Semiconductor device, communication apparatus, imaging system, and radar apparatus

Abstract

A semiconductor device comprising: a reference signal oscillation unit that generates a reference signal; a first injection-locked oscillator and a second injection-locked oscillator that oscillate at a frequency in a terahertz band in synchronization with the reference signal; a first modulation unit connected to the first injection-locked oscillator; a second modulation unit connected to the second injection-locked oscillator; and a signal transmission unit for transmitting a signal outputted from the first modulation unit and a signal outputted from the second modulation unit as electromagnetic waves, wherein the phase of a signal supplied from the first injection-locked oscillator to the first modulation unit is shifted with respect to the phase of a signal supplied from the second injection-locked oscillator to the second modulation unit.

Description

Semiconductor devices, communication devices, imaging systems and radar devices

[0001] The present disclosure relates to semiconductor devices, communication devices, imaging systems, and radar devices.

[0002] Development of oscillators and detectors for wireless communication systems using terahertz waves is underway. Patent Document 1 discloses a phased array radio that transmits and receives signals at frequencies of 300 GHz or higher.

[0003] JP 2023-010120 A

[0004] Generally, as the frequency increases, the wavelength shortens, and so the size of the antenna decreases. In phased array antennas, the antenna spacing must be equal to or less than the wavelength to suppress grating lobes. However, the circuit scale increases due to increased high-frequency loss. Furthermore, complex multiplexing processes such as quadrature-phase modulation (QAM modulation) are used in communication systems. In order to achieve smaller antennas and arrays, the oscillator circuits and modulator circuits must also be made smaller.

[0005] An object of the present disclosure is to provide a technology that is advantageous for communication using terahertz waves.

[0006] In view of the above problems, a semiconductor device according to an embodiment of the present disclosure is a semiconductor device comprising: a reference signal oscillating unit that generates a reference signal; a first injection-locked oscillator and a second injection-locked oscillator that oscillate at a frequency in the terahertz band in synchronization with the reference signal; a first modulation unit connected to the first injection-locked oscillator; a second modulation unit connected to the second injection-locked oscillator; and a signal transmission unit that transmits a signal output from the first modulation unit and a signal output from the second modulation unit as electromagnetic waves, wherein the phase of the signal supplied from the first injection-locked oscillator to the first modulation unit is shifted from the phase of the signal supplied from the second injection-locked oscillator to the second modulation unit.

[0007] According to the present disclosure, it is possible to provide a technology that is advantageous for communication using terahertz waves.

[0008] Other features and advantages of the present disclosure will become apparent from the following description taken in conjunction with the accompanying drawings, in which the same or similar components are designated by the same reference numerals.

[0009] The accompanying drawings are incorporated into and constitute a part of the specification, illustrate embodiments of the present disclosure, and are used to explain the principles of the present disclosure together with the description thereof. A diagram showing a configuration example of a semiconductor device according to this embodiment. A diagram showing a configuration example of a semiconductor device according to this embodiment. A diagram showing a configuration example of a semiconductor device according to this embodiment. A diagram showing a configuration example of a semiconductor device according to this embodiment. A diagram showing a characteristic of injection locking of an RTD oscillator according to this embodiment. A diagram showing a characteristic of injection locking of an RTD oscillator according to this embodiment. A diagram showing a more detailed configuration example of a semiconductor device according to this embodiment. A diagram showing a more detailed configuration example of a semiconductor device according to this embodiment. A diagram showing a more detailed configuration example of a semiconductor device according to this embodiment. A diagram showing a configuration example of a phase adjuster according to this embodiment. A diagram showing a configuration example of a phase adjuster according to this embodiment. A diagram showing a configuration example of a filter circuit according to this embodiment. A diagram showing a configuration example of a filter circuit according to this embodiment. A diagram showing a configuration example of a filter circuit according to this embodiment. A diagram showing a configuration example of a filter circuit according to this embodiment. A diagram showing a configuration example of a filter circuit according to this embodiment. A diagram showing a configuration example of a filter circuit according to this embodiment. A diagram showing an example of a filter circuit according to this embodiment. A diagram showing an example of an equivalent circuit of a semiconductor device according to this embodiment. A diagram showing a configuration example of a semiconductor device according to this embodiment. A diagram showing a configuration example of a semiconductor device according to this embodiment. FIG. 1 is a diagram showing a configuration example of a semiconductor device of this embodiment. FIG. 2 is a diagram showing a configuration example of a semiconductor device of this embodiment. FIG. 3 is a diagram showing a configuration example of a semiconductor device of this embodiment. FIG. 4 is a diagram showing a configuration example of a semiconductor device of this embodiment. FIG. 5 is a diagram showing a configuration example of a semiconductor device of this embodiment. FIG. 6 is a diagram showing a configuration example of a modulation / demodulation unit including a variable impedance element of this embodiment. FIG. 7 is a diagram showing a configuration example of a semiconductor device of this embodiment. FIG. 8 is a diagram showing a configuration example of an imaging system using the semiconductor device of this embodiment. FIG. 9 is a diagram showing a configuration example of a communication system using the semiconductor device of this embodiment. FIG. 10 is a diagram showing an example of an equivalent circuit of the semiconductor device of this embodiment. FIG. 11 is a diagram showing an application example of the semiconductor device of this embodiment. FIG. 12 is a diagram showing an application example of the semiconductor device of this embodiment. FIG. 13 is a diagram showing an application example of the semiconductor device of this embodiment.

[0010] Hereinafter, embodiments will be described in detail with reference to the accompanying drawings. Note that the following embodiments do not limit the scope of the claims. Although multiple features are described in the embodiments, not all of these multiple features are required, and multiple features may be combined arbitrarily. Furthermore, in the accompanying drawings, the same reference numerals are used to designate the same or similar components, and redundant explanations will be omitted.

[0011] Semiconductor devices according to embodiments of the present disclosure will be described with reference to FIGS. 1A to 16D. In the following description, a case where the semiconductor device is used as a transmitter in a communication system (communication device) will be described. However, this is not a limitation, and the semiconductor device of this embodiment can also be used as a receiver in a communication system (communication device). Furthermore, in this disclosure, terahertz waves refer to electromagnetic waves in a frequency range of 100 GHz or higher and 10 THz or lower, more specifically, 300 GHz or higher and 3 THz or lower. In the description of each embodiment, a description of configurations that are the same as those in other embodiments may be omitted. Furthermore, the embodiments can be modified or combined with other embodiments as appropriate.

[0012] A semiconductor device 100 according to a first embodiment will be described with reference to FIGS. 1A to 6. FIG. 1A is a diagram illustrating an outline of the configuration of the semiconductor device 100 according to this embodiment. FIG. 1B is a schematic cross-sectional view of the RTD oscillator 102a of the semiconductor device 100 taken along dashed line A-A' in FIG. 1A. FIG. 1C is a conceptual diagram illustrating the wiring of the functional blocks of the semiconductor device 100. As shown in FIG. 1A, the semiconductor device 100 includes a reference oscillator 101, RTD oscillators 102a and 102b, mixers 103a and 103b, a combiner 104, an antenna 105, and a phase adjuster 106. The semiconductor device 100 is configured to emit terahertz waves TW from the antenna 105. The terahertz waves TW are radio signals (RF signals) generated within the semiconductor device 100 and wirelessly communicated with a receiver (not shown).

[0013] The semiconductor device 100 includes a semiconductor structure for generating terahertz waves TW. More specifically, the semiconductor device 100 includes RTD oscillators 102a and 102b, each of which uses a resonant tunneling diode (RTD) as an oscillator and is combined with a resonant structure. The RTD oscillators in the RTD oscillators 102a and 102b are injection-locked oscillators that oscillate at a terahertz frequency in synchronization with a reference signal supplied from a reference oscillator 101 and output a signal. In other words, the RTDs in the RTD oscillators 102a and 102b are injection-locked oscillators that upconvert at least one of the frequency components of the reference signal supplied from the reference oscillator 101 to a higher frequency component in the terahertz band and output the signal. The RTD oscillator 102a includes two RTD units 110a and 110b, each of which includes an RTD as an oscillator. Similarly, the RTD oscillating unit 102b is provided with two RTD units 110a and 110b, each including an RTD as an oscillator. The RTD oscillating units 102a and 102b may have the same configuration, so the following description will focus on the RTD oscillating unit 102a, but the RTD oscillating unit 102b may also have a similar configuration. Furthermore, the terahertz band signal oscillated by the RTD, which is the oscillator in the RTD oscillating units 102a and 102b, is referred to as the oscillation signal.

[0014] A reference signal is supplied to the RTD oscillator 102a and the RTD oscillator 102b from a (single) reference oscillator 101, which constitutes a reference signal oscillator that generates a reference signal. The reference signal is output from the reference oscillator 101 and branched into two systems by a branching unit 108. A phase delay of π / 2 (90°) is added to one of the two branched reference signals by a phase adjuster 106, which adjusts the phase of the reference signal. Therefore, the phase of the reference signal supplied to the RTD units 110a and 110b arranged in the RTD oscillator 102a and the RTD units 110a and 110b arranged in the RTD oscillator 102b is shifted by π / 2. The RTDs that are the oscillators of the RTD oscillators 102a and 102b oscillate in the terahertz band and output oscillation signals. The oscillation signal is input to mixers 103a and 103b as local oscillation (LO) signals. Mixers 103a and 103b are each a modulation / demodulation unit, and in this embodiment, they are described as modulation units that perform modulation. However, this is not limited thereto. When semiconductor device 100 functions as a receiver as described above, mixers 103a and 103b function as demodulation units. Semiconductor device 100 includes mixer 103a, which is a modulation unit connected to an RTD oscillator arranged in RTD oscillating unit 102a, which is an injection-locked oscillator. It can also be said that semiconductor device 100 includes mixer 103a, which is a modulation unit connected to an RTD oscillator arranged in RTD oscillating unit 102a, which is an injection-locked oscillator.

[0015] Data signals 109a and 109b supplied from a baseband circuit (not shown) are modulated by mixing them with an LO signal by a mixer. Data signals 109a and 109b can be different signals depending on the data supplied to the baseband circuit. Here, the data signals are intermediate frequency (IF) signals. Two modulated signals modulated by two mixers 103a and 103b are combined into one signal by a combiner 104. The combined signal is radiated into space by an antenna 105 connected to the combiner 104 as a terahertz wave TW consisting of the modulated signal. The combiner 104 and the antenna 105 are a signal transmission unit for transmitting the modulated signals output from mixers 103a and 103b into space as electromagnetic waves. The arrangement and order of the components in this configuration are one aspect for realizing the present disclosure, and the connection configuration to which the present disclosure can be applied is not limited to this embodiment.

[0016] As shown in Figures 1A and 1B, the RTD oscillator 102a includes RTD sections 110a and 110b, each of which includes an RTD, a high-frequency oscillator. The RTD sections 110a and 110b each include a resonant conductor 111 and a GND layer 150 that form a resonant structure for oscillating at a frequency in the terahertz band with the RTD, which is an oscillator. A reference signal is supplied to the RTD oscillator 102a from a reference oscillator 101. The oscillation frequency of the RTD sections 110a and 110b is determined by the resonant structure and the inductance and capacitance of the element itself. While the RTD oscillator 102a can self-oscillate at a self-oscillation frequency on its own, the reference oscillator 101 is used as a master signal generator, and the RTD oscillator 102a operates as a slave. As a result, the RTD oscillating unit 102a oscillates in a state of being injection-locked to the oscillation signal of the reference oscillator 101, so the oscillation frequency and phase noise of the slave RTD oscillating unit 102a follow the accuracy of the master reference oscillator 101. In this way, this method makes it possible to stabilize the oscillation frequency of the RTD oscillating unit 102a, which is the LO, and reduce the phase noise. The reference oscillator 101 is a wave source for synchronizing the timing of the oscillation at the terahertz frequency of the RTD oscillating unit 102a. In this way, it can be said that not only the RTD but also the RTD units 110a and 110b that form a resonant structure with the RTD are injection-locked oscillators. Therefore, the reference oscillator 101 receives the oscillation frequency f of the terahertz band oscillation signal as a reference signal. ＴＨｚ In this embodiment, the frequency of the reference signal is the oscillation frequency f of the oscillation signal. ＴＨｚ However, the frequency of the reference signal is not limited to this.

[0017] The power injected from the reference oscillator 101 to the RTD oscillation unit 102a is the output P ＲＴＤ((3 / 16)cosωτ×ΔIΔV) or more. Here, ω is the angular frequency of the oscillation signal in the terahertz band, and τ is the carrier transit time in the semiconductor layer that constitutes the RTD units 110a and 110b. Also, ΔI and ΔV are the current difference and voltage difference, respectively, between the current peak and current valley in the negative resistance region of the RTD. There is no limit to the number of RTDs arranged in the RTD oscillating unit 102a, and the power summed up for the number of RTDs arranged may be injected from the reference oscillator 101 to the RTD oscillating unit 102a. Therefore, the output of the reference oscillator 101 is (number of RTDs arranged in the RTD oscillating unit 102a)×P ＲＴＤ In addition, the output from the reference oscillator 101 may be greater than the transmission loss from the reference oscillator 101 to each RTD in the RTD oscillation section 102a. ＲＴＤ Even if the input signal is smaller than r, locking, which will be described later, is possible. Specifically, even if the input signal is a small signal, about one thousandth of the output of one RTD, the RTD oscillator 102a may synchronize with the signal supplied from the reference oscillator 101. For example, when r=5×10 －４ Injection locking with a small signal of about 1000 kHz is possible. Here, r is called the injection ratio, and V ｉｎｊ / V ｏｓｃ The injection ratio r is expressed as the voltage amplitude (V ｉｎｊ ) and the voltage amplitude of the oscillation signal (V ｏｓｃ ) is the voltage amplitude ratio. It is expected that the effect of reducing phase noise by injection locking will be weakened as the injection ratio r decreases, but it is possible to adjust it to the degree of phase noise required for the semiconductor device 100. Therefore, in a circuit to which the present disclosure is applied, P ＲＴＤ Even when a reference oscillator 101 having an output smaller than ((3 / 16)cosωτ×ΔIΔV) is used, it is possible to synchronize the RTD oscillation unit 102a while maintaining the frequency accuracy output by the reference oscillator 101. Therefore, the signal strength output by the reference oscillator 101 is P ＲＴＤ These settings can be made without being largely dependent on the value of . These settings are similar to those for the RTD oscillator 102b.

