred to as an interaction portion), that is, two arms of a Mach-Zehnder optical waveguide. In the case of a phase modulator, a straight optical waveguide may be used. The CPW traveling wave electrode 4 includes a central conductor 4a and ground conductors 4b and 4c. In FIG. 18, the width of the central conductor 4a is about 6 to 20 μm, and generally around 10 μm is used. On the other hand, a gap (or CPW gap) is formed between the center conductor 4a and the ground conductors 4b and 4c.

In the first prior art, a bias voltage (usually a DC bias voltage) and a high-frequency electric signal (also referred to as an RF electric signal) are superimposed and applied between the center conductor 4a and the ground conductors 4b and 4c. Further, SiO ₂ buffer layer 2 is equivalent refractive index of the high frequency electric signal _{n m} (or microwave equivalent refractive index _{n m)} the optical waveguide 3a, 3b by approximating the effective refractive index _{n o} of the light propagating the light It plays an important role in expanding the modulation band.

  Next, the operation of the LN optical modulator configured as described above will be described. In order to operate this LN optical modulator, it is necessary to apply a DC bias voltage and an RF electrical signal between the center conductor 4a and the ground conductors 4b and 4c.

The voltage-light output characteristic shown in FIG. 19 is the voltage-light output characteristic of the LN optical modulator in a certain state, and _Vb is a DC bias voltage at that time. As shown in FIG. 19, the DC bias voltage _Vb is normally set at the midpoint between the peak and bottom of the light output characteristic.

  When an LN optical modulator is used in an optical transmission system, an optical modulator module in which a chip of an LN optical modulator, an optical fiber, and a microwave connector for electric signals are fixed to a metal casing (package). I have to.

FIG. 20 shows the structure of the optical modulator module. Some numbers in the figure are the same as those in FIGS. 17 and 18. For simplicity, the SiO ₂ buffer layer 2 is omitted.

  Here, 15 is a metal casing, and 16 a and 16 b are core wires of a microwave connector (not shown) fixed to the metal casing 15. Reference numerals 17a and 17b denote cavities formed around the core wires 16a and 16b of the microwave connector. First, the LN substrate 1 is fixed to a pedestal (fixing portion). Here, for convenience of explanation, the inside 20 of the metal casing 15 in FIG. 20 is a pedestal. The CPW traveling wave electrode 4 and the core wires 16a and 16b of the microwave connector are electrically connected.

  Although not shown in FIG. 20, an optical system is set so that light can be input to and output from the LN optical waveguide 3. Next, the metal lid 8 is fixed to the metal housing 15 to complete the LN optical modulator module. Note that the output side of the high-frequency electrical signal (as can be seen by comparing FIGS. 17 and 20, the output side of the high-frequency electrical signal corresponds to the side of 16b and 17b) is electrically terminated by a terminating resistor. In this case, the core wire 16b and the cavity 17b of the microwave connector on the output side are not necessary.

  Hereinafter, the deterioration factors as an optical modulator will be considered from the viewpoint of productivity and electrical characteristics of the LN optical modulator chip. In FIG. 21, the LN wafer is shown as 10 and the chip of the z-cut LN optical modulator is shown as 11. This LN wafer has few optical grade defects, high uniformity, and is extremely expensive.

  As can be seen from the figure, since the width of the chip of the conventional z-cut LN optical modulator is about 2 to 5 mm, the adhesion strength of the metal housing 15 to the base 20 is high (especially if the width is 5 mm). For example, in the case of a 3-inch wafer, only about 10 to 20 chips can be obtained, which is problematic from the viewpoint of productivity, and the cost of the LN optical modulator from the viewpoint of material cost and labor cost. It was one of the major factors that increased

  Next, a microwave dip that significantly deteriorates the yield as an optical modulator from the viewpoint of electrical characteristics will be described. FIG. 22 is an enlarged view of the core wire 16a of the microwave connector and the cavity 17a formed in the metal housing 15 in FIG. Here, EF is a line of electric force of a high-frequency electric signal generated between the core wire 16 a of the microwave connector and the cavity 17 a formed in the metal housing 15.

  As can be seen from FIG. 22, the electric force lines EF of the high-frequency electric signal have an axially symmetric distribution around the core wire 16a of the microwave connector.

  On the other hand, FIG. 23 shows a BB ′ line of a connection portion (referred to as an input feed-through portion) of the traveling wave electrode formed on the z-cut LN substrate 1 of FIG. 20 and the core wire 16a of the microwave connector. A cross-sectional view is shown. Here, DEF represents a line of electric force generated between the center conductor 4a and the ground conductor 4b. As can be seen from FIG. 23, the electric force lines DEF have a planar distribution in BB ′ of the input feedthrough portion.

  In this way, the electric force lines EF having an axially symmetric distribution (referred to as an eigenmode of the microwave connector) around the core wire 16a of the microwave connector, and between the central conductor 4a of the traveling wave electrode 4 and the ground conductors 4b and 4c. There is a mismatch in the distribution shape of the electric force lines DEF of the planar distribution (referred to as the eigenmode of the CPW traveling wave electrode) generated in FIG. 2, and a high frequency electric signal is transmitted from the core wire 16a of the microwave connector to the traveling wave electrode 4 When propagating to the center conductor 4a and the ground conductor 4b, a component that leaks to the high-frequency electric signal is generated.

FIG. 24 shows the shape and dimensions of the z-cut LN substrate 1 constituting the LN optical modulator. In the figure, the x direction corresponds to the width (or lateral width), the z direction corresponds to the thickness, the y direction corresponds to the length, and the respective dimensions are L _x , L _z , and L _y . For the leaked high-frequency electrical signal component, the z-cut LN substrate 1 functions as a dielectric resonator. That is, it is considered that the leaked high frequency electric signal component excites a resonance mode in the z-cut LN substrate 1 and resonates (referred to as dielectric resonance) therein.

  As will be discussed in detail later, according to Patent Document 1, if the length of the diagonal line d in FIG. 25 which is a cross section of the z-cut LN substrate 1 is long, the optical transmission is hindered due to dielectric resonance. That is, when dielectric resonance occurs, much of the energy of the high-frequency electrical signal that should propagate from the core wire 16a of the microwave connector to the center conductor 4a and the ground conductors 4b and 4c of the traveling wave electrode is made of the z-cut LN substrate 1. It is consumed to excite the resonance mode inside the dielectric resonator, and is not effectively used for modulation of light in the interaction optical waveguides 3a and 3b shown in FIGS. Therefore, a drastic drop (referred to as frequency dip) occurs in the modulation index of light with respect to the modulation frequency.

Incidentally, in the idea of Patent Document 1, when the lateral width (L _x ) of the z-cut LN substrate 1 is increased to 5 mm or 6 mm, the length d of the diagonal line in the cross section becomes longer than this lateral width, and the frequency dip is 4 GHz. Appears to a degree. In the case of a modulation rate of 10 Gbit / s, when this frequency dip occurs in the frequency region around 10 GHz or less, or in the case of a modulation rate of 40 Gbit / s, this frequency dip is in the frequency region around 30 GHz or less. If this occurs, it becomes a very serious problem in practical use.

[Second prior art]
As shown in FIG. 21, when the width of the z-cut LN optical modulator chip 11 is large, the number of z-cut LN optical modulator chips 11 that can be taken per wafer is reduced. . Accordingly, FIG. 26 shows a second prior art in which the lateral width of the chip 11 of the z-cut LN optical modulator is narrowed. By doing so, a large number of chips 11 can be obtained per wafer, and the productivity in the process is remarkably improved. Thus, from the viewpoint of productivity, it is very desirable to reduce the lateral width of the chip 11 of the z-cut LN optical modulator.

  Thus, the problem of productivity per wafer could be solved by narrowing the lateral width of the chip 11 of the z-cut LN optical modulator. Furthermore, it shows that the frequency dip which is an electrical characteristic can also be solved by the 2nd prior art. The second prior art improves the influence of the frequency dip on the optical modulation index by removing the resonance frequency as the dielectric resonator in the first prior art from the frequency region necessary for optical modulation to the high frequency side. This is a technique proposed in Patent Document 1. Here, since the same numbers as those in the first prior art shown in FIGS. 17 to 25 correspond to the same function units, the description of the function units having the same numbers is omitted here.

FIG. 27 is a perspective view of the z-cut LN substrate 1 constituting the z-cut LN optical modulator of FIG. 26, and FIG. 28 is a cross-sectional view thereof. According to Patent Document 1, the frequency dip f _c of the modulation index is f _c = c ₀ / (2n · d) (1)
It is said that it is given in. Here, c ₀ is the speed of light in vacuum, n is the equivalent refractive index of the high-frequency electrical signal, and d is an important physical quantity, which is the longest length (usually the length of the diagonal line) in the cross-sectional view shown in FIG. ).

  The product n · d of the equivalent refractive index n of the high-frequency electrical signal and the length d of the diagonal line is made larger than 0.8 mm and smaller than 11 mm, so that the frequency dip has no trouble for optical transmission of 10 Gbit / s. It can be shifted to a higher frequency.

  We actually designed and manufactured an optical modulator using the z-cut LN substrate 1 according to the equation (1). In designing, according to Patent Document 1, the equivalent refractive index of the high-frequency electrical signal was 6.59 in the x direction and 5.36 in the z direction in FIG.

  The LN optical modulator was designed and manufactured based on the requirement condition (1) of the second prior art, and it was confirmed that the frequency dip could be shifted to a high frequency that does not hinder optical transmission. However, an important fact has been found that it is difficult to make an LN optical modulator with a high yield from the viewpoint of mechanical strength.

That is, when the 40GHz frequency dip _{f c} as an example, (1) the length of a diagonal line in the cross-sectional view of the z- cut LN substrate 1 from the equation d becomes 0.63 mm. Therefore, assuming that the thickness (L _z ) of the z-cut LN substrate 1 is 0.5 mm, the lateral width (L _x ) of the z-cut LN substrate 1 is only 0.39 mm.

Usually, since the length (L _y ) of the z-cut LN substrate 1 is several tens of mm, the z-cut LN substrate 1 becomes too small if the diagonal dimension of the cross section of the z-cut LN substrate 1 becomes too small. It has been found that the rigidity of the LN optical modulator becomes extremely weak, and there is a significant problem in the mechanical strength of the LN optical modulator module. Further, since the width is smaller than the thickness and the chip falls down, it is extremely difficult to handle it as a chip or mount it on a package.

Next, the mechanical strength after manufacturing this module will be considered. FIG. 29 shows the adhesive strength of the metal housing 15 shown in FIG. 20 to the base 20 when the area of the bottom surface of the z-cut LN substrate 1 is a variable. As can be seen from the figure, when the lateral width (L _x ) of the z-cut LN substrate 1 is reduced and the area of its bottom surface is reduced, the adhesive strength to the metal casing 15 is significantly deteriorated, and the mechanical strength after mounting is reduced. It was found from the vibration / impact test that the z-cut LN substrate 1 was peeled off.

[Third prior art]
Next, an attempt to reduce the drive voltage (RF drive voltage) at a high frequency of the optical modulator is a third conventional technique, and problems occurring at that time are considered. FIG. 30 shows the relationship between the product Vπ · L of the half-wave voltage Vπ and the length L of the interaction portion in FIG. 17 and the gap W of the CPW. In addition, as the gap W of CPW, about 20-30 micrometers is used now. When the gap W of the CPW is narrowed, the strength of the high frequency electric field that interacts with the light propagating through the interaction optical waveguides 3a and 3b increases. Therefore, as shown in this figure, when the gap W of the CPW is narrowed, the product Vπ · L becomes small. An LN optical modulator with a lower driving voltage can be realized as the product Vπ · L is lower. The driving voltage when driving the LN optical modulator at a speed of 10 Gbit / s or more is about 5 to 6 V, and it is desired that the driving voltage is as low as possible. Therefore, it is desirable that the gap W of the CPW is narrow from the viewpoint of the driving voltage.

It shows the relationship between the microwave equivalent refractive index n _m and CPW gaps W high-frequency electric signal in FIG. 31. It shows the interaction optical waveguides 3a, the equivalent refractive index of the light propagating through _{3b n o (n o ≒ 2.2} ) also in FIG.

When the gap W of the CPW is narrowed, the high frequency electric signal generated between the center conductor 4a and the ground conductors 4b and 4c feels a lot of the SiO ₂ buffer layer 2 having a low relative dielectric constant of about 4, so that the microwave equivalent refractive index n _m can be reduced (the relative dielectric constant of the z-cut LN substrate 1 is about 35).

In general, the microwave equivalent refractive index n _m is greater than the equivalent refractive index n _o of the light, is a major limiting factor in the operation of the LN optical modulator at a high speed and a wide band. Therefore the LN optical modulator is driven at a speed higher than 10Gbit / s becomes essential to close the microwave equivalent refractive index n _m in the light of the equivalent refractive index n _o. From this point of view, it is desirable that the gap W of the CPW is narrow.

As described above, although from the viewpoint while reducing the driving voltage close to the microwave equivalent refractive index n _m in the light of the equivalent refractive index n _o is the CPW gap W was found to be narrow it is desirable, the gap CPW Problems that occur when W is narrowed to generally 25 μm or less, such as 15 μm or less, are described below.

Figure 32 shows the CPW gaps W as a variable for (corresponding to _{Z 39} in the following Figure 34) the center electrode 4a and the ground electrodes 4b, the characteristic impedance Z of the CPW traveling wave electrode 4 made of 4c. When the CPW gap W is narrowed, the characteristic impedance Z is remarkably reduced to 30Ω or less, and impedance mismatching occurs with an external signal source of almost 50Ω. That is, there arises a problem that the power reflectance (so-called S ₁₁ ) of the high-frequency electric signal is deteriorated.

  Next, this will be considered in more detail. FIG. 33 shows a top view of the CPW traveling wave electrode 4 comprising the center conductor 4a and the ground conductors 4b and 4c constituting the third prior art z-cut LN optical modulator.

  Here, I is an input feed-through unit for connecting a core wire (or a gold ribbon or a gold wire) of a connector (not shown) for applying a high-frequency electrical signal from an external signal source (not shown) to the CPW traveling wave electrode 4; II is a connection portion (or input side connection portion) between the input feedthrough portion I and the interaction portion III, III is an interaction portion where an electric signal and light interact, and IV is an interaction with the output feedthrough portion V. It is a connection part (or output side connection part) with the part III. The output feed-through portion V is connected to a core wire (or a gold ribbon or a gold wire) or a terminating resistor (not shown).

