Trench Transistor With Metal Spacers

1. Field of the Invention

The present invention relates to integrated circuits, and more particularly to insulated-gate field-effect transistors.

2. Description of Related Art

An insulated-gate field-effect transistor (IGFET), such as a metal-oxide semiconductor field-effect transistor (MOSFET), uses a gate electrode to control an underlying surface channel joining a source and a drain. The channel, source and drain are located in a semiconductor substrate, with the substrate being doped oppositely to the source and drain. The gate electrode is separated from the semiconductor substrate by a thin insulating layer such as a gate oxide. The operation of the IGFET involves application of an input voltage to the gate electrode, which sets up a transverse electric field in the channel in order to modulate the longitudinal conductance of the channel.

In typical IGFET processing, the source and drain are formed by introducing dopants of a second conductivity type (P or N) into the semiconductor substrate of a first conductivity type (N or P) using a patterned gate electrode as a mask. This self-aligning procedure tends to improve packing density and reduce parasitic overlap capacitances between the gate electrode and the source/drain regions.

Polysilicon (also called polycrystalline silicon, poly-Si or poly) thin films have many important uses in IGFET technology. One of the key innovations is the use of heavily doped polysilicon as the gate electrode in place of aluminum. Since polysilicon has the same high melting point as a silicon substrate, it can be deposited prior to source and drain formation, and serve as a mask during introduction of the source and drain regions by ion implantation. The resistance of polysilicon can be further reduced by forming a silicide on its top surface.

There is a relentless trend to miniaturize semiconductor dimensions. The number of IGFETs that can be manufactured on an integrated circuit chip can be increased by decreasing the horizontal dimensions. Resolution refers to the horizontal linewidth or space that a lithographic system can adequately print or resolve. Lithographic systems include optical projection and step and repeat equipment, and electron beam lithography equipment. In optical systems, for instance, resolution is limited by the equipment (e.g., diffraction of light, lens aberrations, mechanical stability), optical properties of the photoresist (e.g., resolution, photosensitivity, index of refraction), and process characteristics (e.g., softbake step, develop step, postbake step, and etching to step).

Furthermore, scaling down the horizontal dimensions generally requires a corresponding decrease in the vertical dimensions. As IGFET vertical dimensions are reduced and the supply voltage remains constant (e.g., 3V), the maxumum lateral electric field in the channel near the drain tends to increase. If the electric field becomes strong enough, it can give rise to so-called hot-carrier effects. For instance, hot electrons can overcome the potential energy barrier between the substrate and the gate insulator thereby causing hot carriers to become injected into the gate insulator. Trapped charge in the gate insulator due to injected hot carriers accumulates over time and can lead to a permanent change in the threshold voltage of the device.

A number of techniques have been utilized to reduce hot carrier effects. One such technique is a lightly doped drain (LDD). The LDD reduces hot carrier effects by reducing the maximum lateral electric field. Reducing the electric field on the order of 30-40% can reduce hot-electron-induced currents by several orders of magnitude. The drain is typically formed by two ion implants. A light implant is self-aligned to the gate electrode, and a heavy implant is self-aligned to the gate electrode on which sidewall spacers have been formed. The spacers are typically oxides or nitrides. The purpose of the lighter first dose is to form a lightly doped region of the drain (or LDD) at the edge near the channel. The second heavier dose forms a low resistivity region of the drain, which is subsequently merged with the lightly doped region. Since the heavily doped region is farther away from the channel than a conventional drain structure, the depth of the heavily doped region can be made somewhat greater without adversely affecting the device characteristics. The lightly doped region is not necessary for the source (unless bidirectional current is used), however LDD structures are typically formed for both the source and drain to avoid the need for an additional masking step.

Disadvantages of LDDs are their increased fabrication complexity compared to conventional drain structures, and parasitic resistance. LDDs exhibit relatively high parasitic resistance due to their light doping levels. During operation, the LDD parasitic resistance can decrease drain current, which in turn may reduce the speed of the IGFET.

In the manufacture of integrated circuits, the planarization of semiconductor wafers is becoming increasingly important as the number of layers used to form integrated circuits increases. For instance, the gate electrode and/or metallization layers formed to provide interconnects between various devices may result in nonuniform surfaces. The surface nonuniformities may interfere with the optical resolution of subsequent lithographic steps, leading to difficulty with printing high resolution patterns. The surface nonuniformities may also interfere with step coverage of subsequently deposited metal layers and possibly cause open circuits.

Accordingly, a need exists for an IGFET that can be manufactured with reduced horizontal dimensions, that preferably includes an LDD with reduced parasitic resistance as well as a substantially planar top surface. It is especially desirable that the IGFET have a channel length that can be significantly smaller than the minimum resolution of the available lithographic system.

