WO2007019493A2

WO2007019493A2 - Process for making single crystalline flakes using deep etching

Info

Publication number: WO2007019493A2
Application number: PCT/US2006/030870
Authority: WO
Inventors: Sadeg M. Faris
Original assignee: Reveo, Inc.
Priority date: 2005-08-05
Filing date: 2006-08-07
Publication date: 2007-02-15
Also published as: WO2007019487A2; WO2007019493A3; WO2007019487A3

Abstract

A method of forming a plurality of thin strips of semiconductor material includes cutting a wafer having a thickness K from a semiconductor boule. The wafer is masked to form a plurality mask lines of a thickness associated with the desired thickness of the thin strips of semiconductor material. The wafer is etched to form plateaus of semiconductor material beneath the mask line having heights associated with the desired width of the thin strips of semiconductor material. The plateaus are removed thereby providing thin strips of semiconductor material.

Description

Process for Making Single Crystalline Flakes Using Deep Etching

By Sadeg M. Fans, Ph.D. RELATED APPLICATIONS

This application claims the benefit under 35 U.S.C. § 119(e) of U.S. Provisional Application No. 60/705,925 filed on August 5, 2005 entitled "Method and System for Fabricating Milti Layer Devices" which is incorporated by reference herein. BACKGROUND ART

It is well know that microelectronics and integrated circuits rely on semiconductor wafers of high quality, having minimum defects, and minimum impurities. This constraint is one of the key factors affecting the cost of integrated circuits, solar cells, sensors and other devices. The Sliver Cell concept, described in Stocks, MJ. and Weber, K. J. and Blakers, A. W. and Babaei, J. and Everett, V. and Neuendorf, A. and Kerr, M. and Verlinden, P. J. (2003) 65-micron thin monocrystallϊne silicon solar cell technology allowing 12-fold reduction in silicon usage", In Proceedings 3rd World Conference of Photovoltaic Solar Energy Conversion, Osaka, Japan, teaches a useful method to form slivers of wafers. However, improved application and processing may be desirable.

Various problems are often encountered in processing semiconductor wafers, particularly when processing the wafers to form very thin semiconductor based devices. Problem 1:

Silicon wafers of thickness range in the range of 100 microns to 1000 microns are typically cut with a saw. The thickness of the saw causes about 100 microns of the material to be lost, commonly referred to as kerf loss. Problem 2: Advanced high performance integrated circuits technology is relying increasingly on SOI wafers, which require thin Si layers transferred to/bonded to a substrate of Si with a separation of thin SiO₂, or a thin layer bonded to glass or Sapphire (SOS). There are several known techniques for producing SOI wafers (SOITEC, SiGen, O2 Implantation). ^"While these processes may be suitable for certain purposes, they still have limitations of complexity, expensive equipment requirement, and production of a relatively small area of thin film Si per volume of Si cut from the boule or ingot. For instance, in the case of SOS, a thick layer of Si is bonded to Sapphire which is then ground and etched to produce the final desired thin Si on Sapphire- This clearly is wasteful of expensive Si, which is ground off. Problem 3:

The emerging 3D integrated circuits (3D IC) technology (see, e.g., Applicant's PCT Application PCT/US03/37304 entitled "Three Dimensional Device Assembly and Production Methods Thereof filed on November 20, 2003, which is incorporated herein by reference, and IBM U.S. Patent No. 6,355,501) aims to increase the device density by making three dimensional stacks of ultra thin IC wafers of chips. Almost all proposals of 3D IC methods adopt thinning steps to obtain ultra thin Si IC by grinding off most of the Si mechanically and then use chemical etching and polishing. The thinning process steps and the waste of the Si material add to the cost making the desired devices. BRIEF SUMMARY OF THE PΪVΕNTION

It is an object herein to make ultra thin material (e.g., Si) layers, wafers and substrates, and SOI and SOS wafers without the wasteful steps of problems 1-3 described above. In general, a method of forming a plurality of thin, strips of semiconductor material includes cutting a wafer having a thickness K from a semiconductor boule. The wafer is masked to form a plurality mask lines of a thickness associated with the desired thickness of the thin strips of semiconductor material. The wafer is etched to form plateaus of semiconductor material beneath the mask line having heights associated with the desired width of the thin strips of semiconductor material. The plateaus are removed thereby providing thin strips of semiconductor material. BRIEF DESCRIPTION OF THE FIGURES

