Field of the Invention
This invention relates generally to the field of integrated circuits, and more specifically to an improved trench capacitor for high density dynamic random access memory integrated circuits.
Background of the Invention
Recently, integration of circuits on one chip has increased dramatically. In the area of dynamic random access memory (DRAM), memory capacity on one chip has moved beyond the 64 kilobit capacity through the 1 megabit and now into 4 megabits of RAM on one chip. In order to achieve a 4 megabit DRAM on one chip several major problems must be overcome.
Each memory cell in a DRAM at the basic bit level generally comprises one capacitor and one transistor, as shown in FIG. 1, although the actual circuit may vary greatly depending on the desired capacity, materials, etc. In building large capacity DRAMs, capacitors formed in trenches or "trench capacitors" are used in order to reduce the total surface area needed for one cell, thereby to pack the memory cells more densely. Trench capacitors, as known in the art, can be constructed by etching a cylindrical or other shape well or trench into a (usually silicon) wafer substrate, lining the trench with a dielectric layer and filling the remaining volume of the trench with a polysilicon plug. The trench wall and the plug serve as the two plates of the capacitor to store the electrical charge.
There are several tradeoffs required to obtain a high density of trench capacitors. Since the electrical charge of a trench capacitor is stored between the trench wall and the plug, if the trenches are too close together, there may be capacitive coupling between the trench wall of one trench with the trench wall of an adjacent trench. Furthermore, there may also be leakage of current from one trench wall to an adjacent trench wall through the silicon substrate, because a high voltage on one trench wall will tend to cause charge flow through the silicon substrate towards a low voltage on an adjacent trench wall. As a result, trench capacitors generally must be constructed approximately 1.8 microns apart or more.
These problems are well known in the art. In response to them, trench spacing has become a function of substrate doping; that is, the greater the doping of the substrate, the closer together the trenches may be. The high concentration of dopant provides an energy barrier between the trenches. However, if the substrate adjacent to the trench is highly doped, then the substrate under the gate transistor is also highly doped. A high performance pass transistor with low body effect, as is desirable in a high performance memory device, is then very difficult to construct, because the body effect of a transistor increases as the doping of the substrate increases. As a result, the full voltage of the bit line cannot be delivered to the capacitor through a high body effect transistor.
Finally, if one capacitor plate of polysilicon is formed next to a single crystal silicon substrate, a gated diode results, as is well known in the art. Such gated diode generally increases current leakage along and through the sidewall of the trench.
Therefore, it is a general object of this invention to overcome the above-listed problems.
It is a further object of this invention to provide a trench capacitor which requires a minimal amount of etching and fine alignment.
It is a further object of this invention to provide a trench capacitor which may be manufactured using standard processing techniques.
Summary of the Invention
This invention provides an apparatus and method for manufacturing a trench capacitor, preferably for use in a DRAM integrated circuit. The trench capacitor of the present invention comprises two capacitor plates separated by a dielectric formed within a trench formed in a substrate. A first plate may be formed on the wall of the trench and preferably comprises a field shield coupled to ground. The dielectric illustratively comprises silicon nitride. The second plate may illustratively comprise polysilicon. At least some of the layers extend out of the trench and include, outside the trench, a lateral portion. Preferably, several sacrificial layers are established over the trench capacitor, and the multi-layer structure thus formed is etched to form steps outside of the trench. A pass transistor is formed adjacent to the trench capacitor, and a contact layer is established over the steps to couple the trench capacitor to the pass transistor source. Bit lines and word lines are added.
It will be understood that an important aspect of the invention is the structure, fabrication, or use of a trench capacitor that reduces the effects of voltage being stored in the capacitor vis-a-vis the substrate. Much of the prior art has a doped region within the substrate next to the trench, which forms part of the trench capacitor. The present invention avoids this.
The structure of the present invention in some of its aspects therefore includes a DRAM memory cell having a trench capacitor in a trench in a substrate. A transistor is formed beside but now within the trench. The transistor selectively couples data to be stored in the cell to the capacitor. The trench capacitor has an active plate layer, a dielectric layer, and a field plate layer coupled to a reference potential, with the field plate layer also serving to isolate adjacent memory cells, and with the dielectric layer positioned between the plate layers. An insulation layer is located between the field plate layer and the substrate and further located between the field plate and side walls of the trench. The active plate layer is coupled to the transistor, with voltages on the active layer being insulated from the substrate by the substrate by the field plate layer and the insulating layer. The invention also includes the process of forming such a structure.