  A portion that feeds a high-frequency electrical signal in the central conductor of the input feedthrough portion I is referred to as a feeding portion, and a portion that takes out the high-frequency electrical signal in the central conductor of the output feedthrough portion V is referred to as an output portion.

FIG. 34 shows an equivalent circuit of a third prior art z-cut LN optical modulator. Here, 5 and 6 correspond to external circuits, 5 is an external signal source such as an electric driver, and 6 is a load resistance (characteristic impedance is R _g ) of the external signal source. Reference numerals 37 to 41 respectively correspond to equivalent lines from the input feedthrough portion I to the output feedthrough portion V. Specifically, 37 represents an input feedthrough section I, 38 represents an input side connection section II, 39 represents an interaction section III, 40 represents an output side connection section IV, and 41 represents an output feedthrough section V line. . Reference numeral 42 denotes a termination resistor.

Further, Z _{37 to} Z ₄₁ are characteristic impedances from the input feed-through portion I to the output feed-through portion V. Specifically, Z ₃₇ is the input feed-through portion I (or line 37), Z _38. Is the input side connection part II (or line 38), Z ₃₉ is the interaction part III (or line 39), Z ₄₀ is the output side connection part IV (or line 40), and Z ₄₁ is the output feedthrough part V (or This corresponds to the characteristic impedance of the line 41). Z _L is the resistance value of the termination resistor 42.

  Next, with respect to the third prior art z-cut LN optical modulator shown in FIGS. 30 to 34, problems from the viewpoint of impedance mismatch and modulation band will be considered.

In FIG. 34, Z _in is an input impedance when the z-cut LN optical modulator is viewed from the external signal source 5 and the load resistor 6 (impedance R _g ). _{That, Z in} the input feed-through portion characteristic impedance _{Z 37} of the I, characteristic impedance _{Z 38} of the input-side connecting portion II, the interaction portion III of the characteristic impedance _{Z 39,} the output-side connecting portion characteristic impedance _{Z 40} of the IV, the output The characteristic impedance Z ₄₁ of the feedthrough portion V for use and the resistance value Z _L of the termination resistor 42 are synthesized based on the concept of cascade connection of transmission lines in consideration of the length of each portion and the equivalent refractive index of the electric signal propagating through each portion. It can be said to be impedance. 43 in the figure represents the boundary between the external signal source 5 and the load resistor 6 and the input feedthrough I.

Lowering the driving voltage, consider the case where the CPW gaps W in order to close the microwave equivalent refractive index n _m for the light of the equivalent refractive index n _o and narrow as 15μm or less. In this case, the characteristic impedance _{Z 39} of the interaction unit III is as low as 30Ω or less.

In the third prior art, the characteristic impedances of the other lines ₃₇ , ₃₈ , ₄₀ , 41, that is, the characteristic impedance Z _{37 of} the input feed-through part I, the characteristic impedance Z _{38 of} the input side connection part II, and the output side connection The characteristic impedance Z ₄₀ of the part IV, the characteristic impedance Z _{41 of the} output feedthrough part V, and the resistance value Z _L of the termination resistor 42 were all equal to the characteristic impedance Z ₃₉ of the interaction part III (that is, Z _{37 = Z 38 = Z 39 = Z} 40 = Z 41 = Z L).

As a result, the real part Re (Z _in ) of the input impedance Z _{in when} the z-cut LN optical modulator is viewed from the load resistor 6 of the external signal source 5 hardly depends on the frequency f as shown by the solid line in FIG. , consistent with the characteristic impedance _{Z 39} of the interaction portion III, it was as low as 30Ω or less.

Accordingly, the optical modulation index (power modulation index) | m | ² is shown in FIG. 36 due to impedance mismatch between the input impedance Z _in and the load resistance 6 (impedance R _g ) of the external signal source 5. It was extremely difficult to secure 10 GHz as a 3 dB light modulation band because of rapid deterioration with the frequency f.

Assuming that Z ₃₇ = Z ₃₈ = Z ₃₉ = Z ₄₀ = Z ₄₁ = Z _L , when Z _in (in this case, Z _in = Z ₃₉ ) is 40Ω or less, for example, 30Ω, the power of the high-frequency electric signal As shown in FIG. 37, the reflectivity (S ₁₁ ) becomes as high as (−13 dB) (R _g = 50Ω). In an actual experiment, the high-frequency electricity reflected by a slight impedance mismatch Signal will be superimposed and further deteriorated). Here, the power reflectivity (S ₁₁ ) of the high-frequency electrical signal is given by the following equation.
S ₁₁ = | (R _g −Z _in ) / (R _g + Z _in ) | ² (1)
Further, when the reflected high-frequency electric signal returns to the external signal source 5, there is a problem that the jitter of the modulated optical pulse is increased. Here, Z ₃₉ is 30Ω as an example, and the above-described problem is not limited to Z ₃₉ = 30Ω, and particularly occurs when Z ₃₉ is 40Ω or less.

[Fourth Prior Art]
As a technique for improving the impedance mismatch between the characteristic impedance Rg of the external signal source 5 and the characteristic impedance _Z3 of the interaction part III in the _third conventional technique, the fourth conventional technique proposed as Patent Document 2 Will be described. Here, since the same numbers as those in the third prior art shown in FIGS. 30 to 37 correspond to the same function units, the description of the function units having the same numbers is omitted here.

FIG. 38 shows a top view of the CPW traveling wave electrode 4 used in the fourth prior art. Also in the fourth prior art, I is an input feedthrough section, II is an input side connection section, III is an interaction section, IV is an output side connection section, and V is an output feedthrough section. The output feed-through portion V is also connected to a connector core wire (or gold ribbon or gold wire) (not shown) or a terminal resistor. In addition to the same configuration as those of the third prior art, it is added impedance transformation portion VI of length L ₆ in the fourth prior art shown in FIG. 38.

FIG. 39 shows a cross-sectional view as a z-cut LN optical modulator at BB ′ of the interaction part III in FIG. In the fourth prior art shown in FIG. 39, a case is assumed in which the gap W of the CPW is set to be as narrow as about 15 μm or less. If the gap W of CPW thus narrow, microwave equivalent refractive index n _m of the interaction optical waveguides 3a of the high-frequency electrical signal can be reduced and at the same drive voltage as described above, the equivalent refractive index of the light propagating through 3b n _o However, the characteristic impedance Z ₃₉ of the interaction part III is 40Ω or less (for example, 30Ω as described in the first prior art).

  FIG. 40 shows a cross-sectional view of the impedance converter VI as a z-cut LN optical modulator at C-C ′. The CPW gap W ′ in the impedance conversion unit VI is set to be about 50 μm, which is wider than the CPW gap W in the interaction unit III.

FIG. 41 shows an equivalent circuit of the fourth prior art. As in the first prior art shown in FIG. 34, Z ₃₇ is an input feed-through portion I (or line 37), Z ₃₈ is an input side connection portion II (or line 38), and Z ₃₉ is an interaction portion III ( or the line _{39), Z 40} is the output-side connecting portion IV (or the line _40), but _{Z 41} is a characteristic impedance of the output feed-through portion V (or the line 41), the fourth prior art characteristic impedance Z ₄₄ impedance converters VI (or lines 44) are added.

41 in FIG. 41 represents a boundary between the load resistance 6 (characteristic impedance R _g ) of the external signal source 5 and the input feedthrough I. In FIG. 41, Z _in ′ is an input impedance when the fourth conventional technology z-cut LN optical modulator is viewed from the external signal source 5 and the load resistor 6 of the external signal source 5.

_{That, Z in} 'the characteristic impedance _{Z 37} of the input feed-through portion I, the characteristic impedance _{Z 44} of the impedance conversion section VI, the input-side connecting portion II of the characteristic impedance _{Z 38,} the interaction portion III of the characteristic impedance _{Z 39,} the output It can be said that the characteristic impedance Z ₄₀ of the side connection portion IV, the characteristic impedance Z _{41 of the} output feed-through portion V, and the Z _{L of the} termination resistor 42 are combined with the concept of cascade connection of transmission lines.

As described above, since in the fourth prior art by narrowing the CPW gaps W and 15μm or less, such as characteristic impedance Z ₃ of the interaction portion III is for example 30 [Omega, it is lower than 40 [Omega.

Next, the operation of the impedance converter VI will be considered. For simplicity, the characteristic impedance Z ₃₈ of the input side connection part II, the characteristic impedance Z ₃₉ of the interaction part III, the characteristic impedance Z ₄₀ of the output side connection part IV, the characteristic impedance Z _{41 of the} output feedthrough part V, and Assume that Z ₃₇ = Z ₃₉ = Z ₄₀ = Z ₄₁ = Z _L holds for Z _L of the termination resistor 42.

An equivalent circuit in this case is shown in FIG. Here, the characteristic impedance of the transmission line 39 ′ representing the combined portion III ′ formed by combining Z ₃₇ , Z ₃₉ , Z ₄₀ , Z ₄₁ , and Z _L is defined as Z ₃₉ ′. Further, characteristic impedance noted in FIG. 38 indicates the length of the impedance transformation portion VI is Z ₄₄ as L ₆ in FIG. 42.

Here, according to Patent Document 2, the characteristic impedance Z ₃₇ of the input feedthrough I is set to 50Ω, which is the same as the characteristic impedance R _g of the load resistor 6 of the external signal source 5.

As is well known from the theory of cascade connection of transmission lines, when there is an impedance converter having a characteristic impedance Z and an electrical length L between two different characteristic impedances Z _i and Z _j , Z between Z, Z _i , Z _j and L = (Z _i · Z _j ) ^1/2 (2)
L = Λ / 4 (3)
Is satisfied, impedance matching is established between Z _i and Z _j , and there is no electrical reflection. Where λ is the electrical wavelength of the high frequency electrical signal. Z expressed by the geometric mean of Z _i and Z _j is called matching impedance.

In the fourth prior art, Z _{37 corresponds} to Z _i , Z ₃₉ ′ corresponds to Z _j , Z ₄₄ corresponds to Z, and L ₆ corresponds to L. That is, in this case, impedance matching is established between the input feedthrough portion I (Z ₃₇ ) and the combining portion III ′ (Z ₃₉ ′), and electrical reflection is eliminated. In the following, for the sake of simplicity, Z _i in the formula (2) will be described as Z ₃₇ , Z _j as Z ₃₉ ′, Z as Z ₄₄ , and L in the formula (3) as L ₆ .

In the fourth prior art, λ is a wavelength when the high frequency electrical signal from the external signal source 5 propagates through the LN substrate of the impedance converter VI, and means the wavelength of the electromagnetic wave propagating through the LN substrate. Called the wavelength. Specifically, the wavelength in vacuum of the high frequency electric signal and lambda _{0, 'the} lambda When λ = λ _{0 /} n _m' high-frequency electrical signal is an equivalent refractive index at the time of propagating an LN substrate n _m ( 4)
As given.

Next, FIG. 43 shows an example in which the dependence of the electrical power reflectance S _{11 on} the frequency f when Z ₃₇ = R _g = 50Ω and Z ₃₉ ′ = 30Ω is calculated. Here, the characteristic impedance Z ₄₄ of the impedance converter VI is 38.7Ω. The length _{L 6} of the impedance transformation portion VI becomes 6mm an equivalent refractive index _{n m} 'of 2.5. The widths S and S ′ of the central conductors in the interaction part III and the impedance conversion part VI are the same.

As can be seen from FIG. 43, when the impedance converter VI is designed so as to satisfy the expressions (2) and (3), S _{11 is set} to −50 dB or less at a specific frequency such as 5 GHz, 10 GHz, 15 GHz, or 20 GHz. can, electrical power reflection can be eliminated almost completely (Note that in actual experiments, S ₁₁ due electrical reflections from such connecting portion of the cable is never a negative infinity, In the fourth conventional technique, the conditions of the expressions (2) and (3) are satisfied.

As described above, in the fourth prior art, at a specific frequency satisfying the expressions (2) and (3) (here, 5 GHz, 10 GHz, 15 GHz, 20 GHz, etc.) Although it is possible to eliminate the reflected almost completely, usually, very small value such as -50 dB, or less as an electric power reflectance _{S 11} is not required, the minimum -10 dB (actually, about -12dB want ) ˜−15 dB, or −15 dB˜−20 dB is sufficient, so to speak, it can be said that it is an overspec.

On the other hand, as is clear from FIG. 43, at 7.5 GHz, 12.5 GHz, and 17.5 GHz, the electric power reflectivity S ₁₁ reaches the level of −13 dB, which is the envelope (broken line in FIG. 43). As a result, the reflection characteristic deteriorates so that it cannot be used practically like the level of the first prior art.

Furthermore, (in general, 10 [mu] m before and after) the same size as the width S of the center conductor width S 'is the interaction portion III of the center conductor impedance converter VI so narrow, the equivalent refractive index n _m of the high-frequency electrical signals 2 Small, around 5 Therefore, the length of the impedance converter VI becomes relatively long as can be seen from the equations (3) and (4) (6 mm in the fourth prior art). In particular, as shown in FIG. 38, when the impedance conversion unit VI is formed perpendicular to the interaction unit III or with a large angle, the lateral width of the element as the LN optical modulator increases, and one LN substrate is formed. There is also a problem that the number of LN optical modulators that can be taken per wafer is reduced. That is, there is a problem that it is difficult to form the entire impedance conversion unit that requires a considerable length in the LN optical modulator and to reduce the size of the LN optical modulator.

