Active solid-state devices (e.g. – transistors – solid-state diode – Non-single crystal – or recrystallized – semiconductor... – Non-single crystal – or recrystallized – active junction...
Reexamination Certificate
2000-07-14
2001-10-23
Pham, Long (Department: 2823)
Active solid-state devices (e.g., transistors, solid-state diode
Non-single crystal, or recrystallized, semiconductor...
Non-single crystal, or recrystallized, active junction...
C257S529000, C257S530000
Reexamination Certificate
active
06307213
ABSTRACT:
BACKGROUND OF THE INVENTION
(1) Field of the Invention
The present invention relates to the fabrication of integrated circuits, and more particularly to a method for making improved fuse structures on semiconductor integrated circuits such as on Random Access Memory (RAM) devices. This method and structure utilize an etch-stop layer to accurately remove the non-uniformly thick multilayer insulation over the fuse areas across the substrate. This allows for reliable and repeatable repair yields when fuses are opened to disable defective circuits and to replace them with redundant circuits, such as for enabling additional rows of memory cells on a RAM device.
(2) Description of the Prior Art
Advances in semiconductor processing technologies, such as high-resolution photolithography and anisotropic plasma etching, are dramatically reducing the feature sizes of semiconductor devices and increasing the device packing density. Unfortunately, as the density of the semiconductor devices increases and the number of discrete devices increases on the chip, the final product yield for many integrated circuit devices (chip yield) can decrease. For example, one circuit device that can experience this increase in yield loss with increased circuit elements is dynamic random access memory (DRAM) currently having 64 megabits of memory on a chip. After the year 2000 the number of memory cells is expected to increase further to about 1 to 4 gigabits, and high final product yield will be difficult to achieve without utilizing cell redundancy and repair yield methods.
One method of overcoming this lower yield on RAM devices is to provide additional rows of memory cells and fusing each row of cells. Currently lasers are used to routinely open connections (fuses) in the multimegabit RAMs, such as in DRAM or SRAM devices, to disable defective rows of memory cells and to modify the address decoder so that spare rows of memory cells are selected instead.
To better appreciate the problem associated with making and selectively opening the more conventional fuse, a schematic cross-sectional view of a fuse structure is depicted on substrate
10
in FIG.
1
. Typically the fuse
14
is formed as part of a first polysilicon layer
14
at the same time the conductively doped polysilicon layer is patterned to form part of the semiconductor devices, such as the FET gate electrodes on device areas and the local interconnections on the field oxide regions
12
. Only the fuse
14
is shown in
FIG. 1
to simplify the drawing and discussion. The patterned first polysilicon layer
14
with the fuse area (also labeled
14
) is then insulated with a first interpolysilicon oxide (IPO) layer
16
, for example using a silicon oxide (SiO
2
) deposited by chemical vapor deposition (CVD). A conductively doped second polysilicon layer (not shown) is then patterned to form the next level of interconnections (such as bit lines for SRAM and DRAM devices), and a second IPO layer
20
is deposited, such as a CVD SiO
2
, as an insulating layer. Each of the IPO layers (layers
16
and
20
) has contact openings (not shown) to provide interlevel electrical connections between conducting layers. When the semiconductor devices are completed, including the local polysilicon interconnections with fuses, the circuit integration is completed by forming a multilevel of metal interconnections. Four levels of metal are depicted in
FIG. 1
, but the number of metal layers varies between about 3 and 6, and is expected to increase further on future product. Each layer of patterned metal interconnections is formed by depositing metal layers, such as an aluminium/copper alloy, and patterned by plasma etching. Each patterned metal layer is insulated with an interlevel dielectric (ILD) layer in which via holes are etched and filled with metal (plugs). For example, as shown in
FIG. 1
, a first metal layer
22
is deposited and patterned to form the patterned first metal layer
22
(M1). An ILD layer
24
is deposited and planarized in which via holes are etched and filled with metal plugs
26
. In like fashion and repeating the process, a patterned second metal layer
32
(M2) with ILD layer
34
and metal plugs
36
are formed, and a third patterned metal layer
42
(M3) with an ILD layer
44
and via holes with metal plugs
46
are formed. After completing the metal interconnections with a patterned fourth metal layer
52
(M4), the multilevel metallurgy is passivated using a silicon nitride
60
, a spin-on glass
62
, and a thicker silicon oxynitride layer
64
.
