JP2010238722A - Silicon light-emitting element - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a silicon light-emitting element and silicon laser element with high efficiency that include silicon as a basic component on a silicon substrate by a method capable of easily forming in normal silicon processes. <P>SOLUTION: The silicon light-emitting element includes an electrode for implanting an electron, an electrode for implanting a hole, and a light emission portion electrically connected thereto, and has a structure such that a plurality of thin films of silicon crystal, in which a plane orientation is a (001) direction, are arrayed in parallel at the light emission portion. In the silicon light-emitting element, gaps among the respective thin films are filled with oxide films formed by thermal oxidation, and pressure is applied to the silicon thin films. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to a light-emitting element using silicon, and more particularly to an element structure suitable as a high-intensity light-emitting diode or silicon laser.

  Broadband networks that support the Internet industry employ optical communications. Lasers using compound semiconductors such as III-V and II-VI groups are used for transmission and reception of light in this optical communication.

  Various structures have been proposed for compound semiconductor lasers, but double heterostructures are common. The double hetero structure has a structure in which a compound semiconductor having a small band gap is sandwiched between compound semiconductors having a large band gap using two different compound semiconductors. In order to fabricate a double heterostructure, compound semiconductors of conductivity type n-type, undoped i-type, and p-type are continuously epitaxially grown on a substrate and stacked vertically. At that time, it is necessary to pay attention to the band structure of the undoped i-type compound semiconductor sandwiched between them. The band gap is smaller than each of the n-type and p-type compound semiconductors, and the i-type conduction band. It is important that the level is lower than the n-type conduction band level and the i-type valence band level is higher than the p-type valence level. That is, both electrons and holes are confined in the i-type region. Therefore, since electrons and holes are likely to be in the same region, the probability that electrons and holes collide and annihilate increases, and as a result, the luminous efficiency can be increased. In addition, since the refractive index tends to increase as the band gap becomes smaller, light can be obtained by selecting a material in which the refractive index of the i-type compound semiconductor is larger than that of each of the n-type and p-type compound semiconductors. It will be trapped in the type compound semiconductor. The confined light efficiently induces recombination of electrons and holes that form an inversion distribution, leading to laser oscillation.

As described above, a large amount of long-distance information communication is instantaneously performed by optical communication using a compound semiconductor that emits light efficiently. That is, information processing and storage are performed on an LSI based on silicon, and information transmission is performed by a laser based on a compound semiconductor. If silicon can emit light with high efficiency, it is possible to integrate both an electronic device and a light emitting element on a silicon chip, and its industrial value is enormous. Therefore, a great deal of research has been done to make silicon emit light.
However, it is difficult to emit silicon with high efficiency. This is because silicon has an indirect transition type band structure. The indirect transition type band structure refers to a band structure in which the momentum at which the conduction band energy is the lowest or the momentum at which the energy of the valence band is the lowest is not zero. In the case of silicon, the minimum energy point of the valence band is the Γ point where the momentum becomes 0, but the minimum energy point of the conduction band is not between the Γ point but between the Γ point and the X point, and more specifically. Specifically, if the lattice constant is defined as a and defined as k ₀ = 0.85 * π / a, (0, 0, ± k ₀ ), (0, ± k ₀ , 0), (± k ₀ , 0 , 0). This is shown in FIG.
On the other hand, many compound semiconductors are called direct transition type semiconductors because the conduction band and the valence band both have a minimum energy point at the Γ point.

  Next, the reason why the indirect transition type semiconductor has low luminous efficiency and the direct transition type semiconductor has high luminous efficiency will be described. As described above, in order for the semiconductor element to emit light, electrons and holes collide and disappear, and the energy difference between the two must be extracted as light. At that time, both energy and momentum conservation laws must be satisfied. Electrons have energy levels in the conduction band, and holes have energy levels in the valence band where there are no electrons. The difference between the two becomes the energy of light, and the wavelength varies depending on the energy. Therefore, the energy difference between the conduction band and the valence band, that is, the size of the band gap determines the wavelength of light, that is, the color. In this way, no particular difficulty can be found in the establishment of the law of conservation of energy.

  On the other hand, since the collision phenomenon of electrons and holes is involved in light emission, the momentum must also be preserved. According to quantum mechanics, the rule governing the microscopic world, electrons, holes, and photons (light quanta) are all scattered in the form of waves, but the law of conservation of momentum holds. Momentum is qualitatively a measure that quantifies how quickly particles are blown off during a collision. From the light dispersion relationship (ω = ck, where ω is the angular frequency of the light, c is high speed, k is the momentum of the photon) and energy, it can be seen that the momentum of the photon in the crystal is almost zero. This means that even if there is a phenomenon that the material is bounced off by the collision of light, the effect of scattering the material is very small, which is consistent with our intuition.

On the other hand, holes have almost no momentum because the minimum point of energy is at the Γ point. However, in silicon, which is an indirect semiconductor, almost no electrons are present at the Γ point, but at the minimum energy point near the X point, so a large momentum of k ₀ = 0.85 * π / a in magnitude. Have.

