JPH0817230B2 - Semiconductor microfabrication method - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for microfabrication a semiconductor, particularly on a submicron to nanometer scale, and more particularly to a microfabrication method optimal for obtaining a fine embedded structure.

[0002]

2. Description of the Related Art Once the semiconductor microfabrication technology on the scale as described above is established, various functional elements such as so-called quantum wires, new functional elements using the same, and various functional elements that maximize the quantum effect, It is possible to provide a micro-cavity type semiconductor laser with no threshold value or a low threshold value, and it is possible to open an extremely wide range of applications, not only to meet the demand for miniaturization of functional elements. Therefore, various researches and developments have been made in the past regarding such fine processing technology, and the following two can be typically mentioned.

One of them is a report in which the inventor of the present invention is also involved. Prior art 1: Japanese Journal of Applied Physics,
In the method disclosed in Vol. 30, No. 12B, Dec. 1991, pp. 3879-3882, for example, a planar double heterostructure formed on a GaAs substrate is wet-etched.
A fine linear structure including an alternating layered structure of GaAs layers and AlGaAs layers is formed, and then the cladding layer is subjected to liquid phase growth to obtain a buried structure of the linear structure.

The other is a conventional document 2: Applied Physics Letter 60 (3), 20 Jan. 1
992, pp.365-367, in which a GaAs thin film (about 50 Å) for forming an oxide film is formed on an alternating laminated structure of a GaAs layer and an AlGaAs layer epitaxially grown on a GaAs substrate, A patterning step of selectively oxidizing the GaAs layer by electron beam excitation selective oxidation to obtain a desired pattern, and an etching step of etching using the oxidized GaAs pattern as a thin film mask,
Then, for the etched portion, all of the re-growth steps for re-growing crystals are continuously performed in the same apparatus without breaking the vacuum, and this is generally called "vacuum consistent process".

[0005]

In the method disclosed in the above-mentioned prior art document 1, there is a problem in liquid phase selective regrowth for completing the embedded structure with respect to the fine structure portion, and the surface portion of the underlying layer is generally Since it has been oxidized, especially in a semiconductor layer containing Al (for example, an AlGaAs layer) and the like, it is extremely easy to oxidize, and it is difficult to regrow crystals on it (in particular, when the Al composition ratio is 0.4 or more). Further, due to the surface tension of the melt, it is difficult to follow the fine structure portion, and the etching effect due to the meltback effect has been an obstacle to the formation of the fine structure.

On the other hand, in the vacuum consistent process as disclosed in the conventional document 2, the problem of oxidation is unlikely to occur, but a resist film as thick as a micron order to a submicron order as thin as a thin film, which is recognized in a general lithography technique. Instead, it is not desirable to use a thin oxide GaAs film of about several tens of liters as an etching mask and to require a special chemical reaction in the patterning process for selectively forming the oxide GaAs mask. For example, electron beam-excited selective gas etching, electron beam-excited selective oxide film growth, and other special methods required by the technique disclosed in this document, for the time being, determination of control parameters is difficult, and the selectivity ratio of the reaction itself, Alternatively, sufficient pattern accuracy has not yet been obtained due to the influence of mechanical vibration and the like associated with the connection between the etching apparatus and the growth apparatus.

Further, as recognized in Japanese Patent Laid-Open No. 3-49286, a processing method has been proposed in which MOCVD growth is carried out immediately after vapor phase etching in a single apparatus. However, in vapor phase etching, a mask pattern is used. Has a drawback that it is not possible to perform etching in a direction opposite to the crystal orientation because the ability to follow the fine shape is poor and the crystal orientation of the semiconductor to be processed is restricted. For example, a bent linear structure or a linear structure that smoothly draws a curvature cannot be formed,
In other words, it is not possible to perform highly accurate cutting in accordance with an arbitrary desired mask pattern shape.