The present invention provides an insulated-gate field-effect transistor (IGFET) with a gate electrode in a trench (i.e., a trench transistor) and metal spacers in the trench. A gate insulator is disposed on the bottom surface of the trench, the gate electrode is disposed on the gate insulator, metal spacers are disposed between the gate electrode and opposing sidewalls of the trench, and protective insulators are disposed between the gate electrode and the metal spacers. The source, drain and channel are adjacent to the bottom surface of the trench. Preferably, the channel is substantially aligned with the bottom surface of the gate electrode. If the trench length corresponds to the minimum resolution of a lithographic system, then the channel length is significantly smaller than the minimum resolution. Furthermore, lightly doped source and drain regions adjacent to the channel can contact the metal spacers at the bottom surface of the trench, and the gate electrode can be substantially aligned with the top surface of the substrate. In this manner, a highly miniaturized IGFET having an LDD with reduced parasitic resistance can be produced.

A key feature of the invention is the use of metal spacers in the trench, adjacent to the bottom surface and sidewalls of the trench, in contact with the source and drain, and electrically isolated from the gate electrode.

Preferably, the source and drain include heavily doped regions that extend to the top surface of the substrate and contact the metal spacers at the sidewalls, and also include lightly doped regions that contact the metal spacers at the bottom surface. In this manner, the metal spacers reduce the parasitic resistance of the lightly doped regions between the heavily doped regions and the channel. It is also preferred that protective insulators are disposed between the gate electrode and the metal spacers, that substantially all of the gate electrode is within the trench, and that the channel length is less than half the trench length and less than 2000 angstroms. As exemplary materials, the gate electrode is polysilicon, the gate insulator and protective insulators are a continuous layer of silicon dioxide, and the metal spacers are a highly conductive metal such as aluminum, tungsten, titanium, cobalt, or combinations thereof.

Another aspect of the invention is a method of forming an IGFET. The method includes forming a trench with a bottom surface and opposing sidewalls in a substrate, forming a source and drain adjacent to the bottom surface, then forming metal spacers on the bottom surface and in contact with the source and drain, forming a gate insulator on the bottom surface between the metal spacers, forming protective insulators on the metal spacers, and forming a gate electrode on the gate insulator and protective insulators such that the gate electrode is spaced from and electrically isolated from the metal spacers and the substrate. The gate insulator and protective insulators can be formed simultaneously as a continuous insulative layer. The source and drain are implanted and activated before forming the metal spacers in order to avoid melting the metal spacers. Preferably, the metal spacers are formed by depositing a blanket layer of conductive metal over the substrate and then applying an anisotropic etch, and the gate electrode is formed by depositing a blanket layer of gate electrode material over the substrate and then polishing the gate electrode material until the gate electrode is substantially aligned with a planar top surface of the substrate.

A first embodiment of the method includes providing a P-type substrate, implanting an N+ doped layer into the substrate, patterning a photoresist layer over the substrate, etching a trench completely through the doped layer and partially through the substrate to split the doped layer into source and drain regions, stripping the photoresist layer, applying a high-temperature anneal to form the source and drain by activating the source and drain regions and diffusing the source and drain regions beneath the bottom surface, forming metal spacers on the bottom surface and sidewalls of the trench, forming a gate insulator on a central portion of the bottom surface between the metal spacers, forming protective insulators on the metal spacers, depositing a gate electrode material over the substrate, and polishing the gate electrode material to form a gate electrode substantially aligned with a planar top surface of the substrate.

A second embodiment of the method includes providing a P-type substrate, implanting a doped layer that includes an N+ doped upper layer and an N- doped lower layer into the substrate, patterning a photoresist layer over the substrate, etching a trench completely through the doped upper layer and partially through the doped lower layer to split the doped upper layer into heavily doped source and drain regions, stripping the photoresist layer, forming disposable spacers on the bottom surface and sidewalls of the trench, counterdoping the substrate beneath a central portion of the bottom surface between the disposable spacers to form a P-type channel that splits the doped lower layer into lightly doped source and drain regions, stripping the disposable spacers, applying a high-temperature anneal to form the source and drain by activating the lightly and heavily doped source and drain regions, forming metal spacers on the bottom surface and sidewalls of the trench, forming a gate insulator on the central portion of the bottom surface between the metal spacers, forming protective insulators on the metal spacers, depositing a gate electrode material over the substrate, and polishing the gate electrode material to form a gate electrode substantially aligned with a planar top surface of the substrate.

