Eeproms using carbon nanotubes for cell storage   

CROSS-REFERENCE TO RELATED APPLICATIONS

This application is a continuation of and claims priority under 35 U.S.C. §120 to U.S. patent application Ser. No. 11/053,135, now U.S. Pat. No. 7,528,437, filed on Feb. 8, 2005, and entitled “EEPROMS USING CARBON NANOTUBES FOR CELL STORAGE,” which is related to and claims the benefit under 35 U.S.C. §119(e) of U.S. Provisional Patent Application No. 60/543,497, filed on Feb. 11, 2004, entitled “EEPROMS Using Carbon Nanotubes,” which is herein incorporated by reference in its entirety.

This application is also related to and claims the benefit under 35 U.S.C. §120 to the following applications which are assigned to the assignee of this application, and are herein incorporated by reference in their entirety:

“Non-volatile Electromechanical Field Effect Transistors and Methods of Forming Same” (U.S. patent application Ser. No. 10/864,186), filed on Jun. 9, 2004; and

“Nanotube-on-Gate FET Structures and Applications” (U.S. patent application Ser. No. 10/811,373), filed on Mar. 26, 2004.

1. Technical Field

The present invention relates generally to electronically erasable programmable read only memory (EEPROMS) and more specifically to bi-stable non-volatile electromechanical carbon-nanotube based EEPROMS.

2. Discussion of Related Art

Currently, most memory storage devices utilize a wide variety of energy dissipating devices which employ the confinement of electric or magnetic fields within capacitors or inductors respectively. Examples of state of the art circuitry used in memory storage include FPGA, ASIC, CMOS, ROM, PROM, EPROM, EEPROM, DRAM, MRAM and FRAM, as well as dissipationless trapped magnetic flux in a superconductor and actual mechanical switches, such as relays.

Important characteristics for a memory cell in electronic device are low cost, nonvolatility, high density, low power, and high speed. Conventional memory solutions include Read Only Memory (ROM), Programmable Read only Memory (PROM), Electrically Programmable Memory (EPROM), Electrically Erasable Programmable Read Only Memory (EEPROM), Dynamic Random Access Memory (DRAM) and Static Random Access Memory (SRAM).

ROM is relatively low cost but cannot be rewritten. PROM can be electrically programmed but with only a single write cycle. EPROM (Electrically-erasable programmable read-only memories) has read cycles that are fast relative to ROM and PROM read cycles, but has relatively long erase times and reliability only over a few iterative read/write cycles. EEPROM (or “Flash”) is inexpensive, and has low power consumption but has long write cycles (ms) and low relative speed in comparison to DRAM or SRAM. Flash also has a finite number of read/write cycles leading to low long-term reliability. ROM, PROM, EPROM and EEPROM are all non-volatile, meaning that if power to the memory is interrupted the memory will retain the information stored in the memory cells.

EEPROMS are widely used within the computer industry to store a BIOS (basic input-output system) for a computer, sensor, or processing device, allowing it to load data and system instructions from other storage media when the unit receives first power after being in a quiescent state. The size of the BIOS is typically minimized in design because of the high cost of flash memory.

An EEPROM floating gate cell uses the presence or absence of electrons on a floating gate between the control gate and a FET channel to modulate the FET threshold voltage of a double polysilicon gate FET non-volatile storage device.

Non-volatile memory storage in products such as EEPROMs depends on significant charge transfer through the gate oxide of a storage device to and from a floating poly gate structure. Very high electric fields (>10 MV/cm for gate oxides) corresponding to high program/erase voltages in the range of 10 to 20 volts, for example, are required in the thin oxide structures of non-volatile storage devices to achieve charge transfer when cycling EEPROM storage devices between logical “1” and “0” states. By way of comparison, electric fields used in FET devices for volatile memory and logic operation are typically 5 MV/cm, corresponding to operating voltages of 5 volts or less. Trapping of electrons (and holes) in the gate oxide, sometimes referred to as the trap-up phenomenon, inhibits further tunneling injection and causes degradation in erase and program signal margins (difference between signals corresponding to “1” and “0” logic states) with cycling during memory operation, limits endurance, and slows performance [Ashok K. Sharma, “Semiconductor Memories, Technology, Testing, and Reliability”, IEEE Press, 1996, pp. 275-313; pp. 116-120]. Also, the charge transfer mechanism limits programming (write) and erase times to a very slow 10 us to 1 ms range.

