Patent Application
20120280202
The invention relates to heterojunction devices formed of metal oxides or nitrides and silicon and/or germanium and alloys thereof.
In a vertically oriented p-i-n diode formed in an etched pillar of silicon or germanium, it may be advantageous, for reasons of performance and fabrication, to minimize the height of semiconductor material; yet, to reduce reverse leakage current, it may also be advantageous to maximize the height of the intrinsic region.
Alternative ways to form sharp junctions in such devices, then, will be useful.
The present invention is defined by the following claims, and nothing in this section should be taken as a limitation on those claims. In general, the invention is directed to heterojunction devices including a layer of a metal oxide or nitride serving as a p-type or n-type region.
In a first aspect of this invention, a monolithic three dimensional memory array is provided that includes a first memory level formed above a substrate, and a second memory level monolithically formed above the first memory level. The first memory level includes a first plurality of substantially parallel, substantially coplanar conductors extending in a first direction, a second plurality of substantially parallel, substantially coplanar conductors extending in a second direction, the second direction different from the first direction, the second conductors above the first conductors, and a first plurality of devices. Each of the first plurality of devices is disposed between one of the first conductors and one of the second conductors, and includes a resistivity-switching binary metal oxide or nitride compound and a silicon, germanium, or silicon-germanium alloy resistor of a single conductivity type.
In a second aspect of this invention, a MOS device is provided that includes a semiconductor channel region including silicon, germanium, silicon-germanium, or a silicon-germanium alloy, a source region including a conductive binary metal oxide or nitride compound, and a drain region including the conductive binary metal oxide or nitride compound.
In a third aspect of this invention, a nonvolatile memory cell is provided that includes a resistivity-switching binary metal oxide or nitride compound, and a silicon, germanium, or silicon-germanium alloy resistor of a single conductivity type.
Each of the aspects and embodiments of the invention described herein can be used alone or in combination with one another.
The preferred aspects and embodiments will now be described with reference to the attached drawings.
A class of binary metal oxide or nitride compounds is known which can electrically behave as relatively wide-band gap semiconductors. A binary metal oxide or nitride compound is a compound including two elements, where the first is a metal and the second is oxygen or nitrogen. Some of these binary metal oxide or nitride compounds exhibit resistivity-switching behavior, meaning that these materials can be reversibly switched between two or more stable resistivity states. The states include at least a high-resistivity state and a low-resistivity state, wherein the difference in resistivity between the high-resistivity state and the low-resistivity state is at least a factor of three.
Preferred resistivity-switching binary metal oxide or nitride compounds include NixOy, NbxOy, TixOy, HfxOy, AlxOy, MgxOy, CoxOy, CrxOy, VxOy, ZnxOy, ZrxOy, BxNy, AlxNy, where x and y range between 0 and 1. Examples are the stoichiometric compounds NiO, Nb2O5, TiO2, HfO2, Al2O3, MgOx, CoO, CrO2, VO, ZnO, ZrO, BN, and AlN, but nonstoichiometric compounds may be preferred.
Use of these resistivity-switching binary metal oxide or nitride compounds in a nonvolatile memory array is described in U.S. patent application Ser. No. 11/395,995, “Nonvolatile Memory Cell Comprising a Diode and a Resistivity-Switching Material,” now U.S. Pat. No. 7,812,404 (the “'404 patent”), which is a continuation-in-part of U.S. patent application Ser. No. 11/125,939, “Rewriteable Memory Cell Comprising a Diode and a Resistance-Switching Material,” filed May 9, 2005 (the “'939 application”), both hereby incorporated by reference.
In a preferred embodiment of the '939 application, the nonvolatile memory cell shown in
Resistance-switching element 218 can be switched between at least two stable resistance states. When resistance-switching element 218 is in a high-resistance state, very little current flows through the memory cell when a read voltage is applied between top conductor 16 and bottom conductor 12. When resistance-switching element 218 is switched to a low-resistance state, significantly more current flows at the same applied read voltage.
The data state of the memory cell can be stored in the resistance state of the resistance-switching element. For example, a high-resistance state can correspond to a data ‘0’ while a low-resistance state corresponds to a data ‘1,’ or vice versa. The difference in read current allows the data states to be distinguished.
