WO2017145706A1

WO2017145706A1 - Method for bending thin film member

Info

Publication number: WO2017145706A1
Application number: PCT/JP2017/003980
Authority: WO
Inventors: 知也吉田; 榊原　陽一
Original assignee: 国立研究開発法人産業技術総合研究所
Priority date: 2016-02-25
Filing date: 2017-02-03
Publication date: 2017-08-31
Also published as: JPWO2017145706A1; JP6741921B2

Abstract

The present invention addresses the problem of achieving a method such as isotropic etching for bending a thin film member in which the etching stop position differs from the position at which bending from IIB processing begins. Provided is a method for bending a thin film member comprising: a step for laminating a support layer formation starting layer 20' and a thin film member formation starting layer 30' on a substrate 10; a step for forming the thin film member 30 by patterning the thin film member formation starting layer 30'; a step for forming, on the thin film member via a position-controlling layer support layer formation starting layer 60', a position-controlling layer 50 for determining the position at which a cantilever structure of the thin film member begins to bend as a result of ion irradiation; a step for removing a portion of the support layer formation starting layer 20' below the thin film member and a portion of the position-controlling layer support layer formation starting layer 60' so that the circumferential surface 21 of the support layer and the end face 61 of the position-controlling layer support layer 60 are on or inside a line extended downward from the leading edge 51 of the position-controlling material layer to form the cantilever structure 40, which has a free leading end 41, on the thin film member 30; and a step for irradiating ions on a portion of the cantilever structure with the position-controlling layer 50 as a mask to bend a portion of the cantilever structure 40.

Description

Method of bending thin film member

The present invention relates to a fine bending process of a thin film member to which an ion implantation technique is applied, and relates to a method of bending a thin film member that improves the control of the bending position and improves the integration density of the structure.
The technical fields to which the present invention contributes include the manufacture of semiconductor integrated circuits, the manufacture of NEMS / MEMS (microelectromechanical system) devices, the manufacture of optical devices and optical integrated circuits, and the manufacture of integrated devices integrating them. The present invention is applied to a technical field that requires fine processing on the order of micrometers to nanometers of a thin film member, particularly to a technical field that requires a three-dimensional structure instead of a conventional planar structure.

The manufacturing of devices with micrometer to nanometer order dimensions as described above is centered on photolithography technology, semiconductor materials such as silicon, insulator materials such as quartz (silica), and conductors such as metals. It consists of a process in which materials are stacked in layers by repeating film formation and patterning. A deposition technique for depositing the material and an etching technique for patterning the deposited thin film are the basic technologies next to the photolithography technique in this series of processes.

However, with some exceptions, the conventional deposition technique and etching technique are limited in that the dimension of the thin film member to be processed is less than about 1 μm in thickness. For example, in the case of a deposition technique, in order to form a film having a thickness of 200 nm or more, the substrate is warped due to an increase in internal stress of the thin film, the deposited thin film is peeled off, or a uniform thin film is formed on the entire surface of the substrate. There are problems such as difficulty and increase in maintenance frequency of the film forming apparatus. In some insulator materials, film formation with a thickness of 200 nm or more has been put to practical use by overcoming these problems, but the actual situation is that the thickness is still limited to less than about 1 μm.

On the other hand, in the case of an etching technique, in order to carry out etching with a depth of 200 nm or more, there are problems of durability of the resist material, and in the case of fine holes, there are problems such as shape controllability. However, in recent years, deep etching of silicon and the like has been put into practical use and processing of a depth of 100 μm has been put into practical use, but the application field is almost limited to the manufacture of NEMS / MEMS devices and the production of through-hole electrodes of silicon wafers. ing.
That is, it can be said that the existing fine processing technique is a processing technique in an extremely thin planar region of less than 1 μm at most on the surface of the substrate, and is not good at processing a three-dimensional structure having a height exceeding 1 μm. .

In recent years, a thin film member bending technique using an ion implantation technique has been disclosed as a fine processing technique that realizes a three-dimensional structure that is difficult to achieve with only the deposition technique and the etching technique. (See Patent Documents 1 to 4 and Non-Patent Documents 5 to 7)
According to this method, a three-dimensional fine structure of 1 μm to 100 μm, which could not be constructed by the deposition technique or the etching technique, can be formed, and its application is extremely wide.

In this bending process technology by ion implantation (IIB technology (Ion Implantation Bending) or Ion Induced Bending), ions are implanted into a thin film member, and the phenomenon that stress caused by the implanted ions causes deformation of the thin film is used. The thin film member to be processed has a cantilever structure formed in advance, and ion implantation is performed on the cantilever structure. The planar shape of the cantilever structure is not limited to a simple shape such as a rectangle or a triangle, and even a complicated shape including an arc can be freely selected to some extent by using photolithography and etching. That is, the IIB technique is a technique that can be formed into a three-dimensional shape by bending a thin film patterned in two dimensions. In addition, the radius of curvature of the bending process can be a gentle curve on the order of micrometers or a small angle bent vertically on the order of nanometers.

The ion implantation technology that forms the basis of the IIB technology is a core technology used for introducing impurities in the silicon LSI manufacturing process, and is already a mature mass production technology. Since the IIB technology applies ion implantation, which is a mature technology, the processing uniformity and reproducibility are extremely high. Furthermore, it is a technology having high process integration with existing electronic devices such as LSI, NEMS / MEMS devices, optical devices, and optical integrated circuits.

Of course, if it is only a bending process of a thin film member, “micro origami” disclosed in Non-Patent Document 1, a vacuum electron source manufacturing method disclosed in Non-Patent Document 2, and Non-Patent Document 3 is a method for producing a metamaterial, and a processing method in a method for producing a silicon optical coupler disclosed in Non-Patent Document 4. However, these are all techniques for bending by releasing internal stress generated during film formation of the thin film member. In these methods, two kinds of materials having different stresses are laminated, and then the thin film member supporting layer called a sacrificial layer is removed to release the internal stress and to physically self-deform the thin film member.

However, in this method, since the internal stress of the thin film member needs to be controlled with high accuracy, there are few choices of applicable materials, and there are disadvantages in terms of productivity and versatility. Further, the radius of curvature that can be realized is about 64 μm in the example of the film thickness of 2.1 μm, about 20 μm in the example of the film thickness of 0.11 μm, and about 1.1 μm in the example of the film thickness of 0.07 μm. The thickness of the thin film member is as large as 15 to 180 times, and there is a limit to miniaturization. In addition, since the thin film member must be selected with priority given to the control of internal stress, the essential electrical and optical characteristics of the device are sacrificed.

On the other hand, the above-mentioned IIB technique realizes bending of a fine thin film structure using ions generated by implanting ions into the thin film member. That is, the kinetic energy of irradiated ions is the driving force for deformation. Since the stress can be controlled by the kinetic energy of ions and the implantation amount, extremely high in-plane uniform control is possible.
In addition, the radius of curvature that can be realized is 0.1 μm in the case of bending a thin film having a thickness of 0.1 μm, and the thin film bending process is driven by the internal stress of a conventional laminated film represented by the above-mentioned micro origami technology. Compared with the technology, it is possible to perform bending processing with a fine size of 1/15 to 1/180.

The IIB technology is a very versatile technology. Although it has already been disclosed in Patent Document 2, it can be said that the thin film member can be bent almost regardless of the material.
Inferring from its contents, among the materials often used in microfabrication processes, semiconductors such as silicon, molybdenum, tungsten, tantalum, technetium, rhenium, cobalt, nickel, ruthenium, osmium, rhodium, iridium, palladium, Metals such as platinum, copper, silver, gold, lead, germanium, silicon compounds such as tungsten silicide, molybdenum silicide, tantalum silicide, titanium silicide, cobalt silicide, chromium silicide, nickel silicide, carbon, silicon carbide, tantalum carbide, Carbonaceous materials such as titanium carbide, molybdenum carbide, niobium carbide, hafnium carbide, tungsten carbide, vanadium carbide, diamond-like carbon, titanium nitride, aluminum nitride, silicon nitride, tan nitride Nitride such as copper, boron nitride, chromium nitride and zirconium nitride, and transparent conductive films such as indium tin oxide, zinc aluminum oxide, zinc oxide and indium gallium zinc oxide are most often used in the cantilever structure formation process Since it is characterized by high chemical resistance to hydrofluoric acid, it is easily used. It has also been found that a material with a small atomic number, a material with a small mass, and a material with a low density bend better, so that a large deformation angle can be obtained with a small dose.

Regarding the film thickness of the thin film member, as already disclosed in Patent Documents 1 to 4 and Non-Patent Documents 5 to 7, a thin film having a thickness of 20 nm to 240 nm is a main processing target. However, it can be processed if ion irradiation conditions are selected.
Further, regarding the length of the cantilever structure, as already disclosed in Patent Documents 1 to 4 and Non-Patent Documents 5 to 7, a cantilever beam of 0.5 μm to 50 μm can be processed. In principle, as long as a cantilever can be formed, it is clear that it can be processed even shorter or longer.

It has already been disclosed in Patent Document 2 that there are no particular restrictions on the ion species used in the IIB technology. Phosphorous, boron, arsenic, indium, antimony, boron fluoride, aluminum, nitrogen, argon, silicon fluoride, silicon, boron hydride, carbon hydride are used when an ion implantation apparatus is used, and gallium is used when an FIB apparatus is used. It is easy to use because it is generally provided.

In addition, it is not often used in semiconductor processes, but hydrogen, helium, carbon, oxygen, fluorine, neon, magnesium, sulfur, chlorine, scandium, titanium, vanadium, chromium, manganese, iron, cobalt, nickel, copper, zinc, germanium, Krypton, rubidium, zirconium, niobium, molybdenum, ruthenium, palladium, silver, cadmium, tin, iodine, xenon, hafnium, tungsten, iridium, platinum, gold, lead, bismuth, cerium, praseodymium, neodymium, samarium, eurobium, gadolinium, Terbium, dysprosium, erbium, thulium, ytterbium, and the like are ions that can be practically ion-implanted and can be easily used in the IIB technique.

It has also been found that a large deformation angle can be obtained with a small amount of irradiation because the ion species having a larger mass can bend the thin film member better. One type of ion species has higher controllability.
However, when mass spectrometry is difficult or when the IIB technique is carried out with an apparatus not equipped with a mass analyzer, it is possible to use ions in which a plurality of types are mixed.

[Application example to electrical measurement probe]
(Prior art and its problems)
In order to explain the merits of the IIB technology, an application example to an electrical measurement probe will be shown. A scanning probe microscope (SPM) probe such as an atomic force microscope (AFM) or a probe for a silicon LSI device tester uses a three-dimensional, high-aspect ratio sharp structure using etching technology. Has been. A type produced by anisotropic etching of a silicon single crystal is the mainstream, and a pyramid probe with a quadrangular pyramid having a base length of 10 μm to 100 μm and a height of 10 μm to 100 μm is widely used. The reason why single crystal silicon is selected as the material is that a sharp structure with a high aspect ratio is formed in a self-aligning manner by anisotropic etching.

However, since silicon has a higher electrical resistance than metal, deterioration due to repeated current flow is a problem. In addition, when used as a contact type probe, mechanical durability is also low because silicon is lower than metal, which is a problem. Originally, a highly durable metal such as tungsten is desirable as the material for the probe for electrical measurement, but since there is no other manufacturing process for forming a three-dimensional structure, it is currently being used as a substitute for silicon. is there. If a high aspect ratio probe can be formed of a metal such as tungsten, these problems can be overcome, and a probe shape and a probe array arrangement with a higher degree of freedom than before can be realized.

