Semiconductor device manufacturing: process – Making field effect device having pair of active regions... – Having insulated gate
Reexamination Certificate
1993-12-23
2002-03-12
Chaudhari, Chandra (Department: 2813)
Semiconductor device manufacturing: process
Making field effect device having pair of active regions...
Having insulated gate
C438S589000
Reexamination Certificate
active
06355517
ABSTRACT:
The present invention relates to a semiconductor memory and a method for fabricating the same, and more particularly to a semiconductor memory and a method for fabricating the same, which-can increase a memory capacity without increasing the amount of space required for forming the memory.
Since a 1k-bit dynamic random access memory (hereinafter referred to as a dRAM) was first marketed at the beginning of 1970, MOS dynamic memories which is one of the types of semiconductor IC memories have been scaled up four times in memory capacity approximately every three years. A package for accommodating the memory chip primarily uses a 16-pin dual in-line package (DIP) and a cavity size for accommodating the chip is limited. Thus, the size of the memory chip has been increased only about 1.4 times while the integration density has increased four times. Accordingly, a memory cell area per bit has been significantly reduced as the integration density has been increased. Specifically, the memory cell area per bit has been reduced by a factor of three for the increase of the integration density by a factor of four.
Since a capacitance C of the semiconductor memory is represented by C=∈A/t (where ∈ is a dielectric constant of an insulation film, A is an area of a capacitor electrode and t is a thickness of the insulation film between the capacitor electrodes), if the area A is reduced by a-factor of three, the capacitance C is also reduced by a factor of three provided that ∈ and t are kept constant. A signal quantity S derived from a memory capacitor is proportional to a charge quantity Q which is the product of the capacitance C and a voltage V. Thus, if A is reduced, Q is reduced proportionally and the signal quantity S is also reduced accordingly.
A signal to noise ratio S/N decreases as S decreases. This will raise a big problem in circuit operation. Accordingly, the decrease of A is usually compensated by the decrease of t. Thus, as the integration density is increased to 4 kb, 16 kb and 64 kb, a thickness of a typical SiO
2
film is reduced to 100 nm, 75 nm and 50 nm, respectively.
On the other hand, it has been recently found that a charge of approximately 200 fc (femtocoulomb) at maximum is created in a Si substrate by an a particle radiated from a radioactive heavy metal (U, Th, etc.) contained in the package, and this causes an undesirable noise. Accordingly, it is difficult from a standpoint of stable operation to reduce the signal quantity Q below approximately 200 fc. Accordingly, it has been practiced to further reduce the thickness of the insulation film. As a result, another problem of dielectric breakdown of the insulation film has been raised. An electric field for causing dielectric breakdown in the SiO
2
film which is commonly used as the insulation film of the capacitor is 10
7
V/cm at maximum. Thus, an SiO
2
film having a thickness of 10 nm is essentially permanently broken or degraded by an application of a voltage of 10 volts. For long term reliability, it is important to operate the memory at a voltage which is as much smaller than the maximum breakdown voltage as possible.
FIG. 1
shows a configuration of a one-transistor dynamic memory cell which uses an insulated gate field effect transistor (hereinafter referred to as a MOS transistor). It comprises a capacitor
1
for storing a charge and a switching MOS transistor
2
, a drain of the switching MOS transistor
2
is connected to a bit line
3
and a gate thereof is connected to a word line
4
.
In operation, a signal charge stored in the capacitor
1
is read out by the switching transistor
2
. An actual large-scale integration memory is constructed in a memory array by one of the following two major methods.
FIG. 2
shows what is called an open bit line configuration in which bit lines
31
and
32
are arranged on opposite sides of a sense amplifier
5
which differentially senses signals. Only one bit line
31
electrically crosses a word line
41
, and the sense amplifier
5
senses a difference between the signals on the bit lines
31
and
32
.