Diodes 30 provide electrical isolation between adjacent cells in such a memory level. A wire is ohmic, conducting with equal ease in both directions, and with current increasing linearly with voltage. In contrast, a diode is a non-ohmic device. A diode acts as a one-way valve, conducting current more readily in one direction than in the opposite direction. A diode has a turn-on voltage; below this turn-on voltage, little or no current flows. Once the turn-on voltage is reached, current flow increases rapidly.
In an array like that pictured in
If appropriate voltages are chosen for selected and unselected bitlines (top conductors 16) and wordlines (bottom conductors 12), the presence of diodes 30 make it possible to provide a high programming voltage to a selected memory cell in this array without inadvertently programming cells which share a wordline or bitline with the selected cell.
A low-resistivity state for the resistivity-switching binary metal oxides or nitrides will be described as a set state, and a high-resistivity state as a reset state. If a material is placed in more than two resistivity states, the highest resistivity state will be called the reset state, while the other states will be alternative set states.
A set pulse places a material into a lower resistivity set state, while a reset pulse places the material in a higher resistivity reset state. The terms set and reset voltage and set and reset current will be used as well. It has been found that, when in a set state, these switchable materials behave as wide-band-gap semiconductors. Some, such as NixOy, are p-type semiconductors, while others are n-type semiconductors.
In the present invention, a binary metal oxide or nitride compound which is a wide-band-gap semiconductor is formed adjacent to an intrinsic or doped conventional semiconductor material, such as silicon, germanium, or an alloy of silicon and/or germanium, forming a p-n heterojunction. This heterojunction can be used in a variety of devices.
For example,
As compared to the cell of
Diode 32 is arranged in series between bottom conductor 12 and top conductor 16. When nickel oxide is in the reset state, silicon regions 4 and 6 and nickel oxide layer 118 together behave as a high-resistance resistor.
The memory cell of
As described in U.S. patent application Ser. No. 11/148,530, “Nonvolatile Memory Cell Operating by Increasing Order in Polycrystalline Semiconductor Material,” filed Jun. 8, 2005 (the “'530 application) and hereby incorporated by reference, when deposited amorphous silicon is crystallized adjacent only to materials with which it has a high lattice mismatch (silicon dioxide and titanium nitride, for example), the resulting polycrystalline silicon (which will be described in this discussion as polysilicon) tends to include many defects in its crystalline structure, which cause this high-defect polysilicon to be relatively high-resistivity as formed.
A vertically oriented p-i-n diode, like diode 30 of
As described in U.S. patent application Ser. No. 11/015,824, “Nonvolatile Memory Cell Comprising a Reduced Height Vertical Diode,” now U.S. Pat. No. 7,285,464, and hereby incorporated by reference, it has been found that reducing the height of the diode reduces the programming voltage required to transform the high-defect polysilicon from the high-resistivity to the low-resistivity state.
In preferred embodiments, the diodes 30 of
It has been noted that a diode acts as a one-way valve, preferentially conducting in one direction. A p-i-n diode like diode 30 of
By omitting heavily doped p-type region 8 of
The memory cell of
The cell of
For example,
Other devices may be formed employing a p-n heterojunction between a metal oxide or nitride and doped or intrinsic silicon, germanium, or an alloy of silicon and/or germanium.
This device provides the advantage of a very sharp junction between the source/drain regions 210/212 and the channel region 214, and avoids the danger of unwanted diffusion of dopants from the source/drain regions 210/212 to the channel region 214. This device can be formed with channel region 214 in a monocrystalline semiconductor substrate, such as a silicon wafer or a silicon-on-insulator substrate, or can be formed in a polysilicon film as a thin film device. The MOS device could be formed as a memory cell, and multiple memory levels of such devices can be stacked to form a monolithic three dimensional memory array.
Each of the devices described is a heterojunction device comprising a p-n junction. On one side of the p-n junction is a semiconductor element having a first polarity, the semiconductor material comprising silicon, germanium, silicon-germanium, or an alloy of silicon and/or germanium, wherein the semiconductor element is lightly doped or intrinsic; and on the opposite side of the p-n junction is a binary metal oxide or nitride compound, the binary metal oxide or nitride compound having a second polarity opposite the first, and having a resistivity less than 1 megaOhm-cm, preferably less than about 1 kiloOhm-cm, more preferably less than about 1 microOhm-cm.
