JP2006261386A - Electromagnetic wave generating device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an electromagnetic wave generating device in which plasma waves of low intensity are excited. <P>SOLUTION: The electromagnetic wave generating device includes a semiconductor substrate 37, a semiconductor nano wire composed of two or more high-Eg regions 35 and low-Eg regions 34, an insulating film 36, a gate electrode 31, a drain electrode 33, and a source electrode 32. The high-Eg regions 35 are arranged half the period of the wavelength of plasma waves excited in the semiconductor nano wire, so that the strength of the plasma waves is multiplied by the number of the high-EG regions and the number of the gate electrodes. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a semiconductor element that operates in a high frequency region of about 1 THz.

  Conventional semiconductor elements have been improved in operation speed mainly by miniaturizing the element structure. For example, a high electron mobility transistor (HEMT) effectively suppresses impurity scattering by using a two-dimensionally distributed two-dimensional electron gas, but it has been reported that a high-speed operation up to several hundred GHz is reported. The formation of the gate electrode requires an ultrafine processing technique of at least 30 nm or less. This is because charge (electron) movement is used for signal transmission, and the movement speed is limited by the electron saturation speed of the material, so that delay due to channel travel time cannot be avoided.

  As a prior art for this problem, several semiconductor elements that do not directly transfer charges in signal transmission have been proposed. One is an element that utilizes plasma oscillation of a high-concentration electron fluid in an FET channel proposed by M. Shur et al. (See Non-Patent Documents 1 and 2). The electron density wave corresponding to the oscillation frequency of the plasma oscillation is hereinafter referred to as a plasma wave, and an element in which a plasma wave is formed in a channel is defined as a plasma wave element. FIG. 8 shows the structure of the plasma wave element shown in Non-Patent Document 1. It has a structure similar to that of a high electron mobility field effect transistor (HFET, HEMT), and a plasma wave is formed in the two-dimensional electron gas 54. In Non-Patent Document 2, it is observed that a plasma wave that vibrates at about 1 THz is formed in the two-dimensional electron gas 54 of FIG. 8 and a weak (about several nW) electromagnetic wave is radiated.

  Ideshita also proposed an electromagnetic wave amplifying device using the two-fluid instability phenomenon that occurs in a system in which two fluids, a heavy fluid consisting of holes and electrons, or heavy and light electrons, as shown in FIG. (See Patent Document 1). By using the slow wave circuit 61 having a comb electrode structure, the terahertz wave introduced to the input unit 65 is amplified and extracted from the output unit 66.

  On the other hand, since the present invention uses a nanowire structure as shown in FIG. 1 (a), the general technical level of a device using plasmons in the nanowire is shown. A structure (FIG. 10) using nanowires in an optical modulator using plasmons has been proposed (see Patent Document 2). This prior art does not have the object of generating terahertz waves.

Furthermore, since the present invention provides a plurality of nanowires having a periodically high band gap (Eg) region, the general state of the art of nanowire heterostructures is shown. These prior arts do not have the purpose of generating plasma waves. As a nanowire having a periodic high Eg region, a nanowire structure in which Si and Si _1-x Ge _x (0.12 <x <0.25) are alternately arranged by Joan M. Redwing et al. Non-Patent Document 3) has been reported. Further, as shown in FIG. 11 (b), LJ Lauhon et al. Reported a FET structure in which the periphery of a Si nanowire is concentrically surrounded by a Ge layer (see Non-Patent Document 4).
JP-A-8-139306 JP 2003-66386 A Phys. Rev. Lett. 71 (1993) 2465 Appl. Phys. Lett. 84 (2004) 2331 Proc. SPIE 5361 (2004) 52 Nature 420 (2002) 57

  In Non-Patent Documents 1 and 2, although it is possible to operate at a high frequency in theoretical studies, electromagnetic waves are actually very weak or the operating range is limited. In particular, the structure according to Non-Patent Document 2 is the first case where generation of electromagnetic waves at a high frequency is observed without irradiating light or the like from the outside. It can be considered as one of the factors that prevent it. Further, as shown in FIG. 9, Patent Document 1 is not an oscillator but an amplifier, and has a problem that the structure is extremely complicated and it is necessary to introduce an electromagnetic wave for amplification from the outside, which is not practical. It was.

  The reason why the output becomes extremely small is that the free end and the fixed end of the plasma wave are formed at both ends of the gate electrode. In order to solve this problem, a method was devised in which the free end and the fixed end of the plasma wave are formed without using both ends of the gate electrode. In order to explain in detail, problems derived from the above-mentioned problems are shown below.

  The first problem is that the length of the gate electrode is ¼ of the period of the plasma wave. As a result, the length of the gate (corresponding to the gate length of the FET) was shortened to 60 nm and the capacitance formed by the gate electrode was reduced, so the energy of the plasma wave was reduced and the intensity of the electromagnetic wave was reduced. .

  The second problem is that since the length of the gate electrode is as short as 60 nm, it is extremely difficult to form an electrode having a gate width of 100 μm in the width direction without variation in the gate length. For this reason, the frequency of the plasma wave varies, and the intensity of the electromagnetic wave having the target frequency is reduced. Therefore, a method for controlling the frequency of the plasma wave by a method other than the length of the electrode is required.

  Further, as a third problem, when the electromagnetic wave generating element having the structure of Non-Patent Document 2 is operated, it is necessary to short-circuit the gate electrode 51 and the source electrode 52 to create a plasma wave node. . As a result, a sufficient voltage is not applied to the source side of the gate electrode, resulting in a problem that the intensity of electromagnetic waves is reduced. Furthermore, since the velocity of the plasma wave is a function of the applied voltage, the voltage applied to the channel below the gate electrode varies due to a short circuit, which disrupts the wavefront of the plasma wave, especially when the gate width is wide. The frequency of the plasma wave varied, and the intensity of the electromagnetic wave having a specific frequency was lowered.

  As an accompanying problem, the conventional literature has a problem of affecting the environment because a compound semiconductor containing arsenic is used as a component of a device substrate or a crystal body.

  In the following, differences between general technical levels of the present invention and the prior art will be briefly described.

  Patent Document 2 provides an optical waveguide for high-speed switching as shown in FIG. Since the optical waveguide portion is composed of the negative dielectric medium 101 and the positive dielectric medium 102, when light is guided through the optical waveguide portion, surface plasmons are formed inside. When the surface plasmon is used as a carrier and a high-speed modulation signal is applied to the electrode 104 from the outside, the light wave incident on the optical waveguide portion can be modulated at high speed. The object of the present invention is to produce an electromagnetic wave generating element of a type that does not irradiate electromagnetic waves or light from the outside. Therefore, the method is different from the method using such surface plasmons.