Endurance limits refer to the number of times bits may be cycled between “1” and “0” logic states. Assuming that oxide rupture (shorting) due to high voltage is avoided, gate oxide charge-up reduces the difference in amplitude between “1” and “0” logic states until sensing becomes unpredictable. The number of cycles is typically in the 10,000 to 100,000 range, with some products achieving up to 1 million cycles.

Data retention degradation implies a loss of stored charge on the floating gate with time due to oxide damage resulting from a combination of high voltage and charge flow through the gate dielectric. Prevention of slow leaky bits requires oxides of high integrity, and product testing after a specified number of endurance cycles prior to product shipment.

Reading the logical state of the memory requires determining the state of an EEPROM storage device, and does not require gate oxide conduction. Hence, read times may be in the range of 1 to 50 ns, for example. However, trap-up phenomena may change the threshold voltage of the FET, resulting in decreased cell current, and corresponding read time degradation (increase).

The program (write) and erase mechanisms are inherently slow, typically in the 10 us to 1 ms range. Program and erase times degrade with time because of oxide degradation due to trap-up phenomena due to multiple write-verification cycles, such trapped charge alters the electrical characteristics of the FET channel region, reducing the signal difference between “1” and “0” logic states. This degradation results in multiple write-verification cycles to store information in some bit locations and corresponding performance degradation.

EEPROM products are required to operate from a single 5 volts (or lower) power supply source. Accordingly, high voltages in the 10 to 20 volt range must be generated by on-chip voltage generators and applied to array regions. These high voltage requirements stress non-EEPROM devices as well causing a higher failure rate in the supporting CMOS circuits. In addition, such high voltage requirements make it too difficult to embed EEPROM arrays in typical CMOS logic chips.

Consequently, existing technologies are either non-volatile but are not randomly accessible and have low density, high cost, and limited ability to allow multiple writes with high reliability of the circuit\'s function, or they are volatile and complicate system design or have low density. Some emerging technologies have attempted to address these shortcomings.

For example, magnetic RAM (MRAM) or ferromagnetic RAM (FRAM) utilizes the orientation of magnetization or a ferromagnetic region to generate a nonvolatile memory cell. MRAM utilizes a magnetoresistive memory element involving the anisotropic magnetoresistance or giant magnetoresistance of ferromagnetic materials yielding nonvolatility. Both of these types of memory cells have relatively high resistance and low-density. A different memory cell based upon magnetic tunnel junctions has also been examined but has not led to large-scale commercialized MRAM devices. FRAM uses a circuit architecture similar to DRAM but which uses a thin film ferroelectric capacitor. This capacitor is purported to retain its electrical polarization after an externally applied electric field is removed yielding a nonvolatile memory. FRAM suffers from a large memory cell size, and it is difficult to manufacture as a large-scale integrated component. See U.S. Pat. Nos. 4,853,893; 4,888,630; 5,198,994, 6,048,740; and 6,044,008.

Another technology having non-volatile memory is phase change memory. This technology stores information via a structural phase change in thin-film alloys incorporating elements such as selenium or tellurium. These alloys are purported to remain stable in both crystalline and amorphous states allowing the formation of a bi-stable switch. While the nonvolatility condition is met, this technology appears to suffer from slow operations, difficulty of manufacture and poor reliability and has not reached a state of commercialization. See U.S. Pat. Nos. 3,448,302; 4,845,533; and 4,876,667.

Wire crossbar memory (MWCM) has also been proposed. See U.S. Pat. Nos. 6,128,214; 6,159,620; and 6,198,655. These memory proposals envision molecules as bi-stable switches. Two wires (either a metal or semiconducting type) have a layer of molecules or molecule compounds sandwiched in between. Chemical assembly and electrochemical oxidation or reduction are used to generate an “on” or “off” state. This form of memory requires highly specialized wire junctions and may not retain non-volatility owing to the inherent instability found in redox processes.

