U.S. patent application number 11/457203 was filed with the patent office on 2006-11-02 for mos transistor gates with doped silicide and methods for making the same.
Invention is credited to Luigi Colombo, Mark Visokay.
Application Number | 20060244045 11/457203 |
Document ID | / |
Family ID | 34376942 |
Filed Date | 2006-11-02 |
United States Patent
Application |
20060244045 |
Kind Code |
A1 |
Visokay; Mark ; et
al. |
November 2, 2006 |
MOS Transistor Gates with Doped Silicide and Methods for Making the
Same
Abstract
Semiconductor devices and fabrication methods are presented, in
which transistor gate structures are created using doped metal
silicide materials. Upper and lower metal silicides are formed
above a gate dielectric, wherein the lower metal silicide is doped
with n-type impurities for NMOS gates and with p-type impurities
for PMOS gates, and wherein a silicon may, but need not be formed
between the upper and lower metal silicides. The lower metal
silicide can be deposited directly, or may be formed through
reaction of deposited metal and poly-silicon, and the lower
silicide can be doped by diffusion or implantation, before or after
gate patterning.
Inventors: |
Visokay; Mark; (Richardson,
TX) ; Colombo; Luigi; (Dallas, TX) |
Correspondence
Address: |
TEXAS INSTRUMENTS INCORPORATED
P O BOX 655474, M/S 3999
DALLAS
TX
75265
US
|
Family ID: |
34376942 |
Appl. No.: |
11/457203 |
Filed: |
July 13, 2006 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
10674771 |
Sep 30, 2003 |
|
|
|
11457203 |
Jul 13, 2006 |
|
|
|
Current U.S.
Class: |
257/315 ;
257/E21.636; 257/E21.637 |
Current CPC
Class: |
H01L 21/823842 20130101;
H01L 21/823835 20130101 |
Class at
Publication: |
257/315 |
International
Class: |
H01L 29/788 20060101
H01L029/788 |
Claims
1-34. (canceled)
35. A transistor gate structure, comprising: a gate dielectric
formed above a semiconductor body; a first metal silicide on, and
in direct physical contact with, the gate dielectric, the first
metal silicide doped with substantially single polarity impurities
and having a first metal component; and a second metal silicide
having a metal component different from the first metal component,
above the first metal silicide.
36. (canceled)
37. The gate structure of claim 35, further comprising a silicon
layer between the first and second metal silicides.
38. The gate structure of claim 35, wherein the first metal
silicide comprises a refractory metal.
39. The gate structure of claim 38, wherein the refractory metal is
one of molybdenum, tungsten, tantalum, and titanium.
40. The gate structure of claim 38, wherein the second metal
silicide comprises nickel.
41. The gate structure of claim 35, wherein the second metal
silicide comprises nickel.
Description
FIELD OF INVENTION
[0001] The present invention relates generally to semiconductor
devices and more particularly to doped silicide MOS transistor
gates and fabrication methods for making the same.
BACKGROUND OF THE INVENTION
[0002] Field effect transistors (FETs) are widely used in the
electronics industry for switching, amplification, filtering, and
other tasks related to both analog and digital electrical signals.
Most common among these are metal-oxide-semiconductor field-effect
transistors (MOSFETs), wherein a gate contact or electrode is
energized to create an electric field in a channel region of a
semiconductor body, by which electrons are allowed to travel
through the channel between a source region and a drain region of
the semiconductor body. The source and drain regions are typically
formed by adding dopants to targeted regions on either side of the
channel. A gate dielectric or gate oxide is formed over the
channel, and a gate electrode or gate contact is formed over the
gate dielectric. The gate dielectric and gate electrode layers are
then patterned to form a gate structure overlying the channel
region of the substrate.
[0003] In operation of the resulting MOS transistor, the threshold
voltage (Vt) is the gate voltage value required to render the
channel conductive by formation of an inversion layer at the
surface of the semiconductor channel. Complementary MOS (CMOS)
devices have become widely used in the semiconductor industry,
wherein both n-channel and p-channel (NMOS and PMOS) transistors
are used to fabricate logic and other circuitry. For
enhancement-mode (e.g., normally off) devices, the threshold
voltage Vt is positive for NMOS and negative for PMOS transistors.
The threshold voltage is dependent upon the flat-band voltage,
where the flat-band voltage depends on the work function difference
between the gate and the substrate materials, as well as on surface
charge.
[0004] The work function of a material is a measure of the energy
required to move an electron in the material outside of a material
atom from the Fermi level, and is usually expressed in electron
volts (eV). For CMOS products, it is desirable to provide
predictable, repeatable, and stable threshold voltages for the NMOS
and PMOS transistors. To establish Vt values, the work functions of
the PMOS and NMOS gate contact and the corresponding channel
materials are independently tuned or adjusted through gate and
channel engineering, respectively.
[0005] Channel engineering typically involves shallow dopant
implants to the prospective channel regions of the semiconductor
body, sometimes referred to as threshold adjust (Vt adjust)
implants, where the implanted impurities behave as a sheet of fixed
charge located under the gate oxide. A Vt adjust implant for the
NMOS devices introduces boron or other p-type impurities into the
NMOS channel region to raise the channel work function (sometimes
referred to as a VTN implant), and a Vt adjust implant for the PMOS
devices introduces arsenic, phosphorus, or other n-type impurities
to lower the PMOS channel work function (VTP implant). In this
manner, the Vt for the channels can be separately adjusted for NMOS
and PMOS devices. Channel engineering typically includes multiple
implants, for example, a Vt adjust implant, a punch-thru implant to
suppress punch-through, and a channel stop implant, for each of the
NMOS and PMOS devices.
[0006] Gate engineering is employed in combination with channel
engineering to adjust the work function of the gate contact
materials, where different gate work function values are set for
PMOS and NMOS gates. The need to independently adjust PMOS and NMOS
gate work functions has made poly-silicon attractive for use as
gate contact material in CMOS processes, since the work function of
poly-silicon can be easily raised or lowered by doping the
poly-silicon with p-type or n-type impurities, respectively. The
PMOS poly-silicon gates are doped with p-type impurities and NMOS
gate poly-silicon is doped with n-type dopants, typically during
implantation of the respective source/drain regions following gate
patterning. In this way, the final gate work functions are
typically near the Si conduction band edge for NMOS and near the
valence band for PMOS. The provision of dopants into the
poly-silicon also has the benefit of increasing the conductivity of
the gate contact. Poly-silicon has thus far been widely using in
the fabrication of CMOS devices, wherein the gate engineering
(e.g., implants) are conventionally tuned to provide a desired gate
contact conductivity (e.g., sheet resistance value), and the
threshold voltage fine tuning is achieved by tailoring the Vt
adjust implants to change the channel work function.
