JP2012064950A - Add-on layer formation method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an add-on layer formation method which allows deposition and does not have such a defect that a hot operation which is not compatible with a low temperature treatment required for many semiconductor devices, is required.SOLUTION: An add-on layer formation method which allows deposition includes formation of a first semiconductor substrate removing layer, formation of many doping regions on the first semiconductor substrate on the removing layer, and here in formation of many doping layers, doping is made to provide a first conductive type. The method further includes formation of a first doping layer of the first semiconductor substrate on the removing layer, and doping is made to provide a second conductive type against the first conductive type. The method further includes formation of at least an intermediate doping layer at the first semiconductor substrate on the first doping layer, and formation of at least a third doping layer at the first semiconductor substrate on the intermediate doping layer, formation of a first conductive blanket layer on the third doping layer, formation of a second conductive blanket layer on the first conductive blanket layer, and attaching of the first semiconductor substrate to the second semiconductor substrate so that the second conductive blanket layer contacts to a conductive upper part layer corresponding to the second semiconductor substrate.

Description

The present invention relates generally to an add-on layer formation method for use in three-dimensional integrated circuit (IC) structures.

  As shown in FIG. 1, the conventional three-dimensional IC is referred to as a hybrid IC. Conventional hybrid IC implementation methods generally include: The provision of a first IC comprised of a base semiconductor substrate 201 and a dielectric layer 202, the provision of a second IC comprised of a base semiconductor substrate 203 and a dielectric layer 204, the deposition and coupling of these ICs or individual chips, and The implementation of deep vias 255 as shown in US Pat. No. 6,600,173 penetrating the semiconductor substrate or the provision of micro steps as shown in US Pat. No. 6,355,501.

  Still referring to FIG. 1, the devices of the semiconductor substrate 203 to be deposited are usually connected by interconnect lines of a dielectric layer 204 disposed on the substrate 203. Similarly, the interconnect lines of the dielectric layer 202 disposed below the deposition substrate 203 are used to connect to the base substrate 201 device. The apparatus of the deposition substrate 203 does not have a lower electrode, but rather has a contact from the upper side.

  Wafers for bonding in conventional practice require chip positioning marks. The wafer positioning marks are different from the positioning marks used in photographic processing. Therefore, the “hybrid IC technology” used in the three-dimensional IC is considered as another type of multi-chip package (MCP). The main purpose of "Hybrid IC technology" is to reduce the interconnects used in package interconnects to facilitate high speed device operation.

  Also with respect to FIG. 1, the implementation of the prior art 3D IC is generally characterized as follows: each IC layer is implemented by a separation process, and each IC becomes a 3D IC and is combined and deposited. Each IC layer has a semiconductor substrate (for example, 201 and 203), and holds devices (for example, 211 and 212), respectively. Also, in general, the device shares an electrical common substrate 241, 242 or well 243. Conventional implementations using SOI do not have an electrical common substrate, but such implementations have a physically common semiconductor substrate. Further, conventional deposition ICs have dielectric layers, interconnect lines and vias only on the device in each deposition IC.

  Another conventional attempt is formed by a molten polycrystalline or amorphous semiconductor layer that is placed on the dielectric layer using, for example, a laser. The device is then formed using a single crystal semiconductor layer formed from a polycrystalline or amorphous layer. In another conventional attempt, a single crystal epitaxial layer is grown on a dielectric layer having holes partially exposed therethrough to the underlying single crystal layer.

  However, both the laser recrystallization and epitaxial processes described above have the disadvantage of requiring high temperature operation that is incompatible with the low temperature processes required for many semiconductor devices, and further, single crystal semiconductor layers formed by this method Has many disadvantages and therefore these methods are not widely used.

  Briefly, many vertical semiconductor devices are added to a separately fabricated substrate that includes electrical devices and / or interconnects. Many vertical semiconductor devices are physically separated from each other and are therefore not located in the same semiconductor body or semiconductor substrate.

  In one aspect of the invention, many vertical semiconductor devices include thin doped semiconductor regions that are etched to produce a discrete doped stack structure following attachment to the substrate. It is added to the substrate made separately as a layer. In another embodiment of the present invention, many vertical semiconductor devices are fabricated prior to attachment to a separately fabricated substrate.

  In another aspect of the invention, the doped stack structure forms the base of a wide range of semiconductor devices including, but not limited to, diodes, capacitors, n-type MOSFETs, p-type MOSFETs, bipolar transistors, and floating gate transistors. To do.

  In another aspect of the invention, ferroelectric memory devices, ferromagnetic memory devices, chalcogenide phase change devices, and add-on layers that can be deposited for use in conjunction with separately fabricated substrates Formed.

  In yet a further aspect of the invention, the add-on layer that can be deposited includes a minimum of one electrical interconnect line.

It is sectional drawing of the deposition integrated circuit called the three-dimensional integrated circuit by a prior art. It is sectional drawing of the Example of this invention containing a base semiconductor substrate. It is sectional drawing of the Example of this invention without a base semiconductor. 2 illustrates a process flow for forming a three-dimensional IC structure using an SOI layer according to the present invention. 2 illustrates a process flow for forming a three-dimensional IC structure using an SOI layer according to the present invention. 2 illustrates a process flow for forming a three-dimensional IC structure using an SOI layer according to the present invention. 2 illustrates a process flow for forming a three-dimensional IC structure using an SOI layer according to the present invention. The lower part of the SOI layer is a cross-sectional view of an embodiment of the present invention having no directly connected electrodes. 1 is a cross-sectional view of an embodiment of the invention including a multi-deposited SOI layer. FIG. 3 is a cross-sectional view of an embodiment of the present invention having a horizontal rather than vertical device incorporated in an SOI layer. FIG. 3 is a cross-sectional view of an embodiment of the present invention having a horizontal rather than vertical device incorporated in an SOI layer. FIG. 4 illustrates the process flow according to the present invention to produce those devices with directly connected bottom electrodes and a vertical device incorporated in the SOI layer. FIG. 4 illustrates the process flow according to the present invention to produce those devices with directly connected bottom electrodes and a vertical device incorporated in the SOI layer. FIG. 4 illustrates the process flow according to the present invention to produce those devices with directly connected bottom electrodes and a vertical device incorporated in the SOI layer. 1 is a cross-sectional view of an embodiment of the present invention having a vertical device with a flat intermediate electrode. 1 is a cross-sectional view of an embodiment of the present invention having a vertical device with a spacer intermediate electrode. FIG. 4 is a cross-sectional view of an embodiment of the present invention having a vertical device with a spacer intermediate electrode extending to an adjacent dummy vertical device. 1 is a cross-sectional view of an embodiment of the present invention having a vertical device with an intermediate electrode extending to the top of the vertical device. Is a top view of the structure shown in FIG. 18. FIG. 3 is a cross-sectional view of an embodiment of the present invention having a vertical device with a flat etch stop layer for the top contact electrode. FIG. 6 is a cross-sectional view of an embodiment of the present invention having a vertical device with a spacer etch stop layer for the top contact electrode. 1 is a cross-sectional view of an embodiment of the present invention having a vertical pn junction diode. 1 is a cross-sectional view of an embodiment of the present invention having a vertical Schottky diode. 1 is a cross-sectional view of an embodiment of the invention having a vertical Schottky diode with a Schottky contact disposed on an intermediate electrode. FIG. 1 is a cross-sectional view of an embodiment of the present invention having a vertical capacitor formed by a depletion layer region. 1 is a cross-sectional view of an embodiment of the present invention having a vertical capacitor formed by a dielectric layer between electrodes. FIG. 1 is a cross-sectional view of an embodiment of the present invention having a vertical bipolar transistor. FIG. 6 is a top view of an embodiment of the invention having a vertical device for forming a MOSFET with 8 gates or variable gate width for multi-bit operation. FIG. 5 is a cross-sectional view of an embodiment of the present invention having a vertical device for forming MOSFETs with four different sized gates or variable gate widths for multi-bit operation. 1 is a cross-sectional view of an embodiment of the present invention having a vertical inverter. 2 illustrates the top and bottom layout of an SRAM cell according to the present invention. 2 illustrates the top and bottom layout of an SRAM cell according to the present invention. 1 is a cross-sectional view of an embodiment of the present invention having a vertical thyristor as part of an SRAM cell. FIG. 2 is a cross-sectional view of an embodiment of the present invention having both a thyristor and a MOSFET arranged perpendicular to the SOI layer and connected in series. 1 is a cross-sectional view of an embodiment of the present invention having a vertical DRAM cell with SOI layer depletion capacitor and MOSFET vertical connection. FIG. 1 is a cross-sectional view of an embodiment of the present invention having a vertical DRAM cell with a vertical connection of SOI layer dielectric capacitors and MOSFETs. FIG. 1 is a cross-sectional view of an embodiment of the present invention having a vertical non-volatile memory (NVM) cell including a floating gate and a control gate of an SOI layer. FIG. 3 is a cross-sectional view of an embodiment of the present invention having a vertical NVM including a floating gate partially covering the channel region of the SOI layer and a control gate covering the floating gate and the remainder of the channel region of the SOI layer. FIG. 3 is a cross-sectional view of an embodiment of the present invention having a vertical NVM cell with a floating gate, a control gate, and an erase gate in an SOI layer. 1 is a cross-sectional view of an embodiment of the present invention having a vertical NVM with an SOI-layer oxide-nitride-oxide (ONO) gate. FIG. 1 is a cross-sectional view of an embodiment of the present invention having a vertical NVM with a floating gate, a control gate, and a bulk contact all disposed in an SOI layer. FIG. 38 is a top view of the structure of FIG. 37 with eight gates in the SOI layer. 3 illustrates an SOI layer layout with many blocks including different types of devices. 3 illustrates an SOI layer layout with many blocks including different types of devices. 1 is a cross-sectional view of an embodiment of the present invention having a vertical NVM cell including a MOSFET connected in series with a capacitor using a ferroelectric material disposed together in an SOI layer. FIG. FIG. 46 illustrates a standard memory circuit formed with the device illustrated in FIG. FIG. 46 illustrates another standard memory circuit formed with the device illustrated in FIG. FIG. 46 is a cross-sectional view of an embodiment of the present invention having a vertical NVM cell including the structure illustrated in FIG. 48 illustrates a standard memory circuit formed from the structure shown in FIG. FIG. 3 is a cross-sectional view of an embodiment of the present invention that includes a vertical NVM with a MOSFET in the SOI layer and uses a ferroelectric material between the gate and the gate dielectric layer. FIG. 4 is a cross-sectional view of an embodiment of the present invention that includes a vertical NVM with a MOSFET in the SOI layer and uses a ferroelectric material between the floating gate and the control gate. FIG. 4 is a cross-sectional view of an embodiment of the present invention including a vertical NVM having a MOSFET in an SOI layer connected in series to a ferroelectric material below the MOSFET. FIG. 4 is a cross-sectional view of an embodiment of the present invention including a vertical NVM having a SOI layer MOSFET connected in series to a ferromagnetic material on top of the MOSFET. 1 is a cross-sectional view of an embodiment of the present invention including a vertical NVM cell having a MOSFET disposed in an SOI layer connected in series with a resistor formed of a chalcogenide material. FIG. FIG. 6 is a cross-sectional view of an embodiment of the present invention including a vertical NVM having MOSFETs arranged in an SOI layer connected in series in a structure that functions as either fusion or anti-fusion depending on the materials used. 1 is a cross-sectional view of an embodiment of the present invention including a vertical volatile memory cell having a MOSFET in an SOI layer without a body contact. FIG. 38 is a cross-sectional view of an embodiment of the present invention including a vertical NVM cell having a MOSFET connected in series with a floating gate transistor such as that shown in FIG. FIG. 58 is a schematic diagram of an equivalent circuit for the structure shown in FIG. 57 . 1 is a cross-sectional view of an embodiment of the present invention including a high voltage MOSFET disposed in an SOI layer. FIG. 2 is a cross-sectional view of an embodiment of the present invention including a high voltage MOSFET disposed in an SOI layer having a low doping channel region.