Recently, memory devices have been proposed which use nanoscopic wires, such as single-walled carbon nanotubes (CNTs), to form crossbar junctions to serve as memory cells. See WO 01/03208, Nanoscopic Wire-Based Devices, Arrays, and Methods of Their Manufacture; and Thomas Rueckes et al., “Carbon Nanotube-Based Nonvolatile Random Access Memory for Molecular Computing,” Science, vol. 289, pp. 94-97, 7 Jul. 2000. Electrical signals are written to one or both wires to cause them to physically attract or repel relative to one another. Each physical state (i.e., attracted or repelled wires) corresponds to an electrical state. Repelled wires are an open circuit junction. Attracted wires are a closed state forming a rectified junction. When electrical power is removed from the junction, the wires retain their physical (and thus electrical) state thereby forming a non-volatile memory cell.

The use of an electromechanical bi-stable device for digital information storage has also been suggested (c.f. U.S. Pat. No. 4,979,149: Non-volatile memory device including a micro-mechanical storage element).

The creation and operation of a bi-stable nano-electro-mechanical switches based on carbon nanotubes (including mono-layers constructed thereof) and metal electrodes has been detailed in a previous patents and patent application of Nantero, Inc. (U.S. Pat. Nos. 6,643,165, 6,574,130, 6,784,028, 6,706,402, and 6,835,591, and U.S. patent application Ser. Nos. 09/915,093, 10/033,323, 10/341,005, 10/341,055, 10/341,054, 10/341,130, 10/776,059, 10/776,572, and 10/967,858, the contents of which are herein incorporated by reference in their entireties).

SUMMARY

The invention provides EEPROMS Using Carbon Nanotubes for Cell Storage.

Under one aspect of the invention, an electrically erasable programmable read only memory (EEPROM) cell includes cell selection circuitry and a storage cell for storing the informational state of the cell. The storage cell is an electro-mechanical data retention cell in which the physical positional state of a storage cell element represents the informational state of the cell.

Under another aspect of the invention, the storage cell element is a carbon nanotube switching element.

Under another aspect of the invention, an electrically erasable programmable read only memory (EEPROM) cell includes cell selection circuitry and a storage cell for storing the informational state of the cell. The storage includes a nanotube switch and wherein the state of the nanotube switch represents the informational state of the cell.

Under another aspect of the invention, an electrically erasable programmable read only memory (EEPROM) cell includes cell selection circuitry and a storage cell for storing the informational state of the cell. The storage is writable with supply voltages used by said cell selection circuitry.

Under another aspect of the invention, an electrically erasable programmable read only memory (EEPROM) cell includes cell selection circuitry and a storage cell for storing the informational state of the cell. The storage is writable and readable via said selection circuitry with write times and read times being within an order of magnitude.

Under another aspect of the invention, the write times and read times are substantially the same.

Under another aspect of the invention, the storage has no charge storage or no charge trapping.

BRIEF DESCRIPTION OF THE DRAWINGS

In the drawing,

FIG. 1 illustrates a CNT-EEPROM device according to one aspect of the invention;

FIG. 2 illustrates the device of FIG. 1 as the NV Store device, in series with a select transistor;

FIG. 3 illustrates a CNT-EEPROM array with four cells (Cell0-Cell3);

FIG. 4 illustrates waveforms applied to operation of an embodiment of the invention;

FIGS. 5-8 illustrate diagrams of electrical characteristics selected cells according to certain aspects of the invention; and

FIG. 9 illustrates a CNT-NVRAM storage cell using two CNT-EEPROM cells according to certain aspects of the invention.

DETAILED DESCRIPTION

Carbon nanotube fabric-based switches have been described in incorporated references and such description will not be repeated for the sake of brevity.

