Patent Application
20080096360
The present invention now will be described more fully hereinafter with reference to the accompanying drawings, in which a preferred embodiment of the invention is shown. This invention may, however, be embodied in many different forms and should not be construed as limited to the embodiment set forth herein; rather, this embodiment is provided so that this disclosure will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art.
The present inventors have recognized that the forward voltage drop of a semiconductor junction diode device can be improved (i.e., reduced) without the use of a relatively expensive EPI wafer. Instead of an EPI wafer, as described in more detail below, a less expensive, polished non-EPI wafer that undergoes two diffusion processes may be used.
The P-type dopants (e.g., boron) add positive charges to the semiconductor material, while the N-type dopants (e.g., phosphorus) add negative charges to the semiconductor materials. Preferably, the P++ or N++ concentration approximately encompasses the range of 1×1019 to 1×1021 atoms/cm3 and the P+ and N+ concentration approximately encompasses the range of 1×1015 to 1×1020 atoms/cm3. The P− to P (or N− to N) range of dopant concentration approximately encompasses the range of 1×1014 to 1×1016 atoms/cm3. The concentration in the regions may vary depending on what depth of the region is examined. Thus, these concentrations are provided only in approximate ranges.
To reduce the overall resistance of the wafer 200 so that the forward voltage drop can be reduced, an implantation and diffusion process is employed to drive in a dopant into the backside 204 of the wafer 200. The dopant that is employed has the same conductivity as the starting wafer 200. That is, if the starting wafer 200 is initially n-type an n-type dopant is employed and if a p-type starting wafer 200 is employed a p-type dopant is employed in the diffusion process. To drive in the dopant as far as possible into the backside 204 of the wafer 200 a relatively long diffusion process at high diffusion temperatures can be employed. Still, practically speaking, if commonly employed dopant diffusion processes are used, the drive in depth will generally fall between about 200 and 250 microns to provide a low resistivity region 208 (Region II). Thus, if the starting wafer 200 is initially about 250 microns thick and the drive in depth is 200 microns, then a high resistivity region 206 (Region I) will remain, which has a thickness of about 50 microns. The overall thickness of the wafer 200 is therefore limited by the diffusion depth of the dopant through the backside 204 of the wafer 200. Of course, if other implantation and/or diffusion techniques are employed that can drive in dopants by a greater amount, thicker wafers may be employed.
After performing the implantation and diffusion process described above, the wafer 202 has a thin p layer 206 formed on a thicker p+ layer 208 (or, alternatively, n/n+ layers 206 and 208). This resulting dopant concentration profile is shown in
The relationship between the breakdown voltage and dopant concentration can be determined from a consideration of a simple PN junction having a width W0. Without being bound to any particular theory or model, the following analysis is presented to facilitate an understanding of the present invention. As discussed, for example, in chapter 5 of Solid State Electronic Devices, 5th Edition, B. Streeman et al., Pearson Education, Inc., 2004, the contact potential Vo across a pn junction of width W0 can be expressed as:
where ε0 is the electric field across the junction.
The breakdown voltage VB that arises under a certain critical electrical field εC thus can be expressed as
where WB is the depletion width at device breakdown and WB>W0
As shown in Eq. (5-22) of Solid State Electronic Devices the equilibrium width W0 of the junction can be expressed as
and thus
where Na and Nd are the doping concentrations of acceptor ions and donor ions, respectively.
Accordingly, VB can be expressed as:
In the case of a one sided abrupt junction in which one side is more heavily doped (e.g., P+N) Na>>Nd then VB can be well approximated by the following equation:
In the present invention the pn junction is defined by the thin, more lightly doped layer 206 and the more heavily doped thicker layer 208. Accordingly, Nd represents the doping concentration of donor ions in the layer 208, which may therefore be referred to as the bulk concentration of the wafer.
As the preceding equation indicates, since layer 206 has a relatively low dopant concentration, the breakdown voltage of the resulting junction device will be relatively large. To produce a junction device with a lower breakdown voltage, the dopant concentration of layer 206 should therefore be increased. This can be accomplished with a second implantation and diffusion process through the front side 202 of wafer 200. The dopant that is introduced through the front side 202 of wafer 200 will have the same conductivity as the starting wafer 200. The dopant concentration profile of the wafer 200 after both implantation and diffusion processes have been performed through the front and back sides of the wafer 200 is shown in
By controlling the precise dopant concentration of layer 206 in the manner described above, the reverse breakdown voltage of the resulting device can be tailored as desired. Moreover, by controlling the overall dopant concentration profile of the starting wafer 200 (that is, by controlling the overall dopant concentration of layer 208 as well as layer 206) in the manner described above, the reverse breakdown voltage can be tailored while also achieving a low forward voltage drop. The particular dopant concentration profile needed to fabricate a device with some desired target reverse breakdown voltage and forward voltage drop can be determined by techniques such as simulations and the like that are well known to those of ordinary skill in the art. Likewise, appropriate dopants, dopant concentrations and doping parameters (e.g., implantation energy, diffusion times and temperatures) are well known to those of ordinary skill in the art.
Semiconductor junction diode devices were formed on a polished wafer, the inventive wafer (a polished wafer undergoing the doping processes described above) and an Epi wafer. All the wafers had similar reverse voltage performances (VR˜88V). All three wafers were 200 um thick and had n-type conductivity. The polished wafer had a resistivity of 0.45 ohm-sq. Boron oxide was deposited on its surface and underwent a drive-in for 900 minutes. The inventive wafer began with a starting substrate having a resistivity of 20 ohm-sq. Phosphorus oxide was deposited on its back surface and underwent a drive-in for 6000 minutes. Phosphorus was implanted on the front, polished surface with a dosage of 5.0×104 and underwent a drive-in for 4500 minutes. Finally, boron oxide was deposited on its surface and underwent a drive-in for 900 minutes. The EPI Wafer began with an N+ substrate 150 microns thick with a resistivity of 0.005 ohm-sq resistivity. A 50 micron, 3.0×1014 ea/cc EPI layer was formed on the substrate by depositing boron oxide on its surface and underwent a drive-in for 900 minutes.
The reverse voltage performances of all the devices formed on the three wafers were similar and around 80 volts. But the forward voltage drop of both the Epi wafer and the inventive wafer were much better than the polished wafer. In particular, the EPI wafer had a VF of 1.28V, the inventive wafer had a VF of 1.395V and the polished wafer had a VF of 2.985V at a constant current density of 4.0×10−4 A/um. Simulations suggest that at higher current densities a device formed on the inventive wafer may even have superior forward voltage performance.