[0018] The RTD sections 110a and 110b arranged in the RTD oscillator section 102a are supplied with a bias voltage from a bias section (not shown). The bias section supplies power to a bias pattern 112 and supplies a bias voltage to the resonant conductor 111 through a bias supply via 113. The bias pattern 112 is arranged in a dielectric section 151 arranged between the resonant conductor 111 and a GND layer 150. The bias section supplies the power required to drive the RTD sections 110a and 110b and adjusts the bias voltage applied to the RTD sections 110a and 110b. The bias voltage is selected and applied from a voltage that falls within the negative differential resistance region of the RTD. An AC shunt 115 may also be connected to the bias pattern 112. The AC shunt 115 connects a resistance and a capacitance in series and connects them to GND, thereby acting as a filter that shorts out signals in a specific frequency band. The AC shunt 115 can be appropriately arranged in various locations on the bias pattern 112 or in the bias section to suppress parasitic oscillation, which is oscillation at a frequency other than a predetermined frequency in the RTD oscillator 102a. The AC shunt 115 may be any filter that suppresses parasitic oscillation, and may also have other configurations, such as a DC shunt. Furthermore, the bias power applied from the bias section to the RTD sections 110a and 110b to drive the RTDs is direct current. Therefore, the bias section and bias pattern 112 may be directly connected, or may be connected via a low-pass filter (LPF) or the like.

[0019] FIG. 1B is a schematic diagram of the cross section A-A' of the RTD oscillator 102a of FIG. 1A. The resonant structure of the RTD oscillator 102a is a microstrip resonator, and includes a GND layer 150, which is a wide lower conductor, a resonant conductor 111, which is a line-shaped upper conductor, and a dielectric portion 151 located between the lower and upper conductors. The GND layer 150 is also called a reflector layer. The RTD portion 110a is electrically connected between the upper conductor (resonant conductor 111) and the lower conductor (GND layer 150). The RTD portion 110a includes a lower electrode 152 in contact with the GND layer 150, which is the lower conductor; an upper electrode 154 in contact with the resonant conductor 111, which is the upper conductor; and an RTD layer 153 disposed between the lower electrode 152 and the upper electrode 154, which forms the RTD. The lower electrode 152 and the upper electrode 154 are formed using conductive materials including metal layers, high-carrier-concentration semiconductor layers, vias, etc., depending on the applicable process. An insulating layer such as a passivation layer 155 for protecting devices such as the RTD section 110a may be disposed on the resonant conductor 111 and the dielectric section 151. As described above, the bias pattern 112 is disposed in the dielectric section 151 between the resonant conductor 111, which is the upper conductor, and the GND layer 150, which is the lower conductor. The bias pattern 112 is connected to the resonant conductor 111 using vias 113 for bias supply. These structures are formed on a semiconductor substrate 156 using various semiconductor processes.

[0020] In the resonant conductor 111, which is the upper conductor, the line length (resonator length) indicated by the arrow 116 in FIG. 1A is set to the wavelength λ of the signal oscillated by the RTD oscillating portion 102a. ｅｆｆ For λ ｅｆｆ Specifically, the wavelength λ in a vacuum of 0.5 THz may be set to a multiple of λ / 2. ０ is 0.6 mm, and the relative dielectric constant on the resonator is about 2, so λ ０The electrical length of λ / 2 is 0.15 mm. The resonant conductor 111 may be designed based on this line length. As a result, the resonant conductor 111 resonates at a frequency in the terahertz band, and a resonant electric field is generated in the resonant conductor 111 as a standing wave. The resonant electric field is generated in the direction along the arrow 116, with both ends of the resonant conductor 111 forming antinodes (points of maximum amplitude) of the electric field and the center of the resonant conductor 111 forming nodes (points of zero amplitude). In addition, λ ｅｆｆ Two RTD sections 110a and 110b are arranged on a resonant conductor 111 of 1 / 2 at positions facing each other with a node of the resonant electric field at the center. In this configuration, the two RTD sections 110a and 110b oscillate in a push-pull mode. The push-pull mode is an oscillation mode in which the two RTD sections 110a and 110b arranged on the resonant conductor 111 oscillate in sync with each other in a phase-inverted state (opposite phase).

[0021] Furthermore, the thickness of the dielectric portion 151 constituting the RTD oscillation portion 102a is λ ｅｆｆ / 10 or more and λ ｅｆｆ The resonant conductor 111 may also include an open stub 114 extending in the direction of the arrow 117 shown in FIG. 1A. One end of the open stub 114 is connected to the center of the resonant conductor 111, and the other end is open (released). The length of the open stub 114 in the direction of the arrow 117 is λ ｅｆｆ 114, and serves to fix the node of the standing wave oscillating in the resonant conductor 111 to the center of the resonant conductor 111. The width of the bias supply via 113 is smaller than the width of the arrow 116 of the resonant conductor 111.

[0022] Furthermore, the width of the portion (connection portion) of the bias pattern 112 connected to the bias supply via 113 is smaller (thinner) than the width of the arrow 116 of the resonant conductor 111. Furthermore, these widths may be 1 / 10 or less (λ / 10 or less) of the effective wavelength λ of the terahertz band signal standing in the resonant conductor 111. This is because arranging the bias supply via 113 and the bias pattern 112 in dimensions and positions that do not interfere with the resonant electric field of the resonant conductor 111 is suitable for improving resonance efficiency. In addition, the bias supply via 113 may be arranged in the center, which is a node of the standing wave of the resonant conductor 111. By positioning the connection point of the bias supply via 113 at the node of the standing wave in the resonant conductor 111, the impedance of the oscillation signal at the connection point is maximized. Therefore, loss due to leakage of the oscillation signal in the resonant conductor 111 from the bias supply via 113 to the bias pattern 112 can be suppressed. The bias section is adjusted appropriately to efficiently synchronize the reference oscillator 101 with the RTD oscillator sections 102a and 102b.

[0023] The RTD oscillators 102a and 102b oscillate oscillation signals synchronized with a reference signal generated by a reference oscillator 101 and phase-adjusted by a phase adjuster 106. Therefore, the LO signals output by the RTD oscillators 102a and 102b have a phase difference of 90° due to the phase adjustment of the reference signal by the phase adjuster 106. The oscillation signals oscillated by the RTD oscillators 102a and 102b are supplied as local oscillation signals (LO signals) to the mixers 103a and 103b, but there is a phase difference of 90°. Therefore, the signal on the mixer 103a side can be called an In-Phase (I) signal, and the signal on the mixer 103b side can be called a Quadrature-Phase (Q) signal. In this embodiment, to perform 16-quadrature amplitude modulation (16QAM), data signal 109a supplied to mixer 103a is modulated by four-level amplitude modulation using an oscillation signal for the I signal. Meanwhile, mixer 103b performs four-level amplitude modulation using an oscillation signal for the Q signal and data signal 109b. The modulated signal for the I signal (RF signal) and the modulated signal for the Q signal (RF signal) are combined in combiner 104 to form a signal having a total of 16 phase and amplitude combinations (4x4). The combined modulated signal is radiated and output from antenna 105 as an RF signal.

[0024] The conceptual diagram of FIG. 1C is a diagram that schematically illustrates the connections of the above-mentioned functional blocks. The semiconductor device 100 includes a reference oscillator 101, which is a reference signal oscillating unit that generates a reference signal, and RTD oscillators 102a and 102b, which are injection-locked oscillators connected to the reference oscillator 101. Furthermore, the reference oscillator 101 is connected to a branching unit 108 that branches to input a reference signal to the two injection-locked oscillators, and a phase shifter (PS) 106 that shifts the phases of the two injection-locked oscillators. The reference oscillator 101 also includes signal lines for inputting data signals 109a and 109b to mixers 103a and 103b, which are modulation units that receive the oscillation signals of the two injection-locked oscillators as LO signals. Furthermore, the modulation signal (RF signal) modulated by the modulation unit is combined by a combining unit 104 that combines multiple RF signals. The RF signal combined by the combiner is radiated into space by an antenna 105 that transmits the RF signal into space. A semiconductor device 100 can be configured using such circuit blocks. In this embodiment, a phase adjuster 106 is disposed between the reference oscillator 101, which is a reference signal oscillator, and the RTD oscillator 102b, which is an injection-locked oscillator. However, this is not limiting, and the phase adjuster 106 may be disposed between the reference oscillator 101, which is a reference signal oscillator, and the RTD oscillator 102a, which is an injection-locked oscillator. Furthermore, for example, a phase adjuster 106 may be disposed between the reference oscillator 101 and the RTD oscillator 102a, and between the reference oscillator 101 and the RTD oscillator 102b. The phase adjuster 106 may be disposed in any position that generates a predetermined phase difference (e.g., 90°) between the LO signal output from the RTD oscillator 102a and the LO signal output from the RTD oscillator 102b.

[0025] Next, using Figures 2A and 2B, we will explain the influence of the bias voltage supplied to the RTD units 110a and 110b during injection locking performed by the reference oscillator 101 on the RTD units 110a and 110b, which are oscillators. Figure 2A is a graph with the bias voltage applied to the RTD unit 110a on the horizontal axis and the frequency oscillated by the RTD oscillator 102a on the vertical axis. The process of inputting a reference signal from the reference oscillator 101 to the RTD units 110a and 110b to stabilize the frequency of the RTD oscillator 102a is called injection locking. In this embodiment, the RTD oscillator 102a utilizes the harmonic components of the signal output by the reference oscillator 101. The solid line 201 in the graph plots the correlation between the oscillation frequency and bias voltage in the RTD oscillator 102a when there is input from the reference oscillator 101, i.e., when injection locking is performed. In this embodiment, a reference signal of 0.25 THz, which is a subharmonic frequency half that of 0.5 THz, is supplied from the reference oscillator 101 to the RTD oscillator 102a. The dashed line 202 in the graph plots the correlation between the oscillation frequency and bias voltage in the RTD oscillator 102a when injection locking is not achieved. The frequency described in this embodiment is merely an example, and the frequency of the signal output by the reference oscillator 101 may be the same as that of the RTD oscillator 102a or may have a harmonic component.

[0026] First, regarding the self-oscillation frequency of the RTD unit 110a, the dashed line 202 in FIG. 2A indicates a frequency of 0.5 THz when the bias voltage is 0.7 V. It can be seen from the graph that the self-oscillation frequency decreases as the bias voltage decreases, and increases as the bias voltage increases. Next, the solid line 201, where injection locking is performed, is examined. When the bias voltage is 0.7 V, the frequency is the same as the dashed line 202 (0.5 THz) when injection locking is not performed, but even when the bias voltage is reduced to 0.69 V, the frequency remains at 0.5 THz. Similarly, even when the bias voltage is increased to 0.71 V, the frequency remains at 0.5 THz. In this way, the locking of the oscillation frequency of the RTD oscillation unit 102a by injection locking from the reference oscillator 101 is referred to as locking. Conversely, the change from a locked state to an unlocked state of the frequency is referred to as unlocking. In the graph of FIG. 2A , the locked region is the region where the bias voltage is between 0.69 V and 0.71 V, and this region is called the locking range. Within the locking range, the RTD oscillator 102a, which is the injection-locked oscillator, has its oscillation frequency fixed (locked) depending on the frequency of the reference oscillator 101. It is also known that the stronger the signal from the reference oscillator 101 input to the RTD units 110a and 110b, the wider the bias voltage range becomes, and conversely, the weaker the signal, the narrower the voltage range becomes. While FIG. 2A illustrates an example in which a 0.01 V change causes unlocking, the bias voltage value and change amount are not limited to this and can also be smaller or larger (e.g., 0.1 V). These values ​​can be set arbitrarily depending on the RTD, oscillator, and bias unit configuration used.

[0027] FIG. 2B is a graph showing the bias voltage applied to the RTD unit 110a on the horizontal axis and the phase difference between the injected reference signal of the reference oscillator 101 and the oscillation signal of the RTD oscillation unit 102a on the vertical axis when injection locking is achieved. When the bias voltage is 0.7 V, the phase difference is 0°. This indicates that the injection locking frequency and the self-oscillation frequency match (reference signal frequency f1 × N = oscillation frequency f2 of the RTD oscillation unit 102), so the phases of the signal from the reference oscillator 101 and the signal from the RTD oscillation unit 102a match. Lowering the bias voltage gradually shifts the phase difference toward -90°, while increasing the bias voltage gradually shifts the phase difference toward +90°. This phenomenon is understood to occur because the frequency difference between the reference signal from the reference oscillator 101 and the self-oscillation signal from the RTD oscillation unit 102a appears as a phase difference when locked, due to a frequency difference between the two signals. This phase difference can be calculated using the theoretical formula: φ = sin －１ (Q√(P ｏ / P ｉ ) × (f １ -f ０ ) / (f １ where φ is the oscillation phase of the RTD oscillator 102a, and P ｉ is the oscillation power of the RTD oscillator 102a. ｏ is the oscillation power of the harmonic components of the reference signal input from the reference oscillator 101 and synchronized, and Q is the Q value of the RTD oscillator 102a. The sharper the frequency spectrum, the higher the Q value, and is used as an output index of the oscillator circuit. １ is the self-oscillation frequency, and f ０ is the frequency oscillated by injection locking, i.e., the oscillation frequency of the RTD unit 110a. According to a theoretical formula, this phase difference varies within the range of -90° to +90°. That is, the phase difference within the locking range varies within the range of -90° to +90°, and the width of the locking range varies depending on the ratio between the oscillation power and the injection power of the injection-locked oscillator.

[0028] 3A to 3C are block diagrams showing details of the circuit blocks of this embodiment. Fig. 3A shows the components from the reference oscillator 101 to the outputs of the RTD oscillators 102a and 102b. Fig. 3B shows the periphery of the mixers 103a and 103b and the input section for the data signals 109a and 109b. Fig. 3C shows the combiner 104 and antenna 105, which are signal transmission sections.

