The present invention relates generally to semiconductor devices, and more particularly, to complementary field-effect transistors and methods of manufacture.
Size reduction of metal-oxide-semiconductor field-effect transistors (MOSFET), including reduction of the gate length and gate oxide thickness, has enabled the continued improvement in speed, performance, density, and cost per unit function of integrated circuits over the past few decades. To further enhance transistor performance, strain may be introduced in the transistor channel for improving carrier mobilities. Generally, it is desirable to induce a tensile strain in the n-channel of an NMOS transistor in the source-to-drain direction, and to induce a compressive strain in the p-channel of a PMOS transistor in the source-to-drain direction. There are several existing approaches of introducing strain in the transistor channel region.
In one conventional approach, as described in a paper by J. Welser et al., entitled “NMOS and PMOS Transistors Fabricated in Strained Silicon/Relaxed Silicon-Germanium Structures,” published at the December 1992 International Electron Devices Meeting held in San Francisco, Calif., pp. 1000-1002 and incorporated herein by reference, a relaxed silicon germanium (SiGe) buffer layer is provided beneath the channel region. The relaxed SiGe layer has a larger lattice constant compared to relaxed Si, and the layer of Si grown on the relaxed SiGe will have its lattice stretched in the lateral direction, i.e., it will be under biaxial tensile strain. Therefore, a transistor formed on the epitaxial strained silicon layer will have a channel region that is under biaxial tensile strain. In this approach, the relaxed SiGe buffer layer can be thought of as a stressor that introduces strain in the channel region. The stressor, in this case, is placed below the transistor channel region.
This approach is very expensive because a SiGe buffer layer with thickness in the order of micrometers has to be grown. Numerous dislocations in the relaxed SiGe buffer layer exist and some of these dislocations propagate to the strained silicon layer, resulting in a substrate with high-defect density. Thus, this approach has limitations that are related to cost and fundamental material properties.
In another approach, strain in the channel is introduced after the transistor is formed. In this approach, a high stress film is formed over a completed transistor structure formed in a silicon substrate. The high stress film or stressor exerts significant influence on the channel, modifying the silicon lattice spacing in the channel region, and thus introducing strain in the channel region. In this case, the stressor is placed above the completed transistor structure. This scheme is described in detail in a paper by A. Shimizu et al., entitled “Local mechanical stress control (LMC): a new technique for CMOS performance enhancement,” published in pp. 433-436 of the Digest of Technical Papers of the 2001 International Electron Device Meeting, which is incorporated herein by reference.
The strain contributed by the high-stress film is believed to be uniaxial in nature with a direction parallel to the source-to-drain direction. However, uniaxial tensile strain degrades the hole mobility while uniaxial compressive strain degrades the electron mobility. Ion implantation of germanium can be used to selectively relax the strain so that the hole or electron mobility is not degraded, but this is difficult to implement due to the close proximity of the n-channel and p-channel transistors. Therefore, there is a need for an efficient and cost-effective method to induce strain such that the performance characteristics of transistors are enhanced.
These and other problems are generally reduced, solved or circumvented, and technical advantages are generally achieved, by embodiments of the present invention, which provides a strained semiconductor device to improve the operating characteristics of the semiconductor device and a method of manufacture.
In one embodiment of the present invention, a semiconductor device is provided on a substrate wherein the current flow is substantially along the <100> crystal orientation of the substrate. A silicide region is formed substantially along the surface of the substrate beneath the spacers.
In another embodiment of the present invention, a semiconductor device is provided having a first transistor and a second transistor such that the current flow is substantially along a <100> crystal orientation of the substrate. The spacers of the first transistor are larger than the spacers of the second transistor. At least a portion of at least one of the spacers of the first transistor and the spacers of the second transistor is formed over a silicided region.
In still yet another embodiment of the present invention, a method is provided to form a semiconductor device. The method includes the steps of providing a substrate, forming on the substrate a transistor having spacers along the side of a gate electrode, and forming a silicided region substantially along the surface of the substrate such that at least a portion of the silicided region extends beneath the spacers. The transistor is formed such that current flow through the source/drain region of the transistor is substantially along the <100> crystal orientation of the substrate.