Furthermore, as described above, the width S ′ of the center conductor of the impedance converter VI is as narrow as the width S of the center conductor of the interaction section III (generally around 10 μm). There is also a problem that the high-frequency electrical signal is easily attenuated, and the high-frequency electrical signal cannot be sufficiently used for optical modulation in the interaction part III.
JP-A-3-253814 JP 2005-37547 A

As described above, if the width of the LN optical modulator chip is wide, the total number of LN optical modulator chips that can be taken from one wafer is reduced. Therefore, by reducing the width of the LN optical modulator chip, the total number of LN optical modulator chips that can be taken from one wafer is increased, and the cost of the LN optical modulator chip is reduced. The productivity is improved and the second prior art concept, that is, the length (usually the length of the diagonal line) that is the longest in the cross section of the LN substrate is d, and the equivalent refractive index of the high-frequency electric signal is n. obtains a frequency dip _{f c} from _{f c = c 0 / (2n} · d), the cross-sectional structure of the LN optical modulator based on the idea of setting the length d of the diagonal so that the frequency dips _{f c} is sufficiently high As a result, the dimension of the cross section of the LN optical modulator became extremely small, and the mechanical strength of the LN optical modulator as a module deteriorated. In particular, in a mechanical reliability test such as vibration and shock after the manufacture of the optical modulator module, a fatal problem that the chip of the LN optical modulator itself is peeled off from the metal casing of the optical modulator arises. It was. Further, since the width is smaller than the thickness of the LN optical modulator, there is a problem in handling such that the chip falls down when the package is mounted. On the other hand, in the optical modulator according to the third prior art in which each part constituting the CPW traveling wave electrode such as the feed-through part and the interaction part on the input side and the output side has the same characteristic impedance, the drive voltage is reduced. In addition, when the buffer layer is thinned or the gap of the CPW traveling wave electrode is narrowed in order to bring the microwave equivalent refractive index close to the equivalent refractive index of light, impedance mismatching with an external circuit occurs, resulting in electrical There are problems that the frequency characteristic of the power reflectivity is poor and the optical modulation band rapidly deteriorates with respect to the modulation frequency. That is, when the drive voltage is reduced, the characteristic impedance is inevitably lowered, and accordingly, the electrical power reflectance is deteriorated and the jitter of the optical pulse is increased. As a result, it has been difficult so far to reduce the drive voltage and suppress the jitter of the optical pulse. Further, in the fourth conventional technique in which the impedance conversion unit that satisfies the equations (2) and (3) described above is provided between the input feed unit and the interaction unit, electrical reflection is performed at a specific frequency. Although the electrical power reflectance can be made extremely small so that can be almost ignored, the electrical reflection characteristics deteriorate to the level of the third prior art at other frequencies. Then, since the reflected high-frequency electric signal returns to the external signal source, there is a serious problem that the jitter of the finally modulated optical pulse is increased. Further, in general, the length of the impedance conversion unit given by λ / 4 where λ is the guide wavelength is long. Therefore, when this impedance conversion unit is formed on the wafer together with the z-cut LN optical modulator, the z-cut LN light is formed. Since the overall size of the modulator is increased, there is a problem that the number of LN optical modulators that can be produced per wafer is greatly limited. From the above, while increasing the total number of chips of the LN optical modulator can be taken from a single wafer, with removing the frequency dip f _c from the use frequency domain, the optical pulse light modulation band is wide, and to produce LN optical modulators that have low jitter and can withstand mechanical strength as LN optical modulators, especially mechanical tests such as vibration and shock after module fabrication, have not been developed yet, and this is achieved. The technology for this was desired.

  The present invention has been made in view of such circumstances, and is low in cost, high in productivity, strong in mechanical strength, and excellent in electrical reflection characteristics and jitter characteristics of generated light pulses. An object is to provide a modulator.

  In order to solve the above problems, an optical modulator according to claim 1 of the present invention includes a substrate having an electro-optic effect, an optical waveguide for guiding light formed on the substrate, and one of the substrates. An electrode comprising a center conductor and a ground conductor formed on the surface side for applying a high-frequency electric signal for modulating the phase of the light; a fixed portion to which the substrate is fixed; and applying the high-frequency electric signal to the electrode In the optical modulator comprising: an interaction unit that modulates the phase of the light; and a first input feedthrough unit for propagating a high-frequency electrical signal to the interaction unit. There is further provided at least one separate substrate that is juxtaposed and fixed to the portion, and the resonant frequency of the dielectric resonance substantially determined from the thickness of the substrate and the thickness of the separate substrate is the high-frequency electric Trust The thickness of the substrate and the thickness of the separate substrate are set so as to be higher than the frequency of the second substrate, and the separate substrate has a second input feedthrough portion composed of a center conductor and a ground conductor. And the second input feedthrough unit so that the high-frequency electrical signal from the external circuit applied to the second input feedthrough unit propagates to the first input feedthrough unit. The first input feedthrough portion is electrically connected, the characteristic impedance of the second input feedthrough portion, and the connector to be electrically connected to the second input feedthrough portion At least one for reducing impedance mismatch between the characteristic impedance and at least one characteristic impedance of the external circuit and the interaction unit The impedance converter on the separate substrate while the high frequency electrical signal applied from the external circuit to the second input feedthrough unit causes residual reflection. Compared to the case where there is no electrical reflection, the electrical reflection is reduced and propagates to the interaction section.

  The optical modulator according to claim 2 of the present invention is the product of the thickness of the substrate and the equivalent refractive index of the high-frequency electrical signal in the thickness direction, or the thickness of the separate substrate and the high-frequency electrical signal in the thickness direction. The thickness of the substrate and the thickness of the separate substrate are set so that at least one of the products with the equivalent refractive index of the substrate is larger than 0.4 mm and smaller than 15 mm.

  In the optical modulator according to claim 3 of the present invention, the sum of the width of the substrate and the width of the separate substrate is about 1.4 times or more the thickness of at least one of the substrate or the separate substrate. It is characterized by that.

  An optical modulator according to a fourth aspect of the present invention is characterized in that substantially all of the side surfaces of the substrate and the opposing side surfaces of the separate substrate are fixed to each other by an adhesive means.

  The optical modulator according to claim 5 of the present invention is characterized in that a part of a side surface of the substrate and a side surface of the separate substrate facing each other are fixed to each other by an adhesive means.

  The optical modulator according to claim 6 of the present invention is characterized in that the adhesive strength of the substrate to the fixed portion is higher than when the separate substrate is not provided.

  The optical modulator according to claim 7 of the present invention is characterized in that the separate substrate is made of a dielectric made of the same material as the substrate.

  The optical modulator according to claim 8 of the present invention is characterized in that the separate substrate is made of a dielectric material made of a material different from that of the substrate.

  The optical modulator according to claim 9 of the present invention includes a low dielectric constant layer having a relative dielectric constant smaller than that of the substrate, at least below the second input feedthrough portion, and the thickness of the substrate and the thickness of the substrate. The low dielectric constant according to the thickness of the substrate and the thickness of the separate substrate so that the resonance frequency of the dielectric resonance determined from the thickness of the low dielectric constant layer is higher than the frequency of the high frequency electric signal. The thickness of the layer is set.

  The optical modulator according to claim 10 of the present invention is characterized in that at least a part of the low dielectric constant layer is made of air.

  An optical modulator according to an eleventh aspect of the present invention is characterized in that the relative dielectric constant of the low dielectric constant layer is made of a material larger than that of air and smaller than that of the substrate.

  In an optical modulator according to a twelfth aspect of the present invention, at least a part of the impedance converter formed on the separate substrate is characterized by the characteristic impedance of the interaction unit and the characteristic of the second input feedthrough unit. Geometric mean of impedance, geometric mean of characteristic impedance of the interaction portion and characteristic impedance of a connector to be electrically connected to the second input feedthrough portion, or characteristic impedance of the interaction portion and the It has a characteristic impedance different from the geometric mean with the characteristic impedance of the external circuit.

  In an optical modulator according to a thirteenth aspect of the present invention, a part of the impedance conversion unit formed on the separate substrate includes a characteristic impedance of the interaction unit and a characteristic impedance of the second input feedthrough unit. The geometrical mean of the interaction part and the characteristic impedance of the connector to be electrically connected to the second input feedthrough part, or the characteristic impedance of the interaction part and the external It has the same characteristic impedance as at least one of the geometric mean with the characteristic impedance of the circuit.

  In the optical modulator according to claim 14 of the present invention, another impedance converter is provided on the substrate on which the optical waveguide is formed, the second input feedthrough part, the impedance converter, and the first The input feedthrough section and the another impedance conversion section function as an impedance conversion section as a whole, and the characteristic impedance of the impedance conversion section as a whole is the characteristic impedance of the interaction section and the second input. Characterized by having a characteristic impedance different from a geometric mean of the characteristic impedance of the connector to be electrically connected to the feedthrough part for use or a geometric mean of the characteristic impedance of the interaction part and the characteristic impedance of the external circuit To do.

  In an optical modulator according to a fifteenth aspect of the present invention, at least a part of the impedance conversion unit on the separate substrate includes a characteristic impedance of the interaction unit and a characteristic impedance of the second input feedthrough unit. The geometric mean of the interaction part, and the geometric impedance of the characteristic impedance of the connector to be electrically connected to the second input feedthrough part, or the characteristic impedance of the interaction part and the external circuit The characteristic impedance is different from the geometric mean of the characteristic impedance, and the characteristic impedance is within about ± 7Ω.

  In an optical modulator according to a sixteenth aspect of the present invention, at least a part of the impedance converter on the separate substrate is characterized by the characteristic impedance of the interaction unit and the characteristic of the second input feedthrough unit. Geometric mean of impedance, geometric mean of characteristic impedance of the interaction portion and characteristic impedance of a connector to be electrically connected to the second input feedthrough portion, or characteristic impedance of the interaction portion and the The characteristic impedance is different from the geometric mean with the characteristic impedance of the external circuit, and the characteristic impedance is within about ± 15Ω.

  The optical modulator according to claim 17 of the present invention has a power reflectance envelope of between -10 dB and -15 dB at a frequency of at least one point in a frequency range between about 20% and about 30% of the operating bit rate. And having a minimum point where the first derivative of the power reflectance envelope is zero and the second derivative is positive in a frequency range between about 40% and about 70% of the operating bit rate. At least a part of the impedance converter has a characteristic impedance lower than that of the connector or the external circuit so that the power reflectivity is -15 dB or less.

  An optical modulator according to claim 18 of the present invention is such that the power reflectivity at least at one point in a frequency range between about 2 GHz and 4 GHz is between -10 dB and -15 dB, and within a frequency range of about 4 GHz to 9 GHz. Comprising the above-mentioned minimum point.

  The optical modulator according to claim 19 of the present invention is such that the power reflectivity at at least one point in a frequency range between about 5 GHz and 8 GHz is between -10 dB and -15 dB, and within a frequency range of about 10 GHz to 18 GHz. Comprising the above-mentioned minimum point.

  The optical modulator according to claim 20 of the present invention is such that the power reflectance at at least one point in the frequency range between about 9 GHz and 13 GHz is between -10 dB and -15 dB, and within the frequency range of about 17 GHz to 30 GHz. Comprising the above-mentioned minimum point.

  An optical modulator according to claim 21 of the present invention is such that the power reflectivity at least at one point in a frequency range between about 11 GHz and 16 GHz is between -10 dB and -15 dB, and within a frequency range of about 22 GHz to 38 GHz. Comprising the above-mentioned minimum point.

  The optical modulator according to claim 22 of the present invention is such that the power reflectivity at least at one point in the frequency range between about 22 GHz and 33 GHz is between -10 dB and -15 dB, and within the frequency range of about 44 GHz to 77 GHz. Comprising the above-mentioned minimum point.

  The optical modulator according to claim 23 of the present invention is characterized in that the substrate is lithium niobate.

  The optical modulator according to claim 24 of the present invention is characterized in that the substrate is a semiconductor.

According to the present invention, by narrowing the lateral width of the LN optical modulator chip, many thin optical modulator chips can be obtained from an expensive optical grade wafer, so that the material cost for the optical grade wafer is increased. Thus, the cost of the optical modulator can be significantly reduced from the viewpoint of labor costs in a process that is long in time and occupies most of the actual cost. Further, the LN optical modulator chip and the extremely inexpensive SAW grade LN substrate can be regarded as a mechanically integrated LN substrate, and the mechanical strength is maintained without increasing the cost, while the thickness of the integrated LN substrate is reduced. by sufficiently widening the width as well as accurate harmful frequency dip f _c for optical transmission, there is a great advantage that ensures and can be shifted to a higher frequency than readily used frequency domain. In other words, since the impedance conversion unit is formed in the chip of the LN optical modulator, the size of the chip of the LN optical modulator is increased. Since a part or the whole is formed on a separate substrate, the lateral width of an LN optical modulator chip using an optical grade LN substrate can be reduced, and a large number of LN optical modulators can be formed from one optical grade LN substrate. You can take a tip. As a result, the optical modulator manufactured by increasing the bonding area to the package housing as a module and ensuring high mechanical strength, is expensive as a material because it is an optical grade, and requires a long man-hour. can be taken from the wafer can be designed independently of the total number and frequency dips f _c of the chip. Furthermore, since the drive voltage has been reduced, the level required for optical transmission over a wide frequency range, rather than greatly improving the degradation of the band due to the reduced characteristic impedance of the interaction unit only at a specific frequency. In addition, the increase in the jitter of the optical pulse caused by the electrical reflection of the high-frequency electrical signal can be suppressed, so that it is possible to realize a broadband optical modulator with low driving voltage and low bandwidth, as well as an optical pulse with low jitter. It becomes possible to do.

  Hereinafter, embodiments of the present invention will be described. However, since the same numbers as those of the conventional embodiments shown in FIGS. 17 to 43 correspond to the same function units, the description of the function units having the same numbers is omitted here. To do. The optical waveguide, traveling wave electrode, interaction portion, and input feedthrough portion are described as being formed in the same manner as in the prior art, but are not limited thereto.

[First Embodiment]
In the present invention, an expensive optical grade z-cut LN substrate is used in an engine portion as an LN optical modulator in a shape having a narrow lateral width, and a separate substrate made of an inexpensive SAW grade LN substrate is used in combination. This ensures mechanical strength. The relative dielectric constants of the optical grade LN substrate and the SAW grade LN substrate are substantially equal to each other. Therefore, as will be discussed in detail later, when an optical grade LN substrate and a separate SAW grade LN substrate are used side by side (that is, juxtaposed) as in the present invention, electromagnetically Can be considered as an optical grade integral LN substrate.