Therefore, in silicon, the momentum conservation law and the energy conservation law cannot be satisfied at the same time in a process in which electrons and holes simply collide. Therefore, only electron-hole pairs that somehow satisfy the momentum conservation law and the energy conservation law at the same time by absorbing or emitting phonons, which are the quantum of photon oscillation in the crystal, are converted into light. Although such a process does not physically exist, it is a high-order scattering process in which electrons, holes, photons, and phonons collide at the same time, so the probability of such a phenomenon occurring is low. Accordingly, it is known that silicon, which is an indirect transition type semiconductor, has extremely poor luminous efficiency.
On the other hand, many direct transition type compound semiconductors have both a momentum conservation law and an energy conservation law because the conduction band and the valence band have a minimum energy point at the Γ point. Therefore, the luminous efficiency is high in the compound semiconductor. Non-Patent Document 1 below reports a transistor laser element that drives a laser using a compound semiconductor with high luminous efficiency by a bipolar transistor made of the compound semiconductor.

  As described above, silicon is extremely poor in light emission efficiency in the bulk state, but it is known that the light emission efficiency is increased by setting it to a porous state or a nanoparticle state. For example, in Non-Patent Document 2 below, it has been reported that silicon anodized in a hydrofluoric acid solution emits light at room temperature and in the visible light wavelength band due to the porous state. Although the mechanism has not been fully elucidated, it is thought that the quantum size effect generated due to the presence of silicon confined in a narrow region due to the formation of a porous layer is important. In small silicon, the position of electrons is confined in that region, and the momentum is not fixed due to the uncertainty principle of quantum mechanics, so recombination of electrons and holes is likely to occur. It is thought that.

  As another method using silicon, for example, in Non-Patent Document 3 below, a light emitting diode (LED) serving as a light emitting element can be manufactured by implanting Er ions into a pn junction formed on a Si substrate. It is stated that. When Er ions are implanted into the Si substrate, Er creates an impurity level, and the impurity level is a spatially localized level, so the electrons in the Si conduction band become the impurity level created by the Er ion. When trapped, the momentum effectively becomes zero, and it becomes possible to recombine with holes in the valence band and emit light. Since the light emitted through the Er ions has a wavelength of 1.54 μm, light can be propagated without being absorbed by surrounding silicon. In addition, since the wavelength is such that the loss is reduced when using an existing optical fiber, even if Si-based LEDs using Er ions are put into practical use due to future technological innovation, the existing optical fiber network should be used. Therefore, it is expected that there will be no need for large-scale capital investment.

  Further, as another method using silicon, for example, the following Non-Patent Document 4 and Non-Patent Document 5 described below, Er ions are implanted into silicon nanoparticles by combining the above quantum size effect and the idea of Er ions. It is described that it was possible to emit light with increased efficiency.

In the above prior art, in order to make silicon emit light, the band structure of the conduction band of silicon is changed to the bulk band structure, and the momentum is separated from the point of k ₀ by the uncertainty principle. Therefore, it has been considered that silicon should be in a porous state or a nanoparticle state. However, for example, when silicon having a structure like a nanoparticle is formed, there is a problem that the surface of the silicon nanoparticle is oxidized and silicon dioxide is formed on the surface because the silicon surface is very easily oxidized. . Since silicon dioxide is an insulator having an extremely large band gap, when silicon dioxide is formed on the surface, there arises a problem that electrons and holes cannot be injected efficiently. Therefore, in the conventional silicon light emitting device, even if high intensity is obtained by photoluminescence, there is a problem that efficiency is extremely lowered by electroluminescence. In addition, when emitting light, the crystallinity of the material that becomes the light-emitting layer is important. However, the structure is a structure in which nanoparticles are formed by CVD (Chemical Vapor Deposition) or anodized and a large number of holes are formed on the surface. Then, there exists a problem that crystallinity worsens compared with a single crystal. If the crystallinity is poor, light emission through the defect level occurs. However, since light emission using defects is inefficient, there is a problem that an element that can withstand practical use such as information communication cannot be manufactured.

As described above, efforts have been made to make silicon emit light by various techniques such as porous silicon, nanoparticles, and Er doping, but the luminous efficiency has not been high to a practical level.
We have a first electrode part for injecting electrons, a second electrode part for injecting holes, and a light emitting part electrically connected to the first electrode part and the second electrode part, The light emitting portion has a first surface (upper surface) and a second surface (lower surface) opposite to the first surface, and the surface orientation of the first and second surfaces is (100 ) And the thickness of the light emitting portion in the direction perpendicular to the first and second surfaces is reduced, so that it can be easily formed on a substrate such as silicon using a normal silicon process. It has been reported that a light-emitting element that emits light efficiently can be obtained (Patent Document 1 below). This is because when electrons are confined in a very narrow region typified by an ultra-thin single crystal silicon film, in the bulk electronic state, the electrons in the conduction band are silicon-like substances that do not exist at the Γ point. However, it can be understood that it cannot effectively move in the vertical direction of the thin film. This qualitatively indicates that the direction perpendicular to the thin film disappears, so that electrons cannot move in the direction perpendicular to the thin film. In other words, it is considered that an indirect transition type semiconductor is effectively changed to a direct transition in the bulk due to the quantum confinement effect (two-dimensional confinement effect) even when the silicon is made extremely thin. The principle of light emission and its verification results are shown below.
The principle for efficiently illuminating a group IV semiconductor such as silicon or germanium that conforms to it will be described with reference to the drawings.