[01] The foregoing summary as well as the following detailed description of preferred embodiments of the invention will be better understood when read in conjunction with the appended drawings. For the purpose of illustrating the invention, there is shown in the drawings embodiments which are presently preferred. It should be understood, however, that the invention is not limited to the precise arrangements and instrumentalities shown. In the drawings, where:

[02] FIG. 1 is a schematic of a conventional silicon boule;

[03] FIGs.2 and 3 show steps of a method of embodiments of the present invention including deriving relatively thick wafer from a boule;

[04] FIGs.4-6 show steps of treatment of surface of the relatively thick wafer to allow formation of trenches;

[05] FIG. 7 shows formation, of trenches on more than one surface of the relatively fhik wafer;

[06] FIGs. 8A and 8B show vertical slivers detached from the parent slice;

[07] FIG.9 shows slivers mounted or bonded edgewise onto a host substrate; [08] FIG. 10 shows an etch mask which fixes the lengths of the slivers with patterned regions;

[09] FIG. 11 shows processing of a plurality of vertical slivers while they are attached to the slice; and

[10] FIGs. 12A-12H show another process for forming single crystalline silicon flakes or strips.

DETAILED DESCRIPTION OF THE FIGURES

Figure 1 schematically shows the conventional method of cutting a thin slice 12 from a boule 10 (e.g., Si), leaving kerf loss regions 14 between each usable layer 12.

Referring now to Figures 2 and 3, a step in the method of the present invention is shown, whereby a relatively thicker slice 22 (e.g., 0.1 - 10 cm) from a boule 20. The waste associated with the kerf loss regions 24 described above is reduced by a factor of 10 to 100, as less cuts are made throughout the thickness of the boule 20. Preferably, this thick slice 22 is polished on both major surfaces 25, 26 by polishing techniques well known in the art to produce surfaces that are ultra flat and smooth with surface roughness in the sub-nanometer range.

Referring now to Figures 4-6, conventional lithography techniques are used to mask the surface of wafer 22 to allow formation of trenches between mask lines 42. The trenches generally are of length X. If the thick Si slice 22 is circular, many lines of different lengths Xi are produced. If the slice 22 is square, then the lines are of equal length X. The mask lines 42 generally have a width dl protecting the Si from etching, thereby exposing the surface of silicon and a spacing of width d2. The lines are preferably aligned along a crystalline direction to facilitate the preferential etching described herein so that high aspect ratio deep trenches are produced.

Using a suitable etching technique (e.g., including but not limited to reactive ion etching, reactive plasma etching, chemical etching, or any combination thereof), the exposed silicon is etched away. The etching method is optimized to produced preferential etching along chosen crystalline direction to produce very high aspect ratios (depth Z : width dl), preferably ratios of about 10-1000. The result is the production of thin vertical slivers 62.

As an alternative to the above described etching techniques relying on lithographic masking, or in conjunction therewith, substrate machining techniques, such as femto-second laser machining may be employed to create the gaps between slivers.¹ Referring to Figure 7, the process may be repeated, or occur simultaneously, on both surfaces 25, 26 of the slice 22. The final result is the production of thin vertical slivers 62 of Si on both sides of the thick Si slice and attached to the slice. These slivers are rectangular having an area of Xi*Z, and a thickness d2. hi general, the depth Z can range from 10 micron to more than 100 microns, although those of skill in the art will appreciate that the maximum depth Z may increase as deep trench etching technology continues to develop.

Referring now to Figures 8 A and 8B, vertical slivers 62 are now detached from the parent slice. Detaching may be my techniques including but not limited to

¹ See, e.g., US20030152756 — "Ih recent years, ϊtis knownthatthe^mto-seconrf/αserϊs employed fora method of precisely processing a metal or me like. ID this case, a laser having a pulse width of several tens femto-seconds to several hundred femto-seconds is used and Ti: sapphire is typically employed for its light source. It is known that this method is capable of providing a fine and precise process for various materials made of metal, ceramic or others. For instance, see the following papers by the authors of US20030152756, Kumagawa and Midorikawa: Appl. Phys. A 63, 109-115 (1996); Oyo-buturi (Jpn. J. Appl. Phys.) 67(9), 1051 (1998); and O Plus E 21 (9), 1130 (1999)." mechanical cleaving, lateral etching at the bottom of the slivers, or a combination thereof (e.g., mechanical cleaving aided with markings etch at the bottom of the slivers).

