Reduction of reverse short channel effects by implantation of neutral dopants

[0002] This application is a divisional of application Ser. No. 09/539,527 filed Mar. 3, 2000, now U.S. Pat. No. 6,352,912.

[0003] The present invention relates generally to semiconductor devices and specifically to reverse short channel effects occurring in semiconductor devices.

[0004] The trend of decreasing feature size in semiconductor integrated circuits has led to devices with channel lengths approaching 0.05 microns. As the effective channel length (L_eff) decreases, however, the gate voltage at which the conductivity of the channel is inverted and conduction occurs—the threshold voltage—increases above the theoretically predicted level. FIG. 1 shows this increase in voltage, or reverse short channel effect (RSCE), which is generally an undesirable effect. The dashed line in FIG. 1 represents an ideal channel conductivity behavior in a gated device.

[0005] The tendency towards higher threshold voltages with decreased channel length reverses at some point, and the threshold voltage drops off dramatically. This sudden decrease in threshold voltage is referred to as the short channel effect (SCE). Conventionally, as action is taken to reduce the RSCE, the SCE worsens, which is an undesirable collateral effect.

[0006] The RSCE is generally believed to be caused in n-type metal oxide semiconductor field effect transistors (NMOSFETs) by pile-up of threshold boron at the edge of the source and drain, as well as generally uneven boron distribution across the channel region in FETs with short channels. Supplemental implantation of p-type ions in the channel region has been used to attempt to prevent the RSCE by reducing the impact of boron pile-up in the channel region.

[0007] Another technique that has been employed to reduce the RSCE in FETs is the implantation of germanium into the source and drain regions of the FET. FIG. 2 shows a cross-section of an NFET generally at 10, in which shallow germanium implants 22 have been incorporated into the source 18 and drain 20 regions. A p-type silicon substrate 12 comprises a gate 14 disposed on a gate oxide 15 and between sidewall spacers 16. The source 18 and drain 20 diffusions each have shallow germanium implants 22 that are formed to prevent the RSCE.

[0008] Conventional techniques used to reduce the RSCE, however, can require additional processing steps and can cause unwanted collateral effects on the performance of the device. What is needed in the art is a method for fabricating a semiconductor device that does not suffer from RSCEs.

[0009] The present invention is a semiconductor device comprising a semiconductor substrate, a first diffusion region disposed in said substrate, a second diffusion region disposed in said substrate, a channel region disposed between said first diffusion region and said second diffusion region, a gate oxide disposed on said semiconductor substrate over said channel region and overlapping said first diffusion region and said second diffusion region, a gate electrode disposed on said gate oxide, and a neutral dopant diffusion implant disposed throughout said substrate, said neutral dopant diffusion implant having a peak concentration below said first diffusion region and said second diffusion region.

[0010] The process for making said device comprises forming an oxide layer on a semiconductor substrate of a first conductivity type, blanket-implanting a neutral dopant into said substrate to form a neutral dopant implant, forming a gate electrode on said oxide layer, and, implanting source and drain regions into said substrate to a depth less than the depth at which a peak concentration of said germanium implant occurs.

[0011] The present invention will now be described, by way of example only, with reference to the accompanying drawings which are meant to be exemplary, not limiting, and wherein like elements are numbered alike in several FIGURES, in which:

[0012] FIG. 1 is a graph showing the reverse short channel effect and the short channel effect;

[0013] FIG. 2 is a cross-section of FET showing a conventional germanium implant within the source and drain;

[0014] FIG. 3 is a cross-section of a wafer covered with an oxide layer during germanium doping;

[0015] FIG. 4 is a cross-section of the wafer of FIG. 3 after doping;

[0016] FIG. 5 is cross-section of the wafer of FIG. 4 after gate formation and source and drain doping;

[0017] FIG. 6 is a graph showing the relative dopant concentrations in one embodiment of the FET;

[0018] FIG. 7 is a cross-section of a FET undergoing germanium implantation with a gate electrode already formed on the oxide layer; and,

[0019] FIG. 8 is a graph showing the reduction of the RSCE without a consequent worsening of the SCE for a FET with a germanium implant.

[0020] The semiconductor device described herein has a neutral dopant implant, such as germanium, formed with a peak concentration below the source and drain regions. The germanium implant is preferably implanted before source, drain, and gate formation, but implantation after source, drain, and gate formation is possible. The resulting device, which can be a FET, is not subject to reverse short channel effects, and the implant causes no degradation of short channel effects. Although the figures and the following description disclose the invention in an NFET embodiment for clarity, those skilled in the are will realize that the invention is applicable to other semiconductor devices having gate controlled diffusion regions. For example, a PFET can be formed by reversing the doping polarities from the NFET version.

[0021] Referring now to FIG. 3, an NFET has a p-type silicon region 12 on which an oxide layer 23 has been deposited or grown using conventional methods. The p-type silicon region 12 can be a doped monocrystalline wafer such as would be used in an NFET application, or a well of p-type silicon formed by ion implantation of an n-type silicon as would be used in the NFET portion of a CMOS application. The oxide layer 23 is generally formed with an initial thickness of from about 0.04 to about 0.06 microns, with a thickness of about 0.05 microns preferred. The p-type silicon region 12 can be doped with a p-type dopant such as boron to an initial concentration of 1×10¹⁷to about 2×10¹⁸atoms/cm³, with about 3×10¹⁷atoms/cm³preferred.

