US20070111406A1 - FET Channel Having a Strained Lattice Structure Along Multiple Surfaces - Google Patents

Abstract

A channel16of a FinFET10has a channel core24and a channel envelope32, each made from a semiconductor material defining a different lattice structure to exploit strained silicon properties. A gate is coupled to the channel envelope through a gate dielectric. Exemplary materials are Si and Si_xGel_1-x, wherein 78<x<92. The channel core24has a top surface26of width w_cand an upstanding surface28, 30of height h_c, preferably oriented 90° to one another. The channel envelope32is in contact with the top26and upstanding surfaces28, 30so that the area of interface is increased as compared to contact only along the top surface26, improving electrical conductivity and gate18control over the channel16. The height h_ccan be tailored to enable a smaller scale FET10within a stabilized SRAM. Various methods of making the channel16are disclosed, including a mask and etch method, a handle wafer/carrier wafer method, and a shallow trench method. Embodiments and methods for FinFETs with one to four gates are disclosed.

Description

TECHNICAL FIELD
• These teachings relate generally to field effect transistors (FET) disposed on a semiconductor wafer or chip, in particular, to a layered structure as a channel that connects a source and a drain of a FET, wherein one of the layers is characterized by a strained lattice structure.
BACKGROUND OF THE INVENTION
• Semiconductors and integrated circuit chips have become ubiquitous within many products due to their continually decreasing cost and size. Miniaturization in general allows increased performance (more processing per clock cycle and less heat generated) at lower power levels and lower cost. Present technology is at or approaching atomic-level scaling of certain micro-devices such as logic gates, FETs, capacitors, etc. Circuit chips with hundreds of millions of such devices are not uncommon. Further size reductions appear to be approaching the physical limit of trace lines and micro-devices that are embedded upon and within their semiconductor substrates. The present invention is directed to such micro-sized FET devices. A FET is a transistor consisting of a source, a gate, and a drain. The action of the FET depends on the flow of majority carriers along a channel between the source and drain that runs past the gate. Current through the channel, which is between the source and drain, is controlled by the transverse electric field under the gate. More than one gate may be used to more effectively control the channel. The length of the gate determines how fast the FET switches and how fast the circuit can operate, and is generally about the same as the length of the channel (i.e., the distance between the source and drain). State of the art gate lengths are today on the order of 50 nm, and are moving toward 10 nm within the next decade. Such size reductions should enable upwards of a billion devices on a single chip. However, such small scaling necessitates greater control over performance issues such as short channel effects, punch-through, and MOS leakage current.
• Recently, the size of FETs has been successfully reduced through the use of one or more fin-shaped channels, known as fins. FETs employing fins are known as FinFETs. Previously, complementary metal-oxide semiconductor (CMOS) devices were substantially planar along the surface of the semiconductor substrate, the exception being the FET gate that was disposed over the top of the channel. Fins break from that paradigm by using a vertical structure in order to maximize surface area of the channel that is exposed to the gate. The gate controls the channel more strongly because it extends over three sides of the fin shaped channel, rather than only across the top of a more traditional planar channel. An example of a FinFET is shown inFIG. 1, which is a TEM micrograph of a prior art FET with six fins. Each pair of fins comprises an NFET and a PFET.
• One method to enhance fin performance is to fabricate fins in layers of disparate materials. One such exemplary device assigned to the assignee of this invention is described in U.S. Pat. No. 6,252,284 B1, “Planarized Silicon Fin Device”, which is directed to controlling short-channel effects.
• Often, one of the layers of a multi-layered planarized FET fin is strained silicon. It has been previously discovered that biaxially stretching the crystalline lattice structure of silicon can speed the flow of electrons through a transistor, thereby enhancing performance and decreasing power consumption. There is a natural tendency of atoms in layers of disparate materials to align with one another by stretching and/or compressing their mismatched lattice structures along a planar interface of the mismatch. Varying the respective thickness and chemical composition of the materials aids in controlling the extent of extension and compression in either material. As an example,FIG. 2adepicts a silicon lattice structure and a germanium lattice structure, not to scale, in their natural (unstretched) states. When silicon is disposed (or grown) on a thicker layer of germanium, as in the SiGe compound ofFIG. 