The invention also includes a method of operating a DRAM memory cell comprising the steps of: (1) actuating a pass transistor thereby to couple a data signal representing data to a trench capacitor from an electrode of the transistor; (2) impressing the signal on an active plate electrode of the trench capacitor; and (3) isolating the signal from the substrate and side walls of the trench by maintaining a further plate of the capacitor at a reference potential, said further plate being positioned between the active plate electrode and the side walls of the trench, and insulating said further plate from the substrate and the trench side walls by a layer of insulation, so that the signal coupled to the active plate has no effect on the substrate or trench side walls.
Brief Description of the Drawings
The invention, together with its objects and the advantages thereof, may best be understood b reference to the following detailed description taken in conjunction with the accompanying drawings of which:
FIG. 1 is a schematic diagram of a typical DRAM memory cell;
FIG. 2 is a plan view of a region of memory cells built according to the preferred embodiment of this invention;
FIG. 3 is a representational cross-section of the top of a trench at an intermediate step of construction of the preferred embodiment of this invention;
FIG. 4 is a representational cross-section of the trench at a step subsequent to FIG. 3, wherein certain layers have been etched, and additional layers have been established;
FIG. 5 is a representational cross-section of the trench at a step subsequent to FIG. 4 illustrating a multi-layered or stepped structure according to one aspect of this invention;
FIG. 6 is a representational cross-section of the trench at a step subsequent to FIG. 5 illustrating the position of insulating "sticks" as used in the invention;
FIG. 7 is a representational cross-section of parts of a trench capacitor memory cell according to this invention showing the construction of a pass transistor;
FIG. 8 is a representational cross-section of a completed trench capacitor memory cell constructed according to this invention; and
FIG. 9 is an exploded, three-dimensional perspective view of a trench capacitor memory cell constructed according to this invention.
Detailed Description of the Preferred Embodiment
The preferred embodiment of this invention will be described in connection with the simplistic memory cell model of FIG. 1. It is to be understood that the trench capacitor of this invention may be modified to suit the requirements of other memory circuits without departing from the scope of this invention. In FIG. 1, a memory cell is generally denoted by the number 10. Memory cell 10 generally comprises a capacitor 12 and a pass transistor 14. Capacitor 12 is configured to have a first plate coupled to ground and a second plate coupled to a drain of the pass transistor 14. A source and gate of the pass transistor 14 are coupled to the bit line and word line, respectively, as is known in the art. The transistor 14 may comprise an enhancement or depletion type FET or other switching device.
Turning now to FIG. 2, a plan view is shown of a portion of memory array constructed according to the preferred embodiment of this invention, showing four memory cells 10. The array of course contains millions of such cells, and FIG. 2 is merely illustrative. Each memory cell 10 in this array includes a capacitor 12 and a pass transistor 14, constructed according to the preferred embodiment of this invention. Also shown in FIG. 2 are each memory cell's associated bit lines or local interconnect 16 and word lines 18. A trench is shown at 20 with a cell capacitor definition shown at 22 over trench 20. The cell capacitor definition 22 is generally defined by mask at M-1, as will be described below. 24 generally denotes an opening in a mask M-2, as will also be described below. Pass transistor 14 is generally defined by the intersection of 24 and 18. The cell capacitor is defined by 22. Feature 26 is a conductive layer which connects the cell capacitor 12 to the source/drain of transistor 14. Details of this connection are described below.
Referring now to FIG. 3, an illustration of the preferred embodiment of this invention is shown after several steps of processing are substantially complete. The construction of the preferred embodiment of this invention begins with a substrate or wafer 30 formed preferably of single crystal silicon that is P doped, as known in the art. Other substrates can be used, and the P doping could be varied. A trench 20 is formed preferably by etching through the upper surface 32 of the substrate 30 into the substrate 30 using standard processing techniques, forming walls 34 and a floor (not shown in these figures), as known in the art. The dimensions of the trench preferably are 0.7 microns (nominally) by 2 microns by 3 microns deep. After trench 20 is etched, walls 34 may optionally be doped, but in the preferred method and structure, the walls are not doped. The trench 20 is then cleaned as is known in the art, in this embodiment using an oxide etch in buffered H.F.
