This invention relates to Insulated Gate Bipolar Transistors (IGBTs) and more specifically to a trench type IGBT having a depletion stop layer.
Currently, there are two major types of IGBT devices: “punch through” and “non-punch through” devices. The “punch through” IGBT is normally fabricated on an epitaxial wafer, and “non punch-through” IGBTs on float zone (FZ) wafers. Typical epitaxial wafers used for IGBT manufacturing consist of two epitaxially deposited layers on an underlying substrate, a lightly doped top layer on top of a higher concentration buffer layer, on a substrate of opposite dopant type.
The buffer layer in the punch-through IGBT plays an important role in the performance of the device. It acts as a depletion stop layer under reverse bias and controls the injection efficiency of the backside anode in the forward conduction mode. The thickness and concentration of the buffer layer will affect the breakdown, forward conduction and the switching characteristics of the device. In general, the punch-through IGBT will have a better conduction (VCEON) and switching tradeoff than a non punch-through IGBT device for a given technology or manufacturing process. However, epitaxial wafers are more expensive than Float Zone wafers. For higher voltage devices (>600V), it becomes difficult to control the epitaxial layer (epi) resistivity and thickness uniformity.
As described in U.S. Pat. No. 6,707,111, a punch-through IGBT can be formed in a thin FZ wafers with a proton implant to create an N+ buffer layer.
It is also well known that IGBTs with a trench topology can be formed as described in U.S. Pat. No. 6,683,331, and will provide a lower on-state loss compared to a planar cellular or stripe topology.
The well-known figure of merit (FOM) for IGBTs is the trade-off between the conduction losses (VCEON) and the turn-off energy due to minority carrier recombination (tail current). This tail current increases the turn-off energy (EOFF). Any attempt to reduce the tail current amplitude and duration increases the forward voltage drop (VCEON) of the IGBT. Thus, trade-offs adjust IGBTs for different application requirements. PT IGBTs are available with various trade-offs to suit a wide variety of applications (speeds). NPT IGBTs have a low EOFF compared to that of PT IGBTs but, suffer from high VCEON, making them suitable for applications with switching frequencies >10 kHz.
It would be desirable to have an IGBT which has both lower VCEON and EOFF compared to the PT and NPT IGBTs and lower losses across a wide switching frequency range (4 kHz to 30 kHz). It would also be desirable to have an IGBT which can provide improved efficiency in a wide variety of applications with hard switching methods (low VCEON and EOFF) and also with resonant switching methods (low VCEON).
This invention relates to a novel depletion stop IGBT with trench topology made in a float zone wafers and with both VCEON and EOFF lower then those of prior IGBT devices.
A typical trench IGBT, its structure and manufacturing process, as described in U.S. Pat. No. 6,683,331 is fabricated in a float zone wafer. For the 600V device, a resistivity of 22-30 ohm-cm can be used and to 50-80 ohm-cm for 1200V. After the front side structures are patterned and passivated, the wafer is thinned to 60-70 um for a 600V voltage device and 100-140 um for a 1200V device. An N+ depletion stop layer is then formed on the backside of the wafer followed by a weak anode, also formed on the backside of the wafer, as described in U.S. Pat. No. 6,707,111.
The combination of trench topology and non-epi depletion stop technology offers superior VCEON VS. switching energy tradeoff as shown in
(a) Prior Art IGBTS
Referring to
A conductive polysilicon gate lattice 16 then overlies a conventional gate oxide and the invertible channel region between the exterior of source regions 14. An emitter electrode 17 is then formed over the top surface of wafer 10 and is insulated from gate lattice 16 but contacts the base and source regions 14 and 15. A collector electrode 18 contacts the bottom of region 11.
The N+ buffer layer 12 has a thickness and concentration to obtain the desired switching and breakdown characteristics of the device. The N+ buffer 12 controls the injection efficiency of the bipolar transistor 11/12/13. The thickness of the wafer, with P+ substrate 11 permits manufacture without danger of wafer breakage. However, the epi wafer 10 is expensive.