In yet another embodiment of the present invention, a method is provided to form a semiconductor device having a first transistor and a second transistor. The method includes the steps of providing a substrate, forming on the substrate a first transistor having a spacer formed along the side of a first gate electrode such that current flow through a source/drain region of the transistor is substantially along the <100> crystal orientation of the substrate; forming on the substrate a second transistor having a second spacers formed along the side of a second gate electrode, the second spacer being smaller than the first spacer, and forming a silicided region substantially along the surface of the substrate such that at least a portion of the silicided region extends beneath at least one of the first spacer and the second spacer.
For a more complete understanding of the present invention, and the advantages thereof, reference is now made to the following descriptions taken in conjunction with the accompanying drawings, in which:
a-1e are cross-section views of a wafer after various process steps in accordance with one embodiment of the present invention;
a-3d illustrate structures of a semiconductor die of a wafer in accordance with one embodiment of the present invention.
The making and using of the presently preferred embodiments are discussed in detail below. It should be appreciated, however, that the present invention provides many applicable inventive concepts that can be embodied in a wide variety of specific contexts. The specific embodiments discussed are merely illustrative of specific ways to make and use the invention, and do not limit the scope of the invention.
a-1e illustrate a first method embodiment for fabricating strained channel regions of transistors in a semiconductor chip. Embodiments of the present invention illustrated herein may be used in a variety of circuits. For example, embodiments of the present invention may be utilized to form circuits for NOR gates, logic gates, inverters, XOR gates, NAND gates, PMOS transistors for pull-up transistor, NMOS transistor for pull-down transistor, and the like.
Referring first to
In another embodiment, the substrate 110 is a multi-layered structure, wherein the layers have different natural lattice constants. One such type of substrate is a graded silicon-germanium (SiGe) substrate with a strained silicon surface layer. Generally, a graded SiGe layer is formed on a bulk silicon layer, and a relaxed SiGe layer overlies the graded SiGe layer. The relaxed Si1-xGex layer, wherein x is preferably 0.1<x<0.5 has a larger natural lattice constant than that of silicon. Relaxed crystalline silicon is said to be lattice-mismatched with respect to relaxed crystalline SiGe due to the difference in their lattice constants. As a result, a thin layer of silicon that is epitaxially grown on the relaxed SiGe layer will be under biaxial tensile strain because the lattice of the thin layer of silicon is forced to align to the lattice of the relaxed crystalline SiGe layer. In this embodiment, it is preferred that the strained silicon has a crystal orientation of <100>.
Another multi-layered substrate comprises a first layer having a first natural lattice constant. A second layer having a second natural lattice constant is formed on the first layer. The first layer may be formed of an alloy semiconductor, an element semiconductor, a compound semiconductor, or the like. For example, the first layer may be formed of SiGe, and the second layer may be formed of Si or a layer containing Ge/C. In these multi-layered substrates, the surface roughness of the strained surface layer is less than about 1 nm.
Isolation regions, such as shallow trench isolations 112, may be formed in the substrate 110. The STIs 112 are known and used in the art. It is understood that other isolation structures, such as field oxide (e.g., formed by the local oxidation of silicon) may be used. It should also be noted that the STI 112 may induce a tensile stress on the wafer 100.
A gate dielectric layer 114 and a gate electrode 116 are formed and patterned as is known in the art on the substrate 110. The gate dielectric 114 is preferably a high-K dielectric material, such as silicon oxide, silicon oxynitride, silicon nitride, an oxide, a nitrogen-containing oxide, a combination thereof, or the like. Preferably, the gate dielectric 114 has a relative permittivity value greater than about 4. Other examples of such materials include aluminum oxide, lanthanum oxide, hafnium oxide, zirconium oxide, hafnium oxynitride, or combinations thereof.
