Texas Instruments Corporation

Ashwin H Shah

Pallab K Chatterjee

James D Gallia

Shivaling S Mahant-Shetti

Clarence W Teng

Robert R Doering

Richard H Womack

Chu-Ping Wang

David K Liu

William F Richardson

Kwok K Chau

Kueing-Long Chen

Bing W Shen

I-Fay F Wang

Mark Bordelon

An antifuse structure (20) and method of fabrication are provided. A first conductive layer (A) is etched according to a first mask (62a) having a first pattern and according to a second mask (64a) having a second pattern. A first insulative layer (30) is disposed over the first conductive layer (A) and etched according to a third mask (40a) having a third pattern to expose at least one section of the first conductive layer (A). A second insulative layer (26) is disposed adjacent at least one exposed section of the first conductive layer (A). A second conductive layer (1) is disposed over the second insulative layer (26) so that the antifuse structure (20) includes at least one antifuse region (A1) where a section of the second insulative layer (26) is adjacent the first (A) and second (1) conductive layers. The antifuse region (A1) has a sublithographic vertical dimension (t) according to a thickness of the first conductive layer (A). Further, the antifuse region (A1) has a sublithographic horizontal dimension (L) according to an overlap between the first (62a) and second (64a) masks.

Antifuse structure and method of fabrication

BiCMOS circuits are disclosed which achieve high speed operation under a wide range of loading conditions. The circuits are capable of providing a full output voltage swing and dissipate virtually no static power. The BiCMOS circuits are implemented using both CMOS and bipolar transistors. The circuits use their output signal to control the CMOS transistors that overcome bipolar output drops for full swing operation. The same fundamental CMOS and bipolar configurations can be applied to implement complex and simple logic functions such as NAND, NOR, AND, or OR operations.

High performance BiCMOS logic circuits with full output voltage swing up

DRAM cells and arrays of cell on a semiconductor substrate, together with methods of fabrication, are disclosed wherein the cells are formed in pairs or quartets by excavating a trench or two trenches through the cell elements to split an original cell into two or four cells during the fabrication. The cells include vertical field effect transistors and capacitors along the trech sidewalls with word lines and bit lines crossing over the cells.

Vertical DRAM cell and method

An improved memory cell layout (54) is formed including a trench cell (60) formed in a semiconductor substrate (58). The memory cell layout (54) includes a bitline (56) and a wordline (62) for storing and accessing charge. The charge is stored on a capacitor formed from a conductor (68), an insulating region (70) and a semiconductor substrate (58). Bitline (56) is primarily tangential to a trench cell (60), or may surround the periphery thereof. A wordline (62) overlies trench cell (60) and extends therein, and further may be formed of a width narrower than trench cell (60).

Trench memory cell

A static random access memory, wherein power consumption is reduced by using asynchronous edge-triggered power down gates to power up only elements in the critical circuit path for only as long as necessary to access the memory. Thus, power consumption in the memory is reduced to nearly an absolute minimum. This invention uses the address transition clock to provide an asynchronous power up function to various parts of the static RAM so that only the circuit which is propagating the signal is powered up and the power is held high just long enough for the signal to propagate. This is performed using intrinsic timing elements of the RAM critical path so that the timing of the signal and power cycles track each other.

Static random access memory with asynchronous power-down

The described embodiments of the present invention provide structures, and a method for fabricating those structures, which include a memory cell formed within a single trench. A trench is formed in the surface of a semiconductor substrate. The bottom portion of the trench is filled with polycrystalline silicon to form one plate of a storage capacitor. The substrate serves as the other plate of the capacitor. The remaining portion of the trench is then filled with an insulating material such as silicon dioxide. A pattern is then etched into the silicon dioxide when opens a portion of the sidewall and the top portion of the trench down to the polycrystalline capacitor plate. A contact is then formed between the polycrystalline capacitor plate and the substrate. Dopant atoms diffuse through the contact to form a source region on a sidewall of the trench. A gate insulator is formed by oxidation and a drain is formed at the surface of the trench adjacent to the mouth of the trench. Conductive material is then formed inside the open portion of the upper portion of the trench thereby forming a transistor connecting the upper plate of the storage capacitor to a drain region on the surface of the semiconductor substrate.

Dram cell and method

The described embodiments of the present invention provide structures, and a method for fabricating those structures, which include a memory cell formed within a single trench. A trench is formed in the surface of a semiconductor substrate. The bottom portion of the trench is filled with polycrystalline silicon to form one plate of a storage capacitor. The substrate serves as the other plate of the capacitor. The remaining portion of the trench is then filled with an insulating material such as silicon dioxide. A pattern is then etched into the silicon dioxide which opens a portion of the sidewall and the top portion of the trench down to the polycrystalline capacitor plate. A contact is then formed between the polycrystalline capacitor plate and the substrate. Dopant atoms diffuse through the contact to form a source region on a sidewall of the trench. A gate insulator is formed by oxidation and a drain is formed at the surface of the trench adjacent to the mouth of the trench. Conductive material is then formed inside the open portion of the upper portion of the trench thereby forming a transistor connecting the upper plate of the storage capacitor to a drain region on the surface of the semiconductor substrate.