A third embodiment of the method includes providing a P-type substrate, implanting an N+ doped first layer into the substrate, patterning a photoresist layer over the substrate, etching a trench completely through the doped first layer and partially through the substrate to split the doped first layer into heavily doped source and drain regions, implanting an N- doped second layer into the substrate beneath the bottom surface, stripping the photoresist layer, forming disposable spacers on the bottom surface and sidewalls of the trench, counterdoping the substrate beneath a central portion of the bottom surface between the disposable spacers to form a P-type channel that splits the doped second layer into lightly doped source and drain regions, stripping the disposable spacers, applying a high-temperature anneal to form the source and drain by activating and merging the lightly and heavily doped source and drain regions, forming metal spacers on the bottom surface and sidewalls of the trench, forming a gate insulator on the central portion of the bottom surface between the metal spacers, depositing a gate electrode material over the substrate, and polishing the gate electrode material to form a gate electrode substantially aligned with a planar top surface of the substrate.

A primary advantage of the invention is that the IGFET can have a channel length that is significantly smaller than the minimum resolution of the available lithographic system, thereby providing a next generation transistor with the present generation lithography. As further advantages, the source and drain can include lightly doped regions adjacent to the channel and heavily doped regions adjacent to the top surface of the substrate, the metal spacers can reduce the parasitic resistance of the lightly doped regions, and the gate electrode can be substantially aligned with the top surface of the substrate.

These and other aspects, features and advantages of the invention will be further described and more readily apparent from a review of the detailed description of the preferred embodiments which follow.

The following detailed description of the preferred embodiments can best be understood when read in conjunction with the following drawings, in which:

FIGS. 1A-1J show cross-sectional views of successive process steps for forming an IGFET with a gate electrode and metal spacers in a trench in accordance with a first embodiment of the invention;

FIGS. 2A-2K show cross-sectional views of successive process steps for forming an IGFET with a gate electrode and metal spacers in a trench in accordance with a second embodiments of the invention; and

FIGS. 3A-3L show cross-sectional views of successive process steps for forming an IGFET with a gate electrode and metal spacers in a trench in accordance with a third embodiment of the invention.

In the drawings, depicted elements are not necessarily drawn to scale and like or similar elements may be designated by the same reference numeral throughout the several views.

FIGS. 1A-1J show cross-sectional views of successive process steps for forming an IGFET with a gate electrode and metal spacers in a trench in accordance with a first embodiment of the invention.

In FIG. 1A, a semiconductor substrate suitable for integrated circuit manufacture is provided. For instance, substrate 102 includes a P- epitaxial surface layer with a <100> orientation, a boron concentration of 1.times.10.sup.15 atoms/cm.sup.3 and a resistivity of 12 ohm-cm. Preferably, the epitaxial surface layer is disposed on a P+ base layer (not shown) and has a planar top surface 104. Oxide layer 106 with a thickness of 500 angstroms is formed on top surface 104.

In FIG. 1B, the structure is subjected to ion implantation of arsenic and phosphorus, indicated by arrows 108, at a combined dosage in the range of 1.times.10 to 5.times.10.sup.15 atoms/cm and an energy in the range of 2 to 80 kiloelectron-volts. Preferably, the arsenic dosage is about an order of magnitude greater than the phosphorus dosage. This forms N+ doped layer 110 in substrate 102 with a combined arsenic and phosphorus concentration in the range of 1.times.10.sup.20 to 1.times.10.sup.21 atoms/cm.sup.3 and a depth of 1800 angstroms below top surface 104. Doped layer 110 has a graded doping profile in the vertical direction so that the deepest regions are the most lightly doped.

In FIG. 1C, photoresist layer 112 is deposited as a continuous layer on oxide layer 106 and then selectively irradiated using a photolithographic system, such as a step and repeat optical projection system, in which I-line ultraviolet light from a mercury-vapor lamp is projected through a reticle and a focusing lens to project the desired image pattern on photoresist layer 112. Thereafter, photoresist layer 112 is developed and the irradiated portions are removed to provide an opening that defines the lateral boundaries for a trench. The minimum resolution (i.e. linewidth and spacing) of the step and repeat system is 4000 angstroms (0.4 microns). Therefore, to reduce feature sizes, the length of the opening is 4000 angstroms.

In FIG. 1D, an anisotropic etch such as a reactive ion etch is applied using photoresist layer 112 as an etch mask. The etch removes the exposed portion of oxide layer 106, thereby splitting oxide layer 106 into oxide layer regions 106A and 106B. The etch also forms trench 114 completely through doped layer 110, but only partially through substrate 102. Trench 114 has opposing vertical sidewalls 116A and 116B, and planar bottom surface 118. For illustration purposes, trench 114 has a depth (or height) of 2000 angstroms below top surface 104, and a length (between sidewalls 116A and 116B) of 4000 angstroms. The 4000 angstrom trench length represents the minimum resolution (or linewidth) of the photolithographic system. The trench splits doped layer 110 into source region 110A and drain region 110B, which will subsequently be to diffused beneath bottom surface 118 as described below.