In the preferred embodiment in which the gate dielectric layer 114 comprises an oxide layer, the gate dielectric layer 114 may be formed by any oxidation process, such as wet or dry thermal oxidation in an ambient comprising an oxide, H2O, NO, or a combination thereof, or by chemical vapor deposition (CVD) techniques using is tetra-ethyl-ortho-silicate (TEOS) and oxygen as a precursor. In the preferred embodiment, the gate dielectric layer 114 is about 8 Å to about 50 Å in thickness, but more preferably about 16 Å in thickness.
The gate electrode 116 preferably comprises a conductive material, such as a metal (e.g., tantalum, titanium, molybdenum, tungsten, platinum, aluminum, hafnium, ruthenium), a metal silicide (e.g., titanium silicide, cobalt silicide, nickel silicide, tantalum silicide), a metal nitride (e.g., titanium nitride, tantalum nitride), doped poly-crystalline silicon, other conductive materials, or a combination thereof. In one example, amorphous silicon is deposited and recrystallized to create poly-crystalline silicon (poly-silicon). In the preferred embodiment in which the gate electrode is poly-silicon, the gate electrode 116 may be formed by depositing doped or undoped poly-silicon by low-pressure chemical vapor deposition (LPCVD) to a thickness in the range of about 400 Å to about 2500 Å, but more preferably about 1500 Å.
The gate dielectric 114 and the gate electrode 116 may be patterned by photolithography techniques as is known in the art. Generally, photolithography involves depositing a photoresist material, which is then masked, exposed, and developed. After the photoresist mask is patterned, an etching process may be performed to remove unwanted portions of the gate dielectric material and the gate electrode material to form the gate dielectric 114 and the gate electrode 116 as illustrated in
In one embodiment, the width of the gate electrode is different for PMOS devices and NMOS devices. In one such embodiment, the ratio of the gate width of the PMOS devices to the gate width of the NMOS devices is about equal to the ratio of electron mobility to hole mobility in bulk silicon or strained silicon. In another embodiment, the ratio of the gate width of the PMOS devices to the gate width of the NMOS devices is about equal to the square root of the ratio of electron mobility to hole mobility in bulk silicon or strained silicon.
Source/drain regions 118 are a lightly-doped drain (LDD) formed by ion implantation. The source/drain regions 118 may be implanted with an n-type dopant, such as phosphorous, nitrogen, arsenic, antimony, or the like, to fabricate NMOS devices or may be implanted with a p-type dopant, such as boron, aluminum, indium, and the like, to fabricate PMOS devices. Optionally, NMOS devices may be fabricated on the same chip as PMOS devices. In this optional embodiment, it may be necessary to utilize multiple masking and ion implant steps as is known in the art such that only specific areas are implanted with n-type and/or p-type ions.
Optionally, an epitaxial silicon may be grown in the source/drain regions 118. For example, an epitaxial silicon layer about 200 Å may be grown on the wafer. In this situation, the LDD is preferably less than 200 Å above the surface of the substrate to 50 Å below the surface of the substrate.
It has also been found that orienting the transistors, or other semiconductor devices, such that the current flow will be substantially along the <100> crystal orientation of the substrate improves the hole and electron mobility. Accordingly, it is preferred that the masks used to pattern the source/drain regions 118 are such that the current flow through the source/drain regions 118 is substantially along the <100> crystal orientation of the substrate.
Referring now to
The spacer 122, which forms a spacer for the second ion implant, preferably comprises silicon nitride (Si3N4), or a nitrogen containing layer other than Si3N4, such as SixNy, silicon oxynitride SiOxNy, silicon oxime SiOxNy:Hz, or a combination thereof. In a preferred embodiment, the spacer 122 is formed from a layer comprising Si3N4 that has been formed using chemical vapor deposition (CVD) techniques using silane and ammonia as precursor gases.