  The structure shown in Non-Patent Document 3 is a nanowire obtained by alternately growing Si layers 114 and SiGe layers 11 which are semiconductors having different energy band gaps Eg periodically as shown in FIG. A region 115 having a high Ge concentration is formed therein. Further, as shown in FIG. 11B, the structure shown in Non-Patent Document 4 is a concentric undoped i-Ge layer 117, an insulating film 118, and a p-type doped p-Ge layer around Si114. In the case of nanowires, a Ge layer 117 is epitaxially grown on the Si layer 114. A Ge layer 119 is also grown on the insulating layer. However, they only provide basic process technology for fabricating MOSFETs using nanowires, and are not intended for use in devices that generate electromagnetic waves from nanowires. Therefore, the MOSFET characteristics have not been fully evaluated and the generation of plasma waves has not been taken into account at all. Therefore, the basic study results when Si, Ge, and their heterostructures are simply introduced into nanowires are shown. Not too much.

  Hereinafter, the reason why electromagnetic waves due to plasma waves are not detected in the devices manufactured in Non-Patent Document 3 and Non-Patent Document 4 will be described.

  The first reason is that the period and interval between the Si layer and the SiGe layer are not suitable for generating plasma waves. In the present invention, as a result of examining the refractive index of the plasma wave in the channel, it has been clarified that the refractive index is as high as about 300. In Non-Patent Document 3, since one cycle of the Si layer 114 and the SiGe layer 115 is about 2.5 μm, it is considered that a plasma wave of about three wavelengths is excited in the structure of one cycle. It is extremely long, 6 times the half wavelength shown in the invention.

  The second reason is that since the width of the Si layer 114 having a large Eg is as long as about 1.5 μm, it is necessary to exceed a large barrier in order to inject electrons into the next SiGe layer 115. It is considered that no plasma wave is formed in such a nanowire structure.

  As a third reason, even in the SiGe layer 115 in Non-Patent Document 3, since the Si concentration is as high as 70% or more and Si having a large effective mass is used as a base material, a plasma wave is formed. It is hard to be done. In Non-Patent Document 4, a Si nanowire 114 is formed at the center as shown in FIG. 11 (b), and the electron density at the center of the nanowire is reduced by reducing the diameter and reversing the electron distribution. When it is increased, electrons are concentrated in Si having a low electron velocity as in Non-Patent Document 3, and it is considered that a plasma wave is not formed. Thus, when Si is used as a base material, it seems that the intensity of the plasma wave is reduced due to the low velocity of electrons.

  Thus, since the period and width of the high Eg region and the composition of the low Eg region for the purpose of generating plasma waves do not take desirable values, it is clear that generation of plasma waves is not considered.

  Therefore, in the present invention, the high Eg region is arranged in the nanowire with a half period Λ / 2 of the wavelength Λ of the plasma wave, and the length of the high Eg region (gate length L) is 1 nm or more and 5 nm or less. By placing nanowires with small Ge in the low-Eg region, it was clarified that the source electrode and the gate electrode do not need to be short-circuited, indicating that the electromagnetic wave intensity can be increased. Furthermore, as an unexpected effect, it was clarified that electromagnetic waves can be increased in proportion to the number of high-Eg regions and the number of nanowires.

  The present invention has been made in view of the above-mentioned problems of the prior art, and an object thereof is to provide a semiconductor device based on a novel operation principle with high yield by utilizing stably coupled plasma oscillation.

  In order to solve the conventional problem, the electromagnetic wave generator of the present invention is composed of nanowires as shown in FIG. 1, and each nanowire is provided with a plurality of high Eg regions (high Eg regions) 35. The high-Eg region was arranged at the half-cycle length of the plasma wave. As shown in FIG. 2B, the electrons introduced into the nanowire vibrate between the reflection points 39 using the high Eg region 35 as the reflection point 39, so that the plasma wave is high as shown in FIG. As a result, vibration having a node in the Eg region is generated, and the wavelength Λ of the plasma wave is twice the period of the high Eg region. As a result, it has been found that it is not necessary to short-circuit the source electrode and the gate electrode, and the intensity of the plasma wave increases.

  Here, the low-Eg region 34 is made of Ge nanowires, the high-Eg region 35 is made of SiGe or Si, and the width of the high-Eg region 35 has a thickness of 1 nm or more and 5 nm, which is a thickness at which an effect of a band barrier appears and electrons can conduct tunnel conduction As described below, electrons are reflected while being injected, and plasma oscillation can be obtained between the high Eg regions. By forming the nanowire by VLS growth, the position of the high-Eg region can be controlled with extremely high controllability, so that the plasma wave can be generated with the electrode length (gate length) as in Non-Patent Document 2. It was found that extremely high controllability can be obtained in the case of the present invention as compared to the case of controlling the wavelength.

  In addition, by making the diameter of the nanowire less than half the wavelength of the plasma wave, multimode and wavefront disturbance are prevented, and preferably by making it 20 nm or less, the electron distribution is reversed, and the electron density is at the center of the nanowire. It has been found that the plasma wave intensity increases because of abrupt increase.

  Furthermore, as an unexpected effect, it became possible to integrate the intensity of plasma waves by the number of high-Eg regions, and it was found that extremely high-intensity plasma waves can be obtained. That is, Non-Patent Document 2 uses only one gate electrode having a gate length of 60 nm. As a result, only very weak plasma waves were emitted corresponding to the gate length. In the present invention, as shown in FIG. 6, a plurality of nodes of the plasma wave can be formed by using the high Eg region 35 inside the gate electrode, so that the gate electrode can be lengthened regardless of the wavelength of the plasma wave. . As a result, the plasma wave intensity was integrated by the number of high Eg regions.

  The problems described above and their solutions are summarized. By forming a high-Eg region inside the Ge nanowire, the following problems were solved. First, it is expected that not only the electron density will increase by reducing the diameter of the nanowire, but the plasma wave intensity will increase, and the plasma wave intensity will be integrated by the number of nodes by periodically arranging high-Eg materials. The first problem of low output was solved by the effect that was not present. Since high-Eg and low-Eg regions are produced using crystal growth, precise control of the length is possible, and the accuracy of the electrode length is not important, and a long gate electrode is acceptable. As a result, the second problem has been solved. Since the reflection point is created in the high-Eg region without forming the reflection point at the electrode end, it is not necessary to short-circuit the gate electrode and the source electrode, and a sufficiently high voltage can be applied to the gate electrode. The third problem has been solved. By making Ge nanowires that do not affect the environment, arsenic becomes free, and electron mobility similar to that of compound semiconductors can be obtained.

