Method of forming polysilicon layers in a transistor
A semiconductor transistor which is not capable of storing data is formed as follows. An insulating layer is formed over a silicon region. An undoped polysilicon layer is formed over and in contact with the insulating layer. A doped polysilicon layer is formed over and in contact with the undoped polysilicon layer such that at least two edges of the doped polysilicon layer vertically line up with corresponding edges of the undoped polysilicon layer to thereby form sidewalls, and the doped and undoped polysilicon layers form a gate of the transistor. After the doped polysilicon layer is formed, source and drain regions are formed in the silicon region. Dopants from the doped polysilicon layer migrate into the undoped polysilicon layer thereby doping the undoped polysilicon layer.
Latest Hynix Semiconductor, Inc. Patents:
This application is a division of U.S. application Ser. No. 10/870,285, filed Jun. 16, 2004, which is a division of U.S. application Ser. No. 09/994,545, filed Nov. 26, 2001, each of which is incorporated herein by reference in its entirety for all purposes.
BACKGROUND OF THE INVENTION
A simplified conventional process sequence to form memory cell 100 includes: forming tunnel oxide layer over substrate 102; depositing a first layer polysilicon over the tunnel oxide layer; forming an interpoly composite ONO dielectric layer over the first layer polysilicon; depositing a second layer polysilicon over the ONO dielectric; forming a tungsten silicide layer over the second layer polysilicon; and self-aligned mask and self-aligned etch (SAE) to form the gate stack as shown in
The first and second polysilicon layers are deposited by means of Chemical Vapor Deposition (CVD). Both first and second polysilicon layers are in-situ doped (usually by phosphorus P31) to a relatively high level (e.g., 2×1019 to 5×1020 cm−3). The level of polysilicon doping is usually controlled by gas flow rate and pressure of the gas compound containing P31, such as PH3. An example of a set of parameters associated with the polysilicon deposition of a conventional process is provided below.
There are a number of reasons for the high polysilicon doping. First, the high doping prevents or minimizes polysilicon depletion when gate bias is applied to the control gate of the memory cell or to the gate of the MOS transistor. Polysilicon depletion decreases gate capacitance thus reducing gate control in a MOS transistor channel region, and impairs other transistor/cell electrical characteristics. Second, the high doping helps maintain a proper value of polysilicon work function which impacts such important transistor/cell parameters as the threshold voltage. Third, the high doping reduces the world line resistance in the memory array, thus improving the memory performance. Fourth, the high doping reduces time delay associated with the peripheral transistor gate capacitance and resistance.
However, there are also drawbacks to the high polysilicon doping. The high doping leads to higher oxidation rate of polysilicon crystals. Higher oxidation rate in turn leads to a more pronounced “smiling” effect, i.e., an increased gate oxide thickness at the edges of the gate in MOS transistors, and similar increase of tunnel oxide thickness and ONO dielectric at the edges of the cell gate stack as shown in
A further drawback of the high doping is that it leads to a larger polysilicon grain size which in turn leads to a more rugged interface between the gate oxide and the polysilicon gate in MOS transistors, and similarly between each of the tunnel oxide and the floating gate, bottom of the ONO dielectric and the floating gate, and top of the ONO dielectric and the control gate in a memory cell. In extreme cases, it may lead to gate oxide and/or tunnel oxide pinch-off or otherwise impact the integrity and reliability characteristics of the gate oxide in MOS transistors and the tunnel oxide and the ONO dielectric in memory cells.
In conventional processes, the room to achieve the necessary trade-off between the desirable and undesirable effects of the polysilicon doping is limited to only regulating the level of doping and uniformity of the doping profile across the polysilicon layers. Achieving the desired trade off thus often proves to be a difficult task from process and device optimization point of view.
Accordingly, there is a need for polysilicon layers structure and method of forming the same whereby an optimum polysilicon doping profile can be achieved, the depletion of the polysilicon and its associated adverse effects are prevented or minimized, the quality and uniformity of the polysilicon-oxide interface are improved, while the “smiling” effect in the dielectric layers interfacing polysilicon is minimized.
BRIEF SUMMARY OF THE INVENTIONIn accordance with an embodiment of the present invention, a semiconductor transistor which is not capable of storing data is formed as follows. An insulating layer is formed over a silicon region. An undoped polysilicon layer is formed over and in contact with the insulating layer. A doped polysilicon layer is formed over and in contact with the undoped polysilicon layer such that at least two edges of the doped polysilicon layer vertically line up with corresponding edges of the undoped polysilicon layer to thereby form sidewalls, and the doped and undoped polysilicon layers form a gate of the transistor. After the doped polysilicon layer is formed, source and drain regions are formed in the silicon region. Dopants from the doped polysilicon layer migrate into the undoped polysilicon layer thereby doping the undoped polysilicon layer.
In one embodiment, the insulating layer is a gate oxide layer.