  The wave function Ψ (r) representing the state of electrons in a crystal such as silicon can be expressed as (Equation 1) with a very good approximation.

Where k ₀ is the momentum that gives the valley of the conduction band, r = (x, y, z) represents the position in space, and Φ _k0 (r) is the band of the conduction band. The Bloch function at the valley is given, and ξ (r) represents the envelope function. Φ _k0 (r) can be expressed as (Equation 2) using a periodic function u _k0 (r + a) = u _k0 (r) reflecting the periodicity with respect to the unit cell vector a in the crystal.

As is clear from this, it vibrates violently as a function of atomic scale distance. On the other hand, the envelope function ξ (r) represents a component that gradually changes on the atomic scale, and represents the physical shape of the semiconductor and the response to the external field applied from the surroundings. Here, considering the case where Ψ (r) is a wave function in a semiconductor structure having a finite size, which is not necessarily a bulk crystal, a satisfactory expression for ξ (r) is 3) can be led.

Here, ε = ε (k) represents the band structure in the bulk of the conduction band electrons having momentum k, and ε is obtained by substituting the sum of −i ▽ and momentum k _{0 for} the momentum k with the differential operator −i ▽. (k ₀ −i ▽). V = V (r) is the potential felt by the electrons.For example, when an insulator or another type of semiconductor is in contact with the boundary of the semiconductor, a potential barrier is provided and a field effect is applied from the outside. By applying an electric field, the value of V = V (r) can be adjusted. Here, for the sake of simplicity, we focus only on the change of V in the z direction.
Here, in order to facilitate understanding, specifically, assuming, for example, silicon on the (100) plane as a semiconductor, the bulk has a band structure as shown in FIG. 1 as described above. Therefore, the valley of the conduction band existing at (0, 0, ± k ₀ ) in the kz direction can be approximated as (Equation 4).

  Here, m * t and m * l represent effective masses in the silicon crystal obtained from the curvature of the minor axis and major axis direction of the valley of the conduction band having a spheroid shape. Then, (Equation 3) is expressed as (Equation 5).

By setting the direction parallel to the (100) plane as (x, y), the width as W, the length as L, and the envelope function as (Equation 6), (Equation 5) becomes (Equation 7). ).

  Here, ΔE represents the energy in the z direction, and the total energy of electrons measured from the bottom of the conduction band is expressed as (Equation 8).

First, we confirm that (Equation 7) reproduces the bulk electronic state. For that purpose, a solution in a continuous state when V (r) = 0 is obtained. This can be confirmed by assuming that the thickness in the z direction is t, the envelope wave function is (Equation 9), and ΔE is (Equation 10).

  That is, the wave function oscillates violently with the wave function continuously spreading throughout the bulk crystal. At this time, it is natural that the quantum mechanical expectation value of the momentum in the z direction becomes (Equation 11) with the momentum operator in the z direction as kz.

In other words, in an indirect transition type semiconductor such as silicon, most of the electrons are moving with a very large momentum because the probability of being far from the Γ point in the momentum space is overwhelmingly high. Is also shown from the top of the formula.
The basic principle of the present invention is that in the case of an ultra-thin film having a very small thickness t in the z direction, an indirect transition type semiconductor is effectively changed into a direct transition type in the bulk due to the quantum confinement effect. use. Hereinafter, this point will be described in detail.

In order to explain the story concretely and clearly, taking silicon as an example, the thickness t in the z direction is very small, and the upper and lower sides in the z direction are adjacent to each other with a large band gap such as SiO _2. Suppose you are in contact with your body or even a vacuum or atmosphere with a large energy barrier. As a system that can be expected to have the same effect, for example, the same effect can be expected if electrons are confined in a narrow region by a field effect or the like. In these cases, the wave function of electrons in silicon becomes zero at the upper and lower interfaces in the z direction. Strictly speaking, there is exudation of a quantum mechanical wave function, but since the energy barrier is large, the exudation becomes exponentially small with respect to the distance in the z direction, and the approximation is zero at the interface. Is almost strictly correct. Therefore, even if the potential V (r) = 0 applied from the outside, the envelope wave function is completely different from the case where t is thick. In fact, the envelope wave function of electrons and holes confined in such a quantum well is n, where n is an index representing the discrete energy level, and n is an even number. (Equation 12) can be solved, and when n = 1, 3, 5,... Is odd, (Equation 13) is obtained, and the value of the energy level is (Equation 12) regardless of whether n is even or odd. 14).

It goes without saying that the lowest energy state is n = 0. In order to show the envelope wave function, the origin of the z axis is set at the center of the thin film silicon, and an interface with a high energy barrier exists at z = ± t / 2. Here, the property of the envelope wave function χ _n (z) will be described. When n is 0 or an even number, the wave function is symmetric with respect to the sign change of z and has the property χ _n (z) = χ _n (−z). This is called parity even. On the other hand, when n is an odd number, it has a property of χ _n (z) = − χ _n (−z), and the parity is odd.
Since it has a structure reflecting such symmetry, when the contribution to the momentum by the envelope wave function is evaluated, (Equation 15) is obtained.