(Application of IIB technology)
Since the IIB technology can be applied to a wide range of materials, it is possible to easily bend a metal having a high melting point and high hardness, such as tungsten, which is highly difficult to process. The procedure is shown below. First, a triangle having a base of 1 μm and a height of 100 μm is two-dimensionally patterned on a tungsten thin film having a thickness of about 20 nm to 200 nm, using double patterning as necessary to sharpen the tip to the nanometer order. Next, the two-dimensional triangular patterned metal thin film is processed into a cantilever structure using a sacrificial layer etching technique and a supercritical drying technique. Then, by performing ion implantation on the cantilever structure of the triangular tungsten thin film, a high-aspect ratio three-dimensional structure having a height of 100 μm, a thickness of about 20 nm to 200 nm, and a base of 1 μm can be formed. The conventional anisotropic etching technology for silicon has to be a pyramid structure because the angle of crystal orientation to be etched is constant, but the shape can be designed with two-dimensional patterning using IIB technology. Therefore, the footprint can be greatly reduced. As a result, it is possible to form an ultra-high-density probe array with a metal material such as tungsten having excellent electrical characteristics and mechanical durability.
The process of forming a three-dimensional structure having a height of 10 μm to 100 μm by an etching technique as an example of such an electric measurement probe is generally a very special process in which materials and shapes are limited. On the other hand, since the IIB technology is a technology that can bend a wide range of materials, it can be said to be a highly versatile technology.

[Example of application to vacuum electron source]
(Conventional technology)
Next, an application example of the IIB technology to a vacuum electron source will be shown. A vacuum electron source device applied to an ultra-high sensitivity imaging device or the like is a device characterized by having a metal conical structure with a height of 1 μm called an emitter. Conventionally, the three-dimensional emitter structure has been fabricated by a technique called Spindt method using deposition technology.
First, the procedure will be described. First, a substrate in which a metal thin film (about 100 nm) / insulator thin film (about 1 μm) / metal thin film (about 100 nm) is laminated is prepared. Next, a hole having a diameter of about 1 μm is formed in the upper metal thin film. Thereafter, the insulator thin film is etched until the underlying metal thin film is exposed using the metal thin film having the holes formed therein as a mask. Then, a cylindrical hole having a diameter of 1 μm and a depth of about 1 μm is formed in the substrate.

Next, a metal such as molybdenum is vacuum deposited on the substrate. Then, the metal deposited on the upper part of the hole is reduced in diameter as the hole is deposited due to migration in the in-plane direction. Finally, the hole is completely closed by the deposited metal, and a conical structure is formed in the hole in a self-aligning manner accordingly. This method is the oldest method for manufacturing a vacuum electron source. However, many different approaches have been researched and developed since then, but none of the device performance exceeds that of the vacuum electron source based on the Spindt method.

(Prior art issues)
However, this conventional method has not been widely put into practical use.
This is because vacuum deposition of a metal material having a thickness of 1 μm or more is an essential condition for forming a conical shape. The material of the vacuum electron source is preferably a refractory metal such as tungsten or molybdenum. However, the refractory metal is a material that easily generates a large stress because the migration distance on the substrate surface during deposition is small. Moreover, the stress during deposition increases as the film thickness increases. Therefore, the process of depositing the refractory metal with a thickness of 1 μm is a condition in which peeling is very likely to occur because the internal stress generated in the film becomes enormous. Furthermore, in order to ensure device uniformity, the incident direction of flying particles during film formation must be strictly perpendicular to the substrate.

Therefore, a sputtering method in which the incident angle of flying particles has a certain width cannot be used, and a vapor deposition method in a high vacuum with a small amount of residual gas which is a scattering factor of flying particles is selected. However, the vacuum evaporation method has smaller kinetic energy of the flying particles than the sputtering method. Therefore, this is a deposition method in which the migration of flying particles on the substrate surface is small and stress is likely to occur. Thus, since the Spindt method is a special technique that imposes very difficult conditions for the deposition process, it is difficult to put the Spindt method into practical use.
In general, it is recognized that the deposition of a metal having a thickness of 1 μm or more is a difficult technique because of the problem of stress even in the sputtering method.

(Application of IIB technology)
A procedure for manufacturing an emitter using the IIB technique will be described below based on information disclosed in Non-Patent Documents 5 and 7. First, using a double patterning or the like on a molybdenum or tungsten thin film having a thickness of about 20 nm to 50 nm, a triangle having a base of 400 nm and a height of 1 μm is sharply two-dimensionally patterned. Next, the two-dimensional triangular patterned metal thin film is processed into a cantilever structure using an etching technique or the like. Thereafter, ion implantation is performed on the cantilever structure of the triangular metal thin film. When a sufficiently large number of ion implantations are performed, the cantilever structure can be bent completely vertically, and a three-dimensional structure having a high aspect ratio of 1 μm in height, about 20 nm to 50 nm in thickness, and 400 nm in the bottom can be formed.
The conventional method required the deposition of refractory metal with a thickness of 1μm, which caused problems of stress and delamination. However, using the IIB technology, a thin film with a thickness of about 20nm to 50nm was folded to form a three-dimensional structure with a height of 1μm. As a result, the problem of stress and peeling due to thick film deposition does not occur. In addition, since the process constituted by thin film deposition, etching, and ion implantation is highly similar to the silicon LSI manufacturing process, mass production using the equipment of the LSI manufacturing factory is possible.

[Application to silicon photonics]
(Conventional technology)
Next, application examples of IIB technology to silicon photonics will be shown. In recent years, research and development of silicon photonics, in which a silicon thin-wire optical waveguide using a silicon material such as single crystal silicon or amorphous silicon for a core portion of the optical waveguide is a main component, has been actively conducted. Since a large relative refractive index difference is obtained between the silicon core material and the silica-based cladding material, light is not lost even if the optical waveguide is bent with a small radius of curvature, and the optical circuit can be significantly reduced in size. Because.
In addition, since the manufacturing process of silicon LSI can be diverted, low manufacturing costs due to mass production are expected. Silicon photonics is expected as a technology for realizing optical interconnection, and element device performance has already achieved a practical level.
However, the realization of optical coupling technology for mounting external optical components such as optical fibers and light emitting / receiving elements on silicon optical circuits and the realization of wafer level tests essential for mass production have become major issues.

(Prior art issues)
Unlike electrical wiring, optical coupling technology for connecting optical wiring has difficulty due to the light propagation principle that light cannot be transmitted with high efficiency simply by connecting the structures. In the case of metal wiring for electronic devices, electricity flows basically when the metals are in contact with each other. Therefore, if an electrode pad is formed on the chip surface and a metal wire or metal bump is bonded thereto, the vertical direction of the chip An electrical signal can be transmitted to Therefore, it is possible to easily perform mounting with a high degree of integration in which many chips are stacked and a wafer level test in which a probe is applied to a wafer in the middle of a process to inspect electrical characteristics.

On the other hand, in optical wiring, a mechanism for efficiently converting the optical path of light in the vertical direction of the wafer is not easy in the first place. This is because, even when the orientation of the optical wiring is changed by 90 degrees in the plane, the direction change is not so easy that the propagation efficiency is deteriorated unless a gentle curve having a curvature radius of about 3 μm or more is formed. Moreover, due to the characteristics of the conventional microfabrication technology, it has been practically impossible to form an optical wiring that is gently curved in the vertical direction even by making full use of deposition technology and etching technology.

As a result, diffraction grating couplers that utilize the phenomenon that light changes direction with diffraction gratings are becoming the industry standard for silicon photonics surface optical couplers, and development is progressing toward mounting applications and wafer level test applications. It is out. However, since the light diffraction phenomenon has a strong wavelength dependency, incident angle dependency, and polarization dependency in principle, it is a problem that the device has a narrow allowable band of wavelength, angle, and polarization. If a silicon optical waveguide that is three-dimensionally curved in the vertical direction of the wafer can be formed, a surface type optical coupler with a dramatically wider wavelength, angle, and polarization tolerance band than a diffraction grating type coupler can be used. realizable.

(Application of IIB technology)
As shown in Non-Patent Document 6 and Patent Documents 3 to 4, the IIB technique is suitable for forming a silicon optical waveguide shape that is three-dimensionally curved in the vertical direction of such a wafer.
The manufacturing procedure of the surface light input / output device for silicon photonics is shown below. A general silicon photonics circuit has an optical wiring / circuit layer having a thickness of 220 nm formed of silicon on a silicon oxide film having a thickness of 2 μm or more, and a silicon oxide film having a thickness of about 2 μm is further formed thereon. A cladding layer is formed. A cantilever beam having a length of about 5 μm to 50 μm is formed at the terminal portion of such a silicon optical circuit. The method of forming the cantilever is performed by a process of removing the silicon oxide films above and below the silicon optical circuit layer. Thereafter, ions are implanted into the cantilever structure of the silicon optical wiring to form a three-dimensionally curved silicon optical waveguide having a curvature radius of about 3 μm to 30 μm.

Thereafter, the solid curved portion is again filled with the silicon oxide film to complete the device. The structure in which the optical waveguide is curved three-dimensionally can propagate light waves in the direction perpendicular to the substrate on the same principle as a curved optical waveguide formed on a flat surface. Therefore, it is possible to realize a vertical light input / output port and an interlayer optical coupler with low angle dependency and low loss.
Thus, the IIB technique has a great effect when forming a three-dimensional structure that was impossible with the conventional technique. The process of bending a silicon optical waveguide having a thickness of about 220 nm with a radius of curvature of about 3 μm cannot be realized by other bending techniques.

It can be read from the information disclosed in Non-Patent Document 6 that this is not the only merit that the three-dimensionally curved silicon optical waveguide has over the diffraction grating coupler. Since the three-dimensionally curved silicon optical waveguide has its unique three-dimensional shape protruding out of the wafer surface, it realizes a new functional form and contributes to high functionality and high integration of the optical device. Examples of new functional forms include a waveguide core structure formed with a taper in the vertical direction as shown below, a second core formed rotationally symmetrically with respect to the optical axis, and formed at the tip of the waveguide. There is a lens structure.

[Merit of silicon photonics application]
(Merit 1: Waveguide core structure with a taper formed in the vertical direction)
The three-dimensional curved silicon optical waveguide is formed by three-dimensionally bending a waveguide patterned in a two-dimensional plane. Therefore, it is possible to form a waveguide core having a forward taper or a reverse taper formed in the vertical direction, which is very difficult to manufacture by the conventional process. As a result, the spot shape of light propagating in the vertical direction can be freely controlled, and the beam size can be expanded, reduced, flattened, etc., and elements that realize high-efficiency optical coupling with optical fibers, light sources, and optical receivers Technology.

(Merit 2: Second core structure formed rotationally symmetric with respect to the optical axis)
When producing a spot size converter in a normal silicon optical waveguide, a second core responsible for spot size expansion is formed using a deposition technique. Therefore, the second core can be formed only above the waveguide layer. On the other hand, since the three-dimensionally curved silicon optical waveguide has a structure protruding out of the wafer surface, a second core member such as silicon oxynitride can be deposited isotropically by a CVD method or the like. Then, the structure forms a second core that is rotationally symmetric with respect to the optical axis. Such a structure can realize a highly efficient spot size converter.