[0029] The reference oscillator 101 outputs a reference signal at a subharmonic frequency of 0.25 THz. For example, a typical phase-locked loop (PLL) oscillator circuit can be used for the reference oscillator 101. The PLL oscillator circuit is configured to oscillate, at a desired high frequency, a low-frequency oscillation signal generated by an oscillation source with a frequency lower than that of the reference signal, such as a high-precision low-frequency oscillation source using a crystal oscillator. The connection port 120 is a connection port between a semiconductor substrate including the reference oscillator 101 and a semiconductor substrate including the RTD oscillators 102a and 102b and their peripheral circuits. The semiconductor substrate including the reference oscillator 101 and the semiconductor substrate including the RTD oscillators 102a and 102b can be, for example, at least partially stacked. The connection port 120 can be, for example, a three-terminal connection using a coplanar line. In a coplanar waveguide, two parallel GND lines are arranged on either side of a single signal line, sandwiching the signal line. When the reference oscillator 101 is arranged on a semiconductor substrate separate from the RTD oscillators 102a and 102b, these three wiring patterns (GND line, signal line, and GND line) are arranged in parallel and connected to a connection port 120 on the semiconductor substrate including the RTD oscillators 102a and 102b. The connection may be made using bumps, such as a flip chip, or by wire bonding from a common interposer substrate. The connection port 120 is provided with three connection pads, and a signal input to the substrate is transmitted and converted to a microstrip line. The reference signal is a high-frequency signal of several GHz or higher, higher than the signals used in ordinary electrical circuits. Therefore, the connection at the connection port 120 is appropriately implemented using a circuit configuration used for microwaves, millimeter waves, or the like.

[0030] The phase of the reference signal input from connection port 120 is adjusted using, for example, a 90° hybrid coupler 121. Hybrid coupler 121 is a circuit that functions as branching unit 108 and phase adjuster 106 in FIG. 1A. In hybrid coupler 121, a signal input from port 122 is transmitted to ports 124 and 125. A phase delay of 90° occurs in the path from port 122 to port 124, and a phase delay of 180° occurs in the path from port 122 to port 125. As a result, the phase difference between port 124 and port 125 becomes 90°, and the input reference signal is branched and output to two RTD oscillators 102a and 102b. The signal from port 122 can be considered to be always shorted to GND because the signal on the path from port 122 to port 123 and the signal on the path from port 122 to port 124 to port 125 to port 123 cancel each other out at port 123. Therefore, the reference signal input to hybrid coupler 121 is not output from port 123.

[0031] Here, the configuration of the branching unit 108 and the phase adjuster 106 is not limited to the hybrid coupler 121. Figures 4A and 4B show examples of configurations that can be used as the phase adjuster 106 and the branching unit 108 other than the hybrid coupler 121. Figure 4A shows an adjustment circuit using line length delay, and Figure 4B shows an example of an adjustment circuit using a polyphase filter. These circuits have three terminals: an input port 400, an output port 401, and an output port 402, and are designed so that when a signal input from the input port 400 is output from the two output ports 401 and 402, the phases of the signals are shifted by 90° from each other.

[0032] In the line delay type adjustment circuit of Fig. 4A, a signal input from input port 400 is branched into two, one of which passes through λ / 4 strip line 403 and is output from output port 401. The other signal passes through λ / 2 strip line 404 and is output from output port 402. Here, λ is the effective wavelength of the signal when passing through this line. Because the line lengths of the two strip lines 403 and 404 differ by λ / 4, the phase of the signal at output port 402 is output with a 90° delay from the signal at output port 401. Also, for example, it is possible for a λ / 4 strip line to be provided on one path and no strip line to be provided on the other.

[0033] The adjustment circuit of Figure 4B is a combination of filter circuits called a polyphase filter, which uses resistors 405 and 408 and capacitances 406 and 407. The circuit from input port 400 to output port 401 is a low-pass filter (LPF) circuit, and the circuit from input port 400 to output port 402 is a high-pass filter (HPF). By utilizing the fact that the output phase of the LPF lags the input signal and the output phase of the HPF leads the input signal, a phase difference of 90° is achieved between output port 401 and output port 402. As described above, phase adjuster 106 to which the present invention is applied may be any circuit that imparts a phase difference to the signals output from the two output ports, and is not limited to this.

[0034] The RTD oscillator 102a and the RTD oscillator 102b are connected to ports 124 and 125, respectively, of the hybrid coupler 121 shown in FIG. 3A. As described above, the RTD oscillator 102a and the RTD oscillator 102b are injection-locked oscillators that oscillate synchronously at a frequency in the terahertz band using the reference signal supplied from the reference oscillator 101 as a master signal source. Two RTD units, 110a and 110b, are connected to the resonant conductor 111, and oscillate in a push-pull mode in which the phases of the RTD units are reversed by 180°. The RTD oscillator 102b has a similar configuration, but the phase of the reference signal input from the reference oscillator 101 is delayed by 90° from that of the RTD oscillator 102a. The RTD oscillator 102b oscillates in synchronization with the reference signal, so that the phases of the RTD oscillators 102a and 102b are orthogonal (they have a phase difference of 90°), and oscillation signals having two terahertz band frequencies are generated. A connection port 126a to the mixer 103a is connected to the right end of the resonant conductor 111 in FIG. 3A. In other words, the connection port 126a to the mixer 103a is connected to the RTD oscillator 102a, and similarly, the connection port 126b to the mixer 103b is connected to the RTD oscillator 102b.

[0035] Here, the resonant conductor 111 is T-shaped, but may have another shape. It may be configured with a resonant structure including, for example, a resistance R, an inductance L, and a capacitance C that resonates the oscillation signal. Furthermore, the number of RTDs (RTD sections 110a and 110b) connected to the resonant conductor 111 is not limited to two, but may be one, three, or more. When the number of RTDs is small, noise due to RTD variations can be minimized. Furthermore, increasing the number of RTDs can increase the power of the oscillation signal. RTDs capable of oscillating in the terahertz band by generating a negative resistance sufficient to compensate for the resonance of the resonant conductor 111 and the loss in the circuit can be appropriately arranged.

[0036] FIG. 3B is a block diagram including peripheral circuits of mixers 103a and 103b. In this embodiment, a mixer configuration using single-ended mixers (diodes 132a and 132b) is used. The components around the mixer are divided into two systems, one for the I signal and one for the Q signal. In FIG. 3B, the connection configuration of the elements on the upper side with reference numbers ending in "a" is for the I signal, and the connection configuration of the elements on the lower side with reference numbers ending in "b" is for the Q signal. Hereinafter, including the description of the embodiments described below, the elements for the I signal will be mainly described, since the elements for the Q signal may have similar configurations. In other words, when an element for the Q signal is not described, that element may have the same configuration as the element for the I signal. Furthermore, in the above-mentioned RTD oscillators 102a and 102b, etc., when an element for the Q signal is not described, that element may have the same configuration as the element for the I signal.

[0037] As shown in FIG. 3B, the periphery of the mixer 103a includes an LO filter 127a, an IF filter 129a, a matching circuit 130a, an RF filter 131a, a diode 132a, and a matching circuit 133a. The connection ports 126a and 126b are connected to the output terminals of the RTD oscillators 102a and 102b shown in FIG. 3A. The LO signal output from the RTD oscillator 102a connected to the connection port 126a is supplied to the LO filter 127a. The LO filter 127a may be a bandpass filter (BPF) or HPF configured to transmit the LO signal generated by the RTD oscillator 102a and block the data signal 109a. The LO filter 127a may be a stripline open stub, a sectorial radial stub, or a filter circuit with an adjusted inductance L and capacitance C using a wiring pattern, which transmits only specific frequencies.

[0038] The data signal 109a is supplied from a baseband circuit (not shown) via the IF connection port 128a. As mentioned above, IF stands for Intermediate Frequency and is also called intermediate frequency. IF is often used in heterodyne systems to remove noise components from the data signal to be modulated. Furthermore, when directly modulating a data signal, such as in a direct conversion system, the data signal can be used directly. The IF signal supplied from the IF connection port 128a is, for example, a 50 GHz data signal, and is passed through the IF filter 129a to be combined with the LO signal. The IF filter 129a functions as an LPF that separates high-frequency signals such as the LO signal from the IF signal line.

[0039] The signal resulting from combining the IF signal and the LO signal is transmitted to a matching circuit 130a disposed between the mixer 103a and a node connected to a signal line carrying the data signal 109a. The matching circuit 130a adjusts the impedance of the line to match the impedance between the peripheral circuit of the mixer 103a and the peripheral circuit of the RTD oscillator 102a. An RF filter 131a is disposed between the matching circuit 130a and the mixer 103a. The RF filter 131a is disposed to suppress degradation of signal quality caused by the RF signal generated by the mixer 103a being reflected back to the LO signal. The mixer 103a is a single-ended mixer connected to GND via a single diode 132a. A signal is supplied to the anode of the diode 132a, and the cathode is connected to GND. The mixer 103a mixes the input LO signal and IF signal by utilizing the nonlinearity of the diode, and performs frequency conversion to an RF signal.

[0040] The RF signal generated by the mixer 103a is supplied to the mixer output terminal 134a via a matching circuit 133a. The matching circuit 133a is a necessary element for adjusting the impedance between the mixer 103a and a circuit connected downstream of the mixer output terminal 134a, reducing reflected waves to the mixer 103a, and improving the efficiency of RF signal transmission to the antenna 105. In this embodiment, a single-ended mixer is used as the mixer 103a, but any other element having the configuration of a so-called mixer may be used. The mixer 103a may be configured, for example, as a single-balanced mixer using multiple diodes, a double-balanced mixer, or a bidirectional diode mixer also known as a harmonic mixer. Furthermore, mixers using transistors or the like may also be appropriately arranged, and the configuration of the mixer 103a is not limited to the above. In addition, the matching circuits and filter circuits around the mixer 103a and the connection positions with the IF signal can be swapped as appropriate depending on the circuit structure and the configuration of the mixer 103a. The configuration of the Q signal is similar, but the I signal and the Q signal can be set to provide different information to the values ​​of the IF signal input from the two IF connection ports 128a and 128b. By independently arranging the two mixers 103a and 103b to modulate the different signal information of the I signal and the Q signal, it is possible to simplify the modulation by the mixers and improve the quality of the modulated signal.

[0041] FIG. 3C shows the configuration from mixer output terminals 134a and 134b to antenna 105. The circuit in FIG. 3C includes a combiner 104, which is a signal transmission unit, and antenna 105. The combiner 104 is also called a multiplexing circuit. Two types of modulated RF signals output from the two mixers 103a and 103b are expressed as I and Q signals (IQ modulated signals), and are input from mixer output terminals 134a and 134b, respectively. This configuration is called an IQ mixer. The combiner 104 may be, for example, a combiner circuit using a Wilkinson coupler. The combiner 104 has a circuit configuration including a resistor 135 connecting two input terminals and two λ / 4 lines with an output branch located midway between them. The λ / 4 lines of the Wilkinson coupler may be bent as shown in FIG. 3C to reduce the circuit area. By using a Wilkinson coupler, the input ports of the IQ modulated signals can be appropriately separated, and it is possible to prevent the IQ modulated signals from flowing back to a mixer on the opposite side from the original input. The configuration of the combiner 104 is not limited to a Wilkinson coupler, and other appropriate configurations may be used.

[0042] The antenna 105 may be a combination of a feed line 136 and a patch antenna 118. The RF signal combined by the combiner 104 is supplied to the patch antenna 118 from the feed line 136 (also referred to as a feed line). The antenna 105 is a combination of a GND layer 150 (not shown in FIG. 3C ) which is a ground conductor, the patch antenna 118, and a dielectric therebetween. The RF signal input to the antenna 105 is radiated as a terahertz wave TW vertically upward from a surface facing the ground conductor with the patch antenna 118 as a reference. A microstrip antenna or a rectangular patch antenna may be used as the patch antenna 118. The patch antenna 118 radiates an RF signal having an effective wavelength λ of the terahertz wave TW on the patch antenna. Ｒｅｆｆ , in the direction of arrow 119 in FIG. 3C. Ｒｅｆｆ3A, 3B, and 3C on a semiconductor substrate, a transmission circuit can be realized that performs QAM modulation on an LO signal having a frequency in the terahertz band and transmits it as an RF signal.

[0043] In this embodiment, the semiconductor device 100 is shown as a transmitter circuit, but a similar configuration can also be realized as a receiver circuit. A signal received from the antenna 105 is input from the opposite side of the combiner 104, which functions as a branching unit. The RF signal and LO signal are input and demodulated by the mixers 103a and 103b, and data signals are output from the IF connection ports 128a and 128b. The RTD oscillators 102a and 102b function as LO oscillators in the same way as in the oscillator circuit, and a configuration can be realized in which the two mixers 103a and 103b function as IQ mixers. The mixers 103a and 103b can be referred to as demodulators (modulator-demodulators including transmission), and the combiner 104 can be referred to as a branching unit.

[0044] Figures 5A to 5F are examples of the above-mentioned filter circuits. Figure 5A is an example of a microstrip low-pass filter, and Figure 5B is a schematic circuit diagram of the filter in Figure 5A. Figure 5C is a microstrip low-pass filter with a different shape from that in Figure 5A. Figure 5D shows a filter using stubs. Figures 5E and 5F are examples of actual filter circuits.

