Patent Application
20060226491
The invention relates to stacked circuits, and more particularly to a stacked circuit having an inverted device and to a stacked circuit having an inverted device with a different crystal orientation.
SOI (Silicon On Insulator) technology reduces or eliminates bulk CMOS latch-up problems. Additionally, SOI technology reduces junction capacitance and allows circuits to operate at higher speeds. Accordingly, SOI technology allows a higher circuit density to be achieved on a silicon wafer.
A multiple layer device structure can be created using SOI technology. Each level of a multilevel structure created using SOI may be interconnected using vertical plugs. Metal connections can be used at any level. In a multilevel device, NMOS and PMOS devices may each be restricted to a separate level, or NMOS and PMOS devices may be intermixed on a single level. However, having multiple levels where each level requires a set of separate metal connections may result in a large number of metal connections, increasing the complexity and costs of the chip. Furthermore, a layer of a multilevel device may inhibit cooling of adjacent layers by limiting heat transfer therefrom.
For some types of semiconductor devices, it is advantageous to have multiple layers of similar semiconductor material, where each layer has a crystal orientation different from the adjoining layers. When multi-layer devices are formed from a common semiconductor layer they will have the same crystal orientation from one level to the next. This limitation on the crystal orientation across multiple levels of an integrated circuit results from constraints imposed by the fabrication process.
In particular, semiconductor material is traditionally grown from a pre-existing layer of semiconductor material. During the material growth process, the atoms of the newly formed layer have a strong tendency to orient themselves to the pre-existing substrate's crystal structure as they are laid down. Thus, it becomes very difficult to create a new layer of similar semiconductor material having a crystal orientation which is different from the underlying layer.
In a first aspect of the invention, a method of forming a semiconductor device assembly includes forming a first semiconductor device having a first crystal region and a second semiconductor device having a second crystal region. The first semiconductor device is placed in contact with the second semiconductor device.
In another aspect of the invention, the method includes forming a first semiconductor device having a first crystal orientation in a substrate and forming a second semiconductor device having a second crystal orientation in a substrate. The first semiconductor device is placed in contact with the second semiconductor device.
In yet another aspect of the invention, the semiconductor device assembly includes an NMOS device having a first crystal orientation and a PMOS device having a second crystal orientation inverted on a surface of the NMOS device.
In embodiments of the invention, a first CMOS device, such as a pFET, is formed having a first crystal orientation in the active region of the device, and a second CMOS device, such as an nFET, is formed having a second crystal orientation in the active region of the second device. After both devices are formed, the second device is flipped over and bonded to the first device with a bonding layer such as SiO2 forming an inverted stacked or an inverted multi-layered arrangement. At lease one of the contacts of the first device is aligned with the corresponding contact of the second device, and the two contacts are brought into electrical contact with one another during bonding. Because the two devices are bonded in an inverted stacked arrangement, the devices are easily interconnected along the boundary between the two devices. Thus, it is possible to form stacked layers of CMOS devices where each layer has either the same or a different crystal orientation in the active region as a layer below it and/or a layer above it, and the combination of devices has a reduced number of connections.
Referring to
The upper surface of the first Si layer 120 is planarized after the source/drain regions 115 are formed using any of the methods well known in the art to planarize a Si layer. A second SiO2 layer 125 is formed over a surface of the first Si layer 120. A polysilicon gate 130 is formed on top of the second SiO2 layer 125 such that a portion of the second SiO2 layer 125 forms a gate oxide for the first gate 130 of the first device 100.
A CA gate contact 140 is formed on the gate 130 and a metal (M1) contact 155 is formed on the gate contact 140. Source/drain CA contacts 135 and 145 are formed on the source/drains 115. Source/drain M1160 and 150 and gate metal 155 are formed thereon, respectively.
A dielectric 165 is formed over the top of the first device 100, and the top of the dielectric 165 is planarized. Additionally, the dielectric 165 may be etched so that the gate metal 155 and the source/drain metal 150 are slightly higher than the surrounding dielectric 165 to facilitate making positive contact thereon during a later bonding process. Such an etching process may include etching the top surface of the dielectric 165 with an etchant which selectively removes dielectric material, but does not remove the material of the gate metal 155 and source/drain metals 150. Alternatively, the gate metal 155 and drain/source metal 150 may be protected by a mask while the top surface of the dielectric 165 is etched.