(Merit 3: lens structure formed at the tip of the waveguide)
It is difficult to form a lens at the tip of a silicon wire waveguide formed on a normal plane. However, in the case of a three-dimensionally curved silicon optical waveguide protruding out of the wafer surface, lens formation can be easily performed by utilizing the CVD method. By selecting the diameter of the lens and the refractive index of the material, it becomes possible to enlarge / reduce the spot size of the light.

Any of the structures listed above can be realized because of the structure of a solid curved silicon optical waveguide. In other words, the three-dimensionally curved silicon optical waveguide is not limited to the function of merely a mechanism for converting the traveling direction of light into a direction perpendicular to the wafer.
The three-dimensional curved silicon optical waveguide also has a function of forming a base for forming an optical structure for freely controlling the light input / output mode toward the wafer surface.

Moreover, its application is not limited to a simple vertical optical coupler. If a large number of three-dimensional curved silicon optical waveguides that can freely control the mode are arranged, a fumode fiber, a multimode fiber, a multicore fiber, a fumode multicore fiber or a multimode multicore fiber, a laser diode array, a surface emitting laser array, a MEMS mirror, a space It is effective as a vertical light input / output port for optical coupling with an optical device in which a plurality of light propagation modes are spatially gathered, such as a phase modulator, and contributes to an increase in capacity of optical communication technology. In addition, the application field of an array in which a large number of three-dimensionally curved silicon optical waveguides that can freely control modes, such as a vehicle-mounted rider using interference light, is not limited to the communication field.
In other words, IIB technology for silicon photonics can be said to be a revolutionary technology that not only solves conventional problems but also pioneers application fields that were impossible in the past.

JP 2009-252689 A JP2011-140072 JP 2013-178333 A WO / 2014/156233

(Conventional IIB technology)
FIG. 5 schematically illustrates a conventional basic process for bending a thin film member.
As shown in FIG. 5 (A), a starting layer 20 ′ for forming a supporting layer to be a supporting layer for a thin film member is formed on a substrate 10, and a starting layer for forming a thin film member to be a thin film member in the future. 30 ′ is deposited. As shown in FIG. 5B, the thin film member forming starting layer 30 ′ is patterned into a desired planar shape using photolithography and etching techniques to form the thin film member 30.

After this patterning process, only the support layer forming starting layer 20 ′ immediately below the portion of the thin film member 30 to be bent is removed by photolithography and lateral etching, and as shown in FIG. The surface 21 is the support layer 20 at a position retracted inward in the lateral direction from the distal end free end 41 of the thin film member 30, and the patterned thin film member 30 on the support layer 20 is an object to be processed.
That is, the thin film member 30 does not have the support layer 20 below it, and has a cantilever structure 40 that reaches the free end 41 in a floating state. The cantilever structure 40 is substantially subject to bending. The cantilever structure 40 has a shape in which the root is located at a position where the peripheral surface 21 of the support layer 20 is located, and extends from there to the free end 41 of the tip.

Next, the cantilever structure is irradiated with ions to bend the cantilever structure.
As shown in FIG. 5D, the cantilever structure 40 is curved starting from the peripheral surface 21 of the support layer 20. When the ion irradiation is further continued, the cantilever structure 40 finally stands upright with respect to the substrate 10 as shown in FIG.

[Specific examples of issues]
(For electrical measurement probes)
In the case of application to an electrical measurement probe, a metal thin film having a thickness of about 20 nm to 200 nm is formed on a substrate on which an insulator layer (support layer 20) formed of SiO ₂ is formed. The thin film member 30 is formed by patterning. Next, the tip of the thin film member 30 formed from the metal thin film is processed into a cantilever structure 40 having a length of about 10 μm to 100 μm. In the process of forming the cantilever structure, the insulator layer (support layer 20) immediately below the metal material thin film member 30 must be removed by a method that does not damage the metal material thin film member 30.

Therefore, the characteristics of hydrofluoric acid, which dissolves SiO ₂ and does not dissolve metals such as tungsten, are used. When wet etching using hydrofluoric acid is performed, SiO ₂ is isotropically etched, so that the insulator layer (support layer 20) immediately below the thin film member 30 of the metal material can be removed. In addition, the same isotropic etching can be performed by dry etching using hydrogen fluoride gas.

However, such an isotropic lateral etching has a disadvantage in the manufacturing process in that the controllability of stopping the etching at a desired position is low. Normally, in the vertical etching, if a material that does not dissolve in hydrofluoric acid is selected as the material of the substrate, the etching stops when the SiO ₂ is etched and the substrate is exposed. That is, in the vertical etching, the substrate functions as an etch stopper layer. On the other hand, since there is no etch stopper layer for lateral etching, the stop of etching can only be controlled by stopping the supply of hydrofluoric acid.

The operation of stopping the supply of hydrofluoric acid is performed, for example, by a process of substituting with pure water. However, it is very difficult to perform this process strictly in the substrate surface every time. Therefore, there is a difference in the lateral etching distance within the substrate surface, and the difference in the lateral etching distance is a difference in the length of the cantilever structure. As a result, the finished height of the electrical measurement probe varies.
By using IIB technology for probe fabrication, an ultra-high density probe array can be realized. However, if the height variation of these probes is large, it is not useful from the viewpoint of the reliability of the measurement data.

(For vacuum electron source)
Similar challenges exist for vacuum electron source applications. Describing based on the information disclosed in Non-Patent Document 7, in the case of a vacuum electron source application, molybdenum or tungsten having a thickness of about 20 nm to 50 nm on an insulator layer (support layer 20) formed of SiO _2. A thin film member 30 is formed by forming a refractory metal thin film such as the above and patterning it.
Next, the tip of the metal material thin film member 30 is processed into a cantilever structure 40 having a length of about 1 μm. In the process of forming the cantilever structure, the insulator layer (support layer 20) immediately below the metal material thin film member 30 must be removed by a method that does not damage the metal material thin film member 30.

Therefore, also in this case, the characteristics of hydrofluoric acid, which dissolves SiO ₂ and does not dissolve metals such as molybdenum and tungsten, are used. When wet etching using hydrofluoric acid is performed, the etching of SiO ₂ proceeds in the lateral direction, so that the insulator layer (support layer 20) immediately below the thin film member 30 of the metal material can be removed. Of course, the same isotropic etching can be performed by dry etching using hydrogen fluoride gas.

Also in the case of vacuum electron source application, there is the same problem of variation as in the case of electrical measurement probe application. Since the etching in the lateral direction has low controllability to stop the etching at a desired position, the lateral etching distance has a difference in the substrate surface, and the difference becomes the difference in the length of the cantilever structure. As a result, the height of the emitter varies. In addition, the vacuum electron source is a device that is more sensitive to variations than the electrical measurement probe, and even if the variation in the tip position of the emitter is only 100 nm, the device characteristics vary several times.

Vacuum electron sources are devices that are often used as an array in which several to several hundreds of units are grouped, and hundreds to tens of thousands of these groups are arranged. For example, in the case of the above-described ultra-high resolution image sensor, one pixel is formed by 10 to 100, and 1920 × 1080 pixels are laid down in the case of the high vision standard. Since it is necessary in such an application that each emitter has a uniform characteristic, it is a fatal drawback that individual device characteristics have a difference of several times.

(For silicon photonics applications)
Similarly, variations in isotropic etching become a problem even in silicon photonics applications. In the case of silicon photonics application, the tip of an optical waveguide core structure (thin film member 30) made of silicon material having a thickness of about 220 nm and a width of about 400 nm embedded with a quartz (SiO ₂ ) cladding layer is about 5 μm to 50 μm in length. The cantilever structure 40 is formed by being exposed as much as possible. In order to realize this structure, the cladding (support layer 20) immediately below the silicon core structure must be removed by a method that does not damage the silicon core structure.

Therefore, the characteristics of hydrofluoric acid, which dissolves SiO ₂ and does not dissolve silicon, are used. When wet etching using hydrofluoric acid is performed, SiO ₂ is isotropically etched in the lateral direction, so that the clad layer (support) directly under the thin film member 30 that forms the optical waveguide core structure made of silicon material. Layer 20) can be removed. Of course, the same isotropic etching can be performed by dry etching using hydrogen fluoride gas.

Also in the case of silicon photonics application, variation in isotropic etching becomes a problem. The difference in isotropic lateral etching distance is a difference in cantilever structure, and the difference in cantilever structure is a three-dimensionally curved silicon optical waveguide formed by IIB processing (vertical optical coupler or vertical light input). This affects the in-plane variation and height variation of the tip position of the output port.

In general, in optical coupling between a silicon photonics device and other optical devices such as an optical fiber and a light emitting / receiving element, precise alignment of the order of 100 nm or less is required.
Therefore, if the tip position of the three-dimensionally curved silicon optical waveguide varies, it is necessary to align each of the three-dimensionally curved silicon optical waveguides, and it takes time to align whether it is a mounting application or a wafer test application. Cost.

Furthermore, a large number of solid curved silicon optical waveguides formed by IIB processing are arrayed to form optical fibers such as fumode fibers, multimode fibers, multicore fibers, fumode multicore fibers, multimode multicore fibers, and laser diode arrays. And optical devices in which a plurality of light propagation modes are integrated spatially, such as array type light emitting devices such as surface emitting laser arrays, and spatially divided type light control devices such as MEMS mirrors and spatial phase modulators. When used as a vertical light input / output port for optical coupling, if there is in-plane variation and height variation at the tip position of each vertical light input / output port, the optical coupling loss will be different for each, resulting in completely practical use. Don't be.

As shown in the above example, isotropic etching is indispensable for IIB technology to form a cantilever structure. However, it is difficult in principle for isotropic etching to precisely control the etching stop position. It can be seen that the difference in the isotropic etching distance caused by the variation in the etching stop position becomes the difference in the length of the cantilever structure, which causes the variation in shape such as the height and position of the three-dimensional structure after IIB processing. Such a problem of variation is not limited to the applications listed here, such as probes, vacuum electron sources, and silicon photonics.

[Problem 1: Position controllability]
A general description of the problem is given according to FIG. FIG. 6 illustrates the problems regarding the position controllability of the curved structure with reference to the cross-sectional views before (A) and after (I) the ion irradiation with respect to (C) and (D) in FIG. It is a figure for doing.

The variation in the etching stop position (“peripheral surface 21” shown in FIG. 6A) as described above affects the variation in the position of the support point for the bending process in the IIB process.
The IIB processing is a deformation processing in which the free end side of the cantilever structure is greatly displaced and the fixed end is not displaced. Therefore, even if the ion implantation conditions are precisely controlled and the displacement of the beam portion from the fixed end to the free end is precisely controlled, the fixed end position that is affected by the variation in the stationary position of the isotropic etching eventually becomes three-dimensional. This is to define the position of the curved free end.

Another factor is involved in the positional variation of the IIB processed structure caused by isotropic etching. When the state of isotropic etching is viewed from the wafer surface (see FIG. 7A), the support layer just below the position where the cantilever structure starts to bend is eroded by the etchant from both sides and looks like a cape. It is removed while leaving the structure.

When etching is stopped with this cape-shaped tip having a sharp shape on the order of nanometers, and then subjected to IIB processing, the cantilever structure is not supported in a plane but in a point support Therefore, the deformation process becomes extremely unstable. The apex of this cape shape is ideally located exactly in the center of the cantilever structure, and there is no effect if the ion implantation angle can be kept strictly perpendicular to the wafer, but either one is slightly If it is shifted, the bending process of the cantilever beam structure is twisted and the desired shape cannot be obtained. If the three-dimensional structure is different from the desired structure, the three-dimensional cantilever structure varies as a result.