The CNT-EEPROM device of preferred embodiments uses the presence or absence of an electromechanical connection between the control gate/CNT and a polysilicon gate to modulate the FET threshold voltage of a single polysilicon gate FET non-volatile storage device. The CNT connection is used to modulate device capacitance and is not in series with the FET channel for program (write) or read operations. Thus, cell performance is relatively independent of the CNT resistance value. In each case, a select device is used in series with the EEPROM or CNT-EEPROM non-volatile storage device. The structure and operation of the CNT-EEPROM device is described below.

Also, a CNT-NVRAM cell that uses two CNT-EEPROM devices and two select devices per cell is described. Such a four device non-volatile cell can replace the more conventional flip flop with shadow EEPROM devices, typically a total of 10 to 12 devices per cell.

FIG. 1 illustrates a CNT-EEPROM device of certain embodiments. It has internal device capacitive coupling. The voltage VG of gate G is determined by modulating the internal device capacitive network such that the control gate node (node N1 (CG)) is in electrical contact with gate G via a CNT switch, or gate G is capacitively coupled to control gate CG by capacitor CCG-G. When CNT switch is not in contact with gate G, CNT switch is in contact with the insulator of release plate N4(RP). The capacitance network is modulated (activated) by controlling the presence or absence of an FET channel using a diffusion voltage applied to diffusion N2 or N3. This modulation (activation, write) mechanism is well known and is described in the referenced applications. A sufficiently large voltage is applied to the control gate CG, node N1, such that an FET channel is formed. If the diffusion voltage on diffusion N2 or N3 is zero, the channel remains, eliminating CCH-SUB from the capacitance network. If the diffusion voltage on diffusion N2 or N3 is positive, 0.5 to 1.0 volts, for example, then electrons are removed from the channel region, and CCH-SUB capacitance is part of the device (cell) capacitance network. Capacitance values are selected by design. Typical relative capacitance values are CCG-G=0.25, CG-CH=1, and CCH-SUB=0.2, for example.

FIG. 2 illustrates the device of FIG. 1 referred to as the NV Store device, in series with a select transistor. If the CNT switch is programmed (in contact with floating gate G), the cell is in the “1” state. If the CNT switch is released (not in contact with floating gate G), the cell is in the “0” state. The bit line BL is connected to the NV Store device, and is used during both read and write. During read, bit line BL is precharged, word line WL transitions to a higher voltage than the FET threshold voltage, select line SL transitions to a high voltage (VSL=VDD, for example) and turns the select transistor ON. If CG is electrically connected to G by a CNT switch, then VG exceeds the VTH of the FET, current flows, and BL discharges. If CG is capacitively coupled to G, then VG is less than the VTH of the FET, no current flows, and BL is not discharged.

By way of example, if a programmed “1” state is stored, control gate CG is electrically connected to floating gate G by a CNT switch. Assuming VTH of the FET is 0.8 volts and the control gate voltage is 1.5 volts, then voltage VG=VCG=1.5 volts, the channel regions of the NV store and select transistors are both ON, and the precharged bit line BL is discharged to ground. The bit line precharge read voltage may be 0.5 to 1 volt, for example. If a released “0” state is stored, floating gate G is not electrically connected to control gate CG but is capacitively coupled instead. For a control voltage of 1.5 volts, and a capacitance network having typical relative capacitance values of CCG-G=0.25, CG-CH=1, and CCH-SUB=0.2, for example, floating gate voltage VG=0.6 volts, NV store device channel region is OFF, and bit line BL is not discharged. The voltage difference between control voltage and floating gate voltage is 0.9 volts. Therefore CNT switch write (program) threshold voltage VCNT-WR-TH must be greater than 0.9 volts.

FIG. 7 further below illustrates the NV store device electrical characteristic in the write (programmed) “1” state, and the released (erased) “0” state. As illustrated in FIG. 7, if the control voltage VCG is greater than 0.7 of volt but less than 1.8 volts, 1.5 volts for example, bit line BL will be discharged if the NV store device is programmed to a “1” state, but will not discharge the NV store device is released to a “0” state. The read operation is further illustrated in FIG. 8.