The present invention has been made in view of the above-mentioned conventional technical circumstances, and has the problem of oxidation of the crystal regrowth underlayer as observed in the conventional document 1 and the thin problem as observed in the conventional document 2. The problem of the semiconductor oxide mask can be avoided, and it is also possible to escape from the poor followability to the mask pattern and the restriction of the crystal orientation, which are recognized in the technique disclosed in the above-mentioned Japanese Laid-Open Patent Publication, and to obtain any desired mask pattern shape. Accordingly, it is intended to disclose a method for obtaining an extremely fine semiconductor embedded structure with high accuracy.

[0009]

In order to achieve the above object, the present invention provides: (a) An inorganic material film which serves as a mask for etching and regrowth is formed on the surface of a semiconductor to be processed, and the inorganic material film is formed. And (b) carrying the semiconductor to be processed having the patterned inorganic material film for a mask into an ECR etching apparatus without exposing it to the atmosphere. And (c) MOCV provided separately from the ECR etching device without exposing the etched semiconductor to air.
And a re-growth step of re-growing a crystal in an etched portion, which is carried into a D device, and a fine processing method of a semiconductor is proposed.

Note that the material used as the mask in the above-described structure of the present invention is limited to the inorganic material film because the ECR etching rate is slower than that of the compound semiconductor which is the material to be etched, and a sufficient selection ratio is obtained. At the same time, it must be a material that can withstand the high temperature of the subsequent regrowth temperature (generally 700 ° C to 800 ° C). Typically, desirable materials for such an inorganic material include tungsten (W) and silicon oxide (SiO ₂ ), but it is also desirable that these are formed by sputtering. Furthermore, as described in the above-mentioned publication, the MOC
The VD device is also used and the vapor phase etching is also MOCV.
Unlike the case where D growth is also performed, in the present invention, the ECR etching apparatus and the MOCVD apparatus are separate apparatuses, and the semiconductor to be processed is transported between them without being exposed to the atmosphere. This is because the vacuum environment required for MOCVD growth and the vacuum environment for ECR etching are too different, and performing them in a single apparatus rather complicates the apparatus and makes it less practical. is there.

[0011]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 1 shows an experimental example of a semiconductor microfabrication process according to the present invention. First, as shown in FIG. 1 (A), a GaAs substrate 10 is formed, or as shown in FIG. On the AlGaAs layer 11, a tungsten (W) thin film having a thickness of about 100 nm is formed as an etching resist film by sputtering.
After patterning the resist film according to the desired pattern by optical lithography or electronic lithography with a normal lithography dedicated machine, tungsten thin film (etching mask) 12 with the desired pattern is formed by CF ₄ plasma etching.
Left.

After this, the processing device 13, shown in FIG.
That is, the sample introduction chamber 14 and the horizontal MOCVD growth furnace 15
The sample shown in FIG. 1 (A) is loaded into the sample introduction chamber 14 of the processing apparatus 13 provided with the ECR etching apparatus 16 between them, and the sample transfer apparatus (not shown) (known per se in this type of known semiconductor processing apparatus). 1) from the sample introduction chamber to the ECR etching device 16 according to FIG.
As shown in (B), using the tungsten thin film 12 as an etching mask, ECR etching with a depth of about 1 μm was performed.
In the case of the sample having the GaAs layer 11 epitaxially grown on the GaAs substrate 10, the etching was performed to such a depth that the substrate surface was exposed.

Next, without breaking the vacuum (without exposing to the atmosphere), the sample was immediately transferred from the ECR etching chamber 16 into the MOCVD growth furnace 15 by a sample transfer device (not shown), and as shown in FIG. 1 (C). The AlGaAs layer 17 was selectively grown by using the tungsten thin film 12 as a selective growth mask this time.

At this time, when the Al composition is large or when Zn is used as the dopant, a polycrystal grows on the tungsten mask 12 as shown by a virtual line 18 in FIG. 1 (C). Sometimes. If this is unnecessary or inconvenient, the sample may be transferred to the ECR etching chamber 16 again after the regrowth, and the polycrystalline thin film 18 may be removed. In fact, the polycrystalline thin film 18 thus formed is so thin that it can be easily removed. By repeating such regrowth and etching as many times as necessary, the regrowth surface is flattened as shown in FIG. 1 (D), or a multilayer structure or semiconductor of a plurality of layers, which is not shown in FIG. 1, is formed. A multilayer film structure can be obtained.

