The present invention relates to the field of nonvolatile memory devices using a single transistor as the only element in the memory cell and the formation of this transistor with a rare earth manganite thin film ferroelectric material as the gate dielectric.
Nonvolatile ferroelectric random access memory (FeRAM) devices represent an emerging, multibillion-dollar market. The most advanced FeRAMs utilize a 1-transistor/1 capacitor (1T1C) cell technology and a destructive read out (DRO) scheme. These devices compete with EEPROMs, battery backed static RAMs (BBSRAMs), and Flash nonvolatile memory devices.
FeRAM is a type of semiconductor memory, constructed similarly to a Dynamic Random Access Memory (DRAM), but which stores bits of data without the need for power (nonvolatile characteristic). FeRAMs have gained recent interest because of the possibility that they could become the ideal memory of the future, replacing standard mass-produced DRAM. Although the basic fundamentals of ferroelectricity were discovered in the 1920's, developments within the last fifteen years regarding the use of thin ferroelectric films may now make it practical to develop a dense memory with ideal nonvolatile memory performance characteristics. Ferroelectric materials exhibit ferroelectric behavior below a critical temperature, known as the Curie temperature. The Curie temperatures of many ferroelectric materials are above 200° C. allowing them to be used as storage elements for nonvolatile semiconductor memories.
Prior art FeRAMs operate using an array of memory cells, which contain capacitors, built of a special dielectric material (a ferroelectric) sandwiched between two conducting material (electrode) layers. The special ferroelectric material is comprised of a lattice of ions, in which one of the ions in each unit cell has two stable states on either side of the center of the unit cell along an elongated axis as shown in
c shows a schematic of a 1T1C memory cell. When a positive electric field is applied across a negatively polarized ferroelectric capacitor, the center ions of the unit cells switch to positively polarized states. This ion movement can be sensed as a current flow between the electrodes of the ferroelectric capacitor. When this electric field is removed, the polarization settles to the state labeled “+Q0” (See
The prior art for processing FeRAMs requires the fabrication of ferroelectric capacitors after all of the underlying CMOS circuitry has been fabricated just prior to metalization. A typical cross section of a 1T1C cell has a ferroelectric capacitor placed on the field oxide, and is connected to the transistor with a local interconnect. This creates a processing concern, because ferroelectric materials must be activated after deposition at high temperatures. When the underlying CMOS circuitry is heated to high temperatures, hydrogen is released, which degrades the ferroelectric film. Depositing the metalization interconnect layers can also produce hydrogen. Thus, a hydrogen barrier must be added to protect the ferroelectric capacitors. Also, some ferroelectric materials are very sensitive to moisture, which can be formed when hydrogen is released. Finally, when dense memories are fabricated, planarization techniques are commonly used. Most processes require interconnect metalization to be added before the ferroelectric materials are deposited and activated. This interconnect metalization cannot withstand the high temperatures of ferroelectric film activation. All of these problems have slowed the development of dense FeRAMs and have clouded the future for an ideal memory.
The overriding disadvantage of prior art 1T1C cell memories is that under the best of circumstances they cannot be scaled aggressively enough to compete with Flash memory since they require 2 elements per cell compared to a single element in flash. With comparable design rules a prior art FeRAM has a cell size at lest twice the cell size of flash.
A cross section and a diagram of the proposed 1T-cell, the subject of this invention, are shown in
In the 1T-cell, shown in
A read operation does not change the polarization state of the device and, therefore, does not destroy the information (“non destructive read-out” or NDRO). The device does, therefore, not fatigue under a read operation (it only fatigues under a write operation), a major advantage over DRO memories. In addition to a fatigue-free read operation, a NDRO scheme is faster since no restore operation is needed, shortening the write cycle. NDRO also is potentially more reliable, especially in adverse environments (e.g., radiation) as encountered in military and space applications, as there is never an instance where the memory state is unidentified where an upsetting event (e.g., single event upset or SEU) could inadvertently alter the state of the memory.
The cell shown in
The challenges of building such a 1T ferroelectric memory cell is that the ferroelectric material has to be deposited directly on silicon instead of a metallic bottom electrode as in the prior art devices. This means that requirements for the ferroelectric material are quite different. The switched polarization (Pr) should be in a range of 0.1 to 1 μC/cm2 and not as high as possible as in the prior art device. The dielectric permittivity should be as low as possible, ideally less than 20, and the leakage current over the operating voltage range (e.g. +−5 Volt) should be as low as possible. This applies both to electronic and ionic charge motion in the ferroelectric material, but most importantly for the slow moving ionic charge. In order for the information stored in a 1T cell to be preserved, the polarization charge cannot be compensated. Any ionic charge that would slowly drift through the films and accumulate at the ferroelectric interfaces to the electrodes could compensate the polarization charge over time and destroy the information.