  FIG. 1 shows a schematic view of the first embodiment of the present invention. However, for simplicity of explanation, the buffer layer and the electrode pattern are omitted. In the figure, reference numeral 80 denotes an optical grade z-cut LN substrate whose width is narrower than that of the first prior art. As is well known, X-ray topographic observations show that optical grade LN substrates are much better in crystallinity than SAW grade LN substrates. As a result, those skilled in the art use optical grade LN substrates, although they are expensive, for the fabrication of LN optical modulators. On the other hand, a SAW grade LN substrate is generally not applicable to an optical modulator, and is applied to an acoustooptic device that does not form an optical waveguide. 50 is, for example, a SAW grade LN substrate that is a fraction of the optical grade and low in price (hereinafter, 50 is distinguished from the optical grade z-cut LN substrate 80 by using a SAW grade LN substrate, a reinforcing plate, or another Called the body substrate). For example, an alumina substrate, a lithium tantalate substrate, or a quartz substrate may be used as a separate substrate. However, if the thermal expansion coefficient is close to the z-cut LN substrate 80 and the price is low, the substrate is not limited to the SAW grade LN substrate. Needless to say.

  Regarding the orientation of the substrate, the SAW grade LN substrate 50 is also advantageous from the viewpoint of the thermal expansion coefficient if the orientation of the substrate is z-cut. Here, 51 is a contact surface on the side surface of the z-cut LN substrate 80 and the SAW grade LN substrate 50, and the side surface of the z-cut LN substrate 80 and the SAW grade LN substrate 50 is an adhesive (not shown) on the contact surface 51. They are partially or wholly adhered to each other, or are mechanically in close contact or in contact. Note that since the SAW grade LN substrate 50 attached to the z-cut LN substrate 80 is not an optical grade, the price of the original wafer is low, and the cost is not increased from the viewpoint of material costs.

  Furthermore, in practice, the SAW grade LN substrate 50 does not require a long process, and after processing a simple electrode pattern, the SAW grade LN substrate can be cut at a specified width using a programmable dicer. good. Further, after the z-cut LN substrate 80 is bonded to the pedestal 20 that is a fixed portion in the metal casing 15, the SAW grade LN substrate 50 can be bonded to the pedestal 20 with an adhesive in a very short time. Therefore, since the electrode pattern is simple, the cost hardly increases from the viewpoint of labor cost.

  In the present invention, since the lateral width of the z-cut LN substrate 80 is narrow, the number of LN optical modulators that can be obtained from an expensive optical grade wafer can be remarkably increased. Since the process of manufacturing the LN optical modulator is long in time, the labor cost in the process actually occupies a large proportion of the cost in the LN optical modulator. Therefore, if the total number of LN optical modulators that can be manufactured in one process using the present invention can be significantly increased, the cost can be greatly reduced from the viewpoint of labor costs in the process. In other words, if the total number of optical modulators that can be taken from a single wafer is doubled, the labor cost can be halved, and if the total number is tripled, the cost is １ ／. In Patent Document 2, since the entire impedance conversion unit is also formed on an optical grade LN substrate, the size of the LN optical modulator is increased. That is, the idea of Patent Document 2 is not a good idea from the viewpoint of cost.

  Based on the concept of the present invention, the z-cut LN substrate 80 for forming the optical waveguide has been discussed on the premise of the optical grade, but in order to reduce the cost, the SAW grade LN substrate is used instead of the optical grade. Even in the case of using as a modulator, by applying the present invention, the cost related to the labor cost can be greatly reduced as in the case of the optical grade LN substrate. This is true for all embodiments of the invention.

  In the present invention, the set of the z-cut LN substrate 80 and the SAW grade LN substrate 50 shown in FIG. 1 is fixed to the base 20 of the metal casing 15 instead of the z-cut LN substrate 1 in FIG. As a module module.

  FIG. 2A is a cross-sectional view of the set of the z-cut LN substrate 80 and the SAW grade LN substrate 50 shown in FIG. Since the z-cut LN substrate 80 and the SAW grade LN substrate 50 are bonded to or in close contact with, for example, the whole or part of the side surfaces on the contact surface 51 by an adhesive, 80 ′ in FIG. It can be substantially regarded as a substantially integral LN substrate shown as. As will be discussed later, it has been confirmed that an air layer may exist between the z-cut LN substrate 80 and the SAW grade LN substrate 50 from the viewpoint of the resonance frequency.

  Since this integrated LN substrate 80 'has a large lower surface area, the adhesive strength of the metal casing 15 shown in FIG. 20 to the base 20 is high. That is, the bottom surfaces of the z-cut LN substrate 80 and the SAW grade LN substrate 50 in FIG. 2A are bonded to the metal casing 15, and the z-cut LN substrate 80 and the SAW grade LN substrate 50 are bonded (or contacted) to each other. Therefore, this is equivalent to the fact that the integral LN substrate 80 ′ is bonded to the base 20 of the metal casing 15.

  In this manner, according to the present embodiment, a large number of LN optical modulator chips are taken out from a single wafer, thereby significantly reducing material costs and labor costs in the process. The weakness in the vibration / impact test after assembling the module of the second prior art shown can be overcome, and the mechanical strength equivalent to that of the first prior art shown in FIGS. 20 and 21 can be realized.

Next, consider the frequency dips f _c of the optical modulation index in the first LN optical modulator, which has been a problem in the prior art. The cause of the frequency dips f _c of the optical modulation index in the LN optical modulator, and higher-order modes of the high-frequency electrical signal propagating through traveling wave electrode, but although the two resonance as the dielectric resonator of the LN substrate conceivable, Here, in all embodiments of the present invention, the traveling wave electrodes are designed to excite and propagate the fundamental mode.

Thus, in all embodiments of the present invention, the frequency dips f _c of the optical modulation index is determined by the resonance frequency of the substrate as a dielectric resonator. Then, with respect to this frequency dip f _c determined by the resonance frequency of the substrate as a dielectric resonator, the dimension of each side of the substrate is utilized to influence in inverse proportion to their squared.

In the second prior art described above, based on a cross-sectional view shown in FIG. 28, designed based on equation (1) as an expression that gives a frequency dip f _c, the longest in the cross-sectional view length, That is, the diagonal length d is determined.

  On the other hand, in the present invention, we return to the principle / principle formula of the dielectric resonator. In the present invention, since the transmission rate when driving the LN optical modulator is 10 Gbit / s or 40 Gbit / s, the wavelength of the high-frequency electrical signal is as long as centimeter order or millimeter order. Further, the relative dielectric constants of the z-cut LN substrate 80 and the SAW grade LN substrate 50 are substantially equal. Therefore, as described above, for the electromagnetic field propagating through the traveling wave electrode 4 of the LN optical modulator, the z-cut LN substrate 80 and the SAW grade LN substrate 50 are not formed as two separate substrates, but as an integrated LN. Feel as substrate 80 '.

In other words, the resonance frequency of the dielectric resonator consisting of LN substrate 80 'of the integral shown in FIG. 2 (b) considered to correspond to a frequency dip f _c. The frequency dip _{f c} is, SAW grade LN substrate 50 on the same side as the substrate surface of the upper surface (the traveling-wave electrode is formed z- cut LN substrate 80 of integrated LN substrate 80 'in FIG. 2 (b) If there is metal on the lower surface (the substrate surface opposite to the substrate surface on which the traveling wave electrode is formed), f _c = (c _0/2 ) · {(m _x / (n _x · L ^{_{x)) 2 + (m y}} / (n y · L y)) 2
+ ( _{M z} / (n _z · L _z )) ² } ^1/2 (2)
Further, if there is a metal on the upper surface of the integral LN substrate 80 ′ and there is a sufficiently deep void below the lower surface, f _c = (c _0/2 ) · {(m _x / (n _x · L _{x ^{)) 2 + (m y /}} (n y · L y)) 2
+ (( _{M z} −1/2) / ( _{nz z} L _z )) ² } ^1/2 (3)
It becomes.

Here, m _{x, m y,} and m _z are natural numbers representing the order of the resonance, The problem is the resonant frequency of the lowest order (i.e., 1). 2 or more _m x, _{m y,} and _{m z} corresponds to the second- or higher-order modes. n _{x, n y,} and _{n z} is the equivalent refractive index of the high frequency electric signals in each direction of x, y, and z, in accordance with Patent Document 1 _{Here, n x = n y = 6.59} , n z = Set to 5.36.

Further, as can be seen from the comparison between the expressions (2) and (3), in the width direction (x direction) of the z-cut LN substrate 80, there is a metal on one end face, and the end face on the opposite side has a wide space. If you are in contact, the _{m x} may be replaced by _m x -1/2.

Considering the form of the formula in which the lateral width L _x , length L _y , and thickness L _z of the integral LN substrate 80 ′ are used in the formulas (2) and (3), 2 is not the product of the longest length in the cross section (usually the length of the diagonal line) and the equivalent refractive index of the high-frequency electrical signal in that direction. The key point is the product of the length of the shorter side and the equivalent refractive index of the high-frequency electrical signal in that direction. In the present invention, “the length of the side that is the shortest in the cross section” is the thickness L _z of the integral LN substrate 80 ′.

In the present invention, first, in the equation for determining the frequency dips f _c, paying attention to the lengths of the sides of the cross section is in a format that is inversely proportional to the square, the length of the longest side in the transverse plane by intentionally long, extremely small influence on the frequency dips f _c. Then, the frequency determined by the product of by reducing the thickness L _z is the length of the shortest side in the transverse plane intentionally length of shortest side and the equivalent refractive index of the high frequency electric signal in that direction exactly enough frequency region higher than the band to use dip f _c in optical transmission, reliable, is easily shifted. In other words, so in order to shift the frequency dips f _c to a high frequency range, among the cross-sectional dimension, to adopt a structure to reduce the thickness L _z.

The above will be described specifically using equations. In the equations (2) and (3),
L _z << L _y , L _x (4)
Assuming that (2) and (3) are respectively f _c = (c _0/2 ) · m _z / (n _z · L _z ) (5)
_{f c = (c 0/2} ) · (m z -1/2) / (n z · L z) (6)
And can be expressed, the length of the shortest side of the integral of the LN substrate 80 '(here, LN thickness L _z of the _substrate) can be determined frequency dips f _c is the.

Here, the conditions under which it can be assumed that equation (4) holds are considered. Since the length L _y of the integral LN substrate 80 ′ is sufficiently long, L _z << L _y naturally holds. Therefore, next, consider the remaining relationship between L _z and L _x . Since L _x is included in the frequency dip f _c in a form that is inversely proportional to the square of L _x , assuming that L _x is 1.4 or more times larger than L _z , L _x with respect to frequency dip f _c influence of _x becomes sufficiently small as about 1/2 of the effect of _{L z,} it can be considered that _L z << L _x holds. Also, since it may be considered an integral LN substrate 80 'is equivalent to the width L _x is wide, the advantage that sufficient mechanical strength of the optical modulator can be secured occurs.

  In the above description, the thickness of the z-cut LN substrate 80 and the thickness of the SAW grade LN substrate 50 which is a separate substrate are the same, but they may be different. In this case, strictly speaking, the conditions imposed as the present invention are preferably imposed on the thicker z-cut LN substrate 80 or the thicker SAW grade LN substrate 50. Since the shape is also related, it is assumed here that either the thickness of the z-cut LN substrate 80 or the thickness of the SAW grade LN substrate 50 may be satisfied. And this can be said for all embodiments of the invention.

  In order to propagate a high-frequency electrical signal from an external circuit (not shown) in the traveling wave electrode 4, a portion where a core wire of a connector (not shown) is connected is referred to as a feed-through portion. There may be a gap. In addition, this space | gap can be formed by providing a notch in the fixing | fixed part 20 of the housing | casing 15 shown in FIG. As described above, when there is a gap of a finite depth below the integral LN substrate 80 ′, the relative dielectric constant of the integral LN substrate 80 ′ is different from the equation (3) or (6). Since the ratio (its square root can be approximated to the equivalent refractive index) is high, the influence of the integrated LN substrate 80 'is larger than the gap. Therefore, the outline of the discussion here focusing on the integrated LN substrate 80 'is the same.

  In the case where there is a gap of a finite depth below the integral LN substrate 80 ′, in order to design more accurately, the resonance frequency determined from the thickness of the integral LN substrate 80 ′ and the thickness of the void is a high-frequency electric frequency. The gap thickness is set in accordance with the thickness of the integral LN substrate 80 ′ so as to be higher than the signal. In the voids, there is a medium (low dielectric constant layer) having the smallest relative dielectric constant or a smaller dielectric constant than the integrated LN substrate 80 ′, such as the z-cut LN substrate 80 or the SAW grade LN substrate 50. Filled.

As described above, with respect to the frequency dips f _c, the equivalent of the present invention is the second longest length in the "cross section of the prior art (usually diagonal) high-frequency electric signal in the direction shown as patent document 2 Instead of the product of refractive index, the opposite is the product of the length of the shortest side in the cross section and the equivalent refractive index of the high-frequency electrical signal in that direction. The idea is completely opposite.

Specifically, according to the second prior art, in order to shift the frequency dips f _c the higher frequency region will do unless narrowing the width of the integral of the LN substrate 80 '(L _x) but the present invention widely to several mm the width (L _x) of the integral of the LN substrate 80 'Conversely, and a frequency dip f _c than reducing the thickness (L _z) accurate, reliable, easily, The structure is made higher than the optical modulation frequency corresponding to the transmission rate used in actual optical transmission.

  Next, further discussion will be made on “the product of the length of the side that is the shortest in the cross section and the equivalent refractive index of the high-frequency electrical signal in that direction”.

In general, the frequency dips f _c is not sharp with respect to frequency (in other words, not high Q value of the resonance of a dielectric resonator) with a certain frequency band so, when considering the optical transmission of 10Gbit / s is to be set to approximately 13GHz about 30% higher if possible as a frequency dip f _c, the here, the 10GHz to be minimum secured shortest side of the length equivalent refractive high-frequency electric signal in that direction in the "cross-section For a large value of the product of the rate (n _z · L _{z in the} equations (5) and (6)), the value is 15 mm.

On the other hand, with respect to the smaller value of n _z · L _z , there is no problem if the thickness L _z of the z-cut LN substrate 1 is about 0.3 mm in the process that we are currently performing, up to about 0.2 mm. If so, the yield can be secured. Although the equivalent refractive index _nz in the resonance in the thickness direction has been discussed as 5.36, it can be considered up to about 2. Therefore, as a result, “the length of the shortest side in the cross section and the high-frequency electric signal in that direction” A small value of the product of the equivalent refractive index (n _z · L _{z in the} formulas (5) and (6)) is about 0.4 mm.