2b, the lattice of silicon stretches while the lattice of germanium remains substantially unchanged, resulting in strained silicon. This result is achieved as the comparatively greater thickness of the germanium layer yields greater structural integrity and greater resistance to lattice compression. The actual process of lattice stretching may also, or alternatively, entail atoms of Si and Ge intermixed within a lattice structure. This integrates both atoms within a single layer, wherein the larger Ge atoms force the integrated lattice structure to stretch, as compared to a natural (unstretched) silicon lattice structure. The opposite effect from that shown inFIG. 2bcan also be realized, wherein a thinner layer of germanium exhibits lattice compression when bonded to a thicker layer of silicon. Strained silicon has been shown to enhance electron/hole mobility by up to 70% in NFETs, and up to 30% in PFETs. Other materials such as silicon germanium carbon may be used to form the crystalline heterojunction that enhances conductivity. There are several methods used to form strained layers, including chemical vapor deposition (CVD) and molecular beam epitaxy (MBE).
• Strained silicon disposed along a plane has been used in prior art FinFETs. However, current leakage continues to be a limiting factor in further scaling of FETs toward the atomic limit. Current leakage becomes an ascendant concern as miniaturization progresses because shorter length FET gates, which generally track the channel length, have less ability to control electric charge carriers (holes or electrons).
• The unintentional flow of charge carriers when the transistor is off is termed “current leakage.” Current leakage is the primary source of power consumed by an idle transistor. Current leakage may be classified into two types: MOS off current, wherein an unintended current passes through the channel despite the gate attempting to shut off current completely; and gate tunneling leakage current, wherein unintended current follows a parasitic pathway flowing into the channel, diffusions, or silicon body. As FET channel lengths continue to decrease, it is expected that gate tunneling leakage current will become a predominant concern for designers. The fin structure enhances gate control over the channel, but gate control over current is not absolute, even in prior art FinFETs. Compounding the current leakage problem, miniaturization enables ever-lower power levels that require more absolute gate control. Current leakage that escapes the gate's control is less distinguishable from intentional current, particularly at low current levels.
• What is needed is a FET that allows small scaling with enhanced performance for both NFETs and PFETs. Preferably, the FET should improve gate control in a fin structure that exhibits enhanced carrier transport properties.
SUMMARY OF THE PREFERRED EMBODIMENTS
• The foregoing and other problems are overcome, and other advantages are realized, in accordance with the presently preferred embodiments of these teachings. The present invention concerns a channel for electrically connecting a source and a drain of a field effect transistor (FET), commonly called a fin for a FinFET. The channel includes a channel core and a channel envelope. The channel core is coupled to a substrate such as a SIMOX wafer (a wafer with an upper region separated by implanted oxygen) or bonded wafer. The channel core defines a top surface that is spaced from the substrate, and opposed sidewall surfaces between the substrate and the top surface. The channel core is formed from a first semiconductor material defining a first lattice structure.
• The channel envelope is in contact with the opposed sidewall surfaces and the top surface of the channel core. The channel envelope is formed from a second semiconductor material defining a second lattice structure that differs from the first lattice structure. This difference in lattice structure provides enhanced electrical conductivity due to stretching or compressing of the lattice structure. Preferably, the two materials are silicon and a silicon-germanium compound.
• In another aspect of the present invention, the channel core defines a top and an adjoining side surface. In a preferred embodiment, the top surface is the horizontal top and the side surface is one of two upstanding sidewalls of the channel core, although two upstanding surfaces extending from the substrate and meeting at a peak, or two sidewalls joined by a third lateral surface may be used. The channel core comprises a first semiconductor material and the channel envelope comprises a second semiconductor material that differs from the first. At least one of the first or second semiconductor materials exhibits enhanced electrical conductivity due to one of a stretched or compressed lattice structure. The channel envelope is in contact with both the top and side surfaces, providing a larger area of interface between the disparate lattice materials, as compared to prior art channels that provide interface only along a top surface. The exemplary materials noted above are operable to provide the stretched or compressed lattice structure.