In FIG. 1E, a channel implant is provided by boron, as indicated by lines 120, which passes through bottom surface 118 using photoresist layer 112 as an implant mask. The channel implant includes a well implant, then a punchthrough implant, and then a threshold voltage implant The well implant provides a more uniform background doping, the punchthrough implant provides the channel with greater robustness to punchthrough voltages, and the threshold voltage implant sets the threshold voltage of the IGFET to approximately 0.4 to 0.7 volts. The well implant is provided by boron at a dose in the range of 1.times.10.sup.12 to 1.times.10.sup.13 atoms/cm.sup.2 and an energy in the range of 100 to 170 kiloelectron-volts, the punchthrough implant is provided by boron at a dose in the range of 1.times.10.sup.12 to 1.times.10.sup.13 atoms/cm.sup.2 and an energy in the range of 40 to 100 kiloelectron-volts, and the threshold voltage implant is provided by boron at a dose in the range of 1.times.10.sup.12 to 1.times.10.sup.13 atoms/cm.sup.2 and an energy in the range of 2 to 30 kiloelectron-volts. The channel implant has a boron concentration on the order of 1.times.10.sup.17 atoms/cm.sup.3 and provides suitable doping for a channel adjacent to bottom surface 118.

In FIG. 1F, photoresist layer 112 is stripped, and a rapid thermal anneal on the order of 950 to 1050.degree. C. for 10 to 30 seconds is applied to remove crystalline damage and to activate and drive-in the implanted dopants. In source region 110A and drain region 110B, the phosphorus diffuses far more rapidly than the arsenic. Also, as mentioned above, source region 110A and drain region 110B are graded so that the deepest regions are the most lightly doped. As a result, lightly doped source region 122A and lightly doped drain region 122B diffuse from source region 110A and drain region 110B respectively, diffuse beneath trench 114, and are adjacent to the outer portions of bottom surface 118. Lightly doped source and drain regions 122A and 122B are doped N-, with a phosphorus concentration on the order of 5.times.10.sup.17 to 5.times.10.sup.18 atoms/cm.sup.3 since the diffusion is provided mainly by phosphorus (a rapid diffuser) without arsenic (a slow diffuser). Furthermore, source region 110A and drain region 110B remain heavily doped near top surface 104. Advantageously, lightly doped source and drain regions 122A and 122B do not diffuse any substantial amount beneath a central portion of bottom surface 118, between the outer portions, but instead form channel junctions 124A and 124B, respectively, that are substantially aligned with the central portion of bottom surface 118.

In FIG. 1G, a brief wet chemical etch is applied to remove any native oxide in trench 114, a blanket layer of a conductive metal that is 2500 angstroms thick is deposited over substrate 102, and then an anisotropic etch such as a reactive ion etch is applied to the conductive metal to form metal spacers 126A and 126B. Preferred conductive metals include aluminum, tungsten, titanium, cobalt, and combinations thereof The spacer etch is highly selective of the conductive metal with respect to silicon, so only a negligible amount of substrate 102 is removed. Metal spacers 126A and 126B contact and completely cover sidewalls 116A and 116B, respectively, and extend above top surface 104. Metal spacers 126A and 126B also contact the outer portions of bottom surface 118, adjacent to sidewalls 116A and 116B, but leave the central portion of bottom surface 118 exposed. Metal spacers 126A and 126B each have a length of 1200 angstroms (as measured laterally from sidewalls 116A and 116B) along the outer portions of bottom surface 118. Therefore, the central portion of bottom surface 118 has a length of 1600 angstroms (4000-2400 angstroms).

The dimensions of the metal spacers depend on several variables, including the height and length of the trench, the thickness of the blanket layer of conductive metal, and the duration of the spacer etch. The metal spacer dimensions depend on the desired scaling of the trench length to the channel length. However, the channel length depends on various factors including the characteristics of the source and drain regions and the high-temperature anneal. Preferably, the channel length is less than half of the trench length, and is less than 2000 angstroms. Furthermore, as described below, the channel shall be substantially aligned with the bottom surface of a gate electrode.

In FIG. 1H, a thin layer of silicon dioxide, 50 angstroms thick, is deposited over substrate 102 using a low-temperature deposition process that is well below the melting points of metal spacers 126A and 126B. As a result, gate oxide 130 is formed on the central portion of bottom surface 118, protective oxides 132A and 132B are formed on the sides of metal spacers 126A and 126B, respectively, opposite sidewalls 116A and 116B, respectively, and oxide segments 134A and 134B are formed on oxide layer regions 106A and 106B, respectively. Thus, gate oxide 130, protective oxides 132A and 132B, and oxide segments 134A and 134B form a continuous oxide layer over substrate 102.