To access the fuses
14
, a patterned photoresist layer and plasma etching are used to form openings
2
(fuse windows), as depicted by the vertical arrows
4
, to a controlled depth D in the passivation layers
64
,
62
, and
60
, and in the thick multilayer of ILD layers
44
,
34
,
24
, and
20
over the fuses
14
. This opening
2
must be etched to an accurate depth over the fuse to reliably blow (open) the desired fuse, for example, by laser evaporation.
Unfortunately, because of the very thick multilayer of insulating layers, it is difficult to accurately etch the opening
2
to exactly control the thickness H of the remaining insulating layer (SiO
2
) over the fuse
14
. For example, if the thickness of layer
16
is 4000 Angstroms, then variations in the ILD layers and the passivation layers can result in variation of +/−2600 Angstroms in the thickness of the oxide layer
16
in the opening
2
over the fuse
14
. Further exacerbating the problem is the nonuniformity in the ILD layers across the substrate. For example, just a +/−5% variation in the overlying ILD insulating layers and passivation layers (about 8 micrometers) can result in a variation of +/−4000 Angstroms. This makes it very difficult to repeatedly and reliably blow open the required fuses across the substrate
10
.
One method of forming fuses on high performance BiCMOS integrated circuits for programmable devices is described by Keller et al., U.S. Pat. No. 5,457,059. However, the method uses titanium/tungsten fuses for field programmable logic (FPL) and does not address the need for forming reliable and repeatable fuse structures for redundant circuits, such as SRAM and DRAM devices. Another reference for utilizing etch-stop layers to make via holes of various depths is described by Sandhu et al. in U.S. Pat. No. 5,258,096. A method for making buried and shallow contacts using an etch-stop layer is described by Rhodes et al. in U.S. Pat. No. 5,232,874. However, the method for making reliable fuse structures is not addressed.
There is still a strong need in the semiconductor industry to further improve the method for making reliable and repeatable fuse structures for improved repair yields on integrated circuits with redundant circuits, such as on DRAM and SRAM devices and the like.
SUMMARY OF THE INVENTION
It is therefore a principal object of the present invention to provide a method for making an improved fuse structure on semiconductor integrated circuits having multilevels of patterned doped polysilicon and multilayers of metal interconnections having insulating layers therebetween.
Another object of this invention is to provide an etch-stop layer over the fuse portion of the patterned polysilicon layer(s) to accurately etch openings in the multilayers of insulating layers to the etch-stop layer over the fuse areas.
It is another object of this invention to provide fuse structures for Random Access Memory (RAM) devices that improve the repair yield.
Still another objective of this invention is to form these fuse structures without including additional processing steps thereby providing a cost-effective manufacturing process.
The method of this invention begins by providing a semiconductor substrate, typically consisting of a lightly doped single-crystal silicon substrate. Field oxide (FOX) regions are formed on the substrate surface surrounding and electrically isolating devices areas. Semiconductor devices, such as field effect transistors (FETs) used in integrated circuits and
Huang Kuo Ching
Lee Yu Hua
Shih Cheng Yeh
Wu Cheng-Ming
Ying Tse-Liang
Ackerman Stephen B.
Coleman William David
Pham Long
Saile George O.
Taiwan Semiconductor Manufacturing Company
LandOfFree
Method for making a fuse structure for improved repaired... 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 making a fuse structure for improved repaired..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for making a fuse structure for improved repaired... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2597990