• The present invention also includes methods of making a FET channel. In one such method, a substrate is provided with an overlying layer of a first semiconductor material. A first channel core is defined from the overlying layer, such as by a mask and etch technique. The channel core defines a top surface spaced from the substrate and opposed first and second sidewalls between the substrate and the top surface. The method further includes disposing a layer of second semiconductor material to contact at least two of the surfaces (that is, at least two of the top surface, the first sidewall and the second sidewall). Electrical conductivity through the layer of second semiconductor material, which is different from the first, is enhanced by its contact with the channel core. The layer of second semiconductor material may be deposited via vacuum deposition, or may be grown on a carrier wafer and separated therefrom to contact the appropriate surfaces of the channel core.
• Another method of forming the channel includes providing a first layer of a semiconductor material over a substrate. This method includes defining a trench in the overlayer, and the overlayer is then divided into a first section and a second section by the trench. The trench may be filled with a buffer material such as TEOS (also known as tetraethoxysilane, tetraethylorthosilicate, tetraethelorthosilicate, and tetrethoxysilicide). A portion of the second section is then removed to leave a remaining layer of first semiconductor material that has a thickness less than a depth of the trench. This exposes a portion of the trench. The method then includes disposing a layer of a second semiconductor material over the remaining layer and adjacent to the trench. Exemplary semiconductor materials are discussed above. Preferably, the remaining layer is less than about 15 nm. Where the second semiconductor material is Si_xGe_1-x, the relative concentration of germanium may be chosen to promote thermal stability of the channel. Preferably, the germanium content is between about 8% and about 22%, and most preferably between about 10% and about 20%, and a low temperature chemical vapor deposition process is used to form the layer of second semi-conductor material.
• Further, disclosed herein is a channel for electrically connecting a source and a drain of a field effect transistor (FET) comprising: a channel core comprising a bottom surface coupled to a substrate and defining a top surface spaced from the substrate and opposed sidewall surfaces between the bottom surface and the top surface, wherein the channel core is formed from a first semiconductor material defining a first lattice structure; and a channel envelope in contact with at least one of the top surface, the bottom surface and one of the sidewall surfaces (or combinations thereof), wherein the channel envelope comprises a second semiconductor material comprising a strained lattice structure that differs from the first lattice structure, the channel envelope being electrically coupled to a gate electrode. Preferably, the channel is formed of a relaxed Si—Ge lattice, and the envelope is formed of a strained Si lattice.
• An exemplary circuit SRAM circuit using FinFET in accordance with the teachings herein is disclosed.
BRIEF DESCRIPTION OF THE DRAWINGS
• The foregoing and other aspects of these teachings are made more evident in the following Detailed Description of the Preferred Embodiments, when read in conjunction with the attached Drawing Figures, wherein:
• FIG. 1 is a TEM micrograph of a prior art FET with six fins;
• FIGS. 2aand2b, collectively referred to asFIG. 2, are prior art diagrams showing how the lattice structure of silicon is strained when disposed adjacent to germanium (FIG. 2b) as opposed to its natural state (FIG. 2a);
• FIG. 3 is a block diagram depicting a FinFET in which the present invention may be disposed;
• FIG. 4 is a sectional view ofFIG. 3 alongsection lines4′-4′;
• FIGS. 5a-5f, collectively referred to asFIG. 5, are enlarged cross sectional views depicting one method of making a channel according to the present invention, using masking and etching to deposit a channel envelope onto a channel core;
• FIGS. 6a-6h, collectively referred to asFIG. 6, are enlarged cross sectional views depicting an alternative method of making at least one channel using a carrier wafer and a handle wafer;
• FIGS. 7a-7f, collectively referred to asFIG. 7, are enlarged cross sectional views depicting an alternative method of making a channel for a PFET using a shallow trench;
• FIG. 8a-f, collectively referred to asFIG. 8, are perspective views of various embodiments of FinFET devices; and
• FIG. 9 depicts a SRAM circuit and is useful when discussing quantization.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
• FIGS. 1 and 2a-2bare described above and serve as context for understanding the present invention.FIG. 3 depicts in block diagram a FET10. As known in the art, asource12 and drain14 are connected electrically via achannel16, which is crossed by agate18. One ormore channels16 andgates18 may be present in a single FET10. As used herein, the length of thechannel16 is the distance from thesource12 to thedrain14 as depicted inFIG. 3.
