Abstract: A memory cell array for a nonvolatile memory device having single-transistor cells (10). Row lines (15) connect the control gates of each row of cells. Column lines (17) connect the drain regions (11) and source regions (12) of columns of cells, such that pairs of row-adjacent cells share a column line (17). Each shared column line (17) has two junctions for each pair of cells that share the column line. One junction is graded for source regions (12) and the other is abrupt for drain regions (11).

Abstract: In accordance with one embodiment of the invention, a nonvolatile memory array is encased in a P-tank, and the P-tank encased in a deep N-tank, the two tanks separating the memory array from the substrate and from the other circuitry of the integrated memory circuit. The deep N-tank allows application of a negative voltage of perhaps -8 V to the P-tank encasing the memory array. Application of that negative voltage permits the cells of the memory array to be programmed with voltage pulses having a peak value of about +10 V, rather than the +18 V peak value of prior-art memory arrays. Because the external circuitry, such as the wordline driver circuit, need drive the wordlines at +10 V rather than +18 V, the invention permits construction of that external circuitry using thinner gate insulators and space-saving shorter dimensions.

Abstract: The drain-to-source voltage and current for programming a selected nonvolatile memory cell 10 are achieved efficiently by pumping the source 11 of a selected cell 11 to a voltage less than the voltage VSS at the reference-voltage terminal of the memory cell array while, at the same time, pumping the drain 12 of the selected cell 10 to a voltage greater than the voltage VCC, which may be 3 V, at the supply-voltage terminal of the memory cell array. The cell substrate W2 is pumped to a voltage close to the voltage of the source 11 and, optionally, below the voltage of the source 11. One or more simple charge-pump circuits convert the output of the voltage supply VCC to a source-drain voltage and current capable of programming the selected nonvolatile cell 10 by hot carrier injection.

Abstract: In accordance with one embodiment of the invention, an integrated circuit memory capable of receiving address signals has a plurality of array modules. Each array module has a plurality of memory cells arranged in rows and columns and has a means for addressing the rows and columns of the array modules in response to the address signals. The integrated memory includes a circuit for de-coupling an input address signal from array modules and for coupling of a fixed address signal to the array modules, allowing the conversion of the integrated circuit memory to a universal modular memory.

Abstract: The CMOS high-voltage sensor circuit has a voltage reference including, for example, of four N-channel MOS transistors; one pass-gate P-channel transistor; one current-mirror P-channel MOS transistor; and a conventional high-voltage sensor including, for example, of two P-channel MOS transistors and one N-channel MOS transistor. The sensor circuit of this invention generates a high-voltage signal at the output if the input voltage is greater than both the reference voltage plus two P-channel threshold voltages and the supply voltage Vcc plus two P-channel threshold voltages. The power-up or power-down sequence may be in any order without adversely affecting the operation of the circuit of this invention.

Abstract: The power-on reset circuit of this invention includes a current-sensing circuit, a pulse-stretching circuit, and a voltage-reference circuit. The voltage-reference circuit consists, for example, of one N-Channel and one P-Channel MOS transistor. The circuit of this invention uses a static voltage reference comprised of CMOS transistors to detect the power-up condition. The circuit of this invention improves detection of a transient power-supply voltage Vcc loss and detects that power-supply voltage transient on both rising and falling edges.

Abstract: The column-line short detection circuit of this invention includes a special test circuit that turns off wordlines (15), a N-channel transistor (23) for each column line (18), a decoder (19a) that uses only the least significant column address (20d) for input to the test circuit, and a sensor (SA) to detect current between shorted column lines (18). Because the column-line short detection circuit of this invention uses only the least significant address as input for column decoder (19a), it requires a very small number of transistors.

Abstract: The circuit and method of this invention provide for rapid and reliable detection of depleted or nearly-depleted cells in a column. The circuit is formed on the substrate of a nonvolatile, integrated-circuit memory including rows and columns of memory cells. The drain of each memory cell is connected to a drain-column line and the control gate that is connected to a wordline. One input of a sense amplifier is connected to the drain-column line. The other input of the sense amplifier is connected to a current reference formed on said substrate. The wordline is connected to a wordline test voltage and the output of the sense amplifier is coupled to an output pin of the integrated circuit. The current through the drain-column line is compared with the current through the current reference and, if the current through the drain-column line is sufficiently close to the current through said current reference, a signal is transmitted to an output pin of the integrated circuit.