  A three-dimensional IC according to the present invention is shown in FIG. Embodiments of the present invention provide device integration technology.

  References herein to “one embodiment”, “an embodiment” or similar phrases include at least one specific feature, structure, operation or feature described in connection with the embodiment. Means that Thus, the appearances of such idioms or phrases herein are not necessarily all referring to the same embodiment. In addition, various special features, structures, acts, or characteristics may be combined in any suitable manner in one or more embodiments.

(Terminology)
“ASIC” means “application of a special integrated circuit”. “SoC” means “system on chip”, and “SoCs” means plural forms of SoC. The SoC may be an ASIC, but this is not necessarily so. The ASIC may be a SoC, but this is not necessarily so.

  As used herein, the expression “back bias” refers to a voltage applied to the substrate or body of a field effect transistor (FET). Back bias instead refers to substrate bias or reverse bias.

  The expression “conductive type” is well known in the semiconductor manufacturing industry. The conductivity type generally means n-type and p-type. The semiconductor region doped with donor-type impurities forms an n-type region. The semiconductor region doped with acceptor-type impurities forms a p-type region.

  The terms chip, semiconductor device, integrated circuit, LSI device, monolithic integrated circuit, ASIC, SoC, microelectronic device, and similar expressions are sometimes used interchangeably in this industry. Microelectronic devices are considered the broadest term, including others. For these microelectronic devices, signals are typically coupled between these and other circuit elements via physical conductive connections. A connection point is sometimes referred to as an input, output, terminal, line, pin, pad, port, interface, or similar variant or combination.

  As used herein, the term “device” refers to one or more circuit elements having different voltage characteristics. “Devices” include, but are not limited to, FET (n-type and p-type) diodes and varactors.

  The expression “vertical device” means a device having a direction relative to the base substrate such that the current flowing through the devices is substantially perpendicular to the base substrate.

  The FET used here means a metal oxide semiconductor field effect transistor (MOSFET). These transistors are also known as insulated gate field effect transistors (IGFETs). An FET is generally described as a three terminal device having a gate, source, and drain. The FET is further described as a four terminal device when considering the FET body.

  The source and drain terminals refer to the terminals of the FET, during which conduction occurs under the influence of the electric field following the reversal of the semiconductor surface under the influence of the electric field generated from the voltage applied to the gate terminal.

  The acronym “SOI” generally means “silicon on insulator”. As those skilled in the art recognize, the SOI layer can be formed in a variety of ways. Unless otherwise stated, “SOI layer” is used herein to refer to another prefabricated wafer or similar, such that a three-dimensional stack is formed from the SOI layer and a prefabricated wafer or similar type of substrate. Means a relatively thin single crystal portion of a semiconductor wafer that can be walled and bonded to a substrate of this type. In this context, an SOI layer can be considered as an attachment layer or a depositable add-on structure that itself contains minimum devices and / or interconnects and is suitable for bonding to a semiconductor substrate that already contains the devices and / or interconnects. . As a depositable add-on layer, the single crystal layer is doped to have one or more doped regions that are vertically adjacent to each other. For the purposes of this disclosure, doped regions include intrinsic regions as well as p-type and n-type regions. Individual semiconductor structures are formed by etching through a portion of the doped stack to electrically isolate the structures. The space between such individual structures is filled with a dielectric material so as to reform the layer without gaps or spaces therein, thereby providing mechanical stability, and an additional deposited layer Support.

As used herein, the expression three-dimensional IC includes a semiconductor substrate having a device and / or an interconnect structure made thereon, and also at least one SOI layer having the device and / or interconnect, wherein the semiconductor Substrate and SOI layer refer to a three-dimensional integrated circuit that is deposited and bonded together.

  The disclosures of US Pat. No. 6,600,173, US Pat. No. 5,563,084, and US Pat. No. 6,355,501 show the formation of a three-dimensional IC as a compression technique involving individually operating ICs to be deposited. However, embodiments of the present invention do not use individually operating ICs, rather, using junction SOI technology and a thin single crystal semiconductor layer 124 without device formation prior to layer transition as shown in FIG. Provide assembly technology. Since the single crystal semiconductor layer 124 is formed by SOI technology, it simply means SOI here.

The meaning of the thin film including the semiconductor layer 104 of FIG. 3 is a thin film single crystal including a diffusion layer such as p-type, n-type or i (intrinsic) type, and a physically distinguishable layer such as a dielectric layer or a metal layer. It is a semiconductor layer. The meaning of the “non-multi-device formation” semiconductor layer 124 in FIG. 5 is that the thin SOI layer 124 does not have an insulating structure, a metal pattern, a multi-device interconnection, or an isolation impurity region for individual devices. is there.

  As shown in FIG. 2, devices 111, 112, 113 according to the present invention are separated by isolation 135 and have a floating structure in filled dielectric material 133. As used herein, devices 111, 112, 113 refer to floating devices (FLD). Such a floating device instead refers to a doped stack structure or a vertical semiconductor device.