[0007] FIG. 1 illustrates conventional CMOS fabrication process 10
beginning at 12, in which front end processing is performed at 14,
including well formation and isolation processing. At 16 and 18,
channel engineering is performed (e.g., Vt adjust, punch-thru, and
channel stop implants) for PMOS and NMOS regions, respectively. A
thin gate dielectric and an overlying poly-silicon are formed at 20
and 22, respectively, and the poly-silicon is patterned at 24 to
form gate structures for the prospective NMOS and PMOS transistors.
The gate structures are then encapsulated at 26, typically through
oxidation and highly-doped drain (HDD) implants are performed at 28
to provide p-type dopants to prospective source/drains of the PMOS
regions and n-type dopants to source/drains of the NMOS regions,
using the patterned gate structures and isolation structures as an
implantation mask. Sidewall spacers are then formed at 30 along the
lateral sidewalls of the gate structures.
[0008] At 32, the PMOS source/drain regions and the PMOS
poly-silicon gate structures are implanted with p-type dopants to
further define the PMOS source/drains, and to render the PMOS gates
conductive. Similarly, the NMOS source/drain regions and the NMOS
poly-silicon gate structures are implanted at 34 with n-type
dopants, further defining the NMOS source/drains and to rendering
the NMOS gates conductive. Thereafter, the source/drains and
poly-silicon gates are silicided at 36 and back end processing
(e.g., interconnect metalization, etc.) is performed at 38, before
the process 10 ends at 40. In the conventional process 10, the
channel engineering implants at 16 and 18 shift the work functions
of the PMOS and NMOS channel regions, respectively, to compensate
for the changes in the PMOS and NMOS poly-silicon gate work
functions resulting from the source/drain implants at 32 and 34,
respectively. In this manner, the desired work function difference
between the gates and channels, and hence the desired threshold
voltages, may be achieved for the resulting PMOS and NMOS
transistors.
[0009] The gate dielectric or gate oxide between the channel and
the gate contact is an insulator material, typically SiO.sub.2 or
other dielectric, that operates to prevent large currents from
flowing from the gate contact into the channel when a voltage is
applied to the gate electrode. The gate dielectric also allows an
applied gate voltage to establish an electric field in the channel
region in a controllable manner. Continuing trends in semiconductor
product manufacturing include reduction in electrical device
feature sizes (scaling), as well as improvements in device
performance in terms of device switching speed and power
consumption. MOS transistor performance may be improved by reducing
the distance between the source and the drain regions under the
gate electrode of the device, known as the gate or channel length,
and by reducing the thickness of the layer of gate oxide that is
formed over the semiconductor surface.
[0010] However, there are electrical and physical limitations on
the extent to which SiO.sub.2 gate dielectrics can be made thinner.
These include gate leakage currents tunneling through the thin gate
oxide, limitations on the ability to form very thin oxide films
with uniform thickness, and the inability of very thin SiO.sub.2
gate dielectric layers to prevent dopant diffusion from the gate
poly-silicon into the underlying channel. Accordingly, recent
scaling efforts have focused on high-k dielectric materials having
dielectric constants greater than that of SiO.sub.2, which can be
formed in a thicker layer than scaled SiO.sub.2, and yet which
produce equivalent field effect performance. A thicker high-k
dielectric layer can thus be formed to avoid or mitigate tunneling
leakage currents, while still achieving the required electrical
performance equivalent (e.g., capacitance value) of a thinner
SiO.sub.2.
[0011] Another shortcoming of scaled CMOS devices having
poly-silicon gate contacts is known as poly-silicon depletion.
Poly-silicon depletion occurs when annealing or other thermal
back-end processing following the implants at 32 and 34 is
insufficient to drive the implanted impurities down the entire
depth of the poly-silicon gate structures. In this situation, a
bottom portion of the poly-silicon gate contact near the gate
dielectric is "depleted" of charges, and acts as an insulator. The
depleted portion of the gate contact and the gate dielectric
operate as series connected capacitors, resulting in a reduced
effective gate capacitance, which reduces the drive current
capability of the device. Consequently, poly-silicon depletion
causes reduction in device performance which leads to poor
unscalable devices. Simply increasing the implant energy and/or
anneal time to combat poly-silicon depletion has adverse results,
in that the corresponding depths of the concurrently implanted
source/drain regions are increased.
[0012] With the relatively thick gate dielectrics and gate contact
structures of the past, poly depletion was not critical to achieve
desired device performance. However, as gate dielectrics and gate
contacts continue to become smaller through scaling, the
poly-silicon depletion problem is more pronounced, wherein
poly-silicon depletion regions of 2 to 4 angstroms become a
significant fraction of the overall effective gate capacitance.
Thus, while poly-silicon gate contacts have previously offered
flexibility in providing dual work functions for CMOS processes,
the future viability of conventional poly-silicon gate technology
is lessened as scaling efforts continue. Accordingly, attention has
recently been directed again to the possibility of using metal gate
contacts in CMOS products, where the metal gate materials
conceivably do not need doping for conductivity improvement.
Although this approach presumably avoids poly-silicon depletion
issues with respect to gate capacitance, there remains a need for
dual or differentiated work function capability (e.g., for PMOS and
NMOS transistors) in CMOS processes. In this regard, metal work
functions are not shifted as easily by the same amounts as was the
case for poly-silicon. Accordingly, there is a need for improved
CMOS transistor gate designs and fabrication techniques by which
the benefits of scaling can be achieved while avoiding or
mitigating the poly-silicon depletion degradation found in
conventional devices.
SUMMARY OF THE INVENTION
[0013] The following presents a simplified summary in order to
provide a basic understanding of one or more aspects of the
invention. This summary is not an extensive overview of the
invention, and is neither intended to identify key or critical
elements of the invention, nor to delineate the scope thereof.
Rather, the primary purpose of the summary is to present some
concepts of the invention in a simplified form as a prelude to the
more detailed description that is presented later.
[0014] The invention relates to semiconductor devices and
fabrication methods, wherein transistor gate structures are created
using doped metal silicide materials. A first metal silicide is
formed above a gate dielectric and the silicide is doped with
n-type impurities for NMOS gates or with p-type impurities for PMOS
gates. The inventors have appreciated that the gate work function
of the resulting NMOS and PMOS transistors can be successfully
adjusted through appropriate doping the gate silicide, thereby
facilitating elimination or simplification of channel engineering.
The doped first metal silicide may comprise a refractory metal such
as molybdenum, tungsten, tantalum, titanium, or others having a
high melting temperature, and may be formed by direct silicide
deposition or through reacting metal and silicon deposited above
the gate dielectric. In this regard, a single starting material may
be employed to form both PMOS and NMOS gate structures, thus
avoiding process integration problems associated with using
different gate materials. Dopants may be added to the first
silicide through any means, such as direct implantation into a
reacted or deposited silicide or implantation into unreacted metal
and/or unreacted silicon followed by annealing to concurrently
diffuse the dopants into the metal and to create the resulting
metal silicide, where the dopants can be introduced before or after
gate patterning.