FIG. 3
shows what is called a folded bit line configuration in which two bit lines
31
and
32
connected to a sense amplifier
5
are arranged in parallel and a word line
41
crosses to two bit lines
31
and
32
.
Preferred embodiments of the present invention to be described later are primarily implemented in the folded bit line configuration although they may be implemented in the open bit line configuration.
In
FIGS. 2 and 3
, one of major performance indices of the memory array is C
S
/C
D
, where C
D
is a capacitance of a parasitic capacitance
6
of the bit line
32
and C
S
is a capacitance of a capacitor
1
-
2
of the memory cell. The S/N ratio of the memory array directly corresponds to C
S
/C
D
. It is thus important to increase the capacitance of the memory cell and reduce the parasitic capacitance C
D
of the bit line in order to raise the S/N ratio.
FIG. 4
shows a plan structure of a memory cell of the folded bit line configuration, and
FIG. 5
shows a sectional structure thereof taken along a line V—V in FIG.
4
. As seen from
FIGS. 4 and 5
, since a capacitor is formed at a portion in an active region
7
(shown in a dumbbell shape) encircled by a thick field oxidization film
11
which is more than 100 nm thick; the active region
7
is covered by a plate
8
. However, the plate
8
is selectively removed (shown at
80
in
FIG. 4
) at an area at which the switching transistor is formed and at an area of a contact hole
9
through which a bit line
3
is connected to a drain (or source)
15
on a Si substrate. Word lines
4
are deposited on the area
80
and the switching transistor
2
is formed there. In some open bit line memories, the switching transistor
2
has no heavily doped region
15
adjacent to the storage capacitor. The term “transistor” will also include such versions.
The semiconductor memory described above is manufactured in the following manner. For the sake of explanation, the transistor is of n-channel type. If a p-channel type is used, the conductivity types of the Si substrate and the diffusion layer are reversed to those in the n-channel type. A field SiO
2
layer
11
which is approximately 100-1000 nm thick is selectively formed on a p-type Si substrate
10
having a resistivity of approximately 10 &OHgr;-cm by a so-called LOCOS technique which uses Si
3
N
4
as an anti-oxidization mask. Then, a gate oxide film
12
which is 10-100 nm thick is formed on the Si substrate
10
by thermal oxidization. Then, the plate
8
typically formed of poly-Si doped with phosphorus P or arsenic As is selectively deposited and a surface of the poly-Si plate
8
is oxidized to form a first interlayer oxide film
13
. Then, the word lines
4
typically formed of poly-Si, Mo silicide or a refractory metal (Mo or W) are deposited and phosphorus P or arsenic As ions are implanted. Thus, n
+
diffusion regions
15
are formed at areas on which the plate
8
and the word lines
4
are not deposited, to form a source and a drain of the switching MOS transistor
2
. Then, a phospho-silicate glass (PSG)
14
is deposited to a thickness of 500-1000 nm by a so-called CVD method which contains phosphorus and the contact hole
9
is formed at the area at which the bit line
3
typically formed of an Al electrode is to be connected to the diffusion layer
15
. Then, the bit line
3
is selectively deposited thereon.
In the memory cell thus fabricated, the area
16
of the memory capacitor
1
naturally decreases as the size of the memory cell itself decreases. Thus, unless the thickness of the gate oxide film
12
is reduced, the capacitance C
S
is reduced as described above and it raises a big problem in the memory operation.
In order to resolve the above problem, it has been proposed by one of the inventors of the present invention to form a narrow groove in the silicon substrate and to form a memory capacitor on the surface of the groove (Japanese Laid-Open Patent Application 51-130178). Since the propo
Kawamoto Yoshifumi
Kure Tokuo
Sunami Hideo
Antonelli Terry Stout & Kraus LLP
Chaudhari Chandra
LandOfFree
Method for fabricating semiconductor memory with a groove does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method for fabricating semiconductor memory with a groove, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for fabricating semiconductor memory with a groove will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2845457