This, taking the derivative chi _n a (z) with respect to the z direction, since the change and parity had originally chi _n (z), quite generally that becomes zero when taking the integral with respect to z-direction Is characteristic. That is, since the electrons are strongly bound in the z-axis direction, the envelope wave function becomes a standing wave, and it can be seen that the electrons do not move. This is exponential as the envelope wave function in the bulk state is given by (Equation 9), and is in sharp contrast to the fact that electrons move around the entire bulk crystal with momentum. However, the total wave function considering the existence of the Bloch function is obtained by substituting (Equation 2), (Equation 6), and (Equation 13) or (Equation 14) into (Equation 1). It should be noted that the quantum mechanical expectation of the momentum in the direction becomes (Equation 16).

In other words, as a property of the original semiconductor material, when it is bulk, there is no valley bottom of the conduction band at the Γ point, and there is a valley bottom at (0, 0, ± k ₀ ). Is reflected. In this way, even though it was a thin film, it seemed that electrons were moving around with a momentum of ± k ₀ , but I noticed that there was a need for caution. That is, for example, in a material having inversion symmetry as a crystal such as silicon, the valley of (0, 0, + k ₀ ) and the valley of (0, 0, −k ₀ ) are energetically equal, Attention should be paid to degeneration. As described above, when quantum mechanical states having energy levels that are generally degenerated are confined in the same region in space, hybridization occurs between these states. In other words, if there is very little energetic coupling between the valley of (0, 0, + k ₀ ) and the valley of (0, 0, −k ₀ ), the two discrete levels Form an orbit and anti-bonded orbit. For example, a Coulomb interaction between electrons that are not sufficiently included in the band calculation may work strongly between electrons confined in a narrow region. The interaction between electrons is called electron correlation, and has become a major problem in many transition metal oxides including high-temperature superconductivity. In bulk silicon, the sp orbit of the original silicon atom is not sufficient. Reflecting the fact that it has a large orbit, it has never been a big problem. However, when confined in a very narrow region where the quantum mechanical effect is important, the Coulomb interaction works so strongly that the Coulomb interaction between electrons cannot be ignored. If we take the Coulomb interaction properly and calculate the matrix elements of the Hamiltonian, there is a hybrid connecting the valleys of (0, 0, + k ₀ ) and (0, 0, −k ₀ ). And, if the Hamiltonian is diagonalized, it can be seen that it is split into bond orbitals. This is similar to the process by which hydrogen molecules are formed when two hydrogen atoms are brought close to each other, and the method for evaluating such a system is about 70 years before quantum mechanics was formed by Heitler-London. It was understood from. We noticed for the first time that the formation of bonding states understood by Heitler-London is also important for valley coupling when a group IV semiconductor such as silicon is confined in a narrow region. Also, even if there is no such energy coupling, a standing wave that does not move in the z-axis direction can be constructed from the unitary transformation of the two states. This will be explained more specifically. Since the Bloch state has the property of u _−k0 (r) = u _k0 (r) due to the inversion symmetry of the crystal, the (0,0, + k ₀ ) valley and the (0,0, −k ₀ ) the Bloch wave function of the valley, respectively, expressed as _{φ k0 (r) = u k0} (r) e ik0z and _{φ-k 0 (r) =} u k0 (r) e -ik0z. Therefore, it can be seen that attention should be paid to the part of e ^{± ik0z} . In order to construct a new ground state from the sum and difference of these wave functions, it is only necessary to convert (Equation 17) by unitary transformation U.

Therefore, the change of the wave function at the atomic level is the two standing wave waves 2 ^1/2 u _k0 (r) cos (k ₀ z) and 2 ^1/2 u _k0 (r) sin (k ₀ z). It turns out that it can be described by a function. The entire wave function can be expressed as (Equation 18) and (Equation 19).

 The expected value of the momentum in the z-axis direction in the state of (Expression 18) or (Expression 19) is expressed by (Expression 20) reflecting the fact that it is a standing wave.

  That is, it can be seen that electrons do not move at all in the z-axis direction. Note that just changing the base may cause misunderstanding that the expected value of momentum appears to change. Actually, the basis wave functions like (Equation 18) and (Equation 19) are not eigenstates of momentum. That is, the matrix elements of the momentum operator are (Expression 21) using (Expression 18) and (Expression 19), the diagonal matrix element is zero, and the non-diagonal matrix element is a pure imaginary number.

Whether such a basis is physically appropriate depends on the properties of the target system. We assume an ultra-thin single-crystal silicon film. In such a case, the translational symmetry with respect to the z-axis direction collapses, so the momentum eigenstate u _k0 (r) e ^± Rather than using ^ik0z , it is more ^appropriate to use √2u _k0 (r) cos (k ₀ z) and √2u _k0 (r) sin (k ₀ z) which are standing waves. Conversely, when dealing with the bulk state, it is better to use u _k0 (r) e ^{± ik0z} because of translational symmetry. Further, in the bulk state, electrons having momentum ± k ₀ move violently in the crystal, and at that time, they are strongly scattered by phonons, which are the quantum of lattice vibration in the crystal, and the wave function Since the phase changes dynamically, it cannot be expected that a state of momentum + k _{0 and} a state of momentum −k ₀ are combined coherently. In contrast to this, when electrons are confined in a very narrow region, such as an ultrathin single crystal silicon film, which is thinner than the mean free path l that characterizes scattering, even at room temperature. A sufficient wave derivative can form a standing wave with a fixed phase. Qualitatively, it means that while an electron wave travels through a narrow area at high speed, it becomes a steady wave that fits the size of that area.