  For the purpose of increasing the electron density and stabilizing the wavefront using the inversion distribution of electrons, we applied a nanowire structure with a heterostructure to the plasma wave device. As a result, by forming multiple periodic high-Eg regions in the nanowire, the plasma wave intensity increased and the electrode process was simplified.

  As described above, in the electromagnetic wave generator using the nanowire, the present invention forms an periodic high-Eg region in the nanowire below the gate electrode, and as an unexpected effect, the intensity of the plasma wave is the same as the number of high-Eg regions. An increasing electromagnetic wave generating element is provided. Further, by reducing the diameter of the nanowire, the electron distribution in the nanowire is reversed, the electron density is improved, and the plasma wave intensity is increased. Therefore, an extremely high output electromagnetic wave generating element is provided.

  Embodiments of the present invention will be described below with reference to the drawings.

(Embodiment 1) <MOS structure>
FIG. 1 is a schematic diagram of an electromagnetic wave generating element according to Embodiment 1 of the present invention. Also in FIG. 1 and subsequent figures, the same components are denoted by the same reference numerals, and description thereof is omitted.

As shown in the AA ′ cross section of FIG. 1A, semiconductor nanowires 47 are arranged on a substrate 37 with an insulating layer 38 interposed therebetween. Here, since Ge (Eg = 0.66 eV) is used for the low Eg region 34 and Si (Eg = 1.12 eV) is used for the high Eg region 35 in the fabrication of the semiconductor nanowire 47, a so-called heterojunction structure of Si and Ge is used. It has become. Here, as the diameter of the nanowire becomes smaller, the electron confinement becomes tighter, so a SiGe compound was used in the high-Eg region. If the band gap in the high Eg region is too large, Coulomb blockade will occur. Therefore, when the diameter d of the nanowire is 20 nm or less, the Ge composition (x) of the Si _(1-x) Ge _(x) compound was changed as follows.

x (%) = d (nm) × 4 + 20 ± 10 (however, x does not exceed 100%)
A semiconductor nanowire 47 composed of the low-Eg region 34 and the high-Eg region 35 is concentrically surrounded by an insulating film 36 made of SiO ₂ , and the periphery thereof is further concentrically covered with a gate electrode 31 such as poly-Si or Ti. It is a multilayer film body. This nanowire multilayer film body is placed on a substrate 37 having an insulating film 38 deposited on the surface thereof, and after exposing both ends of the semiconductor nanowire 47, a source electrode 32 and a drain electrode 33 are vapor-deposited on both sides. Finally, a gate wiring electrode 30 for electrical connection with the nanowire gate electrode 31 is deposited to obtain an electromagnetic wave generating element.

A method for manufacturing the electromagnetic wave generating element will be described with reference to FIGS. The semiconductor nanowires were grown by VLS (vapor-liquid-single phase) growth method. Gold dots 93 having a diameter of about 10 nm are dispersed at a density of about 10 ⁷ cm ⁻² on a Si substrate 41 on which an insulating film 42 such as SiO ₂ is deposited, and a semiconductor is formed by vapor phase growth (CVD). Create nanowires. Based on hydrogen gas, 2% silane gas and 2% germane gas are used as raw materials. As a doping gas, a phosphine gas diluted with 100 ppm hydrogen was used in order to obtain an n-type. Arsine may be used as a doping gas using arsine. In the case where arsenic is added as a dopant, unlike the case where it is used as a constituent element, the arsenic concentration is below the level that exists in nature, so the influence on the environment is so small that it can be ignored.

The substrate was heated to 300 to 500 ° C., silane gas was supplied in a reduced pressure state of 5 Torr, and Si nanowires were grown using gold dots 93 as a catalyst. The diameter of the nanowire is approximately the same as the diameter of the gold dots dispersed as a catalyst. The growth rate of nanowires was 2 to 10 nm / s. After the Si nanowire is grown as about 1 μm as the high-Eg region 35, the Ge nanowire is continuously grown as 0.4 μm as the low-Eg region 34, the Si nanowire is grown as 2 nm as the high-Eg region, and the Ge nanowire is grown as 0.4 μm as the low-Eg region. Nanowires were obtained (FIG. 3a). Next, this semiconductor nanowire was deposited in another CVD apparatus by using a silane gas and nitrous oxide as raw materials to deposit a 10 nm SiO 2 film at about 350 ° C. Furthermore, 100 nm of titanium and 500 nm of gold were vapor-deposited by an electron beam vapor deposition method to obtain a nanowire multilayer film composed of the insulating film 36 and the gate electrode 31 concentrically on the outer periphery of the semiconductor nanowire. The nanowire multilayer film body was disposed on the substrate 37 on which the insulating film 38 was deposited (FIG. 3b). A view from the right side of FIG. 3b is shown in FIG. 3f. Next, for the purpose of establishing electrical connection with the semiconductor nanowire, the central portion of the nanowire multilayer film band is protected using a resist by a reduction exposure method, and gold is removed by etching with an iodine-based etchant for about 5 minutes. Titanium and SiO ₂ were etched away using acid to expose both ends of the semiconductor nanowire (FIG. 3c). After removing the resist, the resist is applied again, the resist at both ends of the semiconductor nanowire is removed by a reduction exposure method, and a multilayer film made of copper, titanium, and gold is EB-deposited on the exposed semiconductor nanowire region. The source electrode 32, the drain electrode 33, and the gate wiring electrode 30 were formed by lift males to obtain an electromagnetic wave generating element (FIG. 3d). Here, the structure seen from above is shown in FIG.

Below, the matter which should be careful when producing the electromagnetic wave generator of this Embodiment is demonstrated. The semiconductor nanowire 34 has a structure with a diameter of 20 nm or less. Since the wavelength of the plasma wave formed in the nanowire is 800 nm, the single transverse mode condition is sufficiently satisfied. As the nanowire is made thinner, the conducting electrons are strongly affected by surface scattering, and scattering other than electron-electron scattering becomes significant. Therefore, the diameter of the nanowire needs to be 1 nm or more. The periphery of the nanowire 34 was covered with an insulating film 36 made of SiO ₂ or the like to protect particularly susceptible Ge from the influence of ambient humidity. In the present embodiment, Ge is used for the semiconductor layer. This is because when Si is used, the effective mass is large, so that the generation of plasma waves becomes unstable and the output is also small. is there. Therefore, the improvement in characteristics as shown in the present embodiment is not recognized in Si. It has been found that a plasma wave device using a group IV element not containing arsenic can be realized for the first time by using Ge having an effective mass smaller than that of Si.