  The embodiment of the present invention is different from the conventional junction IC having the common well 142 or the substrate 143 in which the electrical common region is disposed. Furthermore, embodiments of the present invention do not have the physical support layer found in conventional SOIIC substrates where all devices of one IC layer are supported by the substrate below the bottom oxide. In the illustrated embodiment of the invention, the second IC layer including FLDs 112 and 113 is referred to as FLDIC layer 102 and the first IC layer including FLD 113 is referred to as FLDIC layer 101. An intermediate dielectric (ILD) layer disposed above and below the FLDIC layer 101 has interconnect lines 132 and vias or contacts 131. Interconnect lines 132 and vias 131 connect the FLD directly or indirectly within the FLDIC layer, or connect the device from one FLDIC layer to another FLDIC layer or to the base substrate 103.

  As shown in FIG. 2, the substrate bonded to the lower dielectric layer 151 is called the base substrate 103, the first FLDIC layer on the base substrate 103 is called the first FLD layer 101, and the next FLDIC layer is the first FLDIC layer. Called the two FLDIC layers 102.

  FIG. 2 shows multiple FLDIC layers 101 and 102 and one base semiconductor substrate 103. Dashed line 134 shows the boundary or interface between the two ILD layers. The first ILD layer 151 has interconnect lines and vias, and the base semiconductor substrate 103 shares these interconnect lines and vias with the first FLDIC layer 101. Some of the vias are connected directly from the first ILD layer 151 to the second FLDIC layer 102. The interconnect lines and bias of the second ILD layer 152 are also shared by the first and second FLDIC layers 101, 102. This sharing of interconnect lines and vias is an advantage of embodiments of the present invention.

  Also, with respect to FIG. 2, the electrode formation for the floating devices 111, 112, 113 is by direct connection to the upper and lower portions of each floating device 111, 112, 113. Also, in various embodiments of the present invention, a floating device having a vertically separated single or multiple intermediate electrodes 123 is constructed. These electrodes can be connected to interconnect lines in the ILD layer that are located above and / or below the FLD layer.

  In FIG. 2, if a logic IC is implemented on the base semiconductor substrate 103, a memory device is implemented on the first FLDIC layer 101, and an image detector is implemented on the second FLDIC layer 102, one semiconductor substrate is Different types of individually optimized devices can be integrated without using difficult and expensive SoC structures or semiconductor processing.

  FIG. 3 shows a three-dimensional IC structure having an FLDIC layer without a mounting base substrate. In one way to obtain the structure of FIG. 3, first the ILD layer 153 is placed on top of the base substrate, then the SOI layer 124 is formed on the ILD layer 153, and then the device uses the SOI layer 124. Then, another ILD layer 154 including interconnect lines 132 and vias 131 is placed on top of the single crystal semiconductor device 104, and then the base substrate is removed from the ILD layer 153. A base substrate (not shown) is a flat substrate with a flat surface, such as plastic, ceramic, glass, metal, or semiconductor material. The base substrate should be able to withstand processing temperatures in the range of 250-650 ° C., which is considered a “non-high temperature semiconductor processing temperature”.

  Also with reference to FIG. 3, an embodiment of the present invention has a pad connected to a package (not shown) disposed below the first IDL layer 153 and / or above the second ILD layer 154. The lower pad 146 can be connected to the package using, for example, solder. The upper pad 145 can be connected to the package using, for example, a wire. Such a pad structure according to the invention reduces the mold area and integration of the package.

  Various embodiments of the present invention do not require a physical support substrate for the floating device. Various embodiments exist with interconnect lines, vias and FLDs without a base substrate.

  Various embodiments of the present invention provide a floating device separated by a dielectric isolation region. These electrically isolated structures do not have parasitic devices commonly found in prior art attempts.

  Various embodiments of the present invention provide floating devices that are directly or indirectly connected.

  In various embodiments of the present invention, combining an SOI layer with a semiconductor substrate does not require the same type of wafer positioning structure as used in photolithography processes; rather, the wafer positioning structure is a wafer positioning mark or step type. Implemented as a positioning structure. Alternatively, without a wafer positioning structure, the SOI layer can be transferred with a simple notch positioning because it does not have a multi-device structure such as an isolation structure or interconnect line in which the transfer SOI layer is horizontally divided. The transition SOI layer has only a few layers divided vertically.

  Various embodiments of the present invention provide interconnection of floating devices both above and below the FLD.

  The prior art generally uses a horizontal MOSFET. In the case of conventional vertical MOSFETs, the implementation of contacts and interconnections is difficult and the process is not compatible between vertical and horizontal MOSFETs. However, embodiments of the present invention can easily implement vertical devices including MOSFETs, and it is easier to implement interconnects and contacts with lower contact resistance than previous attempts.

  In order to implement the logic device in a conventional manner, the individual devices need to be connected. However, in some embodiments of the present invention, the FLD logic device uses discrete devices that are vertically connected without interconnect lines, as embodiments of the present invention include the form of SOI devices and no wells are required. Can be formed.

  In various embodiments of the invention, the FLD can have metal electrodes in direct contact at the top, bottom, and middle regions. The metal electrode region is the same as the upper and lower dimensions of the FLD single crystal semiconductor formed by isolation etching. Therefore, the voltage drop of the device can be reduced.

4-7 describe the process flow for producing a three-dimensional IC like that shown in FIG. After formation of mask positioning marks (not shown) on the substrate 180, a single or multiple ILD layer 133 is formed in the dielectric layer 153 and the interconnect lines 132 and vias 131 on which the conductive material is formed. Here, the boundary line of each ILD layer 133 made of the dielectric layer 153 is shown as a broken line 134. Base substrate 180 should withstand semiconductor processing temperatures in the range of 250 ° C to 650 ° C. Conductors of low electrical resistance materials are formed that carry voltage / current and are metals such as aluminum and copper, refractory metals, silicon, or highly doped low resistance polycrystalline / amorphous semiconductor materials. Once the via 131 connected directly or indirectly to the FLD is formed in the dielectric layer 153, a metal layer used as the bottom electrode 121 of the FLD is placed and, if necessary, another intermediate junction layer 120. A metal layer can be implemented. The metal used for the intermediate bonding layer 120 generally has a lower melting point than the metal layer on the dielectric layer 153. The metal 120 is required to have desirable reflow characteristics at a low temperature for surface flattening in order to prevent a gap due to surface fine roughness in the SOI substrate 190 bonding process. If interconnect line 132 of dielectric layer 153 is aluminum, the metal of intermediate bonding layer 120 should have a melting point in the range of 250 ° C. to 650 ° C. below the melting point of aluminum, 660 ° C. Here are the table and melting point of the metal that can be an intermediate bonding layer, aluminum alloy 204 ° C to 674 ° C, zinc 420 ° C, zinc alloy 377 ° C to 484 ° C, lead 328 ° C, thallium 304 ° C, tellurium 445 ° C, solder 268 ° C. ˜579 ° C. and tin alloy 223 ° C. to 422 ° C.

FIG. 5 shows the SOI substrate 190. In order to perform FLD, a doping layer is formed in the single crystal semiconductor layer 124, where the doping layer includes, but is not limited to, impurity implantation during ion implantation or epitaxial layer growth to form the single crystal semiconductor layer 124. It is formed by an appropriate method. The metal layer 121 is formed on the single crystal semiconductor layer 124, and the intermediate bonding layer 120 is formed on the metal layer 121. In the standard embodiment of the present invention, the metal layer 121 and the bonding layer 120 are formed as a blanket layer formed on the entire surface. The SOI substrate 190 is a single crystal semiconductor substrate and is a material for the FLD single crystal semiconductor layer 124. The SOI substrate 190 is a single source semiconductor such as silicon and germanium or a composite semiconductor such as SiGe, GaAs, GaP, and InP. The SOI substrate 190 is a combination of a single source semiconductor and a compound semiconductor. Before bonding, it is better that the SOI substrate has an intermediate bonding layer having a high reflow rate at a low temperature melting point in order to remove surface roughness.

  The SOI substrate 190 is porous to a desired depth using, for example, Smart Cut (US Pat. No. 5,882,987), ELTRAN (US Pat. No. 5,371,037) or SiGen technology, or It has a removal layer 191 which is a deformation layer. The removal layer 191 is a defect region in the semiconductor lattice, and after bonding with the dielectric layer 153, the SOI substrate 190 is removed except for the single crystal layer 124 that forms the FLD.

6 is a cross-sectional view of the junction of the SOI substrate of the dielectric layer 153 and 5 of FIG. SOI substrate 190 shown in FIG. 5 is vertically reversed, and is bonded onto the dielectric layer 153 of FIG. During the bonding process, pressure is applied with the heat treatment to increase the bonding strength and to remove the gaps between the bonding contact surfaces. Alternatively, eutectic bonding with gold or thermocompression bonding with a soft metal thin film can be used as the intermediate layer bonding process. The metal layer 121 including the intermediate bonding layer 120 used in the bonding process is used as a lower electrode of the FLD.