Actually, “the product of the length of the side that is the shortest in the cross section and the equivalent refractive index of the high-frequency electrical signal in that direction (n _z · L _{z in the} formulas (5) and (6))” is large. For the value 15 mm, the equivalent refractive index _nz of the substrate resonance in the thickness direction may be about 2, and even in the case of a small value of 0.4 mm, the equivalent refractive index _nz of the substrate resonance in the thickness direction. Although the reason is that the resonant frequency of the substrate is sufficiently high even if it is 5.36, the thickness of the z-cut LN substrate 1 becomes too thin in manufacturing an actual LN optical modulator module. From the viewpoint of

  An LN optical modulator was actually manufactured. First, in FIGS. 1 and 2A, the lateral width of the optical grade z-cut LN substrate 80 on which the LN optical modulator is manufactured is set to 1 mm, and the pedestal 20 of the metal casing 15 in the first prior art of FIG. The z-cut LN substrate 80 was bonded to a base of a metal casing (not shown) so that the LN optical modulator was fixed with a conductive adhesive or the like. Further, the lateral width of the SAW grade LN substrate 50 was set to 5 mm, and this was also fixed to the pedestal (not shown) with an adhesive. Further, an adhesive was applied in advance to the side surface of the substrate corresponding to the contact surface 51 of the SAW grade LN substrate 50, and the z-cut LN substrate 80 and the SAW grade LN substrate 50 were fixed to each other. Even if this adhesive protrudes slightly above the contact surface 51 between the z-cut LN substrate 80 and the SAW grade LN substrate 50, there is no problem. The optical grade z-cut LN substrate 80 and the SAW grade LN substrate 50 on which the LN optical modulator is manufactured are fixed to each other in advance on the respective sides, and are fixed to the base 20 of the metal housing 15. Also good.

And in order to confirm the principle of the present invention, the lateral width L _x of the integral LN substrate 80 ′ is 6 mm (that is, SAW grade LN which cannot allow even 10 Gbit / s light modulation in the second prior art. The width of the substrate 50 is 5 mm). The substrate thickness L _z was 0.3 mm, and the substrate length L _y was 50 mm.

  Here, for convenience of explanation, the description has been made assuming that the inside 20 of the metal casing 15 in FIG. 20 is a pedestal, but the inside 20 of the metal casing 15 is not a pedestal, and a pedestal is provided separately. Needless to say, the z-cut LN substrate 80 and the SAW grade LN substrate 50 may be fixed to the pedestal, while being fixed to the metal casing 15. Further, different pedestals may be provided on the z-cut LN substrate 80 and the SAW grade LN substrate 50, respectively, and they may be fixed in a metal casing.

  Thus, by reducing the lateral width of the z-cut LN substrate 80 to 1 mm, for example, the wafer of the optical grade z-cut LN substrate as compared with the first prior art in which the lateral width is 2 to 5 mm. The total number of LN optical modulator chips that can be taken from the present invention can be greatly increased to 2 to 5 times the 10 to 20 chips in the first prior art, that is, 40 to 50 chips. Further, by making the width of the z-cut LN substrate 80 narrower than 1 mm, more LN optical modulator chips can be obtained.

Next, an experiment of observation was performed on the frequency dip. Although the frequency dips _{f c} predicted from equation (1) which is a substrate of the second prior art is 5 GHz, the frequency let alone 20GHz or more, were also observed frequency dips _{f c} In a further 40 GHz. Therefore, without the very small elements as in the second prior art, it was confirmed that it is possible to shift the frequency dips f _c of the modulation index to a frequency higher than 40 GHz.

The major feature of idea here is independent of the width of the frequency dips f _c is z- cut LN substrate 80, in that it depends only on the thickness of the z- cut LN substrate 80 and SAW grade LN substrate 50 There, unlike the second prior art, it can design a total number and frequency dips f _c of the chip that can be taken from one wafer independently.

  Furthermore, as a result of conducting an experiment in which the z-cut LN substrate 80 is peeled off from a metal casing (not shown), the same strength as that of the first conventional LN optical modulator made of the z-cut LN substrate 1 having the same width is obtained. The vibration and impact tests after assembling the LN optical modulator module were cleared without any problems. Thus, this embodiment has a much stronger mechanical strength than the second prior art LN optical modulator in which the lateral width of the z-cut LN substrate 1 is narrow.

As described above, by reducing the width of the LN optical modulator chip, many narrow LN optical modulator chips can be obtained from an expensive optical grade wafer, thereby reducing the cost of the optical modulator. Reduced from the viewpoint of material costs and labor costs in the process that actually occupies a large percentage, and also dramatically increased productivity, and the LN optical modulator chip and the extremely inexpensive SAW grade substrate are mechanically integrated as it regarded as LN substrate maintaining the mechanical strength without increasing the cost, by further sufficiently wide width as well as reduce the thickness of the LN substrate of the integrated, accurate harmful frequency dip f _c for optical transmission There is a great advantage that the frequency can be shifted to a higher frequency than the frequency range of use reliably and easily.

  Furthermore, even when a SAW grade LN substrate is used as the z-cut LN substrate for forming the optical waveguide, the cost related to the labor cost can be greatly reduced by applying the present invention.

  In the above description, FIG. 20 is taken as an example, and z-cut LN substrate 80 and SAW grade LN substrate 50 are fixed using 20 inside metal housing 15 as a pedestal, but this is for convenience of description. Of course, it goes without saying that a separate pedestal may be provided without using the inside 20 of the metal casing 15 as a pedestal, and after attaching the z-cut LN substrate 80 and the SAW grade LN substrate 50 to separate pedestals. Needless to say, it may be fixed inside the metal casing 15.

  Note that it is extremely advantageous from the viewpoint of mechanical strength that the z-cut LN substrate 80 and the separate substrate 50 are fixed to each other on the side surface, but even if they are not fixed to each other on the side surface, The mechanical strength can be increased by the close contact or contact. Furthermore, a separate substrate can be used as a guide when the z-cut LN substrate is fixed to a base (not shown), which is advantageous. Further, the z-cut LN substrate 80 and the separate substrate 50 may be fixed to each other in advance and fixed to a pedestal (not shown).

  In the above discussion, as a separate substrate, a SAW grade LN substrate 50, that is, a material made of the same dielectric material as the LN optical modulator chip is used. The separate substrate may be an LN substrate with different plane orientations such as x-cut and y-cut, or may be another dielectric substrate such as a lithium tantalate substrate, an alumina substrate, an aluminum nitride substrate, or a sapphire substrate. good. Here, since the dielectric substrate means not a metal, in this specification, the term “dielectric substrate” includes a semiconductor substrate such as a GaAs substrate or an InP substrate in addition to a normal dielectric substrate. These can be said for all embodiments of the present invention.

  When a z-cut LN substrate is used as a separate substrate, that is, the SAW grade LN substrate 50, the z-cut LN substrate 80 and the SAW grade LN substrate 50, which are LN optical modulator chips, and the -z plane and + z Although it is desirable that the directions of the surfaces be the same, it is possible to neutralize the charge due to the pyroelectric effect by reversing the directions.

  Now, consider impedance conversion, which is another important pillar of the concept of the present invention. As described above, FIG. 1 shows only the optical grade z-cut LN substrate 80 and the SAW grade LN substrate 50. In an actual LN optical modulator, a buffer layer and a traveling wave electrode are formed on these surfaces. FIG. 3 shows an actual top view in which traveling wave electrodes are formed. However, the buffer layer is omitted for simplicity of explanation. Also, the buffer layer is omitted in the following discussion.

  Here, reference numerals 60a, 60b, 60c, 60d and 60e are CPW electrodes formed on the SAW grade LN substrate 50 and have a function of impedance conversion which is a key of the present invention. 60a and 60d are center conductors, and 60b, 60c, and 60e are ground conductors.

  Here, VIII connects a core wire of a connector (not shown) for applying a high-frequency electric signal from an external signal source (not shown) to an electrode composed of 60a, 60b, and 60c formed on the SAW grade LN substrate 50. This is an input feedthrough section on the SAW grade LN substrate 50. Here, a gold ribbon or a gold wire may be used instead of the connector core wire, but these are collectively referred to as a connector core wire here. IX is an impedance conversion unit on the SAW grade LN substrate 50, and I and VII are an input feedthrough unit and an impedance conversion unit formed on the optical grade LN substrate 80, respectively. X is a connection portion (or output side connection portion) between the output feedthrough portion V and the interaction portion III on the optical grade LN substrate 80, and XI is an output feedthrough portion on the optical grade LN substrate 80. It is. Although it is possible to provide all the electrodes on the SAW grade LN substrate 50 with the impedance conversion function, even the input feedthrough I on the optical grade LN substrate 80 can be provided with the impedance conversion function. It is better to use the electrodes on the optical grade LN substrate 80 as well. XII is an output feed-through portion connected to the terminating resistor. In the central conductor of the input feedthrough portion VIII, a portion where the core wire of the connector is connected is called a power feeding portion, and a portion where a high frequency electric signal is taken out in the central conductor of the output feedthrough portion XII is called an output portion.

  Further, since the input side of the high-frequency electrical signal is important in terms of suppressing electrical reflection, gold ribbons 70a, 70b, and 70c that are wider than gold wires are used here. This gold ribbon is more convenient in the present invention because it can change the characteristic impedance more easily than the gold wire. Since 80a, 80b, 80c, and 80d correspond to the output side of the high-frequency electric signal, they are not as important as the input side, and are gold wires.

FIG. 4 shows an equivalent circuit diagram of the first embodiment of the present invention shown in FIG. The characteristic impedance of the input feedthrough section VIII (line 90) on the SAW grade LN substrate 50 is Z ₉₀ , the characteristic impedance of the impedance conversion section IX (line 91) is Z ₉₁ , and the characteristic impedance on the optical grade LN substrate 80 is The characteristic impedance of the input feedthrough part I (line 92) is Z ₉₂ , the characteristic impedance of the impedance converter VII (line 93) is Z ₉₃ , the characteristic impedance of the interaction part III (line 39) is Z ₃₉ , and the output side connection The characteristic impedance of the part X (line 94) is Z ₉₄ , the characteristic impedance of the output feedthrough part XI (line 95) is Z ₉₅ , and the characteristic of the output feedthrough part XII (line 96) on the SAW grade LN substrate 50 impedance is _{Z 96.} In FIG. 4, Z _in ″ is an input impedance when the z-cut LN optical modulator of this embodiment is viewed from the external signal source 5 and the load resistor 6 of the external signal source 5.

As can be seen from FIG. 3, in this embodiment, the edges of the center conductor and the ground conductor in the output feedthrough portion XI on the optical grade LN substrate 80 have the characteristic impedance of the output feedthrough portion XI as the interaction portion III. The non-linear shape is used in order to match the characteristic impedance with a fairly accurate value, but this is applicable to all embodiments of the present invention. However, this means that a non-linear shape is suitable, and it is not the most important matter that can exhibit the effects of the present invention. Although here is discussed as being equal to the electrical resistance Z _L also interaction portion III of the characteristic impedance Z ₃₉ of the electrical terminations 42, the present invention is naturally not limited to this.

The equivalent circuit of FIG. 4 can be written as shown in FIG. Here, III ″ is a line (or combined portion) 98 obtained by combining the interaction portion III and the subsequent portion, and the characteristic impedance Z ₉₈ has a nonlinear shape on the output side of the high-frequency electric signal as described above. substantially the same as the characteristic impedance _{Z 39} of the interaction portion III.

Also in the present embodiment, in order to reduce the drive voltage, the characteristic impedance Z ₃₉ of the interaction part III (or the characteristic impedance Z of the interaction part III can be more accurately used than the second conventional technique by using a non-linear shape. The characteristic impedances of ₃₉ and the subsequent lines are made to coincide with each other, and the characteristic impedance Z ₉₈ ) of those gold ribbons III ″ is lowered. Again, it is about 30Ω as an example. The object of the present invention is to improve the electrical reflection characteristics, that is, to provide an external signal while maintaining the low driving voltage as an LN optical modulator by utilizing the SAW grade LN substrate 50 and performing impedance conversion. Emphasis is placed on improving the jitter of optical pulses by suppressing the electrical reflections returning to the source.

The first embodiment of the present invention has the impedance conversion unit VII on the optical grade LN substrate 80, but the impedance conversion unit IX is also provided on a separate substrate 50 made of a SAW grade LN substrate. Is formed. In this embodiment, the characteristic impedances Z ₉₁ and Z ₉₃ of the impedance conversion units IX and VII are the characteristic impedance Z ₉₀ of the input feedthrough unit VIII and the characteristic impedance Z ₃₉ of the interaction unit III (or the input feedthrough unit). It is not necessary to be the geometric mean of the characteristic impedance Z _{90 of} VIII and the characteristic impedance Z ₉₈ ) of the gold ribbon III ″. Further, the sum of the lengths of the impedance converters IX and VII need not be λ / 4, where λ is the in-tube wavelength of the bit rate of the high-frequency electric signal.

Characteristics when the feed-through portion VIII and the impedance conversion portion IX for input on the separate substrate 50 and the input feed-through portion I and the impedance conversion portion VII on the optical grade LN substrate 80 are collectively considered as the impedance conversion portion impedance connector characteristic impedance (or an external signal source 5 of the characteristic impedance Rg) neither in the geometric mean of the interaction portion III of the characteristic impedance Z _39, connector characteristic impedance (or characteristic impedance Rg of the external signal source 5) And the geometric mean of the characteristic impedance Z ₉₈ of the gold ribbon III ″.

  Of course, even in this case, the input feedthrough portion VIII and the impedance conversion portion IX on the separate substrate 50, the input feedthrough portion I on the optical grade LN substrate 80, and the impedance conversion portion VII are collectively referred to as an impedance conversion portion. The total length when considered is not required to be λ / 4, where λ is the in-tube wavelength at the frequency of the bit rate of the high frequency electrical signal (for example, 10 GHz when the bit rate of the high frequency electrical signal is 10 Gbit / s). .

Now consider in more detail. In the first embodiment of the present invention, the impedance converter VII is provided on the optical grade LN substrate 80, but the impedance converter IX is also provided on the separate substrate 50 made of the SAW grade LN substrate. Is formed. However, the characteristic impedance _{Z 91} and _{Z 93} of the impedance transformation portion IX and VII in the present embodiment, the characteristic impedance _{Z 90} and the interaction portion III of the characteristic impedance _{Z 39} of the input feed-through section VIII (line 90) (or gold It need not be a geometric mean with the characteristic impedance Z ₉₈ ) of the ribbon III ″.

