SELF-ALIGNED SEMICONDUCTOR DEVICES WITH REDUCED GATE-SOURCE LEAKAGE UNDER REVERSE BIAS AND METHODS OF MAKING

Abstract

A vertical junction field effect transistor (VJFET) having a self-aligned pin, a p+/n/n+ or a p+/p/n+ gate-source junction is described. The device gate can be self-aligned to within 0.5 μm to the source in order to maintain good high voltage performance (i.e. low DIBL) while reducing gate-source junction leakage under reverse bias. The device can be a wide-bandgap semiconductor device such as a SiC vertical channel junction field effect. Methods of making the device are also described.

Description

CROSS-REFERENCE TO RELATED APPLICATIONS

This application is a divisional of U.S. patent application Ser. No. 13/112,075, filed on May 20, 2011, pending, which claims the benefit of Provisional U.S. Patent Application Ser. No. 61/347,928, filed on May 25, 2010. Each of the above-referenced applications is incorporated by reference herein in its entirety.

The section headings used herein are for organizational purposes only and should not be construed as limiting the subject matter described herein in any way.

BACKGROUND

1. Field

This application relates generally to semiconductor devices and methods of making the devices and, in particular, to wide-bandgap semiconductor devices such as SiC vertical channel junction field effect transistors with reduced gate-source leakage under reverse bias.

2. Background of the Technology

To date, vertical channel silicon carbide junction field effect transistors have been proposed as devices with vertical or near vertical sidewalls [1, 2]. In devices with vertical or near vertical sidewalls, however, it can be difficult to achieve uniform p⁺ sidewall doping using ion implantation. In particular, normal incident ion implantation can result in non-uniformly doped sidewalls having a low dopant concentration.

The use of angled ion implantation to dope the sidewalls has been disclosed [1, 3]. Even with this approach, however, it is difficult to achieve an idealized structure having uniform channel width (w_ch). In particular, the use of an angled implantation can still result in heavier doping near the trench bottom and non-uniform doping along the sidewall which reduces device performance. Furthermore, to insure similar doping on both sidewalls the wafer has to be rotated during implantation. For SiC, however, ion implantation requires multiple implants at different energies. Therefore, a process involving rotation of the wafer and angled implantation can add significantly to the complexity and cost of the manufacturing process.

Accordingly, there still exists a need for improved methods of making semiconductor devices such as vertical JFETs with more uniform and well-controlled channel width.

SUMMARY

A semiconductor device is provided which comprises:

a substrate layer of a semiconductor material of a first conductivity type;

a channel layer of a semiconductor material of the first conductivity type on an upper surface of the substrate layer, the channel layer comprising a lower surface and one or more raised regions comprising an upper surface and first and second sidewalls, wherein the first and second sidewalls of the raised regions adjacent the lower surface are tapered inward and form an angle of at least 5° from vertical to the upper surface of the substrate layer, wherein the one or more raised regions comprises an inner portion of a semiconductor material of the first conductivity type and outer portions of a semiconductor material of a second conductivity type different than the first conductivity type, wherein the outer portions are adjacent to the first and second sidewalls;

gate regions of semiconductor material of the second conductivity type in the lower surface of the channel layer adjacent to and contiguous with the outer portions of adjacent raised regions; and

a source layer of a semiconductor material of the first conductivity type on the upper surfaces of the one or more raised regions;

wherein the outer portions of the raised regions are offset from the source layer such that the outer portions of the raised regions do not contact the source layer.

A method is also provided which comprises:

selectively implanting ions into a channel layer of a semiconductor material of a first conductivity type to form implanted gate regions of semiconductor material of a second conductivity type different than the first conductivity type, wherein the channel layer is on an upper surface of a substrate layer and wherein the channel layer comprises a lower surface and one or more raised regions comprising an upper surface and first and second sidewalls, wherein the first and second sidewalls of the raised regions adjacent the lower surface are tapered inward and form an angle of at least 5° from vertical to the upper surface of the substrate, wherein source regions of a semiconductor material of the first conductivity type are on the upper surfaces of the one or more raised regions, the source regions comprising side surfaces adjacent the first and second sidewalls and an upper surface and an implant mask is on the upper surface of the source regions, wherein the implanted gate regions are formed in the sidewalls and in the lower surface of the channel layer and wherein the implanted gate regions are offset from the upper surface of the raised regions; and

removing the implant mask;

wherein the implanted gate regions on the sidewalls are offset from the source layer such that the implanted gate regions on the sidewalls do not contact the source layer.