In order to suppress this phenomenon, the erosion distance from both sides should be made sufficiently long to eliminate the cape-like shape, or the masking design for isotropic etching should be designed so that the cape-like shape is not formed in advance. There is a way to devise. The masking technique is a method in which the etchant erodes in parallel from the position where it becomes the free end of the cantilever structure so that the etchant does not erode from both sides of the cantilever structure.

Figure 8 shows an example of masking techniques. If the masking window region is designed near the tip of the cantilever structure, a cape-like structure will not be formed if lateral etching is performed by isotropic etching. However, at the cost of this, the etching time becomes longer, resulting in a larger etching area and a larger device footprint.

For example, in the case of the above-mentioned silicon photonics application example, in order to obtain a cantilever structure having a length of 5 μm, an isotropic etching having a length of 5 μm is required in the lateral direction, and therefore occupies an area of about 25 μm ^2. This means that a much larger footprint is required than the object to be bent, and as a result, the placement of IIB workpieces at a narrow pitch is limited.

In this way, isotropic etching proceeds not only in the depth direction but also in the plane direction. Therefore, when arranging cantilever structures in a line in a certain direction, it is possible to increase the degree of integration with a design in which adjacent positions are close to each other and the fixed end positions are aligned. Thus, when a cantilever structure is arranged on a two-dimensional plane, each plane etching region interferes to greatly limit the fixed end position.

[Issue 2: Design restrictions]
Here, a specific example of the limitation on the integration density in the design will be described with reference to FIG. FIG. 9A is a top view of a state where the cantilever structure 40 is formed on the thin film member 30. A design that avoids the cape-like structure shown in FIG. 8 is adopted. The length of the desired three-dimensional structure is L, and a state where a cantilever beam structure having a length L is formed on the thin film member 30 is shown. A masking window region indicated by a broken line indicates a window portion of a masking layer for isotropic etching for forming a cantilever structure. The isotropic etching proceeds planarly starting from this masking window region.

In the case of this example, in order to form a cantilever beam having a length L, etching of a length L proceeds in a plane starting from the masking window region. The stop line of the etching region viewed in plan is drawn as the peripheral surface 21.
In FIG. 9, the cantilever structure 40 is drawn in a rectangle, but may of course be a triangle, a circle, or another two-dimensional figure. However, for the sake of simplicity, the description will be continued with a rectangular cantilever structure.

The length of the thin film member 30 other than the cantilever structure 40 is indicated by M. This region becomes a necessary part later as a region for supporting the produced three-dimensional structure. Of course, the shape of M differs depending on the application device. However, it is drawn here as a rectangle for simplicity of discussion.
However, in the region indicated by M, not only the length but also the width is an important parameter, but the width is omitted for the sake of simplicity of discussion. In addition, although it depends on the application device, the region indicated by the length M may be considered including the region necessary for wiring. In any case, in order to form the cantilever structure 40, it must be considered that the supporting portion occupies a certain area.

Next, the integration density in the case where the same three-dimensional structure is arrayed by arranging a plurality of the structures in FIG. 9A on the same substrate will be considered. In order to arrange the same three-dimensional structure, it is important to design an arrangement method including a planar etching region so that the length L of the cantilever is the same.

First, consider a one-dimensional array. When the cantilever beam structure is arranged one-dimensionally, even if the interval g is smaller than L, interference in the planar etching region may not be a problem. That is a method of arranging the positions where the cantilever structure starts to be bent in the case shown in FIG. 9B. In this case, the peripheral edge of the region where the respective etching regions are merged becomes the peripheral surface 21 formed by an actual etching process. Since the positions where the bending starts are aligned, the peripheral surface 21 is aligned at a position of length L from the tip with respect to all the cantilevers.

That is, in the method of arranging the positions where the bends start to be aligned, interference in the planar etching region does not matter. Therefore, it is possible to arrange them at extremely small intervals. In this case, the interval g can be reduced to the resolution limit of photolithography. The arrangement of the vacuum electron source shown in Non-Patent Document 7 and the arrangement of the silicon wire waveguides shown in Non-Patent Document 6 are examples of FIG. 9B.

Further, when the above arrangement method is made circular, a concentric arrangement as shown in FIG. 9C becomes possible. Also in the case of concentric arrangement, the periphery of the region where the etching regions are merged is aligned at a position of length L from the tip of the cantilever structure 40.
Such an arrangement is useful for generating a light wave having a helical equiphase surface in a multiplexing method using optical vortices (or OAM (Orbital Angular Momentum)) in silicon photonics applications, for example. . This is because the light emitting ends formed by the three-dimensional curved optical waveguides are arranged concentrically, and light waves whose phases are controlled can be emitted from each to the space.

However, if the cantilever structure is arranged without aligning the positions where it starts to bend, the arrangement interval is greatly restricted. For example, as shown in FIG. 9D, there is a request to make the interval g between adjacent cantilever structures smaller than the cantilever length L, and a request to arrange the tip positions while shifting the position in the X-axis direction. Are incompatible. In this case, since the peripheral surface 21 (thick line) combined with the planar etching region substantially forms the peripheral surface of isotropic etching, the position where the three cantilever structures shown in FIG. (Bold line) position.

Then, the peripheral surface 21 is formed in the position of the length K from the front-end | tip in the 2nd cantilever structure from the top. Of course, K is larger than L. Similarly, the third cantilever structure from the top has a length J (J is larger than K). That is, when there is a request to make the gap g between adjacent cantilever structures smaller than the cantilever length L and a request to arrange the tip positions while shifting in the X-axis direction, It is impossible to make the length of the structure the same L.

In order to avoid this problem and to arrange one-dimensionally without aligning the positions where the cantilever beam starts to bend, as shown in FIG. 9E, the interval g between adjacent cantilever beam structures is set to the cantilever length. It must be larger than L to avoid interference in the planar etching region. However, the three-dimensional structure cannot obtain a desired integration density.

拡張 Consider such design constraints by extending them to a two-dimensional array. First, FIG. 10A shows a method of arranging a plurality of one-dimensional arrays in FIG. 9B in the X-axis time direction. In this case, the gap g in the X direction is g> 2L + M. That is, a distance of a distance obtained by adding the length M of the support region to twice the length of the desired cantilever structure is necessary.

However, such a two-dimensional array design shown in FIG. 10A can easily double the device integration density depending on the device. As shown in FIG. 10B, two cantilever structures are formed in the interval g where g> 2L + M when the cantilever beam structures are alternately arranged in the right direction and the left direction. can do. If such an arrangement is used, the effective interval g is g> (2L + M) / 2. However, even if such a device is applied, it is impossible to design g to be less than L. In addition, it is a new design issue that the direction of the cantilever structure is staggered.

In the square lattice arrangement shown in FIG. 10A, if the cantilever structure is arranged obliquely with respect to the lattice, the interval in the X-axis direction can be shortened. For example, by tilting 30 degrees, as shown in FIG.

Can be doubled. However, the interval in the Y-axis direction is increased to an interval g> (2L + M) / 2. Similarly, as shown in FIG. 10D, if the cantilever structure is arranged at an angle of 45 degrees with respect to the lattice, the interval in the X-axis direction can be further shortened, but at the same time, the interval in the Y-axis direction is increased. In this case, both X and Y directions

Is required.

In these cases, as shown in FIGS. 10A and 10B, the cantilever structure can be easily arranged by alternately arranging the right and left directions of the cantilever structure. The integration density can be doubled. In this case, the effective interval g is the X-axis direction of FIG.

Y axis direction is g> (2L + M) / 4, and both X and Y axis directions in FIG.

Can be shortened.
However, it is a new design issue that the direction of the cantilever structure is staggered.

Similarly, FIG. 10E illustrates the case where a three-dimensional structure is arranged in a triangular lattice shape. In this case, the gap g of the cantilever structure corresponds to the length of one side of the triangle connecting the lattice points, and the condition is approximately g> 2L. Also in this case, as shown in FIG. 10A and FIG. 10B, the cantilever structure can be easily integrated by alternately arranging the cantilever structures facing right and left. The density can be increased by a factor of 2, and can be effectively reduced to g> L.
However, it is a new design issue that the direction of the cantilever structure is staggered.

(For electrical measurement probe arrays)
A specific example of producing an electrical measurement probe having a height of 50 μm using the IIB technique will be described. In this case, the length L of the cantilever structure is required to be 50 μm. Further, the support region M is assumed to be about 10 μm including wiring. If this is arranged one-dimensionally as shown in FIG. 9B, the interval g can be reduced regardless of the length of L. For example, g can have a pinching pitch of 5 μm or less. It is possible to form a probe array.

On the other hand, considering a two-dimensional array, in the case of FIG. 10C, the interval in the X direction is approximately 95 μm and the interval in the Y direction is approximately 55 μm. Similarly, in the case of FIG. 10D, the interval in the X direction and the Y direction is approximately 78 μm. Similarly, when the array is formed in the triangular lattice shape of FIG. 10E, an interval of 100 μm or more is obtained. These values are equivalent or slightly smaller than existing electrical measurement probe arrays.
As shown in the relationship between FIGS. 10A and 10B, the cantilever structure can be easily arranged in the right direction on the paper and the left direction on the paper alternately to double the integration density of the device. Can be improved.
However, in that case, it becomes a new problem that wiring is complicated.

That is, when a probe array for electrical measurement is manufactured using IIB technology, in addition to the advantage that a probe array can be formed using a material having high wear resistance such as tungsten as described above, There is an advantage that a high-density arrangement becomes possible when arrayed in comparison. However, since the advantage of high density that can be achieved by the IIB technology is small, there is no reason to actively adopt the new technology IIB method. If there is no design restriction as described above, the integration density can be improved dramatically compared to the conventional technology, so a probe array that can dramatically improve the efficiency of various inspections such as wafer tests will be developed. be able to.

(For vacuum electron source)
A specific example of manufacturing a vacuum electron source having a height of 1 μm by using IIB technology will be described. In this case, the length L of the cantilever structure is assumed to be 1 μm, and the length M considering the support area and the wiring routing is assumed to be 2 μm. When these are two-dimensionally arranged, in the case of FIG. 10C, the interval in the X direction is approximately 3.5 μm, and the interval in the Y direction is approximately 2.0 μm. Similarly, in the case of FIG. 10D, the interval between the X direction and the Y direction is approximately 2.8 μm. Similarly, in the case of arraying in the triangular lattice shape of FIG. 10E, the interval is approximately 2.0 μm or more.

As shown in the relationship between FIGS. 10A and 10B, the cantilever structure can be easily arranged in the right direction on the paper and the left direction on the paper alternately to double the integration density of the device. Can be improved. Moreover, in the case of a vacuum electron source application, since an array arrangement in which each cantilever structure is electrically connected in units of 10 to 100 is generally used, there is no need to worry about wiring. In fact, Non-Patent Document 7 discloses such a vacuum electron source array.

However, spatial uniformity is important for vacuum electron source arrays applied to imaging. That is, it is desirable that the distance between the X direction and the Y direction is the same. In order to satisfy the condition, the method of arraying in a triangular lattice shape in FIG. 10E is optimal. However, an interval of at least twice the height is required for an emitter having a height of 1 μm, and the emitter array becomes a device with a low integration density.
That is, when the vacuum electron source array is manufactured using the IIB technology, it becomes possible to manufacture an emitter of a refractory metal material by a process according to the LSI manufacturing process, but there is a problem in terms of integration density. I can say that.

