JP2015195403A - Field effect transistor (fet) and method for fabricating the same - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a suite of novel structures and methods to reduce power consumption in a wide array of electronic devices and systems.SOLUTION: Some of these structures and methods can be implemented largely by reusing existing bulk CMOS process flows and manufacturing technology, allowing the semiconductor industry as well as the broader electronics industry to avoid a costly and risky switch to alternative technologies. Some of the structures and methods relate to a Deeply Depleted Channel (DDC) design, allowing CMOS based devices to have a reduced sigma VT compared to conventional bulk CMOS and can allow the threshold voltage VT of FETs having dopants in the channel region to be set much more precisely. The DDC design also can have a strong body effect compared to conventional bulk CMOS transistors, which can allow for significant dynamic control of power consumption in DDC transistors.

Description

  The present invention relates to electronic devices and systems, and methods for making and using the same.

  Electronic devices are an integral part of everyday life like never before. Systems such as personal computers and mobile phones have fundamentally changed shape depending on how we work, play and communicate. Over the years, new devices such as digital music players, electronic book (e-book) readers and tablets have been introduced, and improvements to existing product families have been introduced. These new devices represent an ever-increasing innovation that continues to change depending on how we live.

  The growing importance of electronic systems to the global economy and modern culture has been realized to a large extent by the semiconductor industry's adherence to Moore's Law. Moore's Law is named after Gordon Moore, the founder of Intel, who first noticed this phenomenon, and can be manufactured in the same area on an integrated circuit (or chip) without cost. It shows that the number of active transistors increases steadily over time. Some industry experts quantify this law, for example, stating that the number of transistors in the same area is roughly doubled approximately every two years. Without the increased functionality presented by Moore's Law and the associated cost and size reduction, many electronic systems that are widely available today have not been put into practical use or have been affordable. Let's go.

  For some time, the semiconductor industry has succeeded in adhering to Moore's Law by fabricating circuits in a chip using bulk CMOS technology. Bulk CMOS technology is particularly “scalable” because bulk CMOS transistors can be made smaller and smaller while optimizing and reusing existing manufacturing processes and equipment to maintain acceptable manufacturing costs. I know that. Historically, as the size of bulk CMOS transistors has been reduced, their power consumption has been reduced, helping the industry to achieve increased transistor density at reduced costs while adhering to Moore's Law . Thus, the semiconductor industry has been able to scale the power consumption of bulk CMOS transistors even at their size, reducing the cost of operating the transistors and the systems to which they belong.

  In recent years, however, it has become increasingly difficult to reduce the power consumption of bulk CMOS transistors while reducing their size. Transistor power consumption directly affects chip power consumption, which in turn affects the cost of operating the system and, in some cases, the utility of the system. For example, if the number of transistors in the same chip area is doubled while the power consumption per transistor is the same or increased, the power consumption of the chip is more than doubled. This is due in part to the cooling of the resulting chip, which requires additional energy. As a result, this makes the energy cost imposed on the end user operating the chip greater than twice. Such increased power consumption can also significantly reduce the usefulness of consumer electronics, for example by reducing battery life of mobile devices. There are other effects, such as increased heat generation and the need for heat dissipation, which can potentially reduce system reliability and adversely affect the environment.

Among semiconductor engineers, the continuous reduction in bulk CMOS power consumption is not feasible, partly because transistor operating voltage V _DD can no longer be reduced as transistor size shrinks. The perception of is spreading. The CMOS transistor is either on or off. The state of the CMOS transistor is determined by the value of the voltage applied to the transistor gate relative to the transistor threshold voltage V _T. When a transistor is switched on, it consumes dynamic power that can be represented by the equation: P _dynamic = CV _DD ² f.

Where V _DD is the operating voltage supplied to the transistor, C is the load capacitance of the transistor when the transistor is switched on, and f is the frequency at which the transistor is operated. When a transistor is switched off, it consumes static power that can be represented by the formula: P _static = I _OFF V _DD . Where I _OFF is the leakage current when the transistor is switched off. Historically, the industry has reduced transistor power consumption primarily by reducing the operating voltage V _DD , which reduces both dynamic and static power.

