Patent Application
20070114616
The invention generally relates to a field effect transistor formed in or on a semiconductor substrate and to a method of manufacturing the field effect transistor. The invention particularly relates to field effect transistors provided in small pitch integrated circuits and/or in DRAM memory devices.
In the field of manufacturing semiconductor devices, particularly DRAM devices (dynamic random access memory), pitches and line widths are continuously decreased in order to improve the integration level of the devices. With regard to field effect transistors generally formed with the semiconductor devices this decrease of dimensions poses several problems when considering the electrical characteristics to be maintained. When entering a sub-em linewidth regime, leakage current effects between gate, source, drain and/or well become important contributors to deficiencies inherent in an integrated circuit.
Accordingly, when designing an integrated circuit these effects have to be considered by adapting corresponding supply voltages, retaining minimum lateral dimensions or layer thicknesses (e.g., of the gate oxide) or applying appropriate doping levels, etc., with respect to the components of a respective transistor. However, continued shrinking of the dimensions will inevitably lead to physical limits as the maximum allowable leakage current will further be reduced. In the case of DRAM memories, where currents or charges being trapped in a conductive filling of a storage node are used to store the information, undesired loss of currents necessitates a more frequent refresh operation of the stored current thus counteracting the gain in speed due to the shrinking process.
Different mechanisms related to leakage currents are known and well studied, such as junction leakage, gate-induced drain leakage (GIDL) and drain-induced barrier lowering (DIBL), etc.
Junction leakage has its origin in minority carrier diffusion and drift near edges of depletion regions (also referred to as channel region throughout this document) of a transistor. Electron hole pair generation may further be responsible for this kind of leakage. Additionally, in the case of heavily doped source/drain regions, band-to-band tunneling may occur.
Gate-induced drain leakage (GIDL) has its origin in the strong field generated under certain circumstances near the drain junction. In case of, e.g., an n-channel field effect transistor (N-MOSFET) the gate voltage may be biased, in order to drive the transistor sufficiently below the threshold voltage (transistor off, at 0.0 V or below). As a result, holes accumulate in the depletion surface region below the gate electrode adjacent to the gate oxide layer thereby forming a channel region, which now acts as a highly doped p-type region within a moderately doped p-type well (substrate), while both the substrate and the channel hold the same bias potential. A strong field is then generated, if the n-type drain region is simultaneously connected to the supply voltage. Minority charge carriers and band-to-band tunneling then lead to a current from n-type drain to p-type well (substrate).
Drain-induced barrier lowering (DIBL) occurs when a high drain voltage is applied to a transistor having a particularly short channel. The profile of voltage potential along the channel is affected and carriers are injected from the source region towards the channel surface adjacent to the gate oxide layer. The channel width is narrowed, which again may influence the effective threshold voltage of the field effect transistor.
Gate-induced drain leakage (GIDL) mainly limits the minimum thickness of the gate oxide layer and the voltage supply to source/drain, while drain-induced barrier lowering (DIBL) limits the channel width.
With regard to the gate-induced drain leakage (GIDL), US Patent Publication No. 2003/0094651 A1, published May 22, 2003, a proposed field effect transistor has a primary gate electrode assisted by an auxiliary electrode formed as a spacer adjacent to an oxide film, which isolates both electrodes from each other, respectively. The primary electrode contacts a gate oxide formed on a substrate, while the auxiliary electrode contacts said oxide film, which is arranged between the auxiliary electrode and each of the source/drain regions of the transistor. Both electrodes are independently supplied with their own voltage sources, respectively.
In operation, when the primary gate electrode is biased to 0.0 V upon a refresh of a DRAM, the auxiliary electrode is simultaneously supplied with the same voltage as the underlying source/drain region in order to suppress the occurrence of a GIDL current.
One aspect of the invention improves the electrical characteristics of an NMOS or PMOS field effect transistor. In particular, an aspect of the invention reduces leakage currents between source/drain, gate and well/substrate of a transistor formed in a semiconductor device.
These and other benefits are obtained by a field effect transistor formed in a semiconductor substrate, comprising a first and a second doped source/drain region, both regions arranged within the substrate on either side of a gate electrode, a channel region formed within the substrate between both doped source/drain regions beneath the gate electrode, a gate oxide layer formed upon the semiconductor substrate, the gate electrode, which contacts a surface of the gate oxide layer and which further comprises at least a first and a second conductive layer, the first and second conductive layers being made of materials having different work functions with respect to each other. The first conductive layer of the gate electrode contacts the gate oxide layer within a first portion of the surface, the second conductive layer contacts the gate oxide layer within a second portion of the surface, and the first conductive layer is further conductively connected to the second conductive layer.
In a preferred embodiment, a field effect transistor (FET) is provided, which may be an n-channel MOSFET or a p-channel MOSFET. The transistor comprises a gate electrode, a first and a second source/drain region, a channel region arranged between the source/drain regions, and a gate dielectric layer, e.g., a gate oxide.
The gate electrode comprises a first conductive layer as well as a second conductive layer. Both conductive layers simultaneously contact the gate dielectric layer in respective first and second portions of a surface of that layer. Both portions, the first and the second portion, preferably overlap with the channel region in order to affect its electrical characteristics during operation.