A detailed example will be provided describing fabrication of a first memory level of a monolithic three dimensional memory array comprising memory cells like the memory cell of
Fabrication details provided in the '939 and '530 applications earlier incorporated, in the '404 patent and in U.S. Pat. No. 6,952,030, “High-Density Three-Dimensional Memory Cell,” may prove to be helpful in fabrication of the memory level to be described. To avoid obscuring the invention, not all of this detail will be described, but it will be understood that no teaching of these applications and patents is intended to be excluded.
Turning to
An insulating layer 102 is formed over substrate 100. The insulating layer 102 can be silicon oxide, silicon nitride, high-dielectric film, Si—C—O—H film, or any other suitable insulating material.
The first conductors 200 are formed over the substrate 100 and insulator 102. An adhesion layer 104 may be included between the insulating layer 102 and the conducting layer 106. A preferred material for the adhesion layer 104 is titanium nitride, though other materials may be used, or this layer may be omitted. Adhesion layer 104 can be deposited by any conventional method, for example by sputtering.
The thickness of adhesion layer 104 can range from about 20 to about 500 angstroms, and is preferably between about 100 and about 400 angstroms, most preferably about 200 angstroms. Note that in this discussion, “thickness” will denote vertical thickness, measured in a direction perpendicular to substrate 100.
The next layer to be deposited is conducting layer 106. Conducting layer 106 can comprise any conducting material known in the art, such as doped semiconductor, metals such as tungsten, or conductive metal silicides; in a preferred embodiment, conducting layer 106 is tungsten.
Once all the layers that will form the conductor rails have been deposited, the layers will be patterned and etched using any suitable masking and etching process to form substantially parallel, substantially coplanar conductors 200, shown in
Next a dielectric material 108 is deposited over and between conductor rails 200. Dielectric material 108 can be any known electrically insulating material, such as silicon oxide, silicon nitride, or silicon oxynitride. In a preferred embodiment, silicon oxide is used as dielectric material 108. The silicon oxide can be deposited using any known process, such as chemical vapor deposition (“CVD”), or, for example, high-density plasma CVD (“HDPCVD”).
Finally, excess dielectric material 108 on top of conductor rails 200 is removed, exposing the tops of conductor rails 200 separated by dielectric material 108, and leaving a substantially planar surface 109. The resulting structure is shown in
For example, the etchback techniques described in U.S. patent application Ser. No. 10/883,417, “Nonselective Unpatterned Etchback to Expose Buried Patterned Features,” now U.S. Pat. No. 7,307,013 (the “'013 patent”), and hereby incorporated by reference in its entirety, can advantageously be used.
Alternatively, conductor rails can be formed by a damascene process, in which oxide is deposited, trenches are etched in the oxide, then the trenches are filled with conductive material to create the conductor rails.
Next, turning to
Semiconductor material that will be patterned into pillars is deposited. The semiconductor material can be, for example, silicon, germanium, or alloys of silicon and/or germanium. The present example will describe the use of silicon, though it will be understood that other materials may be used instead.
In this example, bottom heavily doped region 112 is heavily doped n-type silicon. In a most preferred embodiment, heavily doped region 112 is deposited and doped with an n-type dopant such as phosphorus by any conventional method, preferably by in situ doping. This layer is preferably between about 200 and about 800 angstroms.
Next intrinsic silicon region 114 is formed. In some embodiments a subsequent planarization step will remove some silicon, so an extra thickness is deposited. If the planarization step is performed using a conventional CMP method, about 800 angstroms of thickness may be lost (this is an average; the amount varies across the wafer. Depending on the slurry and methods used during CMP, the silicon loss may be more or less.) If the planarization step is performed by an etchback method, only about 400 angstroms of silicon or less may be removed.
Depending on the planarization method to be used and the desired final thickness, between about 800 and about 3800 angstroms of undoped silicon is deposited by any conventional method; preferably between about 1300 and about 2300 angstroms; most preferably between about 1600 and about 2000 angstroms. If desired, the silicon can be lightly doped with an n-type dopant.