[0015] The gate structures include a second upper metal silicide
located above the doped first metal silicide. A silicon material
may be provided between the first and second metal silicides, or
the second metal silicide may overlie the doped first metal
silicide thereby forming a silicide bilayer above the gate
dielectric. In order to facilitate controlled dopant distribution
within the doped first metal silicide for setting the gate work
function (e.g., particularly at the gate dielectric interface), the
first metal silicide may be relatively thin, wherein the provision
of the upper silicide and/or the intervening silicon can
advantageously reduce the gate contact sheet resistance. In one
example, the first metal silicide is doped MoSi.sub.2 where
poly-silicon is formed over the doped metal silicide. The
poly-silicon may also be doped, for example, via source/drain or
other implantations, to render the poly-silicon conductive. A
second silicide is then formed over the poly-silicon, wherein the
poly-silicon may be reacted with a deposited second metal (e.g.,
nickel) to form the upper silicide. The upper silicide and the
remaining poly-silicon may thus be of lower resistivity than the
first metal silicide. In another example, the entire poly-silicon
is reacted (e.g., consumed) in forming the upper silicide, whereby
the gate stack includes a metal silicide bilayer.
[0016] In accordance with an aspect of the invention, a method is
provided for fabricating PMOS and NMOS gate structures in a
semiconductor device. The method comprises forming a gate
dielectric, forming an n-doped first metal silicide above the gate
dielectric in an NMOS region, and forming a p-doped first metal
silicide above the gate dielectric in the PMOS region. The method
further comprises forming a second metal silicide above the doped
first metal silicide in the NMOS and PMOS regions. The first and
second metal silicides may be different, for example, where the
lower silicide includes a refractory metal (e.g., molybdenum,
tungsten, tantalum, titanium, etc.) and the upper silicide includes
nickel, and the method may further comprise forming silicon between
the first and second metal silicides in the NMOS and PMOS
regions.
[0017] Another aspect of the invention provides a semiconductor
device comprising NMOS and PMOS gate structures. The NMOS gate
structure comprises a gate dielectric above a semiconductor body,
an n-doped first metal silicide structure above the gate
dielectric, and a second metal silicide above the n-doped first
metal silicide. The PMOS gate structure comprises a gate dielectric
above a semiconductor body, a p-doped first metal silicide
structure above the gate dielectric, and a second metal silicide
above the p-doped first metal silicide. In one implementation, the
device further comprises silicon between the first and second metal
silicides in the NMOS and PMOS gate structures. Yet another aspect
of the invention provides a transistor gate structure, comprising a
gate dielectric formed above a semiconductor body, a first metal
silicide above the gate dielectric, the first metal silicide being
doped with n or p-type impurities, and a second metal silicide
above the first metal silicide.
[0018] The following description and annexed drawings set forth in
detail certain illustrative aspects and implementations of the
invention. These are indicative of but a few of the various ways in
which the principles of the invention may be employed.
BRIEF DESCRIPTION OF THE DRAWINGS
[0019] FIG. 1 is a simplified flow diagram illustrating a
conventional poly-silicon gate CMOS fabrication process including
channel engineering for both PMOS and NMOS transistors;
[0020] FIGS. 2A and 2B are partial side elevation views in section
illustrating exemplary MOS transistors having gate structures with
first and second metal silicides in accordance with one or more
aspects of the present invention;
[0021] FIG. 3 is a flow diagram illustrating an exemplary method of
fabricating semiconductor devices having PMOS and NMOS gate
structures in accordance with the invention;
[0022] FIG. 4 is a partial flow diagram illustrating several
techniques for forming MOS transistor gate structures with doped
first metal silicide in the fabrication method of FIG. 3, wherein
the first metal silicide is formed through deposition or reaction
prior to introducing dopants therein;
[0023] FIGS. 5A-5L are partial side elevation views in section
illustrating exemplary NMOS and PMOS transistors undergoing CMOS
processing in accordance with the invention at various stages of
fabrication to form gate structures with first and second metal
silicides and an intervening poly-silicon;
[0024] FIGS. 5M-5O are partial side elevation views in section
illustrating exemplary NMOS and PMOS transistors undergoing CMOS
processing at various stages of fabrication to form gate structures
with first and second metal silicides in a silicide bilayer with no
intervening silicon in accordance with the invention;
[0025] FIG. 6 is a partial flow diagram illustrating other
techniques for forming MOS transistor gate structures with doped
first metal silicide in the fabrication method of FIG. 3, in which
one or both of a first metal and a first silicon are doped prior to
annealing to form the first metal silicide;
[0026] FIGS. 7A-7L are partial side elevation views in section
illustrating exemplary NMOS and PMOS transistors undergoing
processing at various stages of fabrication to form gate structures
with first and second metal silicides and an intervening silicon in
accordance with the invention; and
[0027] FIGS. 7M-7O are partial side elevation views in section
illustrating exemplary NMOS and PMOS transistors undergoing CMOS
processing at various stages of fabrication to form gate structures
with first and second metal silicides in a silicide bilayer with no
intervening silicon.
DETAILED DESCRIPTION OF THE INVENTION
[0028] One or more implementations of the present invention will
now be described with reference to the attached drawings, wherein
like reference numerals are used to refer to like elements
throughout, and wherein the illustrated structures are not
necessarily drawn to scale. The invention relates to semiconductor
devices, as well as MOS transistor gate structures and fabrication
methods therefor, in which first and second metal silicides are
formed in a gate structure and the first metal silicide is
doped.
[0029] FIGS. 2A and 2B illustrate exemplary CMOS semiconductor
devices 50 and 52, with NMOS transistors 58a and 58b, respectively,
having gate structures in accordance with the invention. The
devices 50, 52 are fabricated in or on a silicon substrate 54,
although the invention may be carried out in association with SOI
wafers, epitaxial silicon layers formed over silicon wafers, and
any other semiconductor body. N-wells, p-wells, and/or buried
layers (not shown) may be formed in the substrate 54, and the
substrate 54 may be initially doped with p-type or other
impurities. The transistors 58a and 58b include field oxide (FOX)
isolation structures 56, and n-doped source/drains 60.
[0030] In accordance with an aspect of the invention, the devices
50 and 52 include MOS transistor gate structures, comprising a gate
dielectric layer 62 formed above a channel region of the substrate
54, an n-doped first metal silicide 64 above the gate dielectric 62
having a thickness of about 100 .ANG. or less, and a second metal
silicide 66 above the first metal silicide 64, wherein the
exemplary gate structures of FIGS. 2A and 2B have a total thickness
on the order of about 1000 .ANG. or less. The gate structures have
sidewalls over which an encapsulation oxide or other material 70 is
formed, and along which sidewall spacers 72 are formed. Silicide
source/drain contacts 74 are formed over the source/drains 60. The
transistor 58b in FIG. 2B provides a gate structure having a
bilayer metal silicide with the second silicide 66 situated over
the doped first silicide 64.