Whether the operating voltage V _DD can be reduced depends in part on the ability to accurately set the threshold voltage V _T , which is, for example, random dopant fluctuations (Random Dopant Fluctuation) as transistor dimensions shrink. It becomes increasingly difficult due to various factors including RDF). In a transistor manufactured using a bulk CMOS process, the main parameter that sets the threshold voltage V _T is the amount of dopant in the channel. Other factors affecting the V _T is, halo (halo) injection, which is the source and drain extensions and other factors. In theory, this is the same transistors on the same chip can be carried out accurately so as to have the same V _T, in fact, the threshold voltage can vary significantly. This means that the transistors may not switch all at the same time and some may not switch on in response to the same gate voltage. In the transistor having a channel length of less 100 nm, RDF is typically a major determinant of called _{V T} variation sigma _{V T} or [sigma] v _T, the magnitude of [sigma] v _T caused by RDF, the channel length It only increases as it shortens. Previously among semiconductor engineers, as shown in Figure 1, based on information provided by Intel, approximate experimental data, and keynote presentation at IEEE International Solid-State Circuits Conference 2009 by Kiyoo Itoh (Hitachi) According to the above, σV _T that increases in the nanoscale bulk CMOS provides 1.0 V as the practical lower limit of the future operating voltage V _DD . V _DD is shown as a downward slope function and has an industry goal of reducing to the target area. However, as the device shape (feature) size is shortened, the curve of σV _T increases, and in reality, V _min is increased by RDF. The power function of dynamic and static power is Power = CV _DD ² f + IV _DD . Therefore, the total power increases.

For these and other reasons, engineers semiconductor industry, widely in future process nodes, despite technique for suppressing [sigma] v _T are many known in short channel devices, the must abandon the bulk CMOS I believe. For example, one conventional approach to suppress σV _T in bulk CMOS works to achieve a non-uniform doping profile that increases the dopant concentration in the channel as it goes down in the vertical direction (from the gate to the substrate). It is something to be made. This type of retrograde (retrograde) doping profile reduces sensitivity to doping variations, but increases sensitivity to short channel effects that adversely affect device operation. Because of the short channel effect, these doping parameters are generally not scalable in nanoscale devices, making this approach generally unsuitable for use with nanoscale short channel transistors. As technology moves towards short channel devices formed with 45 nm or even 22 nm process nodes, it is recognized that the benefits of the retrograde approach in such devices will be limited.

Semiconductor engineers who are committed to overcoming these technical obstacles will also use super steep retrograde wells (SSRW) to solve the performance problems associated with scaling to the nanoscale range. I have tried. Similar to retrograde doping of nanometer scale devices, SSRW technology uses a special doping profile to form a heavily doped layer under a lightly doped channel. The SSRW profile differs from retrograde doping in that it has a very steep increase in doping level to reduce channel doping to the lowest possible level. Such a steep dopant profile can result in suppression of short channel effects, increased mobility in the channel region, and less parasitic capacitance. However, it is very difficult to achieve such a structure when these devices are manufactured for large quantities of nanoscale integrated circuits. This difficulty is due in part to out-diffusion of retrograde wells and SSRW dopant species into the channel region, particularly in p-well devices such as NMOS transistors. The use of SSRW is not intended to exclude the problems associated with random dopant density fluctuations may increase to unacceptable levels [sigma] v _T.

In addition to these and other attempts to address the shortcomings of existing bulk CMOS implementations, the industry has focused on CMOS transistors that have no dopant in the channel. Such transistor structures include, for example, fully depleted silicon-on-insulator (SOI) and various FINFET or omega gate devices. An SOI device typically has a group of transistors defined in a thin top silicon layer separated from a silicon substrate by a thin insulating layer of glass or silicon dioxide known as a buried oxide (BOX) layer. FINFET devices use multiple gates to control the electric field in the silicon channel. They can have a suppressed σV _T by having a low concentration of dopant in the silicon channel. This makes it insignificant for atomic level variations in the number and location of dopant atoms implanted in the channel. However, both types of devices require more complex and expensive wafers and associated processes than those used in bulk CMOS.

  In view of the substantial costs and risks associated with the transition to new technologies, semiconductor and electronic system manufacturers have long sought ways to expand the use of bulk CMOS. Those efforts have never been successful. Subsequent reduction of power consumption in bulk CMOS is increasingly seen as an insurmountable problem in the semiconductor industry.