In the preferred embodiment, the ratio of the width of the spacer 122 to the thickness of the dielectric liner 120 is less than 5, and more preferably, less than 3. Furthermore, it should noted that the width of the spacer may vary with the type of device. For example, it has been found that I/O devices may benefit from a larger spacer due to the amount of current the device is expected to handle. PMOS devices may also benefit from larger spacers. In particular, it has been found that larger spacers on PMOS devices help reduce tensile stress on the p-channel. In this example, it is preferred that the larger spacers are at least about 10% larger than the smaller spacers. To fabricate spacers of varying widths, it may be necessary to incorporate additional masking, deposition, and etching steps.
The spacer 122 may be patterned by performing an isotropic or anisotropic etch process. The preferred isotropic etch process uses a solution of phosphoric acid (H3PO4) wherein the dielectric liner 120 acts an etch stop. Because the thickness of the layer of Si3N4 is greater in the regions adjacent to the gate electrode 116, the isotropic etch removes the Si3N4 material on top of the gate electrode 116 and the areas of substrate 110 not immediately adjacent to the gate electrode 116, leaving the spacer 122 as illustrated in
The dielectric liner 120 may be patterned, for example, by performing an isotropic wet etch process using a solution of hydrofluoric acid. Another etchant that may be used is a mixture of concentrated sulphuric acid and hydrogen peroxide, commonly referred to as piranha solution. A phosphoric acid solution of phosphoric acid (H3PO4) and water (H2O) may also be used to pattern the dielectric liner 120.
It should be noted that it is desirable to etch a portion of the dielectric liner 120 from beneath the spacer 122 as illustrated in
It should also be noted that the etching process to create the recess may also remove a portion of the dielectric liner 120 and the gate electrode 116 along the top of the transistors 102 and 104. If desired, a mask may be applied to the top of the transistors 102 and 104 to prevent a recess in the top of the transistors 102 and 104.
After forming the spacers 122, a second ion implant may be performed in the source/drain regions 118 as is known in the art. The source/drain regions 118 may be implanted with an n-type dopant, such as phosphorous, nitrogen, arsenic, antimony, or the like, to fabricate NMOS devices or may be implanted with a p-type dopant, such as boron, aluminum, indium, or the like, to fabricate PMOS devices. Optionally, NMOS devices and PMOS devices may be fabricated on the same chip. In this optional embodiment, it may be necessary to utilize multiple masking and ion implant steps such that only specific areas are implanted with n-type and/or p-type ions. Furthermore, additional ion implants may be performed to create differing graded junction configurations.
Referring now to
Prior to depositing the metal layer, it is preferred to clean the wafer to remove native oxide. Cleaning solutions that may be used includes hydrofluoric acid, sulphuric acid, hydrogen peroxide, NH4OH solution, a combination thereof, or the like.
The silicidation process may be performed by annealing, causing the metal layer to selectively react with exposed silicon regions (e.g., the source/drain regions 118) and the poly-semiconductor regions (e.g., the gate electrodes 116) to form a silicide. In the preferred embodiment in which the metal layer comprises nickel, platinum, palladium, or cobalt, the silicidation process forms nickel silicide, platinum silicide, palladium silicide, or cobalt silicide, respectively. The excess material of the metal layer may be removed, for example, by performing a wet dip in a solution of sulfuric acid, HCl, H2O2, hydrogen peroxide, NH4OH, H3PO4 or the like.
It should be noted that the silicided portion of the silicon extends beneath the spacer 122 due to silicide cap layer thickness or the recess etched into the dielectric liner 120. It has been found that the silicide formed in this manner increases the tensile stress in the channel area of the transistors 102 and 104. As noted above, the tensile stress can enhance the flow of current in the channel area of the transistors, particularly with NMOS transistors.
In an alternative embodiment, one or more of the steps of etching the recess in the dielectric liner 120 and performing the silicidation process are only performed on NMOS devices, thereby enhancing the electron mobility while not affecting the hole mobility of the PMOS devices. In these alternative embodiments, it may be necessary to form a mask over the PMOS devices while performing one or both of these process steps.