• A sectional view of thechannel16 atsection line4′-4′ is depicted atFIG. 4. The FET10 is disposed on asubstrate20 such assilicon21 overlain with a layer of buriedoxide22. Achannel core24 is made from a first semiconductor material disposed over thesubstrate20. Thesubstrate20 is preferably silicon based, such as a SIMOX wafer, a bonded wafer, or CZ silicon (silicon wafer from the Czochralski process) as known in the art. Preferably, thechannel core24 is formed atop the buriedoxide layer22 that forms part of thesubstrate20. Thechannel core24 defines a bottom surface27 (shown as attached to the substrate20), atop surface26 spaced from thesubstrate20 andopposed sidewalls28,30 disposed between thesubstrate20 and thetop surface26. The first semiconductor material from which thechannel core24 is made defines a first natural crystalline lattice structure. As used herein, unless specifically noted otherwise, a natural lattice structure for a particular material (element or compound) refers to the lattice structure of the particular material in question in its natural state, that is, without having been stretched, compressed, or otherwise strained by external manipulation such as was previously described for strained silicon. A compound exhibiting a relaxed lattice structure, wherein the relaxation is due only to annealing or other processing that purposefully removes tensile or compressive forces that would otherwise be present, is not considered herein to be the natural state lattice structure.
• Achannel envelope32 is disposed to substantially cover that portion of thechannel core24 that is not in direct contact with thesubstrate20 when thechannel16 is viewed in cross section, at least at the portion of thechannel16 crossed by thegate18. Preferably, thechannel envelope32 is coupled to thetop surface26 and both opposed sidewalls28,30 of thechannel core24. Thechannel envelope32 is formed from a second semiconductor material that defines a second natural crystalline lattice structure that differs from the first lattice structure. The different natural lattice structures of the first and second semiconductor materials cause either tensile or compressive stresses at least at the boundary of thechannel core24 and thechannel envelope32. This mismatch of natural lattices, herein termed a “heterojunction,” facilitates carrier transport where the gate dielectric contacts the heterojunction stack. However, carrier transport is not necessarily maximized at the heterojunction; the main channel for carrier transport may alternately form within one or the other disparate materials, as known in the art, depending upon several factors. Thechannel core24 defines a core width w_cand a core height h_c. Similarly, thechannel envelope32 defines an envelope width W_eand an envelope height h_e. Any, or all, of these dimensions may be selected to optimize or otherwise control the combined surface area of the opposedsidewalls28,30 and thetop surface26. Where thechannel core24 and/orenvelope32 are not rectangular shaped (as primarily illustrated herein), such as when thechannel core24 is triangular and thechannel envelope32 is disposed over two sides of thechannel core24, the width and height are deemed to be averages for the particular component. Threshold voltage of a FET10 using theinventive channel16 described herein can be selected based on the particular materials of either or both of the first and second semiconductor materials, the presence of a dopant in one or the other, or thecore24 andenvelope32 dimensions.
• Preferably for a PFET, the first semiconductor material that comprises thechannel core24 is silicon, and the second semiconductor material that comprises thechannel envelope32 is a compound comprising silicon and germanium, such as Si_0.7Ge_0.3. For a NFET, the first semiconductor material that comprises thechannel core24 is preferably a compound comprising silicon and germanium that is processed to exhibit a relaxed lattice structure, and the second semiconductor material that comprises thechannel envelope32 is preferably silicon.
• Thegate18, which may be formed of poly or metal or other material as known in the art, is disposed over thechannel16 so as to contact thechannel envelope32 through adielectric layer33, also termed a gate dielectric, disposed about thechannel envelope32. Thedielectric layer33 may include an oxide, oxynitride, or rare earth oxide (e.g., hafnium oxide). Thechannel envelope32 is disposed so as to prevent substantial, and preferably complete, contact between thegate18 and thechannel core24.
• FIG. 5 illustrates a preferred method for making a FET10 in accordance with the teachings herein.FIGS. 5a-fdepict one embodiment, wherein both aPFET channel34 and aNFET channel36 are disposed on asingle substrate20.