(Silicon photonics)
A specific example of manufacturing a vertical photocoupler for silicon photonics having a curvature radius of 3 μm using the IIB technology will be described. In this case, the length L of the cantilever structure is required to be 5 μm. It is assumed that the length M as an area necessary for routing of the support portion and the wiring is 5 μm.
When these are two-dimensionally arranged, in the case of FIG. 10C, the interval in the X direction is approximately 13 μm and the interval in the Y direction is approximately 7.5 μm. Similarly, in the case of FIG. 10D, the distance between the X direction and the Y direction is approximately 10.6 μm. Similarly, when the array is formed in the triangular lattice shape of FIG. 10E, the interval is approximately 10 μm or more. These values are comparable to those of the conventional diffraction grating type coupler, and can be said to have sufficient integration density as a replacement for the prior art.

However, considering the development of silicon photonics, it is desirable to achieve higher density. In the field of optical fibers, development of optical fibers specialized for spatial multiplexing such as fumode fibers, multimode fibers, multicore fibers, fumode multicore fibers, and multimode multicore fibers has already been advanced. In the field of light emitting devices, laser diode and surface emitting laser array technology has been established.
Research and development of array type light control devices such as MEMS modulators and spatial modulators using liquid crystals are also active. In other words, it can be said that various optical devices are being developed in the direction of increasing the integration density in a two-dimensional space.

Among them, although silicon photonics has the feature that the integration density of the optical circuit is very high, the footprint of the optical input / output port for exchanging optical signals with the outside is 10 μm square in the conventional diffraction grating type coupler. Because of the large size, the optical input / output port is a bottleneck in improving the integration density. For example, in the case of a fu mode fiber, there are 3 to 4 modes in a core having a diameter of about 17 μm. Therefore, the footprint of the optical input / output port corresponding to the fu mode fiber is too large at 10 μm square.

In addition, when a solid curved optical waveguide is used as a light output port in the vertical direction, if the interval between the emission positions can be reduced to about half of the wavelength of light, excellent characteristics can be obtained as a device utilizing the light interference effect. . For example, assuming a 1.55 μm band used in optical communication, if the pitch of the vertical light output port can be made between 1.55 μm and 0.775 μm, a rider (LIDAR: Light Detection and Ranging, or Laser) Imaging Detection and Ranging) and other applications.

The problems of position controllability and integration shown so far apply not only to probe arrays, vacuum electron sources, and silicon photonics applications, but also to other applications as specific examples of thin film members. That is, each of the above-mentioned problems is a problem common to the bending process of the thin film member by the IIB technology regardless of the material system or the purpose.

[Problems to be solved by the invention]
It has been found that all of the above problems are caused by the fact that the stop position of isotropic etching or the like is the same as the start position of the IIB processing bend.
Therefore, in order to overcome all of these problems, it is an object of the present invention to realize a thin film member bending method in which an etching stop position such as isotropic etching is different from an IIB processing bending start position. .

Means for solving the above problems are as follows.
(1) A step of laminating a starting layer for forming a support layer and a starting layer for forming a thin film member on a substrate, a step of patterning the starting layer for forming a thin film member to form a thin film member, Forming a position control layer for determining a position at which the cantilever structure of the thin film member starts to bend by irradiating ions directly or via a starting layer for forming the position control layer support layer; A part of the starting layer for forming the supporting layer, a part of the starting layer for forming the supporting layer and a part of the starting layer for forming the position control layer, the peripheral surface of the supporting layer, or the peripheral surface of the supporting layer And a step of removing the end face of the position control layer support layer so as to be on or inside the lower extension line of the tip of the position control layer to form a cantilever structure having a tip free end on the thin film member; The cantilever beam using the position control layer as a mask Irradiated with ions in a part of the concrete, the bending processing method of a thin film member including a step of bending a portion of the cantilever structure.
(2) A step of laminating a starting layer for forming a support layer and a starting layer for forming a thin film member on a substrate, a step of forming a thin film member by patterning the starting layer for forming a thin film member, and on the thin film member Forming a position control layer by anisotropic etching, which determines a position at which the cantilever structure of the thin film member begins to bend when irradiated with ions, directly or through a starting layer for forming a position control layer support layer; A part of the starting layer for forming the support layer under the thin film member, or a part of the starting layer for forming the supporting layer and a part of the starting layer for forming the position control layer supporting layer, the peripheral surface of the supporting layer, or The peripheral surface of the support layer and the end surface of the position control layer support layer are removed by isotropic etching so that the end surface of the position control layer is on or inside the lower extension line of the tip of the position control layer, and the free end of the tip is formed on the thin film member. Form a cantilever structure with Degree and, partially irradiated with ions of the cantilever structure the position control layer as a mask, curving a thin film member including a step of bending a portion of the cantilever structure.
(3) The method for bending a thin film member according to (1) or (2), wherein the cantilever structure constitutes an electrical measurement probe.
(4) The method for bending a thin film member according to (1) or (2), wherein the cantilever structure constitutes an emitter of a field emission device.
(5) The method for bending a thin film member according to (1) or (2), wherein the cantilever structure constitutes an optical waveguide.
(6) A step of laminating a starting layer for forming a support layer and a starting layer for forming an optical waveguide on a substrate, a step of patterning the starting layer for forming an optical waveguide to form an optical waveguide, and on the optical waveguide Forming a position control layer by anisotropic etching to determine a position at which the cantilever structure of the optical waveguide starts to bend by irradiating ions directly or via the position control layer support layer; and A part of the starting layer for forming the supporting layer below, a part of the starting layer for forming the supporting layer and a part of the starting layer for forming the position control layer supporting layer, the peripheral surface of the supporting layer, or the peripheral edge of the supporting layer A cantilever having a free end on the optical waveguide by removing the surface and the end face of the position control layer support layer by isotropic etching so that the end face of the position control layer is on or below the extension line of the tip of the position control layer. Forming the structure; and Part irradiated with ions of the cantilever structure of the control layer as a mask, curved method of processing an optical waveguide including a step, a to the vertical optical coupler by bending a portion of the cantilever structure.
(7) The method for bending an optical waveguide according to (5), wherein the substrate is a silicon optical integrated circuit substrate.
(8) The method for bending an optical waveguide according to (6) or (7), wherein the optical waveguide is an optical waveguide mainly composed of silicon.
(9) The method for bending an optical waveguide according to (8), wherein the constituent material of the position control layer is tungsten or carbon.
(10) In any one of the above (6) to (9), the cantilever structure of the optical waveguide is arranged in a two-dimensional array at an interval smaller than the length of the cantilever structure. A method of bending an optical waveguide as described.
(11) The method for bending an optical waveguide according to (10), wherein the free end of the cantilever structure is disposed at each vertex of a square lattice.
(12) The method of bending an optical waveguide according to (10), wherein the free end of the tip of the cantilever structure is disposed at each vertex of a triangular lattice.
(13) Any one of the above (6) to (9), wherein the cantilever structure of the optical waveguide is two-dimensionally arranged at an interval smaller than the wavelength of the propagation light of the optical waveguide. A method for bending an optical waveguide according to claim 1.
(14) The step of bending a portion of the cantilever structure to form a vertical optical coupler is a step of bending a portion of the cantilever structure upward to form a vertical optical coupler. The method for bending an optical waveguide according to (6) above.
(15) Further, a step of filling the vertical optical coupler curved upward with a clad layer, and a step of laminating a second support layer forming start layer and a second optical waveguide forming start layer on the clad layer Patterning the second optical waveguide forming starting layer to form a second optical waveguide, and directly on the second optical waveguide or via the second position control layer support layer, Forming a second position control layer by anisotropic etching to determine a position at which the second cantilever structure of the second optical waveguide starts to bend when irradiated with ions; and the second optical waveguide. A part of the starting layer for forming the second support layer below or a part of the starting layer for forming the second supporting layer and a part of the starting layer for forming the second position control layer, The peripheral surface of the second support layer, or the peripheral surface of the second support layer and the second position control layer Removing the end face of the position control layer by isotropic etching so that the end face is on or inside the lower extension line of the tip of the position control layer, and forming a cantilever structure having a tip free end in the optical waveguide; Using the second position control layer as a mask, a part of the second cantilever structure is irradiated with ions, and a part of the second cantilever structure is bent downward to form the vertical optical coupler. The method for bending an optical waveguide according to (14), including a step of forming a second vertical optical coupler for optical connection.

According to the present invention, in the bending process of the thin film member by the IIB technique, the cantilever structure of the thin film member is irradiated with ions directly on the thin film member or via the starting layer for forming the position control layer support layer. Since the process of forming a position control layer that determines the position at which to start bending is adopted, position control is performed compared to the case where the cantilever structure is bent from the peripheral surface of the support layer that has been determined by, for example, lateral etching as in the past. Can dramatically improve performance.
Furthermore, since a plurality of cantilever structures can be arranged close to each other, high integration of the processed cantilever structures can also be achieved.

It is drawing which shows typically the bending method of the thin film member which concerns on this invention. It is the SEM image (A) after ion irradiation, and the SEM image (B) after tungsten removal regarding the bending method of the thin film member which concerns on this invention. It is drawing which shows typically the solution of the problem of the bending process of the thin film member which concerns on this invention. It is drawing which shows typically the bending method of the thin film member based on this invention applied to formation of a three-dimensional curved optical circuit. It is drawing which shows typically the bending method of the conventional thin film member. It is drawing explaining the problem of the bending process of the conventional thin film member. It is drawing explaining the problem of the bending process of the conventional thin film member, (A) is a top view, (B) is sectional drawing. It is drawing explaining the problem of the bending process of the conventional thin film member, (A) is a top view, (B) is sectional drawing. It is drawing explaining the problem of the bending process of the conventional thin film member. It is drawing explaining the problem of the bending process of the conventional thin film member.

(Key points of the present invention)
The main point of the present invention is that the irradiated ions are newly shielded in the direction in which the irradiated ions fly as viewed from the cantilever structure, and IIB processing is performed at a position different from the end point position of the lateral etching such as isotropic etching. This is to add a position control layer that can define the position at which the bending starts.
The conditions required for the position control layer are: A: processed by anisotropic etching with higher controllability than isotropic etching, B: whether the tip is in the same plane as the lateral etching stop position, or It is desirable to satisfy the following four conditions: the position extended toward the tip of the cantilever, C: irradiation ions do not penetrate, and D: IIB is not deformed by itself.

However, the position control layer does not have to satisfy all of the above conditions, and even if there are some ions penetrating through C, the energy may be substantially reduced if the energy is greatly reduced. If it is slight, even if it is deformed, there is substantially no influence. However, conditions A and B must be strictly met, and are essential to the present invention.

(Embodiment of the present invention)
1A to 1H are drawings schematically showing a method of bending a thin film member according to the present invention. Hereinafter, the steps will be sequentially described.

(1) A support layer forming starting layer 20 ′ is formed on the substrate 10, and a thin film member forming starting layer 30 ′ is formed thereon. (Fig. 1 (A))
In FIG. 1A, the support layer forming starting layer 20 ′ is drawn as necessary over the entire surface of the substrate, but only the region forming the cantilever 40 is necessary. In other regions, the support layer forming start layer 20 ′ may not be formed under the thin film member forming start layer 30 ′. The support layer forming starting layer 20 ′ is a layer necessary for forming a part of the thin film member forming starting layer 30 ′ into the cantilever structure 40.

