HIGH DENSITY LOCAL INTERCONNECT IN A SEMICONDUCTOR CIRCUIT USING METAL SILICIDE
This invention relates in general to semiconductor circuits and in particular in the performing of high density local interconnect in semiconductor circuits using metal suicides.
With the advent of very large scale integrated circuits, it is desirable to reduce the space required for local interconnections between active circuit elements such as transistors which are located in a semiconductor material adjacent to one another. Conventional local interconnects employ one layer or two layers of metal to interconnect adjacent active circuit elements. Where two layers of metal are used, the two layers are stacked one above the other on top of the substrate of the integrated circuit medium. The two layers are usually separated by an insulating layer except at locations where they are intentionally connected to form circuit connections. One layer metal process is rather restrictive for the local interconnection technology because metal is also used for global interconnection in the integrated circuit. This results in a significant medium. The two-layer metal process is somewhat difficult to control because of surface topology and restriction of low temperature processing after first layer metal interconnect.
Polysilicon has also been used for constructing interconnecting conductor lines for integrated circuits. It is stable at high temperatures, can be oxidized to form silicon dioxide thereon, and is
suitable for etching fine lines. One disadvantage of polysilicon is its relatively high electrical resistance. To reduce resistance, a layer of silicon- rich metal silicide such as tungsten silicide is deposited on a doped polysilicon layer to form a metal silicide/polysilicon sandwich layer having a low resistivity. Such a scheme is described, for example, in U.S. patent 4,443,930, to Hwang et al. Such a sandwich structure is commonly call a polycide. Hwang et al. proposed to use the polycide type sandwich structure for forming the gates of transistors and for forming the interconnecting conductors for integrated circuit devices. In order for the polysilicon layer to act as a conductor, the polysilicon must be doped to become either n type or p type. Hwang et al. described an interconnect scheme apparently possible to be used only for NMOS type devices.
Because of it low power requirements, CMOS type devices have superseded NMOS and other types of integrated circuit implementations in many applications. It is therefore desirable to provide an interconnect scheme which can be used for interconnecting CMOS type devices in integrated circuits.
SUMMARY OF INVENTION One aspect of the invention is directed towards a semiconductor circuit comprising a body of silicon semiconductor material where the body has a plurality of CMOS circuit devices. The circuit further comprises a metal silicide layer in or on the body interconnecting at least two of the devices.
Another aspect of the invention is directed toward a method for interconnecting devices in a body of a silicon semiconductor material where the body includes circuit devices. The method comprises forming a polysilicon layer, a metal silicide layer, and an
electrically insulating layer separating the polysilicon and metal silicide layers at at least one location to interconnect the devices.
BRIEF DESCRIPTION OF THE DRAWINGS Figure 1 is a cross-sectional view of a semiconductor wafer having CMOS devices therein onto which layers including a polysilicon layer are grown or deposited to illustrate the invention.
Figure 2 is a cross-sectional view of the wafer of Figure 1 and, in addition, of a doped CVD oxide layer and an amorphous silicon layer grown or deposited on top of the wafer of Figure 1. Amorphous silicon layer connects to P+ silicon, N+ silicon and polysilicon at the desired location. Figure 3 is a cross-sectional view of the wafer of Figure 2 and of a photoresist mask for doping a portion of the amorphous silicon layer by N+ implantation.
Figure 4 is a cross-sectional view of the wafer of Figure 2 and in addition a photoresist mask for doping another portion of the amorphous silicon layer by P+ implantation.
Figure 5 is a cross-sectional view of the wafer of Figure 2 and in addition a metal silicide local interconnect layer on selected areas of the wafer to illustrated the preferred embodiment of the invention. Figure 6 is a cross-sectional view of the wafer of Figure 5 and, in addition, an undoped CVD oxide layer, a doped glass layer, and a metal interconnect layer on the wafer of Figure 5 forming a completed and locally interconnected circuit to illustrate the preferred embodiment of the invention.
