Semiconductor optical modulator

Abstract

A semiconductor optical modulator, characterized by a high-frequency line with a differential structure, arranged in parallel with a waveguide and used to transmit a high-frequency modulated signal; a connecting pad, continuous with the high-frequency line and formed in the same direction; and a terminal resistor, composed of two rectangular resistor bodies for differentially terminating the high-frequency modulated signal from the connecting pad. These three parts are arranged in a straight line and terminated on the chip, and the high-frequency line of the differential pair is short-circuited after passing through the terminal resistor.

Description

Technical Field

[0001] The present disclosure relates to an ultra-high-speed semiconductor optical modulator for modulating an optical signal with an electrical signal. Background Art

[0002] To meet the growing demand for communications services, high-speed optical modulators that support advanced optical modulation methods are required. In particular, multi-level optical modulators using digital coherent technology have played a significant role in achieving high-capacity transceivers exceeding 100Gbps.

[0003] In order to add independent information to the amplitude and phase of light, these multi-level optical modulators have multiple stages of Mach-Zehnder interferometer optical modulators (hereinafter referred to as MZMs) capable of zero-chirp driving built in parallel.

[0004] In recent years, miniaturization and low drive voltage of optical transmitter modules have become issues, and research and development of semiconductor MZ optical modulators that can be miniaturized and have low drive voltages are being vigorously promoted. In addition, in the research and development of semiconductor MZ optical modulators, support for high baud rates such as 64GBaud and 100GBaud is accelerating, and there is a demand for broadband optical modulators. Among them, the research and development of high-bandwidth coherent driver modulators (HB-CDM) that aims to improve high-frequency characteristics by integrating the driver and optical modulator into a single package and coordinating the design of the driver and optical modulator to achieve miniaturization is accelerating. (Non-patent document 1)

[0005] Ideally, in this HB-CDM structure, since the optical modulator and the differential drive driver are integrated, the modulator itself is also based on a differential drive structure. In the HB-CDM structure, since the driver and modulator are integrated, the design of not only the modulator but also the driver is very important. In particular, in order to achieve low power consumption in HB-CDM, an open collector (or open drain) driver is used. (Non-Patent Documents 1 and 2)

[0006] In open-collector (or open-drain) drivers, the output terminal is open. Therefore, if the high-frequency line of the optical modulator is not properly terminated, the modulated electrical signal can pass through the high-frequency line and return, causing resonance between the driver and the terminating resistor. This can degrade the modulation bandwidth, cause periodic fluctuations in high-frequency characteristics, and lead to poor signal quality. Consequently, the importance of impedance matching between the impedance of the optical modulator's modulation electrode and the impedance of the driver and terminating resistor has increased dramatically in recent years.

[0007] Furthermore, in the structure of an IQ (In-Phase Quadrature) optical modulator, if the RF line length of the optical modulator varies between IQs, this can cause fluctuations in the high-frequency characteristics, leading to variations in the high-frequency characteristics between IQs. Furthermore, achieving high baud rates such as 64GBaud and 100GBaud requires high-precision termination with impedance matching across a wide frequency range (approximately 40GHz for 64GBaud and 60GHz for 100GBaud). The current demand for such high baud rates requires methods to optimize the termination resistors, maintain structural uniformity between IQs across a wide frequency range, and mitigate the effects of impedance mismatch.

[0008] exist Figure 1 , an example of a conventional semiconductor Mach-Zehnder modulator (MZM) is shown.

[0009] (Patent Document 1)

[0010] exist Figure 1 In a conventional semiconductor Mach-Zehnder modulator 100, input light from the lower left input optical waveguide 101a is branched into two arm optical waveguides 101 by a 1×2 MMI coupler 102 on the demultiplexing side. The branched optical signals pass under traveling-wave electrodes 121 of two columns of a capacitance-loading structure (a structure in which T-shaped and inverted-T-shaped horizontal bar electrodes periodically protrude from the electrode body of the differential circuit and are arranged on the arm optical waveguides and face each other), driven by differential modulated electrical signals DATA and / DATA (the superscripts of the electrical signal symbols in the figure are abbreviated as " / " to indicate signals of opposite polarity) input from the left end of the figure, undergoing optical phase modulation.

[0011] The optical signals from the individual optical phase modulated arms are combined and interfered by a 2×1 MMI (multimode interference) coupler 103 on the combining side, and then output from the optical waveguide in the center of the right end. On the output side of the 2×1 MMI coupler 103, two upper and lower dam patterns 110 are provided across the central optical waveguide from which the interference light is output, as a means for emitting high-order mode light from the multimode interference coupler.

[0012] The electrode lines 122 and 123 of the two-column differential high-frequency line 120 through which the differential modulated electrical signals DATA and / DATA pass pass over the 2×1 MMI coupler 103, bend upward and rightward, and are terminated off-chip by a 50Ω terminal resistor.

[0013] In the differential high-frequency line 120 that bends upward to the right and transmits the differential modulated electrical signals DATA and / DATA, due to the difference in curvature between the inner and outer sides of the electrode lines 122 and 123, the line lengths between the two electrode lines forming the differential pair differ, so there is a possibility that the characteristics of the differential signal will be degraded, resulting in the generation of noise and degradation of high-frequency characteristics.

