The present invention relates to power semiconductor devices and a method for the fabrication thereof.
Conventional power MOSFETs use a thick (4 to 10 μm) top layer of metalization for connection to the source regions thereof due to the large currents that the metal has to conduct during the operation of the device. Because of the thickness of the top metal, wet etching is used to pattern the same during fabrication. The use of wet etching requires the metal design rules to be large. Therefore, multiple gate buses in a conventional power MOSFET consume a large area of the semiconductor die, which could otherwise be used for the active part of the device.
When a power MOSFET is configured for flip-mounting onto conductive pads using solder or the like additional issues further lead to the inefficient use of semiconductor area. For example, the gate pad required for flip-mounting is large compared with a wire-bonded device, which wastes more semiconductor area that could be used for the active region of the device. In addition, the layout of a large source pad required for flip-mounting may restrict the use of multiple gate buses.
In a power semiconductor according to the present invention the thick metal layer in a conventional device is replaced by a thin metal layer (e.g. 1-2 μm) which can be dry etched. The thin metal layer is then patterned to obtain a metallic gate bus that is then preferably encapsulated in a hermetic seal, followed by formation of a thick stress relieving buffer body. The buffer body allows the deposition of a thick (e.g. 4-20 μm) second metal layer to be added by preventing stress-related cracking of the hermetic seal during reliability testing such as temperature cycling. The thick second metal layer, which is thick enough to carry current as required by a power semiconductor device, can be then wet etched.
Advantageously, the thick second metal layer can be thicker than conventional front metal bodies in that no gate buses are required to be patterned out of the same.
Other features and advantages of the present invention will become apparent from the following description of the invention which refers to the accompanying drawings.
Referring to
Metallic gate bus 28 is preferably disposed over and ohmically coupled to a polysilicon gate bus 34 which is connected to gate electrodes 20. Polysilicon gate bus 34 is disposed over an insulation body 36 (e.g. SiO2), which is atop semiconductor body 10 to insulate polysilicon gate bus 34 from semiconductor body 10. Metallic gate bus 28 is coupled to a metallic gate pad 29 (
Referring to
Preferably, first metal layer 26 and metallic gate bus 28 are less than 2 microns thick, while second metal layer 32 is between 4 to 20 microns thick. Furthermore, buffer body 30 is made of polyimide, while hermetic sealant body 38 is a stack including a layer of photo silicate glass 40 (PSG), and a silicon nitride layer 42 between PSG 40 and buffer body 30.
Referring next to
A layer of PSG 40 is then deposited followed by the deposition of a layer of silicon nitride 42, both layers covering first metal layer 26, metallic gate bus 28 and the space between gate bus 28 and first metal layer 26 (see
The following are some of the advantages of having two metal layers according to the present invention particularly, but not limited to, on low voltage power MOSFETs.
Thus, the use of two metal layers according to the present invention allows for better active area usage by allowing the area under the gate pad to be utilized for active cells, which may be especially important for flip-mountable devices that require larger gate pads and allowing large shrinkage in the metal space design rules.
A device according to the present invention may also exhibit lower Rdson resulting from a lower metal spreading resistance due to the thick metal stack of the first and second metal layers. For example, 10% reduction in overall silicon RDson has been shown possible.
A device according to the present invention may further exhibit lower controlled Rg because it can include multiple gate buses (not normally feasible in flip-mountable devices that require large source contacts) in the first metal layer that do not consume much active area due to the tighter design rules
Note that although in the preferred embodiment polyimide is used to form buffer body 30 other materials may be used without deviating from the scope and spirit of the present invention. For example, other organic films such as BCB, or even a thick planarised hard dielectric body, such as a stack of TEOS/SOG/TEOS, can be used instead of polyimide.
The seal provided over the metallic gate bus also contributes further to the reliability of the device.
Although the present invention has been described in relation to particular embodiments thereof, many other variations and modifications and other uses will become apparent to those skilled in the art. It is preferred, therefore, that the present invention be limited not by the specific disclosure herein, but only by the appended claims.
Number | Date | Country | |
---|---|---|---|
Parent | 12036718 | Feb 2008 | US |
Child | 14333958 | US |