One mode of writing (programming) the NV Storage device of FIG. 2 using bit line BL requires the device to be in a released (erased) state, with the CNT switch in contact with the insulator of the release plate N4 Therefore, one method of writing (write mode) is release-before-write. Select line (SL) voltage VSL=0 such that shared node N2 is floating. Release is accomplished using a select scheme that releases individual array bits, or groups of bits, as described further below.

Assuming the NV Store device is in the released state, write (programming) is then accomplished by setting VWL=VDDW, the required write (programming) voltage. If VBL=0, then a channel is present in the NV storage device, gate G is tightly capacitively coupled to the channel which is at zero volts, and the voltage between CG and G is sufficiently high that the threshold voltage of the CNT switch VCNT-WR-TH is exceeded. The CNT switch forms an electrical contact between CG and G, and the NV Store device is in a low FET VFET-TH programmed state as shown in FIG. 7. If, however, the bit line voltage is set at a small positive voltage in the range of 0.5 to 1.0 volts, for example, then there is no FET channel, capacitance CCH-SUB is in series with the capacitance CG-CH, gate G is not tightly coupled to ground, VCNT-WR-TH is not exceeded, and the NV Store device remains in the high FET VFET-TH released (erased) state.

By way of example, programming the NV store device to a “1” state occurs as follows. The bit line voltage of BL(N3) shown in FIG. 2 is set to zero volts, modulating the capacitance network formed when the FET gate G of storage device CNT store is activated, such that only relative capacitance values of CCG-G=0.25 and CG-CH=1 are in the capacitance network. Word line WL (N1) shown is connected to control gate CG & CNT, where a CNT switch is integrated in the control gate structure CG & CNT as described in the above references. Control voltage VCG applied to WL(N1) transitions to 1.8 volts, for example, and the capacitance network capacitively couples floating gate voltage VG to 0.2 VCG. The voltage across the CNT switch is the difference between VCG applied to WL(N1) and the voltage 0.2 VCG on floating gate G, or approximately 1.45 volts. If the CNT switch write (programming) threshold voltage VCNT-WR-TH is less than 1.45 volts, then the CNT will switch into contact with gate G, programming the NT store device to the “1” state. If the NV store device is to store a released (erased) “0” state, no switching is needed because of the release-before-write method.

The electrical characteristics of the NT store device illustrate both the non-volatile programmed “1” state and the non-volatile released “0” state as shown in FIG. 7. Programming and releasing the NV Store device is described further below with respect to FIGS. 4, 5, and 6. The release-before-write mode described above is one operating mode. Other write modes that do not require release-before-write are described in the references.

FIG. 3 illustrates a CNT-EEPROM array with four cells (Cell0-Cell3). The array in FIG. 3 may store a unique bit (“1” or “0”) in each of the four cells connected to bit lines BL0, BL1, BL2, and BL3. Release lines RL0 and RL1 connect to corresponding release nodes, word lines WL0 and WL1 connect to control gates, and SL0 and SL1 connect to select device gates. Word lines and select lines are substantially parallel, and are substantially orthogonal to release lines and bit lines. Each cell in the CNT-EEPROM array operates as described in FIG. 2.

FIG. 4 illustrates the waveforms applied to the release lines, word lines, bit lines, and the resulting gate voltage VG of the selected cell (Cell 0 in this example) during release from a “1” and a “0” stored state. The select line voltage VSL=0 and isolates the NT storage device from ground. The word line voltage VWL, which corresponds to control gate voltage VCG associated with control gate node N1(CG) of FIG. 1, is set to zero (VWL=0). If NT store 0 is in a programmed “1” state, the CNT switch connects the CG & CNT node and floating gate G. The release voltage VRL is switched to VDDW. The voltage VDDW must exceed the release threshold voltage of the CNT switch VCNT-RL-TH in order to release (erase) the “1” state. The release threshold voltage may differ from the program (write) threshold voltage VCNT-WR-TH. By way of example, for release voltage VDDW=2.5 volts, the release threshold voltage VCNT-RL-TH maximum is less than 2.5 volts. To prevent adjacent “bit disturb,” VCNT-RL-TH must be greater than 1.8 volts, for example, as shown further below. Once released, the floating gate voltage is not longer at ground level but assumes a voltage between 0 and VDD. If NT store 0 is in a released “0” state, the CNT switch is not connected to VG and is therefore is already in a released (erased) state, and remains in the released state.

