Dual-Transistor Structure and Method of Formation

Related subject matter may be found in the following copending cases:

(1) U.S. patent application Ser. No. 07/902,216, entitled "Self-Aligned Thin Film Transistor and Method of Formation", by Cooper et al., filed on Jun. 22, 1992; and

(2) U.S. patent application Ser. No. 07/925,136, entitled "A Sidewall Resistive Device and Method of Formation", by Sheng et al., filed on Aug. 06, 1992.

The present invention relates generally to semiconductor technology, and more particularly, to a semiconductor metal oxide semonductor (MOS) transistor structure.

Memory cells within integrated circuits (ICs) are continually shrinking in order to achieve increased packing density and future generations of memory ICs. In order to achieve increased packing density, several transistors must be formed within an increasingly smaller area. Several transistors are formed into a smaller surface area by combining functionality (i.e. sharing gates, double gating, sharing source and/or drain electrodes), vertically stacking devices (i.e. using thin film transistors (TFTs) or silicon on insulator (SOI) technology), using more compact transistor technology (i.e. using vertical transistor), and/or the like. For example, thin film transistors (TFTs) are transistors which are formed above the substrate within a polysilicon layer. Thin film transistors are usually stacked on top of other transistors which are formed within the substrate. This stacking achieves a high packing density. Thin film transistors (TFTs) are lithographically formed (i.e. have lithographically defined source, drain, and gate regions). Lithography limits the minimum size to which a device can be formed. Therefore, as memory arrays approach smaller cell sizes, even lithographically-defined undergated and over-gated TFTs will be too large for use within memory cells.

In order to improve density and reduce process complexity, TFTs are formed having a gate which also functions as a gate for another transistor. This method of sharing gate electrodes between two transistors allows for a reduction in the number of interconnect and gate layers and also reduces cell size. In most cases, a lithographically-defined gate which gates a planar formed transistor (i.e. a transistor formed via lithography within the substrate) functions also as a gate for an overlying TFT. Furthermore, a TFT drain of a first TFT may be used to gate a TFT channel of a second TFT and vice-versa. Once again the TFT gates are lithographically aligned/formed and are therefore no less than roughly 0.25 micron to 0.5 micron in dimensions. Lithographically-formed structures and lithographic alignment tolerances can severely limit memory cell size.

In order to further reduce the size of transistors and memory cells, vertical transistors may be used for memory cell formation. Many vertical transistors exist in the art and are usually formed by one of either trench processing or epitaxial/selective growth. Vertical transistors are in some cases difficult to form and difficult to integrate into a memory cell without using complex processing. In most cases, the formation of vertical transistors depends largely on lithography.

An improved structure is therefore required to reduce transistor top-down surface area and to reduce memory cell size.

The previously mentioned disadvantages are overcome and other advantages achieved with the present invention. In one form, the present invention comprises a dual-transistor device. A substrate is provided. A first transistor is formed at least partially within the substrate. The first transistor has a source, a drain, and a sidewall spacer formation functioning as a gate electrode. A second transistor is formed overlying the first transistor and has a source and a drain. The second transistor uses the sidewall spacer formation as a gate electrode.

The present invention will be more clearly understood from the detailed description below in conjunction with the accompanying drawings.

FIGS. 1-8 illustrate, in cross-sectional form, a method for forming a dual-transistor structure in accordance with the present invention;

FIG. 9 illustrates, in cross-sectional form, an etch stop layer for use within the dual-transistor structure of FIGS. 1-8; and

FIG. 10 illustrates, in cross-sectional form, a dual-transistor structure in accordance with the present invention.

Illustrated in FIG. 1 is a structure suitable for the formation of a dual-transistor structure 10. Structure 10 has a substrate 12. Substrate 12 may be made of silicon, gallium arsenide, silicon on sapphire (SOS), epitaxial formations, germanium, germanium silicon, diamond, silicon on insulator (SOI) material, and/or like substrate materials. Preferably, the substrate 12 is made of silicon which is typically single crystalline. An isolation region 11 and an isolation region 13 are illustrated. In the regions 11 and 13 an isolation structure is formed. Local oxidation of silicon (LOCOS), trench isolation, field shield isolation, polysilicon buffered LOCOS (PBL), frame masked PBL, polysilicon encapsulated LOCOS, or like isolation structures may be used.

A conductive layer 16 is formed overlying the substrate 12. Conductive layer 16 is preferably a polysilicon or amorphous silicon layer, but may be a germanium silicon layer, a silicided layer, a salicided layer, a doped silicon layer, an epitaxial material, a selectively grown layer of semiconductive material, or the like. A dielectric layer 18 is formed overlying the conductive layer 16. The dielectric layers described herein may be wet or dry silicon dioxide (SiO.sub.2), a nitride material, tetra-ethyl-ortho-silicate (TEOS) based oxides, boro-phosphate-silicate-glass (BPSG), phosphate-silicate-glass (PSG), boro-silicate-glass (BSG), oxide-nitride-oxide (ONO), tantalum pentoxide (Ta.sub.2 O.sub.5), plasma enhanced silicon nitride (P-SiN.sub.x), titanium oxide, oxynitride, and/or like dielectric materials. Specific dielectrics are noted herein when a specific dielectric material is preferred or required. Preferably, layer 18 is an oxide layer such as a TEOS-based oxide.