[0014] exist Figure 2 An example of a conventional IQ optical modulator is shown in Patent Document 2. Figure 2 In the optical path of the IQ dual system, the light input of the input optical waveguide 10 on the lower left is branched at the coupler 11 and reaches the light output at the output optical waveguide 47 on the right end. Figure 1 The same MZM optical waveguide. Figure 2 In conventional IQ optical modulators, the modulation electrodes (20, 21, 22, and 23) driven by the I, / I, Q, and / Q modulation electrical signals on the left end are terminated off-chip at the lower end of the chip, as indicated by the circle. Consequently, a bent portion is provided in a portion of the modulation electrodes. This bent portion of the modulation electrodes increases the chip size and causes differences in electrode length between the IQ and Q channels, resulting in fluctuations in the high-frequency characteristics of the IQ optical modulator with different periods between the IQ and Q channels.

[0015] Prior art literature

[0016] Patent Literature

[0017] Patent Document 1: Japanese Patent Application Laid-Open No. 2019-194637

[0018] Patent Document 2: Japanese Patent Application Laid-Open No. 2019-45666

[0019] Non-patent literature

[0020] Non-Patent Literature 1: J. Ozaki, et al., “Ultra-low Power Dissipation (<2.4W) Coherent InP Modulator Module with CMOS Driver IC”, Mo3C.2, ECOC, 2018

[0021] Non-patent literature 2: N.Wolf, et al., "Electro-Optical Co-Design to MinimizePower Consumption of a 32GBd Optical IQ-Transmitter Using InP MZ-Modulators", CSICS, 2015 Summary of the Invention

[0022] To realize optical modulators operating at high baud rates and ultra-high speeds, it is necessary to achieve impedance matching across the entire broadband frequency range of the modulated electrical signal, coupled with high-precision termination. With the current demand for such high baud rates, differential high-frequency lines must minimize the presence of bends that could degrade characteristics, optimize the termination resistor structure, and achieve a structure that maintains uniform I / Q differences across a wide bandwidth while minimizing the effects of impedance mismatch.

[0023] The present invention provides a compact termination resistor structure that can precisely terminate the high-frequency modulation circuit of an optical modulator operating at ultra-high speed across the entire broadband, suppressing periodic fluctuations when connected to an open-collector driver. Furthermore, the structure maintains the same period between I and Q signals even when slight periodic fluctuations occur.

[0024] A semiconductor optical modulator, one aspect of the present invention, comprises: a high-frequency line with a differential structure, arranged in parallel with an optical waveguide and used to transmit a high-frequency modulated signal; a connection pad, continuous with the high-frequency line and formed in the same direction; and a terminal resistor comprising two rectangular resistor elements for differentially terminating the high-frequency modulated signal from the connection pad. The high-frequency line, the connection pad, and the terminal resistor are arranged in a straight line, and the high-frequency line is terminated on-chip. The terminal resistor is short-circuited on opposite sides of the connection pad.

[0025] As described above, one aspect of the present invention provides an ultra-wideband semiconductor optical modulator equipped with compact termination resistors that precisely terminate the optical modulator's high-frequency circuit across the entire broadband. When connected to an open-collector driver, for example, this optical modulator can suppress periodic fluctuations in high-frequency characteristics. Furthermore, even when slight periodic fluctuations in the high-frequency characteristics occur, the I / Q period can be maintained uniformly. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 This is a diagram showing an example of a conventional semiconductor Mach-Zehnder modulator.

[0027] Figure 2 This is a diagram showing an example of a conventional IQ optical modulator.

[0028] Figure 3 It is a diagram showing the structure of a semiconductor optical modulator according to an embodiment of the present invention.

[0029] Figure 4 This is an enlarged plan view of the structure of a terminal structure portion of a semiconductor optical modulator according to an embodiment of the present invention.

[0030] Figure 5This is an enlarged top view of the structure of the terminal structure portion of the semiconductor optical modulator according to the embodiment of the present invention.

[0031] Figure 6 (A), (B), (C) are Figure 5 A cross-sectional view of various parts of the terminal structure of a semiconductor optical modulator.

[0032] Figure 7 This is a plan view of an IQ optical modulator according to Embodiment 2 of the present invention.

[0033] Figure 8 This is a plan view of an IQ optical modulator according to Embodiment 2 of the present invention.

[0034] Figure 9 This is a plan view of an IQ optical modulator according to Embodiment 2 of the present invention.

[0035] Figure 10 This is a plan view schematically showing the connection with the driver IC of the IQ optical modulator according to the second embodiment of the present invention.

[0036] Figure 11 This is a diagram showing the frequency characteristics of the reflectance of the semiconductor optical modulator according to the embodiment of the present invention. DETAILED DESCRIPTION

[0037] Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings.

[0038] (Implementation Method 1)

[0039] Figure 3 A plan view of a semiconductor optical modulator according to Embodiment 1 of the present invention is shown.

[0040] exist Figure 3 In the first embodiment of the present invention, an InP Mach-Zehnder modulator fabricated on an InP substrate is described as an example of an optical semiconductor. Since it is known that a semi-insulating (SI)-InP substrate is used in InP Mach-Zehnder modulators that achieve broadband / high baud rate operation, this embodiment describes the case where an SI-InP substrate is used.