<Dependence of plasma wave output on the number of nanowires>
A plurality of nanowire multilayer films may be formed, and the output of electromagnetic waves increases as the number of wires increases. By setting the number of nanowire multilayer films to 3 or more, the Q value is improved by resonance between nanowires. As shown in this application 2 in the table of FIG. 14, when the diameter W of the nanowire is about 4.6 nm, six nanowires are arranged with a period of 1 μm, and the total gate width W _total is 6 μm. In the present embodiment has a 4.6nm nanowire diameter, in this case the number n _n of the nanowires in the same order of 2mW the case of the compound semiconductor to output was required about six. Therefore, it was found that by setting the number of nanowires to 6 or more, it is possible to generate an electromagnetic wave having a higher strength than that of the prior art.

<Diameter dependence of plasma wave output nanowire>
FIGS. 13A to 13C show the electron distribution in the nanowire, and FIG. 3D shows the relationship of the electron density at each location of the nanowire structure of FIGS. As shown in FIG. 3A, when the diameter of the nanowire is 20 nm or more (R _L ), the channel 48 is formed on the side wall of the nanowire 47 on the gate electrode 31 side. On the other hand, as shown in FIG. 13B, by setting the diameter of the nanowire to 20 nm or less (R _S ), the channel 48 leaves the side wall, and the portion where the electron density is maximum becomes the central portion of the nanowire 47. . As a result, the electron density is about 20% larger than that in the case (a). Further, as shown in FIG. 3C, by setting the diameter of the nanowire to 3 nm or less, the electron density is increased as shown in FIG. It was found that the electron density can be increased about twice as compared with the above. Thus, by reducing the diameter of the nanowire, the electron density can be increased, and the output of the electromagnetic wave can be increased in proportion to the electron density.

  Here, in order to increase the electron density, the diameter of the nanowire was set to 20 nm or less. In this case, about six nanowires were required to make the output comparable to that of the compound semiconductor. Here, when the diameter of the nanowire is increased and the number is about one, the output of the electromagnetic wave increases, but on the other hand, the Q value is reduced due to the decrease in the uniformity of the electron density inside the nanowire. descend. Further, as the number of nanowires is reduced to about 1 nm and the number thereof is increased to 30 or more, the phase shift due to the variation in the diameter of the nanowire becomes more significant. Therefore, about 1 to 100 nanowires were appropriate.

By the way, unlike the case of producing a nanowire structure by top-down by dry etching or the like, in the case of a nanowire, it is necessary to align the nanowires one by one. In order to obtain the same high-frequency characteristics, nanowires with the same diameter are required, and the period of the high-Eg region needs to be the same. Therefore, when producing a plasma wave device using nanowires, the number of nanowires cannot be increased unnecessarily, so that a good characteristic can be obtained with one nanowire other than the number of nanowires such as operating voltage. It was necessary to consider the conditions. As shown in the present application 1 in FIG. 14, when a 1 nanowires number n _n is the problem that the output of the plasma wave becomes small and 0.3mW in case where a voltage applied to the gate to 0.5V there were. Therefore, we examined whether a plasma wave output of about 2 mW could be obtained by increasing the gate voltage. As a result, it was found that if the gate voltage was 3.8 V, 2 mW could be obtained. In the case of the present embodiment, since the reflection point is formed in the high Eg region, it is not necessary to short-circuit the gate electrode and the source electrode, so that up to about 20 V can be applied to the gate electrode. Thus, it was found that the increase of the plasma wave can be easily realized by forming the reflection point using the nanowire.

FIG. 15 shows how the frequency fp of plasma changes when the number of nanowires n _n and the gate voltage V _g are changed. Here, the plasma wave frequency at that time is simulated by changing the period L _eff of the high Eg region so that the output of the plasma wave is constant at 2 mW. Here, since the length L _z of the gate electrode is constant at 0.5 mm, the number _ng of high-Eg regions that can be formed decreases as the period L _eff increases. As shown the relationship between the application 1 by white circles, 3.8 V in the case of the 1 nanowires number n _n, when the period L _eff high Eg region was 0.4 .mu.m, obtain plasma wave output of 2mW The gate voltage was required. Frequency f _p of the plasma wave at this time became 2THz. As indicated by white circles in FIG. 15, when the gate voltage is 15 V, the plasma wave frequency f _p can be reduced to about 1 THz. The plasma wave frequency f _p when the gate voltage is 1V was approximately 4 THz. As described above, the plasma wave element using nanowires can easily oscillate about 2 THz to 4 THz, but it is somewhat difficult to obtain an oscillation frequency of 2 THz or less. Conventional devices such as FET and HBT have a problem that the output decreases as the operating frequency increases. In the case of a plasma wave element, the output increases as the frequency increases. This also shows that the plasma wave element is extremely promising as an oscillation element in the THz band.

Figure 15 shows similar results in the case of changing the nanowires number n _n by black circles. In this case, as shown in Application 2 of FIG. 14, the gate voltage was fixed at 0.5 V, and the number of nanowires was changed from 1 to 6. Similarly to the first application, the period L _eff of the high Eg region was changed so that the output of the plasma wave was 2 mW, and the plasma wave frequency at that time was simulated. As a result, when the number of nanowires was one, the plasma wave frequency was about 6 THz. Therefore, when the plasma wave frequency is 6 THz or more, even with one nanowire, about 2 mW was obtained with a gate voltage of 0.5 V. As with the gate voltage dependency, the number of nanowires needed to be increased in order to reduce the frequency of the plasma wave. By using three nanowires, 2 THz can be obtained, but in order to obtain 1 THz, it was necessary to arrange six in parallel. Although it is necessary to solve some difficult process problems to arrange nanowires, the number of nanowires remains at a realistic level compared to the fact that the required gate voltage has risen rapidly. Changing is considered to be a more controllable method.

