Semiconductor fabrication and structure for field-effect and bipolar transistor devices
Semiconductor devices have device regions in which semiconductor properties such as spreading resistivity and its profile are significant. In making a p-type device region on a semiconductor wafer, an initial semiconductor device region is defined by a buried region, and an initial spreading resistivity profile is developed by annealing. After annealing, semiconductor device properties can be enhanced by removing a surface sub-region of the initial device region, and can be further improved by epitaxially growing thereon a monocrystalline film as an improved channel layer for FET devices. Such properties are relevant in MOS as well as bipolar devices.
This is a divisional application of application Ser. No. 09/800,213 filed on Mar. 6, 2001.
TECHNICAL FIELDThe present invention relates to semiconductor devices and, more particularly, to VLSI/ULSI fabrication of MOSFET and bipolar transistors.
BACKGROUND OF THE INVENTIONThe need for scaling metal-oxide-semiconductor (MOS) devices down to below 0.1 μm feature size in very-large-scale integrated (VLSI) circuits has been clearly indicated in the National Technology Road Map for, Semiconductor Technology (1997 Edition), Semiconductor Industry Association, San Jose, Calif. For such circuits, silicon-on-insulator (SOI) MOS devices appear to be promising as described in SOI Technology: Materials to VLSI (2nd edition), Boston, Kluwer, 1997. Such materials are disadvantaged, however, in having a large dose of oxygen ions implanted through the top surface layer of a silicon wafer on which devices are fabricated.
An alternative method for fabricating SOI materials is disclosed in U.S. Pat. No. 5,374,564, issued Dec. 20, 1994 to Bruel and incorporated herein in its entirety. Instead of using ion implantation, an oxide layer is formed by oxidation of the top surface of a silicon wafer, protons are implanted through the oxidized surface, the implanted wafer is annealed to form a hydrogen micro-bubble layer beneath a thin surface layer, a stiffener silicon wafer is attached to the oxidized surface of the annealed wafer, and the resulting structure is heated to expand the micro-bubbles, thereby lifting off the top surface layer which remains attached to the stiffener wafer, forming an SOI wafer. It has been reported further, by K. Henttinen et al., A Mechanically induced Si layer transfer in hydrogen-implanted Si wafers≅, Applied Physics Letters, Vol. 76, No. 17, 24 Apr. 2000, that, after proton implantation beneath a top surface layer of a silicon wafer and annealing to form hydrogen bubbles, the surface layer can be mechanically lifted off by a stiffener wafer at relatively low temperatures.
Incorporated herein in their entirety are U.S. Pat. Nos. 5,198,371 and 5,633,174, issued to Li on Mar. 30, 1993 and May 27, 1997, respectively, disclosing a high-resistivity hydrogen bubble layer or defect layer under a thin surface layer of a silicon wafer after hydrogen implantation and annealing at high temperature. Termed A silicon-on-defect layer≅, the surface layer was found to have improved semiconductor properties such as electron mobility. The structure of the hydrogen bubble or platelet layer is described by J. Grisolia et al., A transmission electron microscopy quantitative study of the growth kinetics of H platelets in Si≅, Applied Physics Letters, Vol. 76, No. 7, 14 Feb. 2000.
Li, Jones, Coleman, Yi, Wallace and Anderson, A Properties of Silicon-on-Defect-Layer Material≅, pp. 745-750 in Materials Research Society Proceedings, Vol. 396, David B Poker et al., Ed., Materials Research Society (MRS), Pittsburgh, Pa., 1996 report on high-temperature annealing after proton implantation resulting in conversion of a top surface layer on a high-resistivity layer from n-type to p-type, thereby forming a p-n junction at the high-resistivity layer. Furthermore, the p-type spreading resistivity was found to decrease steadily from the high-resistivity buried layer to a low resistivity at the surface of the wafer, lower than original wafer resistivity, and the n-type spreading resistivity to decrease steadily to its original value beneath the defect layer, as illustrated there at p. 747 in
Over prior SOI and silicon-on-defect-layer (SODL) device structures, the invention described below results in advantages which are particularly significant in ultra-large-scale integration (ULSI).