The material selected each for the first and second conductive layers, respectively, differs in that the work function, i.e., the energy difference needed to extract an electron from the material to vacuum, is not the same with respect to both layers. Accordingly, the materials selected for the layers relate to different chemical elements or compounds, or alternatively to similar elements or compounds, which, however, have been modified by, e.g., different types of doping, etc., in order to yield distinct conductor characteristics.
Further, the first and the second conductive layers are (electrically) conductively connected with each other. This connection is permanent, i.e., the conductive connection between both layers is not only provided by means of a remote circuit on specific occasions. Rather, the connection is provided within the same transistor by means of a direct or an indirect contact between both layers. According to one embodiment, the conductive connection is provided via a third conductive layer, which may be arranged on top of one of the first or second conductive layers. The third layer may comprise the same or a different chemical composition as one of the first or second conductive layers.
As a consequence of this configuration, the gate contact area on top of the gate dielectric layer is established by two different conductive layers, which—when being connected to the same power or voltage supply as these are conductively connected—exhibit different work functions at the same voltage potential. This portioned gate contact area adversely affects the electrical characteristics at the surface of the channel region on the opposite side of the gate dielectric layer. However, varying the work function acts on the depletion region below as if the voltage potential of the same layer has been changed. As a result, the profile of the depletion or accumulation characteristic along the length direction of the channel region changes according to the work function of the material chosen for the two conductive layers, respectively.
In one embodiment, n-doped or p-doped poly silicon is employed for the first layer. In a further embodiment the second conductive layer may be a material selected from a group of so-called midgap materials. These materials are characterized by moderate values of the work function. However, as there are varying definitions available for midgap materials, a range of work functions provided for this specific embodiment may have a minimum value of 4.4 eV and a maximum value of 4.9 eV. It is noted that the invention is not limited to any of such ranges provided herein or in other literature.
According to another definition of midgap materials, the work function of the midgap materials is larger than that of materials that are similar to heavily n-doped poly silicon, and is smaller than materials that are similar to heavily p-doped poly silicon.
Examples of suitable midgap materials are tungsten (W), titanium nitride (TiN), tungsten silicide (WSiX), nitrogen implanted molybdenum (Mo(N)) and tantalum nitride. However, it is noted that the invention is not limited thereto.
According to another embodiment, the second conductive layer is formed as a vertical spacer at a sidewall of a gate stack, which comprises the first conductive layer. The effect is that the second conductive layer has a reduced footprint with regard to its portion of the gate contact area, and the overall characteristics are dominated by the first conductive layer, which has a larger footprint. A suitable material may be chosen, therefore, in order to set the threshold value. Nevertheless, as the vertical spacer of the second conductive layer is located near the drain/well junction, its work function may, according to the specific design, be selected to reduce leakage effects particularly in that region, despite its small footprint. The occurrence of strong fields near the junctions may thus be mitigated.
A further advantage of the vertical spacer is that, due to its small footprint, the dimension of the gate electrode is not considerably increased.
In case a horizontal layer in a gate stack is formed with respect to the first conductive layer and a vertical spacer is formed at a gate stack sidewall with respect to the second conductive layer, and in case an n- or p-doped poly silicon as well as a midgap material is chosen for the respective conductive layers, advantageous embodiments of the invention relate to both possibilities that the midgap material is selected either for the first or the second conductive layer.
Preferred embodiments of the invention become particularly advantageous with respect to DRAM memory devices. Herein, a specific need for further shrinking exists. Further, it has been found that a drain region suffers from leakage currents only on a single side of the transistor—the drain connected to the bit line.
According to an embodiment of the invention, a single-sided vertical conductive spacer is arranged with respect to this drain region, which is that region connected to a voltage supply, e.g., a bit line.
The complementary source region may be—according to still a further embodiment—connected to a storage node. The storage node according to this embodiment may reside in a trench capacitor or in a stacked capacitor.
Another aspect of the invention is a method of manufacturing a field effect transistor in a semiconductor substrate, comprising providing the semiconductor substrate, depositing a gate dielectric layer on the semiconductor substrate, forming a first conductive layer of a material having a first work function upon the gate dielectric layer, depositing an isolating capping layer, etching the first conductive layer and the isolating capping layer to form a gate stack upon the gate dielectric layer, depositing a second conductive layer of a material having a second work function different from the first work function on a sidewall of the etched gate stack and on the gate dielectric layer to form a conductive vertical spacer, the spacer and the gate stack forming a gate electrode, and implanting the semiconductor substrate where it is not shielded by the gate stack and the vertical conductive spacer to form first and second source/drain regions.
Further advantageous aspects and embodiments are evident from the appended claims.
Other objects and many of the attendant advantages of embodiments of the present invention will be readily appreciated and become better understood by reference to the following more detailed description of preferred embodiments in connection with the accompanied drawings. Features that are substantially or functionally equal or similar will be referred to with the same reference signs.