With reference to FIG. 7 , the SOI substrate 190 is removed after removing the SOI layer 124 on the dielectric layer 153 where the FLD is mounted, and then the isolation structure 135 is implemented on the entire surface or part of the FLD layer. The SOI substrate 190 is removed using the removal layer 191 and a wafer jet is used for SOI substrate removal. Without the removal layer 191, the SOI layer 124 is removed where the floating device is implemented using a bond etchback method (US Pat. No. 5,013,681). This SOI layer is also referred to as an add-on layer that can be deposited. The SOI substrate 190 can be removed by etching or polishing. The handling substrate is also used to bond the SOI substrate 190 to the handling substrate, the SOI substrate 190 is removed from the handling substrate, and the SOI layer 124 is removed, and then the SOI layer is transferred from the handling substrate to the dielectric layer. The handling substrate is the same type of substrate used for the base substrate. The handling substrate uses a vacuum to temporarily hold the single crystal semiconductor layer from the SOI substrate, and then the transfer of the SOI layer to the dielectric layer can be easily performed by releasing the vacuum. It is better that the vacuum surface has a thick dielectric layer that protects the SOI layer. The role of the handling substrate is to transfer the SOI layer from the SOI substrate to the dielectric layer without damage. Also, as described in US Pat. No. 6,355,501, the SOI substrate and the handling substrate can be bonded using polyamide. Once the SOI substrate is separated, chemical mechanical polishing (CMP) is used to reduce the surface roughness of the transferred SOI layer.

Once the single crystal semiconductor layer (i.e., SOI layer) is transferred, isolation is performed to create an individual floating device. Trench technology is used for isolation formation. At the same time, the SOI layer above the top of the marking line will be removed to facilitate subsequent die cutting operations. The lower electrode 121 of the FLD is automatically implemented during the trench isolation process. The method of forming the lower electrode 121 is discussed below in connection with FIG. 12-14. Once the SOI layer has been transferred, the mask positioning marks are exposed on the base substrate 180 by removing the SOI layer on the mask positioning marks, and then using the exposed mask positioning marks, the FLD pattern on the mask. And 131 allow the pattern on the dielectric layer to be positioned. A vertical FLD in which current flows in a vertical direction has an intermediate electrode. Interconnect lines and contacts that connect to the upper electrodes and interconnect lines are formed by conventional semiconductor processing methods.

Using the method described above, multiple FLDs can be deposited, thus increasing IC integration. Various embodiments of the present invention therefore do not require wafer or chip positioning marks or fine steps for wafer positioning when bonding SOI substrates with single crystal semiconductor layers and dielectric layers with interconnect lines and vias. Various embodiments of the present invention can be implemented with mask positioning marks used in conventional photographic processing. The isolation structure is filled with a dielectric material and an intermediate electrode material. The method of forming the dielectric and the intermediate electrode in the isolation structure is illustrated in Figure 15-18. After the process of FIGS. 4-7, in accordance with conventional semiconductor processing, ILD, interconnect lines and vias is performed, and the base substrate is removed, then it becomes a structure shown in FIG.

4 to 7 , the n + layer on the top of the SOI substrate 190 is directly connected to the metal layer 121. However, as seen in FIG. 8 , another dielectric 189 is formed between the top of the SOI substrate 190 and the metal layer 121 and then transferred to the dielectric layer 153. In this case, the lower electrode 121 can be used as a gate electrode having a gate dielectric 189. Another lower electrode 121c is used to connect the lower electrode of the FLD.

The FLDIC layer has one or more SOI layers. The FLDIC layer shown in FIGS. 3 and 5 has a single SOI layer. FIG. 9 shows multi-SOI layers 124 and 128 constituting one FLDIC layer 105. As shown in FIG. 9, there are no vias between the SOI layer 124 and the SOI layer 128. If vias exist between multiple SOI layers as shown in FIG. 2, one FLDIC layer 101 and another FLDIC layer 102 can be separated and separated. Multiple SOI layers are implemented by sequentially adding another SOI layer 128 on the already transferred SOI layer 124. The multi-SOI layers 124 and 128 shown in FIG. 9 have a dielectric layer 138 that electrically isolates the multi-SOI layers 124 and 128. Thus, the multiple SOI layers 124, 128 have different types of devices that are electrically isolated at each SOI layer. For example, one SOI layer has a p-type MOSFET and the other SOI layer becomes a memory device.

FLD is a conventional semiconductor device. MOSFETs, bipolar transistors, diodes, capacitors and resistors, image detectors (eg charge coupled devices (CCD) or active pixel detectors (APS)), or microelectromechanical systems (MEMS). The FLD has a circular column shape (see FIG. 28 ), a rectangular column (see FIG. 29 ), a polygonal column, or a cylindrical column. If the width of the FLD is narrowed, the column structure aspect ratio increases and is removed from the wobbled or bonded dielectric layer. In order to prevent such a phenomenon, the FLD is a trapezoidal shape having a narrow upper portion and a wide lower portion.

  The FLD can be divided into a high temperature apparatus (HT) and a low temperature apparatus (LT) according to the temperature used in the manufacturing process. Similarly, FLD can be divided into vertical device (V) and horizontal device (H) according to the device operating direction, where “V” and “H” mean the current direction of main device is “vertical” and “horizontal” direction. To do.

The FLD processing temperature can be divided into a high temperature above 800 ° C. and a low temperature below 650 ° C. In this disclosure we will refer to equipment manufactured with high temperature processing as HT-FLD and equipment manufactured with low temperature processing as LT-FLD, or simply as FLD, because the advantages of this invention are low processing temperatures. This is because 3D IC is implemented. HF-FLD can be processed at high temperatures for thermal activation of implanted ions and becomes a vertical or horizontal device. To implement HT-FLD, the interconnect 132 and via 131 of the dielectric layer 153 shown in FIGS. 4-7 should be refractory metal such as copper or tantalum, molybdenum or tungsten. The base substrate used in HF-FLD should withstand 800 ° C. or higher.

  LT-FLD or FLD does not require ion implantation, heat treatment for ion implantation, and photographic processing because an impurity layer necessary for device operation is formed on the SOI substrate before transfer to the dielectric layer. If high temperatures are required during the FLD process, the characteristics of the devices present on the other layers will change. Equipment process control based on early prediction of changes is very difficult. Accordingly, embodiments of the present invention can be implemented on top of a base semiconductor substrate having a device with no process changes. An advantage of the present invention is that a low cost process is obtained because ion implantation and photographic processing are not required. Also, various embodiments of the present invention do not require high-temperature processing, refractory metals, and aluminum, have a low melting point, and can use aluminum widely used in semiconductors. Also, LT-FLD can use metal gates and high-k dielectric materials more easily than conventional manufacturing processes.

The standard form of LT-FLD is VFLD (vertical FLD) because the vertical impurity junction is already formed on the SOI substrate and the bottom electrode is easy to implement. However, at low temperatures, HFLD (horizontal FLD) can be performed without ion implantation. HFLD (horizontal FLD) is in the form of a MESFET, MOSFET, diode or horizontal bipolar transistor. As shown in FIG. 10 , once a portion of the SOI layer is etched using PR (photoresist) or hard mask 171, FIG. 10 is formed. FIG. 11 shows a MESFET type HFLD having a metal gate forming a Schottky diode. The lower electrode 121 can be used as a gate electrode. FIG. 11 shows the FLD 113 shown in FIG. If gate 172 has a dielectric layer below it, the FLD becomes a MOSFET. If gate 172 in FIG. 11 has a resistive contact and the n-type region is switched to the p-type region, it becomes a horizontal bipolar transistor. If the n + region is the anode and the p-type region is the cathode, it becomes a horizontal diode from a horizontal bipolar transistor. Also, without a gate, the FLD can become a resistor using only the n-type region.

  VFLD (or LT-FLD) takes the form of MESFETs, MOSFETs, diodes, capacitors, resistors, bipolars, thyristors, or different types of vertical connections of FLD devices that implement circuits on behalf of a single device. The combination of the optimum horizontal device for the base semiconductor substrate and the optimum VFLD and SoC is optimal in terms of performance and price.