On the contrary, as described above, in the present invention, since the input feedthrough portion can also function as a part of the impedance conversion portion, the characteristic impedance of the input feedthrough portion VIII formed on the separate substrate 50 is obtained. characteristic impedance of Z ₉₀ and impedance transformation portion IX _{Z 91,} equally to all the optical grade LN substrate 80 characteristic impedance _{Z 92} of the input feed-through portion I formed on the characteristic impedance _{Z 93} of the impedance conversion section VII (eg The characteristic impedance of the connector (not shown) and the load resistance 6 (Rg) of the external signal source 5 are equal to the geometric mean of the characteristic impedance Z ₃₉ of the interaction part III and the characteristic impedance Z ₉₈ of the gold ribbon III ″). (Of course, everything is different from the geometric mean. It Umate may be any one or more of the geometric mean).

This is because a connector core wire (not shown) may actually be connected to the input feed-through portion VIII (hereinafter, although not shown, a gold ribbon, a gold wire, etc. may be connected. The core wire of this connector (not shown) is a large (or thick) conductor with a diameter of 100 to 300 μm, so that the characteristic impedance characteristic of the input feed-through portion VIII can be obtained by connecting it. the effect of lowering significantly the impedance Z _90. The input feedthrough I formed on the optical grade LN substrate 80 is connected to gold ribbons 70a, 70b and 70c from the impedance conversion unit IX formed on a separate substrate 50. The gold ribbon significantly changes (partially reduces) the characteristic impedance of the connecting portion (for example, the characteristic impedance Z _{92 of the} input feedthrough portion I).

Therefore, the characteristic impedance Z _{90 of} the input feedthrough section VIII formed on the separate substrate 50, the characteristic impedance Z ₉₁ of the impedance conversion section IX, and the input feedthrough section I formed on the optical grade LN substrate 80 The combined impedance obtained by combining all of the characteristic impedance Z ₉₂ and the characteristic impedance Z ₉₃ of the impedance conversion unit VII is the characteristic impedance of the connector (not shown), the load resistance 6 (Rg) of the external signal source 5 and the characteristic of the interaction unit III. This is different (lower) from the geometric mean of the impedance Z _{39 and} the characteristic impedance Z ₉₈ of the gold ribbon III ″.

  That is, the entire portion having the function of impedance conversion (that is, the input feedthrough portion VIII formed on the separate substrate 50, the impedance conversion portion IX, and the input feedthrough portion I formed on the optical grade LN substrate 80). , And the impedance conversion unit VII), the characteristic impedance of these combined units is not the above geometric average value but a lower value.

  In other words, the input feedthrough portion VIII formed on the separate substrate 50 and the input feedthrough portion I formed on the optical grade LN substrate 80 are used only at a specific frequency as in the fourth prior art. Resonant impedance conversion does not occur (in other words, by destroying the resonance condition), and has the effect of shifting the characteristic impedance of the entire part having the function of impedance conversion from the above-mentioned geometric mean, generating a residual reflection and a wide frequency band Impedance conversion for realizing impedance conversion in

  As described above, in this embodiment, the characteristic impedance of the input feedthrough portions VIII and I is intentionally reduced by connecting the core wire of a connector (not shown) or by connecting the gold ribbons 70a, 70b, and 70c. , The entire part having the function of impedance conversion (that is, the input feedthrough part VIII formed on the separate substrate 50, the impedance conversion part IX, the input feedthrough part I formed on the optical grade LN substrate 80, In addition, by making the characteristic impedance of the impedance conversion unit VII) not to be the above-described geometric mean, residual reflection is generated, and as a result, the frequency that can realize the function of impedance conversion is widened.

Further, the characteristic impedance Z ₉₀ of the input feedthrough section VIII formed on the separate substrate 50 is set to about 50Ω, which is the same as the connector (or the characteristic impedance Rg of the external signal source), and the characteristic impedance Z ₉₁ of the impedance conversion section IX. The characteristic impedance Z _{92 of} the input feedthrough I formed on the optical grade LN substrate 80 and the characteristic impedance Z ₉₃ of the impedance converter VII are combined with the characteristic impedance Z ₉₀ of the input feedthrough VIII. Consider a case where a geometrical average of the characteristic impedance Z ₃₉ of III (or the characteristic impedance Z ₉₈ of the gold ribbon III ″ of the interaction part III and the output feed-through part XI) is used.

As described so far, the characteristic impedance Z ₉₀ of the input feedthrough portion VIII to which the connector core wire (not shown) is connected is actually lowered significantly due to the connector core wire which is a large (thick) conductor. Therefore, the characteristic impedance Z _{90 of} the input feedthrough section VIII formed on the separate substrate 50, the characteristic impedance Z ₉₁ of the impedance conversion section IX, and the input feedthrough section I formed on the optical grade LN substrate 80 The characteristic impedance Z ₉₂ and the characteristic impedance Z ₉₃ of the impedance converter VII are combined into an input feed formed on a separate substrate 50 that is actually considerably low due to the core wire of the connected connector. Characteristic impedance Z ₉₀ of the through portion VIII and the interaction portion III (or the characteristic impedance Z _{39 of the} interaction portion III or the characteristic impedance Z ₉₈ of the gold ribbon III ″ between the interaction portion III and the output feed-through portion XI, etc. ) And the geometric mean It must be higher than this geometric average.

In this way, the characteristic impedance Z _{90 of} the input feedthrough section VIII formed on the separate substrate 50, the impedance conversion section IX, the input feedthrough section I formed on the optical grade LN substrate 80, and the impedance The characteristic impedance finally synthesized for the conversion unit VII is combined with the characteristic impedance Z _{90 of} the input feedthrough unit VIII connected to the connector core wire and the characteristic impedance Z _{39 of the} interaction unit III (or the interaction unit III and the output feed). As a result of shifting from the geometric mean of the characteristic impedance Z ₉₈ ) of the gold ribbon III ″ with the through portion V or the like, residual reflection occurs, and as a result, the frequency of the impedance conversion function can be widened. That is, the input feedthrough section VIII formed on the separate substrate 50 operates as a part of the impedance conversion section for widening the frequency of impedance conversion.

  In the present invention, the input feedthrough portions VIII and VIII and the impedance conversion portions IX and VII are referred to as being distinguished from each other. However, as described above, the input feedthrough portions VIII and I also have an impedance conversion function. As a whole, the input converters VIII and I and the impedance converters IX and VII constitute an impedance converter.

  In order to improve the electrical reflection characteristics as much as necessary while breaking the resonance condition in which the electrical reflection characteristics are improved only at a specific frequency and leaving the residual reflection at a wide frequency, the entire portion described above is used. It is set so that the geometric mean according to equation (2) does not hold.

That is, the characteristic impedances Z ₉₀ and Z ₉₂ of the input feedthrough portions VIII and I without the connector core wire and gold ribbon being connected are equal to the characteristic impedances Z ₉₁ and Z ₉₃ of the impedance conversion portions IX and VII. Even in the case of equality or not equality, the characteristic impedances Z ₉₁ and Z ₉₃ of the impedance converters IX and VII may be shifted from the above geometric mean from the beginning, or for input by connecting a connector core wire or a gold ribbon. Needless to say, as a result of lowering the characteristic impedances Z ₉₀ and Z ₉₂ of the feedthrough portions VIII and I, the characteristic impedance of the entire part having the function of impedance conversion may be shifted from the above geometric mean.

  In the present invention, in this way, by shifting the characteristic impedance of the portion having the function of impedance conversion from the geometric mean, the frequency capable of realizing the impedance conversion function is widened while causing residual reflection. These ideas apply to all embodiments of the present invention.

  Further, in the present invention, the sum of the lengths of the impedance converters IX and VII represents the in-tube wavelength at the bit rate of the high-frequency electric signal (for example, the in-tube wavelength at 10 GHz if the bit rate of the high-frequency electric signal is 10 Gbit / s). As in the fourth prior art, it is not necessary to set λ / 4. In other words, in the present invention, the input feedthrough portions VIII and I are also part of the impedance conversion portion. Therefore, if the length of the impedance conversion portion is increased, the input feedthrough portion formed on the separate substrate 50 is used. This is the sum of the electrical lengths of the part VIII, the impedance converter IX, the input feedthrough part I formed on the optical grade LN substrate 80, and the impedance converter VII.

  However, since the above-described geometric mean concept is not used in the first place, it is not necessary to set the sum of the electrical lengths as a whole to λ / 4, where λ is the in-tube wavelength of the bit rate of the high-frequency electrical signal. It seems that the sum of the electrical lengths is preferably longer than λ / 4.

  As described above, since the separate substrate 50 is provided and a portion having the function of impedance conversion is formed thereon, the lateral width of the optical grade LN substrate 80 can be reduced. As a result, the number of LN optical modulators that can be taken from one wafer of an optical grade LN substrate can be extremely increased, which has a significant cost reduction effect.

In this embodiment, in order to reduce the drive voltage, the characteristic impedance Z _{39 of} the interaction unit III (or the lines 94, 95, and 96 after the interaction unit III have substantially the same characteristic impedance as that of the interaction unit III. Therefore, the characteristic impedance Z ₉₈ of the gold ribbon III ″ between them and the interaction part III is the same as the characteristic impedance Z ₃₉ of the interaction part III), which is reduced to about 30Ω.

The power reflectivity S _{11 in} this case (strictly speaking, the envelope of the power reflectivity S ₁₁ ) is shown in FIG. As can be seen from the figure, the power reflectivity S ₁₁ has a residual reflection of about −12 dB in the vicinity of about 3 GHz in the frequency band of 20% to 30% of the bit rate of 11.3 Gbit / s. Thus, in the optical modulator of the present invention, at least one point within the frequency band from 20% corresponding to 30% of the bit rate, the power reflectivity _{S 11} is between -10dB of -15 dB. Then, the power reflectivity _{S 11} to the frequency _{F C} of the frequency band corresponding to from 40% to 70% of the frequency of the bit rate is the minimum value equal to or less than -15 dB.

  In order to make the lines 94 and 95 after the interaction part III and the interaction part III have substantially the same characteristic impedance, the center conductor in the output feedthrough part XI has a non-linear taper shape as shown in FIG. The shape of the edge of the center conductor and the ground conductor was also non-linear. Such a nonlinear shape of the output feedthrough portion XI is applicable to all the embodiments of the present invention.

In this embodiment, (2) since the type or (3) does not consist, the electrical power reflectance S ₁₁ at a particular frequency, such as the fourth prior art shown in FIG. 43 is very well There is no. However, as shown in FIG. 6 for the transmission speed of 11.3 Gbit / s, the electrical power reflectance S ₁₁ is −15 dB or less in the region of 5.6 GHz which is the fundamental frequency of the pulse of 11.3 Gbit / s, In addition, the power reflectivity characteristic S ₁₁ is sufficient in the frequency band of a certain width (strictly speaking, S ₁₁ in FIG. 6 is an envelope of the power reflectivity, and here it is referred to as the power reflectivity. Calling). In this case, the envelope of the power reflectance S ₁₁ becomes a valley (i.e., first derivative of the envelope of the power reflectance S ₁₁ is zero, giving the minimum value second derivative is positive) frequency Fc is important Become.

FIG. 7 shows an optical pulse when the minimum point of the power reflectivity S ₁₁ in FIG. 6 (or the valley of the envelope of the power reflectivity S ₁₁ ) Fc is used as a variable for a transmission speed of 11.3 Gbit / s. It shows jitter having. As can be seen from the figure, if the absolute value of the power reflectivity S ₁₁ is −15 dB or less and the minimum point Fc is in the frequency band between 40% and 70% of the transmission rate (11.3 Gbit / s), jitter Is a value of 2 ps or less with little increase in errors during transmission. FIG. 8 shows the frequency dependence of the light modulation index. As can be seen from the figure, the optical modulation band can be improved by applying the present invention.

In this way, the drive voltage is kept low (as a result, the power reflectivity S ₁₁ becomes worse between −10 dB and −15 dB in the frequency range of about 2 GHz to about 3 GHz. In the case of the format, the frequency spectrum in this region is small), the jitter of the optical pulse can be reduced, or the optical modulation band can be widened, so that high performance as an optical modulator can be achieved.

  The values of 20% to 30% of the bit rate, or 40% to 70% of the bit rate are other bit rates other than 10 Gbit / s and 11.3 Gbit / s, for example, 12.5 Gbit / s, 25 Gbit / s. Even s, 43 Gbit / s, 50 Gbit / s, and 100 Gbit / s are established.

  The value of 20% to 30% is about 2 GHz to 4 GHz when the bit rate is 10 Gbit / s to 12.5 Gbit / s, and about 5 GHz to 8 GHz when the bit rate is 25 Gbit / s. In the case of 43 Gbit / s, it corresponds from about 9 GHz to 13 GHz, in the case where the bit rate is 54 Gbit / s, it corresponds from about 11 GHz to 16 GHz, and in the case where the bit rate is 110 Gbit / s, it corresponds from about 22 GHz to 33 GHz.

  The value of 40% to 70% is about 4 GHz to 9 GHz when the bit rate is 10 Gbit / s to 12.5 Gbit / s, and about 10 GHz to 18 GHz when the bit rate is 25 Gbit / s. In the case of 43 Gbit / s, it corresponds from about 17 GHz to 30 GHz, in the case of a bit rate of 54 Gbit / s, from about 22 GHz to 38 GHz, and in the case of a bit rate of 110 Gbit / s, it corresponds from about 44 GHz to 77 GHz.

Note that in the present invention it is important that the minimum point Fc of the power reflectance _{S 11} is present between 70% to 40% of the transmission rate (for example, 10Gbit / s and 11.3 Gbit / s, etc.), which The first prior art as well as the second prior art cannot be realized.

That is, in order to realize this, the input feedthrough portion VIII formed on the separate substrate 50 by connecting the core wire of the connector or the gold ribbon, or the input feedthrough portion I formed on the optical grade LN substrate 80. It is preferable to use a change in the characteristic impedances Z ₉₀ and Z ₉₂ (generally lower but may be higher).