• InFIG. 5a, asubstrate20 including a buried oxide layer (BOX)22 is provided and overlain with a layer of afirst semiconductor material38, such as silicon. The combination ofsubstrate layer20,BOX layer22, and layer offirst semiconductor material38 may be provided by a SIMOX wafer, may be a bonded wafer, or may be provided by a Czochralski process as known in the art. InFIG. 5b, afirst channel core40 and asecond channel core42 are defined from the layer offirst semiconductor material38 by masking and etching, or by other means known in the art. The length and/or width of thefirst channel core40 may differ from that of thesecond channel core42.FIG. 5cdepicts a mask44 (two masks shown) over thesecond channel core42 and over all areas except the immediate vicinity of thefirst channel core40. Etching or other known processes may be used to remove themask44 from contact with opposed sidewall surfaces of thefirst channel core40, or to prevent themask44 from ever contacting such sidewalls. Thefirst channel core40 will become thePFET channel34.
• FIG. 5ddepicts deposition of a layer ofsecond semiconductor material46 over the entire wafer. Preferably, the layer ofsecond semiconductor material46 is a compound of Si and Ge that is deposited via ultra-high vacuum chemical vapor deposition (UHVCVD), but other chemical deposition processes are also compatible. Most preferably, thelayer46 defines a thickness in the range of 5-10 nm, and the thickness of thelayer46 is determined by the concentration of germanium and criterion related to layer thermal stability for such germanium concentration. The entity ofFIG. 5dis planarized inFIG. 5eto remove that portion of the entire layer ofsecond semiconductor material46 that overlies themask44. The remainder of themask44 is removed inFIG. 5f; leaving thePFET channel34 and theadjacent NFET channel36. More recently developed methods deposit SiGe selectively on silicon. Using such a technique, aSiGe layer46 can be selectively deposited on an exposedsilicon channel core40 and any residual removed via a chemical etch. Where the layer offirst semiconductor material38 is silicon and the layer ofsecond semiconductor material46 is Si_xGe_1-z, thechannel envelope32 formed by the Si_xGe_1-xcompound is under compressive stress due to the smaller natural lattice structure of theunderlying Si layer38.
• An alternative method is depicted inFIG. 6. InFIG. 6a, ahandle wafer48 includes a buriedoxide layer22 as previously described, and an overlying layer offirst semiconductor material38, such as silicon.FIG. 6bdepicts use of masking and etching to define to define one ormore trenches47 adjacent to a remainingportion38aof the layer offirst semiconductor material38. Considering the entire expanse of thehandle wafer48, thetrenches47 and/or the remainingportions38aof the layer offirst semiconductor material38 may or may not be contiguous. Parallel to, or at a separate time from, the processing depicted inFIGS. 6a-6b, aseparate carrier wafer50 is overlain with a layer ofsecond semiconductor material46, such as strained or relaxed SiGe, inFIG. 6c. Thecarrier wafer50 withoverlayer46 is then subjected to ion implantation atFIG. 6c. As known in the art, implantation with ions52 such as hydrogen or boron ions allows theoverlayer46 to be annealed and separated from thecarrier wafer50 at a temperature generally less than about 600° C., while also providing high quality SiGe free from dislocations. Nearly complete (≧95%) strain relaxation of theSiGe layer46 can also be achieved. The above process is generally known in the art as a “smart-cut” technique, and is more particularly described by Lijuan Huang et al, ELECTRON ANDHOLEMOBILITYENHANCEMENT INSTRAINEDSOIBYWAFERBONDING, IEEE Transactions on Electron Devices., vol. 49, no. 9, Sep. 2002, pp 1566-1571, herein incorporated by reference. AtFIG. 6d, masking and etching are used to define one ormore islands49 from the second layer ofsemiconductor material46. The shapes of theislands49 on thecarrier wafer50 are complementary to the shapes of thetrenches47 on thehandle wafer48. Preferably, athin oxide layer51 is disposed over at least theislands49 to facilitate later bonding to thehandle wafer48. Once theislands49 are formed, further processing may be according to known finfet processing techniques, such as, for example, sidewall image transfer.
• FIG. 6edepicts the structure ofFIG. 6dbrought together with the structure ofFIG. 6b. InFIG. 6e, wherein thecarrier wafer50 is inverted over thehandle wafer48. Theislands49 andtrenches47 are precision aligned with one another, preferably accurate to within 0.25 microns. The anneal process described above for the smart-cut technique is employed to remove theislands49 from thecarrier wafer50. Theislands49 then lie within thetrenches47, and are bonded to thehandle wafer48 by thethin oxide layer51. Thecarrier wafer50 is removed, preferably for later reuse as a handle wafer after a high temperature anneal and polishing. In some embodiments, vertical etches through theoverlayer46 and selective ion implantation bounded by those vertical etches is used to combine more than one layer (more than one type of semiconductor material) into anisland49. Such anisland49 may be disposed to match a recessedtrench47 on thehandle wafer48 as previously described, or it may be deposited on an elevated surface of silicon or other semiconductor material on thehandle wafer48. These embodiments result in a heterojunction of more than two layers, as described more fully below.