  As described above, as described in detail using simple mathematical formulas, when electrons are confined in an extremely narrow region typified by an ultrathin single crystal silicon film, electrons in the conduction band in the bulk electronic state. It can be seen that even if it is a silicon-like substance that does not exist at the Γ point, it does not effectively move in the direction perpendicular to the thin film. This qualitatively indicates that the direction perpendicular to the thin film disappears, so that electrons cannot move in the direction perpendicular to the thin film. In other words, even if the bulk moves in the crystal at high speed, the thin film loses the direction to move in the first place, which means that the electrons have to stop.

FIG. 2 illustrates this state using a band diagram. Since the movement in the z-axis direction is no longer possible, the bulk band structure shown in FIG. 1 is projected onto the surface of kz = 0, and when a thin film or a field effect is applied, the band structure as shown in FIG. Become. The band structure as shown in FIG. 2 is the basis for designing a field effect transistor with silicon, and can be said to be the basis for device physics. Such a system confined in two dimensions is called a two-dimensional electron system. In addition, if an ultrafine line structure such as a nanowire is used instead of a thin film, the dimension can be further reduced and a one-dimensional electron system can be formed.
Assuming the band structure as shown in FIG. 2, as described above, in the bulk, the state corresponding to the valley bottom (0, 0, ± k ₀ ) in FIG. 1 comes to the Γ point in FIG. Recognize. As described above, the electrons in this state are not moving in the z-axis direction.

  Returning to the basics of such device physics, the idea is that the electron existing at the Γ point in FIG. 2 should recombine efficiently with the hole and be used as a light emitting element. In other words, by confining the electrons, the electrons cannot move freely, so when they collide with a hole with a small momentum because they are also at the Γ point, the momentum and energy conservation laws are broken. It can be released without any problems. As described above, momentum is a measure of how much impact a particle scatters when it collides with another particle. We have realized that if we make an electron immobile by confining it in a small area, the momentum of the electron will be lost. If the momentum of electrons is reduced, it becomes possible to satisfy the law of conservation of momentum at the time of scattering, which was difficult in the conventional method, so that even a group IV semiconductor such as silicon can shine efficiently.

  Based on such an idea, actually, an ultrathin Si film is partially formed on a substrate with a size of 1 cm square, and the photoluminescence measurement results are shown in FIG. 3, FIG. 4, and FIG. FIG. 3 and FIG. 6 show the intensity of light emission by photoluminescence. This shows that a very strong emission intensity was observed from the ultra-thin Si film. This intensity is several orders of magnitude greater than light emission due to indirect transition of bulk silicon. In other words, it is considered that by confining electrons in a narrow region, a group IV semiconductor such as silicon effectively changed to a direct transition type.

  FIG. 4 shows the wavelength of the peak of the spectrum obtained during this experiment. From this, it was confirmed that a wavelength larger than the band gap of silicon by the energy indicated by (Equation 14) was obtained. This reflects the fact that the band gap is large as the energy becomes discrete due to the quantum confinement effect, which indicates that the above principle is correct. FIG. 5 shows the result of calculating how much the emission wavelength changes as a result of the increased band gap. As described above, it is possible to effectively make the valley of energy a Γ point by thinning the silicon.

  In Patent Document 2, on the basis of the element structure of Patent Document 1, in order to efficiently use a silicon thin film as a light source of a silicon laser, an element in which thin films are arranged in fins at intervals of half the emission wavelength λ and combined with a waveguide The structure is also shown. As a method for manufacturing the fin-like structure, a thin line pattern is drawn on a mask and then deeply etched, and after alternately laminating silicon layers and silicon germanium layers using the MBE method, the silicon germanium layer is selectively formed. A method is shown in which only a fin-like silicon layer is left by etching.

JP 2007-294628 JP 2008-205006

R. Chan, M. Feng, N. Holonyak, Jr., A. James, and G. Walter, Appl. Phys. Lett., 2006, 88, pp. 143508-1 143508-3 L. T. Canham, Applied Physics Letters (Appl. Phys. Lett.), 1990, 57, pp. 1046-1048 S. Coffa, G. Franzo, and F. Priolo, Applied Physics Letters (1996), 69, pp. 2077-2079 F. Iacona, G. Franzo, E. C. Moreira, and F. Priolo, Journal of Applied Phys. (2001), 89, pp. 8354-8356 S. Coffa, IEEE Spectrum, 2005, Oct., pp.44-49 A. W. Fang, H. Park, R. Jones, O. Cohen, M. J. Paniccia, J. E. Bowers, I.E.E.Photonics Technology Letters, 2006, 18, pp. 1143-1145

  In the inventions of Patent Document 1 and Patent Document 2 described above, a light emitting element capable of injecting electrons and holes into an ultrathin silicon film to cause electroluminescence is provided. However, the luminous efficiency of the device is significantly higher than that of bulk silicon, but is still small compared to the luminous efficiency of III-V semiconductors such as GaAs.