Another method to raise the gate metal 155 and source/drain metal 150 above the top surface of the dielectric 165 may include planarizing the dielectric 165, a gate metal 155 and source/drain metal 150, and building up the gate metal 155 and source/drain metal 150 to stand slightly above the top surface of the dielectric 165. Such a build-up process may include masking the surrounding dielectric 165 to leave the gate metal 155 and source/drain metal 150 exposed, and depositing or growing a suitable conductive material on the exposed gate metal 155 and source/drain metal 150.
Referring to
Trenches are etched into the second Si layer 207 and filled with an insulator such as an oxide using any of the methods well known in the art for forming and filling a trench to create shallow trench isolations 210. Source/drain regions 215 are formed on each side of an active region 225. The source/drain regions 215 of the second device 200 are formed by any of the methods well known in the art for forming source/drain regions 215. A portion of the N-well 212 lies between the shallow trench isolation 210 and the source/drain regions 215.
After the source/drain regions 215 are formed in the second device 200, a gate oxide 225 is formed on top of the active region of the second Si layer 220. The gate oxide 225 may be formed by any of the techniques well known in the art for forming a gate oxide. After the gate oxide 225 is formed, a polysilicon gate 230 is formed over the gate oxide 225. A CA contact 240 is formed on the poly gate 230 and a metal (M1) 255 is formed on the CA contact 240. Source/drain M1 contacts 235 and 245 are formed on the source/drains 215. Source/drain metals 260 and 250 are formed thereon, respectively.
A dielectric 265 is formed over the top of the first device. As discussed with the first device, 100, the second device is formed so that the gate metal 255 and source/drain metal 250 protrude slightly above the dielectric 265 surface. For example, the top of the dielectric is planarized and then etched to leave the tops of the gate metal 255 and a source/drain metal 250 slightly protruding above the surrounding surface, depending on the connection.
Referring to
The bonding layer 305 bonds the second device 200 to the first device 100 using methods and materials well known in the art for bonding one device to another. The bonding layer 305 may include any insulating substance capable of bonding a first semiconductor device to a second semiconductor device, such as, for example, a nanocleave method. Other examples of bonding layers include contact adhesive and expoxies. Further examples of bonding layers include certain oxides used for bonding purposes well know in the art.
The bonding layer 305 should be applied so as not to interfere with the electrical connection or gate interconnect 310 between the two gate metals 155 and 255 and the electrical connection or source/drain interconnect 315 between the two source/drain metals 150 and 250. This may include applying the bonding layer 305 so it is clear of the gate metals 155 and 255 and the source/drain metals 150 and 250. Where appropriate, the bonding layer may also be applied in a viscous state so that it flows from the gate metal, 155 and 255, and source/drain metal, 150 and 250, surfaces when the surfaces make contact with one another.
As an alternative to a bonding layer, the first device 100 may be clamped to the second device 200 with an external clamping force. The external clamping force may obviate the need to use a bonding layer between the two devices (also represented as reference numeral 305).
After the first device 100 and the second device 200 are bonded to one another, metal plugs may be formed to the interconnections, 310 and 315, the first and second devices, 100 and 200. For example, the source/drain interconnection 315 may be connected to a metal plug. Similarly the source/drain of the lower device may be connected to a voltage source Vss. The gate interconnection 310 may be connected to a metal plug to receive a voltage input Vin. The separate source/drain metals, 160 and 260 may be separately connected to respective wire or plugs.
In operation, the inverted multilayer device 300 of
In other words,
Other advantages of the inverted multilayer SOI device include a reduction of about 50% of the area needed to create a circuit and using different crystal orientations for different devices. Additionally, where inverted multilayer devices are symmetrical, photolithography masks may be shared.
While the invention has been described in terms of exemplary embodiments, those skilled in the art will recognize that the invention can be practiced with modifications and in the spirit and scope of the appended claims.