A field shield will then be formed both in the trench 20 and on the upper surface 32 of substrate 30 in this embodiment. The field shield may be generally constructed according to the method described in U.S. Pat. No. 4,570,331 to S. Sheffield Eaton, Jr. et al., or variations thereof. Accordingly, a field shield implant is performed in this embodiment on the upper surface 32 of the substrate 30 prior to the etching of the trench 20, which adjusts the threshold voltage of both the active pass transistor 14 and the field shield isolation transistor. The field shield itself preferably comprises two layers. First, a field shield oxide layer 36 is established, preferably by being grown directly on the substrate in a 920.degree. Celsius wet O.sub.2 atmosphere to a thickness of approximately 62 nm. A field shield polysilicon layer 38, doped to greater than 10.sup.20 /cm.sup.3 with phosphorus is then deposited via means known in the art over field oxide layer 36 to a thickness of approximately 0.15 micron in this preferred embodiment.
The field shield will be common to all capacitors and will be coupled everywhere to ground in the preferred embodiment by being coupled to VSS somewhere. However, the field shield may be tied to any source of stable electrical potential on the memory circuit. The field shield provides isolation of the memory cell and prevents leakage of current through the single crystal silicon between adjacent memory cells. The field shield also acts as a first plate of the memory cell capacitor in this embodiment. By this method, a capacitor is formed within a trench, which acts as a mechanical structure instead of an electrical component. Having the plate adjacent to the substrate and held at a constant potential eliminates the gated-diode effect discussed above, thus enhancing electrical isolation of the trench capacitor. Furthermore, the substrate does not have to be doped as heavily as is common in the prior art and discussed above in connection with the background of the invention.
Next, a cell dielectric layer 40 is deposited or grown over the field shield layers. Dielectric 40 preferably comprises silicon nitride, preferably deposited via chemical vapor deposition to a thickness of approximately 0.018 microns. The cell dielectric 40 may vary in composition and/or thickness according to the desired capacitance of the capacitor device, as is known in the art.
Next, in the preferred method an oxidation step is performed which oxidizes the cell dielectric layer 40 to repair any gap in the nitride dielectric layer and to form a silicon oxide layer in order to reduce the conductivity of the stack.
Next, a second plate layer 42 is formed by means known in the art. In this embodiment, second plate layer 42 preferably comprises conductive doped polysilicon deposited by chemical vapor deposition to a thickness of approximately 0.15 microns, preferably in an ASM vertical furnace using disilane (Si.sub.2 H.sub.6) and phosphene (PH.sub.3) which produces a phosphorus doping of greater than 10.sup.20 /cm.sup.3 in the polysilicon. Second plate layer 42 may be connected to the source/drain of the pass transistor 14 which will be described below, in connection with FIGS. 7 and 8.
Next, a stop oxide layer 44 is formed, preferably by deposition. Stop oxide layer 44, as its designation implies, comprises a sacrificial layer that will be used as described below in connection with FIGS. 7 and 8. Stop oxide layer 44 in this embodiment may be an insulating material comprising silicon dioxide deposited to a thickness of approximately 60 nm.
The next layer of this embodiment preferably comprises a stop polysilicon layer 46. However, stop polysilicon layer 46 may not be necessary in other embodiments, depending upon the desired final structure. Stop polysilicon layer 46 comprises undoped polysilicon, deposited to a thickness of approximately 100 nm. This layer effectively fills the trench opening, as illustrated in FIG. 3.
Optionally, any remaining volume of trench may be filled with a plug. The plug may comprise silicon dioxide or polysilicon. The plug would be etched until its upper surface is approximately level with the top of the stop polysilicon layer 46 which includes an exposed lateral region.
Thus it will be seen that in the preferred embodiment as thus far described, a trench is created and filled with layers alternately of dielectric and polysilicon. These layers extend from inside the trench to the surrounding lateral surface regions.