  To create a silicon light emitting device with better performance, that is, a device that can obtain a higher light emission intensity with the same current and voltage, or a device that can obtain an equivalent light emission intensity with a little current and voltage, it is a light source. There is a need for a device that further enhances the luminous efficiency of the silicon thin film.

Also, when designing a silicon laser using silicon as the light source, the stronger the light emission intensity of the light source, the smaller the size of the light confinement mirror and the minimum required voltage / current value. Is born and preferable.
The present invention provides an element structure for further improving the light emission efficiency based on the structure of the light emitting element of Patent Document 2, and a method for manufacturing the element structure.

In order to find a way to improve the luminous efficiency of silicon light emitting devices, we calculate the electronic state of the silicon (001) thin film sandwiched between SiO ₂ from the first principle, and change the film thickness, interface structure, applied pressure, etc. The change in luminous efficiency was estimated.
As a result, it was found that the thinner the film thickness, the better the efficiency, and in particular, the 3 atomic layer, 7 atomic layer, and 11 atomic layer were in this order. However, to make use of this, it is not easy to control the film thickness in atomic layers and make it so thin, and if it is so thin, the energy gap becomes considerably large due to the quantum confinement effect, and the electrons for light emission In addition, there is a problem in that it becomes difficult to inject holes. When the current manufacturing technique for the thin film is used, a film thickness of about 1 to 10 nm is realistic considering cost performance. Furthermore, about 5 to 10 nm is preferable from the viewpoint of stability of control of the thin film. Theoretically, it has been confirmed that light emission occurs when the thickness of the thin film is 100 nm or less.

It was found that the interface structure also has a large effect on the luminous efficiency, but when the silicon film is thinned during the thermal oxidation process, it is a structure that occurs naturally between the oxidized SiO ₂ and the remaining silicon, Control is difficult unless the manufacturing method of the light emitting element is fundamentally changed.
According to the results of calculating the luminous efficiency by compressing or expanding the silicon (001) thin film in various directions, the luminous efficiency increases when compressed in the (001) direction, that is, the direction perpendicular to the film surface. all right. When compression is performed so that the volume shrinks by 2%, the pressure is about 1.5 GPa, and the luminous efficiency increases by about 10 to 30%. The reason for the increase in efficiency is that the degree of change differs depending on the film thickness (number of atomic layers). Since this effect is effective even in a silicon thin film having a film thickness of about 30 atomic layers that can be easily formed, it can be easily used for improving the efficiency of a light emitting element using a silicon (001) thin film.

The present invention provides a silicon light-emitting device that utilizes this pressure effect to enhance the light emission efficiency by adopting an element structure in which pressure is applied to the silicon thin film. Specifically, a plurality of silicon layers (films) sandwiching a hollow layer are arranged and thermally oxidized. When silicon is oxidized to SiO ₂ , the volume increases. As the oxidation progresses, the silicon thin film becomes thinner, but the SiO ₂ layer on the surface becomes thicker, and the hollow space gradually becomes narrower. Ultimately, the SiO ₂ layers grown from the silicon layers on both sides come into contact with each other at the midpoint, and pressure is generated by the SiO ₂ layers trying to grow thicker pressing each other. Thermal oxidation is terminated when the pressure reaches a predetermined level. Since the expansion rate of the SiO ₂ layer is known in advance, if the thickness and spacing of the silicon layer to be prepared first are designed appropriately, the thinness of the silicon layer and the applied pressure after the thermal oxidation treatment can be obtained as planned. Can do.

  When a light emitting element having such a basic structure is used as a light source for a silicon laser, the laser can be obtained by setting the silicon thin film array interval to half the emission wavelength λ (wavelength in the medium of the light emitting element). Convenient for causing oscillation. The silicon layers prepared first may be arranged at this λ / 2 interval. After that, it is only necessary to set the thickness of the initially arranged silicon layer in consideration of the thinness and pressure of the silicon thin film to be finally obtained.

  According to the present invention, a silicon light emitting element that can be easily formed on a substrate of silicon or the like using a normal silicon process can be provided at a lower cost. Similarly, among silicon lasers that can be easily formed on a substrate such as silicon by using a normal silicon process, a laser with increased design freedom can be provided at a low cost by a more powerful light source.