FIG. 5A illustrates cell voltages at selected cells, ½ selected cells, and unselected cells in the array of FIG. 3 in which cell 0 is released, and cells 1, 2, and 3 are unchanged, and where common release line RL1 is at VDD. The NV Store device is designed such that only cells with the full select voltage difference ΔV=VDDW, in this example 2.5 volts, are released. Cells storing a “1” state release when the CNT connection switches from gate G to the oxide of release plate N4 Cells storing a “0” state do not have the CNT switch connected to gate G, so no CNT switching takes place, and they remain in the released state. FIG. 5B illustrates a release operation in which cell 0 is released, and cells 1, 2, 3, and 4 are unchanged, and where common release line RL1 is at zero volts. Cells storing a “0” state do not have the CNT switch connected to gate G, so no switching takes place, and they remain in the released state.

FIG. 6 illustrates writing (programming) from a released state to a “1” state, or remaining at a “0” state, controlled by bit line voltage VBL. If VBL=0, then a “1” state is stored in the NV Store device. By way of example, if the word line voltage transitions to VDDW=1.8 volts, for example, then the capacitance network couples floating gate voltage VG to 0.2×1.8≈0.4 volts. The write (programming) voltage across the CNT switch is therefore 1.4 volts. If VCNT-WR-TH=0.9 to 1.2 volts, for example, then the CNT switch is programmed to the “1” state, in contact with floating gate G, with gate G voltage VG as shown in FIG. 6. If, however, the bit line voltage is positive, 0.5 to 1 volt, for example, then the capacitance network is changed because there is no FET channel in the storage device. If the word line voltage transitions to VDDW=1.8 volts, for example, then the capacitance network couples the floating gate VG to 0.6×1.8≈1 volt. The write (programming) voltage across the CNT switch is therefore 0.8 volts. If VCNT WR-TH=0.9 to 1.2 volts, for example, then the NV store device remains in the released state.

FIG. 7 illustrates an example of the FET I-V characteristic for a NV Store device in the written (programmed) “1” state and the released (erased) “0′ state. The FET threshold voltage VFET-TH is approximately equal to the voltage VCG where the I-V characteristic intersects the horizontal axis. The read voltage applied to VCG is selected to be greater than the VFET-TH for a programmed “1” state, that is, greater than 0.7 volts, and less than the VFET-TH voltage for a programmed “0” state, that is, less than 1.8 volts.

As illustrated in FIG. 8, the bit line voltage VBL will droop if the NV Store device is programmed to the “1” state, but will remain at VDD if the NV Store device is programmed to the “0” state.

FIG. 9 illustrates a CNT-NVRAM storage cell using two CNT-EEPROM cells, such that there are four transistors, two NV Store transistors and two select transistors. By comparison, a typical CNT-NVRAM using a flip flop and shadow EEPROM devices on each node has a total of 8 to 12 transistors. The CNT-NVRAM cell stores the true and complement of each data bit. The cell illustrated in FIG. 9 stores true and complement data in transistors NV Store and NV Store-b. The NV RAM of FIG. 9 uses 1 WL, 1 SL, 1 ground, 1 RL, but 2 bit lines, BL and BL-b. The operation of the CNT-NVRAM cell is similar to the CNT-EEPROM cell operation. The advantage of the CNT-NVRAM over the CNT-EEPROM cell is that one of the bit line pairs will discharge, that is BL or BL-b will discharge. This cell can therefore be sensed using sensing techniques similar to those of SRAM cells. The performance will not be gated by CNT resistance because there is no CNT resistance in series with the write or read paths (CNTs modulate the internal device capacitance). Read time is expected to be similar to the read time of an SRAM cell. Write will be longer than for an SRAM cell because of the addition of the release-before-write requirement. However, the additional delay is for the first write operation. When writing a block of data, such as in Page Mode, the release operation delay can be hidden by using an appropriate architecture. Also, write modes that not require release-before-write described in the above references eliminate the extra delay.