[0045] FIG. 5A shows an example of a low-pass filter (LPF) whose frequency band can be designed by adjusting the width of the microstrip line. The line width between input port 410 and output port 411 is changed. Capacitance sections 412, 414, and 416, which have wide lines, can be regarded as capacitances inserted between the filter circuit and GND. Inductance sections 413 and 415, which have narrow lines, can be regarded as inductances connected in series to the input and output ports. Arrows 421a to 421e shown in FIG. 5A indicate the line lengths of the respective lines, λ １ｅｆｆ It is set at / 8. １ｅｆｆ represents the effective wavelength of the signal to be blocked by this filter. Figure 5B shows an equivalent circuit representation of the capacitance sections 412, 414, and 416 and the inductance sections 413 and 415 of Figure 5A. Another possible low-pass filter pattern is shown in Figure 5C. Figure 5C shows the line pattern and the equivalent circuit superimposed on each other. Coupling of inductance component L and capacitance component C is possible at various points with respect to the T-shaped section 418 extending from line 417. A filter circuit can be constructed by appropriately designing the line and pattern width. Figure 5D shows a filter configuration using an open stub 419 and a radial stub 420. The length, size, and position of the stubs on the microstrip line from input port 410 to output port 411 can be appropriately set to match the effective wavelength of the signal to be filtered. This allows signals having a specific frequency band to be filtered out of multiple passing signals. In the matching circuits 130a, 130b, 133a, and 133b, by appropriately changing the width of the stubs and patterns, their positions on the lines, etc., it is possible to adjust the impedance for a specific frequency passing through the matching circuit and perform impedance matching between the upstream and downstream stages of the matching circuit.

[0046] The configuration shown in FIG. 5A is an example of an IF filter and can be used for the IF filters 129a and 129b shown in FIG. 3B. An IF signal is input through input port 410 and output through output port 411. This circuit is a low-pass filter (LPF), passing low-frequency IF signals (several GHz to several tens of GHz) and blocking high-frequency signals (LO and RF signals). This suppresses signal quality degradation due to noise and crosstalk in the IF signal circuit and increased LO signal loss. In this embodiment, the capacitance sections 412, 414, and 416 may each be set to a capacitance of approximately 20 to 40 fF, and the inductance sections 413 and 415 may each be set to a linewidth of approximately 50 to 100 pH. These values ​​are appropriately set to achieve a configuration that maintains low impedance in the IF signal band while achieving attenuation of less than 1 / 100 (-20 dB) at terahertz band frequencies such as LO signals. Furthermore, in order to form such a filter on the same semiconductor substrate as the RTD oscillators 102a and 102b, the size of the filter must be such that the inductance and capacitance can be formed by a semiconductor process. ＬＯｅｆｆ The line length (arrows 421a to 421e) is set to λ ＬＯｅｆｆ 5A can be used as an LPF that passes IF signals of 100 GHz or less and blocks LO and RF signals of several hundred GHz or more.

[0047] The filter circuit shown in Fig. 5E is a filter configuration according to this embodiment of the LO filter 127a shown in Fig. 3B. The filter circuit includes a capacitive coupling section 427, two capacitance sections 428 and 430, and an inductance section 429 disposed between the two capacitance sections 428 and 430, between an input port 425 and an output port 426. The capacitive coupling section 427 performs capacitive coupling (AC coupling) by overlapping the conductors of the microstrip line with an insulator interposed therebetween. This provides DC isolation between the input port 425 and the output port 426.

[0048] This capacitive coupling unit 427 must pass the LO signal and block the IF signal, which is in a low frequency band. The IF signal reaching the RTD oscillator 102a can adversely affect the stability of the oscillation of the RTD unit 110a of the RTD oscillator 102a, which is an injection-locked oscillator. Therefore, the strength of the IF signal must be lower than the strength of the oscillation signal generated by the RTD oscillator 102a, for example, less than one-hundredth of the strength. Therefore, the capacitive coupling unit 427 and the LO filter are configured to increase the impedance of the IF signal frequency band (several GHz to several tens of GHz) by approximately 10 times, or even 100 times, the frequency band in which the RTD oscillator 102a oscillates. Specifically, the capacitance of the capacitive coupling unit 427 may be approximately 20 fF, and it is configured to be considered conductive with no attenuation in the frequency band of the LO signal generated by the RTD oscillator 102a (RTD unit 110a). The capacitive coupling portion 427 may also be provided at the connection portion that connects the resonant conductor 111 and the LO filter 127a.

[0049] The capacitance section 428, inductance section 429, and capacitance section 430 of the LO filter 127a, together with the capacitive coupling section 427, constitute a bandpass filter (BPF) that performs LC resonance. The LO filter 127a uses a narrow-band filter that passes the LO signal's oscillation frequency of 0.5 THz. The narrow-band LO filter 127a can block multiple frequency signals, including the IF signal supplied as the data signal 109a and the RF signal generated by the mixer 103a. The LO filter 127a thus serves to reduce the strength of unwanted signals returning to the RTD oscillator. For example, a BPF is formed to match the frequency of the LO signal by determining the line width so that the capacitance components of the capacitance sections 428 and 430 are approximately 50 to 100 fF, respectively, and the inductance component of the inductance section 429 is approximately several pH.

[0050] FIG. 5F shows the configuration of the matching circuit 133a shown in FIG. 3B in this embodiment. In this embodiment, the matching circuit 133a also includes a wideband bandpass filter. Capacitive coupling units 433 and 435 are arranged in series between the input port 431 and the output port 432, and short stubs 434 and 436 are arranged between the two capacitive coupling units 433 and 435 and between the capacitive coupling unit 435 and the output port 432, respectively. The capacitive coupling units 433 and 435 are arranged as elements that connect the capacitance component due to capacitive coupling and the inductance component of the line in series. The short stubs 434 and 436 are used as inductance units that have an inductance component between them and GND by adjusting their lengths. Specifically, the matching circuit 133a can be designed to have an optimal inductance component while satisfying a length l<λ / 4, where λ is the effective wavelength of the RF signal passing through the matching circuit 133a. The matching circuit 133a transmits the RF signal output by the mixer 103a to the antenna 105 while filtering out unwanted signals incident from the antenna 105. The matching circuit 133a also removes unwanted harmonic components, such as intermodulation products, generated by the mixer 103a. Furthermore, the matching circuit 133a also serves to remove low-frequency signals, such as IF signals. The matching circuit 133a can also reduce transmission loss by adjusting the impedance between the mixer 103a and the combiner 104. The capacitance components of the capacitive coupling units 433 and 435 are approximately 1 to 30 fF, and the inductance components of the signals flowing through the capacitive coupling units 433 and 435 are approximately several fF to several tens of fF. The inductance components of the short stubs 434 and 436 are approximately 10 to 30 pF and are designed to resonate in the RF frequency band. By connecting multiple similar configurations, a BPF with a wide passband of tens of GHz or more can be constructed.

[0051] An equivalent circuit of the semiconductor device 100 in this embodiment will be described with reference to FIG. 6 . A bias is applied to the RTD oscillators 102a and 102b from a bias unit 600. The RTD oscillator 102a includes two RTD units 110a and 110b, the impedance of the resonant conductor 111 indicated by impedance Z4, and the impedances of the RTD units 110a and 110b indicated by impedance Z5. A reference oscillator 101 is disposed on the far left side of FIG. 6 . A transmission line 601 for a reference signal, which is an output signal of the reference oscillator 101, is connected to the reference oscillator 101 to synchronize the RTD oscillators 102a and 102b. The transmission line 601 has a transmission line impedance Z1 along with parasitic inductances L1 and L2, parasitic capacitances C1 and C2, and parasitic resistances R1 and R2.

[0052] A hybrid coupler 121 is connected between the transmission path 601 and the RTD oscillators 102a and 102b. The hybrid coupler 121 can be expressed as having an impedance Z2 connected in series to the terminal connected to the RTD oscillator 102a, and an impedance Z2 and an impedance Z3 connected in series to the terminal connected to the RTD oscillator 102b. The impedance Z3 is set so that the reference signal of the reference oscillator 101, which the hybrid coupler 121 outputs to the RTD oscillator 102b, is delayed by 90° relative to the reference signal output to the RTD oscillator 102a. The hybrid coupler 121 and the RTD oscillator 102a are connected via a capacitance C3. The hybrid coupler 121 and the RTD oscillator 102b are connected via a capacitance C4. The capacitances C3 and C4 are electrically connected by AC coupling to allow the reference signal of the reference oscillator 101 to pass through. Specifically, it may have a capacitance of about several tens to several hundreds of fF as a bandpass filter for passing a signal of one hundred to several hundreds of GHz from the reference oscillator 101 .

[0053] An LO filter 127a having an impedance Z8 is connected between the output side of the RTD oscillator 102a and the IF filter 129a. The IF signal supplied from the IF connection port 128a is connected to the output side of the LO filter 127a via an IF filter 129a having an impedance Z9. Furthermore, the output side of the LO filter 127a is connected to the mixer 103a via a matching circuit 130a having an impedance Z12 and an RF filter 131a having an impedance Z13. The same is true for the output side of the RTD oscillator 102b. The IQ modulated signals modulated by the mixers 103a and 103b are combined by a Wilkinson coupler shown as combiner 104 and radiated from the antenna 105 connected via a feed line 136 shown as impedance Z18.

[0054] The frequency of the reference signal generated by the reference oscillator 101 is set to half the frequency of the LO signal generated by the RTD oscillator 102a, and the capacitances C3 and C4 can be configured to pass the frequency of the reference signal and block the frequency of the LO signal. Specifically, the capacitances C3 and C4 can have a capacitance of approximately tens to hundreds of fF as a bandpass filter for passing signals from the reference oscillator in the range of 100 to several hundred GHz, as described above. The capacitances C3 and C4 can be realized by using an AC coupling unit with an insulator between the conductor pattern of the signal line connected to the output port of the hybrid coupler 121 and the resonant conductor 111 of the RTD oscillator 102a. The impedances Z8 to Z15 are circuits that serve as filters and impedance matching and are arranged on the signal lines connecting the RTD oscillators 102a and 102b to the mixers 103a and 103b. These filters are arranged to efficiently mix the RF signals output by the mixers 103a and 103b, the IF data signal, and the LO signal input to the mixers from the RTD oscillators. Various filters, such as capacitance, fan-shaped radial stubs, and microstrip lines, are used to pass desired frequencies and block unwanted frequencies. Furthermore, a planar antenna such as a patch antenna is used for the antenna 105, and a signal line (feed line 136) is connected at an inset from the end of the patch antenna to achieve impedance matching with the line. The RF signals output from the RTD oscillators 102a and 102b and modulated by the mixers 103a and 103b form standing waves on the patch antenna 118 of the antenna 105 and are radiated into space. A filter circuit or the like may be appropriately arranged on the strip line connecting the antenna 105 and the feed line 136.

[0055] Furthermore, a filter circuit or the like may be disposed around the RTD oscillation unit 102a as appropriate. ｉｎｊ , the wavelength of the oscillation signal generated by the RTD oscillator 102a is λ ＲＴＤ A filter circuit, λ ｉｎｊBy providing an appropriate short stub at the input end of the RTD oscillator 102a, the stub is ignored at the frequency of the reference signal, allowing only the reference signal to be transmitted. On the other hand, the LO signal, which has twice the frequency of the reference signal oscillated by the RTD oscillator 102a, has a frequency of λ ｉｎｊ The length of / 4 is λ ＲＴＤ Therefore, the LO signal is always connected to GND. ＲＴＤ Therefore, it is possible to prevent the signal from returning to the reference oscillator 101 side. Furthermore, at the output end of the RTD oscillator 102a and the connection side to the modulator, ｉｎｊ An open stub of 1 / 4 may be arranged. The open stub operates in the opposite manner to the short stub, and functions as a filter that cannot transmit the frequency of the reference signal but can transmit an LO signal that has twice the frequency of the reference signal. The stub filter configuration can be based on technology used in microwave and millimeter-wave circuits, so it can be scaled and arranged appropriately to match the terahertz band frequency.

[0056] The reference oscillator 101, mixers 103a and 103b, antenna 105, and associated peripheral circuits of the semiconductor device 100 described in this embodiment do not necessarily need to be arranged on the same semiconductor substrate. For example, a semiconductor device can be configured in which multiple components are integrated on separate semiconductor substrates. In this case, the semiconductor device 100 can be realized by connecting multiple semiconductor substrates using wire bonding, flip chip, or the like during the semiconductor substrate mounting process. The semiconductor device 100 can also be manufactured by bonding and combining semiconductor substrates in a semiconductor process, or by consolidating multiple semiconductor substrates into a single package. Furthermore, the semiconductor device 100 may employ an antenna-in-package (AiP) configuration in which an antenna and semiconductor substrates are integrated into a single package.

[0057] Baluns, conversion circuits, matching circuits, etc. may be appropriately arranged at the boundaries between multiple semiconductor substrates. Furthermore, in the above-described configuration, no active elements are arranged in the paths between the RTD units 110a and 110b (RTD oscillators 102a and 102b), which are injection-locked oscillators, and the mixers 103a and 103b, which are modulation units. However, this is not limited to this. In addition to the above-described components, the peripheral circuits may also include amplifier circuits such as a power amplifier (PA) or a low-noise amplifier (LNA), phase adjustment circuits such as a phase shifter, and switches, as appropriate. The semiconductor substrate may be based on silicon or a compound semiconductor such as indium phosphide (InP) or gallium arsenide (GaAs). For example, the reference signal oscillator (reference oscillator 101) may be formed on a silicon substrate, and the downstream portion (the antenna 105 side) from the connection port 120 shown in FIG. 3A may be formed on a compound semiconductor substrate. In this case, a silicon substrate on which the reference oscillator 101 and other components are formed and a compound semiconductor substrate on which the RTD oscillators 102a and 102b and other components are formed may be stacked to form the semiconductor device 100. However, the configuration and implementation of the semiconductor substrate of the semiconductor device 100 to which the present disclosure can be applied are not limited to those described above. In either case, it is sufficient to impart a phase difference to the output signals (oscillation signals) of the RTD oscillators 102a and 102b, which operate as injection-locked oscillators using the reference signal from the reference oscillator 101. In addition, by modulating the RTD oscillators 102 and 102b with multiple data signals according to their respective phase differences, it is possible to form a semiconductor device 100 to which the present disclosure can be applied, which can realize a transceiver circuit using the RTD oscillators 102 and 103.

[0058] In the configuration of this embodiment, the use of RTD oscillators 102a and 102b enables efficient oscillation of terahertz wave oscillation signals in a small area. Furthermore, by arranging RTD oscillators 102a and 102b near circuits such as mixers 103a and 103b, losses in transmission lines and the like can be minimized. The RTD oscillators 102a and 102b oscillate in synchronization with a reference signal input from a reference oscillator 101, with a phase shift, and output the IQ signals with a 90° phase difference. Modulated signals are generated by modulating the I and Q oscillation signals with data signals, respectively, and multiple modulated signals can be combined and radiated from antenna 105. This configuration eliminates the need for phase modulation by mixers 103a and 103b, even in multilevel modulation such as QAM modulation. This reduces the effects of phase changes due to mixer characteristics and transmission paths, thereby enabling improved signal quality through compactness and improved high-frequency characteristics.