[0041] Figure 3 FIG. 1 shows a plan view of the substrate of the InP Mach-Zehnder modulator according to the first embodiment of the present invention as viewed from the upper surface. Figure 3The InP Mach-Zehnder modulator has: an input optical waveguide 101a for light input; a 1×2 MMI coupler 102 connected to the input optical waveguide 101a; two arm optical waveguides connected to the 1×2 MMI coupler 102; and a 2×1 MMI coupler 103 connected to the two arm optical waveguides. The 2×1 MMI coupler 103 is connected to an output waveguide for light output on the opposite side of the two arm optical waveguides. In addition, dam patterns 110a and 110b for high-order mode radiation are formed above and below (on both sides) the output waveguide of the 2×1 MMI coupler 103. In addition, Figure 3 The InP Mach-Zehnder modulator comprises traveling-wave electrodes 300a and 300b and ground electrodes 300c and 300d, which form a differential high-frequency circuit. The traveling-wave electrodes 300a and 300b have periodically protruding T-shaped and inverted-T-shaped horizontal bar electrode sections 121. These horizontal bar electrode sections 121 are arranged on the two arm optical waveguides. The differentially modulated electrical signals DATA and / DATA are input from the left sides of the traveling-wave electrodes 300a and 300b.

[0042] In the InP Mach-Zehnder modulator of embodiment 1, the structure of the optical waveguide is the same as Figure 1 The conventional Mach-Zehnder modulator (MZM) is basically the same, but the two columns of row-wave electrodes 300a and 300b driven by the differential modulated electrical signal (high-frequency modulated signal Signal: DATA, / DATA) on the substrate surface are clamped on both sides by a group of Gound electrodes 300c and 300d, forming a phase modulation part of the so-called GSSG structure (G: Ground, S: signal).

[0043] exist Figure 3 In FIG, the GSSG structure is described as an example of a differential line structure. However, as long as it is a differential line structure, it can also be a GSGSG structure or a SS structure without a ground.

[0044] like Figure 3 As shown, in the optical modulator of the first embodiment, the traveling wave type electrodes 300a, 300b and the differential high frequency line driven by the differential modulated electrical signal (DATA, / DATA) are connected. Figure 1 Unlike conventional structures, the high-frequency line section (terminal section) after phase modulation has no bends. A pair of differential high-frequency lines is formed by three components: traveling-wave electrodes 300a and 300b; two connecting pads 301a and 301b connected to the terminal sections of traveling-wave electrodes 300a and 300b; and terminal resistors 302a and 302b, consisting of two rectangular resistors connected to the connecting pads 301a and 301b. These three components are arranged linearly and terminate on the chip.

[0045] The terminal resistors 302 a and 302 b are respectively formed on the dam patterns 110 a and 110 b for high-order mode radiation connected to the 2×1 MMI coupler 103 .

[0046] Basically, if there is a dam pattern directly below a high-frequency (RF) line, a p-type or n-type semiconductor layer structure as a conductor exists, which will increase high-frequency loss and deteriorate high-frequency characteristics. In the InP Mach-Zehnder modulator of this embodiment, Figure 3 As shown, except for the 2×1 MMI coupler portion 103, an isolated high mesa structure without a bank pattern is formed.

[0047] The right ends (terminal sides) of the two terminal resistors 302 a and 302 b are short-circuited by a short-circuit portion 303 made of a conductive metal or the like, thereby forming on-chip terminals.

[0048] In addition, Figure 3 In the terminal resistor portion, for simplicity, no ground electrodes are provided on the left and right sides of the signal line. However, a terminal structure with ground electrodes can also be used. In this case, the dam pattern can be enlarged, and the ground electrode can be provided on the dam pattern together with resistors 302a and 302b, or the ground electrode can be provided in the portion without the dam pattern. Furthermore, a GSGSG structure can be used, in which the ground electrode is located between the two resistors 302a and 302b.

[0049] Figure 4 Yes Figure 3 This is an enlarged top view of the 2×1 MMI coupler 103 and the dam patterns 110a and 110b at the terminal end of the differential high-frequency line. Figure 4 In the example, the dam patterns 110a and 110b are Figure 3 An alternative to the dam pattern shown. Figure 4 As shown in FIG. 1 , the 2×1 MMI coupler 103 has two arm waveguides connected to the input side and an output waveguide connected to the output side. Dam patterns 110a and 110b extend from both sides of the connection position of the output waveguide of the 2×1 MMI coupler 103 in an oblique direction relative to the propagation direction of the output light. The dam patterns 110a and 110b are formed to surround the 2×1 MMI coupler 103. Figure 4 In the example, the two dam patterns 110a and 110b connected to the output side of the 2×1 MMI coupler 103 have cutouts and are formed into a feather-like shape. This structure is the best structure for processing the high-order mode radiation light in the MMI. Specifically, Figure 3 The dam patterns 110a and 110b are different. Figure 4The angles of the vertices in the dam patterns 110a and 110b avoid shapes with internal angles such as right angles (90°) and 45°, and are formed by straight lines or curves to form appropriate angles. The angles are designed to be appropriate so that the radiated light does not geometrically return to the MMI side, but is radiated forward. Therefore, the optical characteristics can be maintained by preventing the radiated light of the high-order mode from returning to the MMI side as stray light. However, as long as the high-order mode light of the multimode interference coupler can be appropriately radiated and stray light can be suppressed, the dam pattern is not limited to Figure 4 The structure shown can also be Figure 3 The structure shown, and any other shape.