A conductive layer 20 is formed overlying the dielectric layer 18. The conductive layer 20 is similar to conductive layer 16 in dimensions and composition of material. A dielectric layer 22 is formed overlying the conductive layer 20. The dielectric layer 22 is preferably a nitride material such as silicon nitride.

In FIG. 2, the isolation regions 11 and 13 of FIG. 1 are illustrated as containing field oxide regions 14. Field oxide regions 14 are formed by conventional LOCOS techniques or like isolation techniques.

In FIG. 2, the layers 16, 18, 20, and 22 are masked by a layer or photo-sensitive material (not illustrated), such as photoresist (not illustrated), and etched to form a patterned stack of materials. The etching of the layers 16, 18, 20, and 22 forms sidewalls of the conductive layers 16 and 20. Depending upon the materials chosen for conductive layer 16, etching selective to the substrate 12 may be difficult. Therefore, trenching of the substrate 12 may occur in regions adjacent the layer 16 in FIG. 2. This trenching may have an advantage which is discussed in a subsequent paragraph.

In FIG. 3 a dielectric layer 24 is grown and/or deposited overlying a surface of the substrate 12 and laterally adjacent the sidewalls of conductive layers 16 and 20. An optional spacer (not illustrated) may be formed laterally adjacent the sidewalls of the conductive layers 16 and 20 or adjacent the dielectric layer 24. A conductive layer 26 is formed overlying the dielectric layer 24, and is preferably formed from polysilicon or amorphous silicon. A blanket ion implantation step or in-situ doping may be used to uniformly dope the conductive layer 26 to a specific conductivity and conductivity type (i.e. N-type or P-type). The conductive layer 26 is preferably doped as N-type material and is typically on the order of a few hundred Angstroms in thickness to several thousand Angstroms in thickness.

In FIG. 4, the conductive layer 26 is etched by a reactive ion etch (RIE) or a like etch process to form a sidewall spacer 26a. The spacer 26a will surround the feature defined by the patterned portions of the layers 16, 18, 20, and 22. The circular continuity of the spacer is not beneficial in structure 10. Therefore, a mask and etch process (usually an isotropic etch process) is used to separate the spacer 26a into two portions. Therefore, the sidewall spacer 26a is hereafter referred to as sidewall spacers 26a (i.e. two spacers, one on each side of the patterned portions of layers 16, 18, 20, and 22). The thickness of the spacers 26a may range from 300 Angstroms to over 2000 angstroms depending upon the initial thickness of the conductive layer 26. Preferably, the spacers 26a are made thin in order to reduce the surface area of structure 10.

An ion implant step is used to form drain regions 28 and 30 within the substrate 12 in a self-aligned manner.

In FIG. 5, the dielectric layer 22 is removed by a selective etch process. A dielectric layer 32 is formed over all exposed portions of the sidewall spacers 26a. The sidewall spacers 26a are therefore fully insulated by the dielectric layers 24 and 32. Preferably, the layer 32 is a grown silicon dioxide or a grown polysilicon dioxide (polyox) layer, a deposited oxide layer, or a composite grown/deposited oxide.

In FIG. 6, a conductive layer 34 is formed overlying the dielectric layers 24 and 32. Conductive layer 34 is preferably polysilicon or amorphous silicon and is either insitu doped or blanket ion implanted to a conductivity and conductivity type (i.e. N-type or P-type). Typical dopant atoms are boron, arsenic, and/or phosphorus. The layer 34 is typically doped as an N-type layer of material due to the fact that layer 32 is used, in most cases, to form P channel transistors. Unless the conductive layer 34 is laser recrystallized silicon, epitaxial silicon, or the like, then any transistor formed within conductive layer 34 is considered to be a thin film transistor (TFT) similar to the polysilicon TFTs used in static random access memories (SRAMs).

A dielectric layer 36 is formed overlying the conductive layer 34. Dielectric layer 36 is formed as a TEOS-based oxide or a like layer of material.

In FIG. 7, a masking layer (not illustrated) is formed and patterned overlying the spacers 26a and the dielectric layer 36. The masking layer (not illustrated) is used to mask an ion implant step which is used to form doped regions referred to as drain regions 34c. One or more masking and etch steps are used via at least one masking layer (not illustrated) to form openings 37 through the dielectric layers 36 and 24 and the conductive layer 34. The openings form exposed portions of the drain regions 28 and 30.