[0031] The transistor 58a of FIG. 2A further includes a silicon
layer 68 between the first and second metal silicides 64 and 66.
The intervening silicon 68 may be doped to increase the
conductivity thereof, and to thereby reduce the sheet resistance of
the gate, wherein a portion of the silicon 68 may be reacted with
nickel or other metal in forming the second silicide 66. In this
regard, a poly-silicon (not shown) may be deposited above the first
silicide 64 in the device 52, the entirety of which is consumed in
forming the second silicide 66, leaving the bilayer silicide gate
stack shown in FIG. 2B. The provision of silicon above the first
silicide 64 advantageously provides protection for the first
silicide 64 in fabrication processing, for example, to integrate
the creation of the gate structures of the invention into existing
CMOS process flows tailored for poly-silicon being present during
gate etch processes to define the gate structures for NMOS and PMOS
transistors. For instance, poly-silicon coverage may be desirable
for a case in which the first silicide 64 comprises titanium
silicide, which is reactive to hydrogen fluoride used in wet etch
steps or other processing following formation of the first silicide
64.
[0032] In the illustrated devices 50 and 52, the first and second
metal silicides 64 and 66, respectively, are different, wherein the
first metal silicide 64 comprises a refractory metal such as
molybdenum, tungsten, tantalum, titanium, or others, while the
second metal silicide comprises nickel. Alternatively, the first
and second metal silicides 64 and 66 can be the same type and/or
can comprise any suitable metal within the scope of the invention.
However, the inventors have appreciated that the use of refractory
metal silicides, such as MoSi.sub.2 or others comprising W, Ti, Ta,
etc., for the first silicide 64 provides high melting temperatures
to withstand fabrication processing. In this regard, nickel or
other second silicides 66 provide low resistivity, wherein the
second silicide 66 may be formed later in a fabrication flow, and
thus only be exposed to relatively lower temperature back-end
processing.
[0033] While NMOS transistors 58 are illustrated in FIGS. 2A and
2B, the aspects of the invention are also applicable to PMOS
structures, as described further below with respect to FIGS. 5A-5O
and 7A-7O. The dopants in the first metal silicide 64 of NMOS and
PMOS transistor gate structures can be separately tailored for
setting the respective gate work functions, thereby facilitating
threshold voltage tuning while simplifying or eliminating channel
engineering steps and avoiding poly depletion troubles of
conventional CMOS devices. A single type of doped metal silicide
can be employed in both the NMOS and PMOS regions of a device, thus
simplifying subsequent patterning and other processing steps. N and
p-type dopants are separately introduced into the lower metal
silicide 64 to provide differentiated work functions for NMOS and
PMOS transistors, respectively, either before or after gate
patterning.
[0034] Another aspect of the invention provides methods for
fabricating PMOS and NMOS transistors in a semiconductor device. In
the examples below, an n-doped first metal silicide is formed above
a gate dielectric in an NMOS region of a semiconductor device, a
p-doped first metal silicide is formed above the gate dielectric in
a PMOS region, and a second metal silicide is formed above the
doped first metal silicide in the NMOS and PMOS regions. In
accordance with the invention, the n-doped and p-doped first
silicides can be formed by separate processes for the NMOS and PMOS
regions, respectively. However, the invention also advantageously
provides for forming the doped first silicides using a common
starting material in both NMOS and PMOS regions. In one example, a
single starting first metal silicide is deposited above the gate
dielectric and then selectively doped in the NMOS and PMOS regions.
In another implementation, a common first metal and a first silicon
are deposited in both regions, and are subsequently annealed to
form the first silicide, either before, or after selective doping
of the NMOS and PMOS regions. Thereafter, a second or upper
silicide, doped or undoped, is formed above the first silicide, and
an optional silicon layer may be formed between the first and
second silicides.
[0035] FIG. 3 illustrates an exemplary semiconductor device
fabrication process 100 in accordance with the invention, and FIG.
4 illustrates various exemplary techniques for forming MOS
transistor gate structures with doped first metal silicide in the
fabrication method of FIG. 3, in which the first metal silicide is
formed through deposition or reaction prior to introducing dopants
therein. FIGS. 5A-5O illustrate fabrication of an exemplary CMOS
device according to the processing examples of FIGS. 3 and 4. FIG.
6 illustrates other techniques for forming MOS transistor gate
structures with doped first metal silicide in the fabrication
method of FIG. 3, in which one or both of a first metal and a first
silicon are doped prior to annealing to form the first metal
silicide. FIGS. 7A-7O illustrate fabrication of an exemplary CMOS
device according to the processing examples of FIGS. 3 and 6.
[0036] Although the exemplary method 100 of FIGS. 3, 4, and 6 is
illustrated and described below as a series of acts or events, it
will be appreciated that the present invention is not limited by
the illustrated ordering of such acts or events. For example, some
acts may occur in different orders and/or concurrently with other
acts or events apart from those illustrated and/or described
herein, in accordance with the invention. In addition, not all
illustrated steps may be required to implement a methodology in
accordance with the present invention. Further, the methods
according to the present invention may be implemented in
association with the formation and/or processing of structures
illustrated and described herein as well as in association with
other structures and devices not illustrated.
[0037] The method 100 begins at 102 in FIG. 3, wherein front end
processing is performed at 104. Any front end processing may be
performed within the scope of the invention, wherein the processing
at 104 may include, for example, formation of n and p wells using
diffusion, implantation, or other suitable processing steps, as
well as formation of isolation structures in field regions of a
device wafer, using LOCOS, STI, or any suitable isolation
processing. The front-end processing 104 may also comprise channel
doping (e.g., channel engineering) to adjust the work function of a
channel portion of the semiconductor body in prospective NMOS
and/or PMOS active regions. Such channel engineering may include,
for example, a shallow Vt adjust implant to introduce boron or
other p-type dopants into the NMOS channel regions with the PMOS
regions masked to raise the NMOS channel work function. In
addition, a boron punch-thru implant may be performed to suppress
punch-through, along with a somewhat deeper boron channel stop
implant in the NMOS regions. Similar channel engineering may be
performed to provide n-type dopants (e.g., arsenic, phosphorus,
etc.) to the PMOS regions with the NMOS regions masked.