Referring now to
In another embodiment, PMOS devices may have a compressive-stress film, or no stress film, while NMOS devices have a tensile-stress film. The compressive-stress film results in a compressive strain in the channel region of the p-channel devices in the source-to-drain direction to enhance hole mobility. The process of forming a compressive-stress film on PMOS devices and tensile-stress film on NMOS devices is further described below with reference to co-pending patent application Ser. No. 10/639,170, filed Aug. 12, 2003, entitled, “Strained Channel Complementary Field-Effect Transistors and Methods of Manufacture,” which is incorporated herein by reference.
Referring now to
Thereafter, standard processing techniques may be used to complete fabrication of the semiconductor device, including process steps such as forming metal lines and layers, forming vias and plugs, packaging, and the like.
a and 3b illustrate a layout or shape of semiconductor die 310 that has been sawn off from a wafer having either <100> or <110> notch orientations and may be used in accordance with embodiments of the present invention. It has been found that a wafer 200 having a silicon crystal notch orientation of <100> is more brittle while performing the wafer or die 310 sawing process, substantially along with the die-saw edge, than a wafer having a silicon crystal notch orientation of <110>, which is commonly used for fabricating semiconductor devices. In addition, it has been found that the presence of low-k dielectric may greatly deteriorate those inter-metal dielectrics 332 and/or substrate 110 chipping (or cracking) defect counts during the die 310 sawing process. The low-k materials, such as fluorine-containing or carbon-containing dielectric layers, are being commonly used for inter-metal dielectrics 332, characterized substantially with both lower dielectric constant and lower mechanical strength than conventional silicon oxide dielectric layer. Another finding of this invention is that, regardless the wafer notch orientation is <100> or <110>, the most vulnerable chipping defective locations are the neighboring strip of areas, running substantially in parallel to the die-saw edge 328 longitudinal directions, within about 300 μm to about 500 μm distance from the four die corners 334 of top view of the die 310.
As a result, it is desirable to fabricate the die 310 such that the clearance areas, preferably including 314-a, -b shown in
For a device having 3 to 9 or more layers of multi-layer metal process technology, it has been found that the top metal 336 layer experiences the largest portion of stress amount resulted from the thermal/mechanical combinational effect of those materials including the substrate 110, the top protecting/passivation dielectric 330, inter-metal dielectrics 332, top metal layer 336, inter metal layers 338, the organic/inorganic assembly fillers and the molding compound over the top protecting dielectric 330. For a semiconductor manufacturing process with a fewer number of metal layers, such as less than 3 to 6 metal layers, the preferred clearance area 314 may be a strip of area with about 0.5 μm to about 10 μm in width without covering of the top metal layer or any of the inter-metal layers. In this manner, it may not only help to improve the mechanical stress induced substrate/dielectric chipping (or cracking) problems during the die-saw process, but it may also help to act as a thermal/mechanical stress buffer area to improve the potential dielectrics cracking (or delaminating) reliability problems during the die 310 assembly or packaging process. For semiconductor manufacturing process with more metal layers, such as more than 6 to 9 metal layers, then the preferable clearance area 314 may be a strip of area about 1 μm to about 20 μm in width, without sacrificing a plenty of die area, in order to cope with the getting larger thermal/mechanical stress induced by the ever thicker metal and dielectric stacked layers.
c and 3d are cross-section views of the clearance area 314 illustrating examples of configurations that may be used in accordance with an embodiment of the present invention. Specifically,
While this invention has been described with reference to illustrative embodiments, this description is not intended to be construed in a limiting sense. Various modifications and combinations of the illustrative embodiments, as well as other embodiments of the invention, will be apparent to persons skilled in the art upon reference to the description. It is therefore intended that the appended claims encompass any such modifications or embodiments.
This application claims the benefit of U.S. Provisional Application No. 60/526,133 filed on Dec. 1, 2003, entitled Complementary Field-Effect Transistors and Method of Manufacture, which application is hereby incorporated herein by reference.
Number | Date | Country | |
---|---|---|---|
60526133 | Dec 2003 | US |