(2) The thin film member forming starting layer 30 ′ is patterned to form the thin film member 30. (Fig. 1 (B))
The planar shape of the patterned thin film member 30 depends on the final production target structure. For example, when the final fabrication target structure is an emitter of a field emission device, a planar pattern having a sharp triangular tip is formed, and the final fabrication target structure is a three-dimensionally curved silicon wire waveguide. In some cases, the width is an elongated structure with a submicron order. In addition to these examples, various applications are conceivable. In either case, a generalized cross-sectional view is as shown in FIG.

(3) The patterned thin film member 30 is embedded in the position control layer support layer forming starting layer 60 ′. (Figure 1 (C))
In FIG. 1C, the upper portion of the position control layer support layer forming starting layer 60 ′ is drawn flat, but irregularities reflecting the shape of the thin film member 30 may be generated.
If the processing method for forming the position control layer 50 is not a processing method that substantially leaves a trace on the thin film member 30 as shown in FIG. 1E, the position control layer support layer The formation of the forming starting layer 60 ′ can be omitted.

(4) A position control layer forming starting layer 50 ′ is formed. (Figure 1 (D))
(5) The position control layer 50 is formed by processing the position control layer forming starting layer 50 ′. (Figure 1 (E))
The edge of the tip 51 of the position control layer 50 plays a role of determining a position where the cantilever structure starts to bend when ion irradiation is performed thereafter. For this purpose, the tip 51 needs to be in a position retracted inward in the lateral direction from the tip free end 41 of the cantilever structure 40.
The planar shape of the position control layer 50 depends on the planar shape of the patterned thin film member 30 and the final structure to be created. In either case, a generalized cross-sectional view is as shown in FIG. .

(6) The cantilever structure 40 is formed. (Fig. 1 (F))
The isotropic etching is performed to remove the starting layer 20 ′ for supporting layer formation immediately below the portion to be bent of the thin film member 30 and the starting layer 60 ′ for forming position control layer supporting layer immediately above the portion to be bent, The peripheral surface 21 is the support layer 20 in a position retracted inward in the lateral direction from the free end 41 of the thin film member 30, and the thin film member 30 thereon is a thin film member to be processed.

In FIG. 1F, the position control layer 50 is also drawn so as to have a cantilever structure like the thin film member 30, but the end surface 61 of the position control layer support layer 60 is the tip of the position control layer 50. Even if it is in the same position as 51, the function of the position control layer 50 is not hindered.
Moreover, the peripheral surface 21 of the support layer 20 needs to be in the position extended on the downward extension line of the front-end | tip 51 of the position control layer 50, or inward in the horizontal direction.

The material of the starting layer 60 ′ for forming the position control layer support layer can be selected within the range of materials that can be processed into a shape as shown in FIG. In other words, the position control layer support layer forming starting layer 60 ′ is removed, but the position control layer 50 and the thin film member 30 can be selected within a range in which processing is not affected.
From the viewpoint of facilitating the production by reducing the number of steps, the position control layer support layer forming starting layer 60 ′ is made of the same type or the same material as the support layer forming starting layer 20 ′. desirable. Then, the etching for forming the structure shown in FIGS. 1E to 1F can be integrated into one process.

For example, when both the position control layer support layer forming start layer 60 ′ and the support layer forming start layer 20 ′ are made of silicon dioxide (quartz), they can be removed with hydrofluoric acid, and the thin film member 30 can be made of silicon, molybdenum, tungsten, or the like. , Tantalum, technetium, rhenium, cobalt, nickel, ruthenium, osmium, rhodium, iridium, palladium, platinum, copper, silver, gold, lead, germanium and other metals, tungsten silicide, molybdenum silicide, tantalum silicide, titanium silicide, cobalt Silicon compounds such as silicide, chromium silicide, nickel silicide, and carbon such as carbon, silicon carbide, tantalum carbide, titanium carbide, molybdenum carbide, niobium carbide, hafnium carbide, tungsten carbide, vanadium carbide, diamond-like carbon Materials, nitrides such as titanium nitride, aluminum nitride, silicon nitride, tantalum nitride, boron nitride, chromium nitride and zirconium nitride, and transparent conductive films such as indium tin oxide, zinc aluminum oxide, zinc oxide and indium gallium zinc oxide And the position control material 50 is a metal such as silicon, molybdenum, tungsten, tantalum, technetium, rhenium, cobalt, nickel, ruthenium, osmium, rhodium, iridium, palladium, platinum, copper, silver, gold, lead, germanium, or the like Alloys such as tungsten silicide, molybdenum silicide, tantalum silicide, titanium silicide, cobalt silicide, chromium silicide, nickel silicide and other silicon compounds, carbon, silicon carbide, tantalum carbide, titanium carbide, carbonized Carbon-based materials such as ribden, niobium carbide, hafnium carbide, tungsten carbide, vanadium carbide, diamond-like carbon, photoresist, polyimide, titanium nitride, aluminum nitride, silicon nitride, tantalum nitride, boron nitride, chromium nitride, zirconium nitride, etc. The cantilever structure 40 can be formed with almost no erosion of a transparent conductive film such as a nitride of the above, or indium tin oxide, zinc aluminum oxide, zinc oxide, or indium gallium zinc oxide.

(7) The cantilever structure 40 is bent by ion irradiation. (Fig. 1 (G))
The position where the cantilever structure starts to bend is a position where the cantilever structure 40 intersects with the cantilever structure 40 extending in the direction of the downward extension of the tip 51 of the position control layer 50. However, although ion irradiation is performed from a direction perpendicular to the substrate, the tip 51 of the position control layer 50 becomes a wrinkle and position control is possible even when the ion incident angle is not vertical.

The position control layer 50 needs to have a role to shield the irradiation ions from penetrating. In other words, the ions are shielded when the projected range of ions is smaller than the film thickness.
In other words, a material with a small ion projection range may be selected. Approximately, a higher density material has a smaller ion projection range and higher shielding performance. For example, the shielding performance increases in the order of carbon, silicon, chromium, molybdenum, and tungsten, which is desirable as the position control layer 50.

It is desirable that the position control layer 50 is not deformed by ion irradiation. For this reason, the material and film thickness which are hard to deform | transform are selected. In IIB processing, what are the materials and film thickness that are difficult to deform? In order to show that, first, the ease of IIB processing is summarized.
In IIB processing, the kinetic energy of irradiated ions is a parameter that determines whether the cantilever structure is bent upward or downward with respect to the substrate. Ions penetrate deeper into the cantilever structure as the acceleration energy increases. Collisions with many atoms occur from when ions enter the cantilever structure until they lose kinetic energy and stop in the film. In the process, recoil atoms are generated, and as a result of the recoil atoms being generated, lattice defects such as vacancies and interstitial atoms are generated.

Summarizing the information disclosed in Patent Documents 1 to 4 and Non-Patent Documents 5 to 7, the optimum energy for bending the cantilever structure upward with respect to the substrate in IIB processing is the cantilever structure. It is shown that the peak position of the recoil atom distribution is approximately 30% from the surface with respect to the depth direction of the film to be formed. In addition, when the depth of the peak position of the recoil atom distribution is 10% to 20%, the curve bends upward, but a large amount of irradiation is required because the amount of bending per unit dose is small. % To 30% is not optimal, but a sufficiently practical upward bending process can be carried out. From 30% to 40%, the irradiation becomes too deep and a downward bending force is produced. It is shown that when the amount of bending in the direction is reduced, and when it exceeds 40% to 60%, the downward bending force is superior to the upward bending force.

Replacing these phenomena with the average projected range of ions instead of the recoil atom density, it becomes as follows. First, the best condition for bending upward is that the average projected range is 50% deep with respect to the cantilever structure. If the average projection range is between 20% and 30%, it will bend upward, but since the amount of curvature per unit dose is small, it will be a condition that requires a large dose, and the average projection range will be between 30% and 50%. Although this condition is not optimal, it is a condition that allows a sufficiently practical upward bending process. On the other hand, as the average projection range exceeds 50% and becomes as large as 60% and 70%, it becomes a condition that the downward bending force becomes dominant.

From the above, one of the conditions that are difficult to deform in IIB processing is that the ion irradiation condition where the peak position of recoil atom density is less than 10% deep from the surface relative to the cantilever structure, in other words, the cantilever structure. On the other hand, it can be said that this is an ion irradiation condition in which the average projected range is less than 20% from the surface. The other is that the peak position of recoil atom density is between 40% and 60% deep from the surface with respect to the cantilever structure, and the upward bending force and the downward bending force are equal. Ion irradiation conditions in which the average projected range is between 40% and 60% deep from the surface with respect to the cantilever structure, and the upward bending force and the downward bending force are equal. It can be said that it is a condition.

(Specific examples of silicon photonics applications)
Results of calculation using SRIM code (http://www.srim.org/) about the film thickness of the position control material 50 when bending a 220 nm-thick silicon wire optical waveguide with various ions in silicon photonics applications Indicates.
First, when the optimum kinetic energy of irradiated ions is calculated when the average projected range is 50% of the film thickness, that is, when the depth is 110 nm, for example, in the order of the atomic number, in the case of silicon ions, about 80 keV, About 85 keV for phosphorus ions, about 110 keV for argon ions, about 170 keV for arsenic ions, about 200 keV for krypton ions, about 250 keV for indium ions, about 250 keV for antimony ions, and about 270 keV for xenon ions. .

Next, as a condition of the position control layer 50, when the film thickness at which the average projected range in the calculated kinetic energy of ion implantation is 15% or less of the film thickness is calculated with several materials, for example, density From about 1300 nm to about 1450 nm for photoresist, from about 1030 nm to about 1120 nm for epoxy resin, from about 570 nm to about 670 nm for carbon, from about 740 nm to about 780 nm for silicon, and from about 310 nm to about 330 nm for niobium. Calculations show that the minimum required film thickness is about 270 nm to about 290 nm for molybdenum and about 190 nm to about 210 nm for tungsten.
Of course, in the case where IIB processing is performed under the condition that the irradiation energy may be increased by setting the kinetic energy of the irradiation ions smaller than the optimum value, the position control layer 50 is selected according to the selected kinetic energy. The minimum required film thickness can be made smaller than the above calculation result.

(Specific examples of vacuum electron source applications)
Similarly, the thickness of the position control material 50 in the case of bending a tungsten thin film having a thickness of 20 nm with various ions in a vacuum electron source was calculated using the SRIM code (http://www.srim.org/). Results are shown.
First, the optimal kinetic energy of irradiated ions is calculated when the average projected range is 50% of the film thickness, that is, when the depth is 10 nm. For example, in the case of silicon ions in the order of decreasing atomic number, about 25 keV, About 25 keV for phosphorus ions, about 30 keV for argon ions, about 50 keV for arsenic ions, about 55 keV for krypton ions, about 65 keV for indium ions, about 70 keV for antimony ions, and about 80 keV for xenon ions. .

Next, as a condition of the position control layer 50, when the film thickness at which the average projected range in the calculated kinetic energy of ion implantation is 15% or less of the film thickness is calculated with several materials, for example, density From about 420 nm to about 560 nm for photoresist, about 340 nm to about 450 nm for epoxy resin, about 190 nm to about 250 nm for carbon, about 240 nm to about 290 nm for silicon, and about 110 nm to about 120 nm for niobium. It can be seen that the required film thickness is about 95 nm to about 100 nm for molybdenum and about 70 nm for tungsten.
Of course, in the case where IIB processing is performed under the condition that the irradiation energy may be increased by setting the kinetic energy of the irradiation ions smaller than the optimum value, the position control layer 50 is selected according to the selected kinetic energy. The minimum required film thickness can be made smaller than the above calculation result.