[0038] At 106, a gate dielectric is formed in the NMOS and PMOS
regions using any suitable materials, material thicknesses, and
processing steps, including a single thermal oxidation or
deposition or combinations thereof to form a gate dielectric above
the semiconductor body. The invention may be employed in
conjunction with any gate dielectric material, such as SiO.sub.2,
SiON, high-k dielectrics, and stacks or combinations thereof,
including but not limited to binary metal oxides including aluminum
oxide (Al.sub.2O.sub.3), zirconium oxide (ZrO.sub.2), hafnium oxide
(HfO.sub.2), lanthanum oxide (La.sub.2O.sub.3), yttrium oxide
(Y.sub.2O.sub.3), titanium oxide (TiO.sub.2), as well as their
silicates and aluminates; metal oxynitrides including aluminum
oxynitride (AlON), zirconium oxynitride (ZrON), hafnium oxynitride
HfON), lanthanum oxynitride (LaON), yttrium oxynitride (YON), as
well as their silicates and aluminates such as ZrSiON, HfSiON,
LaSiON, YSiON, etc.; and perovskite-type oxides including a
titanate system material such as barium titanate, strontium
titanate, barium strontium titanate (BST), lead titanate, lead
zirconate titanate, lead lanthanum zirconate titanate, barium
lanthanum titanate, barium zirconium titanate; a niobate or
tantalate system material such as lead magnesium niobate, lithium
niobate, lithium tantalate, potassium niobate, strontium aluminum
tantalate and potassium tantalum niobate; a tungsten-bronze system
material such as barium strontium niobate, lead barium niobate,
barium titanium niobate; and bilayered perovskite system material
such as strontium bismuth tantalate, bismuth titanate and others.
Separate processing may optionally be employed to form different
gate dielectrics in the NMOS and PMOS regions within the scope of
the invention. In the examples illustrated and described herein, a
single thermal oxidation is performed at 106 to create a thin
SiO.sub.2 or gate oxide overlying the substrate in the NMOS and
PMOS regions.
[0039] Gate structures are then formed at 110, with the general
gate fabrication principles being illustrated in FIG. 3, and
several exemplary gate fabrication implementations being
illustrated in FIGS. 4 and 6. An n-doped first metal silicide is
formed above the gate dielectric in the NMOS region at 112 in FIG.
3, and a p-doped first metal silicide is formed above the gate
dielectric in the PMOS region at 114. The silicide may be formed by
any suitable techniques in accordance with the invention, including
but not limited to deposition of a metal silicide, or formation of
metal and silicon layers and reaction thereof to form the first
metal silicide. N-type and p-type dopants are selectively
introduced into the first metal silicide of the NMOS and PMOS
regions using any suitable techniques, including but not limited to
implantation and/or diffusion, wherein the dopants may be
introduced before and/or after silicidation of a first metal with
silicon in the case of reacted silicide.
[0040] A silicon layer is formed at 116 above the first metal
silicide in the NMOS and PMOS regions, some or all of which may
subsequently be consumed in forming a second metal silicide above
the first metal silicide, wherein the silicon may, but need not, be
n-doped or p-doped in the NMOS and PMOS regions, respectively. Any
suitable formation techniques may be employed to create the silicon
at 116, for example, chemical vapor deposition (CVD) processes,
etc. The silicon may be formed at 116 directly over the first metal
silicide or may be formed over one or more intervening material
layers within the scope of the invention. The silicon and first
metal silicide are patterned at 118 to form NMOS and PMOS gate
structures using any suitable techniques, for example, such as
masked etching.
[0041] At 120, the gate structures are optionally encapsulated
using any suitable materials and processing steps. Highly-doped
drain (HDD) implants are performed at 122 to introduce p-type
dopants (e.g., boron, etc.) into prospective PMOS source/drains and
to provide n-type dopants (e.g., arsenic, phosphorus, etc.) to
prospective NMOS source/drains, using any suitable processing steps
and operational parameters, wherein the patterned gate structures
and isolation structures operate as implantation masks. The HDD
implants at 122 also provide n-type dopants to the silicon in the
NMOS regions and p-type dopants to the silicon in the PMOS regions
in the illustrated example, although the gate silicon layers can
alternatively be masked during the implantation operations at 122.
Sidewall spacers are formed at 124 along the lateral sidewalls of
the patterned gate structures, using any suitable process
techniques and materials, for example, by depositing and etching
SiO.sub.2, Si.sub.3N.sub.4, etc.
[0042] Source/drain implants are then performed at 126 and 128 to
further dope the PMOS and NMOS source/drain regions of the
semiconductor body, wherein n-type dopants may also be implanted
into the silicon in the NMOS regions and p-type dopants may be
implanted into the silicon in the PMOS regions. In the illustrated
implementation, a PMOS source/drain implant is performed at 126, to
introduce boron or other p-type impurities into the PMOS region of
the semiconductor body and into the PMOS gate silicon with the NMOS
region masked. Any suitable implantation technique, operational
settings, and implant species may be employed at 126 to render the
PMOS gate silicon conductive and to further define the PMOS
source/drains in the semiconductor body.
[0043] An NMOS source/drain implant is performed at 128 to
introduce arsenic, phosphorus, or other suitable n-type impurities
to the NMOS region of the semiconductor body and to the NMOS gate
silicon with the PMOS region masked. Any suitable implantation
technique, operational settings, and implant species may be
employed at 128 to render the NMOS gate silicon conductive and to
further define the NMOS source/drains, wherein the implants at 126
and 128 may be performed in any order. In addition, activation
anneal operations may be performed following one or both of the
source/drain implants 126, 128, wherein such annealing may also
operate to diffuse dopants in the first silicide.
[0044] A self-aligned silicide process is performed at 130 to
silicide the source/drains and the gate silicon, wherein
silicidation of the gate silicon creates a second metal silicide
above the first metal silicide of the gate stacks in the exemplary
implementation. In one example, nickel or other metal is deposited
at 130 over the gate silicon and over the semiconductor body in the
prospective source/drain regions, and the wafer is annealed to
react the gate silicon and the source/drain region silicon of the
semiconductor body with the deposited nickel to form nickel
silicide. By this technique, the second silicide of the gate stack
may be formed above the doped first metal silicide concurrently
with formation of source/drain silicide contacts, although separate
silicide formation techniques (e.g., reaction or deposition and
etching) may be employed to form the second metal silicide of the
gate structures. In one implementation (e.g., FIG. 2A above), a
portion of the doped silicon remains unreacted after the
silicidation at 130. In another example, substantially all of the
silicon is consumed at 130 in creating the second silicide, leaving
a silicide bilayer gate structure as shown in FIG. 2B above.
[0045] The second silicide can be formed anywhere above the first
metal silicide of the gate structure within the scope of the
invention. For example, the second silicide may be formed directly
over the first metal silicide, directly over an intervening
silicon, or one or more other layers may be situated between the
second silicide and the first silicide or the silicon. As discussed
above, the second silicide may be the same or may comprise
different metal than the first silicide. For example, the second
silicide may advantageously comprise nickel (e.g., low resistivity,
lower melting temperature), where the first metal silicide
comprises a refractory metal (e.g., molybdenum, tungsten, tantalum,
titanium, etc.) having a higher melting temperature. In this
manner, the second silicide may provide improved (e.g., lower) gate
sheet resistance, whereas the first silicide is able to withstand
relatively higher processing thermal budget between the first
silicide formation at 112 and the back-end processing at 132. In
addition, the provision of the intervening silicon over the first
metal silicide (e.g., whether fully consumed at 130 or not) may
advantageously protect the first metal silicide during fabrication
processing. Back end processing (e.g., interconnect metalization,
etc.) is performed at 132, before the exemplary method 100 ends at
134.