[0050] Figure 5 Yes Figure 4 The diagram shows an enlarged plan view of a 2×1 MMI coupler 103 and a terminal portion of a differential high-frequency line formed on bank patterns 110 a and 110 b , including resistors 302 a and 302 b , connection pads 301 a and 301 b extending from differential high-frequency lines 300 a and 300 b , and a short-circuit member 303 .

[0051] Figure 6 (A), (B) and (C) are respectively Figure 5 The shown diagram is a cross-sectional view of the substrate at positions of three substrate cross sections AA', BB', and CC' perpendicular to the guided light.

[0052] Figure 6 The cross-sectional view (A) of the substrate is Figure 5 The substrate below the connection pads 301a and 301b is stacked in the order of i-InP and MQW layers 601, n-InP layer 602, and SI-InP layer (substrate) 603 from the top.

[0053] As mentioned above, the electrodes connecting the pad portions 301a and 301b are formed into a cone (or a shape in which the cone and the rectangle are continuous). For impedance matching, the width of the cone is expanded along the propagation direction of light (and / or is designed to be formed from the area of ​​the rectangle continuous from the cone). The thickness and width of the i-InP and MQW layers 601 and the n-InP layer 602 must also be formed taking impedance matching into consideration.

[0054] The center portion of the substrate cross-sectional view (A) corresponds to the cross-section of the MZM arm optical waveguide. However, no capacitive loading structure 121 is formed on the optical waveguide. The rest of the upper layer is embedded with a dielectric material such as BCB.

[0055] Figure 6 The cross-sectional view (B) of the substrate is Figure 5The substrate cross-section view of the resistors 302a and 302b is taken along the line VIB-VIB'. Similar to the substrate cross-section view (A), the substrate beneath the resistors 302a and 302b is stacked in the order of i-InP and MQW layers 601, n-InP layers 602, and SI-InP layers (substrate) 603. The center portion of the substrate cross-section view (B) corresponds to the cross-section of the 2×1 MMI coupler 103.

[0056] Figure 6 The cross-sectional view (C) of the substrate is Figure 5 The substrate cross-section VIC-VIC' corresponds to the entrance of the connection pads 301a and 301b. The center portion of the substrate cross-section (C) corresponds to the cross-section of the MZM arm optical waveguide. The capacitive loading structure 121 is not formed above the optical waveguide. The remaining upper layer is embedded with materials such as BCB.

[0057] return Figure 3 In the differential high-frequency line for optical phase modulation, a capacitance-loaded traveling-wave electrode structure of a GSSG structure having excellent broadband characteristics is shown as an example, but other differential high-frequency line structures may also be used.

[0058] It is important that the terminal structure from the differential high-frequency lines 300a, 300b of the phase modulation section composed of the capacitor loading structure to the resistors 302a, 302b is as follows: Figure 3 The arrangement is shown as being parallel to the waveguide and arranged in a straight line. This is because if a structure other than a straight line is used, including a bend, the quality of the differential high-frequency signal that contributes to phase modulation will be degraded.

[0059] Also from the perspective of the signal quality of the differential high-frequency signal, it is important that the structure from the capacitor-loaded differential high-frequency lines 300a, 300b is continuous, through the connecting pads 301a, 301b to the two resistors 302a, 302b that terminate the high-frequency signal differential is formed without any bends in the same direction as the propagation direction of the capacitor-loaded structure (on the same straight line in a broad sense).

[0060] In addition, the connection pads 301a and 301b are arranged in the front section of the terminal resistors 302a and 302b, and are the pattern before the differential termination. Therefore, it is ideal that even if the size is set to, for example, less than several tens of μm, it matches the line impedance of the capacity loading structure 300a and 300b of the continuous phase modulation part.

[0061] If a connection pad is used that does not achieve impedance matching or has a shape that is not impedance designed, an impedance mismatch will occur between the connection pad portion and the capacitor loading structure before the differential signal is terminated, causing reflection between them before the termination, and seriously degrading the high-frequency characteristics of this chip (reflection characteristics and transmission characteristics, etc.).

[0062] Therefore, although the connection pads are referred to as connection pads, they are designed as differential high-frequency lines regardless of size, requiring impedance design not only near DC but also within the frequency range up to the frequency band used.

[0063] When viewed from the structure having only the main line (traveling wave electrodes 300 a and 300 b not arranged on the optical waveguide) excluding the capacitive loading structure 121 , higher impedance is seen compared to the structure including the capacitive loading structure 121 .

[0064] This is because the main line alone does not have the electrode pattern (T-shaped horizontal bar portion) arranged on the optical waveguide, which is the basis of the capacitance loading structure 121 and is used to adjust / provide the capacitance component for achieving a desired impedance.

[0065] Therefore, in Figure 3 In the embodiment, the following structure is adopted: by setting the shape of the connecting pad portions 301a and 301b to a taper with a width of about 5 times the width of the main line of a part of the capacity loading structure, the capacitance of the connecting pad portions is increased, and impedance matching with the differential high-frequency line portion of the phase modulation portion is achieved.