[0022] The neutral dopant implant is preferably implanted with enough energy to form a peak neutral dopant concentration in the wafer below the bottom of the source and the drain diffusion implants which are implanted in a later step. Although any neutral dopant can be used, such as silicon or germanium, germanium is the preferred neutral dopant. In one embodiment, the germanium is implanted to form a peak at a depth of about 0.10 to about 0.50 microns, with a depth of about 0.15 to about 0.30 microns preferred, and a depth of about 0.20 to about 0.25 microns especially preferred. A final peak germanium concentration of about 10¹⁹cm⁻³to about 10²¹cm⁻³is preferred, with a concentration of about 10²⁰especially preferred. The germanium concentration at the surface of the p-type silicon region 12 is preferably about 10¹⁷cm⁻³to about 10¹⁹cm⁻³, with a concentration of about 10¹⁸cm⁻³especially preferred. The germanium concentration can vary in any manner between the surface of the p-type silicon region 12 and the peak concentration, but a logarithmic variation is preferred (see, for example, FIG. 6). In order to form the germanium implant at the correct depth and concentration, the germanium ions can be implanted, for example, with from about 230 to about 270 keV at about 10¹³cm⁻²to about 10¹⁶cm⁻², and preferably with about 245 to about 255 keV at about 10¹⁴cm⁻²to about 10¹⁵cm⁻².

[0023] FIG. 4 shows a cross-section of the NFET after implantation of germanium. The distance “x” represents the distance from the surface of the p-type silicon region to the peak concentration of the germanium implant, as described above. The dashed line 26 represents the germanium implant peak concentration depth. The germanium concentration decreases in both directions from the peak concentration depth 26. Distance “x” can be any value that results in an appropriate germanium concentration in the source, drain, and channel region of the NFET, and preferably has a value as described above.

[0024] Referring now to FIG. 5, the NFET is shown after gate formation, and source and drain doping. Gate formation is performed using well-known techniques, and the gate 14 can be a polysilicon gate. Prior to gate formation, the oxide layer 23 is patterned and etched to result in a gate oxide with a thickness of about 4 to about 11 nanometers. A polysilicon layer is then formed, patterned, and etched, to result in a polysilicon gate 14 having a thickness of about 100 to about 200 nanometers.

[0025] Once the gate 14 has been formed, the source 18 and the drain 20 diffusion regions can be doped. Ion implantation of the source 18 and drain 20 with an n-type impurity results in an ion concentration of about 10¹⁹to about 10²¹, with a concentration of about 10²⁰preferred. Source 18 and drain 20 regions are preferably implanted to a depth of less than about 0.15 microns, with a depth of less than about 0.10 microns especially preferred. Side wall spacers 16 comprising oxide or nitride can optionally be formed on the sides of the gate 14 in order to form a second implant (not shown) in the source 18 and the drain 20.

[0026] At this stage, the germanium implant is diffused in the source 18, drain 20, and the channel between the source 18 and the drain 20. Annealing is performed to activate the dopants and restore the crystal structure of the silicon substrate. Annealing can take place between about 600 to about 1200 degrees Celsius. After annealing, NFET fabrication can be completed using conventional metalization and passivation techniques.

[0027] FIG. 6 shows the ion concentrations for the various regions of the NFET. As described above, and as shown in FIG. 6, the peak germanium implant concentration is preferably formed deeper in the wafer than are the source and drain implants. In the NFET described above, the source and drain implants are an n-type ion, and the wafer ion is a p-type ion. The concentrations and depths shown in FIG. 6 and described above are exemplary, and one skilled in the art will realize that alternative doping concentrations and implantation depths are possible and within the scope of this invention.

[0028] Importantly, the germanium can be implanted at any stage of the NFET fabrication process before source and drain formation. For example, the germanium can be implanted before formation of the oxide layer 23, or after formation of the gate electrode 14. FIG. 7 shows the germanium implantation step being performed after the gate 14 has already been formed. In order to effectively implant the germanium under the gate 14 in the channel region, the germanium ions must be implanted at an angle, as shown in FIG. 7. Implantation energy and dose are adjusted to compensate for the angle of implantation. The germanium implant can also be implanted after the source 18 and drain 20 have been formed, and before or after the side wall spacers 16 have been formed, using the implantation technique shown in FIG. 7.

[0029] The blanket implantation of germanium in the p-type substrate inhibits boron pile-up and channel inconsistency, thereby reducing the RSCE by at least 15 percent or more, depending on the device technology. FIG. 8 is a graph that compares the short channel threshold voltage of the germanium implanted NFET of the present invention with a conventional NFET lacking the germanium implant. The NFET with the germanium implant is represented by the dotted line. The reverse short channel effect is reduced to close to an ideal level for the germanium implanted NFET. The short channel effect, however, does not worsen with the germanium implant.

[0030] The NFET described above has the advantage of a significantly reduced reverse short channel effect, without commensurate degradation of the short channel effect or other critical features of the NFET device. The single germanium implantation step allows easy incorporation of the germanium implant in standard NFET and CMOS applications.

[0031] While preferred embodiments have been shown and described, various modifications and substitutions may be made thereto without departing from the spirit and scope of the invention. Accordingly, it is to be understood that the present invention has been described by way of illustration only, and such illustrations and embodiments as have been disclosed herein are not to be construed as limiting to the claims.