The present invention is a memory device with a memory cell composed of a single transistor. This transistor is formed with a rare earth manganite as the thin film ferroelectric material that forms the gate dielectric of the ferroelectric transistor. This ferroelectric gate material is deposited directly on silicon; instead of a metallic or conductive oxide bottom electrode as in prior art ferroelectric devices. By its nature, the ferroelectric material has to be inserted early in the CMOS process. It has, therefore, to be able to withstand higher temperatures, most likely up to 950° C., (e.g. the activation temperature for the implanted source and drain junctions).
The requirements for the ferroelectric gate material are quite different from the ones in a 1T1C cell. In addition to the high temperature requirements and the low mobile ionic charge, it has to have a low dielectric constant, preferably less than 30. The reason for this is that any ferroelectric material deposited on silicon forms an interfacial layer with a rather low dielectric constant, most likely around 3.9, the value for SiO2. A voltage applied to the gate of the transistor is, therefore, divided between the ferroelectric material and the interfacial layer. If the capacitance of the interfacial layer is much smaller than the capacitance of the ferroelectric layer, most of the voltage will drop over the interfacial layer and is not available for switching the ferroelectric material. It is, therefore, not possible to operate the device at low voltages. The ferroelectric material of the present invention has a low relative dielectric permittivity around 10 and forms an interfacial layer with a relative dielectric permittivity larger than that of SiO2, which makes it ideal for a 1T cell.
The ferroelectric material of the present invention is quite different from the materials such as lead zirconate titanate (PZT) and strontium bismuth tantalate (SBT), which are the preferred materials in prior art devices. The device in a preferred embodiment of the current invention includes a substrate such as silicon, a thin film of a rare earth manganite as the ferroelectric material deposited directly onto this substrate and an electrode formed on top of the ferroelectric material. Source and drains are formed either before or after the ferroelectric layer deposition. The device is provided with contacts to the source and drain regions and to the gate electrode. Standard metalization (e.g. aluminum) is used to connect these contacts to other elements in an integrated circuit such as a CMOS circuit and to the outside world.
The ferroelectric material of the present invention is deposited directly on silicon and is composed of any rare-earth manganite with the formula A3+MnOx (where A=Y, Ce, Pr, Nd, Pm, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, Yb, or Lu) and x is between 1 and 3 inclusive.
The rare earth manganites used in the present invention have the desired properties for a successful operation of a 1T memory cell. They have dielectric permittivities of less than 20, they form a stable interfacial layer with a relatively high dielectric constant of ˜30, they have a high transition temperature (Curie temperature) of >600° C. and they have inherently low mobile charges (e.g. oxygen vacancies). The ferroelectric polarization is in the ideal range of 0.1 to 2 μC/cm2.
The ferroelectric materials of the present invention are deposited preferably by metallorganic chemical vapor deposition techniques (MOCVD), although other methods such as PECVD or MOD could be used. The MOCVD process preferably uses a liquid source delivery and flash evaporation method, since the precursors preferably used typically have low vapor pressure, making other methods such as vapor delivery with bubblers difficult. Examples of precursors used for CeMnO3, a special embodiment of the current invention, are Cerium (IV) tris-tetramethylheptanedionate (thd) for Cerium (Ce) and Manganese (III) tris-tetramethylheptanedionate for Manganese (Mn). These materials are solid powders. They can be brought into solution with tetrahydrofuron (THF) as the solvent with tetraglyme added to prevent any early evaporation problems.
The MOCVD process produces rare earth manganite thin films with excellent uniformity and compositional control and good step coverage. Most importantly, it produces a well controlled interfacial layer of AxOy that is enhanced by the subsequent anneal steps in O2.
a (PRIOR ART) shows the crystal lattice of a ferroelectric material.
b (PRIOR ART) is a hysteresis loop.
c (PRIOR ART) is a schematic diagram of a prior art 1T1C memory cell.
a shows a cross section of an embodiment of the present invention.
b shows a schematic diagram of an embodiment of the present invention.
c shows a hysteresis loop for an embodiment of the present invention.
d shows the signals on the PL and WL line during the and reading of ‘0s’ and ‘1s’ for an embodiment of the present invention.
a shows a graph of an RBS spectrum for a ferroelectric thin film of a rare earth manganite deposited on a silicon substrate.
b shows the voltage drops over the ferroelectric thin film and the interfacial layer, shown in schematic form.