<Period dependence of plasma wave power in high Eg region>
In FIG. 15, the period of the high-Eg region is changed in order to keep the plasma wave output constant. Description will be made using the conditions of the second application in FIG. As shown in FIG. 2, the high Eg region was arranged at 0.3 μm, which is a length Λ / 2 = L _eff which is half the period Λ of the plasma wave of about 0.6 μm. Within the range of the gain of the plasma wave, the wavelength of the plasma wave changes so as to coincide with the period by changing the period Λ / 2 of the high Eg region. Therefore, the frequency and intensity of plasmons can be changed by adjusting the period of the electrodes. The width of the high Eg region was 3 nm so that a tunnel current could flow. As a result, it has been clarified that plasma waves having nodes in each high Eg region are formed, and the intensity of the plasma wave increases in proportion to the number _ng of the high Eg region.

  The difference between the present invention and the conventional oscillation element will be described in detail with reference to FIG. In the case of the structure of the oscillation element shown in Non-Patent Document 1 as shown in FIG. 8, there is one gate electrode 51, and a plurality of orders of plasma waves are generated below the gate electrode 51. The plasma wave output in this case is as follows.

P = CU ₀ ² Wν ₀ α ² (1)
Where C is the capacitance under the gate electrode, U ₀ is the DC component of the voltage applied to the channel under the gate electrode, W is the diameter of the nanowire, ν ₀ is the velocity of the electrons, and α is the voltage applied to the gate electrode. U = U ₀ + U ₁ exp (-iωt)
Is defined by a modulation degree α = U ₁ / U ₀ (= AC component / DC component).

By the way, in the case of Formula (1), it turns out that the output P does not show gate length L dependence. However, in this embodiment, by describing the expression of the output P as a function of the gate length L as follows, it is clarified that the output can be described as having a gate length dependency. I was able to. That is,
P = (CU ₀ ² WL) · (ν ₀ / L) · α ² = E · n _m · α ² (2)
Here, the first term of the product of the right side of the equation (2) is energy E = CU ₀ ² WL stored in the lower part of the gate electrode, and the second term is the number of round trips of electrons traveling between the high-Eg regions is n _m = ν ₀ / L _eff . Since the length L of the electrode is constant even if a high Eg region is provided, the energy E stored in the electrode is constant. On the other hand, by increasing the number _ng of the high Eg regions, the period between the high Eg regions becomes L _eff = L / _ng , and the electron reciprocation phenomenon is simultaneously performed independently in all the high Eg regions. As a result, the number of electron reciprocations is n ′ _m = n _g · ν ₀ / L. as a result,
P ′ = E · n ′ _m · α ² = n _g · P
Thus, it was found that the output of the electromagnetic wave was _ng times. Here, in order for the electrons to reciprocate accurately between the high Eg regions, the electrons need to be reflected to form a node in each high Eg region. Therefore, the high Eg region was arranged at a half interval of the plasma wave wavelength Λ. As shown in FIG. 6, by periodically arranging a plurality of high Eg regions 35, electrons reciprocate independently in each high Eg region to generate plasma waves. The intensity P ′ can be made _ng times that is the number of high Eg regions.

As a result of evaluating the refractive index of the plasma wave, it has been clarified that the refractive index n _eff is about 290 when the electron concentration n _s in the semiconductor layer is 10 ¹² cm ⁻² . This is an extremely large value of about 100 times because the refractive index of light in the semiconductor is about 3. This knowledge has led to the development of capturing the nanowire structure of the plasma wave device as a waveguide of an optical system, and the basis of the basic idea of the present application has been completed. The electron concentration n _s is about 3 × 10 ¹² cm ⁻² , which is about three times that of normal because of using Ge nanowires. When this electron concentration is increased to 4 × 10 ¹² cm ⁻² , the dependence of the refractive index on the electron concentration becomes Δn _eff / Δn _s = −6 × 10 ^−12, and therefore n _eff decreases to 284. To do. Thus, since the relationship between the refractive index and the electron concentration has been clarified, when the frequency of the plasma wave in the semiconductor layer is set to 1 THz or more, the period Λ of the plasma wave is 0.6 μm, and the period of the high-Eg region is increased. It was found that Λ / 2 is about 300 nm. Even in the case of InAlAs / InGaAs HEMT that has achieved the highest speed in the past, if the gate length is 300 nm, ft and fmax are as low as about 300 GHz. The plasma wave generating element of this embodiment has an advantage that an ultra-high frequency can be generated by a simple process because an electromagnetic wave of 1 THz or more can be generated even when the gate length is 300 nm.

FIG. 14 shows the results of performance comparison between the electromagnetic wave generating element of the present embodiment and the elements shown in Non-Patent Documents 1 and 2. When Ge is used, the effective mass m * is about 20% larger than that of InGaAs used in Non-Patent Document 1. However, since both Non-Patent Documents 1 and 2 use arsenic (As), when using arsenic free, the effective mass is slightly large, but the use of Ge is desired. In the case of Non-Patent Document 2, the effective gate length L _eff is 0.1 μm or less, the band gap is small, and since it is a direct transition, avalanche breakdown occurs and the voltage U ₀ can be applied only about 0.08V. Here, the effective gate length is _defined as L _eff = L + 2δ including the thickness δ of the depletion layer formed around the length L of the gate electrode. For this reason, the output of electromagnetic waves is smaller than that of Non-Patent Document 1, although the effective mass is as extremely small as 0.042. Here, in Non-Patent Document 1 and Non-Patent Document 2, the electromagnetic wave outputs P are 2 mW and 0.5 mW, respectively, which are the highest outputs that can be obtained ideally. Actually, only ^10-6 times the output is obtained.

On the other hand, since Ge is an indirect transition, operation does not become unstable even when a voltage of about 0.5 V is applied. As a result, the non-patent document 1 and the present embodiment have substantially the same maximum value of the electron drift velocity v ₀ . The gate width W is conventionally about 100 μm because it is based on a normal HEMT device, but in the case of the present embodiment, it is made nanowire of 4.6 nm. FIG. 14 shows the result in the case of W = 4.6 nm in this application 1. The degree of modulation α is about 0.5. In the case of nanowires of about 4.6 nm, the sheet carrier density n _s increases to about 3 × 10 ¹² cm ^{−2 by} about three times the sheet carrier concentration. The period L _eff of the high Eg region was set to 400 nm in consideration of the growth rate. The length L _z of the gate electrodes was 500 [mu] m. As a result, the number of high Eg regions _ng is 350 per nanowire. In addition, the number _n n of the nanowire was the one. As a result, in this embodiment (this application 1), an output comparable to 2 mW of the output P of the non-patent document was obtained. Further, the Q value indicating the performance of the resonator is 1.5 in the case of the present embodiment, which is increased compared to the conventional 1.2. From the above results, the plasma wave intensity increased with respect to the prior art by setting the number of nanowires to one or more.