The following list of reference symbols can be used in conjunction with the figures:
1 MOSFET 28 source/drain regions
4 gate electrode 30 LDD
10, 102, 104, 106 1st conductive layer 32, 322, 323 isolation spacer
12, 122, 124, 126 2nd conductive layer 38 sidewall of gate stack
14 3rd conductive layer 40, 42 portions of gate contact area
16 capping layer 50 electrically conductive connection
18 sidewall oxidation layer 500 voltage supply
20, 202, 204, 206 nitride liner 70 storage node
22 gate dielectric layer 75 sense amplifier
24 semiconductor substrate 80-100, 922, 924, 926 method steps
26 channel region, depletion region
The gate electrode 4 comprises a gate stack with a sequence of layers 10, 14, 16—from bottom to top—starting with n-doped poly silicon (first conductive layer 10), tungsten silicide (third conductive layer 14) and silicon nitride (isolating capping layer 16). The first conductive layer 10 of poly silicon provides the actual gate electrode, while the third conductive layer 14 of tungsten silicide provides for a conductor having a low ohmic resistance, which may further serve, e.g., as a word line in a DRAM. The third conductor 14 is coupled to a voltage supply 500 selectively providing levels of a voltage potential.
The gate stack has sidewalls 38. A portion of the sidewalls 38 provided by the lowermost first conductive layer 10 of poly silicon is covered with a sidewall oxidation layer 18. Further, a vertical spacer formed by a second conductive layer 12 of a midgap material such as tungsten or tungsten silicide is arranged adjacent to both sidewalls 38 of the gate stack. A nitride liner 20 covers the vertical spacers on both sides of the gate electrode 4. As indicated by arrow 50, an electrical connection is established between the first 10 and the second 12 conductive layers via the third conductive layer 14.
The first conductive layer 10 of poly silicon is doped to have a work function of 4.1 eV, and the second conductive layer of tungsten, or tungsten silicide has a work function of 4.6 eV, or 4.7 eV respectively—with respect to vacuum.
The gate electrode 4 has a device length of 90 nm, of which 65-70 nm are contributed by the extension of the first conductive layer 10, and 10 nm are provided by each of the two vertical spacers on either side of the gate stack. A surface 40, or gate contact area is divided into a first portion 40 provided by the footprint of the first conductive layer 10, and into a second portion(s) 42 provided by the footprint of the second conductive layer 12, i.e., the vertical spacer shown in
Within substrate 24, first and second highly n-doped source/drain regions 28 are each formed by means of implantation. Lightly n-doped drains 30 (LDD) absorb strong field gradients with respect to a p-type channel region (well) with respect to the source/drain regions 28.
The reduction of field gradients due to the LDDs is further supported by portions 42 due to the second conductive layer. As the work function is larger in this case, accumulation of holes near the surface of the depletion or channel region 26 is reduced when the voltage of the gate electrode 4 has dropped to 0.0 V or below near the junction edges of the depletion region.
Source/drain regions 28 will occasionally—and depending on the design of the circuit—be connected with conductive contacts, which are not shown here for simplicity and to illustrate the principles of embodiments according to the invention.
Referring now to
The single layer gate stack shown in
The overhang is formed by first depositing the first conductive layer 102 upon the gate dielectric layer 22, and then the isolating capping layer 16 is formed above (if there are further layers in the stack) or directly on top of the first conductive layer 102. Next, the anisotropic etch is performed to define the stack. Then, an isotropic etch is performed to recess the stack material of the first conductive layer below the isolating capping layer 16 in a horizontal direction, with a selectivity of the etch process of the conductive layer material as compared with the isolating capping material.
The second conductive layer 122 is further covered with a nitride liner 202, followed by an isolation spacer 32 of, e.g., silicon dioxide.
Also indicated is the connection of a first of the source/drain regions 28 with a storage node 70, and a second of the source/drain regions 28 with a sense amplifier 75. The sense amplifier in operation may apply a voltage to the respective source/drain region when reading out or writing in information from/to the storage node device, wherein the information is represented by charge carriers.
Layers 10, 14, 16 of poly silicon, tungsten and silicon nitride are sequentially deposited upon the gate oxide (step 86). A second lithographic step (88) is performed to structure the gate stack having a desired width of, e.g., 65-70 nm. Lithographic steps as referred to herein may comprise coating the substrate with a resist, exposing and developing the resist, removing portions of the resist and etching the stacked layers using the remaining resist as an etch mask.
In case tungsten is employed for the third conductive layer 14 and poly silicon is employed for the first conductive layer 10, an optional barrier layer of TiN or WN having a thickness of, e.g., 4-7 nm may be arranged between the first and third conductive layers.
Next, the poly silicon portion of a resulting sidewall 38 of the gate stack is oxidized to yield a thin oxidation layer 18 (step 90). Adjacent to the sidewalls 38 and upon the oxidation layer 18 is formed a second conductive layer 12 of, e.g., a midgap material such as tungsten or tungsten silicide (step 92), see also
In one embodiment this layer 12 is first deposited and then exposed to an anisotropic etch process, wherein only vertical portions of the layer 12 are retained, thus yielding the vertical spacers.
In another embodiment, illustrated in
