This application is directed, in general, to semiconductor integrated circuits and to methods for their fabrication.
Semiconductor integrated circuits are often fabricated by creating raised topographic features upon a substrate. Then a dopant species is introduced into the substrate with the raised topographic features serving to mask a portion of the substrate. For example, in the fabrication of semiconductor integrated circuits using field effect transistors (FETS), a gate stack (typically including a gate oxide with an overlying body of polysilicon) is formed upon a silicon substrate. Then a dopant species is introduced into a silicon substrate by diffusion or ion implantation to create the source and drain regions on both sides of the gate stack. As the dopant species is introduced, the gate stack serves as a self-aligned mask shielding the channel under the gate from the dopant species.
Of course, during the above-described dopant introduction, the gate stack is subjected to the same environment as the to-be-formed source and drain regions are subjected. For example, if ion implantation techniques are employed to create the source and drain, the gate stack is exposed to ion implantation of the same dopant species as the to-be-formed source and drain regions.
In the past, exposure of the gate stack to ion implantation species has not generally created a problem because the implanted species have been completely absorbed by the gate polysilicon. However, as integrated circuit geometries have continued to shrink, the thickness of gate stacks has also shrunk. If the thickness of the gate is too low relative to the implant dose energy, the implanted species may penetrate through the gate.
Penetration of the implanted species through the gate is often termed “channeling.” If the energy of the implanted species is great enough and the polysilicon grains are oriented with the direction of the implant species, then the range of implanted species becomes greater than the thickness of the gate stack, and the implanted species may arrive at the gate oxide-silicon interface with enough energy to penetrate into or perhaps through the gate oxide. Thus, channeling depends upon the size and orientation of the polysilicon, as well as the energy of the implant species. A single large grain, if oriented parallel to the implant direction, can permit channeling.
When channeling occurs, the silicon surface beneath the gate may be inverted, leading to transistor leakage and/or shifts in the threshold voltage. Another adverse affect of channeling is gate oxide degradation. In addition, channeling may cause flat band voltage shifts in polysilicon capacitors in the same integrated circuit. Heretofore, the channeling problem has not posed a serious obstacle to integrated circuit designers because gate stacks in previous generation integrated circuits have been thick enough to prevent channeling.
These problems are alleviated by the present invention which illustratively includes: forming a dielectric layer upon a substrate; forming a conductive layer upon the dielectric layer; forming a material layer overlying the conductive layer; forming a patterned resist upon the material layer; at least partially etching the material layer to form a raised feature; removing the resist; using the raised feature as a mask, anisotropically etching the conductive layer and the dielectric layer, thereby forming a gate; forming a source and drain region; and removing the mask.
Reference is now made to the following descriptions taken in conjunction with the accompanying drawings, in which:
In
Reference numeral 13 denotes an oxide layer which may typically have a thickness between 30 Å and 300 Å.
Reference numeral 15 denotes a polysilicon layer which may or may not be doped. The thickness of polysilicon layer 15 is typically desirably between 200 Å and 5000 Å.
Reference numeral 17 denotes a doped silicon dioxide layer. The thickness of layer 17 is typically desirably between 100 Å and 4000 Å.
Desirably, layer 17 may be formed as a single layer or sometimes as a bilayer. For example, layer 17 may be formed from BPSG having approximately 4% boron and 4% phosphorous by weight. Alternatively, layer 17 may be formed from BPSG, having approximately 1% boron and 5% phosphorus. Furthermore, layer 17 may; be formed from PSG having a doping of approximately 2% or greater phosphorous. Other suitable materials for layer 17 are BPSG, plasma enhanced doped or undoped oxide, spin-on glass, silicon nitride (LPCVD or plasma enhanced CVD), or silicon oxynitride. Generally, layer 17 may be a doped silicon dioxide formed from a variety of precursors such as TEOS, silane, DADBS, etc.