In FIG. 1I, a blanket layer of doped polysilicon 136 with a thickness of 4000 angstroms is deposited by low pressure chemical vapor deposition over substrate 102. Polysilicon 136 contacts gate oxide 130, protective oxides 132A and 132B, and oxide segments 134A and 134B, and fills the remaining space in trench 114.

In FIG. 1J, the structure is planarized by applying chemical-mechanical polishing in the presence of an abrasive slurry. The polishing grinds down polysilicon 136, completely removes oxide segments 134A and 134B, removes portions of oxide layer regions 106A and 106B, metal spacers 126A and 126B, and protective oxides 132A and 132B above top surface 104, but is discontinued before reaching top surface 104. After polishing occurs, the remaining portion of polysilicon 136 forms polysilicon gate electrode 138. The top surfaces of metal spacers 126A and 126B, protective oxides 132A and 132B, oxide layer regions 106A and 106B and polysilicon gate electrode 138 are aligned and form a planar surface. Polysilicon gate electrode 138 extends merely 50 angstroms above top surface 104, and therefore is substantially aligned with top surface 104. Furthermore, substantially all of polysilicon gate electrode 138 is within trench 114, and no portion of polysilicon gate electrode 138 overlaps top surface 104. Of importance, polysilicon gate electrode 138 is spaced from and electrically isolated from metal spacers 126A and 126B and substrate 102.

Junction 124A is substantially aligned with a first side of polysilicon gate electrode 138 where it contacts gate oxide 130, and junction 124B is substantially aligned with a second side of polysilicon gate electrode 138 where it contacts gate oxide 130. Thus, junctions 124A and 124B define a channel, substantially aligned with the bottom surface of gate electrode 138, with a length of approximately 1600 angstroms, for an N-channel MOSFET controlled by polysilicon gate electrode 138 that includes a source provided by source regions 110A and 122A, and a drain provided by drain regions 110B and 122B.

Moreover, heavily doped source and drain regions 110A and 110B contact metal spacers 126A and 126B, respectively, at sidewalls 116A and 116B, respectively, of the trench 114, whereas lightly doped source and drain regions 122A and 122B contact metal spacers 126A and 126B, respectively, at bottom surface 118. Metal spacers 126A and 126B have far lower electrical resistance than lightly doped source and drain regions 122A and 122B. As a result, metal spacers 126A and 126B significantly lower the resistance between heavily doped source and drain regions 110A and 110B and channel junctions 124A and 124B, respectively, thereby reducing the parasitic resistance of lightly doped source and drain regions 122A and 122B, respectively.

FIGS. 2A-2K show cross-sectional views of successive process steps for forming an IGFET with a gate electrode and metal spacers in a trench in accordance with a second embodiment of the invention. It may desirable to implant the source and drain regions to a greater depth than the bottom surface of the trench, particularly if the implanted source and drain regions may not diffuse far enough to form channel junctions substantially aligned with the bottom surface of the gate electrode. The primary difference between the second embodiment and the first embodiment is that in the second embodiment, the doped layer is implanted deeper than the trench, disposable spacers are formed in the trench, the channel implant counterdopes the doped layer beneath the central portion of the bottom surface using the disposable spacers as an implant mask, and the doped layer diffuses only slightly. Unless otherwise noted, the elements for the second embodiment (e.g., substrate 202, gate oxide 230, etc.) are similar to elements of the first embodiment (e.g., substrate 102, gate oxide 230, etc.), and the description of related elements and process steps need not be repeated.

In FIG. 2A, semiconductor substrate 202 suitable for integrated circuit manufacture is provided. Substrate 202 includes a P- epitaxial surface layer with a planar top surface 204. Oxide layer 206 is formed on substrate 202.

In FIG. 2B, the structure is subjected to ion implantation of arsenic, indicated by arrows 240, at a first dosage in the range of 1.times.10.sup.15 to 5.times.10.sup.15 atoms/cm at a first energy of about 30 kiloelectron-volts, and a second dosage in the range of 1.times.10.sup.13 to 5.times.10.sup.14 atoms/cm.sup.2 at a second energy of about 60 kiloelectron-volts. This forms doped layer 242, consisting of N+ doped upper layer 244 with an arsenic concentration in the range of 1.times.10.sup.20 to 1.times.10.sup.21 atoms/cm.sup.3 and a depth of 1800 angstroms below top surface 204, and N- doped lower layer 246 with an arsenic concentration in the range of 5.times.10.sup.17 to 5.times.10.sup.18 atoms/cm.sup.3 and a depth of 500 angstroms below doped upper layer 244. Doped layer 242 differs from doped layer 110 in several respects. First, doped layer 242 is deeper than doped layer 110 by 500 angstroms. Second, doped layer 242 is doped solely by arsenic. Third, doped layer 242 includes a heavily doped upper layer above a lightly doped lower layer.