  Unlike the prior art vertical device 212 shown in FIG. 1, it is easy to implement and connect electrodes, contacts, and interconnect lines to the VFLD. In this disclosure, “electrode” means an electrical component that is directly connected to a device or gate with a gate dielectric material. “Contact” means a connecting piece between a normally vertical electrode and an interconnect line. The vertical devices in US Pat. No. 5,414,288, US Pat. No. 6,027,975, US Pat. No. 6,337,247 and US Pat. No. 6,449,186 provide space for source / drain and contact formation. It should have a horizontally extending doping region that is used for this purpose. Thus, in the prior art, the extended source / drain region increases resistance and parasitic capacitors. As shown in FIG. 2, the electrodes are formed on the upper part 122 and the lower part 121 of the FLD 111. In the case of VFLD, the intermediate electrode 123 is connected to the interconnect line at the top and bottom of the FLD. Furthermore, the intermediate electrode is used as a partial interconnect. This very flexible interconnection mechanism for 3D ICs is not possible with conventional attempts to form 3D ICs.

  The lower part of the FLD is connected to the metal layer 121 that is also directly connected to the via 131 of the ILD layer 151. Therefore, the lower part of the FLD already has pre-formed electrodes and contacts. Since the lower electrode 121 of the FLD 111 is connected via the via 131 of the dielectric layer 151, they need to be positioned. The positioning mechanism used in this technique is performed by a conventional photographic positioning mark (no display). However, photographic processing has a misalignment margin, and the lower electrode 121 and via 131 should be positioned within the alignment margin. In general, as shown in FIG. 2, interconnect lines are connected through vias 131 at different levels in the ILD layer, so the width of the interconnect lines needs to be wider than the dimensions of the vias 131. Photoprocessing by photographic mask and etching is necessary to form interconnect lines 132 and vias 131.

12 and 13 , the formation of the bottom electrode 121 used for positioning between the lower portion 124z of the FLD and the via 131 uses a self-positioning technique and therefore does not require a photographic mask mold for processing. Part of the metal layer used in the SOI substrate bonding process is an extension of the lower part of the FLD, and the other part of the metal layer becomes the lower electrode 121. Using etching mask 173 as shown in FIG. 12 , the portions of layers 122 and 124, which are indicated by broken lines, are etched away. FIG. 13 shows a spacer type etching mask that allows the lower electrode 121 to be wider than the via 131. If the etching mask is placed higher than the FLD and etched by, for example, a dry etching process, the width of the lower electrode 121 is, for example, more than twice the height of the FLD. The width of the lower electrode 121 is controlled by the thickness of the hard mask, the height of the FLD, the width of the FLD, and the etching amount of the spacer 182. If the width of the FLD is larger than the photographic processing margin, the wider lower electrode 121 is not necessary.

15-18 , the intermediate electrode according to the present invention can be implemented as follows. First, there is a flat intermediate electrode method or a flat electrode method. After the CMP operation for electrode material positioned and flattened, dry etching is performed to provide a planar electrode 123 shown in FIG. 15. The pattern formation of the flat electrode 123 is performed before and after the dry etching process. The installed electrode material is thicker than the height of the VFLD. Also at this point in the process, the etch stop layer 122 is necessary on top of the FLD to prevent damage to the SOI layer 124. The etch stop 122 is typically a combination of a multi-oxide, nitride or metal layer. In FIG. 15 , the dielectric material 133a is placed in a manner similar to the formation of the flat electrode 123, planarized and dry etched. The dielectric material 133 a reduces the parasitic capacitance between the lower electrode 121 and the flat electrode 123.

The second is a method of using the spacer 123 as shown in FIG. 16. If the spacer intermediate electrode or the spacer electrode is wide, it is easy to obtain electrical contact with the spacer electrode. However, it is difficult to achieve high integration. If the width is narrow, the spacer method does not require photographic or CPM processing.

The third method is a spacer method using a dummy FLD (that is, an FLD that does not operate as a device). As shown in FIG. 17 , the dummy FLD 124 a is installed close to the FLD and increases the width of the spacer used for the intermediate electrode 123. Since the contact 123a connected to the intermediate electrode 123 is installed on the upper part of the dummy FLD 124a, a margin for forming the contact increases. As shown in FIG. 17 , the space between the FLD and the dummy FLD 124a should be less than twice the thickness of the spacer film.

A fourth method is a thin spacers method of extending the intermediate electrode 123 to the top of the FLD, as shown in FIG. 18. Covering the contact forming region on the intermediate, and the remaining area is etched, after the placement of the intermediate electrode material, we obtain a structure in FIG. 18. This method is good for thin spacer thicknesses. A thick dielectric layer is used on the top electrode to reduce the parasitic capacitance between the top electrode and that in the middle.

  The intermediate electrode surrounds all or part of the intermediate region of the VFLD. The multi-intermediate electrode can be formed in one FLD.

After the SOI layer is transferred from the SOI substrate, the electrode material 122 is disposed to the SOI layer, and after applying the pattern to the FLD, the upper electrode is carried out as shown in FIG. 12. If the size of the contact 122a is smaller than the dimension of the upper electrode 122, a conventional semiconductor photo / etching techniques may be used as shown in FIG. 18. However, if the FLD width is less than the photographic process deviation margin for forming the contact 122a, or if the size of the contact 122a is larger than the FLD region, the photo / etch process for the contact 122a will cause a short to the intermediate electrode. . This disclosure thus describes several structures according to the present invention that increase the processing error margin for photo / etching during contact 122a formation. The first is to increase the thickness of the upper electrode forming material in order to provide an etching process margin. In Second, as shown in FIG. 20 planarization technique is to use an etch stop layer 184. Third, using an etch stop layer 184 with spacer technology, as shown in FIG. 21 , where the etch stop layer 184 has a slower etch rate than the dielectric layer 133c during contact 122a formation. For example, if the dielectric layer 133c is an oxide film, the etch stop layer 184 is a nitride.

In this disclosure we annotate VFLD which is performed at low temperature and has vertical operation as follows. That is, MOSFET VMFLD, MESFET VMEFLD, diode VDFLD, resistor VRFLD, capacitor VCFLD, bipolar VBFLD, and thyristor VTFLD.

VDFLD is implemented as a vertical p-n or p-i-n junction diode as shown in FIG. 22. FIG. 23 also shows a vertical Schottky diode having a vertical Schottky junction with a Schottky junction between the upper electrode 122 and the SOI 124. As shown in FIG. 24 , the metal intermediate electrode 123 is used for a three-dimensional Schottky diode. The VDFLD shown in FIG. 22 has a capacity that is twice that of the current driving capacity of FIG. 23 because the current travels from the anode of the intermediate electrode 123 to the cathodes of the upper and lower electrodes.

There are two types of VCFLD. One is a MOS capacitor type or a depletion layer capacitor using a depletion layer region formed in a single crystal semiconductor, and the other or a dielectric capacitor accumulates electric charge on a dielectric contact surface without a depletion layer region. If the semiconductor doping concentration is low, a depletion layer is present in the semiconductor region. If the doping concentration is high, it exists in the semiconductor region. If the doping concentration is high, it becomes a dielectric capacitor, VCFLD without a depletion layer. VCFLD are shown in Figures 25 and 26. In FIG. 25 , there is a gate dielectric semiconductor surrounding the n-type single crystal and an electrode connecting the n-type semiconductor. Since the total capacitance is generally proportional to the electrode area, the surrounding gate 123b increases the total capacitance of the VCFLD. Without a gate dielectric layer, the metal gate forming the Schottky diode can be used as a capacitor with reverse bias.

If the semiconductor has a pillar structure, the capacitance of the VCFLD increases due to the increased semiconductor and gate contact area. Also, as shown in FIG. 26 , the gates 123b, 123c and the gate dielectric layer are repeatedly deposited on the VCFLD, and the deposition capacitor and VCFLD are connected in parallel and then the capacitance is increased. This type of capacitor has the same structure as the deposited capacitor used in DRAM. A contact 121a in FIG. 26 connects the gate of the deposited capacitor and the lower electrode 121.

Bipolar VBFLD shown in FIG 7. Impurity regions composed of collectors 124c, 124d, base 124b, and emitter 124a are implemented in the SOI substrate and then transferred. Electrodes constituting the emitter 124a and the collector 124d are formed in the lower part 121 and the upper part 122, and the base 124b electrode 123 is formed in the middle of the FLD. Even though the emitter 124a is located above or below the VBFLD, the emitter is at the bottom of the VBFLD in the illustrated embodiment. In this case, the emitter is implemented on the top of the SOI substrate before the single crystal semiconductors 124a to 124d are transferred. Thus, precise junction control is possible when the emitter 124a and base 124b regions are formed. SiGe heterojunction bases are also possible, and polycrystalline semiconductors can be used as part of the emitter region. Furthermore, since the emitter is disposed below the VBFLD, the emitter can avoid thickness variation during the planarization process after the SOI layer transfer process. If the handling substrate is used for SOI layer transition, the emitter is placed on top of the FLD.