(Specific examples of probe array applications)
Similarly, in the probe array for electrical measurement, the SRIM code (http://www.srim.org/) is used for the film thickness of the position control member 50 when a tungsten thin film having a thickness of 100 nm is bent with various ions. The calculated result is shown.
First, when the optimum kinetic energy of irradiated ions is calculated when the average projected range is 50% of the film thickness, that is, when the depth is 50 nm, for example, in the order of the atomic number, in the case of silicon ions, about 140 keV, About 150 keV for phosphorus ions, about 180 keV for argon ions, about 300 keV for arsenic ions, about 325 keV for krypton ions, about 450 keV for indium ions, about 450 keV for antimony ions, and about 500 keV for xenon ions. .

Next, as a condition of the position control layer 50, when the film thickness at which the average projected range in the calculated kinetic energy of ion implantation is 15% or less of the film thickness is calculated with several materials, for example, density From about 2200 nm to about 2500 nm for photoresist, from about 1800 nm to about 2100 nm for epoxy resin, from about 950 nm to about 1150 nm for carbon, from about 1250 nm to about 1350 nm for silicon, and from about 530 nm to about 560 nm for niobium. It can be seen that the minimum required film thickness is about 450 nm to about 480 nm for molybdenum and about 320 nm to 340 nm for tungsten.
Of course, in the case where IIB processing is performed under the condition that the irradiation energy may be increased by setting the kinetic energy of the irradiation ions smaller than the optimum value, the position control layer 50 is selected according to the selected kinetic energy. The minimum required film thickness can be made smaller than the above calculation result.

When the position control layer 50 is selected from materials used for semiconductor processes, it is convenient because molybdenum and tungsten can easily perform their functions with a relatively small film thickness.
On the other hand, carbon-based materials such as carbon, photoresist and polyimide have low ion shielding performance per unit thickness, but can easily form a thick film on the order of 1 μm. As desirable. The fact that carbon-based materials can be easily removed by oxygen ashing is another factor that simplifies the process.

(8) When ion irradiation is continuously performed, more ions are implanted, and the cantilever structure 40 is bent vertically. (Fig. 1 (H))
There are no fundamental restrictions on the ions used in the method of bending a thin film member according to the present invention. When using an ion implantation apparatus, phosphorus, boron, arsenic, indium, antimony, boron fluoride, aluminum, nitrogen, argon, silicon fluoride, silicon, boron hydride, carbon hydride, etc. are generally provided. Therefore it can be easily used.
When a focused ion beam (FIB) apparatus is used, gallium is generally provided so that it can be easily used.

In addition, hydrogen, helium, carbon, oxygen, fluorine, neon, magnesium, sulfur, chlorine, scandium, titanium, vanadium, chromium, manganese, iron, cobalt, even ionic species that are not widely used in the semiconductor industry , Nickel, copper, zinc, germanium, krypton, rubidium, zirconium, niobium, molybdenum, ruthenium, palladium, silver, cadmium, tin, iodine, xenon, hafnium, tungsten, iridium, platinum, gold, lead, bismuth, cerium, praseodymium , Neodymium, samarium, eurobium, gadolinium, terbium, dysprosium, erbium, thulium, ytterbium and the like have been established as ion irradiation techniques and can be easily used.

It is desirable from the viewpoint of controllability that bending can be performed with any one of the ion species selected from the above. However, due to the principle of IIB technology, it is possible to process with high accuracy even when a plurality of types of ions are mixed. Therefore, the present invention can also be implemented with an inexpensive ion irradiation apparatus that omits the mass analyzer.

(Point of embodiment)
The positioning of the tip 51 of the position control layer 50 is important. When the processing of the position control layer 50 is performed by anisotropic etching, higher positioning accuracy can be obtained as compared with the isotropic etching of the conventional method.
In dry etching processes such as RIE (Reactive Ion Etching) and ion milling, a thin film can be etched anisotropically and vertically, so that a fine pattern formed in a photoresist can be accurately transferred to the thin film.
On the other hand, in the case of wet etching, it is impossible to transfer the fine pattern formed on the photoresist as it is to the thin film because the etching process isotropically proceeds due to the effect of the chemical solution soaking.

(Features of the present invention)
As shown in FIG. 1 (F), the position control layer 50 for determining the position where the cantilever structure starts to bend in the direction in which the ions fly as viewed from the cantilever structure 40 exists at least at the time of ion irradiation. That is. When the position control layer 50 is selected so as not to be deformed when the cantilever structure 40 is subjected to ion irradiation for bending processing, the cantilever structure 40 is selected from the tip 51 of the position control layer 50. The position of the extended line down to the structure 40 determines the position where the cantilever structure 40 begins to bend.

As is apparent from the bending principle in the present invention described later, the position control layer 50 does not have to be a single layer film. Similarly, the position control layer support layer 60 for supporting the position control layer 50 may be omitted as is apparent from the bending principle in the present invention described later.

(Example)
An example of manufacturing a three-dimensional curved optical waveguide for silicon photonics will be described with reference to FIG.
(1) An SOI wafer is prepared in which the substrate 10 is single crystal silicon, the support layer forming start layer 20 ′ is a 2 μm thick silicon thermal oxide film, and the thin film member forming start layer 30 ′ is 220 nm thick single crystal silicon. To do. (Fig. 1 (A))
(2) A silicon optical circuit (thin film member 30) is produced by patterning single crystal silicon having a thickness of 220 nm by photolithography and RIE. (Fig. 1 (B))
(3) A SiO ₂ layer having a thickness of about 2 μm was formed as a cladding layer on the silicon optical circuit by the CVD method. In this state, the silicon optical circuit is completed. The cladding layer corresponds to the starting layer 60 ′ for forming the position control layer support layer. (Figure 1 (C))
(4) Tungsten having a film thickness of 200 nm is deposited by sputtering as the starting layer 50 ′ for forming the position control layer. (Figure 1 (D))
(5) The position control layer 50 is formed by patterning tungsten having a thickness of 200 nm by photolithography and RIE. (Figure 1 (E))
(6) a SiO ₂ layer is 'and a is SiO ₂ layer supporting layer forming starting layer 20' Position control layer supporting layer forming starting layer 60 with a buffer hydrofluoric acid solution to lateral etching, the silicon optical circuit The end portion is formed as a cantilever structure 40. (Fig. 1 (F))
(7) Irradiate a silicon ion beam with an acceleration energy of 80 kev to bend the end portion of the silicon optical circuit having a cantilever structure in the vertical upward direction of the substrate. (Fig. 1 (G), (H))
It should be noted that by devising the irradiation of the silicon ion beam, the end portion of the silicon optical circuit having a cantilever structure can be curved in the downward direction perpendicular to the substrate, if necessary.
This is applied to a structure in which, for example, when a plurality of silicon optical circuits are stacked, the end portion of the silicon optical circuit in the upper layer is bent downward and utilized.

In the above embodiment, tungsten having a thickness of 200 nm is used for the position control layer 50. However, a photoresist having a thickness of 2450 nm or more, an epoxy resin having a thickness of 1120 nm or more, carbon having a thickness of about 670 nm or more, and a thickness of about 780 nm. Replacement with a material that satisfies the conditions that do not cause deformation in the ion implantation process, such as silicon, niobium with a thickness of about 330 nm or more, molybdenum with a thickness of about 290 nm or more, and tungsten with a thickness of about 210 nm or more. Is possible.

FIG. 2A shows an SEM image after ion irradiation. Although the silicon optical waveguide is curved, it can be seen that the position where the cantilever structure starts to bend is determined by the edge position of the tungsten thin film which is the position control layer.
Thereafter, if the position control layer 50 made of tungsten is unnecessary, it is selectively removed by wet etching with an acid such as a sulfuric acid solution.

FIG. 2B shows an SEM image after removing tungsten.
A silicon optical waveguide with a curved tip is revealed, and the position where the cantilever beam structure begins to bend is determined by the peripheral surface of the support layer in the prior art, but in this embodiment it is more than that. It can be confirmed that the turn starts from the left side position.
Compared with the SEM image of FIG. 2A, it can be seen that the position where the bending starts is aligned with the position where the edge of the tungsten thin film originally existed.

In the conventional IIB technology, the peripheral surface determined by wet etching, which has low in-plane uniformity and reproducibility of position control, determines the position of the curved process, so even if it can be used for lab-level device demonstration, mass production It was a difficult technology.
The present invention overcomes this problem and enables a process in which the peripheral surface of the position control layer determined by dry etching, which has high in-plane uniformity and reproducibility of position control, determines the position of the bending process.

Furthermore, according to the present invention, since a plurality of cantilever structures can be arranged close to each other, high integration of the processed cantilever structures can also be achieved.
The effect of high integration will be described with reference to the schematic diagram shown in FIG.
FIG. 3A is a top view of a state in which the cantilever structure 40 is formed on the thin film member 30 according to the present invention. The length of the desired three-dimensional structure is L, and a state where a cantilever structure of length L + α is formed on the thin film member 30 for that purpose is shown. A tip 51 indicated by a solid line indicates an edge portion of the position control layer 50 in the cross-sectional view shown in FIG.
By subjecting this structure to ion irradiation, the cantilever structure 40 is bent. The position where the cantilever structure 40 begins to bend is a position where it extends in the direction of the downward extension of the tip 51 of the position control layer 50 and intersects the cantilever structure 40. That is, the portion indicated as the length L in FIG. 3A is the portion to be subjected to the solid bending process.

First of all, if the present invention is adopted, a design that avoids the structure like the cape shown in FIG. 7 is not necessary. Therefore, in the step of removing the support layer 20 and the position control layer support layer 60 by isotropic etching, the lateral etching at a distance corresponding to the length L of the desired three-dimensional structure is not performed. A stop line of the shortened planar etching region is drawn as the peripheral surface 21. In FIG. 3, the cantilever structure 40 is drawn as a rectangle, but it may be a triangle, a circle, or another two-dimensional figure.

In FIGS. 9 and 10, the length of the thin film member 30 other than the cantilever structure 40 (the portion necessary as a region for supporting the produced three-dimensional structure) is indicated by M, and this portion is also included. In order to avoid interference with the planar etching region, it has been an important design item. However, when the present invention is used, such interference in the planar etching region is allowed to some extent. The portion of the length M shown in FIGS. 9 and 10 is omitted.
Consider the integration density in the case where a plurality of such structures in FIG. 3A are arranged on the same substrate to form the same three-dimensional structure.

The interference between the planar etching regions of the two cantilevers will be considered with reference to FIG. When the cantilever structure is arranged at the interval g <L without aligning the positions at which the bend starts, interference does not occur because the planar etching region is reduced compared to FIG. Therefore, only cantilever beams adjacent to each other are bent at the same length L.
When the distance g is further reduced and the distance between the planar etching regions interferes, as shown in FIG. 9D, the lengths of adjacent cantilever structures are L + α on one side, L + β on the other side, (Α is smaller than β). However, the position where the cantilever beam structure 40 begins to bend is a position where it extends in the direction of the downward extension of the tip 51 of the position control layer 50 and intersects with the cantilever beam structure 40. Even if they are different from each other, the curved lengths of adjacent cantilever structures can be the same L. This is an essence that overcomes the limitations of integrated design, which was a drawback of the IIB technology according to the present invention. Hereinafter, this is expanded into a square lattice and a triangular lattice.

