Abstract: In a power converter, an input power source (98) is intermittently coupled to provide a current flow (Icoil) in consecutive cycles (T) to generate an output voltage (Vo). The coupling durations (Tp) are adjusted in conjunction with a cycle skip count (112CNT) which is the count of cycles in which no coupling occurs. In some embodiments, the adjustments are performed to keep the coupling durations near the maximum efficiency range (between TLOW and THIGH), and the skip count is adjusted at the same time to obtain the desired output voltage in the presence of load current variations. The coupling frequencies are kept in a desired range to avoid interference with other circuit elements.

Abstract: To fabricate a semiconductor memory, one or more pairs of first structures are formed over a semiconductor substrate. Each first structure comprises (a) a plurality of floating gates of memory cells and (b) a first conductive line providing control gates for the memory cells. The control gates overlie the floating gates. Each pair of the first structures corresponds to a plurality of doped regions each of which provides a source/drain region to a memory cell having the floating and control gates in one or the structure and a source/drain region to a memory cell having floating and control gates in the other one of the structures. For each pair, a second conductive line is formed whose bottom surface extends between the two structures and physically contacts the corresponding first doped regions. In some embodiments, the first doped regions are separated by insulation trenches. The second conductive line may form a conductive plug at least partially filling the region between the two first structures.

Abstract: To fabricate a semiconductor memory, one or more pairs of first structures are formed over a semiconductor substrate. Each first structure comprises (a) a plurality of floating gates of memory cells and (b) a first conductive line providing control gates for the memory cells. The control gates overlie the floating gates. Each pair of the first structures corresponds to a plurality of doped regions each of which provides a source/drain region to a memory cell having the floating and control gates in one or the structure and a source/drain region to a memory cell having floating and control gates in the other one of the structures. For each pair, a second conductive line is formed whose bottom surface extends between the two structures and physically contacts the corresponding first doped regions. In some embodiments, the first doped regions are separated by insulation trenches. The second conductive line may form a conductive plug at least partially filling the region between the two first structures.

Abstract: Semiconductor dies are bonded to contact pads formed in a substrate's cavity. Vias through the substrate open into the cavity. Conductive lines passing through the vias connect the contact pads in the cavity to contact pads on another side of the substrate. A passage in the substrate opens into the cavity and provides an escape or pressure relief path for material filling the cavity. The passage can also be used to introduce material into the cavity.

Abstract: Nonvolatile memory wordlines (160) are formed as sidewall spacers on sidewalls of row structures (280). Each row structure may contain floating and control gates (120, 140), or some other elements. Pedestals (340) are formed adjacent to the row structures before the conductive layer (160) for the wordlines is deposited. The pedestals are formed in the area of the contact openings (330.1) that will be etched in an overlying dielectric (310) to form contacts to the wordlines. The pedestals raise the top surface of the wordline layer near the contact openings, so the contact opening etch can be made shorter. The pedestals also increase the minimum thickness of the wordline layer near the contact openings, so the loss of the wordline layer during the etch of the contact openings becomes less critical, and the photolithographic tolerances required for patterning the contact openings can be relaxed. The pedestals can be dummy structures (they may have no electrical functionality).

Abstract: A photoresist layer is exposed two or more times. At least one exposure is conducted through a regular mask, and at least one exposure through a modified mask with a clear region overlapping the position of a non-clear region of the first mask. The radiation dose used with the modified mask is insufficient by itself to create a resist pattern on the substrate. The exposure through the modified mask alleviates the resist underexposure in concave corners of the opaque pattern of the regular mask. Instead of the modified mask, an exposure without a mask can be performed.

Abstract: To reduce the edge roll off in a semiconductor wafering process, the wafer (110) is subject to a plasma etch with an edge underetch. The edge underetch is achieved by means of a wafer holder (410) that emits gas towards the wafer (e.g. a gas vortex) to draw the wafer towards the holder's body (460). The plasma impinges on the wafer surface (110.1) opposite to the body. Some of the gas emitted by the holder wraps around the wafer edge and dilutes the etchant near the wafer edge. Consequently, the etch proceeds slower near the edge (the edge is underetched). In some embodiments, the wafer is rotated around an axis (440) passing through the wafer to increase the underetch.

Abstract: A nonvolatile memory array can be erased by sectors, or alternatively a chip erase operation can be performed to erase all the cells of the array in parallel.

Abstract: A photoresist layer is exposed two or more times. One exposure is conducted through a regular mask, and one exposure through a modified mask with a non-clear region extending beyond a convex boundary of the non-clear region of the regular mask. The exposure through the modified mask allows one to reduce the exposure dose used with the regular mask, and thus alleviates the resist overexposure near convex areas of the non-clear pattern of the regular mask. Other embodiments are also provided.

Abstract: Semiconductor dies are bonded to contact pads formed in a substrate's cavity. Vias through the substrate open into the cavity. Conductive lines passing through the vias connect the contact pads in the cavity to contact pads on another side of the substrate. A passage in the substrate opens into the cavity and provides an escape or pressure relief path for material filling the cavity. The passage can also be used to introduce material into the cavity.