[0066] exist Figure 3 In the embodiment, the connection pads 301a and 301b have a tapered shape with a length of 50 μm in the light propagation direction, followed by a rectangular portion with a length of 50 μm in contact with the resistor.

[0067] From the viewpoint of impedance matching, a long taper length of the connection pad portion means that a line with higher impedance than desired will continue. Therefore, the taper length is preferably 100 μm or less.

[0068] However, the taper alone cannot sufficiently increase the capacitance, and therefore, the following structure may be adopted: a rectangular electrode pattern is provided after the tapered portion, and at least a portion of the lower portion thereof is formed as shown in FIG. Figure 6 As shown in FIG. 5 (B), an n-type semiconductor layer 602 formed on a SI-InP substrate and an undoped semiconductor layer 601 formed by regrowth are provided to provide strong capacitance in the stacking direction and achieve impedance matching.

[0069] When adjusting the capacitance, the length of the rectangular electrode portion having an n-type semiconductor layer and an undoped semiconductor layer, which is located at the bottom of the rectangular electrode pattern after passing through the above-mentioned conical portion, is lengthened, or the width of the above-mentioned semiconductor layer is expanded to the entire width of the above-mentioned rectangular electrode portion rather than the width of a part thereof, thereby adjusting the amplitude of the semiconductor layer, thereby adjusting the capacitance.

[0070] In the structure of this embodiment, a semiconductor layer is not provided in the tapered portion of the connection pad, for example, a structure is formed on a low dielectric 604 such as BCB (Benzocyclobutene), but as long as the shape can achieve impedance matching, a structure in which a semiconductor layer is formed at the bottom of the tapered portion can also be used.

[0071] In the above embodiment, the n-type semiconductor layer 602 is used as the semiconductor layer, but it can also be composed of a p-type semiconductor layer. Ideally, the thickness of the n-type semiconductor layer or the p-type semiconductor layer is about 50 nm or more to fully obtain the effect at high frequencies.

[0072] It can also be, except Figure 6 The undoped semiconductor layer 601, outside the waveguide portion and located below the pads and resistors, is formed of a dielectric other than a semiconductor. For example, a low-dielectric dielectric such as BCB can be used. However, in this case, the dielectric constant is lower than that of an undoped layer of a semiconductor such as InP. Therefore, the dielectric constant of the area where the line is formed tends to decrease, which means that the capacitance tends to decrease. This is not ideal from the perspective of increasing capacitance.

[0073] While the embodiments simply include only one n-type semiconductor layer, the structure may be formed from one or more n-type or p-type semiconductor layers and one or more dielectric layers. Multiple n-type or p-type semiconductor layers and dielectric layers may be present. For example, a pin structure may be formed, where a typical phase modulation portion is formed by mixing an n-type semiconductor layer, a p-type semiconductor layer, and an undoped MQW or InP layer. Furthermore, a dielectric such as a glass film may be formed on top of the pin structure.

[0074] (Material of the terminal resistor)

[0075] Furthermore, considering the manufacturing process, it is ideal to select the same resistor material for the terminal resistor as for the heater (not shown) that forms the electrode for adjusting the operating point of the modulator. By using the same resistor material, the terminal resistor can be manufactured simultaneously with, for example, the heater electrode for phase adjustment, in the same process.

[0076] However, in this case, the resistance value for the terminal is affected by the design value of the resistor of the operating point adjustment electrode, and therefore it is difficult to select a high-resistance resistor material such as that used in general analog ICs.

[0077] Of course, it is also possible to select a resistor material with a higher resistance than the heater so that the length of the resistor in the propagation direction is as short as possible. This is ideal from the perspective of high-frequency characteristics. However, in this case, since an additional process of setting up the resistor for the terminal occurs in a process different from the process of the heater for the action point adjustment electrode, the production process is increased / complexed, and the production cost increases, which is not ideal.

[0078] Furthermore, from the perspective of preventing oxidation, the top portion of the resistor is ideally covered with a dielectric such as glass. However, even without a dielectric such as glass, desired characteristics can be achieved, so this is not essential. In the drawings of Embodiment 1 of the present invention, such a glass film covering the resistor is omitted for simplicity.

[0079] In terms of high frequencies, it is more ideal to terminate with a smaller area. Considering the influence on the reflection characteristics, the ideal length of the resistor used for termination in the propagation direction is 200 μm or less.

[0080] Furthermore, from the perspective of resistance value manufacturing variation and stability, the ideal resistor width is 5 μm or greater. If the width is less than 5 μm, slight variations in the width during manufacturing will cause significant resistance value variation, which can lead to uneven reflection characteristics or degradation.

[0081] The terminal resistor section, consisting of the connection pads and the resistor element, can be considered a differential transmission line. Therefore, it should be designed as a high-frequency line. Ideally, the terminal resistor element should be rectangular to allow high-frequency signals to propagate in a straight line. While the resistor element can be formed in a zigzag (snaking) shape other than a straight line, such a shape, when considered a differential transmission line, would create bends and other problems, leading to degraded signal quality.