As shown in FIG. 1 (E), the GaAs layer is formed on the regrown AlGaAs layer 17 flattened as described above.
19 or the like can be further provided.

When the fine structure thus formed was actually observed by a scanning electron microscope, the polycrystalline thin film 18 was grown on the tungsten thin film (tungsten mask) 12, but it was thin and therefore was re-etched. It was easy to remove and the tungsten thin film 12 and GaAs layer 19
There was no adverse reaction with him either. Also,
The verticality of the side surface of the microstructure was sufficiently high, and high precision microfabrication results were confirmed.

By the way, the characteristics of the semiconductor laser have been gradually improved since the 1970s, but the power consumption per element is still several milliwatts or more even if it is low. In the future, it will be necessary to reduce this to at least tens of microwatts in order to obtain integration densities comparable to electronic integrated circuits.

As a matter of fact, one of the conventional semiconductor lasers that is small and highly efficient is the so-called quantum well type semiconductor laser. In this case, the geometrical dimensions of the quantum well layer or the active layer in which laser oscillation actually occurs are tentatively reduced to a thickness of about 10 nm, a width of about 2 μm, and a length of about 200 μm. It is 4 μm ³ . But this is not enough.

On the other hand, when the above-described fine processing method of the present invention is applied, the width can be accurately reduced to about 0.2 μm, and the reflectance can be improved by a device described below. The length can be reduced to about 20 μm. In this case, the volume of the active layer and the threshold current are one-hundredth of those of the above-mentioned existing lasers, and the threshold value is usually 3 to 10 mA to 0.03 to 0.1 mA.
In addition, the volume is reduced to about 0.04 μm ³ .

Furthermore, if a structure is formed in which the periphery of the active layer is covered with a regrown multilayer reflective layer so that spontaneous emission light cannot freely go out of the resonance cavity, the apparent oscillation threshold It is possible to make a laser without a laser. That is, in the conventional semiconductor laser, spontaneous emission light is first emitted as carriers are injected, and a part of it is amplified as an optical resonator mode, and when the amplification factor exceeds the loss, the optical resonance mode becomes dominant. Therefore, the stimulated emission follows the principle that efficient electron-light conversion is possible. In other words, there is a constant threshold current for efficient light emission. On the other hand, in the micro-semiconductor laser manufactured according to the present invention, if the periphery of the active layer is further covered with a three-dimensional high-reflectance mirror, the mode allowed in the optical resonator is reduced and the discretization is performed. The spontaneous emission light is also emitted as a limited mode. The spontaneous emission light in such a limited mode is randomly generated in time, and thus their temporal coherency cannot be maintained.
Since the emission distribution and the wavelength are determined by the cavity parameters, they are virtually indistinguishable from the laser light. Therefore, such a semiconductor laser can be regarded as having substantially no threshold value or an extremely low threshold value and can realize highly efficient laser oscillation at an extremely low current. Incidentally, regarding the apparent thresholdless or low-threshold laser, the conventional literature 3: “Microcavity laser: Present state and prospect”, Applied Physics, Vol. 61, No. 9 (1992), pp.890-901.

Although the present invention is useful for producing such a minute semiconductor laser, FIG. 3 shows an example of producing a laser according to the present invention which further adopts the distributed feedback principle. n type
On the GaAs substrate 10, an n-type lower clad layer 21, a pair of upper and lower green layers 22 and 22, and a p-type upper clad layer 24 sandwiching a laser active layer or a quantum well layer 23 therebetween are formed by a normal epitaxial growth technique. After sequentially forming the layers, as described above with reference to FIG. 1, using the existing dedicated lithography machine,
The tungsten thin film is patterned into a desired shape. In the illustrated case, the patterned tungsten thin film 12
Has a width dimension of about 200 nm and a length of 30 μm or less. Further, its planar shape is such that the side surface periodically has irregularities in the length direction. The period of the side surface unevenness is one quarter of the wavelength in the medium of light, and the central portion in the length direction has a convex portion with a half wavelength so that the phase shifter 29 is formed underneath. They are continuous and are designed to realize an optical resonator at the Bragg reflection wavelength. In terms of corresponding to the process example of FIG. 1, the n-type lower clad layer 21, the laser active layer or the quantum well layer 23, the pair of upper and lower green layers 22 and 22, and the p-type upper clad layer 24 are described. It may be considered that the laminated structure consisting of all of them collectively corresponds to the epitaxial layer 11 on the substrate 10 shown in FIG.