It is a figure which shows the band structure in the bulk state of the silicon | silicone for demonstrating the principle of this invention. It is a figure which shows the band structure in the thin film state of silicon, or a gate electric field application state for demonstrating the principle of this invention. It is a figure which shows the experimental data which demonstrate the principle of this invention, and the emitted light intensity from an ultra-thin silicon layer. It is an experimental data demonstrating the principle of this invention, and is a figure which shows the light emission wavelength from an ultra-thin silicon layer. It is a figure which shows the ultrathin silicon layer film thickness dependence of the light emission wavelength based on the principle of this invention. It is a figure which shows the dependence of the light emission wavelength and intensity | strength based on the principle of this invention on the ultrathin silicon layer film thickness. 1 is a plan layout illustrating an integrated light-emitting device according to Example 1 of the present invention. The element cross-section figure explaining the manufacturing process of the integrated light emitting element of Example 1 by this invention. The element cross-section figure explaining the manufacturing process of the integrated light emitting element of Example 1 by this invention. The element cross-section figure explaining the manufacturing process of the integrated light emitting element of Example 1 by this invention. The element cross-section figure explaining the manufacturing process of the integrated light emitting element of Example 1 by this invention. The element cross-section figure explaining the manufacturing process of the integrated light emitting element of Example 1 by this invention. The element cross-section figure explaining the manufacturing process of the integrated light emitting element of Example 1 by this invention. The plane layout explaining the integrated light emitting element of Example 2 by this invention. The element cross-section figure explaining the integrated light emitting element of Example 2 by this invention. The element cross-section figure explaining the manufacturing process of the integrated light emitting element of Example 2 by this invention. The element cross-section figure explaining the manufacturing process of the integrated light emitting element of Example 2 by this invention. The element cross-section figure explaining the manufacturing process of the integrated light emitting element of Example 2 by this invention. The element cross-section figure explaining the manufacturing process of the integrated light emitting element of Example 2 by this invention. The element cross-section figure explaining the manufacturing process of the integrated light emitting element of Example 2 by this invention. The element cross-section figure explaining the manufacturing process of the integrated light emitting element of Example 2 by this invention. The element cross-section figure explaining the manufacturing process of the integrated light emitting element of Example 2 by this invention.

  Embodiments will be described below in detail with reference to the drawings.

A procedure for preparing a fin-like silicon thin film array with a hollow layer sandwiched between them and manufacturing a light-emitting element using the thin-film array will be described in accordance with the method for manufacturing a light-emitting element described in Example 1 of Patent Document 2. Note that the present embodiment is different from Patent Document 2 in that the structure between the silicon thin films is finally filled with SiO ₂ .

  FIG. 7 shows a planar layout for forming an integrated light emitting device according to the present invention. 8 to 13 are device cross-sectional structure diagrams showing a process of forming an integrated light emitting device. The left side of each figure shows the A-A ′ section of the planar layout (FIG. 7), and the B-B ′ section of the planar layout (FIG. 7) on the right side.

An SOI wafer having a silicon oxide film (1900) having a thickness of 1 μm on a silicon support substrate (1100) and a single crystal silicon (1120) having a thickness of 100 nm on the silicon oxide film is thermally oxidized to obtain a film having a thickness of 20 nm. An oxide film (1910) is formed (FIG. 8). Using the active region pattern 1150 shown in FIG. 7, a silicon thin film region (fin) and a contact region are formed (FIG. 9). A thin-film single crystal region formed in a direction perpendicular to the substrate surface is hereinafter referred to as a fin. Here, thermal oxidation of the silicon surface exposed on the side surface of the fin is performed until the oxide film (1920) fills the hollow region between the fins and a predetermined pressure (1 GPa to 2 GPa) is applied between the fins. The thickness of the fin before oxidation is set in advance so that the resulting silicon layer has a predetermined thickness (1 nm to 5 nm). Further, the crystal orientation of the SOI wafer is set in advance so that the exposed silicon surface becomes the (001) plane (FIG. 10). A resist mask (1800) is formed with the 1850 hole pattern of FIG. 7, and an N-type impurity diffusion layer (1200) is formed by doping arsenic with an acceleration energy of 25 keV at 2 × 10 ¹⁵ cm ⁻³ by ion implantation (FIG. 7). 11). After that, although not shown, a resist mask in which the pattern 1850 is inverted is formed, and boron is implanted at 2 × 10 ¹⁵ cm ⁻³ at 5 keV by an ion implantation method, thereby forming a P-type impurity diffusion layer region (1300). . This forms a diode with a PN junction. Although an inversion mask is used here, the electric field can be set to a desired value by adjusting the distance between PN and providing an intrinsic region between PN.

A silicon nitride film is deposited to a thickness of 300 nm by CVD, and the waveguide 1500 is processed using the waveguide pattern (1550) of FIG. It is possible to obtain a configuration in which the fins whose gaps are filled with the oxide film are covered with the waveguide 1500. Here, the term “waveguide” is not limited to a single wavelength or the like, but is used as a path for transmitting light in a broad sense (FIG. 12). A contact hole (FIG. 7, 1650) is opened in the oxide film 1910, and a metal wiring 1600 is formed. (Omitted in FIG. 7)
Thereby, a forward bias can be applied to the PN junction by wiring in the P and N regions. Thereby, light emission can be obtained at the joints in the fins arranged in parallel. When a plurality of fins are arranged, the light emission can be effectively enhanced by arranging the fins at half wavelength intervals. Moreover, laser oscillation can be achieved by covering this structure with a reflective film layer.

  In the present embodiment, a method for realizing a light emitting element to which a similar pressure application is performed by laminating a silicon thin film in a direction parallel to the substrate surface will be described in accordance with an example in Embodiment 2 of Patent Document 2. In addition, it differs from the element of Patent Document 2 in that it has a structure in which the silicon thin film is finally filled with an oxide film.