These and other features of the present teachings are set forth herein.

BRIEF DESCRIPTION OF THE DRAWINGS

The skilled artisan will understand that the drawings, described below, are for illustration purposes only. The drawings are not intended to limit the scope of the present teachings in any way.

FIG. 1A is a schematic of a vertical junction field effect transistor having sloped sidewalls wherein the n+ source region is offset 0.5 μm or less from the p+ implanted gate region by a n, n−, p− or p region.

FIG. 1B illustrates the conduction band energy as a function of the distance from the source electrode for an SIT device and a power VJFET device.

FIG. 1C is a schematic illustrating a method of making VJFET device having sloped sidewalls.

FIG. 1D is a schematic illustrating residual lattice implant damage at the p+/n+ gate-source junction of a device manufactured according to the method depicted in FIG. 1C.

FIG. 1E is a schematic illustrating the minimization of the p+/n+ gate-source junction in a device manufactured according to the method depicted in FIG. 1C using zero degree implantation.

FIG. 2 is a schematic illustrating a method of making a device as set forth in FIG. 1A using a conformal mask layer.

FIG. 3A illustrates the doping profile for a device having a p+/n+ junction.

FIG. 3B illustrates the doping profile for a device having a p+/p/n+ junction made using a conformal mask showing reduced electric field and less implant damage.

FIG. 4 is a schematic illustrating a method of making a device as set forth in FIG. 1A using a non-conformal mask layer.

FIG. 5 is a schematic illustrating a method of making a device as set forth in FIG. 1A using thermal oxidation of the sidewall prior to sidewall implantation.

FIG. 6 is a schematic illustrating a method of making a device as set forth in FIG. 1A using an implant mask that overhangs the sidewalls.

FIG. 7 is a schematic illustrating a method of making a device as set forth in FIG. 1A using a multilayer implant mask comprising a layer with a high lateral oxidation rate.

FIG. 8 is a schematic illustrating a method of making a device as set forth in FIG. 1A by recessing the n+ source layer by oxidation such that the n+ source layer does not contact the p+ implanted regions.

FIG. 9 is a schematic of a vertical junction field effect transistor having sloped sidewalls wherein the n+ source region is offset from the p+ implanted gate region by an n source region.

DESCRIPTION OF THE VARIOUS EMBODIMENTS

A power junction field effect transistor (JFET) should remain in the off-state even with very large biases applied to the drain terminal (e.g., 600 V-10 kV). Accordingly, the power JFET device should have a minimal “drain-induced barrier lowering” (commonly referred to as “DIBL”). In the DIBL phenomenon, the applied drain voltage lowers the energy barrier between the source and drain, thus allowing undesirable leakage current to flow through the device.

To minimize the DIBL effect and thus enable the power transistor to block large voltages (e.g., 600 V-10 kV), the off-state energy barrier should occur near the source electrode and there should be a “long channel” separating the drain from the source. In effect, the energy barrier (which is modulated by the bias applied to the p+ gate) should be as far away from the drain as possible to minimize DIBL. This is accomplished by locating the narrowest part of the channel near the source, as is the case with a JFET that has sloped sidewalls as disclosed in U.S. patent application Ser. No. 12/613,065 or by a device having a non-uniform channel doping profile in which the doping concentration near the source is lower than the rest of the channel as disclosed in U.S. patent application Ser. No. 12/117,121.

Since the energy barrier should be very close to the source and should be modulated by the gate in order to control conduction through the device, the p+ gate should necessarily be located in very close proximity to the n+ source. The process by which the p+ gate is formed should also be self-aligned to the channel/finger. This is the case when a SiC vertical JFET is formed by etching a finger and implanting the p+ gate using the same mask. Other SiC vertical transistors such as the static induction transistor (SIT) are not designed to block large drain voltages, therefore the channel design requirements are less stringent and structures with a large, non-self-aligned separation between n+ source and p+ gate are permitted.