<Reason for making nanowire diameter 400nm or less>
In the above description, the increase in the carrier concentration due to the semiconductor layer having a nanowire structure has been described. In this case, a great effect was obtained by setting the nanowire diameter to 20 nm or less. However, even when the nanowire diameter is 400 nm or less, it has a high advantage over the conventional case where the gate width is 100 μm. Yes. The advantages of having a nanowire structure of 400 nm or less will be described below. In the present embodiment, the improvement in output was confirmed by using the high-Eg region periodically arranged. However, when this periodic electrode structure is used in Non-Patent Documents 1 and 2 that do not have a nanowire structure. As shown in FIG. 12, the wavefront of the plasma wave is greatly disturbed. The reason will be described below. As shown in FIG. 12A, by applying an electric field to the drain 33 and the source 32 to obtain a steady current, the gate voltage U ₀ applied to the semiconductor layer is low on the source side 32, and the drain electrode 33 It gets bigger as you go to the side. As a result, the velocity s of the plasma wave is
s = (eU ₀ / m) ^1/2 (3)
In as indicated, showing the same tendency as U _0. The frequency fp (wavelength) of the propagating plasma wave is
fp = s / 4L (4)
Although fp does not change in the nanowire, it is necessary to increase L as s increases. As a result, as shown in FIG. 12B, the wavefront of the plasma wave (indicated by a dotted line) is bent toward the drain electrode 33 as indicated by the arrow in the direction of the wavefront. Since the plasma wave reciprocates in the two-dimensional electron gas 54, plasma waves having a plurality of wave fronts are mixed in the two-dimensional electron gas, and the modes interfere with each other as shown in FIG. There will be many modes. As described above, when a plurality of modes exist at the same time, there is a problem that a resonance phenomenon hardly occurs, the Q value is lowered, and the plasma wave becomes unstable. However, as shown in FIG. 12D, by making the diameter of the nanowire narrow to about 400 nm, which is a half wavelength, and forming the nanowire shape 47, only one mode in the lateral direction can exist. Also, as shown in FIG. 12 (d), the wavefront cannot be bent. Therefore, when viewed from the whole nanowire, the plasma wave coupling coefficient partially decreases, but multiple modes do not occur. As a result, stable plasma waves can be generated. As described above, in the present embodiment, the case where the diameter of the nanowire is 4.3 nm is shown. However, by making the diameter of the nanowire less than or equal to half of the wavelength of the plasma wave, it is not affected by the nonuniformity of the gate voltage. Plasma waves can be generated, and plasma waves can be increased by a plurality of electrodes.

<Relationship between number of nanowires and Q value>
By the way, when several nanowire is arrange | positioned in parallel, the electromagnetic waves in each nanowire will couple | bond together by a gate electrode and a drain electrode. As a result, plasma waves with the same phase are formed throughout the nanowire, and a significantly high Q value is obtained. That is, an electromagnetic wave having a very narrow line width is generated at a frequency corresponding to the period of the gate electrode. In the case of the electromagnetic wave generating elements shown in Non-Patent Documents 1 and 2, since the semiconductor layer is not divided, the structure is not synchronized with a plurality of resonators. In equation (2), the integrated intensity with respect to the frequency of the electromagnetic wave is shown as the intensity of the electromagnetic wave. When such a high Q value is obtained, the peak intensity P _P of the electromagnetic wave at a specific frequency is the Q value. It becomes extremely large when the value is low. For example, since it is considered that when n _n nanowires are synchronized, the intensity of P _P ∝√n _n is obtained, and in the case of 100 nanowires, the peak intensity is one in comparison with the case of one. It was found that 10 times the electromagnetic wave peak intensity was obtained.

In this application 2 of FIG. 14, the Q value is expressed as Q = sτ√n _n / L _eff
As shown in Fig. 2, the output P and Q values are about 2.5 and 1.2 times higher by taking injection locking between the drain electrodes in the electromagnetic wave generating element in which the high-Eg region is periodically arranged for the six nanowires. I was able to improve. In addition, the frequency fp of the plasma wave is 1 THz, which is about the same value as that of Non-Patent Document 1, and even if the gate length is 500 μm, an electromagnetic wave of 1 THz or more can be generated. I knew it was possible.

  Note that although the gate electrode is made of copper or titanium in this embodiment mode, other metal such as a multilayer film containing gold, chromium, or silver may be used as long as it is a material that easily generates plasma waves.

<Modification 1 of Embodiment 1> Resonant tunnel structure In FIG. 1, Si nanowires in which the high-Eg region 35 on the source side is arranged at a length of 0.4 μm from the source electrode 32 are used, but as shown in FIG. In addition, the high-Eg region 35 on the source side may have a resonant tunnel structure. Since electrons having high energy are supplied from the resonant tunneling structure into the semiconductor nanowire, the plasma wave intensity increases. When producing this structure, instead of supplying silane gas at the beginning of nanowire growth, germanium gas is supplied to grow Ge nanowire about 0.4 μm, gas type is changed from germane gas to silane gas, and the film thickness is 2 nm. A Si layer, a Ge layer having a thickness of 1 nm, and a Si layer having a thickness of 2 nm are grown while sequentially switching gases, and then Ge nanowires are grown while supplying a germane gas.

In this way, most are made of Ge nanowires, and since the Si layer uses only a few nm, the electron velocity is high, and the thermal saturation rate is 2.7 × 10 ⁷ cm of the normal MOS structure ^It increased about 30% from ^-3 . Furthermore, it was found that by using a resonant tunneling structure, the thermal saturation rate was about 2 × 5 × 10 ⁷ cm ⁻³ , and the plasma wave intensity was increased 1.8 times to obtain 3.6 mW.

<Modification 2 of Embodiment 1> Vertical Plasma Wave Device
If gold dots are dispersed directly on the Si substrate without depositing an insulating film on the Si substrate 37, Si nanowires having conductivity with the Si substrate can be formed. As a result, as shown in FIG. 7, a vertical plasma wave device can be manufactured. In this case, since nanowires can be formed at the density of dispersed gold dots, an extremely large number of nanowires can be formed. As a result, the intensity P of the plasma wave is P∝√n _n as a function of the number n _n of nanowires. Therefore, when a structure using 100 nanowires is fabricated, the output of the plasma wave is one. 10 times as much.