  Techniques that can reduce power consumption in electronic devices are provided.

According to one aspect, the field effect transistor (FET) comprises a gate having a gate insulator and a conductive electrode, and 1 × 10 ¹⁸ atoms / cm ³ and 1 × 10 ²⁰ atoms / cm ³ over a minimum thickness of 5 nm. Less than 5 × 10 ¹⁷ atoms / cm ³ between the shielding region and the gate insulator, and a shielding region that is doped to extend between the shielding region and the gate insulator. A substantially undoped channel region having a dopant concentration, a channel region having a minimum thickness of 5 nm, a source and drain between which the channel region extends, and a bias voltage selectively applied to the transistor body Including body taps.

It is a figure which shows an example of the tendency of the electric power limit regarding a device scaling, and a (sigma) VT limit. FIG. 3 illustrates a field effect transistor having a deep depleted channel (DDC) according to one embodiment. FIG. 3 illustrates a channel having a deep depleted region according to one embodiment. FIG. 4 is a diagram illustrating another example of a channel having three regions with different doping concentrations according to one embodiment. FIG. 6 is a diagram illustrating another example of a channel having a deep depletion region according to one embodiment. 6 is a graph illustrating dopant concentration versus channel depth according to one embodiment. 6 is a graph illustrating the change in dopant concentration versus device depth according to one embodiment. FIG. 6 is a diagram illustrating an example of a statistical display in which different threshold voltages from various devices according to one embodiment are plotted against power supply voltage. FIG. 6 illustrates an example of an improved σV _T according to one embodiment. It is a figure which shows an example of the bulk CMOS transistor manufactured according to the conventional process and structure. FIG. 7B illustrates a DDC transistor according to one embodiment having a significantly deeper depletion region compared to the conventional bulk CMOS device of FIG. 7A. It is a figure which shows an example of FET corresponding to the conventional bulk CMOS structure shown to FIG. 7A. It is a figure which shows an example of FET corresponding to the new deep well structure shown to FIG. 7B. It is a figure which shows an example of the universal mobility curve of an NMOS device. It is a figure which shows an example of the comparison between the threshold voltage of a DDC structure, and a body bias with respect to a uniform channel. Of [sigma] v _T for the body bias of DDC structure shows a comparison for the uniform channel. FIG. 6 shows an example of a comparison between a new DDC structure profile and a conventional bulk CMOS profile with SSRW. FIG. 6 illustrates an example of a comparison of a conventional CMOS device with a structure configured according to embodiments disclosed herein. FIG. 5 illustrates an example process flow for manufacturing a device having a channel with a DDC doping profile. FIG. 5 illustrates an example process flow for manufacturing a device having a channel with a DDC doping profile. FIG. 5 illustrates an example process flow for manufacturing a device having a channel with a DDC doping profile. FIG. 5 illustrates an example process flow for manufacturing a device having a channel with a DDC doping profile. FIG. 5 illustrates an example process flow for manufacturing a device having a channel with a DDC doping profile. FIG. 5 illustrates an example process flow for manufacturing a device having a channel with a DDC doping profile. FIG. 5 illustrates an example process flow for manufacturing a device having a channel with a DDC doping profile. FIG. 5 illustrates an example process flow for manufacturing a device having a channel with a DDC doping profile. FIG. 5 illustrates an example process flow for manufacturing a device having a channel with a DDC doping profile. It is a figure which shows an example of the mechanism which applies the body bias voltage to a body and the body which has a screen area | region highly doped. between the n-channel DDC devices and conventional n-channel devices is a diagram showing an example of comparison of the threshold voltage V _T with respect to the bias voltage V _BS. It is a figure which shows an example of how the dispersion | variation in the threshold voltage for every device produces a big spread of delay time in the conventional device. FIG. 6 is a diagram illustrating an example of an improved delay time specific to a DDC device according to one embodiment. 