SIT structures typically have the off-state barrier much closer to the drain terminal and typically have shorter channels (for high frequency operation) than a power JFET as shown in FIG. 1B. See, for example, Nishizawa et al., IEEE Trans. Electron Devices, Vol. 4 (2000), pg. 482. Both of these features make the SIT unsuitable for high voltage (e.g., 600 V-10 kV) applications. By design, a high voltage power JFET will be more effective at blocking high voltages if the off-state barrier is located close to the source terminal as shown in FIG. 1B.

As described in U.S. patent application Ser. No. 12/613,065, which is incorporated by reference herein in its entirety, a device having sloped sidewalls can be made by depositing an implant mask layer (e.g., SiO₂) on an epitaxially grown SiC layer structure, patterning and etching the implant mask layer and the SiC fingers and implanting the self-aligned p+ gate regions using the implant mask. This process is illustrated in FIG. 1C. As can be seen from FIG. 1C, the implant mask in this process does not completely cover the n+ region. Lateral implant straggle may also result in implanted species extending under the implant mask. Both of these effects can result in a p+/n+ gate-source junction.

As described above, a high voltage power JFET should have a long channel with gate-modulated electron barrier close to source. For the implanted gate vertical JFET, this may result in a p+/n+ gate-source junction as shown in FIG. 1C. However, a p+/n+ junction will have high leakage when a reverse bias is applied to the gate (e.g., V_gs=−15 V) to turn off the device. This can result in an undesirable static power dissipation during the off-state. High doping on both sides of the junction results in a narrow depletion region, and hence, a high electric field under reverse voltage bias which results in undesirable gate-source junction leakage. A PiN (or p+/n/n+, p+/p/n+) junction would result in a reduced electric field and hence lower leakage. Also because the junction receives a large implant dose during the p+ gate formation, residual lattice damage that is not removed during the implant activation process will may lead to enhanced generation-recombination leakage current during reverse bias. Implant damage at the p+/n+ gate-source junction is shown in FIG. 1D.

However, multi-slope devices with zero degree implant as described in U.S. patent application Ser. No. 12/613,065 minimize the p+/n+ junction problem because the implant mask prevents a heavy p+ dose from being implanted at the edge of the n+ region as shown in FIG. 1E. In contrast, the use of tilted or angled implantation in devices with vertical sidewalls results in high doping on both sides of the gate-source junction.

Vertical junction field effect transistors are provided having self-aligned pin (or p+/n/n+, p+/p/n+) gate-source junctions. The p+ gate can be self-aligned to within 0.5 μm of the n+ source in order to maintain good high voltage performance (i.e. low DIBL) while reducing gate-source junction leakage under reverse bias. The p+ implant region can be offset from the n+ source either during the implant or afterwards through additional post-implant processing. According to some embodiments, the p+ and n+ regions have peak doping concentration greater than 1×10¹⁹ cm⁻³ to minimize contact resistance. According to some embodiments, the region between the n+ and p+ regions has lower doping (n−, n, p−, or p) of less than 1×10¹⁹ cm⁻³.

Both the device structure and methods for fabricating the structure are provided. The methods are applicable to the manufacture of vertical, single-slope, or multi-slope VJFET fingers with both tilted and zero degree gate implantation.

The p+ implant region can be offset from n+ source by either a self-aligned mask layer during ion implantation or by the self-aligned recessing of n+ source layer after ion implantation. A VJFET device is shown in FIG. 1A. As shown in FIG. 1A, the n+ source is offset from the p+ implanted gate regions by a distance of 0.5 μm or less.

According to a first embodiment, conformal deposition of an implant blocking layer (e.g., SiO₂) over an existing implant mask can be used to offset the implanted gate regions from the source regions. This method is illustrated in FIG. 2. As shown in FIG. 2, the n+ source is offset from the p+ implanted gate regions by an n−, n, p− or p region.

FIG. 3A illustrates the doping profile for a device having a p+/n+ junction. FIG. 3B illustrates the doping profile for a device having a p+/p/n+ junction which was made using a conformal mask as set forth in FIG. 2. As can be seen from FIG. 3B, the device shows reduced electric field and less implant damage than the device of FIG. 3A. The approximate dopant concentrations are shown in FIGS. 3A and 3B.