[0059] Next, a semiconductor device 700 according to a second embodiment of the present disclosure will be described with reference to FIGS. 7A to 7C, 11, and 14. FIG. 7A is a circuit block diagram of the semiconductor device 700 according to this embodiment, and FIG. 7B is a schematic diagram of the cross-sectional structure taken along line A-A' in FIG. 7A. FIG. 7C shows a conceptual diagram of the circuit block of the semiconductor device 700. Unlike the first embodiment described above, the semiconductor device 700 does not have a mixer. Also, the configuration of bias patterns 161a and 161b for supplying bias to the RTD oscillation units 160a and 160b is different from that of the first embodiment.

[0060] As shown in FIG. 7A , two bias patterns 161 a and 161 b are independently arranged to supply bias to the RTD oscillators 160 a and 160 b, respectively. A data signal line 162 is electrically connected to the bias pattern 161 a via an AC coupling unit 163. As shown in FIG. 7B , the AC coupling unit 163 may have a metal-insulator-metal (MIM) structure. Specifically, a portion of the dielectric unit 151 is disposed between the conductor pattern of the data signal line 162 and the bias pattern 161 a, thereby capacitively coupling the data signal line 162 and the bias pattern 161 a. The data signal line 162 is connected to the signal line of the data signal 109 a at a position on the substrate (not shown), and a data signal for modulation is supplied to the bias pattern 161 a. Specifically, the bias supplied to the RTD oscillation section 160a is supplied as a direct current (DC) component from the bias section through a bias pattern 161a, and the data signal 109a is supplied as an alternating current (AC) component through a data signal line 162.

[0061] Here, the capacitance of the AC coupling unit 163 is approximately several tens to several hundreds of fF and is configured to pass data signals of several GHz or higher. This capacitance can be set to a value that sufficiently ensures the bandwidth of the data signal relative to the low impedance of the bias pattern 161a while sufficiently eliminating crosstalk due to unnecessary low-frequency components in the bias unit and data signal 109a. In the configuration shown in FIG. 7B , the data signal line 162 is arranged within the dielectric unit 151 on a layer above the bias pattern 161a using a microstrip line or a coplanar line. However, this is not limited to this. The data signal line 162 may be arranged on the same metal layer as the resonant conductor 111 and connected to a conductor pattern arranged in the AC coupling unit 163 using a via. Furthermore, the AC coupling unit 163 does not have to have the MIM structure shown in FIG. 7B . For example, in the layer in which the bias pattern 161a is arranged, the AC coupling unit 163 can also form capacitive coupling by reducing the distance between the bias pattern 161a and the data signal line 162. Therefore, superimposing the AC component of the data signal line 162 on the bias voltage of the bias pattern 161a can be achieved using various methods and is not limited to the configuration disclosed herein.

[0062] Changes in the AC component of the voltage (hereinafter sometimes referred to as the bias voltage) in the bias pattern 161a change the amplitude of the signal oscillated by the RTD oscillator 160a, which is injection-locked to the reference signal of the reference oscillator 101. This changes the amplitude of the signal oscillated by the RTD oscillator 160a, functioning as a modulation unit (modulation / demodulation unit) that performs amplitude modulation. The RTD units 110a and 110b in the RTD oscillator 160a have the characteristic that the amplitude of the signal they output changes depending on the bias voltage applied. The RTD oscillator 160b oscillates in synchronization with a reference signal that is delayed by π / 2 phase from the reference signal injected into the RTD oscillator 160a by the phase adjuster 106, generating an oscillation signal that is orthogonal to the oscillation signal of the RTD oscillator 160a. The RTD oscillator 160b has the same bias pattern and data signal line configuration as the RTD oscillator 160a, and outputs a modulated signal that is amplitude-modulated using the data signal 109b. The modulation section of this embodiment shows a form including the characteristics of the AC coupling section 163 provided on the bias patterns 161a and 161b, and the RTD oscillation section whose signal output such as amplitude changes depending on the bias voltage.

[0063] The RTD oscillators 160a and 160b synchronize with the reference signal supplied from the reference oscillator 101 and modulate in response to changes in the bias voltage of the bias patterns 161a and 161b. Therefore, the injection power of the reference signal supplied from the reference oscillator 101 must be strong enough to prevent the lock from being lost due to modulation. Specifically, in FIG. 2A, the bias voltage of the locking range was 0.69 [V] to 0.71 [V]. Therefore, the amplitude of the data signal is set so that the amplitude of the bias voltage on which the data signal is superimposed is 0.7±0.01 [V]. Because this locking range can be controlled by the injection power, it is necessary to appropriately set the locking range of the reference signal and the amplitude of the data signal.

[0064] The manner in which the data signals 109a and 109b modulate the oscillation signals of the RTD oscillators 160a and 160b is not limited to this embodiment. A typical bias-T circuit can also be formed and connected to the bias unit and bias patterns 161a and 161b. In the bias-T circuit, the direct current (DC) side can be connected to a bias power supply, the alternating current (AC) side can be connected to the data signal, and the AC / DC output side can be connected to the RTD oscillator. Furthermore, the data signal line 162 may be electrically connected directly to the resonant conductor 111, rather than to the bias pattern 161a. When the data signal line 162 is connected to the resonant conductor 111, a modulation signal can be transmitted by providing an AC coupling section between the conductor pattern of the resonant conductor 111 and the conductor pattern of the data signal line 162. In this case, the AC coupling unit can be connected in a position or a connection form having a capacitance component that makes the impedance different in two frequency bands, namely, the LO frequency band in which the RTD oscillator 102a oscillates and the IF signal frequency band of the data signal. Furthermore, modulation is also possible in the impedance adjustment unit, for example, by using the connected data signal line 162 to change the impedance of the resonant conductor 111. The impedance adjustment unit can control the impedance from the external data signal 109a by using an impedance variable element such as a varactor diode, thereby changing the impedance of the resonant conductor 111.

[0065] 11 shows a block diagram of a modulation unit 1100 (which may also be a demodulation unit or modulation / demodulation unit) when an impedance adjustment unit using a variable impedance element is used. The modulation unit 1100 includes impedances Z20 and Z21, which are λ / 4 lines corresponding to the terahertz band signal wavelength oscillated by the RTD oscillation unit 160a, connected in series. The terminal of impedance Z20 opposite the terminal connected to impedance Z21 is open and functions as an open stub. The terminal of impedance Z21 opposite the terminal connected to impedance Z20 is connected to the resonant conductor 111 via AC coupling (capacitive coupling). A variable impedance circuit 1107 is connected to the connection between Z20 and Z21. In the variable impedance circuit 1107, a capacitor C20, an impedance Z22, and a varactor diode Cv, which is a variable capacitance, are connected between GND and a node where impedance Z20 and impedance Z21 are connected. In order to adjust the capacitance of the varactor diode Cv, the signal line that supplies the data signal 109a passes through an inductor L20 and is connected to the node that connects the impedance Z22 and the varactor diode Cv.

[0066] The modulation unit 1100 is connected to the RTD oscillator 160a and can adjust the impedance of the resonant conductor 111. Changing the impedance of the resonant conductor 111 changes the frequency, phase, and intensity of the oscillation of the RTD oscillator 160a, allowing the modulation unit 1100 to perform modulation. A variable impedance element such as a variable resistor can be used as the variable capacitance of the varactor diode Cv, and a variable impedance circuit using a transistor can also be used as the modulation / demodulation unit of the RTD oscillator. In either configuration, the oscillation signal of the RTD oscillator 160a can be modulated from the data signal 109a. Furthermore, modulation of the RTD oscillators 160a and 160b can be achieved not only by amplitude modulation but also by frequency or phase modulation. In either method, the oscillation signals generated by the RTD oscillators 160a and 160b can be modulated by inputting and modulating the data signals 109a and 109b to the RTD oscillators 160a and 160b. Other circuit configurations may be the same as those in the first embodiment described above, and therefore will not be described here.

[0067] The conceptual diagram of FIG. 7C schematically illustrates functional blocks of this embodiment. In the semiconductor device 700, modulation units 701a and 701b are arranged to include two injection-locked oscillators. The modulation units 701a and 701b are connected to signal lines that input different data signals 109a and 109b, respectively, and are configured to modulate and output signals oscillated by the RTD units 110a and 110b, which are injection-locked oscillators. The RF signals modulated by the multiple modulation units 701a and 701b are combined by a combiner 104 that combines multiple RF signals, and the combined RF signal by the combiner 104 is radiated into space by an antenna 105 that radiates electromagnetic waves into space. In this embodiment, the semiconductor device 700 can be configured using such circuit blocks.

[0068] An example of the circuit configuration around the RTD oscillator in this embodiment will be described using the equivalent circuit in FIG. 14 . The connection port 120 is a three-terminal connection terminal. The central pad 170 is a signal line connection terminal, and GND line pads 711 and 712 that determine the reference voltage are arranged above and below it. A hybrid coupler 121 is connected to the pad 170 via a capacitor C30. The capacitor C30 is arranged to remove DC components from the reference signal supplied from the reference oscillator 101. An RTD oscillator 713a is connected to port 124 of the hybrid coupler 121, and an RTD oscillator 713b is connected to port 125. A patch antenna 716 is connected to the terminals of the two RTD oscillators 713a and 713b opposite the hybrid coupler 121. The patch antenna 716 also serves as a combiner that combines signals input from the two RTD oscillators 713a and 713b on the patch antenna. The combined RF signal is radiated into space by patch antenna 716 .

[0069] The two RTD oscillators 713a and 713b each include two RTD sections 717a and 718b. In the RTD oscillator 713a, the RTD section 717a, strip line 714a, resistor R30, strip line 715a, and RTD section 718a are connected to the connection terminal of the hybrid coupler 121. One terminal of each of the two RTD sections 717a and 718a and resistor R30 is connected to GND. The strip lines 714a and 715a function as resonant conductors 111, and the overall frequency of the RTD oscillator 713a is determined by resonating the oscillation signals generated by the RTD sections 717a and 718a, which act as oscillators. The resistor R30 has a DC shunt structure and serves to suppress unwanted parasitic oscillations in the RTD oscillator 713a caused by the RTD sections 717a and 718a. The RTD oscillator 713b has a similar structure.

[0070] In the RTD oscillator 713a, a signal line 719a is connected to the node connecting the resistor R30 and the striplines 714a and 715a. The signal line 719a is a signal line in which a bias DC component is superimposed on a data signal. The signal line 719a is connected to a connection port 721a via a stripline 720a. The connection port 721a is connected to another substrate via a three-terminal pad, similar to the connection port 120.

[0071] Similarly, in the RTD oscillator 713b, a signal line 719b is connected to the node connecting the resistor R31 and the striplines 714b and 715b. Unlike the RTD oscillator 713a, no DC bias component is superimposed on the signal line 719b; only the AC component of the data signal is supplied. The DC component of the signal input from the connection port 721b is filtered using a capacitor C31 disposed between the stripline 720b and the connection port 721b. In the RTD oscillator 713a, the signal input from the signal line 719a to the RTD oscillator 713a is modulated by the nonlinearity of the RTD sections 717a and 718a, converted into an RF signal, and output. At the same time, bias for the RTD sections 717a and 718a is also supplied from the signal line 719a. In the RTD oscillator 713b, a signal input from a signal line 719b to the RTD oscillator 713b is modulated by the nonlinearity of the RTDs 717b and 718b, converted into an RF signal, and output. The bias for the RTDs 717b and 718b is supplied from a signal line 719a.

[0072] While FIG. 14 illustrates an example of a transmitter circuit envisioned by this embodiment, the transmitter configuration to which the present invention is applied is not limited to this. The hybrid coupler 121 may be a stripline or rat-race circuit that provides different phase differences between the RTD oscillator 713a and the RTD oscillator 713b. The number of RTDs may be one or three or more. The more RTDs there are, the stronger the oscillation intensity becomes, but it is necessary to address issues such as power consumption, inter-element variation, and suppression of parasitic oscillation. The patch antenna 716 is not limited to a rectangular patch antenna; slot antennas, dipole antennas, loop antennas, and various other planar antennas can also be used. Resistors R30 and R31 form a DC shunt, but an AC shunt structure in which a resistor and a capacitor are connected in series may also be used. Although the signal line 719a simultaneously supplies the data signal and bias to the RTD oscillator 713a, for example, the data signal and bias may be separately supplied to the RTD oscillator 713a. Furthermore, although mixing is performed using the RTD oscillator 713a as a mixer, other modulation methods may be used in which the data signal is superimposed on the LO signal of the RTD oscillator 713a. In either case, a terahertz wave transmitter (receiver or transceiver) can be fabricated using this embodiment.

[0073] The semiconductor device 700 of this embodiment can integrate the RTD section (RTD oscillation section), which is an injection-locked oscillator, and the modulation section. This allows the semiconductor device 700 to reduce the scale of the terahertz wave generation and modulation circuits. Although this embodiment requires multiple RTD oscillation sections and multiple modulation sections, the reduced circuit scale allows the semiconductor device 700 to be miniaturized.

[0074] Next, a semiconductor device 800 according to a third embodiment of the present disclosure will be described with reference to FIGS. 8A and 8B . In this embodiment, the semiconductor device 800 includes a plurality of reference oscillators in its reference signal oscillation unit. FIG. 8A is a schematic diagram of the circuit blocks of the semiconductor device 800, and FIG. 8B is a conceptual diagram showing the wiring of the functional blocks of the semiconductor device 800. The reference signal oscillation unit of the semiconductor device 800 includes a plurality of reference oscillators 801 and 802. The reference oscillator 801 is an oscillator for supplying a reference signal to the RTD oscillation unit 102a. The reference oscillator 802 is an oscillator for supplying a reference signal to the RTD oscillation unit 102b. The reference oscillators 801 and 802 are controlled to oscillate at the same frequency. The two reference oscillators 801 and 802 are connected using a synchronization line 803, thereby synchronizing the phases of their oscillation signals. The reference oscillator 802 generates a reference signal to be supplied to the RTD oscillator 102b based on the signal output by the reference oscillator 801 (the reference signal supplied to the RTD oscillator 102a). The reference signal output from the reference oscillator 802 is delayed in phase by π / 2 (90°) by the phase adjuster 106. As a result, the reference signals input to the RTD oscillator 102a and the RTD oscillator 102b are shifted by π / 2. Therefore, the RTD oscillator 102a and the RTD oscillator 102b can oscillate at frequencies in the terahertz band in quadrature.