The tungsten thin film 12 having the above-mentioned pattern
Then, the sample with the etching mask as the etching mask is introduced into the ECR etching chamber 16 of the etching and regrowth processing apparatus 13 shown in FIG. 2 for etching, and immediately thereafter, it is transferred to the MOCVD regrowth furnace 15 without being exposed to the atmosphere. , The tungsten thin film 12 was used as a selective mask for regrowth this time, and the entire periphery of the remaining portion was regrown i
Filled with a type (ie, semi-insulating) AlGaAs peripheral cladding layer 17. In this embodiment, both end faces in the length direction, which are the light emitting end faces, are also covered with the AlGaAs cladding layer 17 in the same manner as the side face side, and the laser structure portion is completely embedded. This is a desirable structure especially for a laser that oscillates a laser beam having a short wavelength. This is because an effect of suppressing recombination at the light emitting end face can be expected.

After this, an appropriate metal electrode, preferably Au electrode 25 is formed on the tungsten pattern 12 to complete the fine distributed feedback semiconductor laser. In order to reduce the device resistance, as described above, after the polycrystalline AlGaAs that may be deposited on the tungsten thin film 12 by etching through the regrowth step is removed by etching,
It is also effective to selectively grow a doped AlGaAs layer or a GaAs layer.

In the actual manufacturing of the semiconductor laser shown in FIG. 3, the pattern shape of the tungsten thin film 12 as a minute pattern has an accuracy of several tens of nanometers, which is as high as one quarter of the wavelength in the medium of light. Formed by. According to the present invention, if the patterning accuracy can be further improved to a few nm accuracy in the future, the quantum level of the quantum wire can be controlled by the shape in the lateral direction orthogonal to the length direction. The level can be made smaller than that of the waveguide, and the light absorption in the waveguide can be further reduced.

In the fine distributed feedback laser manufactured by the above-described method according to the present invention, the quantum well width and the like are determined by the initial uniform crystal growth, and the side surface shape is for dry etching regardless of the crystal orientation. Since it is determined by an arbitrary pattern of the plane shape of the mask, the direction of the laser stripe can be freely set.

Therefore, the present invention is not limited to such a laser manufacturing example, and in a sense, the present invention is useful for manufacturing an electron waveguide as a basic functional element, which is also a quantum wire having an arbitrary plane shape or path pattern. Is. The quantum wire structure is basically the same or similar to the cross-sectional structure of the semiconductor laser shown in FIG. 3, but in the GaAs system, at least the AlGaAs layer which is the lower clad layer on the GaAs substrate, A GaAs layer having a thickness on the order of electron wavelengths (about 10 nm) as a substantial waveguide layer or core layer, and AlGa being an upper clad layer.
Any As layer may be formed by stacking As layers and patterning them linearly in plan view, and then re-growing AlGaAs layers on both side surfaces to form left and right cladding layers. Therefore, it can be manufactured by effectively utilizing the method of the present invention described above. Further, according to the present invention, when such a quantum wire is formed, the pattern itself is formed into a Y-shaped branch shape at the time of patterning the etching and regrowth mask 12, for example. A V-shaped electron or optical branch path can be formed, and a so-called Mach-Cendar interference element or the like can also be manufactured by patterning into a shape such that it splits into two and is once again grouped into one line or waveguide. Furthermore, a four-terminal directional coupler can be realized by patterning one main waveguide so that the sub-waveguide approaches the middle of the length and leaves the main waveguide again. In other words, according to the present invention, an extremely fine and highly accurate device can be used to manufacture various electronic or optical passive functional devices that realize a specific function depending on the characteristics related to the patterning shape of the waveguide. Can be provided.