According to some embodiments, the p+ implanted gate regions can be offset from the n+ source regions using a non-conformal deposition of an implant blocking layer (e.g., SiO₂) over existing an implant mask. This method is illustrated in FIG. 4. As shown in FIG. 4, the non-conformal mask layer is thicker on the sides of the implant mask than on the sidewalls of the raised channel regions.

According to a some embodiments, the p+ implanted gate regions can be offset from the n+ source regions by the formation of an implant blocking layer (e.g., SiO₂) on the sidewalls using thermal oxidation prior to implantation. This method is illustrated in FIG. 5.

According to a some embodiments, the p+ implanted gate regions can be offset from the n+ source regions by using SiO₂ and SiC etch processes to form implant mask that overhangs finger sidewall. This method is illustrated in FIG. 6.

According to a some embodiments, the p+ implanted gate regions can be offset from the n+ source regions by using a multilayer implant mask consisting of one layer with high lateral oxidation rate at temperatures that will not significantly oxidize SiC (e.g. SiO₂/Si/SiO₂, SiO₂/Ge/SiO₂, SiO₂/poly-Si/SiO₂, SiN/Ge/SiN). The multilayer implant mask can be etched and oxidized using conditions which cause negligible oxide growth on SiC (e.g., <1000° C. in O₂). This method is illustrated in FIG. 7.

As set forth above, the p+ implant region can be offset from n+ source by using a self-aligned mask layer during ion implantation. Alternatively, the p+ implant region can be offset from n+ source by self-aligned recessing of the n+ source layer after ion implantation. In particular, the n+ region can be recessed after p+ implantation to remove the overlap of the n+ source and the p+ gate regions. The differential oxidation rates of SiC crystal can be used to laterally oxidize the n+ region preferentially to the p+ region. The (0001) surface of SiC oxidizes slowly compared to the other surfaces of SiC. Von Munch et al., J. Electrochemical Soc., vol. 122, pg. 642 (1974). In addition, for the (1120) surface, the oxide thickness on n+ SiC was about twice the thickness on p-type SiC. These differential oxidation rates may be used to selectively oxidize the n+ source layer thus removing the p+/n+ overlap between gate and source.

A method of offsetting the p+ implanted gate regions from the n+ source regions by the self-aligned recessing of the n+ source layer is shown in FIG. 8. As shown in FIG. 8, the (0001) surfaces of SiC oxidize slowly whereas the (11 20) face oxidizes quickly, especially when the (11 20) face is n+. As a result, the source is recessed from the edges of the sidewalls such that the n+ source regions no longer contact the p+ gate regions.

FIG. 9 is a schematic depicting an alternative embodiment wherein a non-uniformly doped source layer is used to offset the n+ source from the p+ implanted gate regions. As shown in FIG. 9, the source layer comprises a lower source layer in contact with the p+ gate regions adjacent an upper source layer with a higher doping concentration. The lower source layer create a pin junction with the gate regions whereas the higher doping in the upper layer can reduce contact resistance. The upper source layer can have a doping concentration of 1×10¹⁹ to 1×10²⁰ cm⁻³ and the lower source layer can have a doping concentration of less than 1×10¹⁹ cm⁻³. According to some embodiments, the thickness of each layer can be 0.25 μm. The thickness and doping concentration of the upper and lower source layers can be varied to obtain desirable operating characteristics.

While the foregoing specification teaches the principles of the present invention, with examples provided for the purpose of illustration, it will be appreciated by one skilled in the art from reading this disclosure that various changes in form and detail can be made without departing from the true scope of the invention.

REFERENCES

• [1] U.S. Patent Application Publication No. 2007/0187715 A1, “Power Junction Field Effect Power Transistor with Highly Vertical Channel and Uniform Channel Opening.”
• [2] U.S. Pat. No. 5,903,020, “Silicon Carbide Static Induction Transistor Structure.”
• [3] U.S. Pat. No. 7,479,672 B2, “Power Junction Field Effect Power Transistor With Highly Vertical Channel And Uniform Channel Opening.”