A method for manufacturing this vertical plasma wave device will be described below. Semiconductor nanowires were grown by VLS growth method. Gold dots having a diameter of about 10 nm are directly dispersed on the Si substrate 41 at a density of about 10 ⁷ cm ⁻² , and semiconductor nanowires are manufactured by a vapor deposition method (CVD method). Based on hydrogen gas, 2% silane gas and 2% germane gas are used as raw materials. As a doping gas, a phosphine gas diluted with 100 ppm hydrogen was used in order to obtain an n-type. The substrate is heated to 300 to 500 ° C., silane gas is supplied in a reduced pressure state of 5 Torr, and Si nanowires are grown using gold dots as a catalyst. The diameter of the nanowire is approximately the same as the diameter of the gold dots dispersed as a catalyst. The growth rate of nanowires was 2 to 5 nm / s. Since an insulating film is not used, Si is deposited also on the substrate 37. However, since the growth rate of VLS growth is larger, nanowires grow relatively. However, the growth rate of nanowires was reduced to about half. After the Si nanowire is grown as about 1 μm as the high-Eg region 35, the Ge nanowire is continuously grown as 0.4 μm as the low-Eg region 34, the Si nanowire is grown as 2 nm as the high-Eg region, and the Ge nanowire is grown as 0.4 μm as the low-Eg region. Nanowire 47 was obtained. Next, 10 nm of SiO 2 film 36 is deposited at about 350 ° C. using silane gas and nitrous oxide as raw materials in another CVD apparatus. Furthermore, 100 nm of titanium and 500 nm of gold are vapor-deposited by an electron beam vapor deposition method to obtain a nanowire multilayer film body including the insulating film 36 and the gate electrode 31 concentrically on the outer periphery of the semiconductor nanowire. An SOG film is deposited and buried so as not to collapse the nanowire multilayer film body, the upper part of the SOG film is polished by CMP, the high Eg region 35 is exposed, and the drain electrode 33 is evaporated. The SOG film is left only in a necessary region, the SOG film in other regions is removed, and the source electrode 32 and the gate wiring 43 are formed by EB vapor deposition and lift-off to produce the electromagnetic wave generating element shown in FIG. Can do.

(Embodiment 2) <Heterostructure>
In the first embodiment, the insulating film 36 is inserted between the gate electrode 31 and the semiconductor nanowire. In this case, the electron density in the channel formed at the interface between the insulating film and the semiconductor depends on the dielectric constant and film thickness of the insulating film. Because it depends, you can not get a large value. On the other hand, as shown in FIG. 1C, by replacing the portion of the insulating film 36 with Si, a high Eg region having a large band gap with respect to Ge constituting the low Eg region 34 is formed. Can do. By performing high concentration doping (δ doping) 46 on a part of the high Eg region, it is possible to supply high concentration electrons from the high concentration high Eg region 43 to the low Eg region 34 as in HEMT. Become. Δ doped region 46 doped with phosphorus or arsenic in Si by 3 × 10 ¹² cm ⁻² or more and 10 ¹³ cm ⁻² or less, preferably 5 × 10 ¹² cm ⁻² δ by supplying a doping gas after interrupting growth. Form. The δ-doped region 46 and the low-Eg region 34 are separated by 3 nm. Here, arsenic is used, but since the concentration of arsenic with respect to the entire chip is lower than the concentration existing in nature, there is no concern for the environment.

  A manufacturing method is shown below. Since the manufacturing method of the semiconductor nanowire 47 is the same as that of the first embodiment, the description thereof is omitted. The semiconductor nanowires obtained in FIG. 3a are deposited in the same CVD apparatus, the substrate is set to 450 ° C., silane gas is supplied at 50 Torr, and an undoped Si layer 34 is deposited to 3 nm. Here, the growth temperature was increased from 5 Torr to 50 Torr in order to grow the undoped Si layer 34 concentrically on the outer periphery of the semiconductor nanowire. Once the growth is interrupted, 1% phosphine or arsine as a doping gas is supplied for 3 minutes, and phosphorus or arsenic is δ-doped. Thereafter, doping was stopped, and only the silane gas was supplied to grow an undoped Si layer 34 by another 20 nm. Titanium was deposited to 100 nm, chromium was deposited to 100 nm, and gold or platinum was deposited to 500 nm by an electron beam vapor deposition method to obtain a nanowire multilayer film composed of concentric Schottky electrodes 31 on the outer periphery of the semiconductor nanowire. This nanowire multilayer film body was placed on the substrate 37 on which the insulating film 38 was deposited (FIG. 3b). Since the subsequent manufacturing method is the same as that in Embodiment Mode 1, description thereof is omitted.

By forming the high-concentration high-Eg region 43, electrons of 5 × 10 ¹² cm ⁻² are supplied to the low-Eg region 34. Therefore, an electromagnetic wave having a strength 1.7 times higher than that of the plasma wave device of the first embodiment is generated. It turns out that it is obtained.

Embodiment 3 <Doping Periodic Structure>
In the first embodiment, the plasma wave node is formed by forming the high-Eg region. However, the plasma wave node can also be formed by forming the highly doped region. FIG. 5 shows a schematic diagram of the electromagnetic wave generating element in the third embodiment. A highly doped region 44 was formed on the source side of the semiconductor nanowire, and a highly doped region 44 having a thickness of 2 nm was formed at a position half a cycle away from the plasma wave. The other region is an undoped region 34. In the highly doped region 44, since the electron velocity is remarkably reduced, a plasma wave node is formed, and it has been found that plasma wave oscillation can be obtained as in the first embodiment.

A manufacturing method is shown below. Semiconductor nanowires were grown by VLS (vapor-liquid-single phase) growth method. Gold dots having a diameter of about 10 nm are dispersed at a density of about 10 ⁷ cm ⁻² on a Si substrate 41 on which an insulating film 42 such as SiO ₂ is deposited, and a semiconductor nanowire is formed by vapor phase growth (CVD). Is made. Based on hydrogen gas, 2% germane gas is used as a raw material. As a doping gas, phosphine or arsine diluted with 100 ppm hydrogen was used in order to obtain n-type. The substrate is heated to 300 to 500 ° C., germane is supplied in a reduced pressure state of 5 Torr, and Ge nanowires are grown using gold dots as a catalyst. The diameter of the nanowire is approximately the same as the diameter of the gold dots dispersed as a catalyst. The growth rate of nanowires was 2 to 10 nm / s. After the Ge nanowire doped with phosphorus 10 ¹⁸ cm ^{−3 is} grown as about 1 μm as the highly doped region 44, the undoped Ge nanowire is continuously grown as 0.4 μm as the undoped region 34, and the supply of germane as the highly doped region 44 is stopped, and only phosphine Ge nanowire having a doping region doped with phosphorus or arsenic at 5 × 10 ¹² cm ⁻² at 3 × 10 ¹² cm ^{−2 to} 10 ¹³ cm ^{−2 after} supplying for 3 minutes, preferably 1 nm to 5 nm Form 2 nm. Further, an undoped Ge nanowire 34 was grown as a lightly doped region by 0.4 μm to obtain a semiconductor nanowire (FIG. 3a). The subsequent steps are the same as those in the first embodiment, and will be omitted.