Layer 17 may be formed as a bilayer, as mentioned above. For example, layer 17 may be one of the forms of doped silicon oxide mentioned above formed over an undoped silicon oxide. Alternatively, layer 17 may be a single silicon oxide layer whose doping gradually increases from bottom to top. Layer 17 may also be a layer of silicon nitride with an underlying layer of silicon oxide which serves as an etch stop during subsequent etching steps.
Layer 21 is a patterned photoresist layer.
Turning to
In any case, after layer 17 has been subjected to the etch process for an appropriate period of time, resist 21 may be removed and the portion 117 of layer 17 beneath resist 21 may be used as a mask for subsequent etching which ultimately defines gate 23 shown in
In
Turning to
Turning to
Wet etching formulas based upon HF tend to attack doped glass more quickly than undoped glass. However, such processes nevertheless do etch undoped glass and may cause undesirable reduction of the bird's beak, leading to transistor leakage.
Layer 17 may also be removed utilizing NH4OH/H2O2. The use of NH4OH/H2O2 is termed an ammonium peroxide (AP) clean. The preferred formula is eight parts H2O, two parts H2O2 (30% concentrated), and one part concentrated NH4OH at approximately 80° C. Dry etch recipes may also be employed to remove layer 17. P-glass may be removed by unbuffered HF or NH4OH/H2O2.
If silicon nitride is used as layer 17, it can be removed in hot phosphoric acid or in plasma using chemistries selective to oxide. In such an event a protective oxide layer may be previously formed on top of layer 15 to protect it from an attack by the plasma. Alternately, plasmaless dry etching using gas phase fluorides such as chlorine trifluoride, bromine trifluoride, iodide pentafluoride and xenon difluoride can be used.
If silicide is not desired upon gate stack 23 or over junctions 37 and 39, conventional processing may begin at this point. For example, a dielectric may be blanket deposited, windows opened to expose junctions 37 and 39, and first level metallization formed.
Layer 17 has prevented channeling through the gate which consists of layers 13, and 15. Furthermore, layer 17 has been removed without risk of damage to the gate, the substrate, or the field oxide.
If silicide is desired, either upon gate stack 23 or over junctions 37 and 39, a variety of processing options are available. The next few paragraphs will explain how silicide may be formed upon the gate 23 and junctions 37 and 39.
Turning to
Alternatively, if it is desired to form a silicide over junctions 37 and 39 without forming a silicide over gate stack 23, a slightly different process may be employed. Starting from
After appropriate heat treatment, silicide regions 51 and 55 in
Should it be desired to form a silicided gate without silicided source or drain, the structure of
Next, turning to
Because oxide 17 is doped, it may be singly removed without a risk of damaging thermal oxide 57.
Turning to
Should a silicided gate be desired with silicided source or drains, the procedure initially depicted in
Turning to
The presence of layer 17 upon gate stack 77 serves to protect the silicide from ion implantation. If a spacer 200 is formed, it will protect the silicide in further processing, e.g., HF cleans where the silicide is titanium-silicide.
In
The present invention may also be employed to form a transistor without a silicided source or drain region. In
This Application is a Continuation of prior application Ser. No. 12/114,589 filed on May 2, 2008 to Sailesh Chittipeddi, et al. entitled, “TRANSISTOR FABRICATION METHOD” which is a continuation of prior application Ser. No. 10/224,220 filed on Aug. 20, 2002, now abandoned, which is a Divisional of application Ser. No. 08/587,061 filed on Jan. 16, 1996, now U.S. Pat. No. 6,498,080 issued on Dec. 24, 2002. The above-listed Applications are commonly assigned with the present invention and is incorporated herein by reference as if reproduced herein in its entirety under Rule 1.53(b).
Number | Date | Country | |
---|---|---|---|
Parent | 08587061 | Jan 1996 | US |
Child | 10224220 | US |
Number | Date | Country | |
---|---|---|---|
Parent | 12114589 | May 2008 | US |
Child | 12689749 | US | |
Parent | 10224220 | Aug 2002 | US |
Child | 12114589 | US |