Considering not only the impedance conversion units IX and VII but also the input feedthrough units VIII and VIII as impedance conversion units, the characteristic impedance of the combined portion of the input feedthrough units VIII and I and the impedance conversion units IX and VII A load resistance Rg of the external signal source 5 (also called a characteristic impedance, an output impedance, an impedance, which is generally desirable to be equal to a characteristic impedance of a connector not shown), a characteristic impedance Z ₃₉ of the interaction unit III, It is effective to make it different from the geometric mean of the characteristic impedance Z ₉₈ of the gold ribbon III ″ of the interaction part III and the output feedthrough part.

The impedance converting section IX in the present invention, VII (wherein input feed-through section VIII, also good as the impedance conversion section I) was the one-stage configuration, minimum point Fc of the power reflectance S ₁₁ in the present invention Is set between 40% and 70% of the transmission speed (for example, 10 Gbit / s or 11.3 Gbit / s), and the power reflectivity S _{11 at the} minimum value F _C is set to be −15 dB or less. Therefore, it is a feature that the modulation band and especially the jitter of the optical pulse are reduced. Therefore, it is needless to say that a multi-stage configuration of two stages or three stages or more may be used instead of the one-stage configuration. It goes without saying that this multi-stage configuration may be configured on an optical grade LN substrate, a separate SAW grade substrate, or may be configured separately on both.

  In addition, the condition of having a minimum point Fc that is −10 dB to −15 dB at a frequency of at least one point of 20% to 30% of the bit rate frequency, or −15 dB or less within a frequency of 40% to 70% is as follows: Needless to say, this is preferable as an embodiment of the invention, but is not an essential requirement of the present invention.

The characteristic impedance of the connector is generally 50 [Omega, because the characteristic impedance _{Z 98} of the characteristic impedance of the interaction portion III _{Z 3} gold ribbons III '' is a 30 [Omega, these geometric mean becomes 38.7Omu. In order to exhibit the effect of the present invention that the frequency capable of exhibiting the impedance conversion function is widened while causing residual reflection, the values of the characteristic impedances Z ₉₁ and Z ₉₃ of the impedance conversion units IX and VII are set to the geometric average. It is effective to make it different from the value. Therefore, the input feedthrough part I and the impedance conversion parts IX and VII are, for example, 44Ω higher by about 5Ω (or 33Ω lower by about 5Ω may be used) in the sense that it is different from the geometric mean.

  However, as described above, a core wire of a connector (not shown) is fixed to the input feedthrough portion VIII formed on the separate substrate 50, and the core wire of the connector (not shown) has a diameter of about 100 to 300 μm. Consists of large (or thick) conductors. Therefore, the characteristic impedance of the input feedthrough portion VIII considering the core wire of the connector (not shown) is 29Ω (when the impedance conversion portion VII is 33Ω, the input feedthrough portion VIII considering the core wire of the connector not shown) The characteristic impedance is set to 25 Ω) and is much different from the above-mentioned geometric mean value. This also applies to the input feed-through portion I formed on the optical grade z-cut LN substrate 80 to which the gold ribbons 70a, 70b, and 70c are connected.

Here, the characteristic impedance Z _{90 of} the input feedthrough section VIII formed on the separate substrate 50, the characteristic impedance Z ₉₁ of the impedance conversion section IX, and the input feedthrough section I formed on the optical grade LN substrate 80. Consider the case where the characteristic impedance Z ₉₂ and the characteristic impedance Z ₉₃ of the impedance converter VII are 38.7Ω, which is the above-mentioned geometric average. The input feed-through section VIII for being connected to the core wire of the connector (not shown), the effective characteristic impedance _{Z 90} is the impedance conversion section IX, the characteristic impedance of _{VII Z 90, Z 93 (38.7Ω} ) Is actually much lower (around 10Ω).

As a result, the characteristic impedance Z _{90 of} the input feedthrough portion VIII (to be exact, it can be said that the effective characteristic impedance Z ₉₀ is lowered due to the core wire of the connector of the input feedthrough portion I), the impedance conversion portion IX Characteristic impedance Z ₉₁ , the characteristic impedance Z _{92 of} the input feed-through part I lowered due to the gold ribbons 70 a, 70 b, 70 c, and the characteristic impedance Z _{93 of the} whole part of the impedance conversion part VII of the impedance conversion part VII Is a value lower than the geometric mean (38.7Ω). Further, the effective characteristic impedance thus lowered may be lower than the characteristic impedance Z ₃₉ of the interaction part III (or the characteristic impedance Z ₉₈ of the synthesis part, here about 30Ω).

  As described above, the impedance composed of the input feedthrough section VIII, the impedance conversion section IX, the input feedthrough section I formed on the optical grade LN substrate 80, and the impedance conversion section VII formed on the separate substrate 50. By shifting the characteristic impedance of the entire conversion site from the geometric mean described above, it is possible to exhibit the effect of the present invention in that the frequency at which the impedance conversion function can be exhibited is broadened while residual reflection occurs.

And a part lower than the value of the characteristic impedance of the interaction part III (or the characteristic impedance Z ₉₈ of the gold ribbon III ″ including the interaction part III) between the connector (not shown) and the interaction part III. There may be, and conversely, it is convenient. This is because an important point in the present invention is that the minimum point Fc of the envelope of the power reflectivity S ₁₁ is the bit rate of the high-frequency electric signal (for example, 10 Gbit / s, 11.3 Gbit / s, 25 Gbit / s, 50 Gbit / s, 100 Gbit / s). s etc.) (for example, 10 GHz, 11.3 GHz, 25 GHz, 50 GHz, 100 GHz, etc.) between 40% and 70%, and the power reflectance S _{11 at the} minimum point Fc is −15 dB or less. This concept is not disclosed in the fourth prior art and cannot be realized by the fourth prior art. In other words, this is because it can be realized only under a condition where Expressions (2) and (3) do not hold.

  Furthermore, in the present invention, since the above-described geometric mean value is not used for the overall characteristic impedance composed of the input feedthrough portion and the impedance conversion portion formed on the separate substrate and the optical grade LN substrate, the residual reflection is completely eliminated. This is completely different from the idea of the fourth prior art that attempts to suppress the noise at a specific frequency, but at a specific frequency, it is not improved as much as the fourth prior art, but it is intentionally necessary to cause residual reflection in a wide frequency range and is necessary. It can be said that the residual reflection value is suppressed to a certain level.

[Second Embodiment and Third Embodiment]
FIG. 9 shows a perspective view of the second embodiment of the present invention. In this figure, traveling wave electrodes are omitted for the sake of simplicity. FIG. 10 is a cross-sectional view taken along the line DD ′ of FIG. 50 is a separate substrate made of a SAW grade LN substrate, and 52 is an adhesive layer such as an ultraviolet curable resin having a thickness G. Normally, even when the z-cut LN substrate 80 and the separate substrate 50 are not almost completely adhered, the thickness of the adhesive layer 52 is as thin as about 5 to 10 μm. This includes the case where the thickness of the agent layer 52 is increased to about 30 to 50 μm and further to 100 μm.

  In the present embodiment, the adhesive layer 52 is included in the entire gap between the z-cut LN substrate 80 and the separate substrate 50. However, the resonance of the substrate is a traveling wave (not shown) from the core wire of the connector (not shown). Since it is caused by a high-frequency electrical signal that leaks and radiates when moving to an electrode, it mainly occurs at an input feedthrough portion that connects a core wire of a connector (not shown) and a traveling wave electrode (not shown). Accordingly, DD ′ traverses the vicinity of an input feedthrough portion (not shown), and FIG. 10 may be considered as a cross-sectional view in the vicinity. However, the optical waveguide 3 is omitted.

  FIG. 11 shows a perspective view of the third embodiment of the present invention. A cross-sectional view taken along line EE ′ of this perspective view is shown in FIG. However, the optical waveguide 3 is omitted. Here, the separate substrate 50 composed of the z-cut LN substrate 80 and the SAW grade LN substrate is partially fixed by the adhesive layer 30. DD ′ crosses the vicinity of an input feedthrough portion (not shown) in the sixth embodiment, and FIG. 12 may be considered as a cross-sectional view in the vicinity. That is, the air layer 53 is interposed between the z-cut LN substrate 80 and the separate substrate 50 in the cross section EE ′.

  Also in the case of the third embodiment, the z-cut LN substrate 80 and the separate substrate 50 partially fixed to each other by the adhesive layer 52 are further fixed to a pedestal inside the housing (not shown). , The bonding area can be widened. Therefore, compared to the case where only the z-cut LN substrate 80 is fixed to the pedestal inside the housing (not shown), the z-cut LN substrate 80 can be firmly fixed to the pedestal inside the housing (not shown). Become.

  FIG. 13 shows a case where the thickness G of the adhesive layer 52 is a variable in the second embodiment of the present invention shown in FIGS. 9 and 10, and the z-cut LN substrate 80, the adhesive layer 52, and the separate layers The z-cut LN substrate 80 in the case where the resonance frequency of the structure composed of the substrate 50 is a solid line and the thickness G of the air layer 53 is a variable in the third embodiment of the present invention shown in FIGS. , The resonance frequency of the structure including the air layer 53 and the separate substrate 50 is indicated by a broken line. Note that both the z-cut LN substrate 80 and the separate substrate 50 have a thickness and a width of 1 mm. Here, optical transmission of 10 Gbit / s or 11.3 Gbit / s is assumed as a bit rate.

  As can be seen from the figure, the thickness G of the adhesive layer 52 and the air layer 53 in the present invention is G = 0, that is, approximately the resonance frequency in the case of an integral z-cut LN substrate having a thickness of 1 mm and a width of 2 mm. There is no demerit that the resonance frequency changes by adopting the structure of the present invention as compared with the case of the integral z-cut LN substrate.

  When the z-cut LN substrate 80 and the separate substrate 50 are formed of an integral z-cut LN substrate, an electromagnetic field simulating a high-frequency electric signal leaking from an input feedthrough portion (not shown) is an integral z. -It is clear that it also propagates in the width direction of the cut LN substrate, and was actually confirmed by electromagnetic field analysis.

  When this electromagnetic field simulating a high-frequency electrical signal leaking from an input feedthrough (not shown) is excited inside the separate substrate 50, the gap G between the separate substrate 50 and the z-cut LN substrate 80 is 10%. Of course, in the case of ˜30 μm, even in the case of 50 μm, not only 100 μm, and even 200 μm, the high frequency electric signal penetrates the adhesive layer 52 or the air layer 53 from the separate substrate 50 and z− Propagate to the cut LN substrate 80.

  That is, when the gap G between the z-cut LN substrate 80 and the separate substrate 50 is as shown in FIG. 11, the adhesive layer 52 is present in the gap between the z-cut LN substrate 80 and the separate substrate 50. However, even if the air layer 53 is present, the electromagnetic field passes through, and the z-cut LN substrate 80 and the separate substrate 50 function as if they were an integral z-cut LN substrate (80 'in FIG. 2B). .

  Thus, by narrowing the width of the optical grade z-cut LN substrate 80, a large number of LN optical modulator chips can be obtained from one wafer and the overall bonding area can be increased, so that the narrow LN light is obtained. While the effect of the present invention that enables the modulator chip to be firmly fixed to the pedestal of the housing (not shown) is exhibited, it can be seen that there are no problems accompanying it.

[Fourth Embodiment]
FIG. 14 shows a top view of the CPW traveling wave electrode 4 used in the fourth embodiment of the present invention. However, in this figure, only the traveling wave electrode formed on the optical grade LN substrate is shown in the figure. Of course, it goes without saying that this electrode pattern configuration may be formed on a separate substrate which is a SAW grade LN substrate. In the present embodiment, XIV is a first impedance converter, XV is a second impedance converter, and XVI is a third impedance converter. Here, all of the characteristic impedances of the first impedance converter XIV, the second impedance converter XV, and the third impedance converter XVI may be different from the geometric mean described above, for example, the characteristics of the second impedance converter XV The impedance may be the geometric average described above, the characteristic impedance of the first impedance converter XIV may be higher than the geometric average, and the characteristic impedance of the third impedance converter XVI may be lower than the geometric average (or vice versa). ). Regardless of this combination, various combinations of characteristic impedances of the first impedance conversion unit XIV, the second impedance conversion unit XV, and the third impedance conversion unit XVI may be used, or any two of them may be a geometric mean. good. In addition, in the present invention, including this embodiment, it is also possible to utilize the fact that the characteristic impedance of this portion changes significantly by connecting a gold ribbon (not shown) to the input feedthrough portion I as described above. It is effective from the viewpoint of shifting.

[Fifth Embodiment]
FIG. 15 shows a top view of the CPW traveling wave electrode 4 used in the fifth embodiment of the present invention. However, this drawing also shows the drawing focusing only on the traveling wave electrode formed on the optical grade LN substrate. Of course, it goes without saying that this electrode pattern configuration may be formed on a separate substrate which is a SAW grade LN substrate. In the present embodiment, XVIII is a first impedance converter, and XIX is a second impedance converter. In the case of this embodiment, the length of the interaction part III is ensured sufficiently long by once forming the impedance conversion parts XVIII and XIX by folding them back in the opposite direction with respect to the interaction part III. This concept can be applied to other embodiments including the fifth embodiment of the present invention.

[Sixth Embodiment]
FIG. 16 shows a top view of a CPW traveling wave electrode 4 comprising a central conductor 4a and ground conductors 4b and 4c used in the sixth embodiment of the present invention. In the present embodiment, XX is a first impedance converter, and XXI is a second impedance converter. In the present embodiment, as in the fifth embodiment shown in FIG. 15, the first impedance converter XX and the second impedance converter XXI are once folded in the opposite direction with respect to the interaction unit III. By making the distance in the longitudinal direction of the substrate from the starting point of the interaction portion to the light incident end face shorter than the distance in the longitudinal direction of the substrate from the high frequency electric signal feeding portion to the light incident end face, FIG. The length of the interaction part III is ensured longer than the fifth embodiment shown in FIG. Further, the output feed-through part XVII may be turned back from the end point of the interaction part III, and this can be said for all the embodiments of the present invention. The idea of increasing the interaction length described in the sixth embodiment is not limited to the case where the impedance conversion unit is two stages, but may be one stage, three stages or more, and all implementations of the present invention. Applicable to form. Also, in this embodiment, by connecting a connector core wire (not shown) to the input feedthrough portion I, the fact that the characteristic impedance of this portion is greatly reduced is utilized from the viewpoint of shifting from the geometric mean described above. It is valid.