Turning now to FIG. 4, a trench capacitor constructed according to the preferred embodiment of the present invention is shown after further processing. It will be seen from FIG. 5 that the alternating layers will be defined at two locations, M-1 and M-2, on the "lateral" portions of the layers rather than in the trench itself.
A first step 48 (FIG. 4) which corresponds to 22 (FIG. 2) is formed adjacent to the trench. In the preferred method, a mask is formed of photoresist (not shown) as is known in the art over the structure of FIG. 3, wherein M-1 in FIG. 4 denotes the edge of the mask. Stop polysilicon layer 46 is etched using an anisotropic dry etch well known in the art that will stop on oxide 44. Stop oxide layer 44 next is etched using an anisotropic dry etch that will stop on polysilicon, which in this embodiment comprises second plate polysilicon layer 42. Polysilicon layer 42 is then etched via means essentially similar to that used to etch stop polysilicon layer 46 such that the etch will stop on the silicon nitride of cell dielectric layer 40, thus forming a first "step" 48. The photoresist used to form the mask is then stripped.
An oxide layer 50 may then be grown or, as in the preferred embodiment, deposited by means known in the art to a thickness of approximately 0.2 microns. The structure thus created may then be subjected to densification by standard techniques, as described above. The resulting structure is shown in FIG. 4.
Turning now to FIG. 5, the structure created in FIG. 4 is further masked and etched. First, oxide layer 50 is masked with photoresist so that trench 20 is covered with photoresist that extends beyond the trench 20 to line M-2, which denotes the edge of the mask. Layer 24 (FIG. 2) corresponds to the opening in photoresist mask M2. It will be noted that M-2 is closer to the trench than M-1. The distance from M-2 to M-1 is approximately 0.7 microns in the preferred embodiment. Next an etch of the exposed portions of oxide layer 50 is performed, stopping on the stop polysilicon layer 46 in a first region 52 defined by lines M-1 and M-2. This etch also etches in a second region 54 (adjacent to region 52) through oxide 50 and then through the cell dielectric 40, stopping on the top surface of field shield polysilicon layer 38. After the photoresist is stripped, the result is shown in FIG. 5. Thus, it will be seen that a second step (at M-2) has now been formed between the first step (at M-1) and the side of trench 20.
The next operation is preferably an anisotropic plasma polysilicon etch, which etches two different polysilicon member simultaneously in the preferred embodiment. One polysilicon member that is etched is the stop polysilicon layer 46 exposed between lines M-1 and M-2. The other is the exposed field shield polysilicon layer 38 to the right of line M-1. At this time, field shield oxide 36 still covers the top surface 32 of silicon substrate 30. Thus, the etch stops on oxides 44 and 36.
A further oxide layer (not shown) may be deposited by means known in the art over the entire structure thus formed to a thickness of 0.20 microns, illustratively. The further oxide layer and the oxides 44 and 36 exposed in the prior step may then be anisotropically etched using a low-silicon-damage etch, so that only a pair of spacer sticks 56 and 58 (FIG. 6) are left at the steps created at M-1 and M-2. Thus, each oxide spacer stick 56, 58 has a substantially vertical sidewall nearest the trench, i.e., at M-1 and M-2, respectively. The structure thus created may again be subjected to densification depending on the type of oxide deposition used. In the preferred embodiment, densification is not needed because the oxide used is deposited at a high temperature on the order of 800.degree. C. The resulting structure is shown in FIG. 6.
It will be seen that oxide layer 36 extends to the outer edge of spacer stick 58. The body of stick meets and insulates the outer edges of first plate 38, dielectric 40, and second plate 42, all at the "first step" M-1. Similarly, oxide 44 extends to the outer edge of spacer stick 56, the body of which meets and insulates the edges of poly 46 and oxide 50 at second step M-2. Between the first and second steps, a lateral expanse of second plate 42 is exposed; the remaining structure is covered by oxide, except for part of the substrate 30.
Turning now to FIG. 7, a gate oxide layer 60 is grown or otherwise established on exposed silicon to a thickness of 20 nm. Next, a gate polysilicon layer 62 is deposited over the entire surface to a thickness of approximately 0.2 microns. Next, a gate poly oxide layer 64 is deposited to a thickness of approximately 0.2 microns over gate polysilicon layer 62. This structure is then densified at 920.degree. C. for 10 minutes in a dry O.sub.2 atmosphere in the preferred embodiment.