[0075] The synchronization line 803 directly connects the reference oscillator 801 and the reference oscillator 802 as shown in FIGS. 8A and 8B . However, this is not a limitation, and the synchronization line 803 may be a signal line commonly routed from a control circuit that controls the two reference oscillators 801 and 802. The configuration shown in FIGS. 8A and 8B is not required as long as the reference oscillators 801 and 802 can oscillate reference signals at the same frequency and phase. Alternatively, instead of the phase adjuster 106, the synchronization line 803 may be used to shift the phases of the reference oscillators 801 and 802, thereby adjusting the phase difference between the reference signals. In this case, the phase adjuster 106 is disposed on the synchronization line 803. These configurations are for generating reference signals that allow the RTD oscillators 102a and 102b to oscillate in quadrature phase, and are not limited to those of this embodiment.

[0076] The conceptual diagram of FIG. 8B schematically illustrates the above-described functional blocks. Unlike the above-described embodiments, the semiconductor device 800 includes a reference oscillator 801 and a reference oscillator 802, which are connected and synchronized by a synchronization line 803. The semiconductor device 800 also includes an RTD oscillator 102a (RTD units 110a and 110b) that is an injection-locked oscillator connected to the reference oscillator 801, and an RTD oscillator 102b (RTD units 110a and 110b) that is an injection-locked oscillator connected to the reference oscillator 802. Furthermore, a phase adjuster 106 that adjusts the phases of the two injection-locked oscillators is connected between the reference oscillator 802 and the RTD oscillator 102b. In this manner, two injection-locked oscillators are configured to generate two systems of oscillation signals with different phases, and the respective oscillation signals are input to mixers 103a and 103b to generate RF signals. In this embodiment, a semiconductor device 800 can be configured using such circuit blocks.

[0077] The semiconductor device 800 of this embodiment can easily optimize the circuit layout because the reference oscillator 801 and the reference oscillator 802 can be arranged separately. Furthermore, because there is no need to branch the output of the reference oscillator, it is possible to transmit a large amount of power to the RTD oscillation units 102a and 102b with less loss than in the above-described embodiments.

[0078] Next, a semiconductor device 900 according to a fourth embodiment of the present disclosure will be described with reference to Figures 9A and 9B. In the above-described embodiments, phase adjustment is performed on the signals supplied to the RTD oscillators 102a and 102b, but in this embodiment, phase adjustment is performed on the signal output from the RTD oscillator 102b. Figure 9A is a schematic diagram of the circuit blocks of the semiconductor device 900, and Figure 9B is a conceptual diagram showing the wiring of the functional blocks of the semiconductor device 900.

[0079] The reference signal output from the reference oscillator 101 is branched by a branching unit 108 and supplied to RTD oscillating units 102a and 102b (RTD units 110a and 110b), which are injection-locked oscillators. The RTD oscillating units 102a and 102b oscillate in synchronization with the reference signal, and therefore oscillate at the same frequency and phase in the terahertz band. The signal oscillated by the RTD oscillating unit 102a is input as is as an oscillation signal to the mixer 103a as an LO signal. The signal oscillated by the RTD oscillating unit 102b has its phase adjusted by a phase adjuster 106 and is then input to the mixer 103b as an LO signal. The input LO signals to mixer 103a and mixer 103b are shifted by π / 2 by adjustment by phase adjuster 106, so that two mixers 103a and 103b operate as mixers that respectively modulate quadrature LO signals.

[0080] The conceptual diagram of FIG. 9B schematically illustrates the functional blocks described above. Unlike the above-described embodiments, the semiconductor device 900 supplies the output of the reference oscillator 101 in phase to two injection-locked oscillators, namely, RTD oscillators 102a and 102b (RTD units 110a and 110b). The injection-locked oscillators oscillate terahertz-band oscillation signals in phase in synchronization with the reference signal. The output of the RTD oscillator 102a is supplied directly to the mixer 103a, and the output of the RTD oscillator 102b is phase-adjusted by the phase adjuster 106 and then supplied to the mixer 103b. LO signals with different phases are modulated by the mixers 103a and 103b, and then the same operation as in the above-described embodiments is performed. In this embodiment, the semiconductor device 900 can be configured using such circuit blocks.

[0081] In this embodiment, the RTD oscillator 102a and the RTD oscillator 102b can be made to oscillate in the same phase. Furthermore, the transmission path of the reference signal for injection locking from the reference oscillator 101 to the RTD oscillators 102a and 102b can be simplified. This results in improved signal quality of the signals generated by the RTD oscillators 102a and 102b, and improved quality of the reference signal due to reduced line loss from the reference oscillator 101 to the RTD oscillators 102a and 102b.

[0082] Next, a semiconductor device 1000 according to a fifth embodiment of the present disclosure will be described with reference to FIGS. 10A and 10B . The semiconductor device 1000 of this embodiment is configured to combine modulated signals not on a circuit but in the space where the signals are radiated. FIG. 10A is a schematic diagram of the circuit blocks of the semiconductor device 1000, and FIG. 10B is a conceptual diagram showing the wiring of the functional blocks of the semiconductor device 1000. In this embodiment, the signal transmission unit of the semiconductor device 1000 includes an antenna 1001a connected to a mixer 103a, which is a modulation unit, and radiates a signal output from the mixer 103a, and an antenna 1001b connected to a mixer 103b, which is a modulation unit, and radiates a signal output from the mixer 103b. The two mixers 103a and 103b generate modulated signals by modulating LO signals with data signals 109a and 109b. The modulated signal from mixer 103a is then supplied to antenna 1001a, which radiates the terahertz wave TWa into space, and the modulated signal from mixer 103b is supplied to antenna 1001b, which radiates the terahertz wave TWb into space.

[0083] Antennas 1001a and 1001b are arranged close to each other on the same substrate. The distance between the antennas may be equal to or less than the wavelength of the signals emitted by RTD units 110a and 110b, which are injection-locked oscillators, or even equal to or less than half the wavelength. Terahertz waves TWa and TWb emitted from antennas 1001a and 1001b, which are close to each other, are combined as they propagate through the same space to become a combined terahertz wave TWm. At this time, terahertz waves TWa and TWb are combined in the near-field (Fresnel region) of antennas 1001a and 1001b, and propagate as a combined terahertz wave TWm in the far-field region (Fraunhofer region). A receiving unit (not shown) can receive and demodulate this combined terahertz wave TWm. Furthermore, the phase difference between multiple antennas can be adjusted more flexibly by using a phase adjuster, in which case beamforming can be realized, which changes the beam direction of the radiated signal. For RF signals with two orthogonal phases, increasing the phase delay of one of the RTD oscillators can tilt the plane on which the orthogonal phases appear. This makes it possible to support beamforming, which tilts the beam in the direction of the antenna radiating the phase-delayed signal.

[0084] The conceptual diagram of FIG. 10B schematically illustrates the above-described functional blocks. Unlike the above-described embodiments, the semiconductor device 1000 has mixers 103a and 103b, which are modulation units, connected to separate antennas 1001a and 1001b, respectively. The combining unit 104 can be said to be located not between the mixers and the antennas, but in the space of the terahertz waves emitted by the two antennas 1001a and 1001b. This generates a combined RF signal (composite terahertz wave TWm) modulated from two injection-locked oscillators with different phases. In this embodiment, the semiconductor device 1000 can be configured using such circuit blocks.

[0085] In this embodiment, the combiner 104 shown in FIG. 1A is not required on the circuit, improving the layout flexibility of the circuit from the mixers 103a and 103b to the antennas 1001a and 1001b. Furthermore, by directly connecting the outputs of the mixers 103a and 103b to the antennas 1001a and 1001b and radiating them, RF signal loss on the circuit can be reduced. Because the RF signals output from the mixers 103a and 103b are broadband, the number of components used from the mixers 103a and 103b to the antennas 1001a and 1001b can be minimized, allowing for efficient placement of only broadband components. As the frequency of the terahertz-band antennas 1001a and 1001b increases, the antenna size becomes smaller, and the modulation section may become larger. Therefore, this embodiment can be selected based on the size and wiring flexibility of the circuit.

[0086] Next, a semiconductor device 1200 according to a sixth embodiment of the present disclosure will be described with reference to FIG. 12 . The semiconductor device 1200 according to this embodiment has a configuration including more than two RTD oscillators. FIG. 12 is a block diagram illustrating an outline of the semiconductor device 1200 according to this embodiment. A reference signal output from a reference oscillator 101 is branched into four signals by a branching unit 1201. The four branched reference signals are supplied to the four RTD oscillators 102a to 102d, respectively. The reference signal supplied to the RTD oscillator 102a is input directly to the RTD oscillator 102a without any phase adjustment. The reference signal supplied to the RTD oscillator 102b is input to the RTD oscillator 102b via a phase adjuster 1202a, which applies a delay of π / 2 to the reference signal. The reference signal supplied to the RTD oscillator 102c is input to the RTD oscillator 102c via a phase adjuster 1202b, which imparts a delay of π to the reference signal. The reference signal supplied to the RTD oscillator 102d is input to the RTD oscillator 102d via a phase adjuster 1202c, which imparts a delay of 3π / 2 to the reference signal. The frequencies of the RTD oscillators 102a, 102b, 102c, and 102d are equal to each other because they are injection-locked to the reference signal supplied from the reference oscillator 101, and the phases are shifted by π / 2 to oscillate. The oscillation signals output from the RTD oscillators 102a to 102d are input to the mixers 103a to 103d, respectively, as LO signals. Furthermore, the mixers 103a to 103d are supplied with four types of data signals 109a to 109d, and output RF signals obtained by modulating the oscillation signal and the data signal. The outputs of mixers 103a and 103b are combined in combiner 104a, and the outputs of mixers 103c and 103d are combined in combiner 104b. Furthermore, the outputs of combiner 104a and 104b are combined in combiner 104c, and the combined signal is radiated from antenna 105 as a terahertz transmission signal.

[0087] In this embodiment, the phase adjustments in the phase adjusters 1202a to 1202c are all performed with different shift amounts, but the same value can also be used for multiple phase adjusters. Since the signal strength of the terahertz waves can be increased by arranging multiple RTD oscillators 102a to 102d, it is possible to input the terahertz waves to multiple RTD oscillators with the same phase, for example, by performing no phase adjustment on two RTD oscillators and a π / 2 delay phase adjustment on the other two RTD oscillators. Furthermore, as described in the third embodiment above, multiple reference oscillators 101 may be provided, and a reference signal may be supplied from each reference oscillator 101 to each RTD oscillator. Furthermore, in this embodiment, four different data signals 109a to 109d are respectively supplied to the mixers 103a to 103d. However, a redundant configuration is also possible, for example, by inputting the same data or inverted data as the data signal 109a and the data signal 109c and modulating them. Because the RTD oscillator 102a with no phase delay and the RTD oscillator 102c with a phase delay of π are out of phase by 180 degrees, it is possible to invert and modulate one of the data signals, and then invert and combine the other of the modulated RF signals. To minimize the impact of changes in the data signal, modulation is performed using both positive (+) and negative (-) data transitions, which makes it possible to stabilize the current in the circuit and reduce crosstalk.

[0088] Furthermore, instead of modulation by the mixers 103a to 103d, modulation can also be performed using a switch that switches the signal output on and off. By switching the signal output on and off using a data signal, phase modulation, which switches between four phases and outputs, is possible. Furthermore, modulation can be performed using an LO signal from a common RTD oscillator in multiple modulators. Furthermore, as in the fifth embodiment, multiple antennas can be provided for the antenna 105 and combiners 104a to 104c, and some or all of the RF signal combinations can be performed in space. This embodiment demonstrates that the configuration of the RTD oscillator and modulator is not limited to two configurations, and that multiple redundant and efficient distribution and combinations are possible using the circuit disclosed herein. In either case, complex modulation schemes can be realized by generating oscillation signals with different phases using multiple injection-locked oscillators, modulating them with different data signals, and combining them.

[0089] In this embodiment, it is possible to reduce the variation in values ​​that can be taken by a modulation unit such as a mixer. This simplifies the configuration of the modulation unit, and improvement in signal quality can be expected. Furthermore, since it is possible to arrange multiple RTD oscillators for one antenna, an improvement in signal-to-noise ratio (SNR) can also be expected as the signal strength increases.

[0090] Next, a semiconductor device 1500 according to a seventh embodiment of the present disclosure will be described with reference to FIGS. 15A, 15B, and 16A to 16D. The semiconductor device 1500 of this embodiment has a configuration in which a plurality of active antennas, including the transmitter circuits and patch antennas of the above-described embodiments, are arranged in a two-dimensional array. FIG. 15A is a schematic diagram of the circuit block of the semiconductor device 1500 of this embodiment, and FIG. 15B is a top view of the semiconductor device 1500. FIGS. 16A, 16B, 16C, and 16D are cross-sectional views taken along lines A-A', B-B', C-C', and D-D', respectively, shown in FIG. 15B.

[0091] FIG. 15A shows a block diagram of two of the four active antennas mounted on a semiconductor device 1500. Two active antennas, AA1 and AA2, are shown in FIG. 15A. Active antennas AA3 and AA4 are omitted from FIG. 15A. Active antenna AA1 includes modulation sections 701a and 701b including injection-locked oscillators, a synthesis section 104, and an antenna 105, and its configuration is similar to that shown in FIG. 7C. Modulation sections 701a and 701b including injection-locked oscillators include RTD oscillation sections 160a and 160b (RTD sections 110a and 110b) as injection-locked oscillators, respectively. Active antennas AA2 to AA4 have the same configuration as active antenna AA1, and a common signal line is connected to these active antennas AA1 to AA4.