According to the present invention, to obtain a low collector series resistance, the VBFLD does not require a buried layer and a highly doped collector region connecting the collector contact and the buried layer. Various embodiments of the present invention provide a lower collector series resistance compared to prior attempts. In addition, the base series resistance can be lowered without the high doping intrinsic base region because the surrounding base electrode 123 formed in the middle of the VBFLD has a wide contact surface in the base region. Furthermore, VBFLD does not have a parasitic capacitor that prevents high speed operation. Further, since VBFLD does not have a substrate, the parasitic bipolar transistor of the base collector substrate does not exist in the embodiment of the present invention. While conventional implementations require deep and shallow trench isolation, VBFLD only requires one isolation structure 135. In FIG. 25 , if the base intermediate electrode 123 extends from the base region to the collector region, the lightly doped collector region 124C forms a Schottky diode with a base electrode that allows high speed operation of the VBFLD.

MOSFET type VMFLD is shown in FIGS. 15-18 and 20-21. The vertical MOSFET has a high degree of integration in a small space. The channel length of the MOSFET according to the invention is not limited by photography and etching process limitations, but rather is determined by the thickness of the doping layer. Further, VMFLD can have a high driving current because the channel width can be easily increased by the surrounding gate as compared with the conventional structure having the same channel length.

  However, prior art vertical MOSFETs are not used due to many drawbacks. The vertical transistors shown in US Pat. No. 5,414,288 and US Pat. No. 6,027,975 are formed by epitaxial growth of exposed single crystal regions. This technique is not good for low temperature semiconductor processing because it requires difficult manufacturing techniques and high temperature operation for epitaxial growth.

  US Pat. No. 6,337,247, and US Pat. No. 6,449,186, columnar surrounding gate transistors (SGTs) are difficult to coexist with optimal horizontal devices, and shadow effects during ion implantation by columnar transistors Wake up. SGT also does not have a high integration density because it has problems with electrode formation in the source / drain and gate regions. Therefore, these attempts are not suitable for SoC formation.

  The VMFLD has a directly connected bottom electrode that reduces current reduction due to voltage drop and parasitic resistance. The VMFLD can easily be in a complete or partial depletion layer mode by controlling the FLD width, where the depletion layer mode can also be controlled by the operating voltage and the gate dielectric constant. Because the surface removed from the SOI substrate becomes a highly doped source / drain region, unlike the prior art horizontal devices, there are few gate oxide quality, device operation, and manufacturing, even with small surface defects. There is no effect.

  The VMFLD has a gradient of impurity distribution in the channel region, and an electric field is formed in the channel region due to stepped impurities, where the induced electric field accelerates the flow of current, and stepped impurities become short channel effects (SEC). ). Stepped impurities cannot be easily formed by ion implantation or epitaxial processing. Increasing the impurity concentration in the channel region from the source side to the drain side performs an asymmetric operation. Furthermore, LDD (low doping drain) can be selectively formed only on the drain side. Due to the difficulties in high tilt ion implantation and device placement, it is difficult to implement a stepped channel of a horizontal MOSFET in the prior art.

A MOSFET type VFLD or VMFLD has a gate dielectric layer implemented below 650 ° C. as shown in US Pat. No. 5,330,935 and US Pat. No. 5,443,863. The dielectric layer is a thermal oxide, a deposited oxide, an oxynitride, or a combination of oxides and nitrides such as ONO and NO (nitrogen oxide). A suitable dielectric material can be used except for a high-temperature processing film requiring 650 ° C. or more. Another advantage of embodiments of the invention includes, but is not limited to, Al ₂ O ₃ , ZrO ₂ , HfO ₂ , Y ₂ O ₃ , La ₂ O ₃ , Ta ₂ O ₅ , TiO ₂ and BST in the gate dielectric layer. It is easy to use such a high dielectric constant (high k). In conventional fabrication of MOSFETs, high temperature thermal activation operation is necessary after source / drain ion implantation. At this point, the properties of the high-k material can be changed. However, VMFLD processing does not require high temperature processing, so high k materials can be used in a stable state. Also, if ALD (atomic layer arrangement) is used to provide the gate dielectric layer, a substantially uniform layer is obtained.

  According to the present invention, the threshold voltage can be controlled by changing the thickness of the gate dielectric and / or the width of the FLD. If different gate dielectric thicknesses are used or different dielectric constant materials are used in VMFLD, many operating voltages and threshold voltages can be implemented in the same SOI layer, which is valid for SoC. Also, since VMFLD is manufactured at low temperature and an enclosing gate is used, it is easier to use a metal gate compared to prior art manufacturing attempts.

In the prior art, for digital applications, a MOSFET is in either an “off” or “on” state depending on the voltage or current state. The VMFLD shown in FIG. 28 is a multi-level (ML) VMFLD in which many gates share one source / drain and have many states. The current driving capability of VMFLD is proportional to the gate region. Thus, simply multiple gates of the same gate size are used for current ramping. Also, many gates with the same gate dimensions of VMFLD are used for ML-VMFLD. FIG. 29 shows an ML-VMFLD having two “W” size gates and two “3W” size gates, where “W” is a constant and “3W” means three times the “W” value. To do. Using these four different gate combinations, the ML-VMFLD can have nine different current values from “0” to “8”. If the gate of the same dimensions is used for ML-VMFLD, 8 pieces of gates are required nine different values as shown in FIG. 28. ML-FLD is used for application to storage devices or digital logic devices. The multilevel intermediate electrode is used as a base electrode in bipolar transistors.

Three-dimensional ICs including FLDs not only have the form of a single device, such as a MOSFET or bipolar transistor, but also have many devices formed in a single FLD layer. FIG. 30 shows a single inverter type VFLD. Configure the inverter. The p-MOSFET and n-MOSFET do not require different wells, so this inverter has a high degree of integration. A contact 123f that connects the gates of the p-MOSFET and the n-MOSFET together serves as an input terminal of the inverter. The drains of the P-MOSFET and n-MOSFET are connected together and connected to electrode 123g and contact 123h. In FIG. 30 , the p + -p-p + type p-MOSFET is a depletion layer mode MOSFET. The p-MOSFET is of the p + -n-p + type and in this case the n region requires a reference voltage. As shown in FIG. 30 , the contacts through the dielectric layer used in the FLD isolation structure are connected to interconnect lines above or below the FLD layer.

In addition to the FLD inverter shown in FIG. 30 that uses the SOI layer of only one, it is possible to FLD inverter implemented using two SOI layer as shown in FIG. In this case, one SOI layer has an n-MOSFET and the other SOI layer has a p-MOSFET.

  According to the present invention, the memory device can be implemented using many FLDs.

Using two inverters and two pass transistors on the base semiconductor substrate, six-transistor SRAM cell can be implemented as shown in FIG. 32. The two inverters are VFLD and the two transistors with word lines and bit lines are on the base semiconductor substrate. 31 and 32 show the interconnect lines for the upper and lower contacts, respectively. Two FLD inverters are latched with the input of each inverter connected to the output. The counterpart of one inverter contact 122a, 123h, 123f, 131 is indicated by underlined 122a, 123h, 123f, 131. Therefore, such an SRAM cell has a high degree of integration. In particular, since the base semiconductor substrate does not require an n-well for the p-MOSFET in the SRAM cell, the degree of integration of the base semiconductor substrate is high. If 4 pass transistors are used, a 2-port SRAM can be implemented.

  There are many ways to implement the SRAM according to the invention. The first method is by using four n-MOSFETs and two p-MOSFET type FLDs on a base semiconductor substrate. The second method is based on the arrangement of two p-MOSFETs and four n-MOSFET type FLDs on the base semiconductor substrate. The third method is by placing two p-MOSFET type FLDs on the SOI layer and four n-MOSFET type FLDs on another SOI substrate. The fourth method has either four n-MOSFET type FLDs or four n-MOSFETs on a base semiconductor substrate. With the use of a 4-transistor SRAM cell and a resistor is formed on the FLD layer, or a polycrystalline semiconductor resistor is used.