DETAILED DESCRIPTION OF THE INVENTION Figure 1 is a cross-sectional view of a silicon wafer in which a pair of transistors forming a CMOS pair has been fabricated. Thus wafer 10 comprises a substrate 12 containing an N-well 14 and a P-well 16. The N- and P-wells are doped appropriately to form P+ regions 18, 20 and N+ regions 22, 24. A thick field oxide layer 32 and a thin gate oxide layer 34 are then provided on selected areas of the wafer as shown in Figure 1. A polysilicon layer is also provided on selected areas on the oxide layers as shown in Figure 1. Wafer 10 is composed of the above described components. As shown in Figure 1, a portion 42 of the polysilicon layer, together with P+ regions 18, 20, are part of a P-CH transistor 50. Another portion 44 of the polysilicon layer is formed on top of a portion of the field oxide 32 as shown in Figure 1 to interconnect devices on the wafer. Yet another portion 46 of the polysilicon layer, together with N+ regions 22, 24, form a part of a N-CH transistor 52. Transistors 50, 52 together form a CMOS pair. The methods for fabricating the various regions and layers in wafer 10 shown in Figure 1 are known to those skilled in the art and will not be described in detail herein. The process for manufacturing an integrated circuit employing the high density local interconnect using metal silicide is illustrated in a sequence of the following numbered steps:
1. As explained above, standard integrated circuit wafer fabrication techniques are used to make the wafer 10 of Figure 1, until the drain and source regions 18-24 have been implanted and annealed.
2. Steps 2-4 are illustrated in reference to Figure 2 which is a cross-sectional view of the wafer 10 of Figure 1 and, in addition, of an undoped CVD oxide and an amorphous silicon layer as shown in Figure 2.
First a layer of undoped CVD oxide is deposited onto wafer 10 by means of a conventional chemical vapor deposition method and annealed at high temperatures.
3. A photomasking step is performed to define via cuts in certain regions to connect N+, P+ regions 18-24, and portions of the polysilicon layer 44. The CVD oxide layer 62 is etched away from these regions.
4. A thin layer 64 of amorphous silicon is deposited on the entire surface of the wafer.
5. Steps 5 and 6 are illustrated in reference to Figure 3 which is a cross-sectional view of wafer 10 of Figure 1 together with the layers added shown in Figure 2 and, in addition, of a masking layer 72 to illustrate the invention. The photomasking step is performed to provide a photoresist layer 72, leaving open certain regions, such as those between points 72a, 72b; and 72c, 72d.
6. High dose N+ ions are implanted into the wafer at regions that have been left unmasked by layer
72, where the energy of the implant is adjusted such that the peak of the implant is in the silicon substrate but near the interface of the thin amorphous silicon film 64 and N+ regions 22, 24 or polysilicon 44. After the implant, the photoresist layer 72 is removed by plasma ash and a chemical process using H2So4/H202.
7. Steps 7 and 8 are illustrated in reference to Figure 4 which is a cross-sectional view of the wafer of Figure 3 except that photoresist 72 is replaced by a different photoresist layer 74, leaving unmasked the region between points 74a, 74b. After photoresist 72 has been removed, another photomasking step is performed to provide photoresist layer 74, leaving unmasked certain regions. 8. Now high dose P+ ion implantation is performed. The energy of the implant is adjusted such
that the peak of the implant is in the silicon substrate that near the interface of the thin amorphous silicon film 64 and P+ regions 18, 20. After the implant, photoresist 74 is removed by plasma ash and a chemical process using H2So4/H202.
9. Step 9 is illustrated in reference to Figure 5 which is a cross-sectional view of the wafer of Figure 4 except that the photoresist 74 has been removed and a metal silicide layer 82 is shown instead. A film of metal silicide such as tungsten silicide is deposited after the removal of photoresist 74. This is accomplished by a photomasking step followed by a plasma etching of silicide at desired regions. This silicide film interconnects the P+ region 20 to N+ region 22 and polysilicon strip 44 through the amorphous silicon layer 64.
10. Steps 10-12 are illustrated in reference to Figure 6 which is a cross-sectional view of the wafer of Figure 5 and, in addition, of an undoped CVD oxide layer, a doped layer of glass, and a metal interconnect layer. First a layer of undoped CVD oxide layer 92 is deposited. This is followed by deposition of a doped glass layer 94. A very short high temperature heat cycle is performed using rapid thermal annealing process (RTA) . This process is used to activate the dopings, lower the sheet resistance of the metal silicide, and flow the glass film.
11. A photomasking step is now performed to define connect holes where metal connects to the metal silicide, polysilicon and N+/P+ silicon regions.
12. This is followed by standard metal deposition and definition step whereby metal layer 96 interconnects the metal silicide, polysilicon, and N+/P+ silicon regions. In Figure 6, however, the metal layer is shown only connected to P+ and N+ regions through the silicide and amorphous silicon layers; it will be
understood that the metal layer may be used to interconnect to the polysilicon portion 44 as well through the silicide and amorphous silicon layer if desired. The high density local interconnect system for interconnecting active circuit elements of the semiconductor circuit 100 is then complete.