In another embodiment, a thickness of the doped polysilicon layer is greater than a thickness of the undoped polysilicon layer by a factor in the range of two to four.
In another embodiment, insulating spacers are formed along the sidewalls of the doped and undoped polysilicon layers.
In another embodiment, the doped polysilicon layer is formed by depositing an in-situ doped polysilicon layer.
In another embodiment, the transistor is any one of a NMOS transistor, PMOS transistor, enhancement transistor, and depletion transistor.
In yet another embodiment, the doped polysilicon layer has a doping concentration and a thickness greater than a thickness of the undoped polysilicon layer so as to prevent polysilicon depletion in the gate.
In another embodiment, a conductive layer comprising tungsten is formed over and in contact with the doped polysilicon layer.
BRIEF DESCRIPTION OF THE DRAWINGS
In accordance with an embodiment of the present invention, conventional semiconductor device structures wherein a doped polysilicon layer comes in contact with an insulating layer, such as silicon-dioxide, are modified so that the polysilicon layer comprises a doped and an undoped polysilicon layer with the undoped polysilicon layer interfacing with the insulating layer. In this manner, the drawbacks of the prior art structures wherein doped polysilicon layers are in direct contact with insulating layers are minimized or eliminated while the advantages of a doped polysilicon are maintained, as discussed in more detail below.
In one embodiment, the structure of
After deposition of polysilicon layers 206-a, 206-b, interpoly ONO dielectric 208 is formed in accordance with conventional methods. Next, in forming control gate 210, two successive polysilicon deposition steps are performed. First, an undoped polysilicon deposition step is carried out, followed by an in-situ doped polysilicon deposition step.
The tables below show the temperature, gas flow rate and pressure, doping concentration, and polysilicon thickness for each of the doped and undoped polysilicon layers in accordance with an exemplary embodiment of the present invention. This table reflects a thickness ratio of doped polysilicon to undoped polysilicon of in the range of 2:1 to 5:1, with a preferred ratio of 3:1. Note that the values in these tables are merely illustrative and not intended to be limiting. Varying these values to achieve the target parameters and the desired cell performance would be obvious to one skilled in this art in view of this disclosure.
After deposition of the polysilicon layers in forming control gate 210, a tungsten (WSix) layer 212 is optionally deposited in accordance with conventional methods. In some processes, an ARC oxynitride layer (not shown) is deposited over the tungsten layer to complete gate layer formation. This is followed by gate mask and gate etch to form the control gate of the memory cell and the gate of peripheral transistors, and then self-aligned mask and self-aligned etch (SAE) is carried out to form the gate stack as it appears in
During thermal oxidation and anneal cycles, such as ONO steam anneal (after ONO deposition), polysilicon re-oxidation after gate stack formation, and source/drain oxidation cycle(s), the top and bottom oxide layers in ONO dielectric 208 at the periphery of the gate stack (side walls) grow at a lower rate due to the lower oxidation rate of undoped polysilicon layers which interface the two oxide layers. Thus, ONO dielectric “smiling” effect is substantially reduced. The tunnel oxide “smiling” effect can similarly be reduced by including another undoped polysilicon layer as the bottom polysilicon layer of the floating gate. This is shown in the
Other than the reduction in “smiling” effect, the smaller grain size of undoped polysilicon yields a polysilicon-oxide interface which is more uniform leading to improved tunnel oxide and ONO dielectric quality and integrity. Further, by selecting proper doping concentration in the doped polysilicon layers and proper thickness ratio between adjacent doped and undoped polysilicon layers, by the end of the thermal cycles, a uniform and high enough doping concentration can be achieved throughout the whole floating gate and control gate so as to prevent polysilicon depletion effects.
By the end of the oxidation/anneal thermocycle, depending on the thermal budget, the impurity (e.g., phosphorus) profile in the undoped polysilicon layers may be of diffusion character.
After the thermal cycles, as shown by dashed line 516, the impurity concentration is highest in the doped polysilicon layer 306-a and gradually reduces at the boundaries between the doped and undoped polysilicon layer and through the undoped polysilicon layers 306-b, 306-c, and reaches its lowest concentration level at the interface between the undoped polysilicon layers 306-b, 306-c and the corresponding tunnel oxide 304 and ONO dielectric layer 308. The thickness of the polysilicon layers and the thermocycles need to be optimized such that the final polysilicon doping concentration and its gradient at the polysilicon-dielectric interface is high enough to prevent or minimize polysilicon depletion effects.
Note that despite the high final doping concentration at the polysilicon-dielectric interface, the benefits of using undoped polysilicon layers are maintained. The diffusion of dopants from the doped polysilicon layer to the undoped polysilicon layers occurs slowly during the thermocycles. Thus, because the doping concentration at the polysilicon-dielectric interface is relatively low during a significant part of the oxidation processes, a reduced smiling effect is achieved. At the same time, the undoped polysilicon layers retain smaller size and more uniform grain structure, resulting in better uniformity and quality of polysilicon-dielectric interface.