[0082] In addition, in order to reduce reflection at the operating frequency, it is ideal to make the length of the terminal resistance part composed of the connecting pad and the resistor body in the propagation direction sufficiently small. It is ideal to make the length of the terminal resistance part composed of the connecting pad and the resistor body in the propagation direction less than 1 / 4 of the wavelength of the high-frequency signal within the operating frequency of the differential high-frequency line transmission, and if possible, 1 / 8.

[0083] On the other hand, the width and length of the resistor body are also determined by the impedance required of the resistor as a high-frequency circuit. Therefore, when the resistor body is considered a high-frequency circuit, the degree of freedom in impedance design is significantly reduced compared to conventional high-frequency circuits formed of metal. Therefore, similar to the connection pad portion described above, at least a portion of the lower portion of the resistor body is provided with an n-type semiconductor layer formed on a Si-InP substrate and an undoped semiconductor layer formed by regrowth, significantly increasing the degree of freedom in capacitive / high-frequency circuit design.

[0084] In this case, considering the thickness of the resistor, it is desirable to form a stable pattern with no level differences and a flat surface to eliminate the risk of disconnection. Therefore, unlike the conventional pad portion, it is ideal to form the above-mentioned n-type semiconductor layer and undoped semiconductor layer under the entire resistor area.

[0085] Specifically, in order to reliably form a resistor on the above-mentioned n-type semiconductor layer and undoped semiconductor layer, from the perspective of achieving a stable process, the n-type semiconductor layer and the undoped semiconductor layer are arranged in a manner that is 5 μm or more wider than the resistor in the width direction.

[0086] This is because if the n-type semiconductor layer and the undoped semiconductor layer are smaller than the above width, the resistor will protrude from the semiconductor layer, flatness cannot be ensured, and there is a risk of disconnection in some areas.

[0087] From the perspective of miniaturizing the size of the optical modulator and efficiently integrating the terminal resistor on the chip, it is possible to effectively utilize Figure 1 (and Figure 3 ) shows a dam pattern 110 of a high-order mode light emitting element in a multimode interference coupler (1×2, 2×1, 2×2, etc.). By forming a terminal resistor on the shape of this dam pattern, a compact terminal resistor can be realized without generating unnecessary patterns. In particular, from the perspective of miniaturization, the multimode interference coupler is ideally designed for multiplexing (2×1, 2×2, etc.) in terms of high-frequency line structure.

[0088] Moreover, ideally, Figure 3 As shown in the figure, the differential pair is short-circuited immediately after passing through the resistor. This is because if a structure without short-circuiting is adopted, the signal component that is not completely terminated by the resistor will be transmitted as is, degrading the reflection characteristics.

[0089] By performing a short circuit, the differential signals are completely canceled out at the midpoint of the differential pair (the midpoint of the short-circuiting metal member 303 forming the short circuit), and therefore, short circuiting is ideal.

[0090] For example, when the results of a case where a short circuit is performed and a case where a short circuit is not performed are compared through simulation, it is found that there is a difference of at least several dB in the reflection characteristics.

[0091] (Implementation Method 2)

[0092] exist Figure 7 , a schematic diagram of an IQ optical modulator according to a second embodiment of the present invention is shown in the case where a driver PAD is additionally provided in the IQ.

[0093] Figure 7 Basically, two Figure 3 The optical modulators of the first embodiment of the present invention are arranged side by side. The driver driving voltage is applied from the driver driving voltage application PADs 700I and 700Q located in the center of the short-circuit components at the terminal end of each optical modulator, and modulated light is output from the optical waveguide below each PAD (not shown). Figure 7 In the IQ optical modulator, the two differential electrode lines use a GSSG structure. However, other differential line structures can also be used, such as an SS structure without a ground or a GSGSSG structure with a ground between SS lines. It should be noted that if four optical modulators of Embodiment 1 are arranged side by side, a Twin-IQ optical modulator is formed, which can handle multiple polarizations.

[0094] exist Figure 8 , an example of the schematic diagram of the IQ optical modulator of the second embodiment is shown in FIG. 1 , where the driver PAD is integrated in the IQ. The driver driving voltage application terminals of the two optical modulators are combined into Figure 8 The driver driving voltage is applied with PAD900 in the lower right corner, and the IQ is aggregated into one in the channel. It should be noted that when four optical modulators are arranged side by side to cope with polarization multiplexing, the IQ can be aggregated separately for X polarization and Y polarization, or the IQ can be separated and aggregated into one for X polarization and Y polarization. Figure 8 In the IQ optical modulator, the GSSG structure is used, but other differential line structures such as the SS structure without ground or the GSGSG structure with ground between SSs may also be used.

[0095] Figure 9 This is another schematic diagram showing an IQ optical modulator according to Embodiment 2. It shows an example of the structure of an optical waveguide composed of MZ modulators forming a nested mother-and-child structure.