A heat cycle is used to drive dopant atoms out of conductive layer 16 into the substrate to form a ground contact or source region 38. Simultaneously, a diffusion region referred to as one of either a power supply region or a source region 34a is formed via the same heat cycle. Region 34a and regions 34c define boundaries of two channel regions 34b within layer 34. Therefore, layer 34 forms two silicon on insulator transistors or TFTs. One transistor overlies one of the spacers 26a and the other transistor overlies the other spacer 26a. The spacers 26a function as gates for the transistors formed within layer 34.

Two transistors are also formed within the substrate 12. One of the spacers 26a forms a gate electrode between the drain region 30 and the source region 38. Another of the spacers 26a forms a gate electrode between the drain region 28 and the source region 38. Channel regions are defined between the source region 38 and the drain region 28 and between the source region 38 and the drain region 30. It is known in the art that source and drain electrodes are usually symmetrical and may be interchanged without loss of functionality. The four transistors illustrated in FIG. 7 may each be formed as either N-channel or P-channel transistors. In many cases, the two transistors formed within the substrate will be formed as N-channel transistors and the two transistors formed from conductive layer 34 will be P-channel transistors or vice versa.

In FIG. 8, openings 37 are at least partially filled with a conductive material 40. Conductive material 40 functions as an interconnect layer. Material 40 may be titanium nitride, a metal, a polysilicon layer, a silicided layer, a salicided layer, or a like conductor or semiconductor. In some cases, the layer 40 may be a composite material of several conductive layers, barrier layers, etch stop layers, polish stop layers, and/or the like. Material 40 may be selectively deposited or blanket deposited and subsequently masked and etched. Chemical mechanical polishing (CMP) may be used to form plug regions, planarized contacts, Damascene interconnects, or the like. An oxide layer (not illustrated) may be optionally formed and optionally reflowed or planarized before etching the openings 37 in FIG. 7. This extra oxide layer (not illustrated) will produce improved isolation and planarity.

The conductive material 40 connects the drain regions 34c to the drain regions 30 and 28, respectively. Any PN junctions resulting from electrical contacts between regions 24c, 30, and 28 may be electrically short circuited by using a material 40 which contains a metallic compound. In another form a sidewall spacer or sidewall oxidation process may be used to isolate the drain regions 34c from the drain regions 38 and 30.

Structure 10 of FIG. 8 may be used to form a compact SRAM cell. To form an SRAM cell, the two transistors within the substrate 12 are formed as N-channel transistors and function as latch transistors for an SRAM cell. The two transistors formed within the conductive layer 34 are formed as P-channel TFTs or P-channel transistors and function as pull-up transistors in the SRAM cell. Each of the two regions formed by the conductive material 40 is connected to a different pass transistor (not illustrated). The conductive layer 20 supplies a power supply potential, and the conductive layer 16 supplies a ground potential.

In general, the trenching which occurs in the substrate 12 in FIG. 2 may be used to contain/minimize the diffusion of dopant atoms from the conductive layer 16 (i.e. contain the lateral diffusion of source region 38). The trenching will also increase the channel length of the pass transistors formed within the substrate 12 to ensure that the source region 38 and drain regions 28 and 30 do not electrically short circuit to one another.

FIG. 9 illustrates that an etch stop layer 42 having an opening (not labeled) may be used to form a device similar to that illustrated in FIG. 2 without resulting in trenching of the substrate 12. The etch stop layer 42 is typically a grown silicon dioxide layer and is on the order of 50-350 Angstroms thick.

FIG. 10 illustrates a dual-transistor structure wherein the source and drain regions are trenched into the substrate. This trenching increases the channel length of the substrate formed transistors. This increase in channel length may further stabilize an SRAM cell. Furthermore, thermal cycle constraints placed on back-end processing are relaxed due to the fact that diffusions formed within the substrate 12 are separated by a greater distance.

While the present invention has been illustrated and described with reference to specific embodiments, further modifications and improvements will occur to those skilled in the art. For example, the transistors taught herein may be formed as either N-channel or P-channel transistors. SOI transistors, TFT transistors, or polysilicon-formed transistors may be formed. The structure described herein may be used for other applications besides SRAM cells, such as logic circuits, inverters, and the like. The order in which operations are performed in the above description may be altered and still result in the same functional structure. For example, diffusions could be formed by a heat cycle before or after forming contacts to sources and drains. Various oxides in the structure listed above are used for gate oxides and others are used for isolation purposes. The thickness of the isolation oxides may be adjusted and optimized based upon an application in order to provide adequate isolation. The thicknesses of the gate oxides may be any thickness within the approximate range of 50 Angstroms to 400 Angstroms. It is to be understood, therefore, that this invention is not limited to the particular forms illustrated and that it is intended in the appended claims to cover all modifications that do not depart from the spirit and scope of this invention.