Accordingly, by providing a combination of doped and undoped polysilicon layers in each of the floating gate and the control gate, a more flexible process is obtained whereby much of the adverse effects associated with the trade-offs in the polysilicon doping concentration present in conventional processes is eliminated. By providing undoped polysilicon at the dielectric interfaces, an ONO dielectric and a tunnel oxide layer having uniform thicknesses and improved dielectric quality and integrity are achieved while a high enough doping concentration in most of the floating gate and the control gate is maintained. Also, after all the thermal cycles, a relatively homogeneous polysilicon doping across the whole floating gate and control gate is obtained. Further, the tunnel oxide and the ONO dielectric are more uniform both in terms of their geometrical thickness and in terms of their dielectric quality and integrity. High uniformity of tunnel oxide and ONO result in better gate control over the channel, higher coupling ratio between the control gate and the floating gate, enhanced program, erase, and read efficiency, tighter erase distribution, and allow use of lower operating voltages. Further the improved quality of the ONO dielectric and tunnel oxide results in improved charge retention characteristics and overall reliability of the memory cell. Thus, a memory cell with a much improved electrical and reliability characteristics is achieved.
All subsequent processing steps are carried out in accordance with conventional methods.
As shown, by using the doped/undoped polysilicon layers, the “smiling” effect is reduced not only at the outer edges of gate oxide 404 near the drain and source regions (
The MOS transistor polysilicon gate can be formed simultaneously (i.e., using the same mask step) with the control gate of the memory cell. That is, the same two successive deposition steps in forming an undoped polysilicon layer followed by a doped polysilicon layer may be carried out to simultaneously form the control gate of the memory cells and the gate of periphery transistors. In another embodiment, the MOS transistor polysilicon gate can be formed simultaneously with the floating gate rather than with the control gate. In the memory cell embodiment wherein the floating gate comprises three polysilicon layers (
The present invention is not limited in application to MOS transistors and stacked gate non-volatile memories. Any structure wherein doped polysilicon comes in contact with an insulating layer can benefit from the doped/undoped multi-layer polysilicon approach described herein. Examples of other structures include N-channel or P-channel non-volatile memory cells such as ROM, EPROM, EEPROM, and flash EEPROM cells, volatile memory cells such as DRAM and SRAM cells, NMOS and PMOS transistors, and depletion and enhancement transistors. Further, the present invention is not limited to any specific parameters or values indicated herein. For example, the values indicated in the tables above correspond to one particular process and set of targets, and may be varied to accommodate other processes and cell technologies.
While the above is a complete description of preferred embodiments of the present invention, it is possible to use various alternatives, modifications, and equivalents. Therefore, the scope of the present invention should be determined not with reference to the above description but should, instead, be determined with reference to the appended claims, along with their full scope of equivalents.
Claims
1. A method of forming a semiconductor transistor, comprising:
- forming an insulating layer over a silicon region;
- forming an undoped polysilicon layer over and in contact with the insulating layer;
- forming a doped polysilicon layer over and in contact with the undoped polysilicon layer such that at least two edges of the doped polysilicon layer vertically line up with corresponding edges of the undoped polysilicon layer to thereby form sidewalls, the doped and undoped polysilicon layers forming a gate of the transistor; and
- after the doped polysilicon layer forming act, forming source and drain regions in the silicon region,
- wherein dopants from the doped polysilicon layer migrate into the undoped polysilicon layer thereby doping the undoped polysilicon layer, and the semiconductor transistor is not capable of storing data.
2. The method of claim 1 wherein the insulating layer is a gate oxide layer.
3. The method of claim 1 wherein a thickness of the doped polysilicon layer is greater than a thickness of the undoped polysilicon layer by a factor in the range of two to four.
4. The method of claim 1 further comprising:
- forming insulating spacers along the sidewalls of the doped and undoped polysilicon layers.
5. The method of claim 1 wherein said doped polysilicon layer forming act comprises depositing an in-situ doped polysilicon layer.
6. The method of claim 1 wherein the transistor is any one of a NMOS transistor, PMOS transistor, enhancement transistor, and depletion transistor.
7. The method of claim 1 wherein the doped polysilicon layer has a doping concentration and a thickness greater than a thickness of the undoped polysilicon layer so as to prevent polysilicon depletion in the gate.
8. The method of claim 1 further comprising:
- forming a conductive layer comprising tungsten over and in contact with the doped polysilicon layer.
Type: Application
Filed: Jul 13, 2006
Publication Date: Nov 9, 2006
Applicant: Hynix Semiconductor, Inc. (Kyoungki-do)
Inventors: Peter Rabkin (Cupertino, CA), Hsingya Wang (San Jose, CA), Kai-Cheng Chou (San Jose, CA)
Application Number: 11/487,093
International Classification: H01L 21/336 (20060101); H01L 21/22 (20060101);