From the above description, it will be noted that there are three layers of interconnects in the final product wafer 100 of Figure 6. First, the polysilicon layer forms strips 42, 44, 46, where strips 42, 46 serve as gates of transistors 50, 52 and strip 44 is a polysilicon interconnect for connecting other devices (not shown) of the wafer. The second layer of interconnect is formed by the metal silicide layers 82. However, before the interconnect system using layer 82 is described, it is first necessary to describe in more detail the composition of the amorphous silicon layer 64.
In reference to Figure 3, it will be noted that only strips of silicon layer 64 between points 72a,
72b and between points 72c, 72d are implanted by the N+ implants so that only these portions are electrically conductive. Since the polysilicon strip 44 also has ϊ~+ implants, the N+ implants in layer 64 in contact with polysilicon layer 44 are compatible there with. In reference to Figure 4, only the portion of layer 64 between points 74a, 74b are implanted by P+ implants, so that only such section and those described above implanted with N+ implants are electrically conductive. In reference to Figure 5, the portion of the amorphous silicon layer 64 immediately above the polysilicon strips 42, 46 have been etched away. Thus, in reference to Figure 6, the portion of layer 64 in contact with region 18 electrically connects region 18 to metal contact 96; since such portion of layer 64 has P+ implants, it is compatible with region 18. Similarly,
the portion of layer 64 in contact with region 20 is compatible therewith and electrically connects the region to metal silicide region 82. And the N+ doped portions of layer 64 are compatible with regions 22, 24 and polysilicon strip 44 and electrically connect these portions to the silicide layer.
The portion of layer 64 in contact with polysilicon strip 44 is separated from the portion of layer 64 connected to regions 20 by an undoped portion between points 72a, 74a in reference to Figures 3, 4 and 6. It will be noted that the portion of layer 64 in contact with strip 44 has N+ implants, so that, by separating such portion from the P+ implanted portion of layer 64 in contact with P+ region 20 by an undoped portion, a PN junction is prevented. The electrical connection between polysilicon layer 44 and region 20 is accomplished through the metal silicide layer 82. The portion of amorphous silicon layer 64 in contact with region 22 has been implanted with N+ implants, where such portion of layer 64 is isolated from the portion of the same layer connected to polysilicon strip 44 by an undoped portion between points 72b, 72c. By selectively implanting ions in the manner described in reference to Figures 3 and 4, contact channels are provided to N+ and P+ division regions in the substrate through the amorphous silicon layer without creating PN junctions; this also permits selective contacts to the N+, P+ regions and the polysilicon layer through the metal silicide layer above. The objectives of the invention are therefore achieved. A local interconnect system employing a metal silicide layer is described herein. Since the silicide layer can be fabricated at higher density than the metal layer in the conventional one or two metal layer interconnect systems, the silicon area required for the interconnect is reduced so that the die size can be
reduced for the same number of devices, or the number of devices fabricated on the same size die can be increased.
As described above, the amorphous silicon layer 64 is separated from the polysilicon layer by an undoped CVD oxide layer 62 at most locations so that the metal silicide layer can perform an interconnecting role entirely independent from the polysilicon layer. If desired, portions of the silicon layer 64 may be made to contact the polysilicon interconnect 44, as shown in the figures, to permit selective connections between the metal silicide and the polysilicon layers.
In the example described above, the amorphous silicon layer 64 is substantially identical in coverage to the metal silicide layer 82. It will be understood, however, that this is not necessary, so that the amorphous silicon layer 64 may be used for connecting the polysilicon strip 44 to the N+ region 22, for example. In other words, if the portion of layer 64 between points 72b, 72c is also implanted with N+ implants, then polysilicon strip 44 will be electrically connected to region 22 through the silicon layer 64, even if a portion of the silicide layer immediately above has been removed. In this manner, the amorphous silicon layer 64 performs an interconnect function independent from the metal silicide layer 82. By leaving a portion of the amorphous silicon undoped, it is always possible to prevent the formation of a PN junction. Preferably, the metal silicide layer is composed of a refractory metal silicide, so that the layer is stable at high temperatures.
When the invention has been described in reference to the preferred layouts and methods, it will be . understood that various modifications may be made without departing from the scope of the invention which is to be limited only by the appended claims.