• FIG. 6fshows smoothing of the exposed surfaces of theislands49 and of the remainingportions38a, where these surfaces are opposite thehandle wafer48. Smoothing and polishing is used to achieve a substantially uniform height.
• FIG. 6gdepicts a step wherein both theisland49 and theremainder38aare masked and etched, or otherwise processed, to define one or morePFET channel cores53 and preferably also one or moreNFET channel cores55. While it is preferable to fabricate the PFET and NFET channels simultaneously as described herein, the present invention does not so require. As shown inFIG. 6h, an epitaxial layer of a semiconductor material different from the first semiconductor material is grown or disposed on thePFET channel cores53 to formchannel envelopes32, creating a strained silicon or other hetero-layer.
• The material of thechannel envelope32 may be strained or unstrained, depending upon the desired properties of theresultant channel16. Maximum lattice discontinuity occurs along the lines defined by either of the opposed sidewall surfaces28,30 and thebottom surface27/top surface26 of thechannel core24.
• Preferably, thePFET channel34 orNFET channel36 are characterized by a width of about 100-150 Å and a height of about 500-600 Å. These dimensions may change with scaling. Preferably, the strained lattice is disposed along both thesidewalls28,30 and the top26 of thePFET channel34 orNFET channel36.
• One particular application in which the carrier pathway along the strained sidewall can be exploited is in a FET10 in a static random access memory (SRAM). SRAM is a type of memory that does not need to be refreshed like dynamic RAM (DRAM), so SRAM is generally much faster (typically about 10 ns for SRAM versus about 60 ns for DRAM) and more reliable. In addition, the cycle time (a measurement of how quickly two back-to-back accesses of a memory chip can be made) of SRAM is much shorter than that of DRAM because it does not need to pause between accesses. The design of SRAMs generally assumes FETs in several fixed sizes (quantized). However, the width of thechannel16 is critical for stability in SRAM, and thereby imposes a trade off between size and stability. Rather than select a FET with a quantized channel width that may be larger in the x-y plane (the plane ofFIG. 3) than desirable, a FET10 employing a fin according to the present invention may be used to modulate the height of the channel to enable a smaller FET10 that still ensures SRAM stability. During fabrication, the starting silicon would be at least as high as the highest estimated fin. Selective masking and etching of certain devices within a cell tailors fin height to achieve the desired SRAM stability. Preferably, a slow etch is used to ensure uniformity and fine control over fin height. Additionally, effective channel width can be adjusted by using multiple fins, preferably in a side-by-side relation.
• FIG. 7 is a series of block diagrams showing process steps in forming a PFET.FIG. 7 depicts a shallow trench isolation (STI) that is imposed between a first semiconductor material, such as SiGe, and a second semiconductor material, such as Si. AtFIG. 7a, asubstrate20 supports a buried oxide layer (BOX)22, which is overlain with a layer of afirst semiconductor material38, such as silicon. The substrate/BOX/overlayer combination70 may be a SIMOX wafer, a bonded wafer, or a CZ wafer as previously described and known in the art. AtFIG. 7b, atrench58 is formed and filled with an oxide or other insulator such as TEOS, as is known in the art. It is noted that the trenching and filling represented byFIG. 7bis performed in several steps, and thetrench58 divides the layer offirst semiconductor material38 into afirst region62 and aPFET region64. Alternatively, thetrench58 may be etched, but not filled, without departing from the advantages of the present invention. AtFIG. 7c, amask layer44 is selectively disposed over thetrench58 and thefirst region62 of the first semiconductor material. InFIG. 7d, thePFET region64 is then etched back to athin layer60, preferably to a thickness on the order of 10 nm. InFIG. 7e, either before or after themask44 is removed, a layer ofsecond semiconductor material46 is disposed over thethin layer60. As shown inFIG. 7e, thethin layer60 includes ahorizontal surface67, and a substantiallyvertical surface68.
• Preferably, thesecond semiconductor material46 is SiGe of a moderate concentration. For thermally stable fins, 10%-20% Ge concentration is preferable when the thickness of thelayer46 is between about 10-30 nm. It has been found thatthicker layers46 of SiGe, and/or higher concentrations of Ge, will be metastable and may require further thermal processing, though not annealing. Any embodiment of the present invention preferably includes an enveloping layer of gate dielectric or gate oxide. Achannel16 according to the present invention may be part of a FinFET10, as inFIG. 3, and the FinFET10 may preferably be part of an integrated circuit100. Thechannel16 may include achannel core24 that has substantiallyupstanding sidewalls28,30, atop surface26, and abottom surface27, as inFIG. 4, or may be fabricated to define only ahorizontal surface67 and substantiallyvertical surfaces68 as inFIG. 