FIG. 4 shows another semiconductor microlaser, which is a microcavity laser manufactured according to the present invention. Like the semiconductor laser described above, the laser function portion has a laser structure portion in which the GaAs quantum well layer 23 as an active layer is sandwiched by upper and lower clad layers.
Furthermore, a semiconductor multilayer film, especially n-type in this case, is formed above and below
A multilayer film of AlGaAs (for example, an alternating laminated structure of AlGaAs layers having an Al composition ratio of 0.2 and 0.7) and a multilayer film 27 having an alternating laminated structure of p-type AlGaAs having the same different Al composition ratio. Sandwiched between. Also, re-growth provided on both left and right sides
The AlGaAs clad layer is also i-type and has a different Al composition ratio.
It is an AlGaAs multilayer film. Thus, by embedding the left and right upper and lower sides of the active layer or the core layer with the semiconductor multilayer film, high reflection characteristics can be obtained. Since the free emission of the radiated light can be limited, as far as the spatial emission pattern and the emission wavelength are concerned, it is possible to obtain a characteristic close to a threshold value.

Although the microcavity laser shown in FIG. 4 is constructed as an end face emission type, the length direction in the drawing is shortened to the same extent as the lateral width (that is, several hundred nm).
However, by increasing the number of layers and increasing the reflectance in the thickness direction, so-called surface-emitting lasers that emit laser light in the direction perpendicular to the substrate should also have characteristics close to thresholdless. Can be realized. Of course, although the present invention is effective for producing a microstructure of a semiconductor, particularly a compound semiconductor,
As a matter of course, the structure can be applied to a structure having a slightly larger size. Therefore, the semiconductor laser having the structure shown in FIG. 4 is also formed to have an existing class size, that is, a width of about 2 μm and a length of about 100 μm. You may.

FIG. 5 also shows the integrated structure of a surface-emitting laser advantageously produced by the method according to the invention. Since each laser has the same configuration, it can be explained by focusing on one,
Generally, in a surface emitting laser, it is desirable to have a structure in which a large number of quantum wells are stacked in order to obtain a laser gain in the thickness direction. However, it has been difficult to uniformly inject minority carriers into the active layer in the pn junction formed by simply stacking layers as described above. Further, if the active layer portion is simply filled with the p-type and n-type cladding layers, current leakage occurs at the boundary between the cladding layers, which deteriorates the efficiency.

On the other hand, the use of the fine processing method of the present invention makes it possible to obtain the desirable structure shown in the drawing with high accuracy. That is, in the illustrated structure, the laser structure portion laminated on the GaAs substrate 10, particularly the lateral portion of the active layer composed of a plurality of quantum well layers 23 (multiple quantum well structure) laminated, The current blocking layers 33 and 34 having a thin reverse pn junction are buried so as to be separated between the p-type cladding layer 31 and the n-type cladding layer 32. Therefore, the lateral injection current can be concentrated in the active layer having the multiple quantum well structure as expected,
The efficiency can be greatly improved.

The structure shown in FIG. 5 can be manufactured by the following process according to the present invention. First, a MOCV stack structure is formed on a GaAs substrate to form a vertical optical resonator composed of the multiple quantum well structure and the semiconductor multilayer film described above.
Epitaxial growth is performed by D or MBE. Next, a tungsten thin film 12 (not shown in this drawing: see FIG. 1) is formed on the tungsten thin film 12 according to the process shown in FIG.
The tungsten thin film in the area where the type cladding layer 32 is to be formed and the area where the n-type current blocking layer 33 is to be formed are removed.

This sample was placed in the processing apparatus 13 for the in-situ etching / regrowth integrated process shown in FIG. 2, and after etching in the ECR etching chamber 16, the n-type cladding layer 32 and the n-type current were set in the growth furnace 15. The block layer 33 is grown. Then, the sample is returned to the dedicated lithography machine again, and the tungsten thin film 12 is patterned so as to remove the area portions where the p-type cladding layer 31 and the p-type current blocking layer 34 are to be formed. After the etching in the ECR etching chamber 16 by the etching / regrowth integrated process, the p-type cladding layer 31 and the p-type current blocking layer 34 are grown in the growth furnace 15. In this way, a lateral current injection type distributed feedback micro surface emitting laser with a lateral current blocking layer can be realized. By the way, the thickness of the effective portion (columnar portion including the active layer) of this semiconductor laser is about 3
The diameter of the circular cavity opening surface, which is the light emitting surface when viewed two-dimensionally, is about 1 μm.