One embodiment of the present invention includes a vertical inverter. A layer of P-type material is formed on the surface of an N+-type substrate, followed by formation of an N+ layer, a P+ layer, an N- layer, and a P+ layer. (Of course different doping configurations may be used and remain within the scope of the invention.) A trench is then etched along one side of the stack thus formed and a connector is formed to the middle P+ and N- layers. Another trench is then formed where a gate insulator and a- gate are formed. The gate serves as the gate for both the N-channel and P-channel transistors thus formed.

Vertical inverter circuit

A scheme for addressing memory cells in random access memory arrays includes bit lines divided into a plurality of segments. Each pair of bit lines has a sense amp at each end coupled to both bit lines in the pair. Word lines address memory cells coupled to each bit line of the pair. When a pair of memory cells is accessed, the bit lines are electrically divided so that one memory cell is coupled to one sense amp through one bit line, and the other memory cell is coupled to the other sense amp through the other bit line. The memory cells can be coupled to the bit lines through segment lines, with each segment line connecting a subset of the memory cells to a bit line, in order to reduce capacitances presented to the sense amps. An alternating linear array of sense amps and bit line pairs can be used to increase overall density of the memory array by allowing sense amps to access more than one bit line pair. The bit lines are addressed so that each sense amp receives data from one one bit line pair at a time. Segment lines having no currently addressed memory cells can be coupled to the sense amps in order to better balance input capacitances presented thereto. Selecting the bit line sections, segments, and memory cells in the proper order minimizes the effect of noise due to stray capacitances by causing them to appear as a common mode signal across the bit line pairs.

Dual ended folded bit line arrangement and addressing scheme

Constant pulse width generator having applicability to a static random access memory (SRAM) where a constant width output pulse is desired, regardless of the address line activity, until reset, for powering up peripheral circuits of the static random access memory when an input address changes. An exclusive-NOR circuit has address inputs including the address line and the address line delayed. The constant pulse width generator comprises a monostable delayed feedback loop which is provided on the output of the exclusive-NOR circuit, with the output of the loop changing only upon receipt of a change of state from the exclusive NOR circuit, otherwise remaining stable until the delay resets the output. The output of the constant pulse width generator is a pulse as wide as the delay introduced in the address input signal.

Constant pulse width generator

A semiconductor memory includes a memory array (10) that is operable to be addressed in either the page mode or the static column decode mode. A column address transparent latch (20) is provided which is controlled to either directly input a column address to a column decoder (26) or to latch the address in response to the generation of the column address strobe. A sequence detect circuit (30) detects the sequence to the row address strobe and the column address strobe to determine whether the page mode or the static column decode mode is generated. The sequence detect circuit (30) generates a Y-enable signal in a circuit (31) for control of the latch (20).

Semiconductor memory with static column decode and page mode addressing

Column decoding is performed using multiply decoded subsets of column address bits bussed across the array to multiple first stage and second stage column multiplexers. That is, for example, in an 8k by 9 memory wherein each subarray contains 16 selectable columns at each bit position, two of the address bits would be fully decoded to provide four buss lines across the chip. Each column has a primary sense amplifier, controlled by one of these four decoded lines. The outputs of each set of four primary sense amplifiers are multiplexed into a secondary sense amplifier, (preferably on a local three-scale buss) and the output of each secondary sense amplifier is selected or deselected by four buss lines which are the decoded signals corresponding to the other two address bits which select one of 16 columns. Preferably multiplexing of the output of the secondary sense amplifiers is accomplished by a three-state buffer, so that the output of these buffers can be accomplished as a wired-or function.

Memory decoding circuit

Static random access memory having an edge-triggered power up architecture. Each element of the signal path is powered up only during the period when it is expected to be active. Separate delays are provided to tailor the delay of the power-up pulses for each separate circuit component, and separate 1-shot pulse generators, with the pulse width tailored to the power-up duration required by each circuit element, are provided for each circuit element.

SRAM with constant pulse width

DRAM cells and arrays of cells on a semiconductor substrate, together with methods of fabrication, are disclosed wherein the cells are formed in pairs or quartets by excavating a trench or two trenches through the cell elements to split an original cell into two or four cells during the fabrication. The cells include vertical field effect transistors and capacitors along the trench sidewalls with word lines and bit lines crossing over the cells.

A memory decoder wherein a power-up device is interposed between a NOR decoder and ground (VSS), rather than between the decoder and VDD. Preferably the signal to the power-up transistor is itself decoded, so that the power-consuming NOR circuits are inactive over a majority of the chip, even during power-up conditions.

Growing community of inventors

Dallas, TX, United States of America

Ashwin H Shah