7e. As a further alternative, shown inFIG. 7f, thechannel16 may have twosidewall surfaces74 that extend from thesubstrate22 and join at an acute or roundedpeak71 so as to define a substantially triangular cross section75, as opposed to the rectangular cross sections illustrated herein, such as the cross section of thethin layer60. It is anticipated that a channel defining a non-orthogonal cross section would be optimized using strained silicon on the top and in contact with the gate oxide. Thechannel envelope32 may contact any two of thesurfaces26,27,28,30 of thechannel core24, or allsuch surfaces26,28,30 not contacting thesubstrate20. In any embodiment above, a broader surface of the strained lattice structures that carry charge (electrons or holes) is better exposed to thegate18 as compared to prior art FinFETs, enabling greater control by thegate18 and more efficient current transport.
• FIG. 8 represents various additional embodiments of FinFET structures in accordance with the teachings herein. In the embodiments shown inFIG. 8, the structures include a relaxed Si—Ge layer with an overlying strained Si layer. This combination provides for improved electron-hole mobility.
• Thechannel16 shown in FIGS.8A-B, is formed of achannel core24 and is overlain with thechannel envelope32, as is shown inFIG. 4. As shown inFIG. 8, thesubstrate20 includes the buriedoxide layer22, and theSi layer21. Each structure depicted inFIG. 8 includes thesource12, adrain14, thegate18, and thechannel16. Also shown is adielectric layer33 disposed between thechannel16 and the various gates. As thechannel16 is internal to the structures shown in FIGS.8C-E, and is therefore not visible, thechannel16 is only depicted in FIGS.8A-B and8F. Regardless, each of the embodiments ofFIGS. 8A-8F may include thechannel core24 andchannel envelope32 as previously described.
• InFIG. 8A, asingle gate FinFET90 is shown. In this embodiment, thechannel16 includes a relaxed Si—Ge layer, and a strained Si layer. Athin oxide layer33 is disposed between thechannel16 and asingle gate95.FIG. 8B represents adouble gate91 embodiment of a FinFET. In addition to thesingle gate95 shown inFIG. 8A, asecond gate96 is presented. In this embodiment, thesecond gate96, is buried within a trench in the buriedoxide layer22.FIG. 8C represents atriple gate FinFET92, wherein atriple gate97 is formed over thethin oxide layer33, which is disposed over the channel16 (shown in FIGS.8A-B).FIG. 8D represents aquadruple gate FinFET93, wherein aquadruple gate98 effectively surrounds thechannel16. InFIG. 8D, a portion of thequadruple gate98 is submerged in the buriedoxide layer22. InFIG. 8E, aFinFET94 having aPi gate99 is shown. ThePi gate99 extends downward, and at least partially into the buriedoxide layer22.
• FIG. 8F is another embodiment of adouble gate FinFET87, defining afirst gate88 and asecond gate89. The FinFET ofFIG. 8F may be formed, for example, by polishing the FinFET ofFIG. 8C until the portion of thetriple gate97 ofFIG. 8C that overlies thechannel16 is removed, leaving the twogates88,89 ofFIG. 8F. Preferably, the channel includes strained silicon germanium. One advantage in the gate arrangement ofFIG. 8F is that eachgate88,89 may be independently controlled so that, for example, thefirst gate88 is a typical FinFET gate and thesecond gate89 may apply variable voltage. In this manner, thesecond gate89 is a back gate to control a depletion region, when current between thesource12 and drain14 is switched between on and off. Such asecond gate89 better controls leakages and performance of thefirst gate88 by use of the variable voltage; so that only the minimum voltage necessary to control leakage current is applied.
• Depositing the strained Si layer on thequadruple gate FinFET93 and thePi gate FinFET94, wherein theFinFET93,94 are PFET (and not NFET), improves mobility within the PFET, while preserving beneficial aspects of the PFET and the NFET. Further it is noted that depositing strained Si layers in the PFET in the [110] direction also improves the mobility. The [110] direction is at a 45° angle to the crystalline axes.
• FIG. 9 depicts one embodiment of a six transistor SRAM unit cell. For SRAM FinFET quantization, preferably, PFET quantization is carried out on the smallest device in a six-transistor conventional SRAM (e.g., NFET pass gate77 (Pg) and pull-up PFET76 (PL) is smaller in size). Thus, fin height quantization can be based on individual devices, or bucketing pass gate NFET and pull up PFET into one quanta height and larger NFET78 (Cc) into another. Alternatively, a smaller fin height, which is employed in multiples of NFET and PFET devices of the embodiment shown in FIGS.8A-F, may be used. This quantization will help to make a robust and stable cell.
• While described in the context of presently preferred embodiments, various modifications of and alterations to the foregoing embodiments can be made with normal skill in the art, and all such modifications and alterations remain within the scope of this invention and the ensuing claims. Examples herein are considered to be illustrative and not exhaustive of the teachings of this invention.