The present invention directly determines whether or not the semiconductor multilayer film sandwiching the quantum well is provided, and in which position in the height direction with respect to the substrate a plurality of laminated structures of the quantum well are provided. It is not specified but depends on the design. For example, in the semiconductor laser as shown in FIG.
When the p-type and n-type clad layers are alternately provided in the circumferential direction as shown in FIG. 6, a surface emitting laser with a beam deflecting function can be realized.

In the case of the structure shown in FIG. 6, four pillars are formed around the substantial light emitting surface portion (cavity opening surface) 40 of the columnar body including the quantum well layer as described above. p-type clad layers p1 to p4, n-type clad layer n1
~ N4 are provided. Therefore, assuming that a dedicated electrode is provided in each clad layer and, for example, only the first p-type clad layer p1 and the second n-type clad layer n2 are energized, the second and fourth p-type clad layers p2, p4 , First and fourth n-type cladding layers
n1 and n4 each act as the current blocking layer described above, and among the several possible current paths shown in the drawing, the current at this time passes through the path indicated by the arrow A. On the other hand, if the electrode attached to the first p-type cladding layer p1 is
When a current is applied to the electrode attached to the mold cladding layer n3, the current path is as shown by arrow B. Obviously, such a difference in the current path causes a change in the light emission pattern on the cavity opening surface 40, and the emission pattern can be modulated even in the far field image.

The present invention is also effective for manufacturing a so-called micro FET. As already known, the mobility increases in an electron waveguide in which electrons are confined in a one-dimensional direction. Further, in order to control a very small channel current, even in a FET having a MOS structure or a structure similar thereto, it is effective to shorten the channel width. However, the conventional electron flow path in the MOS structure or the like is determined by the doping concentration and the electric field of the space charge layer.
It is not a geometrically closed area. Therefore, conventionally, it has been difficult to definitively limit the channel size to 0.2 to 3 μm or less, which is generally called the device length.

On the other hand, if the present invention is adopted, the width of the electron channel can be relatively easily set to 20 while geometrically defining the electron channel.
Since it can be less than or equal to nm, in addition to the branch path and Mach-Cendar interferometer mentioned above, the channel path between the source and the drain and the gate line are made with a quantum wire to have an extremely small width. You can

One example of the manufacturing process of such a micro FET or quantum wire FET is shown in FIG. First, as shown in FIG. 7A, on a suitable semiconductor substrate or semi-insulating substrate such as a GaAs substrate (for simplicity, the substrate is not shown), by a known existing method, for example, The non-doped or n-type AlGaAs layer 41 having a thickness of about 100 nm, the non-doped GaAs or non-doped InGaAs layer 42 having a thickness of about several nm, and the non-doped or n-type AlGaAs layer 43 having a thickness of about 100 nm are again MBE or M.
Laminated by OCVD. These laminated layers 41,
The group of 42 and 43 corresponds to the epitaxial growth layer 11 in FIG. 1 in correspondence with the basic embodiment described in FIG.

Next, although this may be done by a known existing lithography-dedicated machine, a tungsten thin film 12 is formed and patterned into a desired shape. FIG. 7 (B) shows a state in which this patterning has been completed. The tungsten thin film 12 has a source forming portion 44S, a channel forming portion 44C, and a drain forming portion underneath.
44D is a series of first pattern parts, a part 44G that is close to and opposite to the channel forming part 44C, and a gate will be formed underneath it in the future 44G, and is electrically and geometrically connected to the gate. The second pattern portion is formed by connecting the portion 44GT for forming the.