SUMMARY OF THE INVENTIONIn making a p-type device region on a semiconductor wafer, an initial semiconductor device region is defined by a buried region, and an initial spreading resistivity profile is developed by annealing. I have discovered that, after annealing, semiconductor device properties can be enhanced by removing a surface sub-region of the initial device region, and can be further improved by epitaxially growing thereon a monocrystalline film as an improved channel layer for FET devices. Such properties are relevant in MOS as well as bipolar devices.
BRIEF DESCRIPTION OF THE DRAWINGThe Figures illustrate preferred embodiments of the invention, as to processing as well as to resulting semiconductor structures and devices.
An n-type silicon wafer having an original bulk spreading resistivity of about 40 ohm-cm was implanted with 180 keV protons to a dose of about 3×1016 protons/cm2 and annealed in a nitrogen-hydrogen atmosphere at 900 degrees C. for 10 seconds to develop a buried hydrogen bubble or platelet layer, as further described in the Li patents. The implanted wafer was annealed by heating to 1180 degrees C. for 20 minutes for conversion of the top surface layer from n-type to p-type conductivity as further described in the 1996 MRS article. The surface of the annealed wafer was subjected to plasma etching to reduce the thickness of the top surface layer overlying the buried layer to approximately 0.1 μm. Other suitable means for thickness reduction include chemical etching and chemical-mechanical polishing (CMP).
For the resulting structure,
While exemplary processing as described herein-above involved two separate steps of annealing, benefits of the invention can be realized also with a single annealing step, e.g. at 900 degrees C. for 4 hours, and other suitable temperature-time profiles are not precluded. Benefits include control of the spreading resistivity profile on an active layer, in MOSFET as well as bi-polar devices.
As shown in
The CMOSFET device shown in
Advantageously, the doped p+ source and drain regions 7a-7d can be fabricated by implanting dopants such as boron and phosphorus, utilizing plasma doping as described by M. J. Goeckner et al., “Plasma doping for shallow junctions”, Journal of Vacuum Science and Technology B, Vol. 17, No. 5, September/October 1999, pp. 2290-2293. The resulting shallow regions are particularly suited for use with the profiled surface layer as described in connection with
In operation, I found that the CMOS device, and in particular the NMOS portions, had improved performance compared to conventional CMOS devices. Moreover, the high-resistivity p-layer had significantly improved cut-off and transconductivity characteristics. Although the exact theory is not known, the SR profile appears to provide a conductivity channel through which both lateral and vertical electric fields can substantially penetrate and control charge carriers. Substantial depletion of the charge carriers was obtained, thereby improving the threshold voltage and sub-threshold slope of the source-drain current. Best results are obtained by etching back the initial top surface layer until the spreading resistivity is equal to or greater than the original resistivity of the wafer.
Referring to
Referring to
The PMOS and NMOS regions further include p+ and n+ doped source and drain regions 17a and 17b and 17c and 17d respectively, having metallic electrodes 18a-18d fabricated in surface layer 11. As described in connection with
In operation, the present CMOSFET device showed improved transconductance and lower latch-Lip than conventional CMOSFET. Putatively, the improvement may be related to the combination of high-resistivity defect layer 12, which extends under all the devices, and trench 19, which electrically isolate adjacent PMOS and NMOS devices. In addition, improved transconductance may be attributed to the gettering action of defect layer 12 and by the etch-back step for reducing the thickness of the top surface layer. Furthermore, the spreading resistivity profile that resulted from the implantation, annealing and etch-back procedures functioned to confine the conduction of charge carriers to a thin layer 11′ of the top surface layer 111 into which both normal and lateral electric fields could penetrate, thereby improving cut-off and other properties of the FET. In addition, I found that using wafers with minimum oxygen content enabled the defect layer 11 to getter metals and other impurities otherwise present in the top surface layer.