[0046] Referring now to FIGS. 3, 4, and 5A-5O, an exemplary CMOS
device 302 is illustrated in FIGS. 5A-5O at various stages of
fabrication processing generally according to the method 100 above.
In this example, the gate fabrication for the NMOS and PMOS
transistors (e.g., 110 in method 100) is illustrated in further
detail in FIG. 4, in which a first metal silicide (e.g., MoSi.sub.2
312 in the exemplary device 302) is formed through deposition or
reaction prior to introduction of n or p-type dopants therein.
FIGS. 5A-5L illustrate exemplary NMOS and PMOS transistors in the
device 302 undergoing processing at various stages of fabrication
in accordance with the invention to form gate structures with first
and second metal silicides 312 and 344 and an intervening
poly-silicon 324. In FIGS. 5M-5O, an alternative device 302a is
illustrated having gate structures with first and second metal
silicides in a silicide bilayer with no intervening silicon in
accordance with the invention.
[0047] As illustrated in FIG. 5A, the CMOS device 302 includes a
silicon substrate semiconductor body 304 with a p-well 306 formed
in an NMOS region and an n-well 308 formed in a PMOS region, as
well as field oxide (FOX) isolation structures 310, where the wells
306, 308 and the field oxide 310 are formed during front-end
processing (e.g., at 104 in FIG. 3). A gate dielectric 316 is then
formed (e.g., at 106 in FIG. 3) over the surface of the substrate
304 in NMOS and PMOS active regions of the device 302 between the
FOX isolation structures 310, for example, a thermally grown
SiO.sub.2 oxide 316 or any other dielectric material formed by any
suitable technique in the illustrated device 302. FIG. 4
illustrates various alternative gate structure fabrication
approaches wherein a first metal silicide is formed over the gate
dielectric 316 through deposition or silicide reaction of metal and
silicon (e.g., "DEP" or "REACT" at 200), and wherein poly-silicon
is formed above the first metal silicide either before or after
doping (e.g., "POLY FIRST" or "DOPE FIRST" at 204).
[0048] In the exemplary device 302, the first metal silicide 312 is
formed through deposition (e.g., "DEP" at 200). In FIG. 5B, a
refractory metal silicide 312 (molybdenum disilicide, MoSi.sub.2)
is deposited (e.g., 202 of FIG. 4) over the gate dielectric 316 in
NMOS and PMOS regions of the device 302 via a deposition process
314, for example, sputtering, CVD, ALD, or other suitable
deposition technique. The first metal silicide 312 may be deposited
or otherwise formed to any suitable thickness within the scope of
the invention, wherein the silicide 312 in the exemplary device 302
has a thickness of about 100 .ANG. or less. Compared with reacting
the first metal silicide (e.g., 230-234 below), deposition of the
first metal silicide 312 through sputtering, CVD, etc., may allow
tighter control over the final silicide composition.
[0049] A layer of silicon 324 is then deposited in FIG. 5C over the
first metal silicide 312 before doping (e.g., "POLY FIRST" at 204)
using a CVD deposition process 326 (206 in FIG. 4). The silicon 324
may be formed to any desired thickness within the scope of the
invention. Alternatively, dopants may be directly implanted into
the first metal silicide prior to silicon deposition (e.g., 216-220
below). In FIG. 5D, a mask 321 is formed to cover the PMOS regions
and expose the NMOS regions, and n-type dopants (e.g., P, As, etc.)
are introduced via an implantation process 323 into the exposed
silicon in the NMOS regions, thus creating n-doped poly-silicon
324a over the first metal silicide in the NMOS regions (208 in FIG.
4). The range of the implantation can also be adjusted to dope the
poly-silicon near to the metal silicide, or to dope the silicide
layer itself.
[0050] The mask 321 is then removed and a mask 320 is formed in
FIG. 5E to cover the NMOS regions and expose the PMOS regions. An
implantation process 322 is performed to provide boron or other
p-type dopants to the poly-silicon in the PMOS region (210 in FIG.
4), thereby creating p-doped poly-silicon 324b over the first
silicide 312 in the PMOS region, after which the mask 320 is
removed. Alternatively, the first silicide 312 and the poly-silicon
324 can be patterned before the implantations at 208 and 210.
Thereafter in FIG. 5F, an anneal process 325 is performed (212 in
FIG. 4) to diffuse n and p-type dopants from the NMOS and PMOS
poly-silicon 324a and 324b, respectively, into the underlying first
metal silicide 312 in the NMOS and PMOS regions, thereby creating n
and p-doped first metal silicide 312a and 312b in the NMOS and PMOS
regions, respectively.
[0051] In FIG. 5G, a mask 328 is formed, and the n and p-doped
poly-silicon 324a and 324b, respectively, as well as the doped
first metal silicide are patterned (214 in FIG. 4) via an etch
process 330 to define patterned NMOS and PMOS gate structures. In
FIG. 5H, the mask 328 is removed and the patterned gate structures
are encapsulated (120 in FIG. 3) by forming a single or multi-layer
encapsulation structure 332 along the sidewalls and top of the gate
structures. Following encapsulation, HDD implants 334 are performed
(122 in FIG. 3) to introduce p-type dopants into prospective PMOS
source/drains 336 and to introduce n-type dopants into prospective
NMOS source/drains 338, with the field oxide 310 and patterned gate
structures operating as an implantation mask.
[0052] In FIG. 5I sidewall spacers 340 are formed (e.g., 124 in
FIG. 3) along the lateral sidewalls of the patterned gate
structures. A mask 320 is formed over the NMOS region, and a p-type
source/drain implant 342a is performed (126 in FIG. 3) to implant
boron or other p-type impurities into the PMOS region, which
further defines the PMOS source/drains 336 and provides additional
dopants to the PMOS gate stack poly-silicon 324b. The mask 320 is
removed in FIG. 5J, and a mask 327 is formed over the PMOS region.
An n-type source/drain implant 342b is performed (128 in FIG. 3) to
implant arsenic, phosphorus or other n-type impurities into the
NMOS region, thus further defining the NMOS source/drains 338 and
providing additional dopants to the NMOS gate stack poly-silicon
324a.