After such a patterning step, the sample is introduced into the processing apparatus 13 capable of in-situ etching and regrowth processing by the vacuum consistent process as shown in FIG. Etching (in this case, electron beam excitation dry etching is also possible) is performed, and as shown in FIG. 7C, the portion of the tungsten thin film 12 other than the pattern is etched to a predetermined depth.

After this etching, according to the present invention, the sample was transferred into the MOCVD growth furnace without breaking the vacuum, and the non-doped AlGaAs layer 17 was formed around the portion left unetched.
Re-grow. As a result, as shown in FIG. 7D, an embedded structure of the ultrafine quantum wire FET 50 having the source S, the channel C, the drain D, and the gate G is obtained by the quantum wire. In such a structure, a gate length of 20 nm and a channel size of 10 nm × 10 nm can be practically used, and the channel length can be shorter than the electron mean free path. Not only is the speed high, but the operation is characterized in that the channel conductance changes depending on the tunnel probability of the channel barrier that changes depending on the magnitude of the gate electric field.

In FIG. 8, three quantum wire FETs 50 shown in FIG. 7 are used as shown individually by reference numerals 51, 52 and 53, respectively.
The structural example integrated so that an inverter circuit may be shown is shown. The fabrication process itself has already been described with reference to FIG. 7, and the three quantum wire FETs 51, 52, 53 constructed on the GaAs semi-insulating substrate and the quantum wire parts connected to them are regrown non-doped AlGaAs layers. Embedded in 17.
However, the tungsten patterning thin film which may remain is not shown. One of the characteristic features of the configuration is that the line portion 56 from the channel quantum wire 56 of the previous stage FET to the drain is directly used as the gate quantum wire 55 of the next stage, which minimizes the parasitic capacitance. Can be
High-speed operability can be guaranteed. The buried layer is a source S and a drain D at each stage (for the first and second stage FETs 51 and 52, each drain has a tunnel junction 5 which will be described later).
It is connected to the ground electrode 58 via 7, 57, and the drain D at the final stage also serves as an output terminal), but basically the device can be operated without being doped.

However, if the dimension is less than several hundreds nm, it is difficult to form a reproducible resistance due to the fluctuation of the conductance due to the electron traveling path. Therefore, as the load resistance, as shown in FIG. It is better to use the joints 57 and 57.

Speaking of a tunnel junction, as shown in FIG. 9, a pair of quantum wires 61 and 62 are arranged on a GaAs semi-insulating substrate 10 according to the present invention so that their ends are closely opposed to each other. There are two tunnels, one is arranged between the quantum dots 63 and each of the quantum wires 61 and 62, and a cylindrical quantum dot 63 is provided between the ends of the quantum dots 63 and the surroundings are filled with the regrowth layer 17. It is also possible to construct a functional element having a barrier formed. Although the cross-sectional shapes of the quantum wires 61 and 62 and the cylindrical quantum dot 63 are not shown in detail in this figure, they are shown in FIG.
A laminated structure 41, 42, 43 similar to that shown in
The same applies to the regrown buried layer 17. Such elements are
Usually, it can be called a double barrier tunnel element. By combining it with an active element such as the quantum wire FET described above,
There are applications as a memory utilizing the Coulomb blockade phenomenon and as a negative resistance characteristic (particularly N-type negative resistance characteristic) element.

Furthermore, an electron-optical coupling device such as a combination of the quantum wave device represented by the FETs shown in FIGS. 7 and 8 and the microcavity laser shown in FIG.
According to the present invention, it can be practically manufactured. FIG. 10 shows such an example, which may be similar to the microcavity already described with reference to FIG.
Surrounds the quantum wave device 50 including the quantum dots 63. Therefore, in the microcavity 71, the electron system and the photon system can strongly interact with each other, and the number of states that light can take is limited to several or less, so that the photon and the electron-hole pair are reversible. It can be converted. That is, if the quantum wave element 50 is designed to perform a logical operation and the result is expressed as a change in the charge distribution of the quantum dot 63 or the built-in electric field, the quantum level in the quantum dot 63 becomes a resonance of the microcavity 71. A specific photon can be efficiently emitted at the moment when the wavelength matches the wavelength.