  The concept shown in FIGS. 14 to 16, that is, the impedance conversion in which the impedance conversion unit is formed on a separate substrate, such as having a plurality of stages of impedance conversion, or extending the length by folding back one end in the longitudinal direction of the optical waveguide. Needless to say, it may be applied to the department.

[About each embodiment]
In the above description, the CPW electrode has been described as an example of the traveling wave electrode. However, it goes without saying that various traveling wave electrodes such as an asymmetric coplanar strip (ACPS) and a symmetric coplanar strip (CPS), or a lumped constant electrode may be used. . In addition to the Mach-Zehnder type optical waveguide, it goes without saying that other optical waveguides such as directional couplers and straight lines may be used as the optical waveguide. It goes without saying that a ridge structure may be used as the LN optical modulator. In the above description, the number of separate substrates is one, but two or more may be used. In this case, the chip of the LN optical modulator in which the optical waveguide is formed may be sandwiched between separate substrates, or the chip of the LN optical modulator in which the optical waveguide is formed on the side surface of the separate substrate arranged side by side. It may be arranged. Needless to say, the impedance converter may be formed only on a separate substrate.

  Further, although it has been described that there is a core wire of the microwave connector on the output side of the high-frequency electrical signal, this is not essential in the present invention, and the electrical resistance is generated in the housing package using the terminating resistor. Needless to say, it may be terminated. In the above description, a gold wire is used for electrical connection, but a gold ribbon may be used. Furthermore, although the case has been described as a metal, the idea of the present invention can also be applied when the case is made of a non-metallic material such as plastic.

  The adhesive as an adhesive means for adhering the side surface of the LN optical modulator chip and the side surface of a separate substrate may be an ultraviolet curable adhesive or a thermosetting adhesive, or a conductive adhesive such as a silver paste. Or soldering agent.

  In the above embodiment, the substrate has an x-cut, y-cut or z-cut plane orientation, that is, a crystal x-axis, y-axis or z-axis in a direction perpendicular to the substrate surface (cut plane). In addition, the plane orientation in each of the embodiments described above may be used as the main plane orientation, and other plane orientations may be mixed as the sub-plane orientation.

  As described above, the optical modulator according to the present invention is advantageous in that it is inexpensive, has high mechanical strength, and can improve the RF modulation performance, and is useful as an optical modulator having a high driving speed and a low driving voltage. It is.

1 is a perspective view of an LN optical modulator according to a first embodiment of the present invention. (A) A cross-sectional view of the LN optical modulator according to the first embodiment of the present invention, and (b) a cross-sectional view of an integrated LN substrate for explaining the principle of the first embodiment of the present invention. 1 is a top view of an LN optical modulator according to a first embodiment of the present invention. 1 is an equivalent circuit diagram according to the first embodiment of the present invention. 1 is an equivalent circuit diagram according to the first embodiment of the present invention. The figure explaining the characteristic of the 1st Embodiment of this invention The figure explaining the characteristic of the 1st Embodiment of this invention The figure explaining the characteristic of the 1st Embodiment of this invention The perspective view of the LN optical modulator which concerns on the 2nd Embodiment of this invention Sectional drawing in the DD 'line of FIG. The perspective view of the LN optical modulator which concerns on the 3rd Embodiment of this invention Sectional drawing in the EE 'line of FIG. The figure explaining the operation principle of 2nd Embodiment and 3rd Embodiment of this invention Top view of an LN optical modulator according to the fourth embodiment of the present invention Top view of an LN optical modulator according to a fifth embodiment of the present invention Top view of an LN optical modulator according to the sixth embodiment of the present invention 1 is a perspective view of an optical modulator according to a first conventional technique. Sectional drawing in the AA 'line of FIG. The figure explaining operation | movement of the optical modulator based on 1st prior art. 1 is a perspective view of an optical modulator module according to a first conventional technique. The top view which shows the wafer of LN based on 1st prior art, and the optical modulator chip | tip formed in it The figure explaining distribution of the electric line of force of the high frequency electric signal in the microwave connector part of the optical modulator module concerning the 1st prior art The figure explaining the distribution of the electric force line | wire of the high frequency electric signal in the feed-through part for input shown by the BB 'line | wire of FIG. 1 is a perspective view of a z-cut LN substrate according to a first prior art. Cross-sectional view of z-cut LN substrate according to first prior art The top view which shows the wafer of LN based on 2nd prior art, and the optical modulator chip | tip formed in it Perspective view of z-cut LN substrate according to second prior art Cross-sectional view of z-cut LN substrate according to second prior art The figure explaining the problem of the prior art about mechanical adhesive strength The figure explaining the relationship of V (pi) * L and W of the optical modulator which concerns on a 3rd prior art. The figure explaining the relationship between _nm and W of the optical modulator which concerns on a 3rd prior art. The figure explaining the relationship between Z and W of the optical modulator which concerns on a 3rd prior art Top view of the traveling wave electrode of the third prior art Equivalent circuit diagram of optical modulator according to third prior art The figure explaining the relationship between _Zin and f of the optical modulator which concerns on a 3rd prior art The figure explaining the relationship between | m | ² and f of the optical modulator according to the third prior art View for explaining the relationship of the third according to the prior art optical modulator S ₁₁ and f Top view of traveling wave electrodes according to the fourth prior art Sectional view taken along line B-B 'of FIG. Sectional view taken along line C-C 'of FIG. Equivalent circuit diagram of optical modulator according to second prior art Equivalent circuit diagram of optical modulator according to second prior art View for explaining the relationship between S ₁₁ and f of the optical modulator according to a second prior art

Explanation of symbols

1, 50, 80, 80 ': z-cut LN substrate (substrate, LN substrate)
2: SiO ₂ buffer layer (buffer layer)
3: Optical waveguide 3a, 3b: Interaction optical waveguide (optical waveguide, interaction part)
4: Traveling wave electrode (electrode)
4a, 60a, 60d: center conductor 4b, 4c, 60b, 60c, 60e: ground conductor 5: external signal source 6: load resistance 8: metal lid 10: wafer 11: chip of LN optical modulator 15: metal housing 16a, 16b: Core wire of the microwave connector 17a, 17b: Cavity around the core wire of the microwave connector 20: Base (fixed portion)
50: SAW grade LN substrate (separate substrate)
51: Contact surface (adhesion surface)
70a, 70b, 70c: Gold ribbons 80a, 80b, 80c: Gold wires 90: Lines representing the feedthrough portion VIII for input on the SAW grade LN substrate 91: Lines representing the impedance converting portion IX on the SAW grade LN substrate 92: A line representing the impedance converter VII on the optical grade LN substrate 93: A line representing the input feed-through portion I on the optical grade LN substrate 94: The output side connection portion X on the optical grade LN substrate Line 95: Line representing the output feedthrough portion XI on the optical grade LN substrate 96: Line representing the output feedthrough portion XII on the SAW grade LN substrate

Claims (24)

A substrate having an electro-optic effect, an optical waveguide for guiding light formed on the substrate, and a high-frequency electric signal formed on one surface side of the substrate and modulating the phase of the light An electrode composed of a central conductor and a ground conductor, a fixed portion to which the substrate is fixed, an interaction portion that modulates the phase of the light by applying the high-frequency electric signal to the electrode, and the interaction portion An optical modulator comprising a first input feedthrough section for propagating a high-frequency electrical signal;
At least one separate substrate fixed in parallel with the fixed portion together with the substrate is further provided, and resonance of dielectric resonance substantially determined from the thickness of the substrate and the thickness of the separate substrate The thickness of the substrate and the thickness of the separate substrate are set so that the frequency is higher than the frequency of the high-frequency electrical signal,
The separate substrate has a second input feedthrough portion including a center conductor and a ground conductor, and the high-frequency electrical signal from an external circuit applied to the second input feedthrough portion is the first input feedthrough portion. The second input feedthrough portion and the first input feedthrough portion are electrically connected so as to propagate to one input feedthrough portion, and the second input feedthrough portion Characteristic impedance of a connector to be electrically connected to the second input feedthrough section, or impedance mismatch between at least one characteristic impedance of the external circuit and the interaction section Comprising at least one impedance converter for reducing,
The high-frequency electrical signal applied from the external circuit to the second input feedthrough section causes residual reflection, and is electrically reflected as compared with the case where the impedance conversion section on the separate substrate is not provided. And the light modulator propagates to the interaction unit.   At least one of the product of the thickness of the substrate and the equivalent refractive index of the high-frequency electrical signal in the thickness direction, or the product of the thickness of the separate substrate and the equivalent refractive index of the high-frequency electrical signal in the thickness direction is 0. 2. The optical modulator according to claim 1, wherein the thickness of the substrate and the thickness of the separate substrate are set to be larger than 4 mm and smaller than 15 mm.   3. The sum of the width of the substrate and the width of the separate substrate is about 1.4 times or more the thickness of at least one of the substrate or the separate substrate. An optical modulator according to 1.   4. The optical modulator according to claim 1, wherein substantially all of a side surface of the substrate and a side surface of the separate substrate facing each other are fixed to each other by an adhesive unit. .   4. The optical modulator according to claim 1, wherein a part of a side surface of the substrate and a side surface of the separate substrate facing each other are fixed to each other by an adhesive unit. .   6. The optical modulator according to claim 1, wherein an adhesive strength of the substrate to the fixing portion is high as compared with a case where the separate substrate is not provided.   The optical modulator according to claim 1, wherein the separate substrate is made of a dielectric made of the same material as the substrate.   The optical modulator according to claim 1, wherein the separate substrate is made of a dielectric material made of a material different from that of the substrate.   A dielectric having a low dielectric constant layer having a relative dielectric constant smaller than that of the substrate at least below the second input feedthrough portion and determined from the thickness of the substrate and the thickness of the low dielectric constant layer The thickness of the low dielectric constant layer is set according to the thickness of the substrate and the thickness of the separate substrate so that the resonance frequency of resonance is higher than the frequency of the high-frequency electrical signal. The optical modulator according to claim 1.   The optical modulator according to claim 9, wherein at least a part of the low dielectric constant layer is made of air.   The optical modulator according to claim 9, wherein the low dielectric constant layer is made of a material having a relative dielectric constant larger than that of air and smaller than that of the substrate.   At least a part of the impedance conversion unit formed on the separate substrate is a geometric mean of the characteristic impedance of the interaction unit and the characteristic impedance of the second input feedthrough unit, the characteristic of the interaction unit A characteristic different from the geometric mean of the impedance and the characteristic impedance of the connector to be electrically connected to the second input feedthrough part, or the geometric mean of the characteristic impedance of the interaction part and the characteristic impedance of the external circuit The optical modulator according to claim 1, wherein the optical modulator has an impedance.   A part of the impedance conversion unit formed on the separate substrate is a geometric mean of the characteristic impedance of the interaction unit and the characteristic impedance of the second input feedthrough unit, the characteristic impedance of the interaction unit And at least one of the geometric mean of the characteristic impedance of the connector to be electrically connected to the second input feedthrough part, or the geometric mean of the characteristic impedance of the interaction part and the characteristic impedance of the external circuit, The optical modulator according to claim 1, wherein the optical modulators have the same characteristic impedance.   On the substrate on which the optical waveguide is formed, another impedance conversion unit is provided, the second input feedthrough unit and the impedance conversion unit, and the first input feedthrough unit and the other impedance conversion. And the characteristic impedance of the impedance conversion unit as a whole should be electrically connected to the characteristic impedance of the interaction unit and the second input feedthrough unit. 14. A characteristic impedance different from a geometric mean of a characteristic impedance of a connector or a geometric mean of a characteristic impedance of the interaction part and the characteristic impedance of the external circuit. The optical modulator according to one.   At least part of the impedance conversion unit on the separate substrate is a geometric mean of the characteristic impedance of the interaction unit and the characteristic impedance of the second input feedthrough unit, the characteristic impedance of the interaction unit, A geometric mean of the characteristic impedance of the connector to be electrically connected to the second input feedthrough part, or a geometric mean of the characteristic impedance of the interaction part and the characteristic impedance of the external circuit; 15. The optical modulator according to claim 1, wherein the optical modulator has a characteristic impedance whose difference is within about ± 7Ω.   At least a part of the impedance conversion unit on the separate substrate is a geometric mean of the characteristic impedance of the interaction unit and the characteristic impedance of the second input feedthrough unit, the characteristic of the interaction unit And a geometric mean of the impedance and the characteristic impedance of the connector to be electrically connected to the second input feed-through part, or a geometric mean of the characteristic impedance of the interaction part and the characteristic impedance of the external circuit The optical modulator according to claim 1, wherein the optical modulator has a characteristic impedance whose difference is within about ± 15Ω.   At at least one frequency in a frequency range between about 20% and about 30% of the operating bit rate, the power reflectance envelope is between -10 dB and -15 dB, and about 40% to about 40% of the operating bit rate. In the frequency range between 70%, there is a minimum point where the first derivative of the envelope of the power reflectivity is zero and the second derivative is positive, and the power reflectivity at the minimum point is -15 dB or less. The optical modulator according to any one of claims 1 to 16, wherein at least a part of the impedance converter has a characteristic impedance lower than a characteristic impedance of the connector or the external circuit.   The power reflectance at at least one point in a frequency range between about 2 GHz and 4 GHz is between −10 dB and −15 dB, and the minimum point is provided in a frequency range of about 4 GHz to 9 GHz. The optical modulator according to claim 1.   The power reflectivity at at least one point in a frequency range between about 5 GHz and 8 GHz is between -10 dB and -15 dB, and the minimum point is provided in a frequency range of about 10 GHz to 18 GHz. The optical modulator according to claim 1.   The power reflectance at at least one point in a frequency range between about 9 GHz and 13 GHz is between -10 dB and -15 dB, and the minimum point is provided in a frequency range of about 17 GHz to 30 GHz. The optical modulator according to claim 1.   The power reflectance at at least one point in a frequency range between about 11 GHz and 16 GHz is between −10 dB and −15 dB, and the minimum point is provided in a frequency range of about 22 GHz to 38 GHz. The optical modulator according to claim 1.   The power reflectance at at least one point in a frequency range between about 22 GHz and 33 GHz is between −10 dB and −15 dB, and the minimum point is provided in a frequency range of about 44 GHz to 77 GHz. The optical modulator according to claim 1.   The optical modulator according to any one of claims 1 to 22, wherein the substrate is lithium niobate. The optical modulator according to claim 1, wherein the substrate is a semiconductor.

Images