The above described technique produces a large number of ultra thin slivers of silicon in a more efficient manner than conventionally known and with minimum wasted Si material. The number of slivers per side of the slice 22 may be represented by:

N = D/(dl+d2), where D is the diameter of the slice 22 and dl and d2 are as described above.

The total area of thin slivers produced by this method is

^ = ∑(Z-Xi)/(dl+d2)

Xi

At can be orders of magnitude higher than the area

from a wafer of

diameter X and a thickness T of a thin Si layer produced by conventional methods.

Further, the new method produces the thin slivers of the same area as prior art by saving

(T+K)/(dl+d2) of the expensive Si material, where K is the kerf attributed to the width of the cutting saw.

The above material savings can be exemplified numerically.

Example 1: Based on the following assumptions, increased yield and material savings is shown: dl = d2 =1 micron T= 500 micron

K = 100 micron Z = 200 microns

Material savings ratio: sliver method to conventional thin layer fabrication:

(T+K)/(dl+d2) = (500+100)/(l+l) = 300/1

Increased area per slice Assuming slice 22 is a square having side length X = 200,000 micron, N = D/(dl+d2) = 200,000/2 = 100,000 slivers. Therefore, total area based on sliver method = NZX = (100,000)(200μm)(200,000 μrα>= 4x10¹² μm², compared to X² for conventional cutting methods, whereby the area is 4x10¹⁰ μm². Accordingly, the surface area yield is 100 times larger that if produced by conventional method using the same dimension X.

Example 2: Even with a more conservative approach, increased yield and material savings is still demonstrated: dl = d2 =2 micron T= 200 micron

K = 100 micron Z = 50 microns

Material savings ratio: sliver method to conventional thin layer fabrication: (T+K)/(dl-hl2) = (200+100)/(2+2) = 75/1 Increased area per slice

Assuming slice 22 is a square having side length X = 200,000 micron, N = D/(dl+d2) = 200,000/4 = 50,000 slivers. Therefore, total area based on sliver method = NZX = (50,000)(50μm)(200,000 μm)= 5x10^u μm², compared to X² for conventional cutting methods, whereby the area is 4x10¹⁰ μm². Accordingly, the surface area yield is 12.5 times larger that if produced by conventional method using the same dimension X.

Referring now to Figure 9, N slivers 62 may be mounted or bonded edgewise onto a host substrate 64. The substrate 64 may be any suitable substrate, depending, for example, on the desired device or processing conditions. The substrate 64 may be a relatively inexpensive substrate (as compared to the semiconductor material of the slivers 62), such as glass, fused quartz, or metal substrates. The slivers 62 may be bonded to the substrate by any known methods.

In certain embodiments, the selective bonding methods taught in applicant's copending U.S. Patent Application Serial No.09/950,909 filed on 9/12/2001 entitled "Thin films and Production Methods Thereof, which is incorporated herein by reference, may be used. The substrate used therein, referred to as a selectively bonded multiple layer substrate, allows for processing of multiple useful devices on a wafer as is known, but allows the device layer of the wafer to be readily removed, preferably without mechanical grinding or other etch-back techniques (e.g., by peeling).

The slivers produced herein may be used to process a variety of useful devices, including but not limited to integrated circuits, memory circuit, power circuit, other micro-electronic circuit, optical device, photovoltaic device, micro-electro-mechanical device, microfluidic device, or a combination of any of the above.