Therefore, if a plurality of microcavities 71 having the above-mentioned structure are connected to each other by an optical waveguide 72, the arithmetic and storage operations are electronic systems, and the transmission is optical. , This has been proposed in the past,
In other words, the method of sharing roles can be performed even in a micro device. In the electron waveguide, since electrons are fermions, only one electron can be accommodated in one state, and 12.9 KΩ per mode even in the ideal state.
However, if photons, which are poson particles, are used for transmission, there is no limit to the number of particles that can be accommodated in the fundamental mode of the optical waveguide, and the transmission speed is significantly improved.

Although the present invention and some of the examples of devices which can be produced by the present invention have been described above in detail, various modifications are possible as long as they are in accordance with the gist of the present invention. Further, in the above embodiment, the mask layer for etching and regrowth, which is provided on the growth layer formed on the semiconductor or the semi-insulating substrate and is patterned first, is the tungsten thin film, but the present invention is not limited to this. In particular, silicon oxide and the like can be advantageously used like tungsten.

[0047]

According to the present invention, the problems of the prior art can be solved and an extremely fine semiconductor embedded structure can be realized by a relatively simple process. When the present invention is used for microfabrication of a structure containing a heterostructure, this embedded structure is
Since the hetero interface does not include an oxide film, it is not necessary to consider the depletion layer due to the surface level, and ultrafine electron or optical waveguides, micro distributed feedback lasers, microcavity lasers, lateral injection type surface emitting ultrafine lasers, etc. It can be manufactured, and a quantum wire FET or the like can be realized. Furthermore, it is possible to realize a multi-functional element which is itself a plurality of integrated fine photo-electron coupling elements, which can contribute to the provision of an ultra-high performance information processing device.

[Brief description of drawings]

FIG. 1 is an explanatory diagram of an example of basic steps of a method of the present invention.

FIG. 2 is a schematic configuration diagram of an example of an apparatus that can be used in the method of the present invention.

FIG. 3 is a schematic configuration diagram of a micro distributed feedback laser that can be manufactured by the method of the present invention.

FIG. 4 is a schematic configuration diagram of an edge-emitting microcavity laser that can be manufactured by the method of the present invention.

FIG. 5 is a schematic configuration diagram of a surface emitting type distributed feedback laser that can be manufactured by the method of the present invention.

FIG. 6 is an explanatory diagram of a main part of a surface emitting microlaser with a beam deflecting function that can be manufactured by the method of the present invention.

FIG. 7 is an explanatory diagram of an example of a manufacturing process of a quantum wire FET that can be manufactured by the method of the present invention.

FIG. 8 is a schematic configuration diagram of a fine inverter circuit that can be manufactured by the method of the present invention.

FIG. 9 is a schematic configuration diagram of a fine double barrier tunnel element that can be manufactured by the method of the present invention.

FIG. 10 is a schematic configuration diagram of an example of an electron-optical coupling element that can be manufactured by the present invention.

[Explanation of symbols]

 10 semiconductor or semi-insulating substrate, 11 epitaxial growth layer, 12 etching and regrowth mask, 13 etching and regrowth processing equipment, 14 sample introduction chamber, 15 MOCVD growth furnace, 16 ECR etching chamber, 17 regrowth layer.

Claims (3)

[Claims] To 1. A workpiece semiconductor on the surface, forming an inorganic material film as a mask for etching and regrowth, the inorganic <br/> material film, according to the planar shape of the microstructure to be formed A step of patterning into a planar shape; an ECR etching apparatus without exposing the semiconductor to be processed having the patterned inorganic material film for a mask to the atmosphere
And carrying out etching to the substrate ; exposing the etched semiconductor to the atmosphere
M provided separately from the above ECR etching device
Carry it in the OCVD equipment and bond it to the above-mentioned etched part.
Semiconductor microfabrication method comprising forming Rukoto comprise; a regrowth step of regrowing crystal. 2. The method according to claim 1, wherein the inorganic material film is a tungsten thin film formed by sputtering on the surface of the semiconductor to be processed. 3. The method according to claim 1, wherein the inorganic material film is a silicon oxide thin film formed by sputtering on the surface of the semiconductor to be processed.