6 is a graph illustrating a set of static _VT values for a device according to one embodiment. FIG. 4 illustrates an example of a plurality of groups of transistors having individual bodies according to one embodiment. It is a figure which shows an example of the n-channel four terminal transistor according to one Embodiment. FIG. 4 is a diagram illustrating an example of an n-channel four-terminal transistor having a shallow P-well (SPW) according to one embodiment. FIG. 3 is a diagram illustrating an example of a dynamic multimode transistor having a body access transistor according to one embodiment. FIG. 5 is a diagram illustrating another example of a dynamic multimode transistor having partial trench isolation (PTI) according to one embodiment. FIG. 3 illustrates an example of a four terminal transistor having a PTI according to one embodiment. FIG. 4 illustrates an example of a three terminal transistor having a local interconnect according to one embodiment. It is a figure which shows another example of the 3 terminal transistor provided with PGC which connects the body to a gate according to one Embodiment. FIG. 5 is a diagram illustrating another example of a three-terminal transistor with a body contact created in an active region extending under a gate extension according to one embodiment. It is a figure which shows another example of the 3 terminal transistor provided with the body contact according to one Embodiment. FIG. 3 is a diagram illustrating an example of a programmable four / three terminal transistor according to one embodiment. It is a figure which shows an example of the circuit which can perform dynamic mode switching using the 4-terminal transistor according to one Embodiment. It is a figure which shows an example of the dynamic mode switching circuit using the 4-terminal transistor according to one Embodiment. It is a figure which shows an example of the circuit which can switch the dynamic mode according to one Embodiment. It is a figure which shows an example of the cross section of the circuit block of FIG. 32A. It is a figure which shows an example of the circuit which can switch the dynamic mode according to one Embodiment. It is a figure which shows an example of the cross section of the circuit block of FIG. 33A. It is a figure which shows an example of the circuit comprised using various common parts. It is a figure which shows an example of the transistor group using the body access poly according to one Embodiment. It is a figure which shows an example of the transistor group using the body access transistor according to one Embodiment. FIG. 6 is a diagram illustrating an example of a group of transistors using body access transistors with separate taps according to one embodiment. It is a figure which shows an example of sectional drawing corresponding to FIG. 34D. FIG. 6 illustrates an example of a multi-mode switching circuit that uses a mixed legacy device and a new device according to one embodiment. It is a figure which shows an example of the other multimode switching circuit based on a legacy method. FIG. 3 is a diagram illustrating an example of a multi-mode switching circuit based on partially depleted (PD) SOI technology according to one embodiment. FIG. 3 is a diagram illustrating an example of a 6T SRAM cell according to one embodiment. FIG. 39 is a diagram showing an example of the layout of the 6T SRAM of FIG. 38. It is a figure which shows the example of sectional drawing of the layout of FIG. FIG. 40 is a diagram illustrating an example of a perspective view of a 6T SRAM cell corresponding to FIG. 39. It is a figure which shows an example of the upper side figure of the well corresponding to FIG. FIG. 6 illustrates an example of a 6T SRAM cell laid out to form a 2 × 2 array according to one embodiment. FIG. 3 is a diagram illustrating an example of a tap cell layout that can be used with the embodiments described herein. It is a figure which shows an example of sectional drawing corresponding to FIG. FIG. 43 is a diagram showing an example of a top view of the tap cell of FIG. 42. FIG. 3 illustrates an example of forming a 2 × 2 SRAM array according to one embodiment. FIG. 3 is a diagram illustrating an example of a 4 × 4 SRAM array using tap cells for SPW isolation according to one embodiment. Is a diagram illustrating an example of a 6T SRAM circuit _{V SS} for each row in accordance with one embodiment. FIG. 48 is a diagram showing an exemplary layout of an SRAM cell corresponding to FIG. 47. FIG. 49 is a diagram illustrating an example of SPW and SNW in the SRAM layout corresponding to FIG. 48. Is a diagram showing a 2 × 2SRAM array with V _SS technique for each row in accordance with one embodiment. It is a diagram showing a 4 × 4SRAM array having a V _SS technique for each row in accordance with one embodiment. FIG. 48 is a diagram showing another example of the layout of the SRAM cell corresponding to FIG. 47. FIG. 52 is a diagram illustrating an example of an SPW and SNW of an SRAM layout corresponding to FIG. 50. FIG. 3 is a diagram illustrating an example of a 2 × 2 SRAM array having V _SS for each row according to one embodiment. It is a diagram showing a 4 × 4SRAM array having a _{V SS} for each row in accordance with one embodiment. FIG. 2 illustrates a DDC device described herein and a system application of an embodiment. FIG. 2 illustrates a DDC device described herein and a system application of an embodiment. FIG. 2 illustrates a DDC device described herein and a system application of an embodiment.