[0092] The common signal lines are for two systems of data signals 109a and 109b and a reference signal from reference oscillator 101. Data signal 109a is supplied to modulation units 701a and 701c, and data signal 109b is supplied to modulation units 701b and 701d. The reference signal from reference oscillator 101 is supplied to modulation units 701a and 701c including injection-locked oscillators without passing through phase adjuster 106, and is supplied to modulation units 701b and 701d including injection-locked oscillators via phase adjuster 106. Active antenna AA1 and active antenna AA2 are configured to receive the same signal, and therefore can radiate similar RF signals as terahertz waves TW into space.

[0093] As shown in Figure 16A, inside the active antenna AA1, in the RTD section 110a, a semiconductor substrate 156, a GND layer 150 which is a conductor pattern, a lower electrode 152a, an RTD layer 153a, an upper electrode 154a, and a conductor pattern 1512 are laminated in this order. The upper electrode 154a is formed inside the dielectric sections 151a to 151d, and the conductor pattern 1512 and the RTD layer 153a are electrically connected via the upper electrode 154a. Here, if the width of the upper electrode 154a is too large, the resonance characteristics of the patch antenna will deteriorate and the radiation efficiency will decrease due to an increase in parasitic capacitance. For this reason, the width of the upper electrode 154a is set to a dimension that does not interfere with the resonant electric field, typically a dimension that does not exceed the oscillation frequency f that exists in the active antenna AA1. ＴＨｚ The width of the upper electrode 154a may be configured to be equal to or smaller than 1 / 10 of the effective wavelength λ of the terahertz wave. The width of the upper electrode 154a may be small enough not to increase the series resistance, and may be reduced to, for example, approximately twice the skin depth. To reduce the series resistance to a value not exceeding 1 Ω, the width of the upper electrode 154a may typically be in the range of 0.1 μm to 20 μm.

[0094] As shown in FIG. 16B , the conductor pattern 1512 is electrically connected to the bias pattern 1510, which is a common wiring, via the via 1511, and is electrically connected to the bias unit. The bias unit may also be called a power supply circuit. The bias pattern 1510 is disposed between the dielectric portion 151b and the dielectric portion 151a. The bias pattern 1510 is drawn out from each of the conductor patterns 1512. The common bias pattern 1510 ensures sufficient wiring width, thereby reducing variations in operating voltage between antennas due to variations in wiring resistance. This stabilizes synchronization even when the number of active antennas is increased. Another advantage is that the structure around the antenna can be made symmetrical, preventing the radiation pattern from collapsing.

[0095] The vias 1511 connecting the conductor pattern 1512 and the bias pattern 1510 are connection parts for electrically and mechanically connecting the bias pattern 1510 to the conductor pattern 1512. In this way, a structure electrically connecting upper and lower layers is called a via. The GND layer 150 and the conductor pattern 1512 constitute a patch antenna. In addition to serving as components, by being connected to these vias, they also serve as electrodes for injecting current into the RTD layer 153a. In this embodiment, the vias are made of a material having a resistivity of 1×10 －６ A conductive material having a resistance of Ω·m or less can be used. Specifically, metals and metal compounds such as Ag, Au, Cu, W, Ni, Cr, Ti, Al, an AuIn alloy, and TiN can be used as the via material.

[0096] The width of the via 1511 connecting the conductor pattern 1512 and the bias pattern 1510 is smaller than the width of the conductor pattern 1512. Here, the width of the conductor pattern 1512 is the width in the electromagnetic wave resonance direction (i.e., the A-A' direction) within the active antenna AA1. Also, the width of the portion (connection portion) of the bias pattern 1510 connected to the via 1511 is smaller (thinner) than the width of the conductor pattern 1512 (active antenna AA1). Also, these widths are set according to the oscillation frequency f of the oscillation signal standing in the active antenna. ＴＨｚ The effective wavelength λ of the terahertz wave can be set to 1 / 10 or less (λ / 10 or less) of the effective wavelength λ of the terahertz wave. This is because the radiation efficiency can be improved by arranging the via 1511 and the bias pattern 1510 in a size and position that does not interfere with the resonant electric field in the active antenna.

[0097] The position of the via 1511 is determined based on the oscillation frequency f ＴＨｚ In this case, the via 1511 and the bias pattern 1510 can be arranged at a node of the electric field in the standing wave of the oscillation frequency f ＴＨｚ In other words, the via 1511 and the bias pattern 1510 are configured so that the impedance is sufficiently higher than the absolute value of the negative differential resistance of the RTD layer 153a, which is an RTD, in a frequency band around the oscillation frequency f ＴＨｚIn this case, the active antenna AA is connected to the antenna so that it has high impedance as seen by the RTD at frequency f ＴＨｚ , the path via the bias pattern 1510 for bias supply is isolated. As a result, the oscillation frequency f induced in each active antenna passes through the bias pattern 1510 and the bias section. ＴＨｚ The current of the active antenna does not affect the adjacent antenna. ＴＨｚ Interference between the electric field of the power supply member and these power supply members is suppressed.

[0098] The bias pattern 1510 is a bias supply wiring common to multiple active antennas. A bias unit can be arranged outside the chip to supply a bias to the RTD layer 153a of each active antenna. The bias unit includes a stabilization circuit for suppressing low-frequency parasitic oscillation. The stabilization circuit is configured to have an impedance lower than the absolute value of the negative resistance corresponding to the gain of the RTD layer 153a in the frequency band from DC to 10 GHz. For stabilization of relatively high frequencies from 0.1 to 10 GHz, an AC short circuit consisting of a resistor and a capacitor connected in series may be arranged for each active antenna. In this case, the capacitance is large within the above-mentioned frequency range, and in one example, is a capacitance of approximately several pF.

[0099] In this embodiment, focusing on the active antenna AA1, for example, as shown in FIG. 15B , a reference signal is injected from the reference oscillator 101 to port 1 via a control line IL1. Furthermore, a reference signal obtained by shifting the phase of the output of the reference oscillator 101 by a phase adjuster 106 is injected to port 2 via a control line IL2. In the example of FIG. 15B , a stub 1513 is configured to extend from port 1 to the opposite side of the RTD layer 153a. Here, the stub 1513 is set to a length that is one-quarter of the wavelength of the terahertz wave and is grounded to GND via an MIM capacitor 1514. Furthermore, the reference signal injected to port 1 is injected into the RTD layer 153a via a control line RL that extends to the opposite side of the stub 1513. The length of the control line RL is designed to be one-quarter of the wavelength of the terahertz wave so as to function as an RTD resonator.

[0100] The data signal 109a is supplied to the RTD unit 110b via a control line DL1. The control line DL1 transmits only the AC component of the data signal, so it may be connected to the RTD unit 110b using a frequency filter according to the data signal or capacitive coupling for DC cutting. The data signal 109b is supplied to the RTD unit 110a via a control line DL2. In this embodiment, the same control line DL2 parallel to the B-B' direction is connected to the active antennas AA1 and AA3. This is because the frequency f of the data signal Ｄ This is because the difference in the line length of the control line DL2 is sufficiently short compared to the wavelength of the data signal, since the frequency of the control line DL2 is several tens of GHz. On the other hand, by designing the wiring path using equal-length wiring, it is possible to suppress the variation in the signal input to each active antenna, and it is also possible to wire each active antenna independently.

[0101] As shown in FIG. 16D , conductor pattern 1515 is disposed between dielectric portion 151d and dielectric portion 151c. Conductor pattern 1515 is used as a wiring pattern for control line IL1, control line IL2, stub 1513, control line RL, etc., and is provided so that a reference signal can be injected into each active antenna. For the sake of explanation, FIG. 15B illustrates control lines IL1, IL2, control line RL, and stub 1513 as separate parts. However, these are a continuous conductor and are not necessarily separated as shown. Stub 1513 is conductor pattern 1515 shown in FIG. 16D , but is connected to conductor pattern 1516 and grounded to GND layer 150, which is a conductor pattern, via MIM capacitance 1517. As shown in FIG. 16D , control line RL extending to the RTD portion of conductor pattern 1515 is connected to upper electrode 154d and configured to inject a reference signal into RTD layer 153d.

[0102] As shown in FIG. 16C , a conductor pattern 1518 is disposed between the dielectric portion 151c and the dielectric portion 151b. The conductor pattern 1518 is used as a wiring pattern for the control lines DL1 and DL2, and is provided so that data signals for modulation can be input to each active antenna. The control lines DL1 and DL2 may be provided with capacitive coupling sections for DC blocking and filters for passing signals in relatively low frequency bands, such as data signals, at various locations. In this case, as shown in FIG. 16D , an MIM capacitor grounded to the GND layer 150 may also be disposed. The conductor pattern 1518 is connected to the upper electrode 154c of the RTD portion 110c and configured to perform modulation using the RTD layer 153c.

[0103] FIG. 15B is a top view of semiconductor device 1500. Semiconductor device 1500 forms an antenna array, with four active antennas AA1 to AA4 arranged in a 2×2 matrix. These active antennas AA1 to AA4 each radiate an RF signal represented by a terahertz wave TW into space. The number of active antennas is not limited to four; the active antennas may be arranged in a 3×3 matrix or a 2×4 matrix. A transmitter to which the present disclosure is applied can be configured as long as there are M×N active antennas (M and N are natural numbers). The RF signals radiated from these active antennas AA1 to AA4 are signals oscillated by RTD oscillators synchronized by reference oscillator 101. Therefore, the RF signals from each active antenna can be synchronized with high precision, resulting in the radiation of terahertz waves into space with higher intensity than a single active antenna. This improves the gain of the antenna array.

[0104] Although a common bias pattern 1510 is used in this embodiment, a bias may be supplied to each active antenna via a separate bias pattern. In this case, overcrowding of the wiring pattern can be avoided by forming through-vias in a semiconductor substrate or the like and stacking the RTD with a CMOS integrated circuit. As shown in FIG. 2B , the phase of the RTD oscillator can be varied within the locking range by changing the bias voltage. That is, the phase of each active antenna can be varied depending on the frequency difference between the actual oscillation frequency locked by the reference signal of the reference oscillator and the self-oscillation frequency determined by the bias voltage of the RTD oscillator. For example, by applying different bias voltages to the RTD oscillators 160a and 160b arranged in active antenna AA1 and the RTD oscillators 160c and 160d arranged in active antenna AA2, the RTD oscillators can be operated at different self-oscillation frequencies. This enables beamforming in active antenna arrays arranged in one-dimensional and two-dimensional planes.

[0105] In this embodiment, the functions of the injection-locked oscillator and the modulator can be integrated into the active antenna, making it possible to realize a compact transmitter. An IQ mixer can be used with one active antenna, and the active antennas can be arranged in an array. This has the added benefit of providing high-intensity power that cannot be output by individual active antennas, and sharp directivity. This allows the overall transmitter configuration to be made more compact.

[0106] Other Embodiments Although the embodiments of the present invention have been described above, the present invention is not limited to these embodiments, and various modifications and changes are possible within the scope of the gist of the present invention.

[0107] For example, while the above-described embodiments have been described using a transmitter in a wireless communication device as an example, the application is not limited thereto and may also be, for example, a receiver or radar device. A radar device requires constant sweeping using a chirp signal, but the circuit of the present invention, such as that shown in FIG. 1, can be used to sweep changes in a modulated data signal. This makes it possible to determine the delay of the received signal, and by performing absolute value and phase conversion using an IQ signal, it is possible to use the circuit to accurately detect distance and azimuth. In other words, a wave radiated from a transmitter equipped with the semiconductor device of each of the above-described embodiments and reflected by an object can be detected by a receiver, and the distance and azimuth to the object can be measured from the radiated wave and reflected wave.

[0108] In addition, although the above-described embodiments use an RTD oscillator having a resonant structure with an RTD as an injection-locked oscillator, elements or circuits with gain other than an RTD can also be used as long as the oscillator oscillates in the terahertz band. Furthermore, even in the resonant structure, a resonant circuit may be configured by appropriately arranging lumped constant circuits such as inductors and capacitors to control oscillation. Furthermore, although the above-described embodiments use a square patch antenna as the terahertz wave antenna, the shape of the antenna is not limited to this. For example, a patch conductor having a rectangular, triangular, polygonal, circular, or elliptical shape, a planar antenna such as a loop antenna, a log-periodic antenna, or a Vivaldi antenna, or a horn antenna may also be used as the antenna.

[0109] Furthermore, for example, in the above-described embodiments, an example in which two RTDs are arranged for one RTD oscillator is described, but it is also possible to arrange only one or more RTDs and oscillate them within the resonator. Using only one RTD allows oscillation to be controlled without considering RTD variations, and arranging multiple RTDs allows for the construction of a high-power oscillator. Furthermore, while the description has been given with one mixer associated with one RTD oscillator, connections such as 1:N, N:1, or N:N (N is a natural number) may also be used. Components such as combining sections and branching sections may also be appropriately arranged in various locations in the circuit in a redundant configuration. Furthermore, the inductance and capacitance constants set in the filter circuits used in the above-described embodiments may be selected appropriately depending on the frequency of the signals to be passed or blocked.

[0110] Furthermore, although the structure of the RTD oscillator using a laminated structure has been described, the present invention is not limited to this, and the above discussion can also be applied to oscillator devices that do not use a laminated structure.

[0111] Furthermore, the following combinations may be used as materials for the RTD: GaAs / AlGaAs and GaAs / AlAs, InGaAs / GaAs / AlAs formed on a GaAs substrate InGaAs / InAlAs, InGaAs / AlAs, InGaAs / AlGaAsSb formed on an InP substrate InAs / AlAsSb and InAs / AlSb formed on an InAs substrate SiGe / SiGe formed on a Si substrate The above-mentioned structures and materials can be selected appropriately depending on the desired frequency, etc.

[0112] Furthermore, circuits with frequencies relatively lower than the terahertz band, such as reference oscillators and data signals, can be connected to semiconductor substrates using general Si-CMOS circuits for transmission. Also, antennas and the like can be patterned on different semiconductor substrates or connected via connectors to mechanical structures fabricated by machining techniques, which can be used as a method for reducing material costs and manufacturing costs.