A description of a 1 transistor (1T) memory cell 10 in accordance with the present invention is shown in
The ferroelectric layer 30 may be patterned using standard photolithography and etch techniques. Source and drain regions 40 and 41 are formed adjacent to the patterned ferroelectric layer using standard techniques used in semiconductor processing. These regions 40 and 41 can either be formed before or after deposition of the ferroelectric layer 30. The most common scheme for source and drain formation is by ion implantation of, for example boron for p+ regions (p- channel transistors), or phosphorous or arsenic for n+ regions (n-channel transistors). Either type is part of the invention. On top of the ferroelectric layer, a conductive gate material 50 is deposited using standard techniques (e.g. CVD, sputtering), completing the basic ferroelectric transistor. They conductive gate material may be composed of platinum, gold or any other noble metal, iridium, rhodium, ruthenium or oxides thereof, doped polycrystalline silicon, or a metal silicide such as platinum silicide. The doping of polycrystalline silicon can occur during deposition of the polycrystalline silicon layer (in situ) or can be performed as a separate processing step after the polycrystalline silicon is deposited.
For a commercially useful transistor as past of an integrated circuit (e.g. CMOS), additional processing steps are required. An insulating layer 60 (e.g. SiO2) is deposited on the gate material, contact windows 70, 71 are etched to source and drain and also to the gate electrode, respectively, and connections 80 are provided by standard metalization (e.g. aluminum). Finally, the device is properly passivated with a suitable passivation 90, e.g. polyimide, plasma oxide or plasma nitride.
The ferroelectric thin film 30 is a rare earth manganite with the general formula A3+MnOx (where A=Y, Ce, Pr, Nd, Pm, Sm, Eu, Gd, Th, Dy, Ho, Er, Tm, Yb, or Lu) and x is between 1 and 3 inclusive.
Most rare earth magnates are uniaxial ferroelectrics in hexagonal crystal formation, with some in a cubic crystal lattice. These rare earth manganites possess properties that are ideal for 1 T cell type memory cells, the topic of the present invention. For example, their transition temperatures (Curie temperatures) are high relative to the maximum operating temperature of a circuit, in the range of 500 to 700° C. This property allows insertion of the ferroelectric film into a manufacturing process early in the process flow, for example prior to source/drain formation. The material can withstand the processing temperatures of >850° C. that is required to, for example, activate the junctions.
Rare earth manganites have low dielectric permittivties, in the range from 10 to 50, due to the closely packed crystal lattice in a hexagonal structure. The interfacial layer 31, on the other hand, has relatively high dielectric permittivities, in the range of 10 to 30, because it is a mixture of the rare earth oxide AxOz and the silicon oxide SivOw with x,y,v,w in the range of 1 to 3. For example Ce2O3, Y2O3 and Pr2O3, preferred embodiments of the present invention, have relative dielectric permittivties of about 25, 20 and 30, respectively. The interfacial layer 31 either forms automatically during CVD deposition of the rare earth manganite, enhanced by the subsequent anneals in O2, or is deposited deliberately, preferably by MOCVD in the same processing chamber and prior to the ferroelectric deposition. Using liquid delivery methods of the precursors allows easy deposition of this interfacial oxide. It can also be deposited by other means, for example by sputtering, e-beam evaporation or others.
The ferroelectric layer is preferably deposited by MOCVD, although other techniques such as MOD, sol-gel, PECVD and others can be used and are subject to this invention. The preferred embodiment of depositing the ferroelectric layer by MOCVD is by using a liquid source delivery method as described in U.S. Pat. No. 5,887,117, “Flash evaporator,” the contents of which are incorporated by reference. An important element of this method is using at suitable precursor chemistry. A preferred embodiment is using Cerium (IV) tris-tetramethylheptanedionate (thd) for Cerium (Ce), Manganese (III) tris-tetramethylheptanedionate for Manganese (Mn) and tetrahydrofuron (THF) as the solvent.
Table I lists the preferred deposition conditions for depositing CeMnO3.