Abstract: An article which is being processed with plasma is moved during plasma processing so that the motion of the article comprises at least a first rotational motion, a second rotational motion, and a third rotational motion which occur simultaneously. The apparatus that moves the article comprises a first arm rotatable around a first axis, a second arm rotatably attached to the first arm and rotating the article around a second axis, and a rotational mechanism for inducing a rotational motion of the article in addition to, and simultaneously with, the rotation of the first and second arms.

Abstract: A silicon nitride layer (120) is formed over a semiconductor substrate (104) and patterned to define isolation trenches (130). The trenches are filled with dielectric (210). The nitride layer is removed to expose sidewalls of the trench dielectric (210). The dielectric is etched to recess the sidewalls away from the active areas (132). Then a conductive layer (410) is deposited to form floating gates for nonvolatile memory cells. The recessed portions of the dielectric sidewalls allow the floating gates to be wider at the top. The gate coupling ratio is increased as a result. Other features are also provided.

Abstract: To fabricate back side contact pads that are suitable for use in a vertical integrated circuit, vias are made in the face side of a wafer, and dielectric and contact pad metal are deposited into the vias. Then the wafer back side is etched until the metal is exposed. When the etch exposes the insulator at the via bottoms, the insulator is etched slower than the wafer material (e.g. silicon). Therefore, when the dielectric is etched off and the metal is exposed, the dielectric protrudes down from the wafer back side around the exposed metal contact pads, by about 8 &mgr;m in some embodiments. The protruding dielectric portions improve insulation between the wafer and the contact pads when the contact pads are soldered to an underlying circuit. In some embodiments, before the contact pads are soldered, additional dielectric is grown on the wafer back side without covering the contact pads.

Abstract: A clock distribution network (110) is formed on a semiconductor interposer (320) which is a semiconductor integrated circuit. An input terminal (120) of the clock distribution network is formed on one side of the interposer, and output terminals (130) of the clock distribution network are formed on the opposite side of the interposer. The interposer has a through hole (360), and the clock distribution network includes a conductive feature going through the through hole. The side of the interposer which has the output terminals (130) is bonded to a second integrated circuit (310) containing circuitry clocked by the clock distribution network. The other side of the interposer is bonded to a third integrated circuit or a wiring substrate (330). The interposer contains a ground structure, or ground structures (390, 510), that shield circuitry from the clock distribution network. Conductive lines (150) in an integrated circuit are formed in trenches (610) in a semiconductor substrate.

Abstract: A data synchronizer (210) transfers data from a data sending circuit (120) to a data receiving circuit (130). The data sending circuit is synchronous with a first clock (SCLK), and a data receiving circuit is synchronous with a second clock (RCLK). The two clocks have equal frequencies but may be out of phase. The synchronizer includes a circular FIFO. The FIFO entries (FF0-FF3) are written synchronously with the first clock (SCLK). The entries' outputs are connected to a multiplexer (230) whose select signals (RSEL0-RSEL3) are generated synchronously with the second clock (RCLK). Multiple entries make their data items available to the multiplexer at the same time. The sender (120) writes a data item and a data valid flag to the FIFO in each cycle of the first clock. The receiver (130) reads the FIFO in each cycle of the second clock.

Abstract: A memory performs a hidden refresh only at the end of a read operation or when the memory is disabled but is supposed to retain data in the disabled state. When the memory is in the enabled state, the refresh is not performed at the end of any operation other than read. This is done to ensure that execution of any memory access command will not be delayed by a refresh as long as the user follows certain timing rules. Other embodiments arc also provided.

Abstract: In some embodiments, a fabrication method comprises: forming a structure that has one or more substrates, wherein the one or more substrates are either a single substrate or a plurality of substrates bonded together, wherein the structure comprises a non-electronically-functioning component which includes at least a portion of the one or more substrates and/or is attached to the one or more substrates; wherein the one or more substrates include a first substrate which has: a first side, an opening in the first side, and a conductor in the opening; wherein the method comprises removing material from the structure so that the conductor becomes exposed on a second side of the first substrate. In some embodiments, the second side is a backside of the first substrate, and the exposed conductor provides backside contact pads.

Abstract: In fabrication of integrated circuits, trenches (184) are formed in a dielectric (170), then a metal (e.g. tungsten or copper) is deposited. The metal (194) is removed from the top surface of the dielectric by a polishing process (e.g. CMP). The metal remains in the trenches. The inventor has discovered that the erosion of the structure in the polishing process does not strongly depend on the size of the structure. Therefore, the erosion of a large structure (440) can be estimated by measuring the erosion of a smaller test structure (450). The erosion of the test structure is measured by a probe instrument (230), e.g. a stylus profilometer or a scanning probe microscope. Use of the test structure reduces the probability of damaging the larger structure by the probe. Other embodiments are also provided.

Abstract: Circuit elements (e.g. transistor gates) formed over a semiconductor substrate are protected by adjacent dummy structures during mechanical or chemical mechanical polishing of an overlying dielectric.

Abstract: A first level packaging wafer is made of a semiconductor or insulating material. The bumps on the wafer are made using vertical integration technology, without solder or electroplating. More particularly, vias are etched part way into a first surface of the substrate. Metal is deposited into the vias. Then the substrate is blanket-etched from the back side until the metal is exposed and protrudes from the vias to form suitable bumps. Dicing methods and vertical integration methods are also provided. Solder or electroplating are used in some embodiments.