Abstract: An electrically-erasable, electrically-programmable, read-only-memory cell array is formed in pairs at a face of a semiconductor substrate (22). Each memory cell includes a source region (11) and a drain region (12), with a corresponding channel region between. A Fowler-Nordheim tunnel-window (13a) is located over the source line (17) connected to source (11). The source line (17) consists of alternating buried N+ windows (17a) and source regions (11). A floating gate (13) includes a tunnel-window section. A control gate (14) is disposed over the floating gate (13), insulated by an intervening inter-level dielectric (27). The floating gate (13) and the control gate (14) include a channel section (Ch). The channel section (Ch) is used as a self-alignment implant mask for the source (11) and drain (12) regions, such that the channel-junction edges are aligned with the corresponding edges of the channel section (Ch).

Abstract: An electrically-erasable, electrically-programmable, read-only-memory cell array is formed in pairs at a face of a semiconductor substrate (22). Each memory cell includes a source (11) and a drain (12), with a corresponding channel (Ch) between. A control gate (14) is disposed over the floating gate (13), insulated by an intervening inter-level dielectric (27). The floating gate (13) and the control gate (14) include a channel section (Ch). The channel section (Ch) is used as a self-alignment implant mask for the sources (11) and drains (12), such that the channel-junction edges are aligned with the corresponding edges of the channel section (Ch). Each memory cell is programmed by hot-carrier injection from the channel to the floating gate (13), and erased by Fowler-Nordheim tunneling from the floating gate (13) to the source (11).

Abstract: The structure and method of this invention provide, for example, electrical isolation between active elements in adjacent rows and/or columns of an integrated circuit by use of a self-aligned field-plate conductor formed over and insulated from the substrate regions that are bounded by the channel regions of field-effect transistors in adjacent rows and that are bounded by the bitlines forming those transistors in a column. The field-plate conductor is formed, for example, in a strip that extends over the isolation areas and thermal insulator regions between row lines of the memory cell array. The field-plate conductor strip is connected to a voltage supply that has a potential with respect to the potential of the semiconductor substrate which causes the isolation areas to be nonconductive. Component density may be increased over that of prior-art structures and methods.

Abstract: An integrated circuit structure and process relating to a self-aligned window at the recessed junction of two insulating regions formed on the surface of a semiconductor body. The window may include a trench forming an isolation region between doped semiconductor regions, or may include an electrical conductor connected to a doped semiconductor region, or may include an electrical conductor separated from doped semiconductor regions by an electrical insulator. Embodiments include, but are not limited to, a field-effect transistor, a tunnelling area for a floating gate transistor, and an electrical connection to a doped area of the substrate.

Abstract: A nonvolatile memory cell having separate regions for programming and erasing. The cells are formed in an array at a face of a semiconductor body, each cell including a source that is part of a source-column line and including a drain that is part of a drain-column line. Each cell has first and second sub-channels between source and drain. The conductivity of the first sub-channel of each cell is controlled by a field-plate, which is part of a field-plate-column line, positioned over and insulated from the first sub-channel. The conductivity of each of the second sub-channels is controlled by a floating gate formed over and insulated from the second sub-channel. Each floating gate has a first tunnelling window positioned over the adjacent source-column line and has a second tunnelling window positioned over the adjacent drain-column line. Row lines, including control gates, are positioned above and insulated from the floating gates of the cells for reading, programming and erasing the cells.

Abstract: A nonvolatile memory has pairs of cells in which each cell includes a control gate, a floating gate and a source/drain diffusion. A first cell in each of the pairs is producible to have one value of floating-gate to diffusion capacitance. A second cell in each of the pairs is producible to have a second value of floating-gate to diffusion capacitance different from the first value. The memory includes a first circuit for applying a first erasing pulse to the control gates and the diffusions of the first cells of the pairs and includes a second circuit for applying a second erasing pulse to the control gates and the diffusions of the second cells of the pairs. The first erasing pulse is adjustable to have a different magnitude than the second erasing pulse in order to narrow the margin of erased threshold voltages and thereby compensate for misalignment.