After deposition, the ferroelectric films have to be annealed in O2 at temperatures around 800 to 950° C. in order to form the proper ferroelectric phase. This anneal also slightly increases the thickness of the interfacial layer, an added advantage, since it improves the quality and reliability of this important oxide layer. The relatively high anneal temperature, which would be detrimental in prior art ferroelectric memory devices, is not a problem since the ferroelectric films are inserted into the process flow early, for example prior to source and drain formation.
The combination of a low relative dielectric permittivity ferroelectric material on top of a high relative dielectric permittivity oxide interfacial layer is an important element of the present invention and is explained in more detail with reference to
Vg=Vf+Vo.
The voltage Vf is given by the equations below.
∈fVfto=∈oVotf
Vf=(∈o/∈f)(tf/to)Vo=[1/(1+(∈f/∈o)(to/tf))]Vg
where tf is the thickness of the ferroelectric layer and to the thickness of the oxide layer. The effective voltage Vg applied to the ferroelectric film is reduced by the ratio of dielectric permittivities of the interfacial layer and the ferroelectric films. In order to maximize the available voltage applied to the ferroelectric film, the voltage available for switching, the ratios (∈f/∈o) and (to/tf) should be as low as possible. The rare earth manganites, the subject of the present inventions, satisfy this requirement.
The gate electrode 50 is formed of a conducting material consistent with CMOS processing. In prior art FeRAMs noble metals such as Pt are typically used in order to prevent unwanted oxides to form at the interfaces. In the present invention, oxide interfacial layers are not critical due to the low dielectric permittivity of the rare earth manganites. A much wider range of gate electrode materials can be used including noble metals, conductive oxide and, most importantly, doped polycrystalline silicon or metal suicides. These latter materials are particularly attractive since they are inexpensive and compatible with current CMOS processing.
The ferroelectric transistor of the present invention can be integrated into a standard CMOS manufacturing process as shown in
It should be clear that a number of aspects of the present invention have been disclosed, and any of them may have independent value and significance apart form the teachings as a whole. Further, it should be clear the present invention relates to a device and method of making and using such a device.
This applications is a 371 of PCT/US02/13095 filed Aug. 24, 2001, which claims benefit of provisional 60/227,858 filed Aug. 24, 2000 And claims benefit of provisional 60/272,986 filed Mar. 2, 2001.
Filing Document | Filing Date | Country | Kind | 371c Date |
---|---|---|---|---|
PCT/US02/13095 | 8/24/2001 | WO | 00 | 2/21/2003 |
Publishing Document | Publishing Date | Country | Kind |
---|---|---|---|
WO02/082510 | 10/17/2002 | WO | A |
Number | Name | Date | Kind |
---|---|---|---|
3832700 | Wu et al. | Aug 1974 | A |
4860254 | Pott et al. | Aug 1989 | A |
5046043 | Miller et al. | Sep 1991 | A |
5060191 | Nagasaki et al. | Oct 1991 | A |
5070385 | Evans, Jr. et al. | Dec 1991 | A |
5146299 | Lampe et al. | Sep 1992 | A |
5198994 | Natori | Mar 1993 | A |
5227855 | Momose | Jul 1993 | A |
5302842 | Chan | Apr 1994 | A |
5307305 | Takasu | Apr 1994 | A |
5345414 | Nakamura | Sep 1994 | A |
5365094 | Takasu | Nov 1994 | A |
5378905 | Nakamura | Jan 1995 | A |
5384729 | Sameshima | Jan 1995 | A |
5418389 | Watanabe | May 1995 | A |
5434811 | Evans, Jr. et al. | Jul 1995 | A |