The device material and the substrate material may be the same or different materials. While commonly referred to herein as semiconductor materials, the present invention may apply to other materials useful for forming devices ("device material"). These materials may include materials including, but not limited to, semiconductor, plastic (e.g., polycarbonate), metal, insulator, monocrystalline, amorphous, noncrystalline, biological (e.g., DNA based films) or a combination comprising at least one of the foregoing types of materials. For example, specific types of materials include silicon (e.g., monocrystalline, polycrystalline, noncrystalline, polysilicon, and derivatives such as Si₃N₄, SiC, SiO₂), GaAs, InP, CdSe, CdTe, SiGe, GaAsP, GaN, SiC, GaAlAs, InAs, AlGaSb, InGaAs, ZnS, AlN, TiN, other group HIA-VA materials, group HB materials, group VIA materials, sapphire, quartz (crystal or glass), diamond, silica and/or silicate based material, or any combination comprising at least one of the foregoing materials. Of course, processing of other types of materials may benefit from the process described herein to provide slivers of desired composition. Materials which are particularly suitable for the herein described methods include semiconductor material (e.g., silicon) as the device material, and semiconductor material (e.g., silicon) as the substrate, other combinations include, but are not limited to; semiconductor (device) on glass (substrate); semiconductor (device) on silicon carbide (substrate) semiconductor (device) on sapphire (substrate); GaN (device) on sapphire (substrate); GaN (device) on glass (substrate); GaN (device) on silicon carbide (substrate);plastic (device) on plastic (substrate), wherein substrate and device layers may be the same or different plastics; and plastic (device) on glass (substrate).

The methods described herein for producing IC and other devices relies on the fact that large continuous Si layers are not needed. The devices have sizes much less than the diameter of the wafers (Z).

In certain embodiments, since very large quantities of slivers 62 may be produced, automated techniques for placing and attaching the slivers 62 onto the host substrate 64 are employed. For example, conventional substrates are based on a boule, typically cylindrical in shape thus producing circular slices. However, the substrate that may be used herein can take the form of a continuously fed roll, thereby facilitating certain processing and handling operations of the devices on the substrate. In another example, a substrate may have pre-formed indentations configured to fit the slivers 62 In certain embodiments, and referring to Figure 10, an etch mask 72 is provided which fixes the lengths of the slivers 76 with patterned regions 74. For example, the mask used in steps of Figures 4-6 may have a pattern as exemplified in Figure 10, or alternatively have a striped pattern (e.g., as in Figure 4), or any other desired pattern. Accordingly, processing may readily result in greater than 90, preferably greater than 99% of the slivers have substantially the same lengths (e.g., within 1% or less tolerance, in certain applications within less than 0.1% tolerance), even if the thick slice is circular. Thus, a very narrow length distribution is achieved. This will facilitate automated processing described above.

For certain manufacturing processes, and referring now to Figure 11, it may be desirable to process the plurality of vertical slivers 62 while they are attached to the slice 22. After etching or otherwise forming the slivers, the gaps therebetween may be filled with a plug region 68 of a suitable material that may subsequently be removed while minimizing or elimmating damage to the slivers 62. When the slice is solid, the slivers may be detached from the mother slice in a conventional manner. For example, the plural slivers 62 may be removed in one step by fracture, etching perpendicular to the Z direction, or other methods. Further, as another alternative, prior to etching steps of Figures 4-6, the surface(s) of slice 22 may be ion implanted to the desired depth Z to weaken the bond thereby facilitating pulling the plural slivers from the mother slice 22. Further, while the plural slivers 62 are on the handler, they may be polishing (e.g. CMP polishing) to achieve uniform Z. This will facilitate automated processing described above. It should be noted that an optimized etching method are desired that achieve Z values in the cm range. Thus, slivers of areas Z= 1 cm by X =1 cm will be possible. Nonetheless, the present invention should not be limited, in that slivers of areas on the order of Z=IOs of microns may be provided of any desired length X.

Referring to Figures 12A-12H, another process for forming single crystalline silicon flakes or strips is provided. Referring to Figure 12A, deep trenches are formed in a layer, e.g., 10 microns to about 100 microns, as described above. Referring to Figure 12B, a removable material 80 is introduced at the bottom surface of the deep trenches. This material 80 is resistance to oxidation. Referring to Figure 12C, the walls of the deep trenches are oxidized 82. Referring to Figure 12D, with a portion of the walls of the deep trenches oxidized, deep trench etching continues to form an even deep trench. Advantageously, by providing a portion of the walls of the deep trenches oxidized, undue etching of those portions of the wall is limited.