[0113] Next, a case where the semiconductor device of each of the above-described embodiments is applied to a terahertz camera system (imaging system) will be described. The following description will be made with reference to FIG. 13A . The terahertz camera system 1300 includes a transmitter 1301 that emits terahertz waves TW and a detector 1302 that detects the terahertz waves TW. The terahertz camera system 1300 further includes a controller 1303 that controls the operation of the transmitter 1301 and the detector 1302 based on an external signal, and processes or externally outputs an image based on the detected terahertz waves. The semiconductor device of each of the embodiments may be used as the transmitter 1301 or the detector 1302. The terahertz waves emitted from the transmitter 1301 are reflected by an object 1305 and detected by the detector 1302. A camera system including such a transmitter 1301 and detector 1302 may also be called an active reflection imaging camera system. The camera system can also be applied to a transmission imaging system in which the transmitter 1301 and the detector 1302 are arranged opposite each other and a subject is placed between them, thereby observing terahertz waves that have passed through the subject. Furthermore, in a passive camera system in which the transmitter 1301 is not provided, the semiconductor device of each of the above-described embodiments can be used as the detector 1302. By using the semiconductor device of each of the embodiments that can modulate and demodulate IQ signals, the detection sensitivity of the camera system can be improved, and high-quality images can be obtained.

[0114] Furthermore, a case where any of the devices according to the above-described embodiments is applied to a terahertz communication system (communication device) will be described below with reference to FIG. 13B . The semiconductor device can be used as a component of the communication system. Possible communication systems include a simple ASK system, a superheterodyne system, and a direct conversion system. A superheterodyne communication system includes, for example, an antenna 1400, an amplifier 1401, a mixer 1402, a filter 1403, a mixer 1404, a converter 1405, a digital baseband modulator / demodulator 1406, and local oscillators 1407 and 1408. The RTD oscillator 102a of the above-described embodiment can be used as the local oscillator 1407, the antenna 105 as the antenna 1400, and the mixer 103a of FIG. 1 as the mixer 1402.

[0115] In the case of a receiver, terahertz waves received via antenna 1400 are converted into an intermediate frequency signal by mixer 1402. Then, mixer 1404 converts the signal into a baseband signal, and converter 1405 converts the analog waveform into a digital waveform. The digital waveform is then demodulated at baseband to obtain a communication signal. In the case of a transmitter, the communication signal is modulated, and then converted from a digital waveform into an analog waveform by converter 1405. Then, the signal is frequency-converted via mixers 1404 and 1402, and output from antenna 1400 as a terahertz wave.

[0116] The direct conversion communication system includes an antenna 1400, an amplifier 1411, a mixer 1412, a modulator / demodulator 1413, and a local oscillator 1414. In the direct conversion system, during reception, the mixer 1412 directly converts the received terahertz wave into a baseband signal. During transmission, the mixer 1412 converts the baseband signal to be transmitted into a terahertz signal. The other configurations are similar to those of the superheterodyne system. The devices according to the above-described embodiments can simplify the local oscillator and mixer configurations of the semiconductor device and can support QAM modulation, such as generation of IQ signals. Therefore, by using the semiconductor device according to the above-described embodiments capable of IQ modulation, wireless quality, such as the signal-to-noise ratio, can be improved in the communication system, enabling large-capacity information transmission over a wide coverage area at low cost. Furthermore, if frequency sweeping can be performed using the same local oscillator for transmission and reception, it will be possible to measure distance from the delay and phase information of the transmitted signal and the reflected wave as a radar device. By applying this invention, wireless quality such as the signal-to-noise ratio can be improved, and distance measurement accuracy can be improved, all at low cost.

[0117] The technical ideas derived from this disclosure are not limited to the disclosed exemplary embodiments, but are intended to encompass various modifications to the exemplary embodiments, or the replacement of equivalent structures or functions, etc. The scope of the following claims should be accorded the broadest interpretation so as to encompass all such modifications and equivalent structures and functions.

[0118] This application claims priority based on Japanese Patent Application No. 2024-146771, filed August 28, 2024, the entire contents of which are incorporated herein by reference.

Claims

1. A semiconductor device comprising: a reference signal oscillator that generates a reference signal; a first injection-locked oscillator and a second injection-locked oscillator that oscillate at a frequency in the terahertz band in synchronization with the reference signal; a first modulator connected to the first injection-locked oscillator; a second modulator connected to the second injection-locked oscillator; and a signal transmission unit that transmits the signal output from the first modulator and the signal output from the second modulator as electromagnetic waves, wherein the phase of the signal supplied from the first injection-locked oscillator to the first modulator is shifted from the phase of the signal supplied from the second injection-locked oscillator to the second modulator.

2. The semiconductor device of claim 1, wherein each of the first injection-locked oscillator and the second injection-locked oscillator includes a resonant tunneling diode.

3. A semiconductor device according to claim 1 or 2, characterized in that a phase adjuster for adjusting the phase of the reference signal is disposed between the reference signal oscillating section and the second injection-locked oscillator.

4. A semiconductor device according to claim 3, characterized in that a phase adjuster other than the phase adjuster for adjusting the phase of the reference signal is arranged between the reference signal oscillating section and the first injection-locked oscillator.

5. A semiconductor device according to claim 1 or 2, characterized in that a phase adjuster for adjusting the phase of the reference signal is disposed between the second injection-locked oscillator and the second modulation section.

6. The semiconductor device according to any one of claims 3 to 5, wherein the phase adjuster includes a hybrid coupler, a rat-race circuit, a line delay type phase shifter, or a polyphase filter.

7. The semiconductor device according to any one of claims 1 to 6, wherein each of the first modulation section and the second modulation section includes a mixer.

8. The semiconductor device of claim 7, wherein the mixer comprises a single-ended mixer, a single-balanced mixer, a bidirectional diode mixer, or a double-balanced mixer.

9. A semiconductor device according to claim 7 or 8, characterized in that a first signal line for supplying a data signal from a baseband circuit is electrically connected between the first injection-locked oscillator and the first modulation section, and a second signal line for supplying a data signal from the baseband circuit is electrically connected between the second injection-locked oscillator and the second modulation section.

10. A semiconductor device according to claim 9, characterized in that a matching circuit is arranged between the node to which the first signal line is connected and the first modulation section, and between the node to which the second signal line is connected and the second modulation section.

11. A semiconductor device according to any one of claims 7 to 10, characterized in that a filter circuit is arranged between the first injection-locked oscillator and the first modulation section, and between the second injection-locked oscillator and the second modulation section.

12. A semiconductor device according to any one of claims 1 to 11, wherein each of the first injection-locked oscillator and the second injection-locked oscillator comprises a resonant conductor for oscillating at a frequency in the terahertz band.

13. The semiconductor device of claim 12, further comprising a bias pattern connected to the resonant conductor of the first injection-locked oscillator and the resonant conductor of the second injection-locked oscillator in order to supply power to the first injection-locked oscillator and the second injection-locked oscillator.

14. The semiconductor device according to claim 13, wherein each of the resonant structures including the first injection-locked oscillator and the second injection-locked oscillator includes the resonant conductor and a lower conductor, and a dielectric portion in which the bias pattern is arranged is arranged between the resonant conductor and the lower conductor.

15. The semiconductor device according to claim 14, wherein an AC shunt is connected to the bias pattern.

16. The semiconductor device according to any one of claims 1 to 6, characterized in that: each of the first injection-locked oscillator and the second injection-locked oscillator comprises a resonant conductor for oscillating at a frequency in the terahertz band; the first modulation unit includes a first bias pattern connected to the resonant conductor of the first injection-locked oscillator for supplying power to the first injection-locked oscillator, and a first data signal line capacitively coupled to the first bias pattern and supplied with a data signal from a baseband circuit; and the second modulation unit includes a second bias pattern connected to the resonant conductor of the second injection-locked oscillator for supplying power to the second injection-locked oscillator, and a second data signal line capacitively coupled to the second bias pattern and supplied with a data signal from the baseband circuit.

17. The semiconductor device according to claim 16, wherein each of the resonant structures including the first injection-locked oscillator and the second injection-locked oscillator includes the resonant conductor and a lower conductor, and a dielectric portion in which the first bias pattern and the second bias pattern are arranged is arranged between the resonant conductor and the lower conductor.

18. A semiconductor device according to any one of claims 1 to 6, characterized in that each of the first injection locked oscillator and the second injection locked oscillator comprises a resonant conductor for oscillating at a frequency in the terahertz band, the first modulation unit includes a first impedance adjustment unit that is capacitively coupled to the resonant conductor of the first injection locked oscillator and that changes the impedance of the resonant conductor of the first injection locked oscillator in response to a data signal supplied from a baseband circuit, and the second modulation unit includes a second impedance adjustment unit that is capacitively coupled to the resonant conductor of the second injection locked oscillator and that changes the impedance of the resonant conductor of the second injection locked oscillator in response to a data signal supplied from the baseband circuit.

19. A semiconductor device according to any one of claims 12 to 18, characterized in that each of the resonant structures comprising the first injection-locked oscillator and the second injection-locked oscillator is a microstrip-type resonator.

20. A semiconductor device according to any one of claims 12 to 19, characterized in that the resonant conductor comprises an open stub of λ / 4.

21. A semiconductor device according to any one of claims 1 to 20, wherein the reference signal is a signal having a subharmonic frequency of the frequency at which the first injection-locked oscillator and the second injection-locked oscillator oscillate.

22. A semiconductor device according to any one of claims 1 to 21, characterized in that the first injection-locked oscillator and the second injection-locked oscillator up-convert at least one of the frequency components of the reference signal to a higher frequency component and output the up-converted component.

23. A semiconductor device according to any one of claims 1 to 22, characterized in that the signal transmission section includes a combining section that combines the signal output from the first modulation section and the signal output from the second modulation section, and an antenna that radiates the signal combined by the combining section.

24. The semiconductor device according to claim 23, wherein the combining section includes a Wilkinson coupler.

25. A semiconductor device according to any one of claims 1 to 22, characterized in that the signal transmission section includes an antenna connected to the first modulation section and radiating the signal output from the first modulation section, and an antenna connected to the second modulation section and radiating the signal output from the second modulation section.

26. A semiconductor device as described in claim 25, characterized in that the distance between the antenna connected to the first modulation section and the antenna connected to the second modulation section is equal to or less than the wavelength of the signals oscillated by the first injection-locked oscillator and the second injection-locked oscillator.

27. A semiconductor device according to claim 25 or 26, characterized in that the signal radiated from the antenna connected to the first modulation section and the signal radiated from the antenna connected to the second modulation section are combined in a Fresnel region.

28. The semiconductor device according to any one of claims 23 to 27, wherein the antenna comprises a patch antenna.

29. The semiconductor device according to any one of claims 1 to 28, further comprising a first semiconductor substrate on which the reference signal oscillator is arranged, and a second semiconductor substrate on which the first injection-locked oscillator and the second injection-locked oscillator are arranged, wherein at least a portion of the first semiconductor substrate and the second semiconductor substrate are laminated.

30. A semiconductor device according to any one of claims 1 to 29, characterized in that no active elements are arranged in the path between the first injection-locked oscillator and the first modulation section, and in the path between the second injection-locked oscillator and the second modulation section.

31. A semiconductor device according to any one of claims 1 to 30, comprising: a third injection-locked oscillator and a fourth injection-locked oscillator that oscillate at a frequency in the terahertz band in synchronization with the reference signal; a third modulation unit connected to the third injection-locked oscillator; a fourth modulation unit connected to the fourth injection-locked oscillator; and a signal transmission unit separate from the signal transmission unit for transmitting a signal output from the third modulation unit and a signal output from the fourth modulation unit as electromagnetic waves, wherein the third injection-locked oscillator and the fourth injection-locked oscillator operate such that the frequency of self-oscillation differs from the frequency of self-oscillation of the first injection-locked oscillator and the second injection-locked oscillator.

32. A semiconductor device according to any one of claims 1 to 31, characterized in that the reference signal oscillating section includes a reference oscillator common to the first injection locked oscillator and the second injection locked oscillator for supplying the reference signal to the first injection locked oscillator and the second injection locked oscillator.

33. A semiconductor device according to any one of claims 1 to 31, characterized in that the reference signal oscillating section includes a first reference oscillator that supplies the reference signal to the first injection-locked oscillator, and a second reference oscillator that supplies the reference signal to the second injection-locked oscillator based on a signal output by the first reference oscillator.

34. A semiconductor device comprising: a reference signal oscillator that generates a reference signal; and a first injection-locked oscillator and a second injection-locked oscillator, each of which includes an oscillator that oscillates at a frequency in the terahertz band in synchronization with the reference signal, wherein the first injection-locked oscillator outputs a first signal obtained by modulating a first data signal supplied from a baseband circuit; and the second injection-locked oscillator outputs a second signal obtained by modulating a second data signal supplied from the baseband circuit; and wherein the phase of the first signal and the phase of the second signal are shifted.

35. The semiconductor device of claim 34, wherein the first injection-locked oscillator and the second injection-locked oscillator each include a resonant tunneling diode.

36. A semiconductor device as described in claim 34 or 35, further comprising a patch antenna for transmitting the first signal and the second signal, wherein the reference signal oscillator is disposed on a first semiconductor substrate, and a plurality of active antennas each including the first injection-locked oscillator, the second injection-locked oscillator, and the patch antenna are disposed in a two-dimensional array on a second semiconductor substrate.

37. A communication device comprising: a transmitter comprising a semiconductor device according to any one of claims 1 to 36; and a receiver for detecting a signal emitted from the transmitter.

38. An imaging system comprising a semiconductor device according to any one of claims 1 to 36, a transmitter for emitting a signal towards a subject, and a detector for detecting the signal reflected from or transmitted through the subject.

39. A radar device comprising: a transmitter having a semiconductor device according to any one of claims 1 to 36; and a receiver that detects waves emitted from the transmitter and reflected by an object, wherein the radar device measures distance from the emitted waves from the transmitter and the reflected waves.

Images