In order to avoid the higher cost of wafers with epitaxially deposited layers, it is known that the DMOS pattern of
The P− weak anode 21 may be implanted or may be an amorphous silicon layer. Such devices are described in U.S. Pat. No. 6,753,580, issued Jun. 22, 2004; and U.S. Pat. No. 6,242,288, issued Jun. 5, 2001.
The device of
The device of
The conventional way of forming N+ buffer 30 is by the implant of phosphorus or arsenic atoms into the back of the wafer after the thinning operation, followed by an activation anneal to activate the dopant. These implants require an implant energy of 600 keV to 2 MeV to reach the desired N+ region depth. This requires an expensive high energy implanter, and added handling of the fragile wafer. Further, the anneal temperature must be kept below the deposition temperature of the top surface passivation layers (350° C. to 425° C.). However, the preferred anneal temperature of phosphorus or arsenic is above about 700° C. Since a lower temperature must be used, only a small portion of the N+ implant dopant 30 will be annealed and the amount of the anneal varies greatly with small temperature change.
The N+ region 30 of
Thus, hydrogen ions can be implanted with an energy in the range of 100 to 500 KeV at doses in the range of 1E12/cm2 to 1E16/cm2. Good results have been obtained using an energy of 170 KeV at specific doses of 5E13/cm2 to 5E14/cm2 of hydrogen ions.
The wafer is then annealed in a forming gas for 30 to 60 minutes at 300° C. to 400° C., followed by a P− ion implant, or by a P doped amorphous silicon layer deposited by PECVD or by sputtering. A contact is next formed by sputtering of the following metals, in sequence: pure aluminum (1000 Å); titanium (1000 Å); nickel-vanadium (7% V) (4000 Å); silver (6000 Å). An in-situ annealing process prior to aluminum deposition drives off any residual moisture from the wafer surface and ensures good contact between the aluminum and the silicon.
As pointed out previously, trench type IGBTs are also well known, such as the devices shown in our U.S. Pat. No. 6,683,331, the full disclosure of which is incorporated herein by reference.
The structure of
Trenches 131 and 132 extend through N+ emitter regions 140 and 141 respectively which are very deep (2 microns to 4 microns) and have a very short lateral extension, for example (1.5 microns to 3 microns). Note that emitter regions 140 and 141 which have shallow shelf contact regions 142 and 143 respectively, which have a lateral extension of about 0.2 microns to 0.5 microns.
A P+ contact region 150 extends into P− base 37 and between emitter regions 140 and 141. The polysilicon gates 135 and 136 are covered by a suitable insulation oxide 151 and the top surface of the device receives an aluminum or other suitable emitter contact 152. The backside of the device contains a P+ diffusion 154 which receives collector contact 153.
The use of the very deep trench (6.5 microns) and very deep P− base 37 (7 microns) permits the use of the very deep, but narrow emitter regions 140 and 141 while still leaving a sufficiently long invertible channel below the emitter regions (for example, 2 microns) to permit the P regions 137 to support a reasonable voltage and so that the N− body 126 can be optimized. Further, when the device operates in avalanche, a hole current flows from P+ region 154 and up and under the emitters 140 and 141 and through the effective resistance RB, under the very narrow emitter regions 140 and 141. This resistance is very low to avoid the turn on of the NPN transistor 140, 137, 126, for example, and to avoid latching on the IGBT structure.
(b) The Trench IGBTS of the Invention
In accordance with the invention, the N− region 126 of
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.
Number | Name | Date | Kind |
---|---|---|---|
5661314 | Merrill et al. | Aug 1997 | A |
5795793 | Kinzer | Aug 1998 | A |
6242288 | Francis et al. | Jun 2001 | B1 |
6246090 | Brush et al. | Jun 2001 | B1 |
6482681 | Francis et al. | Nov 2002 | B1 |
6683331 | Francis et al. | Jan 2004 | B2 |
6707111 | Francis et al. | Mar 2004 | B2 |
6753580 | Francis et al. | Jun 2004 | B1 |
Number | Date | Country |
---|---|---|
2002-314083 | Oct 2002 | JP |
2003-533047 | Nov 2003 | JP |
Number | Date | Country | |
---|---|---|---|
20070096167 A1 | May 2007 | US |