Claims (14)

1-18. (canceled)

19. A Field Effect Transistor disposed on a substrate comprising:

a source;

a drain;

a fin connecting the source to the drain and defining a channel core and a channel envelope;

a gate coupled through a gate dielectric to at least two surfaces of the fin, wherein the channel core defines at least two surfaces extending from the substrate and comprises a first semiconductor material, and the channel envelope is in contact with the at least two surfaces and comprises a second semiconductor material, and wherein at least one of the first or second semiconductor material exhibits one of a stretched and a compressed lattice structure.

20. An integrated circuit comprising at least one field effect transistor according toclaim 19.

21. A field effect transistor (FET) comprising:

a source, a drain, a channel, a gate electrode, and a gate dielectric,

wherein the channel comprises a channel core defining a bottom surface and a top surface spaced from the bottom surface by laterally opposed sidewall surfaces disposed between the bottom surface and the top surface,

wherein the channel core comprises a first semiconductor material defining a first lattice structure; the channel further comprising a channel envelope in contact with at least the top surface of the channel core,

wherein the channel envelope comprises a second semiconductor material defining a second lattice structure that differs from the first lattice structure,

wherein one of the first and second lattice structures is one of stretched and compressed; and,

wherein the gate electrode is coupled through the gate dielectric to the channel envelope only at a top surface of the channel envelope that is opposed to the top surface of the channel core.

22. The FET ofclaim 21, wherein the first lattice structure is relaxed relative to the second lattice structure.

23. A field effect transistor (FET) comprising:

a source, a drain, a channel, a gate electrode, and a gate dielectric;

wherein the channel comprises a channel core defining a bottom surface and a top surface spaced from the bottom surface by laterally opposed sidewall surfaces disposed between the bottom surface and the top surface,

wherein the channel core comprises a first semiconductor material defining a first lattice structure; the channel further comprising a channel envelope in contact with at least the top surface of the channel core,

wherein the channel envelope comprises a second semiconductor material defining a second lattice structure that differs from the first lattice structure; and,

wherein the gate electrode is coupled through the gate dielectric to the channel envelope only at surfaces of the channel envelope that are opposed to the top and bottom surfaces of the channel core.

24. The FET ofclaim 23, wherein the first lattice structure is relaxed relative to the second lattice structure.

25. A field effect transistor (FET) comprising:

a source, a drain, a channel, a gate electrode, and a gate dielectric;

wherein the channel comprises a channel core defining a bottom surface and a top surface spaced from the bottom surface by laterally opposed sidewall surfaces disposed between the bottom surface and the top surface,

wherein the channel core comprises a first semiconductor material defining a first lattice structure; the channel further comprising a channel envelope in contact with at least the top surface and the sidewall surfaces,

wherein the channel envelope comprises a second semiconductor material defining a second lattice structure that differs from the first lattice structure; and,

wherein the gate electrode is coupled through the gate dielectric to the channel envelope only at surfaces of the channel envelope that are opposed to the top and sidewall surfaces of the channel core.

26. The FET ofclaim 25, wherein the first lattice structure is relaxed relative to the second lattice structure.

27. A field effect transistor (FET) attached to a substrate comprising:

a source, a drain, a channel, a gate electrode, and a gate dielectric;

wherein the channel comprises a channel core defining a bottom surface and a top surface spaced from the bottom surface by laterally opposed sidewall surfaces disposed between the bottom surface and the top surface,

wherein the channel core comprises a first semiconductor material defining a first lattice structure; the channel further comprising a channel envelope in contact with at least the top surface,

wherein the channel envelope comprises a second semiconductor material defining a second lattice structure that differs from the first lattice structure; and,

wherein the gate electrode is coupled through the gate dielectric to the channel envelope at surfaces opposed to the top surface, sidewall surfaces and bottom surface of the channel core.

28. The FET ofclaim 27, wherein the first lattice structure is relaxed relative to the second lattice structure.

29. A field effect transistor (FET) attached to a substrate comprising:

a source, a drain, a channel, a gate electrode, and a gate dielectric;

wherein the channel comprises a channel core defining a bottom surface and a top surface spaced from the bottom surface by laterally opposed sidewall surfaces disposed between the bottom surface and the top surface,

wherein the channel core comprises a first semiconductor material defining a first lattice structure; the channel further comprising a channel envelope in contact with at least the top surface,

wherein the channel envelope comprises a second semiconductor material defining a second lattice structure that differs from the first lattice structure; and,

wherein the gate electrode is coupled through the gate dielectric to the channel envelope at surfaces of the channel envelope that are opposed to the top, sidewall, and bottom surfaces of the channel core.

30. The FET ofclaim 29, wherein the first lattice structure is relaxed relative to the second lattice structure.

31-35. (canceled)

Images