[0096] Figure 9The IQ optical modulator shown is a so-called nested MZM, where each arm of the mother MZM is composed of a daughter MZM 100. Two differential electrode lines (IchDATA, Ich / DATA, QchDATA, Qch / DATA) are provided along the two arms that constitute the daughter MZM optical waveguides, respectively, to phase-modulate the optical signal propagating through the MZM optical waveguide by applying a modulation signal. Figure 9 In the IQ modulator, the driver driving voltage application PAD 900 is arranged approximately in the middle of the I channel and the Q channel. The driver driving voltage application PAD 900 may be arranged at any position between the I channel and the Q channel, or may be arranged at a position offset to the Q channel side instead of between the I channel and the Q channel (see Figure 8 ) or a position offset to the I channel side. Figure 9 As shown, the two differential electrode lines are SS structures without ground. The two differential electrode lines can also be GSSG structures or GSGSG structures with ground electrodes. Figure 9 As shown, the width (length in a direction perpendicular to the propagation direction of the modulated signal) of the extended portion of the electrode line 122 constituting the two differential electrode lines is wider than the portion provided with the traveling-wave electrode 121. The width of the extended portion may be the same as or narrower than the width of the portion provided with the traveling-wave electrode 121.

[0097] (Connection between driver IC and IQ optical modulator)

[0098] exist Figure 10 , an example of connection between an open collector or open drain driver IC and the IQ optical modulator according to the second embodiment of the present invention is shown.

[0099] exist Figure 10 , an example is shown in which a driver IC 1001 having two sets of output pads arranged in a GSGSG arrangement on the left end of the figure is connected to a 2-channel IQ optical modulator 1000. Although the output destinations of the central G pads of the two sets of output pads arranged in a GSGSG arrangement of the driver IC 1001 are not shown, they can be connected to a ground electrode (not shown) provided between the arms of the optical modulators of the IQ optical modulator 1000 or to any location with a ground voltage.

[0100] exist Figure 10 Although the connection with the driver IC is shown as wire connection, connection via a wiring substrate or flip-chip mounting via bumps may also be employed.

[0101] Considering the connection between the IQ optical modulator and the driver IC, power is supplied from the IQ optical modulator's driver-driving voltage application pad 1100 to the driver IC 1001 via a terminating resistor and a phase modulation unit. Ideally, voltage is applied from the midpoint of the short-circuited electrode that terminates the optical modulator's differential signal line, resulting in complete differential signal cancellation.

[0102] This is because, when the driver IC driving voltage is applied from other parts, a differential signal component that is not completely canceled out remains, so that noise may be present in the driver IC driving voltage, which may cause unstable driving of the driver IC.

[0103] Moreover, if Figure 9 As shown, in the case of an IQ optical modulator composed of two or more Mach-Zehnder modulators arranged in parallel, a wiring may be extended from the midpoint of the short-circuit conductor 303 of the terminal resistor and then integrated with the wiring for applying driver voltages to other channels.

[0104] In these configurations, the driver IC's driving voltage is applied via the terminating resistors of the optical modulator's modulated signal differential line. Consequently, the driver IC's driving current flows through the resistors in the terminating resistors. Consequently, the resistors generate significant heat during driver connection operation, affecting the operational and long-term stability of the Mach-Zehnder modulator. Ideally, the heating element (i.e., the resistor) should be at least 10 μm away from the optical waveguide of the optical modulator.

[0105] Furthermore, from the perspective of suppressing the influence of heat on the optical waveguide, it is ideal to provide a separation groove for separating heat next to the optical waveguide, but this is not an essential structure.

[0106] Furthermore, when considering connection with an open collector or open drain driver IC, the output terminal of the driver IC is open, so high-frequency signals that are not completely terminated by the terminal resistor are reflected from the terminal resistor side to the driver side.

[0107] After that, the output end of the driver is open, so it is reflected again at the driver output end. This reflection is repeated multiple times, and thus the high-frequency characteristics produce periodic fluctuations due to the length of the signal line.

[0108] If the periodic fluctuation of the high-frequency characteristics differs between channels (between the I and Q channels of the IQ modulator), smooth signal processing may not be performed, and transmission characteristics may deteriorate, which is not desirable.

[0109] For example, in Figure 2The conventional IQ optical modulator shown is based on an off-chip termination structure. Consequently, different lengths of the bends between the IQs lead to different transmission line lengths, resulting in differences in the period of fluctuations in the high-frequency characteristics between the IQs, a major factor in degrading transmission characteristics.

[0110] On the other hand, in the structure of the embodiment of the present invention, to achieve on-chip termination, a meandering line structure is employed that avoids degrading transmission characteristics, resulting in a straight high-frequency line structure without any bends. As a result, a line structure with the same length can be used across all channels, preventing periodic fluctuations in high-frequency characteristics that vary between channels. Even if fluctuations do occur, the period remains the same across all channels due to the uniform length.

[0111] In the configuration of the embodiment of the present invention, as described above, impedance matching and smooth high-frequency signal transmission are achieved across all parts of the high-frequency line constituting the modulator and the operating frequency band. Consequently, fluctuations in the reflection characteristics of the modulated signal incident from the driver side can be controlled to below -15 dB, the threshold for preventing the aforementioned fluctuations.

[0112] (Reflection characteristics of optical phase modulator with respect to frequency)

[0113] exist Figure 11 Figure 2 shows the frequency-dependent reflection characteristics of an optical phase modulator using the above-described terminal resistor structure according to an embodiment of the present invention. This optical modulator with the terminal resistor structure demonstrated excellent reflection suppression, achieving a reduction of less than -20 dB in the modulation frequency range up to approximately 53 GHz and less than -15 dB in the modulation frequency range up to 65 GHz. Therefore, it can be said that the optical modulator according to an embodiment of the present invention has terminal resistors suitable for ultra-high-speed operation exceeding 64 GBd.