In FIG. 2C, photoresist layer 212 is deposited as a continuous layer on oxide layer 206 and then selectively irradiated using a step and repeat system. Thereafter, photoresist layer 212 is developed and the irradiated portions are removed to provide an opening that defines the lateral boundaries for a trench.

In FIG. 2D, an anisotropic etch is applied using photoresist layer 212 as an etch mask to form trench 214 completely through doped upper layer 244, partially through doped lower layer 246, and thus partially through substrate 202. Trench 214 includes opposing vertical sidewalls 216A and 216B, and planar bottom surface 218. Trench 214 splits doped upper layer 244 into heavily doped source region 244A and heavily doped drain region 244B. In addition, oxide layer 206 is split into oxide layer regions 206A and 206B.

In FIG. 2E, photoresist layer 212 is stripped, a blanket layer of silicon nitride is conformally deposited over substrate 202, and then an anisotropic etch such as a reactive ion etch is applied to the silicon nitride layer to form nitride spacers 248A and 248B on sidewalls 216A and 216B, respectively, and on outer portions of bottom surface 218. Nitride spacers 248A and 248B leave exposed a central portion of bottom surface 218.

In FIG. 2F, a channel implant is provided by subjecting the structure to ion implantation of boron, as indicated by arrows 220. The channel implant differs from the channel implant in the first embodiment in several respects. First, the channel implant does not pass through all of bottom surface 218, instead, nitride spacers 248A and 248B provide an implant mask for the outer portions of bottom surface 218, and the channel implant passes through only the central portion of bottom surface 218. Second, the channel implant counterdopes doped lower layer 246 beneath the central portion of bottom surface 218, effectively splitting doped lower layer 246 into lightly doped source region 246A and lightly doped drain region 246B. Third, the boron dosage is about twice the boron dosage used in the first embodiment to assure adequate counterdoping of doped lower layer 246 beneath the central portion of bottom surface 218.

In FIG. 2G, nitride spacers 248A and 248B are stripped, and a rapid thermal anneal on the order of 950 to 1050.degree. C. for 10 to 30 seconds is applied to remove crystalline damage and to activate and drive-in the implanted dopants. Since heavily doped and lightly doped source and drain regions 244A, 244B, 246A and 246B are doped solely by arsenic, very little diffusion occurs. As a result, lightly doped source and drain regions 246A and 246B form channel junctions 224A and 224B, respectively, that are substantially aligned with the central portion of bottom surface 218.

In FIG. 2H, a blanket layer of conductive metal is conformally deposited over substrate 202, and then an anisotropic etch is applied to form metal spacers 226A and 226B on sidewalls 216A and 216B, respectively, and on the outer portions of bottom surface 218. Metal spacers 226A and 226B leave exposed the central portion of bottom surface 218.

In FIG. 2I, an oxide layer is deposited over substrate 202 to form gate oxide 230 on the central portion of bottom surface 218, protective oxides 232A and 232B on metal spacers 226A and 226B, respectively, and oxide segments 234A and 234B on oxide layer regions 206A and 206B, respectively.

In FIG. 2J, a blanket layer of doped polysilicon 236 is deposited over substrate 202. Polysilicon 236 contacts gate oxide 230, protective oxides 232A and 232B, and oxide segments 234A and 234B and fills the remaining space in trench 214.

In FIG. 2K, the structure is planarized by applying chemical-mechanical polishing in the presence of an abrasive slurry. The polishing grinds down regions of polysilicon 236, but is discontinued before reaching top surface 204. The remaining portion of polysilicon 236 forms polysilicon gate electrode 238, which is aligned with metal spacers 226A and 226B, protective oxides 232A and 232B, and oxide layer regions 206A and 206B. Polysilicon gate electrode 238 is substantially aligned with top surface 204, and substantially all of polysilicon gate electrode 238 is within trench 214. Moreover, junctions 224A and 224B are substantially aligned with the bottom surface of polysilicon gate electrode 238.

FIGS. 3A-3L show cross-sectional views of successive process steps for forming an IGFET with a gate electrode and metal spacers in a trench in accordance with a third embodiment of the invention. It may desirable to implant lightly doped source and drain regions through the bottom surface of the trench, particularly if heavily doped source and drain regions may not diffuse far enough to form channel junctions substantially aligned with the bottom surface of the gate electrode. The primary difference between the third embodiment and the first embodiment is that in the third embodiment, a lightly doped layer is implanted through the bottom surface of the trench, disposable spacers are formed in the trench, the channel implant counterdopes the lightly doped layer beneath the central portion of the bottom surface using the disposable spacers as an implant mask and the source and drain regions diffuse only slightly. Unless otherwise noted, the elements for the third embodiment (e.g., substrate 302, gate oxide 330, etc.) are similar to elements of the first embodiment (e.g., substrate 102, gate oxide 130, etc.), and the description of related elements and process steps need not be repeated.

