Abstract:
A multi-state current-switching magnetic memory element has a magnetic tunneling junction (MTJ), for storing more than one bit of information. The MTJ includes a fixed layer, a barrier layer, and a non-uniform free layer. In one embodiment, having 2 bits per cell, when one of four different levels of current is applied to the memory element, the applied current causes the non-uniform free layer of the MTJ to switch to one of four different magnetic states. The broad switching current distribution of the MTJ is a result of the broad grain size distribution of the non-uniform free layer.
Abstract:
One embodiment of the present invention includes multi-state current-switching magnetic memory element including a stack of two or more magnetic tunneling junctions (MTJs), each MTJ having a free layer and being separated from other MTJs in the stack by a seeding layer formed upon an isolation layer, the stack for storing more than one bit of information, wherein different levels of current applied to the memory element causes switching to different states.
Abstract:
An oscillator circuit residing internally to a semiconductor device for generating a clock signal for use by digital circuits. The oscillator circuit includes a voltage regulator circuit responsive to frequency selection signals for selecting a predetermined frequency and a supply voltage. The voltage regulator circuit is operative to generate a voltage reference signal having a voltage level being adjusted to compensate for variations due to temperature, process and supply voltage variations. The oscillator circuit further includes a ring oscillator circuit responsive to the voltage reference signal for generating a clock out signal having a particular frequency based upon the voltage level of the voltage reference signal. Wherein the frequency of the clock out signal remains substantially constant despite temperature, process and supply voltage variations in the semiconductor circuit.
Abstract:
A flash EPROM cell has a reduced cell size by providing vertical coupling between the floating gate and the bit line during programming. The erase operation is done by tunneling of electrons from the sharp tip of the Poly spacer to the control gate. The cell is adapted so that the source for each cell within the array is the source of an adjacent cell and the drain is the drain to another adjacent cell. The cell is formed by forming the drain regions into the substrate through openings in a first insulator that is preferably the field oxide. A second insulator is deposited over the first insulator, over the substrate and along the side walls of the openings and is preferably a thin layer so that the opening is covered with a thin insulating layer. The insulated opening is filled with a metal, preferably tungsten or a tungsten alloy. The field oxide is selectively removed. A gate oxide is grown and a first polysilicon layer is formed and then etched to form spacers along the edges of the metal/second insulator structure. The first polysilicon is selectively etched and a tunneling insulator layer is formed thereover. A second polysilicon layer is formed over the tunneling insulator.
Abstract:
A memory array is organized into rows and columns of resistive elements and is disclosed to include a resistive element to be read or to be written thereto. Further, a first access transistor is coupled to the resistive element and to a first source line and a second access transistor is coupled to the resistive element and to a second source line, the resistive element being coupled at one end to the first and second access transistors and at an opposite end to a bit line. The memory array further has other resistive elements that are each coupled to the bit line. The resistive element is written to while one or more of the other resistive elements are being read.
Abstract:
An embodiment of the present invention includes a non-volatile storage unit comprising a first and second N-diffusion well separated by a distance of P-substrate. A first isolation layer is formed upon the first and second N-diffusion wells and the P-substrate. A nano-pillar charge trap layer is formed upon the first isolation layer and includes conductive nano-pillars interspersed between non-conducting regions. The storage unit further includes a second isolation layer formed upon the nano-pillar charge trap layer; and at least one word line formed upon the second isolation layer and above a region of nano-pillar charge trap layer. The nano-pillar charge trap layer is operative to trap charge upon application of a threshold voltage. Subsequently, the charge trap layer may be read to determine any charge stored in the non-volatile storage unit, where presence or absence of stored charge in the charge trap layer corresponds to a bit value.
Abstract:
A memory array is organized into rows and columns of resistive elements and is disclosed to include a resistive element to be read or to be written thereto. Further, a first access transistor is coupled to the resistive element and to a first source line and a second access transistor is coupled to the resistive element and to a second source line, the resistive element being coupled at one end to the first and second access transistors and at an opposite end to a bit line. The memory array further has other resistive elements that are each coupled to the bit line. The resistive element is written to while one or more of the other resistive elements are being read.
Abstract:
Fabrication methods for MRAM are described wherein any re-deposited metal on the sidewalls of the memory element pillars is cleaned before the interconnection process is begun. In embodiments the pillars are first fabricated, then a dielectric material is deposited on the pillars over the re-deposited metal on the sidewalls. The dielectric material substantially covers any exposed metal and therefore reduces sources of re-deposition during subsequent etching. Etching is then performed to remove the dielectric material from the top electrode and the sidewalls of the pillars down to at least the bottom edge of the barrier. The result is that the previously re-deposited metal that could result in an electrical short on the sidewalls of the barrier is removed. Various embodiments of the invention include ways of enhancing or optimizing the process. The bitline interconnection process proceeds after the sidewalls have been etched clean as described.
Abstract:
A method of making a magnetic random access memory cell includes forming a magnetic tunnel junction (MTJ) on top of a wafer, depositing oxide on top of the MTJ, depositing a photo-resist layer on top of the oxide layer, forming a trench in the photo-resist layer and oxide layer where the trench has a width that is substantially the same as that of the MTJ. Then, the photo-resist layer is removed and a hard mask layer is deposited on top of the oxide layer in the trench and the wafer is planarized to remove the portion of the hard mask layer that is not in the trench to substantially level the top of oxide layer and the hard layer on the wafer. The remaining oxide layer is etched and the MTJ is etched to remove the portion of the MTJ which is not covered by the hard mask layer.