[0053] In FIGS. 5K and 5L, a second metal silicide 344a is formed
(130 in FIG. 3) above the first metal silicide 312 and the
poly-silicon 324, and source/drain silicide contacts 344b are
formed over the source drains 336 and 338. In the illustrated
example, a layer of nickel 341 is deposited over the device 302 in
FIG. 5K, whereby the nickel 341 overlies the poly-silicon 324 of
the patterned gate stacks and also overlies the doped source/drains
336 and 338 of the substrate 304. A thermal anneal 343 is performed
to react the nickel 341 with the poly-silicon 324 of the gate
stacks, thereby forming a second metal silicide 344a above the
remaining poly-silicon 324 as seen in FIG. 5L. In addition, the
anneal 343 concurrently reacts the nickel 341 with the source/drain
silicon 336 and 338 to form nickel silicide source/drain contacts
344b, after which any remaining (e.g., unreacted) nickel 341 is
removed. Interconnect metalization and other back-end processing
(not shown) may thereafter be performed to complete the fabrication
of the device 302 (132 in FIG. 3). The second metal silicide 344a
and the source/drain silicide contacts 344b may alternatively be
formed by separate processes within the scope of the invention.
[0054] FIGS. 5M-5O illustrate an alternative device 302a in
accordance with the invention, having MOS gate structures
comprising first and second metal silicides 312 and 344a in a
silicide bilayer with no intervening poly-silicon. In FIG. 5M, a
relatively thick first metal silicide 312 is formed over the gate
dielectric 316 (e.g., MoSi.sub.2 having a thickness greater than
about 100 .ANG.) and somewhat thinner poly-silicon 324 is deposited
over the first metal silicide 312. Following gate stack and
source/drain fabrication as described above, FIG. 5N illustrates
deposition of nickel 341 over the patterned NMOS and PMOS gate
structures and the source/drains 336, 338. An anneal process 343 is
performed to react the nickel 341 with the poly-silicon 324 of the
gate stacks, thereby forming a second metal silicide 344a and
concurrently forming silicide source/drain contacts 344b. In the
device 302a, the relative deposited thicknesses of the first
silicide 312, the poly-silicon 324, and the nickel 341 are selected
such that the anneal causes the silicide 344a to consume
substantially all of the poly-silicon 324 in the gate structures,
as seen in FIG. 5O. Thus, a bilayer (e.g., first and second
silicides 312 and 344a) is provided in the gates above the gate
dielectric 316. As with the above example, a single silicide
process may be used to concurrently form the second metal silicide
344a of the gate structures as well as the source/drain contacts
344b, or these may be formed by separate processes within the scope
of the invention.
[0055] In the above example of FIGS. 4 and 5A-5O, the doped first
metal silicide 312 is formed by depositing first metal silicide
material 312 above the gate dielectric in the NMOS and PMOS
regions, introducing n-type dopants into the first metal silicide
material 312 in the NMOS region, and introducing p-type dopants
into the first metal silicide material 312 in the PMOS region.
However, alternative implementations are possible where the first
(e.g., lower) metal silicide 312 is formed prior to introduction of
dopants therein, some of which are illustrated in FIG. 4. In one
alternative, the first metal silicide 312 may be formed by reaction
of deposited silicon and metal (e.g., "REACT" at 200). In this
case, a metal (e.g., refractory or other suitable first metal
material) is deposited over the gate dielectric at 230 to any
thickness using any suitable process. A first silicon is then
deposited at 232 over the deposited metal, and an anneal is
performed at 234 to form the first metal silicide 312.
[0056] This reacted silicide can then be provided with dopants
through diffusion from an overlying doped poly-silicon (e.g., at
206-212 as described above), or be directly implanted prior to
poly-silicon formation. In this case ("DOPE FIRST" at 204), the
first metal silicide of the NMOS region is implanted at 216 with
n-type dopants and the first metal silicide of the PMOS region is
implanted with p-type dopants at 218, using appropriate masks.
Thereafter an anneal may optionally be performed to diffuse the
implanted dopants, and poly-silicon is deposited over the doped
first metal silicide at 220. As one alternative, the silicon
deposition at 220 (e.g., CVD deposition) may provide suitable
thermal heating to diffuse the dopants to the desired depth,
concentration, profile, etc. within the first metal silicide,
whereby an extra (e.g., dedicated) anneal may not be required.
Thereafter, the gate structures are patterned at 214 as described
above. In the various implementations illustrated and described
herein, the gate structures can alternatively be patterned at other
points in the fabrication process, for example, prior to
implantation and/or diffusion of dopants into the first metal
silicide. Other possible alternatives include forming poly-silicon
over undoped first metal silicide (reacted or deposited), and
implanting dopants through the poly-silicon and into the underlying
first metal silicide, with or without subsequent diffusion
annealing.
[0057] Referring now to FIGS. 3, 6, and 7A-7O, another possible
implementation involves doping one or both of a first metal and a
first silicon prior to annealing to form the first metal silicide.
This approach may advantageously facilitate location of dopants at
a desired concentration at the bottom of the first metal silicide
near the interface with the gate dielectric. For example, one
exemplary implementation illustrated and described below provides
for formation of a doped silicon overlying a deposited metal, where
an anneal process concurrently diffuses dopants from the silicon
into the metal while reacting the metal with the silicon to form
the first metal silicide over the gate dielectric. In another
alternative implementation, dopants are implanted through the
silicon and into the underlying metal, after which a post-implant
anneal is performed to react the metal and the silicon and thereby
form the first silicide. The gate structures may be patterned at
any suitable point in these implementations, such as before or
after the anneal steps.
[0058] FIGS. 7A-7L illustrate another exemplary CMOS device 502 at
various stages of fabrication processing generally according to the
method 100 above. Fabrication of the NMOS and PMOS gate structures
in the device 502 (e.g., 110 in FIG. 3) is illustrated in further
detail in FIG. 6. A first metal 511 (e.g., refractory metal such as
Mo, W, Ti, Ta, etc.) is deposited over the gate dielectric, along
with an overlying silicon 524, wherein dopants are introduced into
one or both of these materials prior to or concurrently with
silicidation thereof. The completed transistor gates include first
and second metal silicides 512 and 544a and an intervening silicon
524. FIGS. 7M-7O illustrate an alternate device 502a having gate
structures with first and second metal silicides in a silicide
bilayer with no intervening silicon in accordance with the
invention.
[0059] The device 502 (FIG. 7A) includes a silicon substrate
semiconductor body 504 with a p-well 506, an n-well 508, and field
oxide isolation structures 510 formed during front-end processing
(104 in FIG. 3). A gate dielectric 516 is then formed (106 in FIG.
3) over the surface of the substrate 504 in NMOS and PMOS active
regions of the device 502, which can be a thermally grown
SiO.sub.2oxide 516 or any other dielectric material formed by any
suitable technique. Gate fabrication then proceeds in accordance
with processing indicated generally as 110b in FIG. 6. In FIG. 7B,
a refractory metal 511 (e.g., Mo, W, Ti, Ta, etc.) is deposited
(e.g., 402 of FIG. 6) over the gate dielectric 516 in NMOS and PMOS
regions of the device 502 via a deposition process 514, for
example, sputter deposition or other suitable process. The first
metal 511 can alternatively be a non-refractory metal and may be
deposited or otherwise formed to any suitable thickness within the
scope of the invention. In FIG. 7C, a layer of silicon 524 is
deposited over the first metal 511 using a CVD deposition process
526 (404 in FIG. 6) to any desired thickness. In one possible
implementation, the deposition process 526 is preferably adjusted
such that the metal layer 511 and the deposited silicon layer 524
do not react to form a silicide layer at this process step.