Referring to Figure 12E, the bottom of the trenches are again protected to prevent oxidation. Figure 12F shows another oxidation step of the inside of the walls of the trenches. Figure 12G shows another continued deep trenching step. Figure 12H shows a portion of s surface having ultra deep trenches.

These trenches then may be removed as described above, and used for further processing as strips, or broken or etched into flakes of desired dimensions.

While preferred embodiments have been shown and described, various modifications and substitutions may be made thereto without departing from the spirit and scope of the invention. Accordingly, it is to be understood that the present invention has been described by way of illustrations and not limitation.

Claims

What is claimed is:

1. A method of forming a plurality of thin strips of semiconductor material comprising: cutting a wafer having a thickness K from a semiconductor boule; masking the wafer to form a plurality mask lines of a thickness associated with the desired thickness of the thin strips of semiconductor material; etching the wafer to form plateaus of semiconductor material heneath the mask line having heights associated with the desired width of the thin strips of semiconductor material; and removing said plateaus thereby providing thin strips of semiconductor material.

2. A method of forming a thin strip of semiconductor material having a length Ls, width Ws and thickness Ts, the method comprising: cutting a wafer having a thickness K from a semiconductor boule; masking the wafer to form a mask line of a thickness Tm associated with the desired thickness Ts; etching on opposing sides of the mask line to form a plateau of a height Hp of semiconductor material beneath the mask line associated with the desired width Ws; and removing said plateau thereby providing the thin strip of semiconductor material having a length Ls, width Ws and thickness Ts.

3. A method of a making a precursor to a thin strip of semiconductor material having a length Ls₅ width Ws and thickness Ts, the method comprising: cutting a wafer having a thickness K from a semiconductor boule; masking the wafer to form a mask line of a thickness Tm associated with the desired thickness Ts; and etching on opposing sides of the mask line to form a plateau of a height Hp of semiconductor material beneath the mask line associated with the desired width Ws.

4. A method of forming a precursor to a plurality of thin strips of semiconductor material comprising: cutting a wafer having a thickness K from a semiconductor boule; masking the wafer to form a plurality mask lines of a thickness associated with the desired thickness of the thin strips of semiconductor material; and etching the wafer to form plateaus of semiconductor material beneath the mask line having heights associated with the desired width of the thin strips of semiconductor material.

5. A method as in one of claims 1-4, wherein etching is selected from group of etching techniques consisting of reactive ion etching, reactive plasma etching, chemical etching.

6. A method as in one of claims 1-4, wherein etching comprises substrate machining techniques such as femto-second laser machining.

7. A method as in one of claims 1-4, wherein etching is optimized to produced preferential etching along chosen crystalline direction to produce very high aspect ratios (depth Z : width dl) of about 10-1000.

8. A method as in one of claims 1-2, further comprising detaching vertical slivers form the parent slice.

9. A method as in one of claims 1-2, further comprising detaching vertical slivers form the parent slice by mechanical cleaving, lateral etching at the bottom of the slivers, or a combination thereof (e.g., mechanical cleaving aided with markings etched at the bottom of the slivers).

10. A method as in one of claims 1-4, wherein etching is based on a pattern optimized to produce slivers having predetermined lengths X.

11. A method as in one of claims 1-4, wherein etching is based on a pattern optimized to produce a plurality of slivers having lengths X, wherein at least 90% of said slivers have lengths X within at least 1% tolerance.

12. A method as in one of claims 1 or 2, wherein the thin strips of semiconductor material are supported on a substrate for processing or a useful device.

13. A method of forming a plurality of thin strips of semiconductor material comprising: providing a wafer having a thickness K; masking the wafer to form a plurality mask lines of a thickness associated with the desired thickness of the thin strips of semiconductor material; first step etching the wafer to form plateaus of semiconductor material beneath the mask line; protecting a bottom surface of the trench formed in the first step etching from oxidation; oxidizing inner walls of the trench formed in the first step etching; second step etching the trench formed in the first step etching to form deeper trenches; removing said plateaus thereby providing thin strips of semiconductor material.

14. The method as in claim 13, further comprising, prior to removing said plateaus, protecting a bottom surface of the deeper trench formed in the second step etching from oxidation; oxidizing inner walls of the trench formed in the first step etching and the deeper trench formed in the second step etching; third step etching the deeper trench formed in the second step etching to form even deeper trenches.