Next, regions 114 and 112 are patterned and etched into pillars 300. Pillars 300 should have about the same pitch and about the same width as conductors 200 below, such that each pillar 300 is formed on top of a conductor 200. Some misalignment can be tolerated.
The photolithography techniques described in U.S. patent application Ser. No. 10/728,436, “Photomask Features with Interior Nonprinting Window Using Alternating Phase Shifting,” now U.S. Pat. No. 7,172,840; or U.S. patent application Ser. No. 10/815,312, Photomask Features with Chromeless Nonprinting Phase Shifting Window,” filed Apr. 1, 2004, both owned by the assignee of the present invention and hereby incorporated by reference, can advantageously be used to perform any photolithography step used in formation of a memory array according to the present invention.
A dielectric material 108, for example an HDP oxide such as silicon dioxide, is deposited over and between pillars 300, filling gaps between them. Next the dielectric material on top of the pillars 300 is removed, exposing the tops of pillars 300 separated by dielectric material 108, and leaving a substantially planar surface. This removal of dielectric overfill and planarization can be performed by any process known in the art, such as CMP or etchback. For example, the etchback techniques described in the '013 patent can be used. At this point each pillar 300, including silicon regions 112 and 114, is a resistor. The resulting structure is shown in
Turning to
Next, in preferred embodiments barrier layer 123 is deposited on nickel oxide layer 118. Layer 123 is preferably titanium nitride, though some other appropriate conductive barrier material may be used instead. An advantage of barrier layer 123 is that it allows an upcoming planarization step to be performed on barrier layer 123 rather than nickel oxide layer 118. In some embodiments, layer 123 may be omitted.
Layers 123 and 118 are patterned and etched to form short pillars 302, ideally directly on top of pillars 300 formed in the previous pattern and etch step. Some misalignment may occur, as shown in
If nickel oxide layer 118 is sputter etched, a thickness of overlying layer 123 will be removed. The thickness of layer 123 should be adjusted accordingly. Gaps between short pillars 302 are filled with dielectric material 108, then another planarization step, for example by CMP or etchback, removes dielectric overfill and exposes tops of pillars 300, which now include short pillars 302, as shown in
Turning to
The resulting structure, shown in
In operation, the resistance state of the switching element of each memory cell is switched by applying voltage or flowing current between one of the bottom conductors 400 and one of the top conductors 400 through the heterojunction diode of the memory cell. The cells are rewriteable memory cells. The array further comprises circuitry adapted to individually switch the resistivity-switching binary metal oxide or nitride compound of each memory cell between a stable low-resistivity state and a stable high-resistivity state.
Additional memory levels can be monolithically formed above this first memory level. In some embodiments, conductors can be shared between memory levels; i.e. top conductor 400 would serve as the bottom conductor of the next memory level. In this case a CMP step would remove dielectric overfill, exposing top conductors 400 at a substantially planar surface.
In other embodiments, an interlevel dielectric is formed above the first memory level of
A monolithic three dimensional memory array is one in which multiple memory levels are formed above a single substrate, such as a wafer, with no intervening substrates. The layers forming one memory level are deposited or grown directly over the layers of an existing level or levels. In contrast, stacked memories have been constructed by forming memory levels on separate substrates and adhering the memory levels atop each other, as in U.S. Pat. No. 5,915,167, “Three dimensional structure memory.” The substrates may be thinned or removed from the memory levels before bonding, but as the memory levels are initially formed over separate substrates, such memories are not true monolithic three dimensional memory arrays.
A monolithic three dimensional memory array formed above a substrate comprises at least a first memory level formed at a first height above the substrate and a second memory level formed at a second height different from the first height. Three, four, eight, or indeed any number of memory levels can be formed above the substrate in such a multilevel array.
Each memory level in the monolithic three dimensional memory array formed in the example provided is a first memory level of nonvolatile memory cells, each cell comprising: a heterojunction diode, each heterojunction diode comprising: a) a switching element comprising a binary metal oxide or nitride compound, wherein the switching element is switchable between a low-resistance state and a high-resistance state, and wherein, when in the low-resistance state, the switching element is a first terminal of the heterojunction diode having a first polarity; and b) a semiconductor element of silicon, germanium, silicon-germanium, or an alloy thereof, the semiconductor element comprising a second terminal of the heterojunction diode, the second terminal having a second polarity opposite the first, wherein the switching element contacts a lightly-doped or intrinsic region of the semiconductor element.