[0060] Doping of the ultimately formed first metal silicide may be
accomplished using any suitable technique within the scope of the
invention, including direct implantation or diffusion ("DIRECT
IMPLANT" or "DIFFUSION" at 406 in FIG. 6). In the exemplary device
502, diffusion is employed (e.g., 408-414 in FIG. 6). In FIG. 7D, a
mask 521 is formed covering the PMOS regions and exposing the NMOS
regions, and an implantation process 523 provides n-type dopants to
the NMOS silicon, thereby creating n-doped silicon 524a over the
first metal 511 (408 in FIG. 6). In FIG. 7E, the mask 521 is
removed and a mask 520 is formed to cover the NMOS regions and
expose the PMOS regions. An implantation process 522 is performed
to introduce p-type dopants into the silicon in the PMOS region
(410 in FIG. 6), thus creating p-doped silicon 524b over the first
metal 511 in the PMOS region and the mask 520 is then removed. In
FIG. 7F, an anneal process 525 is performed (412 in FIG. 6) to
diffuse n and p-type dopants from the NMOS and PMOS silicon 524a
and 524b, respectively, into the underlying first metal 511, and to
concurrently react the silicon 524 and the metal 511 to form doped
first metal silicide 512 in the NMOS and PMOS regions. As a result,
n and p-doped first metal silicides 512a and 512b are created in
the NMOS and PMOS regions, respectively, as shown in FIG. 7F.
[0061] A mask 528 is formed in FIG. 7G and the silicon 524a, 524b
and the first metal silicide 512 are patterned (414 in FIG. 6) via
an etch process 530 to define patterned NMOS and PMOS gate
structures, after which the mask 528 is removed. Alternatively, the
layers can be patterned before annealing, although the illustrated
example advantageously avoids gate structure shape changes during
annealing by patterning after the anneal. In FIG. 7H, the patterned
gate structures are encapsulated (120 In FIG. 3) with an
encapsulation structure 532, and HDD implants 534 are performed
(122 in FIG. 3) to provide p-type dopants in prospective PMOS
source/drains 536 and n-type dopants in prospective NMOS
source/drains 538. Sidewall spacers 540 are formed in FIG. 7I along
the lateral sidewalls of the patterned gate structures (124 in FIG.
3), as well as a mask 520 covering the NMOS region. A p-type
source/drain implant 542a is performed (126 in FIG. 3) to implant
p-type impurities into the PMOS region, further defining the PMOS
source/drains 536 and providing additional dopants to the PMOS gate
stack silicon 524b.
[0062] In FIG. 7J, the mask 520 is removed and a mask 527 is formed
over the PMOS region. An n-type source/drain implant 542b is
performed (128 in FIG. 3) to implant arsenic, phosphorus or other
n-type impurities into the NMOS region, thus further defining the
NMOS source/drains 538 and providing additional dopants to the NMOS
gate stack silicon 524a. In FIGS. 7K and 7L, a second metal
silicide 544a is formed (130 in FIG. 3) above the doped first metal
silicide 512 and the silicon 524, and source/drain silicide
contacts 544b are formed over the source drains 536 and 538. In
FIG. 7K, a layer of nickel 541 is deposited over the silicon 524
and over the doped source/drains 536 and 538 of the substrate 504.
An anneal 543 is performed to react the nickel 541 with the silicon
524 of the gate stacks, thereby forming a second metal silicide
544a above the remaining silicon 524 as seen in FIG. 7L, wherein
the anneal 543 concurrently reacts the nickel 541 with the
source/drain silicon 536 and 538 to form nickel silicide
source/drain contacts 544b. Thereafter, any remaining (e.g.,
unreacted) nickel 541 is removed. Interconnect metalization and
other back-end processing (not shown) is thereafter performed to
complete the fabrication of the device 502 (132 in FIG. 3). As with
the above examples, the second metal silicide 544a and the
source/drain silicide contacts 544b may alternatively be formed by
separate processes within the scope of the invention.
[0063] In FIGS. 7M-7O, an alternative device 502a is illustrated,
having MOS gate structures comprising first and second metal
silicides 512 and 544a in a silicide bilayer with no intervening
silicon. In FIG. 7M, a first metal 511 is formed over the gate
dielectric 516 and silicon 524 is deposited over the first metal
511. Following gate stack and source/drain fabrication as described
above (e.g., including reaction of the metal 511 with the silicon
524 to form the first silicide 512), FIG. 7N illustrates deposition
of nickel 541 over the remaining silicon 524 of the patterned NMOS
and PMOS gate structures and over the source/drains 536, 538. An
anneal 543 is performed to react the nickel 541 with the silicon
524 of the gate stacks, thereby forming a second metal silicide
544a, wherein the anneal consumes substantially all of the silicon
524 in the gate structures, as shown in FIG. 7O.
[0064] Referring again to FIG. 6, as an alternative to diffusion,
the first metal 511 and/or the overlying silicon 524 may be doped
by direct implantation (e.g., "DIRECT IMPLANT" at 406 in FIG. 6)
following silicon formation within the scope of the invention. At
420, n-type dopants are implanted through the silicon and into the
underlying first metal in the NMOS region, using any suitable
masking and implantation processes. At 422, p-type dopants are
implanted through the silicon and into the underlying first metal
in the PMOS region, after which a post implant anneal is performed
at 424 to react the silicon and the doped first metal, thus
creating doped first metal silicide. The silicon may be partially
or entirely consumed in the silicidation anneal at 424 within the
scope of the invention. Thereafter, the first metal silicide and
any remaining silicon are patterned at 426, and further fabrication
processing is performed as described above.
[0065] Although the invention has been illustrated and described
with respect to one or more implementations, alterations, and/or
modifications may be made to the illustrated examples without
departing from the spirit and scope of the appended claims. In
particular regard to the various functions performed by the above
described components or structures (assemblies, devices, circuits,
systems, etc.), the terms (including a reference to a "means") used
to describe such components are intended to correspond, unless
otherwise indicated, to any component or structure which performs
the specified function of the described component (e.g., that is
functionally equivalent), even though not structurally equivalent
to the disclosed structure which performs the function in the
herein illustrated exemplary implementations of the invention. In
addition, while a particular feature of the invention may have been
disclosed with respect to only one of several implementations, such
feature may be combined with one or more other features of the
other implementations as may be desired and advantageous for any
given or particular application. Furthermore, to the extent that
the terms "including", "includes", "having", "has", "with", or
variants thereof are used in either the detailed description and
the claims, such terms are intended to be inclusive in a manner
similar to the term "comprising".
* * * * *