An example of arrangement in a square lattice is shown in FIG. In FIG. 3E, the free end of the tip of the cantilever structure is arranged at each vertex of the square lattice.
It can be seen that the interval g can be easily reduced to g <L in both the X-axis and Y-axis directions.
Similarly, as shown in FIG. 3 (F), it can be seen that the gap g can be easily reduced to g <L even when arranged in a triangular lattice. In FIG. 3F, the free end of the cantilever structure is arranged at each vertex of the triangular lattice.
Furthermore, if the number is limited, the interval g can be reduced to g <L / 2 as shown in FIG. Such a two-dimensional arrangement with a narrow pitch is not possible with the prior art.

The present invention is applied to an example in which an electrical measurement probe having a height of 50 μm is manufactured using IIB technology. In this case, the length L of the cantilever structure is required to be 50 μm. If this is arranged as shown in FIG. 3E, the distance between the X direction and the Y direction is less than about 50 μm. Similarly, when the array is formed in the triangular lattice shape of FIG. 3F, the interval is less than about 50 μm. Similarly, when the array is formed in the triangular lattice shape of FIG. 3G, the interval is less than about 25 μm. These values are smaller than those of the existing electrical measurement probe array, and the merit of fine integration can be obtained by introducing the IIB technology.

The present invention is applied to an example in which a vacuum electron source having a height of 1 μm is manufactured using IIB technology. In this case, the length L of the cantilever structure is 1 μm. If this is arranged two-dimensionally, the arrangement in the X direction and the Y direction will be less than about 1 μm if arranged as shown in FIG. Similarly, when the array is formed in the triangular lattice shape of FIG. 3F, the interval is less than about 1 μm. Similarly, when the array is formed in the triangular lattice shape of FIG. 3G, the interval is less than about 0.5 μm. These values are smaller than those of the existing vacuum electron source array, and it can be said that the introduction of the IIB technology in the manufacture of the vacuum electron source array can also provide a merit of significant increase in density.

The present invention is applied to an example in which a vertical photocoupler for silicon photonics having a curvature radius of 3 μm is manufactured using IIB technology. In this case, the length L of the cantilever structure is required to be 5 μm. When the two-dimensional array is arranged as shown in FIG. 3E, the interval in the X direction and the Y direction is less than about 5 μm. Similarly, when arrayed in the triangular lattice shape of FIG. On the other hand, when an array is formed in a triangular lattice with a limited number as shown in FIG. 3G, an interval of less than 2.5 μm is possible.

These values are values that cannot be achieved with conventional silicon optical couplers. Therefore, spatially multiple light propagation such as fumode fiber, multimode fiber, multicore fiber, fumode multicore fiber or multimode multicore fiber, laser diode array, surface emitting laser array, MEMS mirror, spatial phase modulator, etc. This is effective when optical coupling is performed with an optical device in which modes are integrated, and a dramatic reduction in device size can be realized.

3) The arrangement method shown in FIG. 3G can ultimately achieve a pinching pitch of about 0.5 μm. In that case, there is a great merit in the application of interference devices. For example, assuming a 1.55 μm band used in optical communication, the meaning of setting the interval between vertical light output ports to about 0.5 μm means that the pitch of the light source can be made about half of the wavelength, Application to a rider (LIDAR: Light Detection and Ranging) or Laser ImagingDetectDetection and Ranging) will enable dramatic technological innovation.

(3D curved optical circuit)
In the case of application to a three-dimensionally curved optical waveguide, bending is performed in a downward direction as shown in FIG. It can also be used to optically connect the silicon optical circuit and the upper silicon optical circuit.
This will be described in detail with reference to FIG.
In accordance with the present invention, a silicon optical waveguide having a vertical optical coupler that is three-dimensionally curved upward is filled with a cladding layer 70 to obtain a lower layer silicon optical circuit.
Next, a start layer for forming a support layer and a start layer for forming an optical waveguide are laminated thereon, and an optical waveguide is formed by patterning the start layer for forming an optical waveguide, and is supported directly or on the position control layer on the optical waveguide. A position control layer for determining a position where the cantilever structure of the optical waveguide starts to bend by irradiating ions through the layer is formed by anisotropic etching, and one of the starting layers for forming the support layer under the optical waveguide is formed. Or a part of the starting layer for forming the support layer and a part of the starting layer for forming the position control layer, the peripheral surface of the support layer, or the peripheral surface of the support layer and the end surface of the position control layer supporting layer are positioned. The cantilever structure 40 having the tip free end 41 is formed in the optical waveguide by removing by isotropic etching so as to be on the lower extension line of the tip of the control layer or on the inner side thereof. (Fig. 4 (A))

Next, a part of the cantilever structure 40 is irradiated with ions using the position control layer 50 as a mask, and a part of the cantilever structure is bent downward to form a second vertical optical coupler. (Fig. 4 (B))
At this time, the lower vertical optical coupler in the lower layer and the second vertical optical coupler in the upper layer are arranged to face each other so as to be optically connected.
Finally, by removing the position control layer 50 and filling with the clad layer 70, an inter-layer connected solid curved optical circuit is obtained. (Fig. 4 (C))
Such connection between the upper and lower optical circuits cannot be realized without the highly accurate position control technique according to this patent.

As described above, the embodiments and examples disclosed in this specification are illustrated for facilitating the understanding of the method of bending a thin film member according to the present invention, and the present invention is not limited thereto. is not.
That is, it is needless to say that the present invention can be appropriately changed in design when performing the method of bending a thin film member without departing from the scope of the claims.

DESCRIPTION OF SYMBOLS 10 Substrate 20 Support layer 20 'Starting layer 21 for supporting layer formation Peripheral surface 30 Thin film member 30' Starting layer 40 for forming thin film member Cantilever structure 41 Free tip end 50 Position control layer 50 'Starting layer 51 for position control layer formation Tip 60 Position control layer support layer 60 ′ Position control layer support layer forming starting layer 61 End face 70 Clad layer

Claims

A step of laminating a starting layer for forming a support layer and a starting layer for forming a thin film member on a substrate; a step of patterning the starting layer for forming a thin film member to form a thin film member; and a direct or position on the thin film member Forming a position control layer for determining a position where the cantilever structure of the thin film member starts to bend by irradiating ions through the starting layer for forming the control layer support layer; and the support under the thin film member A part of the starting layer for layer formation or a part of the starting layer for forming the supporting layer and a part of the starting layer for forming the supporting layer are arranged on the peripheral surface of the supporting layer or the peripheral surface of the supporting layer and the position control. Removing the end face of the layer support layer so as to be on or inside the lower extension line of the tip of the position control layer to form a cantilever structure having a tip free end on the thin film member; The cantilever beam structure using the control layer as a mask Part irradiated with ions, curving a thin film member including a step of bending a portion of the cantilever structure.
A step of laminating a starting layer for forming a support layer and a starting layer for forming a thin film member on a substrate; a step of patterning the starting layer for forming a thin film member to form a thin film member; and a direct or position on the thin film member Forming a position control layer by anisotropic etching for determining a position at which the cantilever structure of the thin film member begins to bend by irradiating ions through the starting layer for forming the control layer support layer; and the thin film member A part of the starting layer for forming the supporting layer below, a part of the starting layer for forming the supporting layer and a part of the starting layer for forming the position control layer, the peripheral surface of the supporting layer, or the supporting layer The cantilever having a free end at the tip is removed by isotropic etching so that the peripheral surface and the end face of the position control layer support layer are on or below the extension line of the tip of the position control layer. Process for forming beam structure , Part irradiated with ions of the cantilever structure the position control layer as a mask, curving a thin film member including a step of bending a portion of the cantilever structure.
3. The method for bending a thin film member according to claim 1, wherein the cantilever structure constitutes an electrical measurement probe.
3. The method of bending a thin film member according to claim 1 or 2, wherein the cantilever structure constitutes an emitter of a field emission device.
The method for bending a thin film member according to claim 1 or 2, wherein the cantilever structure constitutes an optical waveguide.
A step of laminating a starting layer for forming a support layer and a starting layer for forming an optical waveguide on a substrate; a step of patterning the starting layer for forming an optical waveguide to form an optical waveguide; and a direct or position on the optical waveguide. Forming a position control layer by anisotropic etching to determine a position at which the cantilever structure of the optical waveguide starts to bend by irradiating ions through the control layer support layer; and under the optical waveguide, A part of the starting layer for forming the support layer or a part of the starting layer for forming the supporting layer and a part of the starting layer for forming the position control layer are arranged on the peripheral surface of the supporting layer or the peripheral surface and the position of the supporting layer. The end face of the control layer support layer is removed by isotropic etching so that it is on the lower extension line of the tip of the position control layer or inside thereof, thereby forming a cantilever structure having a tip free end in the optical waveguide. And the position system Layer ion is irradiated to a portion of the cantilever structure as a mask, curved method of processing an optical waveguide including a step, a to the vertical optical coupler by bending a portion of the cantilever structure.
The method for bending an optical waveguide according to claim 6, wherein the substrate is a silicon optical integrated circuit substrate.
The method of bending an optical waveguide according to claim 6 or 7, wherein the optical waveguide is an optical waveguide mainly composed of silicon.
The method for bending an optical waveguide according to claim 8, wherein the constituent material of the position control layer is tungsten or carbon.
10. A plurality of two-dimensional arrays of the cantilever structure of the optical waveguide are arranged at intervals smaller than the length of the cantilever structure. A method of bending an optical waveguide as described.
11. The method of bending an optical waveguide according to claim 10, wherein the free end of the cantilever structure is arranged at each vertex of a square lattice.
11. The method of bending an optical waveguide according to claim 10, wherein the free ends of the cantilever structure are arranged at the respective apexes of the triangular lattice.
10. The plurality of two-dimensional arrays of the cantilever structures of the optical waveguide are arranged at intervals smaller than the wavelength of propagating light of the optical waveguide. A method for bending an optical waveguide according to claim 1.
The step of bending a part of the cantilever structure to form a vertical optical coupler is a process of bending a part of the cantilever structure upward to form a vertical optical coupler. 7. A method for bending an optical waveguide according to item 6.
A step of filling an upwardly curved vertical optical coupler with a clad layer; a step of laminating a second support layer forming start layer and a second optical waveguide forming start layer on the clad layer; and Patterning the second optical waveguide forming starting layer to form a second optical waveguide, and the second optical waveguide directly on the second optical waveguide or via the second position control layer support layer Forming a second position control layer by anisotropic etching for determining a position where the second cantilever structure of the optical waveguide starts to bend by irradiating ions; and under the second optical waveguide, A part of the starting layer for forming the second support layer or a part of the starting layer for forming the second supporting layer and a part of the starting layer for forming the second position control layer supporting layer are used as the second support. The peripheral surface of the layer, or the peripheral surface of the second support layer and the edge of the second position control layer support layer Is removed by isotropic etching so as to be on or inside the lower extension line of the tip of the position control layer to form a cantilever structure having a tip free end in the optical waveguide; and Using the position control layer of 2 as a mask, a part of the second cantilever structure is irradiated with ions, and a part of the second cantilever structure is bent downward to make optical connection with the vertical optical coupler. The method for bending an optical waveguide according to claim 14, further comprising a step of forming a second vertical optical coupler.