Claims

1-26. (canceled)

27. A method comprising:

selectively implanting ions into a channel layer of a semiconductor material of a first conductivity type to form implanted gate regions of semiconductor material of a second conductivity type different than the first conductivity type, wherein the channel layer is on an upper surface of a substrate layer and wherein the channel layer comprises a lower surface and one or more raised regions comprising an upper surface and first and second sidewalls, wherein the first and second sidewalls of the raised regions adjacent the lower surface are tapered inward and form an angle of at least 5° from vertical to the upper surface of the substrate, wherein source regions of a semiconductor material of the first conductivity type are on the upper surfaces of the one or more raised regions, the source regions comprising side surfaces adjacent the first and second sidewalls and an upper surface and an implant mask is on the upper surface of the source regions, wherein the implanted gate regions are formed in the sidewalls and in the lower surface of the channel layer and wherein the implanted gate regions are offset from the upper surface of the raised regions; and

removing the implant mask;

wherein the implanted gate regions on the sidewalls are offset from the source layer such that the implanted gate regions on the sidewalls do not contact the source layer.

28. The method of claim 27, wherein an implant blocking layer is on the sidewalls of the raised regions, on side surfaces of the source regions and on side surfaces of the implant mask prior to selectively implanting, the method further comprising removing the implant blocking layer after selectively implanting.

29. The method of claim 28, wherein the implant blocking layer comprises SiO2.

30. The method of claim 28, wherein the implant blocking layer is thicker on the side surfaces of the implant mask than on the sidewalls of the raised regions.

31. The method of claim 27, further comprising oxidizing the sidewalls of the raised regions to form oxidized sidewalls prior to selectively implanting, wherein the oxidized sidewalls inhibit implantation in the offset region such that the implanted gate regions are offset from the upper surface of the raised regions by an offset region of semiconductor material of the first conductivity type having a lower doping concentration than the source layer.

32. The method of claim 27, wherein at least a portion of the implant mask is wider than the upper surface of the source regions such that the implant mask overhangs the sidewalls.

33. The method of claim 32, wherein the implant mask comprises a layer of a first material adjacent the source regions and a layer of a second material different than the first material adjacent the layer of the first material, wherein the second material has a higher lateral oxidation rate than the first material, the method further comprising oxidizing the implant mask prior to selectively implanting such that the layer of the second material overhangs the sidewalls.

34. The method of claim 33, wherein the first material is SiO2 and wherein the second material is Si, GE, or poly-Si or wherein the first material is SiN and the second material is Ge.

35. The method of claim 33, wherein the implant mask is oxidized at temperatures less than 1000° C.

36. The method of claim 27, wherein the side surfaces of the source regions oxidize more rapidly than the gate regions, the method further comprising oxidizing the source and gate regions after selectively implanting, wherein oxidizing recesses the side surfaces of the source layer such that the source layer is separated from the gate regions.

37. The method of claim 36, wherein the semiconductor material of the first conductivity type is an n-type semiconductor material and wherein the semiconductor material of the second conductivity type is a p-type semiconductor material.

38. The method of claim 37, wherein the source regions and/or the gate regions have a doping concentration of at least 1×1019.

39. The method of claim 27, wherein the ions are implanted into the channel layer at an angle of +/−2° from vertical to the upper surface of the substrate.

40. The method of claim 27, wherein the first and second sidewalls adjacent the upper surfaces of the one or more raised regions are oriented at an angle of <5° from vertical to the upper surface of the substrate layer.

41. The method of claim 27, wherein the first and second sidewalls adjacent the upper surfaces of the one or more raised regions are oriented at an angle of <2° from vertical to the upper surface of the substrate layer.

42. The method of claim 27, further comprising a drift layer of a semiconductor material of the first conductivity type between the substrate and the channel layer.

43. The method of claim 27, further comprising a buffer layer between the substrate and the channel layer.

44. The method of claim 42, further comprising a buffer layer between the substrate and the drift layer.

45. The method of claim 27, wherein the device comprises a plurality of raised regions, wherein the plurality of raised regions are elongate and are arranged in spaced relationship as fingers.

46. The method of claim 27, wherein the semiconductor material of the first conductivity type is an n-type semiconductor material and wherein the semiconductor material of the second conductivity type is a p-type semiconductor material.

47. The method of claim 27, further comprising:

forming a gate contact on the lower surface of the channel layer;

forming a source contact on the source layer; and

forming a drain contact on the substrate layer opposite the channel layer.