  In order to reduce power consumption in a wide array of electronic devices and systems, a set of new structures and methods are provided. Some of these structures and methods can generally be realized by reusing existing bulk CMOS process flows and manufacturing technologies, replacing the semiconductor industry and the wider electronics industry with cost and risk It makes it possible to avoid switching to technology.

As described below, some of the structures and methods relate to deeply depleted channel (DDC) designs that are deeply depleted. DDC enables CMOS devices with reduced σV _T compared to conventional bulk CMOS, and allows the threshold voltage V _{T of} FETs with dopants in the channel region to be set much more accurately. obtain. The DDC design can also have a strong body effect compared to conventional bulk CMOS transistors, which can allow for significant dynamic control of the power consumption of the DDC transistors. There are numerous approaches to configuring the DDC to achieve various benefits, and the additional structures and methods presented here can be used alone or in conjunction with the DDC to generate additional benefits. It is possible.

  For example, advantageous methods and structures for integrating transistors on a chip are also provided, including implementations that can utilize DDC to provide improved chip power consumption. Further, the transistors and integrated circuits in some embodiments may provide a variety of other benefits including low heat dissipation, increased reliability, miniaturization, and / or better manufacturing economy. A variety of approaches exist to enhance some or all of the advantages of the new transistor structure, both static and dynamic. Many of the developments at the integrated circuit level also provide benefits in the absence of the new transistors described herein. Many of the methods and structures may also be useful in device types other than bulk CMOS transistors, including, for example, other types of transistors having dopants in the channel and / or body.

  Provides benefits including improved system level power consumption, improved system performance, improved system cost, improved system manufacturability, and / or improved system reliability in some implementations Thus, methods and structures are also provided for incorporating the innovation described herein into a system, such as an electronic product, or for use within the system. As illustrated later, in some embodiments, for example, a personal computer, a mobile phone, a television, a digital music player, a set top box, a laptop or palmtop computing device, an electronic book reader, a digital camera, a GPS The innovation can be advantageously employed in a wide range of electronic systems including systems, flat panel displays, portable data storage devices, and household equipment such as tablets, and a variety of other electronic devices. In some of these implementations, the transistors and integrated circuits can substantially enhance the overall operation of the electronic system and thus the commercial suitability. In some embodiments, the innovative transistors, integrated circuits, and systems that include them described herein may also allow for a more environmentally friendly implementation than other approaches.

  In one embodiment, a new field effect transistor (FET) structure is provided that has a precisely controlled threshold voltage compared to conventional short channel devices. The FET structure may also have improved mobility and other important transistor characteristics. This structure and its manufacturing method may allow FET transistors with lower operating voltages compared to conventional devices. In addition, or alternatively, it may be possible to dynamically control the threshold voltage of such devices during operation. The FETs in some implementations provide the designer with the ability to design an integrated circuit with a dynamically tunable FET device during operation of the integrated circuit. The FET structure in an integrated circuit can be designed to have a nominally identical structure in some embodiments, and in addition or alternatively, at different operating voltages in response to different bias voltages. It can be designed to be able to be controlled, adjusted or programmed to operate. These structures may allow the circuit to statically specify and / or dynamically change the operating mode in an efficient and reliable manner. Also, in some implementations, these structures can be set after manufacture for various applications within a circuit.

  These and other benefits provide advancements in digital circuits that meet the numerous requirements of designers, manufacturers and consumers. These benefits can provide a system with a new structure that allows for further development of integrated circuits, thereby resulting in devices and systems with improved performance. In some implementations, bulk CMOS can continue to keep Moore's Law pace for additional periods, and further innovations in circuits and systems based on bulk CMOS may continue to increase the speed of advanced performance. it can. Embodiments and examples are described herein with reference to transistors, integrated circuits, electronic systems, and related methods, and new structures and methods are described in various levels of manufacturing processes, including those for end users of electronic products. And highlight the features and benefits offered in the commercial chain. It will be appreciated that the application of concepts specific to the structures and methods of manufacturing integrated circuits and electronic systems in these examples is extensive. Accordingly, as will be appreciated, the spirit and scope of the invention is not limited to these embodiments and examples, but only by the claims appended hereto and the related application of the same applicant. .