Abstract: A wordline-decode system of a nonvolatile memory array is split into three smaller decoding subsystems (a Read-Mode Decode Subsystem, a Program/Erase-Mode Decode Subsystem and a Segment-Select Decoder Subsystem). The segmented array has small bitline capacitance and requires few input connections to each decoding subsystem. The Read-Mode Decoder circuitry and the Program/Erase-Mode Decoder circuitry are separated, allowing the Read-Mode Decoder circuitry to be desired for high speed access and allowing the Program/Erase-Mode Decoder circuitry to be desired for high voltage operation. Buried-bitline segment-select transistors reduce the area required for those transistors. Erasing may be performed after first checking each row of a segment to determine the present of any over-erased cells. Programming may be performed by allowing the common source-column lines of the selected segment to float and by placing preselected voltages on the appropriate wordline and drain-column line.

Abstract: A circuit for driving a wordline or group of wordlines in a floating-gate type EEPROM cell array includes a read-driver subcircuit for switching positive read voltages, a program-driver subcircuit for switching positive programming voltages and, optionally, a subcircuit for switching negative erasing voltages. The read-driver subcircuit may be constructed using relatively short-channel transistors for relatively high speed operation when connected to high-capacitance wordlines. On the other hand, the program-driver subcircuit may be constructed using relatively long-channel transistors and those long-channel transistors may be located on the memory chip remotely from the memory cells and from the read-driver circuit. P channel isolating transistors are used to isolate unused circuitry during operation. A voltage translator in the program-driver subcircuit has a transistor configuration that lessens the probability that the breakdown voltages of those transistors will be exceeded.

Abstract: A structure and method for improving the sense margin of nonvolatile memories is disclosed. An improvement to the sense margin of nonvolatile memories is accomplished by improving the margin both for "ones" at low control gate voltage Vcc and for "zeros" at high control gate voltage Vcc. Improvement in sensing at low control gate voltages Vcc is accomplished by skewing the sense amplifier response characteristics by forming the channel length of the reference memory cell to have a longer channel length than the memory cells of the array.

Abstract: A diffusionless source/drain conductor, electrically-erasable, electrically-programmable read-only memory cell is formed at a face of a semiconductor layer (38) of a first conductivity type and includes a source conductor (10), a drain conductor (12), a channel region (18), and a tunnel region (22). Source conductor (10) and drain conductor (12) are disposed to create inversion regions, of a second conductivity type, opposite said first conductivity type, in the source inversion region (14) and drain inversion region (16) of semiconductor layer (38) of the layer semiconductor, upon application of voltage. Thin oxide tunneling window (22) is disposed adjacent source conductor (10). A floating gate (24) disposed adjacent tunneling window can be charged or discharged by Fowler-Nordheim tunneling when a voltage is applied between the inversion created in source inversion region (14) and a control gate (26) insulatively adjacent floating gate (24).

Abstract: A method for making a NMOS self-aligned contact in CMOS circuits without an extra mask is described. The maskless contact technique makes use of the fact that the blanket N-type implant, self-aligned to exposed field-oxide edge, will not change the P+ diffusion to N-type. The net P+ concentration in the contact region is reduced slightly but does not degrade the PMOS device performance.

Abstract: The dielectric between the floating gate and the control gate, in an EEPROM or other floating gate memory is made by forming an oxide/nitride stack over the (first polysilicon) control gate. This dielectric not only provides a very high specific capacitance, which is desired to provide tight coupling of the control to the floating gate, but also provides excellent dielectric integrity. Moreover, the thickness of this dielectric layer does not exhibit any uncontrolled increase during exposure to second gate oxidation. Thus, the polysilicon-to-polysilicon dielectric is not only of high specific capacitance and high integrity, it is also very uniform.