  Here, while Si nanowires were partially used in the embodiment, Ge nanowires are all used in the present embodiment, so that the electron velocity is injected into the nanowire in a high state, It has been found that there is an advantage that the intensity of the plasma wave increases. Further, germane and doping gas are sufficient, and there is an advantage that it is not necessary to use silane. Furthermore, since the Coulomb blockade is less likely to decrease as in the case of forming a high Eg region, the source-drain voltage can be reduced.

  The electromagnetic wave generating element according to the present invention has a semiconductor nanowire and a periodic high bandgap (Eg) region, and the number of high Eg regions is added by making the period of the high Eg region half the period of the plasma wave. It is useful as an oscillation element or the like of about 1 THz from which an electromagnetic wave output can be obtained.

Schematic diagram of an electromagnetic wave generating element in Embodiments 1 and 2 of the present invention FIG. 3 is a relational diagram between the plasma wave of the electromagnetic wave generating element and the period of the high Eg region in the first embodiment of the present invention. Explanatory drawing of the manufacturing method of the electromagnetic wave generator in Embodiment 1-3 of this invention FIG. 5 is a relational diagram between the plasma wave of the electromagnetic wave generating element and the period of the high Eg region in the first embodiment of the present invention Schematic of the electromagnetic wave generating element in Embodiment 3 of the present invention FIG. 3 is a relationship diagram of the plasma wave of the electromagnetic wave generating element and the period of the high Eg region in the first embodiment of the present invention Schematic of a vertical electromagnetic wave generating element in a modification of Embodiment 1 of the present invention Schematic diagram of conventional electromagnetic wave generator Schematic diagram of conventional electromagnetic wave generator Schematic diagram of conventional electromagnetic wave generator Schematic diagram of conventional electromagnetic wave generator Explanatory drawing of the wave front of the plasma wave inside the electromagnetic wave generating element in Embodiment 1 of the present invention Explanatory drawing of the electron density of the electromagnetic wave generating element in Embodiment 1 of this invention Characteristics diagram of electromagnetic wave generating element according to Embodiment 1 of the present invention Characteristics diagram of electromagnetic wave generating element according to Embodiment 1 of the present invention

Explanation of symbols

30 Gate wiring 31 Gate electrode 32 Source electrode 33 Drain electrode 34 Low Eg region 35 High Eg region 36 Insulating film 37 Substrate 38 Insulating film 39 Fixed end 40 Free end 41 Substrate 42 Insulating film 43 High concentration Eg region 44 Highly doped region 45 SOG
46 δ doped region 47 semiconductor nanowire 48 channel 51 gate electrode 52 source electrode 53 drain electrode 54 two-dimensional electron gas 55 substrate 56 cladding layer 57 barrier layer 61 slow wave circuit 62 electron layer contact 63 hole layer contact 64 two fluid layer 65 input Portion 66 Output Portion 101 Negative Dielectric Structure 102 Positive Dielectric Structure 103 Semiconductor Substrate 104 Electrode 111 Gate Electrode 112 Source Electrode 113 Drain Electrode 114 Si
115 SiGe
116 gold dots 117 i-Ge
118 Insulating film 119 p-Ge

Claims (10)

A semiconductor layer comprising: a substrate; a semiconductor layer held on the substrate; an insulating film surrounding the semiconductor layer; a gate electrode surrounding the insulating film; a drain electrode; and a source electrode. In the cross section perpendicular to the major axis direction, the width orthogonal to each other is equal to or less than the wavelength of the plasma wave, and in the major axis direction, a plurality of high Eg regions made of a high Eg material and a plurality of low Eg regions made of a low Eg material. The junction consisting of the high Eg region and the low Eg region is located below the gate electrode, and the interval between the high Eg regions is half the period of the wavelength of the plasma wave. Electromagnetic wave generator. 2. The electromagnetic wave generator according to claim 1, wherein a plurality of high Eg regions are periodically arranged from the source electrode side, and the intensity of the plasma wave is multiplied by the number of high Eg regions. 3. The electromagnetic wave generator according to claim 1, wherein the electromagnetic wave generator has a resonant tunnel structure formed of a high Eg region on the source electrode side. 4. The electromagnetic wave generator according to claim 1, wherein a width of the high Eg region is 1 nm or more and 5 nm or less. The semiconductor layer of the undoped low-Eg material is surrounded by a semiconductor layer of a high-Eg material including a δ-doped region that is concentrically doped to 3 × 10 ¹² cm ^-2 or more and 10 ¹³ cm ^-2 or less. The electromagnetic wave generator according to claim 1, wherein: The semiconductor substrate includes a semiconductor substrate, a semiconductor layer whose cross-sectional width is equal to or less than the wavelength of the plasma wave, an insulating film surrounding the semiconductor layer, a gate electrode, a drain electrode, and a source electrode, It is composed of a high concentration region highly doped in the major axis direction and a low concentration region that is not doped, and a junction formed by the high concentration region and the low concentration region is located below the gate electrode, An electromagnetic wave generating device, wherein the concentration region is periodically arranged from the source electrode side, and the period of the high concentration region is half the period of the wavelength of the plasma wave. A plurality of high concentration regions are arranged, the high concentration regions are δ-doped to 3 × 10 ¹² cm ⁻² or more and 10 ¹³ cm ⁻² or less, and the intensity of the plasma wave is multiplied by the number of the high concentration regions. The electromagnetic wave generator according to claim 6. 8. The electromagnetic wave generator according to claim 6, wherein a width of the high concentration region is 1 nm or more and 5 nm or less. 9. The electromagnetic wave generating device according to claim 1, wherein a semiconductor layer having a cross-sectional orthogonal width equal to or smaller than a plasma wave wavelength is composed of a semiconductor nanowire having a diameter of 1 nm or more and 20 nm or less. If the high Eg material is Si or Si _(1-x) Ge _(x) , the low Eg material is Ge, and the orthogonal width of the semiconductor layer is 20 nm or less, x (%) = d (nm) 6. The electromagnetic wave generator according to claim 1, wherein x4 + 20 ± 10.

Images