The structure thus created is masked with a photoresist and etched. First, gate poly oxide layer 64 is etched, stopping on the polysilicon layer 62, the photoresist is then stripped and the gate polysilicon layer 62 is etched, stopping on gate oxide 60 and field oxide 50.
Next, a 50 nm first spacer oxide layer (not shown) is deposited over the surface of the structure. A lightly doped drain (LDD) region 66 is then defined via masking. The LDD region 66 in the preferred embodiment of this invention is an N+ doped region which results from implanting phosphorous at a dose of approximately 1.times.10.sup.14 per cm.sup.2 and an energy level of 60 keV through the first spacer oxide layer. A second spacer oxide layer (not shown) is then deposited to a thickness of approximately 0.1 microns. Both spacer oxide layers are then etched anisotropically leaving sticks 68 and 70 as shown in FIG. 7.
Thus, at this point in the process, a gate electrode has been created with an insulated side wall (facing the steps at M-1 and M-2). Between the gate electrode and the steps is an exposed upper surface of the substrate 30, in which an LDD implant has been done.
Source-drain regions for both n-channel and p-channel transistors are then formed on other regions of the wafer using methods which are known in the art.
Turning now to FIG. 8, a 20 nm titanium layer is next deposited over the entire structure of FIG. 7. A titanium nitride layer 72 may then be formed over the non-silicon regions while a titanium silicide layer 74 is formed under the titanium nitride where the titanium contacts polysilicon 42 or substrate 30. A 50 nm titanium nitride layer is then sputtered onto the structure. A masking silicon nitride (Si.sub.3 N.sub.4) is then deposited by chemical vapor deposition, to a thickness of 50 nm. The structure is masked with photoresist covering area 26 (FIG. 2) corresponding to the extend of 72 (FIG. 8). Next, the exposed areas of the silicon nitride are etched with a dry, isotropic plasma etch, stopping on the underlying titanium nitride layer. The photoresist layer is stripped and the silicon nitride is used as a mask while etching the exposed portions of the titanium nitride layer, stopping on the titanium silicide layer 74.
By forming a memory cell according to the preferred embodiment of this invention, sticks 56, 58, 68 and 70 prevent the contact/barrier layer or region of titanium nitride (which is conductive) from contacting and hence electrically connecting to certain other conductive layers--specifically the field shield layer which is tied to ground and the gate polysilicon layer 62 which is the word line 18. However, the titanium silicide and titanium nitride operate to couple the second capacitor plate electrode 42 (which was exposed between the two steps) to the source/drain region 66. This corresponds to FIG. 1 where the upper capacitor plate is coupled to the source/drain of FET 14.
Consequently, the contact/barrier layer may be deposited without etching contact windows. By using sacrificial layers formed into steps in combination with insulating sticks, a self-aligned contact layer has been established.
Next, nitride can be deposited to a depth of approximately 0.03 microns. A BPSG layer 76 (FIG. 9) may then be deposited to a thickness of approximately 0.6 microns over the entire structure, as is commonly done in the art. Bit lines 16, as are known in the art, are then formed.
Turning now to FIG. 9, a memory cell 10 built according to the preferred embodiment of this invention is shown in an exploded, three dimensional view. The trench 20 and lower layers are generally shown on the left and the pass transistor 14 and upper layers are shown on the right. It is to be understood that the various layers on the right fit over top of the layers on the left.
It will be seen that in FIG. 9 the side walls of the trench are not shown as precisely parallel but rather at a slight angle. It will be understood therefrom that in the practice of this invention, an opposing pair of side walls of the trench can be perpendicular to the substrate upper surface as depicted in FIGS. 3 to 8, and thus exactly parallel to each other, or they may depart from being exactly parallel to one another. It will also be understood that the trench may contemplate a pair of rounded walls, which also are illustrated in FIG. 9.
Although the present invention has been described herein terms of the preferred embodiment, it is envisioned that the scope and spirit of the present invention encompasses such changes and minor alterations as would normally be apparent to one skilled in the art and familiar with teachings of this specification.