Semiconductor device manufacturing: process – Coating with electrically or thermally conductive material – To form ohmic contact to semiconductive material
Reexamination Certificate
1998-08-03
2003-04-01
Chaundhuri, Olik (Department: 2814)
Semiconductor device manufacturing: process
Coating with electrically or thermally conductive material
To form ohmic contact to semiconductive material
C438S686000
Reexamination Certificate
active
06541375
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention pertains to methods and apparatus involving smooth electrodes and thin film ferroelectrics for use in integrated circuits. More particularly, a bottom electrode is DC-sputter deposited in a special carrier gas mixture to improve the memory retention characteristics of a ferroelectric capacitor.
2. Statement of the Problem
Thin film ferroelectric materials are used in a variety of nonvolatile random access memory devices. For example, U.S. Pat. No. 5,600,587 issued to Koike teaches a ferroelectric nonvolatile random access memory using memory cells consisting of a ferroelectric capacitor and a switching transistor. U.S. Pat. No. 5,495,438 issued to Omura teaches a ferroelectric memory that is formed of ferroelectric capacitors connected in parallel. The capacitors have ferroelectric materials of different coercive field values and, consequently, can use or store multi-value data. U.S. Pat. No. 5,592,409;issued to Nishimura et al. teaches a nonvolatile memory including a ferroelectric layer that is polarized by the impressed voltage between two gates. The polarization or memory storage state is read as a high or low current flow across the ferroelectric layer, which permits nondestructive readout. U.S. Pat. No. 5,539,279 issued to Takeuchi et al. teaches a high speed one transistor one capacitor ferroelectric memory that switches between two modes of operation including a dynamic random access memory (“DRAM”) mode and a ferroelectric random access memory (“FERAM”) mode.
Ferroelectric memories are nonvolatile because the ferroelectric materials polarize in the presence of an applied field and retain the polarization even after the applied field is removed.
FIG. 1
depicts an ideal polarization hysteresis curve
100
for ferroelectric thin films. Side
102
of curve
100
is produced by measuring the charge on a ferroelectric capacitor while changing the applied field from a positive value to a negative value. Side
104
of curve
100
is produced by measuring the charge on the ferroelectric capacitor while changing the applied field E from a negative value to a positive value. The points −E
c
and E
c
are conventionally referred to as the coercive field that is required to bring polarization P to zero. Similarly, the remanent polarization Pr or −Pr is the polarization in the ferroelectric material at a zero field value. The Pr and −Pr values ideally have the same magnitude, but the values are most often different in practice. Thus, polarization measured as 2Pr is calculated by adding the absolute values of the actual Pr and −Pr values even though these values may differ in magnitude. The spontaneous polarization values Ps and −Ps are measured by extrapolating a linear distal end of the hysteresis loop, e.g., end
106
, to intersect the polarization axis. In an ideal ferroelectric, Ps equals Pr, but these values differ in actual ferroelectrics due to linear dielectric and nonlinear ferroelectric behavior. A large, boxy, substantially rectangular central region
108
shows suitability for use as a memory by its wide separation between curves
102
and
104
with respect to both coercive field and polarization.
Ferroelectric memories are fast, dense, and nonvolatile. Even so, ferroelectric memories do not enjoy widespread commercial use, in part, because the polarization of a thin film ferroelectric material degrades with repeated use. Actual thin film ferroelectrics do not perform as ideal ferroelectrics. Deviation from the ideal behavior of
FIG. 1
is observed as ferroelectric imprint and fatigue. These deviations are so common and severe that it is nearly impossible to find thin film ferroelectrics which meet commercial requirements. The best materials for integrated ferroelectric devices are switched using a coercive field that can be obtained from conventional integrated circuit operating voltages, i.e., three to five volts (“V”). The materials should have a very high polarization, e.g., one exceeding twelve to fifteen micro coulombs per square centimeter (“&mgr;C/cm
2
”) determined as 2Pr, to permit the construction of memories having sufficient densities. Polarization fatigue should be very low or nonexistent over hundreds of millions of switching cycles. Furthermore, the ferroelectric material should not imprint, i.e., the hysteresis curve should not shift to favor a positive or negative coercive field.