Prior art SRAM cells using thyristors have a complex structure with vertical thyristors and horizontal MOSFETs on the same semiconductor substrate. This SRAM is therefore process incompatible with other devices and it is not good for SoC applications. FIG. 33 shows a VTFLDSRAM cell having a gate 123j according to the present invention. Intermediate electrode gate 123j is used for word line 2 and the upper electrode is connected to the reference voltage. The VTFLD is connected to a horizontal access transistor on the base semiconductor substrate so that each device is optimized and high integration is provided for SoC applications. The gate of the access transistor is used for the word line. FIG. 34 shows an SRAM cell having another structure shown in FIG. 33 , which vertically connects the access transistor 161c and the thyristor, and finally forms a VFLDSRAM cell. The thyristor gate 123j and the access transistor gate 123i are all intermediate electrodes. 33 is the same as the SRAM cell shown in US Pat. No. 6,225,165B and US Pat. No. 6,172,899. A dynamic random access memory (DRAM) cell according to the present invention has one transistor and one capacitor, where the transistor is a VMFLD on the base semiconductor substrate or on the FLDIC layer, and the floating source of the transistor is on another FLDIC layer. Connected to the VCFLD. VCFLD are shown in Figure 25-26. In addition, one SOI layer having a transistor and another SOI layer having a capacitor are connected from a multi-SOI layer constituting one FLDIC layer to form a DRAM structure. Another VFLD DRAM structure has a series connection of transistors and capacitors in the SOI layer. FIG. 35 shows a DRAM structure having an n-type MOSFET and a depletion layer capacitor connected in series therewith. The upper electrode 122 is connected to the bit line, and the intermediate electrode 123 is connected to the word line. In Figure 35, a depletion layer region formed between the floating n + source 124e and p region 124f is connected to the lower electrode 121 has a width greater than n-type MOSFET, a wider semiconductor region herein 12-14 Can be carried out using spacer technology without additional photographic processing operations as shown in FIG. In FIG. 36 , MOSFETs having a floating source and a dielectric capacitor are connected in parallel, where the floating source p region is connected to a reference voltage source (not shown). In FIG. 36 , the lower electrode 121 is connected to the bit line, and the intermediate electrode 123 is connected to the word line.

Showing a nonvolatile FLD memory structure according to the invention. FIG 37-42. FIG. 37 has two gates, where one floating gate surrounds the p-type channel region with a gate dielectric layer 183b, and the control gate 123 connected to the via has a floating gate 123k with another gate dielectric layer 183c. surround. FIG. 38 shows a split gate non-volatile memory device where the floating gate 123k includes a portion of the p-type channel region, and the remainder of the channel region and the floating gate 123k are surrounded by the control gate 123. FIG. FIG. 39 has three gates: a floating gate 123k, a control gate 123, and an erase gate 323 designed for erasing data. FIG. 40 shows a floating gateless non-volatile memory device VFLD with an ONO gate dielectric layer 183, where information is stored at different locations 30 according to current flow. FIG. 41 shows a flash memory FLD structure having a bulk contact 122 c on a p-type bulk region 124. The VMFLD has a bulk contact without a gate dielectric layer on one side and a gate contact with a gate dielectric layer on the other side.

One of the advantages of the embodiment of the present invention is ML-VMFLD in which the non-volatile memory stores multi-bit information in the FLD. As shown in FIG. 42 , an FLD having one source / drain has eight isolation gates, and then one FLD has eight multi-bit memory cells. FIG. 42 has a bulk contact 122c and the remaining contacts are connected to the source / drain of the SOI layer forming the FLD. In FIG. 41 , the dotted line “756” indicates the boundary line of the FLD bulk region exposed from the upper FLD. The remaining contacts 122a on the SOI region 124 are connected to the source / drain. If the non-volatile memory, source and drain of FIGS. 37 to 18c have different doping concentrations, a multi-bit non-volatile memory can be achieved by device operation similar to ETOX.

  In one embodiment of the invention, the FLD memory device has redundancy on the same or different FLDIC layers.

One embodiment of the present invention has a block area in the FLDIC, as shown in FIGS. 43-44 , where each block has a different type of FLD. 43 and 44 are top views of the FLDIC, and each chip 441 is divided by a marking line. For example, one FLDIC layer has four blocks 413a-413d, where the first block has a programmable FPGA, the second block has a flash memory, the third block has a bipolar device, and The fourth block has SRAM. Each block requires a different impurity junction for different device types, where the impurity junction should be formed prior to the SOI layer transfer process in the case of LT-FLD. In order to form the block FLD, wafer positioning marks are required on the SOI substrate and the base substrate. In this case, it is better to have an overlay error correction area (OECA) 412 to account for wafer misalignment, where OECA has a gap of a few microns to a few hundred microns.

FIG. 45 shows a nonvolatile memory cell 700 having a capacitor using a ferroelectric film 710 and a VMFLD connected in series to the capacitor. Nonvolatile memory using the ferroelectric film 700 is called FRAM (ferroelectric random access memory). Conventional ferroelectrics are (PbZr) TiO ₃ (referred to as PZT) and SrBi ₂ Ta ₂ O ₉ (referred to as SBT). And YMnO ₃ . If an electric field is applied to such a ferroelectric, the ferroelectric has polarization properties. In FIG. 45 , the FRAM cell 700 directly connects a ferroelectric capacitor and VMFLD. The VMFLD gate 123 is the word line (WL), the drain is the bit line (BL), the source is connected to the ferroelectric capacitor, and the other electrode 122a is connected to the drive line (DL or plate line). Is done.

FIG. 46 shows an equivalent circuit for FRAM memory cell 700, where the logic device for sense amplifier 770 is typically implemented on a base substrate, and FRAM cell 700 including VMFLD is implemented in the SOI layer.

FIG. 47 shows one memory bit using the two FRAM cells shown in FIG. 45 , where the logic device for the sense amplifier 700 is generally implemented on the base substrate 103, and the FRAM cell 700 containing VMFLD is in the SOI layer. Will be implemented.

FIG. 48 shows a non-volatile memory cell 730 having a capacitor using a ferroelectric film 710 and a VMFLD connected in parallel to the capacitor. Parallel connected FRAMs operate at higher speeds and have lower power consumption compared to series connected FRAM cells. One intermediate electrode 123 is WL. Another intermediate electrode 123a has an applied reference voltage and maintains a constant current state for parallel connection of the ferroelectric capacitor and VFLD.

FIG. 49 is an equivalent circuit of the FRAM cell 730. FRAM cells 730 are concatenated to form a byte.

Capacitor using a ferroelectric film 710 is disposed on top of VFLD in FIGS. 45 and 46. However, a capacitor using the ferroelectric film 710 can be installed under the VFLD. VFLD may also be a MOSFET, bipolar, or other type of transistor,

50 and 51 show a non-volatile VMFLD 750 having a ferroelectric film 710 as part of a VFLD structure. 50 , the FRAM has a ferroelectric film 710 set between a gate dielectric layer 183 and a gate electrode 123. This is called metal ferroelectric insulating silicon (MFIS). Gate dielectric layer 183 is a standard MOSFET gate dielectric layer and can be formed of, for example, silicon dioxide or nitric oxide compounds. If there is no gate dielectric layer 183 in FIG. 50 and ferroelectric film 710 is used as the gate dielectric layer, the device is a metal ferroelectric silicon (MFMIS) type FRAM.

In FIG. 51 , a ferroelectric film 710 is disposed between the floating gate 123k and the control gate 123 of the VMFLD, which forms a metal ferroelectric metal-insulated silicon (MFMIS) type FRAM 760FLD.

Ferroelectric layer 710 used in the illustrated embodiment of FIGS. 45-51 should be carried out at below 660 ° C. for low temperature FLD.

FIG. 52 shows a non-volatile MRAM magnetoresistive random access memory 800 cell structure using VMFLD and series-connected MJT magnetic tunnel junction stack 810. In FIG. 52 , the MJT 810 is formed and installed under the ILD 133 and the FLD 124.

FIG. 53 also shows an MRAM cell 850 using the MJT 810. The MJT 810 is formed and installed above the FLD 124.

The MJT 810 has various electric resistance characteristics depending on the applied magnetic field, and the electric resistance changes depending on the polarization of the MJT 810. The MJT 810 is composed of many thin film layers. In general, one magnetic file is a free layer that is polarized by an applied magnetic field. The other magnetic film is a pinned layer and is used with an exchange layer that is generally an antiferromagnetic layer. The pinned layer is polarized by the applied magnetic field. The film stack is therefore called a magnetic tunnel junction stack (MJT). The MJT is not limited to a structure having two magnetic films and a dielectric film. The MJT has a combination of different thin layers. The MJT is classified into two types according to the deposited layer, one is a giant magnetoresistance (GMR) using a non-magnetic material, and the other is a tunnel magnetoresistance (TMR) using a dielectric layer such as an oxide layer. It is. The VFLD shown in FIGS. 52 and 53 is a MOSFET, bipolar, or MESFET.