  FIG. 14 shows a planar layout, and FIG. 15 shows a B-B ′ element cross-sectional structure of FIG. 14. The element manufacturing process will be described with reference to FIGS. 16 to 22 shown in the A-A ′ cross section of FIG. 14.

  In FIG. 16, thin films of silicon germanium 1121 and silicon 1120 having a thickness of about 3 nm to 10 nm are alternately grown on the SOI substrate 1120 by MBE. It is preferable to arrange the silicon thin film so that the period of the arrangement of the silicon thin film is exactly half of the emission wavelength λ, because it has an effect of strengthening light. When using it as a silicon laser light source, it is important to arrange it with the exact cycle. In addition, the distribution of the thickness of the silicon layer and the thickness of the silicon germanium layer in the period of λ / 2 is determined by removing the silicon germanium layer later and thermally oxidizing the silicon layer, so that the oxide film is hollow. Allocation is performed so that a predetermined silicon layer thickness (1 nm to 5 nm) is obtained when participation is performed until pressure is generated by filling the portion.

  In FIG. 17, the laminated film is etched using the active region pattern 1165 of FIG. 14 to obtain an etched laminated film (active region) 1155.

  In FIG. 18, a PN junction is formed using the ion implantation mask 1850 of FIG. 14 and its inverted pattern.

  In FIG. 19, the silicon germanium crystal layer is selectively etched using the thin film mask pattern 1165 of FIG. 14, thereby obtaining a hollow thin film structure of a silicon thin film having a PN junction.

  In FIG. 20, the thermal oxidation of the fin-shaped silicon thin film is performed until the oxide film (1920) fills the hollow region between the fins and a predetermined pressure (1 GPa to 2 GPa) is applied between the fins. Since the fin structure is prepared in advance so that the thickness of the oxidized silicon layer is a predetermined thickness (1 nm to 5 nm), a silicon thin film having a thickness suitable for light emission can be obtained. The crystal orientation of the SOI wafer is set in advance so that the surface direction of the silicon film is the (001) plane.

  In FIG. 21, a silicon nitride film is deposited and processed using a waveguide pattern (FIG. 14, 1550). At this time, a structure filled with a silicon nitride film is formed in the hollow region formed by the silicon thin film.

  In FIG. 22, by forming the interlayer insulating film and the metal wiring, an integrated light emitting element in which thin films are integrated in the vertical direction can be obtained.

1150,1550,1650,1850,1165 ... mask pattern,
1100,1120 ... single crystal silicon,
1900,1910,1920 ... silicon oxide film,
1800 ... photoresist,
1200,1300 ... impurity diffusion layer,
1500 ... waveguide,
1600 ... metal layer,
1121… Silicon germanium crystal,
1155 ... Active region.

Claims (11)

An insulating layer provided on a semiconductor substrate;
A light emitting portion provided on the insulating layer;
A first electrode for injecting electrons provided on a surface of the light-emitting portion and electrically connected to the light-emitting portion, and a second electrode for injecting holes;
The light emitting portion has a structure in which main surfaces of two or more thin films having a film thickness that emits light by injecting the electrons and holes are arranged in parallel so as to face each other,
The thin film is made of silicon crystal, and the plane orientation of the main surface of the thin film is a (001) plane or a plane orientation equivalent thereto,
A silicon light emitting device characterized by having a structure in which a space between the thin films arranged in parallel so as to face each other is filled with a silicon oxide film. A plurality of the thin films are periodically arranged at predetermined intervals so as to face each other,
2. The silicon light emitting device according to claim 1, wherein the predetermined interval is half of the wavelength of light emitted in the thin film within the thin film.   3. The silicon light emitting device according to claim 2, wherein the thin film has a multilayer structure in which a plurality of thin films are arranged in a direction perpendicular to the main surface of the insulating layer.   The silicon light emitting element according to claim 3, wherein the thickness of the thin film is 100 nm or less.   4. The silicon light emitting device according to claim 3, wherein the thickness of the thin film is in the range of 5 to 10 nm.   4. The silicon light-emitting element according to claim 3, wherein the thickness of the thin film is any one of a three-atom layer, a seven-atom layer, and an eleven-atom layer.   3. The silicon light emitting device according to claim 2, wherein the thin film has a multilayer structure in which a plurality of the thin films are arranged in a direction parallel to the main surface of the insulating layer.   8. The silicon light emitting device according to claim 7, wherein the thickness of the thin film is any one of a three atomic layer, a seven atomic layer, and an eleven atomic layer.   The silicon oxide film is provided between one thin film and one adjacent thin film, and is formed so as to apply compressive stress to each of the one thin film and another adjacent thin film. The silicon light emitting device according to claim 1, wherein:   2. The silicon light emitting device according to claim 1, wherein the silicon oxide film is a film formed by thermal oxidation between one thin film and one adjacent thin film.   The silicon oxide film thermally oxidizes silicon constituting the thin film between one of the thin films and one adjacent to the thin film, and the oxide films grown from both main surfaces of the thin films are mutually connected. 2. The silicon light-emitting element according to claim 1, wherein the silicon light-emitting element is a film formed such that a predetermined pressure is applied between the thin films by continuing thermal oxidation after contact.

Images