FIG. 2
depicts the effects of environmental stress on hysteresis curve
100
. Curve
200
shows the effect of fatigue on curve
100
. Fatigue reduces the separation between curves
102
and
104
defining central region
108
. Central region
108
progressively becomes smaller and smaller with additional fatigue. This change in separation is primarily due to the creation of point charge defects arising in the ferroelectric material as a consequence of polarization switching together with the associated screening effect of the charge defects on the applied field. Thus, fatigue causes the ferroelectric material to wear out over time due to repeated polarization switching.
U.S. Pat. No. 5,519,234 issued to Araujo et al. teaches that the fatigue problem of curve
200
is substantially overcome by the use of layered superlattice materials, such as the “layered perovskite-like” materials described in Smolenskii et al. “Ferroelectrics and Related Materials,” Gordon and Breach (1984). The use of thin film layered superlattice materials in integrated circuits was unknown prior to Dr. Araujo's work. The layered superlattice materials are reported to provide a thin film ferroelectric material wherein the polarization state may be switched up to at least 10
9
times with less than thirty percent fatigue. This level of fatigue endurance provides a significant advance in the art because it is at least about three orders of magnitude better than the fatigue endurance of other ferroelectrics, e.g., lead zirconium titanate (“PZT”) or lead lanthanum zirconium titanate (“PLZT”). Prior layered superlattice material work has been done primarily with the use of a Pt/Ti bottom electrode and layered superlattice material films on the order of 1800 Å thick. The titanium is used as an adhesion layer to prevent peeling of the electrode from the substrate.
According to section 15.3 of the Smolenskii book, the layered perovskite-like materials or layered superlattice materials can be classified under three general types:
(A) compounds having the formula A
m−1
Bi
2
M
m
O
3m+3
, where A=Bi
3+
, Ba
2+
, Sr
2+
, Ca
2+
, Pb
2+
, K
+
, Na
+
and other ions of comparable size, and M=Ti
4+
, Nb
5+
, Ta
5+
, Mo
6+
, W
6+
, Fe
3+
and other ions that occupy oxygen octahedra;
(B) compounds having the formula A
m+1
M
m
O
3m+1
, including compounds such as strontium titanates Sr
2
TiO
4
, Sr
3
Ti
2
O
7
and Sr
4
Ti
3
O
10
; and
(C) compounds having the formula A
m
M
m
O
3m+2
, including compounds such as Sr
2
Nb
2
O
7
, La
2
Ti
2
O
7
, Sr
5
TiNb
4
O
17
, and Sr
6
Ti
2
Nb
4
O
20
.
Smolenskii observed that the perovskite-like layers may have different thicknesses, depending on the value of m, and that the perovskite AMO
3
is in principal the limiting example of any type of layered perovskite-like structure with m=infinity. Smolenskii also noted that if the layer with minimum thickness (m=1) is denoted by P and the bismuth-oxygen layer is denoted by B, then the type I compounds may be described as . . . BP
m
BP
m
. . . . Smolenskii further noted that if m is a fractional number then the lattice contains perovskite-like layers of various thicknesses, and that all the known type I compounds are ferroelectrics.
Despite the tremendous improvements in low fatigue ferroelectrics attributable to layered superlattice materials, there remains an imprint problem that is typified by curve
202
of FIG.
2
. Curve
202
shows that environmental stresses can imprint curve
100
by shifting it to
Arita Koji
Hayashi Shinichiro
Chaundhuri Olik
Patton & Boggs LLP
Peralta Ginette
LandOfFree
DC sputtering process for making smooth electrodes and thin... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with DC sputtering process for making smooth electrodes and thin..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and DC sputtering process for making smooth electrodes and thin... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3003547