  FIG. 54 shows an Ovonic integrated memory (OUM) 900 cell structure using a phase change film (RSPCF) 910 with an inverted structure and a series-connected VFLD. 54, the RSPCF 910 is implemented after the FLD 124 is formed and is disposed above the FLD 124. The RSPCF 910 is performed before the formation of the FLD 124 and is installed below the FLD 124 (no display). The RSPCF 910 has an amorphous or polycrystalline layer depending on the amount of current and time, in other words, the temperature applied to the RSPCF, where the polycrystalline has a low electrical resistance.

RSPCF910 is an alloy of chalcogenide and the VI element of the periodic table. Therefore, RSPCF 910 is Ge-Sb-Te, GaSb, InSb, InSe, Sb ₂ Te ₃ , GeTe, Ge ₂ Sb ₂ Te ₅ , InSbTe, GaSeTe, SbSb ₂ Te ₄ , InSbGe, AgInSbTe, (GeSnTe, TeS ) Or an alloy of Te ₈₁ Ge ₁₅ Sb ₂ S ₂ . The electrode 910a connected to the RSPCF 910 is TiAIN or TiW which is stable at 650 ° C. 54 is a MOSFET, a bipolar device or a MESFET.

  FIG. 55 shows a programmable read-only memory (PROM) cell structure using a fused or antifused layer 310 and series connected VFLD. In FIG. 55, the fused or antifused layer 310 is formed on the FLD 124 after the FLD is formed. Also, the fused or antifused layer 310 is formed under the FLD 124 before the FLD is formed. The anti-fusion layer 310 has a high electrical resistance. However, if a high programming voltage / current is applied to the antifusion layer, it will have a low electrical resistance. PROMs are generally not reprogrammable.

  PROMs are used in application-specific integrated circuit (ASIC) field programmable gate arrays (FPGA), or programmable logic arrays (PLA).

  The anti-fusion layer is formed from an ONO layer, a metal oxide layer, a chalcogenide layer, or an undoped amorphous silicon layer, but is not limited to these materials. The fusion layer is formed from nichrome or polycrystalline silicon, but is not limited to these materials. The electrodes 301a and 301b for fusion or anti-fusion are made of TiW that is stable at high temperatures.

  In FIG. 55, VFLD is a MOSFET, bipolar transistor, MESFET, or diode.

  FIG. 56 shows a DRAM 400 cell structure with VFLD 124 only. The VMFLD 124 using the SOI layer has a floating body p region as shown in FIG. 56 without an applied bias, and charges are accumulated in the floating body for a short time (ie, update time). The charge becomes readable or writable data.

FIG. 57 shows an electrically erasable programmable read only memory (EEPROM) 500 cell structure with VMFLD and non-volatile VFLD memory connected in series. The nonvolatile memory has two gates, a floating gate and a control gate. However, this is a silicon oxide nitrogen oxide (SONOS) type nonvolatile memory. A MOSFET connected to the selection line as shown in FIG. 57 is arranged on the nonvolatile memory. However, the position of these devices can be reversed.

FIG. 58 is an equivalent circuit of one EEPROM cell.

FIG. 59 shows an output VMFLD 600 operating at a high voltage. Compared to conventional low power VMFLD, the output VMFLD has an SOI layer thickness in the range of several μm to several hundred μm, and the thickness of the gate dielectric layer has a range of 1/10 nm to several thousand nm. The operating voltage ranges from 7V to 1000V. The FLD also has a trapezoid that helps extend the depletion region and reduce the electric field, thus increasing the operating voltage.

  The power VMFLD 600 has many advantages over horizontal MOSFETs. Conventional horizontal MOSFETs need to have a long channel length in order to increase the operating voltage. However, it is expensive due to the low integration. However, since the channel length of the power VMFLD 600 is determined by the vertical height of the SOI layer, the integration degree is not changed. Also, because the power VMFLD has a surrounding gate, it has a low resistance and its current drive capability is more than twice that of a conventional horizontal MOSFET. Thus, the power VMFLD of FIG. 59 replaces other conventional power devices such as outer double diffused MOS (LDMOS) and trench MOS. Also, by combining the low voltage device of the base substrate and the power VMFLD into one chip, smart power or smart MOS that handles analog and digital signals with one chip can be achieved.

If lever power VMFLD in FIG. 59 has a double diffused drain, it becomes an apparatus as shown in FIG. 60. The double diffusion region prevents depletion region expansion to the highly doped drain region and allows the device to operate at high voltages.

59 and 60 show that if the gate dielectric layer has a combination of low temperature thermal oxide, high k dielectric and CVD dielectric, the reliability of the device is increased, and between the semiconductor 124 and the gate dielectric layer 183. Contact surface traps are reduced. Also, current drive capability increases and on-resistance decreases.

  It is to be understood that the invention is not limited to the embodiments described above, but encompasses any and all embodiments within the scope of the additional claims.

101, 102, 105: FLDIC layer 103: base substrate 104: single crystal semiconductor devices 111, 112, 113: FLD
120: Intermediate bonding layer 121, 122: Lower electrode 122, 122a: Upper electrode 123: Intermediate electrode 124, 128: SOI layers 123a, 123f, 123h, 131: Contacts 123a, 123f, 123h, 131: Contacts (the above-mentioned counterparts)
123b: Etching stop 123c: Gate 123j: Thyristor electrode 123i: Transistor electrode 124a: Emitter 124b: Base 124c, 124d: Collector 124e: Floating source 131: Contact / bias 132: Interconnect line 133, 133a: Dielectric material 134: Boundary 135 : Isolation 138, 133c: dielectric layer 142: common well 143: substrate 145: upper pad 146: lower pads 151, 152, 153: ILD layer 161c: transistor 171: hard mask 172: gate 173: etching mask 180: base substrate 182 : Spacers 183, 183b, 183c, 189: Dielectric layer 184: Etching stop layer 190: SOI substrate 191: Removal layer 201, 203: Base semiconductor substrates 202, 204, 241, 24 : Dielectric layer 211: retention device 243: well 255: via 301a, 301b: melting / anti molten electrode 310: melt / anti molten layer 323: the erase gate 400: DRAM
412: Correction area (OECA)
413a, 413b, 413c, 413d: Block FLDIC layer 441: Chip 500: EEPROM
600: VMFLD
700, 710: Ferroelectric film 730: FRAM cell 760: Insulator silicon type FRAM (MFMIS)
770: Sense amplifier 800: Cell structure 810: MJT
850: MRAM cell 910: RSPCF

Claims (7)

  Formation of the removal layer of the first semiconductor substrate, formation of many doping regions in the first semiconductor substrate above the removal layer, wherein formation of many doping layers is doped and removed to have the first conductivity type Forming a first doping layer of a first semiconductor substrate over the layer, doped to have a second conductivity type relative to a first conductivity type, and forming a minimum intermediate doping layer on the first semiconductor substrate over the first doping layer , And formation of at least a third doping layer on the first semiconductor substrate on the intermediate doping layer, formation of a first conductive blanket layer on the third doping layer, and second conductivity on the first conductive blanket layer Formation of the blanket layer and attachment of the first semiconductor substrate to the second semiconductor substrate such that the second conductive blanket layer contacts the corresponding conductive upper layer of the second semiconductor substrate Consisting depositable add - one layer forming method. The method of claim 1 , further comprising removing a portion of the first substrate on the side of the removal layer opposite the side on which the first doping layer is disposed such that the first doping layer is exposed. The method of claim 2 , further comprising forming a third conductive blanket layer on the exposed portion of the first doping layer. Etch a portion of the third conductive blanket layer, thereby exposing the surface of many underlying doped regions, and exposing portions of the many doped layers that are not masked by unetched portions of the third conductive blanket layer 4. The method of claim 3 , further comprising: etching a plurality of individually doped stack structures to expose portions of the first conductive blanket layer that are not masked by the individually doped stack structures. Individually such that the first layer of dielectric material surrounds the third doped stack layer and a portion of the intermediate doping layer of each individually doped stack, and a minimum portion of the intermediate doping layer is exposed. Disposing a first layer of dielectric material between the doped stack structures to form at least one conductive line electrically connected to at least one exposed intermediate doping layer of many individually doped stack structures; The method of claim 4 further comprising: 6. The method of claim 5 , further comprising disposing a second layer of dielectric material on the first layer of dielectric material. 5. The method of claim 4 , further comprising forming a gate dielectric layer around a minimum portion of at least one doped stack structure and forming a gate electrode adjacent to the minimum portion of the gate dielectric layer.