In FIG. 3A, semiconductor substrate 302 suitable for integrated circuit manufacture is provided. Substrate 302 includes a P- epitaxial surface layer with a planar top surface 304. Oxide layer 306 is formed on substrate 302.

In FIG. 3B, doped layer 310 is implanted into substrate 302 by subjecting the structure to ion implantation of arsenic, as indicated by arrows 308. Doped layer 310 differs from doped layer 110 in that doped layer 310 is doped solely by arsenic, instead of arsenic and phosphorus.

In FIG. 3C, photoresist layer 312 is deposited as a continuous layer on oxide layer 306 and then selectively irradiated using a step and repeat system. Thereafter, photoresist layer 312 is developed and the irradiated portions are removed to provide an opening that defines the lateral boundaries for a trench.

In FIG. 3D, an anisotropic etch is applied using photoresist layer 312 as an etch mask to form trench 314 completely through doped layer 310 and partially through substrate 302. Trench 314 includes opposing vertical sidewalls 316A and 316B, and planar bottom surface 318. Trench 314 splits doped layer 310 into heavily doped source and drain regions 310A and 310B. In addition, oxide layer 306 is split into oxide layer regions 306A and 306B.

In FIG. 3E, the structure is subjected to an ion implant of arsenic, indicated by arrows 350, at a dosage in the range of 1.times.10.sup.13 to 5.times.10.sup.14 atoms/cm.sup.2 and an energy in the range of 2 to 80 kiloelectron-volts using photoresist layer 312 as an implant mask. This forms N- doped layer 352 in substrate 302 beneath and adjacent to the bottom surface 318. Doped layer 352 is substantially aligned with sidewalls 316A and 316B.

In FIG. 3F, photoresist layer 312 is stripped, a blanket layer of silicon nitride is conformally deposited over substrate 302, and then an anisotropic etch provided by a reactive ion etch is applied to the silicon nitride layer to form nitride spacers 348A and 348B on sidewalls 316A and 316B, respectively, and on outer portions of bottom surface 318. Nitride spacers 348A and 348B leave exposed a central portion of bottom surface 318.

In FIG. 3G, a channel implant is provided by subjecting the structure to ion implantation of boron, as indicated by arrows 320. The channel implant differs from the channel implant in the first embodiment in several respects. First, the channel implant does not pass through all of bottom surface 318, instead, nitride spacers 348A and 348B provide an implant mask for the outer portions of bottom surface 318, and the channel implant passes through only the central portion of bottom surface 318. Second, the channel implant counterdopes doped layer 352 beneath the central portion of bottom surface 318, effectively splitting doped layer 352 into lightly doped source region 352A and lightly doped drain region 352B. Third, the boron dosage is about twice the boron dosage used in the first embodiment to assure adequate counterdoping of doped layer 352 beneath the central portion of bottom surface 318.

In FIG. 3H, nitride spacers 348A and 348B are stripped, and a rapid thermal anneal on the order of 950 to 1050.degree. C. for 10 to 30 seconds is applied to remove crystalline damage and to activate and drive-in the implanted dopants. Since heavily doped and lightly doped source and drain regions 310A, 310B, 352A and 352B are doped solely by arsenic, very little diffusion occurs. However, source regions 310A and 352A merge to form a source, and drain regions 310B and 352B merge to form a drain. Furthermore, lightly doped source and drain regions 352A and 352B form channel junctions 324A and 324B, respectively, that are substantially aligned with the central portion of bottom surface 318.

In FIG. 3I, a blanket layer of conductive metal is conformally deposited over substrate 302, and then an anisotropic etch is applied to form metal spacers 326A and 326B on sidewalls 316A and 316B, respectively, and on the outer portions of bottom surface 318. Metal spacers 326A and 326B leave exposed the central portion of bottom surface 318.

In FIG. 3J, an oxide layer is deposited over substrate 302 to form gate oxide 330 on the central portion of bottom surface 318, protective oxides 332A and 332B on metal spacers 326A and 326B, respectively, and oxide segments 334A and 334B on oxide layer regions 306A and 306B, respectively.

In FIG. 3K, a blanket layer of doped polysilicon 336 is deposited over substrate 302. Polysilicon 336 contacts gate oxide 330, protective oxides 332A and 332B, and oxide segments 334A and 334B and fills the remaining space in trench 314.