5479317 | Ramesh | Dec 1995 | A |
5515311 | Mihara | May 1996 | A |
5519235 | Ramesh | May 1996 | A |
5523964 | McMillan et al. | Jun 1996 | A |
5536672 | Miller et al. | Jul 1996 | A |
5541870 | Mihara et al. | Jul 1996 | A |
5541871 | Nishimura et al. | Jul 1996 | A |
5541873 | Nishimura et al. | Jul 1996 | A |
5559733 | McMillan et al. | Sep 1996 | A |
5563081 | Ozawa | Oct 1996 | A |
5578846 | Evans, Jr. et al. | Nov 1996 | A |
5621681 | Moon | Apr 1997 | A |
5623439 | Gotoh et al. | Apr 1997 | A |
5736759 | Haushalter | Apr 1998 | A |
5739563 | Kawakubo et al. | Apr 1998 | A |
5744374 | Moon | Apr 1998 | A |
5757042 | Evans, Jr. et al. | May 1998 | A |
5768185 | Nakamura et al. | Jun 1998 | A |
5789775 | Evans, Jr. et al. | Aug 1998 | A |
5808676 | Biegelsen et al. | Sep 1998 | A |
5822239 | Ishihara et al. | Oct 1998 | A |
5825317 | Anderson et al. | Oct 1998 | A |
5858533 | Greuter et al. | Jan 1999 | A |
5872739 | Womack | Feb 1999 | A |
5877977 | Essaian | Mar 1999 | A |
5886920 | Marshall et al. | Mar 1999 | A |
5887117 | Desu et al. | Mar 1999 | A |
5919515 | Yano et al. | Jul 1999 | A |
5946224 | Nishimura | Aug 1999 | A |
5953061 | Biegelsen et al. | Sep 1999 | A |
5955213 | Yano et al. | Sep 1999 | A |
5962884 | Hsu et al. | Oct 1999 | A |
5977577 | Evans, Jr. | Nov 1999 | A |
5998819 | Yokoyama et al. | Dec 1999 | A |
6025735 | Gardner et al. | Feb 2000 | A |
6027947 | Evans et al. | Feb 2000 | A |
6031754 | Derbenwick et al. | Feb 2000 | A |
6066868 | Evans, Jr. | May 2000 | A |
6067244 | Ma et al. | May 2000 | A |
6069381 | Black et al. | May 2000 | A |
6087688 | Furuta et al. | Jul 2000 | A |
6091621 | Wang et al. | Jul 2000 | A |
6104049 | Solayappan et al. | Aug 2000 | A |
6121648 | Evans, Jr. | Sep 2000 | A |
6130103 | Cuchiaro et al. | Oct 2000 | A |
6140672 | Arita et al. | Oct 2000 | A |
6144579 | Taira | Nov 2000 | A |
6147895 | Kamp | Nov 2000 | A |
6150184 | Evans et al. | Nov 2000 | A |
6151241 | Hayashi et al. | Nov 2000 | A |
6151242 | Takashima | Nov 2000 | A |
6165802 | Cuchiaro et al. | Dec 2000 | A |
6171934 | Joshi et al. | Jan 2001 | B1 |
6194751 | Evans, Jr. | Feb 2001 | B1 |
6201731 | Kamp et al. | Mar 2001 | B1 |
6207465 | Cuchiaro et al. | Mar 2001 | B1 |
6225156 | Cuchiaro et al. | May 2001 | B1 |
6225654 | Evans, Jr. et al. | May 2001 | B1 |
6225656 | Cuchiaro et al. | May 2001 | B1 |
6236076 | Arita et al. | May 2001 | B1 |
6245451 | Kamisawa et al. | Jun 2001 | B1 |
6245580 | Solayappan et al. | Jun 2001 | B1 |
6246602 | Nishimura | Jun 2001 | B1 |
6255121 | Arita et al. | Jul 2001 | B1 |
6256220 | Kamp | Jul 2001 | B1 |
6285577 | Nakamura | Sep 2001 | B1 |
6307225 | Kijima et al. | Oct 2001 | B1 |
6310373 | Azuma et al. | Oct 2001 | B1 |
6319542 | Summerfelt et al. | Nov 2001 | B1 |
6322849 | Joshi et al. | Nov 2001 | B1 |
6326315 | Uchiyama et al. | Dec 2001 | B1 |
6335550 | Miyoshi et al. | Jan 2002 | B1 |
6339238 | Lim et al. | Jan 2002 | B1 |
6358758 | Arita et al. | Mar 2002 | B1 |
6362068 | Summerfelt et al. | Mar 2002 | B1 |
6365927 | Cuchiaro et al. | Apr 2002 | B1 |
6370056 | Chen et al. | Apr 2002 | B1 |
6372518 | Nasu et al. | Apr 2002 | B1 |
6373743 | Chen et al. | Apr 2002 | B1 |
6396093 | Nakamura | May 2002 | B1 |
6396095 | Shimada et al. | May 2002 | B1 |
6623985 | Igarashi | Sep 2003 | B1 |
6674110 | Gnadinger | Jan 2004 | B1 |
Number | Date | Country |
---|---|---|
408055919 | Feb 1996 | JP |
WO9813300 | Feb 1998 | WO |
Number | Date | Country | |
---|---|---|---|
20030183859 A1 | Oct 2003 | US |
Number | Date | Country | |
---|---|---|---|
60272986 | Mar 2001 | US | |
60227858 | Aug 2000 | US |