[0114] Furthermore, the structure of the embodiment of the present invention has an advantage in that the chip size can be reduced because it does not have a curved structure like the conventional structure.

[0115] Of course, unlike the above embodiment, it is also possible to directly connect the capacitor-loading structure to the resistor without the connection pad. However, in this case, the capacitance of the capacitor-loading structure is insufficient, resulting in a high impedance. Therefore, it is necessary to add further capacitance to be supplemented by the connection pad. In this case, by reducing the distance between the two resistors or increasing the area of ​​the n-type or p-type semiconductor below the resistor, the overall structure of the terminal resistor can also achieve impedance matching with the capacitor-loading structure.

[0116] Furthermore, in the embodiments of the present invention, the transmission characteristics primarily focus on the differential mode, which contributes to modulation. Therefore, the termination structure for the in-phase mode is not included in the embodiments of the present invention, resulting in a structure that implements only differential mode termination. To terminate the in-phase mode, after differential termination and short-circuiting the high-frequency line, a resistor for in-phase mode termination can be further provided.

[0117] While this embodiment describes an SI-InP substrate, substrates other than semi-insulating substrates, such as n-InP and p-InP, may also be used. Unlike the case of using a semi-insulating substrate, when a semi-insulating substrate is not used, the substrate itself is n-type or p-type, eliminating the need to stack one or more n-type or pnp-type semiconductor layers on the substrate.

[0118] The present invention is not limited to InP-based materials, and for example, a material system that matches a GaAs substrate, a Si substrate, etc. can also be used.

[0119] Industrial applicability

[0120] As described above, the semiconductor optical modulator according to the embodiments of the present invention can realize an ultra-wideband semiconductor optical modulator equipped with compact termination resistors capable of accurately terminating the high-frequency line of the modulated signal on-chip across the entire broadband. When connected to an open-collector driver, for example, this optical modulator can suppress periodic fluctuations in high-frequency characteristics and further suppress fluctuations in high-frequency characteristics.

Claims

1. A semiconductor light modulator, characterized in that: have: A high-frequency line with a differential structure is arranged in parallel with the optical waveguide and is used to transmit high-frequency modulated signals; a connection pad formed continuously with the high-frequency line and in the same direction; and a terminal resistor having two rectangular resistor bodies for differentially terminating the high-frequency modulated signal from the connection pad; The high-frequency line, the connection pad, and the terminal resistor are arranged in a straight line, and the high-frequency line is terminated on-chip. The side of the terminal resistor opposite to the connection pad is short-circuited, The connection pad and the resistor forming the terminal resistor are formed on one or more layers of n-type or p-type semiconductor and one or more layers of undoped semiconductor. The optical waveguide is formed in an independent mesa shape, The semiconductor optical modulator has a high-order mode light emitting structure that emits high-order mode light of a multi-mode interference coupler of a Mach-Zehnder interferometer formed of a high mesa shape. The resistor is arranged on the high-order mode light radiating structure.

2. The semiconductor optical modulator according to claim 1, wherein At least a portion of the connection pad and the entire resistor body of the terminal resistor are formed on one or more n-type or p-type semiconductor layers and one or more dielectric layers formed on the semiconductor layers, which are sequentially formed on an insulating semiconductor substrate.

3. The semiconductor optical modulator according to claim 2, wherein The semiconductor layer below the resistor is wider than the resistor by 5 μm or more. The resistor is arranged at a position more than 10 μm away from the optical waveguide.

4. The semiconductor optical modulator according to claim 1, wherein The optical waveguide is formed in an isolated mesa shape and is connected to a multimode interference coupler. The multimode interference coupler is a multimode interference coupler for combining waves having a high-order mode light radiation structure. The resistor body of the terminal resistor is arranged on the high-order mode light radiating structure.

5. The semiconductor optical modulator according to claim 4, wherein The high-order mode light radiating structure is in the shape of arrow feathers.

6. The semiconductor light modulator according to any one of claims 1 to 5, characterized in that The high-frequency line of the semiconductor optical modulator having a differential structure for transmitting the high-frequency modulated signal is connected to an open-collector or open-drain driver IC via flip-chip mounting using bumps, leads, or a wiring substrate. A voltage for driving the driver IC is applied to a midpoint of the semiconductor optical modulator that is short-circuited after being terminated by the resistor body.

7. The semiconductor light modulator according to any one of claims 1 to 5, characterized in that The operating point adjustment electrode of the semiconductor optical modulator is of a heater-driven type and is composed of the same resistor as the terminal resistor. The upper surface of the resistor constituting the operating point adjustment electrode is covered with a dielectric.

8. The semiconductor light modulator according to any one of claims 1 to 5, characterized in that The propagation direction length of the resistor is less than 1 / 4 of the wavelength in the substrate of the operating frequency, the propagation direction length of the resistor is less than 200 μm, and the width of the resistor is greater than 5 μm.

9. The semiconductor light modulator according to any one of claims 1 to 5, characterized in that Two or more semiconductor optical modulators are arranged in parallel to form one or more IQ optical modulators. All wiring for applying voltage for driving the driver IC is integrated into one.

Images