Many details of fabrication of this memory level, or of its structure, can be varied, and it is impractical to detail all possible variations. A few preferred alternatives will be discussed, however.
As described in the '939 application, nickel oxide layer 118 can be formed as part of the top conductors 400, or even as an unpatterned blanket layer between the top conductors 400 and the pillars 300. In general, nickel oxide is formed in a relatively high-resistivity state. If nickel oxide layer 118 is sufficiently high-resistivity, it will not provide an unwanted conductive path shorting adjacent memory cells.
When programming voltage is applied to switch the nickel oxide 118 to a low-resistivity state, resistivity switching only takes place in a localized switching region between each pillar 300 and top conductor 400, and only this small region becomes low-resistivity. The remainder of the layer remains in the original high-resistivity state.
In the example provided, nickel oxide layer 118 and its associated barrier layer is patterned and etched in a separate step from the pattern and etch step that forms the pillars comprising silicon regions 114 and 112. If desired, these could be patterned and etched using a single patterning step. For example, nickel oxide layer could be used as a hard mask during etch of the semiconductor pillar below.
In another alternative embodiment, it may be desirable to form the silicon resistor (regions 112 and 114) of low-defect polysilicon crystallized adjacent to a silicide which provides an advantageous crystallization template, such as titanium silicide. As described in the '530 application, and in U.S. patent application Ser. No. 10/954,510, “Memory Cell Comprising a Semiconductor Junction Diode Crystallized Adjacent to a Silicide,” now U.S. Pat. No. 7,176,064, and hereby incorporated by reference, when amorphous deposited silicon is crystallized adjacent to, for example, titanium silicide, the polysilicon is lower in defects and lower resistivity.
In contrast, when crystallized adjacent only to materials with which it has a high lattice mismatch, the resulting polysilicon is higher resistivity. Application of a relatively high-amplitude electrical pulse through the polysilicon changes the high-defect, high-resistivity polysilicon, leaving it lower resistivity. When formed adjacent to a silicide, the high-amplitude pulse is not required to reduce the resistivity of this high-resistivity polysilicon.
To crystallize polysilicon regions 112 and 114 adjacent to a layer of titanium silicide, regions 112 and 114 are deposited, patterned, and etched as described, gaps between them filled, and top surfaces exposed by planarization. Next, a thin layer of titanium and thin layer of titanium nitride are deposited. A low-temperature anneal reacts the titanium with the silicon of each pillar, forming a disk of titanium silicide at the top of each pillar. A wet etch removes the titanium nitride layer and strips away any unreacted titanium.
Next, a higher temperature anneal crystallizes silicon layers 114 and 112, which will be low-defect, low-resistivity polysilicon. The titanium silicide is then removed in a wet etch, and fabrication continues as in embodiment described above, with deposition of nickel oxide layer 118.
Embodiments of the present invention have been described in the context of memory cells, and of a monolithic three dimensional memory array. It will be understood, however, that the invention is limited neither to memory nor to monolithically stacked devices, and can be used to advantage in other contexts.
Detailed methods of fabrication have been described herein, but any other methods that form the same structures can be used while the results fall within the scope of the invention.
The foregoing detailed description has described only a few of the many forms that this invention can take. For this reason, this detailed description is intended by way of illustration, and not by way of limitation. It is only the following claims, including all equivalents, which are intended to define the scope of this invention.
This application is a division of U.S. patent application Ser. No. 13/007,812, filed Jan. 17, 2011, now U.S. Pat. No. 8,227,787, which is a division of U.S. patent application Ser. No. 11/395,419, filed Mar. 31, 2006, now U.S. Pat. No. 7,875,871, each of which is incorporated by reference herein in its entirety for all purposes.
| Number | Date | Country | |
|---|---|---|---|
| Parent | 13007812 | Jan 2011 | US |
| Child | 13553963 | US | |
| Parent | 11395419 | Mar 2006 | US |
| Child | 13007812 | US |