The top surface layer 91 was fabricated by controlling the annealing steps and etch back steps, such as described in connection with
In any of the embodiments discussed above, the top surface layer on the defect layers which comprise microbubbles or platelets after the first annealing step, may be removed from the Si wafer by attaching the top surface layer to an oxidized (stiffener) Si wafer and by heating in order to raise the pressure of the microbubbles sufficiently to exfoliate the top surface layer as described by the Bruel '564 patent. Alternatively and in contrast, the top surface layer can be mechanically sheared off as described by Henttinen et al.
However, the top surface layer is converted to p-type conductivity with a high spreading resistance profile, thereby producing a SOI p-channel with high-resistivity and a predesigned gradient. The stiffener wafer may be n- or p-type. In addition, a gate electrode may be applied to the top oxide layer which subsequently becomes a buried SOI layer before the lift-off or exfoliation step. After the lift-off step, the boundary of the buried layer becomes the new top surface of the top surface layer. This surface may be polished or etched to remove the boundaries of the bubbles or platelets and to adjust the total thickness of the new p-type high-resistivity layer.
In yet another embodiment of the invention,
Conventional fabrication techniques, operating conditions and applications, such as described in U.S. Pat. No. 5,461,245, issued Oct. 24, 1995 to Gribnikov et al. which is incorporated herein in its entirety, may be used along with the novel process procedures disclosed herein.
In operation, the mobility of the charge carriers in the base was found to be significantly increased compared with those in conventional bipolar and floating devices. Furthermore, when used in applications, such as logic circuits as described in the Gribnikov '245 patent, substantially improved operating characteristics were found. High mobility floating base layer 101 and its defect layer 102 in contact with collector layer 103 provided unique p-n junction characteristics not heretofore available.
In addition, the improved composite epi-layer 108 on annealed and etched layer 101 on defect layer 102 was found to provide improved performance for other semiconductor devices such as, for example, described above in
The epitaxial layer is on the annealed and etched p-layer with the desired surface of the p-layer, after irradiating, being annealed and etched so as to produce the desired spreading resistivity. Combinations of such conductivity types include n-epi on p/peak/p or on p/peak/n, and p-epi on p/peak/p or on p/peak/n, where “peak” means the high resistivity peak which is produced by the initial defect layer. In operation I found that, although it is necessary to create the defect layer by ion implantation in order to getter impurities and increase the resistivity above the original resistivity of the wafer substrate, it is not necessary for a detectable defect layer to be present after annealing provided impurities are anchored at microscopic platelets or even a microscopic layer of dislocations. The crystal structure of the device region after irradiating, annealing and etching, in which devices are fabricated is improved when compared to the wafer substrate because impurities are lower in the initial surface region. Consequently the epitaxial layer, which faithfully follows its substrate structure, also has an improved crystal structure and lower impurities than available by other processes.
In operation, devices fabricated using the structure shown in
Claims
1. A semiconductor device comprising a p-type device region on at least a portion of a surface of a semiconductor substrate having an original bulk resistivity, said device region having a spreading resistivity profile that increases from a surface region to a peak value at a depth below said surface region, and said spreading resistivity value in said device region being greater than said original bulk resistivity of said wafer.
2. A semiconductor device comprising a composite device region on at least a portion of a surface of a semiconductor substrate having an original bulk resistivity, said composite device region comprising an epitaxial layer grown on the surface of a recrystallized p-type layer, and said p-type layer having a spreading resistivity profile that increases from the surface region to a peak value at a depth below the surface region.
3. The device of claim 2, wherein said peak value is greater than said original bulk resistivity.
4. The device of claim 2, wherein said epitaxial layer is n-type.
5. The device of claim 4, wherein said substrate is p-type.
6. The device of claim 4, wherein said substrate is p-type.
7. The device of claim 2, wherein said epitaxial layer is p-type.
8. The device of claim 7, wherein said substrate is p-type.
9. The device of claim 7, wherein said substrate is n-type.
Type: Application
Filed: Jun 11, 2003
Publication Date: Mar 17, 2005
Inventor: John Coleman (Locust Valley, NY)
Application Number: 10/460,436