This invention relates generally to a gate electrode stack and the use of the gate conductor stack.
A gate electrode stack in a conventional DRAM device can comprise the following layers (from substrate (bottom) up):
The polysilicon layer and the W/WN/Ti (or other materials such as WSix) layer comprise the gate conductor (GC) stack in the gate electrode stack. A thin Ti flash layer in the W/WN/Ti metal stack is used to guarantee good contact properties between the metal stack and the polysilicon layer, since Ti silicide is formed at the interface after full processing.
In principle such a gate electrode stack is described in U.S. Pat. No. 6,716,734 B2, which is incorporated herein by reference.
The etching of a GC stack is challenging since after the etching of the metal stack, an over-etch into the polysilicon layer must be performed. This over-etch is difficult to control since the end-point of the over-etch is primarily controllable only through fixed time. There is no end-point signal generated when the over-etch is performed into the polysilicon layer.
Typical etch chemistries for plasma dry-etching of a W/WN/Ti stack are Cl2, NF3, O2 and HBr, which are quite aggressive chemistries, while Cl2, O2 and NF3 are typically used etch chemistries for WiSix etch.
In one aspect, the invention provides a design of a gate electrode stack that is easy to produce. In another aspect, the invention provides a process for the manufacturing of a gate electrode stack. The use of a gate electrode stack is also disclosed.
Within the gate electrode stack, according to embodiments of the invention, a polysilicon layer and a poly-Si1−xGex layer form a GC-stack, the relative position of the layers, i.e., which is above the other one, can vary. The introduction of the poly-Si1−xGex layer has the effect that end-point detection can be achieved between the etching of the polysilicon layer and the poly-Si1−xGex. The physical mechanism for end-point detection is related to the optical emission from excited molecules, which identifies directly or indirectly Ge. The poly-Si1−xGex layer (x<0.8) has similar electrical and structural properties as polysilicon and is compatible with the overall processing.
Properties of poly-SiGe and utilization of poly-SiGe as a gate material can be referred to Dongping Wu's PhD thesis ‘Novel concepts for advanced CMOS: Materials, process and device architecture’ (ISRN KTH/EKT/FR-2004/3-SE and ISSN 1650-8599), which thesis is incorporated herein by reference.
Given the GC-stack according to embodiments of the invention it is possible to reduce the total thickness of the poly layers. Furthermore this improves the uniformity of the etch process.
Other features and advantages of the invention become apparent upon reading of the detailed description of the invention, and the appended claims provided below, and upon reference to the drawings.
For a more complete understanding of the present invention, and the advantages thereof, reference is now made to the following descriptions taken in conjunction with the accompanying drawing, in which:
FIGS. 1 to 6 show process steps in the manufacturing of a first embodiment of the invention;
In a conventional DRAM fabrication process, it is known that a W/WN/Ti/polysilicon (or WSix/polysilicon) gate conductor stack is usually deposited by a physical vapor deposition (PVD) after gate oxide is grown on the silicon substrate. An insulation cap, which is usually nitride, is then deposited atop. A typical GC (gate conductor) structuring process then follows: the cap nitride layer is structured by lithography and nitride etch. The structured cap nitride layer is then served as a hard mask for the subsequent gate stack etch. The metal stack is first etched and a fixed time over-etch into polysilicon is then performed. An encapsulation liner, usually silicon nitride, is deposited and structured if W/WN/Ti metal stack is used. No encapsulation liner is necessary if the WSix is used. Finally, the remaining polysilicon is etched with end-point detection on the underlying gate oxide.
In the following FIGS. 1 to 6 the process steps for manufacturing a first embodiment of the invention are described.
The substrate is covered by a thin layer of gate dielectric 2, preferably a gate oxide.
This first embodiment will have a dual layer poly gate conductor 3, 4 within the gate electrode stack 10. Therefore, a polysilicon layer 3 is positioned on the gate oxide layer 2. The polysilicon layer 3 has thickness in the range of 3 to 100 nm, preferably 30-50 nm. On the polysilicon layer 3 a poly Si1−xGex layer 4 is positioned with a thickness in the range of 3 to 100 nm, preferably 30-50 nm.
Above the Si1−xGex layer 4 the metal layer W/WN/Ti (or WSix) 5 is positioned. The Si1−xGex layer 4 and the metal layer W/WN/Ti (or WSix) 5 are covered with a cap layer 21, here made of silicon nitride.
The thickness of Ti flash layer is preferably in the range of 1 to 15 nm. The thickness of W/WN layer is in the range of 10-100 nm, preferably 30-50 nm. In principle, there is no strict limitation for the layer thicknesses in the usual range.
The next step (
Now (
Finally a polysilicon etch is performed using the cap layer 21 as hard mask (
Given this first embodiment the GC etching process can detect an end-point between the Si1−xGex layer 4 and the polysilicon layer 3. Due to the high etch selectivity between the Si1−xGex layer 4 and the polysilicon layer 3 an effective etch stop is achieved. Consequently, an improved dry etch process window can be obtained and etch uniformity and controllability can be improved.
The other embodiments, depicted in
The difference of the second embodiment (
Since this embodiment has a Si1−xGex layer 4/gate oxide layer 2 interface, the p-type poly gate depletion is improved. This effect is described in connection with a very specific gate dielectric in the article by Lu et al. “Improved Performance of Ultra-Thin HfO2 CMOSFETs Using Poly-SiGe Gates”, IEEE 2002 Symposium on VLSI Technology, which article is incorporated herein by reference.
The third embodiment (
The third embodiment has a triple layer structure (from bottom up):
The thicknesses of the layers are 3-100, 3-100 and 3-100 nm, respectively.
This embodiment keeps the benefits from the first embodiment, while the interface between the poly and metal stack is still Ti/Si instead of Ti/Si1−xGex as is the case in the first embodiment. This removes the possible risk due to complicated Ti−Si1−xGex interaction.
The fourth embodiment depicted in
The thicknesses of the layers are 3-100, 3-100, 3-100 and 3-100 nm, respectively. (Again there is no strict limitation.)
This embodiment inherits the benefits from the second and third embodiment, while enabling the possibility to trim poly gate length. The trimming can be realized by isotropic etching of the poly Si1−xGex layer 41, which is selective to the polysilicon layer 31 and the underlying gate oxide layer 2.
In general the process flow for producing the embodiments is similar to the case of metal/polysilicon gate stack, which is described earlier. The main difference lies in the metal stack over-etch into the polysilicon. Taking the first embodiment as an example: the poly-Si1−xGex layer is etched during the metal stack over-etch. The end-point signal can be observed when poly-Si1−xGex layer is etched away and the underlying polysilicon starts to be etched. Since the dry etch rate of polysilicon is usually much lower than that of the Si1−xGex layer, the polysilicon layer can serve as an etch stop. The improvement of the gate electrode in terms of uniformity and controllability can therefore be obtained because of the capability of an end-point detection during metal over-etch into the poly layer and the etch rate difference between the poly Si1−xGex and polysilicon layers.
One application for an embodiment of the invention is a dual workfunction DRAM.