An aspect of the present embodiment (carbon nanotube fabric-based—CNT-EEPROMs is to solve these problems. The CNT-EEPROM uses electromechanical switching in the device to modulate an FET threshold voltage. This method of programming uses lower voltages and is faster than prior art EEPROM programming using hot electrons.

CNT-EEPROM storage device can be used to build EEPROMs and NRAMs using CNT-on-Gate. The CNT-EEPROM device is used with a select device to form the basic cell. The cell can be programmed and erased faster and at lower voltages than conventional EEPROM cells.

Preferred embodiments speed up write (program) and erase operation. They also reduce or eliminate endurance limitations, data retention degradation, read, program, and erase time degradation. They also reduce or eliminate the high program and erase voltage requirements, and enable embedded EEPROMs with CMOS logic circuits.

Preferred embodiments eliminate the charge transfer storage mechanism to and from a floating gate embedded in a gate oxide, as is done in conventional EEPROM. More specifically, to replace the charge transfer storage mechanism to and from a polysilicon non-volatile storage gate, preferred embodiments use a non-volatile nanotube storage device. Non-volatile storage using a nanotube mechanical state enables EEPROMs that can program (write) and erase (release) at voltage levels of 5 volts and below. Also, such EEPROM operate such that read, write (program) and erase times are substantially the same, 1 to 50 ns, for example. Also, supporting CMOS circuits can operate at standard CMOS voltage levels of 5 volts and below. Also, such EEPROM devices can be embedded in CMOS logic circuits.

The following patent references refer to various techniques for creating nanotube fabric articles and switches and are assigned to the assignee of this application, each of which is hereby incorporated by reference in their entirety: U.S. patent application Ser. No. 10/341,005, filed on Jan. 13, 2003, entitled Methods of Making Carbon Nanotube Films, Layers, Fabrics, Ribbons, Elements and Articles; U.S. patent application Ser. No. 09/915,093, filed on Jul. 25, 2001, entitled Electromechanical Memory Array Using Nanotube Ribbons and Method for Making Same; U.S. patent application Ser. No. 10/033,032, filed on Dec. 28, 2001, now U.S. Pat. No. 6,784,028, entitled Methods of Making Electromechanical Three-Trace Junction Devices; U.S. patent application Ser. No. 10/033,323, filed on Dec. 28, 2001, entitled Electromechanical Three-Trace Junction Devices; U.S. patent application Ser. No. 10/128,117, filed on Apr. 23, 2002, now U.S. Pat. No. 6,835,591, entitled Methods of NT Films and Articles; U.S. patent application Ser. No. 10/341,055, filed Jan. 13, 2003, entitled Methods of Using Thin Metal Layers to Make Carbon Nanotube Films, Layers, Fabrics, Ribbons, Elements and Articles; U.S. patent application Ser. No. 10/341,054, filed Jan. 13, 2003, entitled Methods of Using Pre-formed Nanotubes to Make Carbon Nanotube Films, Layers, Fabrics, Ribbons, Elements and Articles; U.S. patent application Ser. No. 10/341,130, filed Jan. 13, 2003, entitled Carbon Nanotube Films, Layers, Fabrics, Ribbons, Elements and Articles; U.S. patent application Ser. No. 10/776,059, filed Feb. 11, 2004, entitled Devices Having Horizontally-Disposed Nanofabric Articles and Methods of Making The Same; and U.S. patent application Ser. No. 10/776,572, filed Feb. 11, 2004, entitled Devices Having Vertically-Disposed Nanofabric Articles and Methods of Making the Same.

It will be further appreciated that the scope of the present invention is not limited to the above-described embodiments but rather is defined by the appended claims, and that these claims will encompass modifications and improvements to what has been described.