In FIG. 3L, the structure is planarized by applying chemical-mechanical polishing in the presence of an abrasive slurry. The polishing grinds down regions of polysilicon 326, but is discontinued before reaching top surface 304. The remaining portion of polysilicon 336 forms to polysilicon gate electrode 338, which is aligned with metal spacers 326A and 326B, protective oxides 332A and 332B, and oxide layer regions 306A and 306B. Polysilicon gate electrode 338 is substantially aligned with top surface 304, and substantially all of polysilicon gate electrode 338 is within trench 314. Moreover, junctions 324A and 324B are substantially aligned with the bottom surface of polysilicon gate electrode 338.

Further processing steps in the fabrication of IGFETs typically include forming a thick oxide layer over the active regions, forming contact windows in the oxide layer above the source, drain and gate electrode, forming appropriate interconnect metallization in the contact windows, and then forming a passivation layer. In addition, either earlier or subsequent high-temperature process steps can be used to supplement or replace the drive-in step to provide the desired anneal, activation, and diffusion for the implanted dopants, although it is highly desirable to avoid melting the metal spacers since this may lead to metal spiking. These further processing steps are conventional and need not be repeated herein. Likewise the principal processing steps disclosed herein may be combined with other steps readily apparent to those skilled in the art.

At completion of the process, it is preferred that the channel length be precisely controlled. However, to the extent that the lateral locations of the channel junctions with the source and the drain are not (or can not be) precisely controlled, it is far preferable to have a slight overlap between the junctions and the gate electrode (where the gate electrode contacts the gate insulator) as opposed to a lateral displacement or gap. While a slight overlap leads to capacitive effects (known as "Miller capacitance") that tend to reduce switching speeds, a lateral displacement or gap may prevent formation of a conductive channel between the source and the drain during operation. Furthermore, decreasing the channel length increases the drive current. Therefore, the overlap between the gate electrode and the source and drain involves a tradeoff between switching speed and drive current.

The present invention includes numerous variations to the embodiments described above. For instance, various features of the above embodiments can be combined with others. The channel implant may not be essential and can provide enhancement-mode doping if desired. The source and drain can be spaced from the sidewalls, as long as they are adjacent to the bottom surface of the trench. Likewise, the metal spacers need not necessarily extend to the top surface of the substrate, as long as the gate electrode remains electrically isolated from the metal spacers and the substrate. The gate electrode can be patterned to extend beyond the trench length over the substrate. The disposable spacers, if any, need not have similar shapes or sizes as the metal spacers. The metal spacers can include several metal layers. For instance, a thin barrier metal such as titanium nitride can be deposited on the substrate, a thick metal layer can be deposited on the barrier metal, and then the anisotropic etch can be applied to form the metal spacers. The gate insulator, or a lower portion thereof, can be grown on the bottom surface between the disposable spacers. In this manner, the gate insulator can include a grown layer beneath a deposited layer. The gate insulator and protective insulators can be various dielectrics including oxides, nitrides, and oxynitrides. The N-type and P-type dopants can be reversed. Suitable P-type dopants include boron B.sub.10, boron B.sub.11, and BF.sub.x species such as BF.sub.2.

Further details regarding trench transistors are disclosed in U.S. application Ser. No. 08/739,595 filed concurrently herewith, entitled "Trench Transistor With Insulative Spacers" by Gardner et al.; U.S. application Ser. No. 08/739,592 filed concurrently herewith, entitled "Trench Transistor With Localized Source/Drain Regions Implanted Through Voids In Trench" by Gardner et al.; U.S. application Ser. No. 08/739,596 filed concurrently herewith, entitled "Trench Transistor With Localized Source/Drain Regions Implanted Through Selectively Grown Oxide Layer" by Gardner et al.; U.S. application Ser. No. 08/739,597 filed concurrently herewith, entitled "Trench Transistor In Combination With Trench Array" by Fulford, Jr. et al.; U.S. application Ser. No. 08/739,566 filed concurrently herewith, entitled "Trench Transistor And Isolation Trench" by Gardner et al.; and U.S. application Ser. No. 08/739,567 filed concurrently herewith, entitled "Trench Transistor With Source Contact In Trench" by Gardner et al.; the disclosures of which are incorporated herein by reference.

The invention is particularly well-suited for fabricating N-channel MOSFETs, P-channel MOSFETs, and other types of IGFETs, particularly for high-performance microprocessors where high circuit density is essential. Although only a single IGFET has been shown for purposes of illustration, it is understood that in actual practice, many devices are fabricated on a single semiconductor wafer as widely practiced in the art. Accordingly, the invention is well-suited for use in an integrated circuit chip, as well as an electronic system including a microprocessor, a memory, and a system bus.

Those skilled in the art will readily implement the steps necessary to provide the structures and methods disclosed herein, and will understand that the process parameters, materials, and dimensions are given by way of example only and can be varied to achieve the desired structure as well as modifications which are within the scope of the invention. Variations and modifications of the embodiments disclosed herein may be made based on the description set forth herein, without departing from the scope and spirit of the invention as set forth in the following claims.