  Nanoscale field effect transistors (FETs) having a gate length of less than 90 nm are provided to have threshold voltages that are more precisely controllable than conventional nanoscale FET devices. Further benefits include improved carrier mobility and reduced threshold variation due to RDF. One embodiment includes a nanoscale FET structure operable to have a depletion zone or depletion region that extends to a depth below the gate that is set to be greater than 1/2 the gate length. The FET structure has at least two regions with different doping concentrations that help define DDC in this depletion zone or depletion region below the gate. In one example, the first region near the gate has a lower dopant concentration than a second region spaced from the first region and disposed at a distance below the gate. This allows a second doped screening (screening) that can act to define the DDC by terminating the electric field arising from the gate when a threshold voltage or greater is applied to the gate. A first lightly doped channel region (typically a substantially undoped epitaxial growth channel layer) is provided that pairs with the region. Regions that are deeply depleted can be referred to alternatively as DDCs or deep depleted zones and can vary in spatial extent and characteristics depending on the transistor structure and electrical operating conditions. There are many variations on the exact geometry and arrangement of these structures and regions, some of which are described in detail below.

  These structures, and methods of manufacturing the structures, enable FET devices that have both low operating voltages and low threshold voltages compared to conventional nanoscale devices. Furthermore, they allow the threshold voltage of such devices to be controlled dynamically during operation. Ultimately, these structures, and methods of manufacturing the structures, provide for designing integrated circuits having FET devices that can be dynamically adjusted while the integrated circuit is in operation. Thus, the transistors in an integrated circuit are designed to have nominally the same structure, but operate at different operating voltages in response to different bias voltages, or differ in response to different bias voltages and operating voltages. It can be controlled, adjusted or programmed to operate in an operating mode. They can also be set after manufacture for different applications in the circuit.

  Here, specific embodiments and examples are described with reference to transistors, highlighting the features and benefits that the new structures and methods provide for transistors. However, the applicability of concepts specific to the structures and integrated circuit fabrication methods in these embodiments is extensive and is not limited to transistors or bulk CMOS. Accordingly, as will be understood in the art, the spirit and scope of the invention is not limited to these embodiments and examples, or the claims appended hereto and the related application of the same applicant, but other digital It can also be advantageously applied to circuit situations.

  In the following description, numerous specific details are given of some of the preferred ways in which the invention may be implemented. As will be readily apparent, the present invention may be practiced without these specific details. In other instances, well-known circuits, components, algorithms and processes have not been shown in detail or are shown in schematic or block diagram form in order not to obscure the invention in unnecessary detail. Also, for the most part, materials, tools, processing timing, circuit layout and die design are considered to be within the understanding of those skilled in the art, so that a full understanding of the invention may be obtained. Omitted unless detailed details are required. Throughout the following description and claims, specific terminology is used to refer to specific system components. Similarly, as will be appreciated, components may be referred to by different names, and the description herein is not intended to distinguish between components that differ in name but not function. In the following description and claims, the terms “including” and “having” are used as endless, open-ended, and thus mean, for example, “including but not limited to” It should be interpreted.

  Various embodiments and examples of the methods and structures described above are described. As will be appreciated, this detailed description is merely exemplary and is not intended to be in any way limiting. Other embodiments are suggested to those skilled in the art who benefit from this disclosure. Reference will now be made in detail to embodiments illustrated in the accompanying drawings. The same reference numbers are used throughout the drawings and the following detailed description to refer to the same or like parts.

  For the sake of clarity, not all of the normal functions of the implementations and embodiments described herein are illustrated and described. As will be appreciated, of course, in the actual implementation of the present invention, a number of implementation specific decisions will typically be made to achieve the specific purpose of the developer. It will also be appreciated that such development efforts can be complex and time consuming, but nevertheless, with normal engineering work for those skilled in the art who benefit from this disclosure. I will.

  In addition, the concentration of atoms implanted or otherwise present in a semiconductor substrate or crystal layer to change the physical and electrical properties of the semiconductor is dependent on the physical and functional regions or layers (layers). ). These can be understood by those skilled in the art as a collection of three-dimensional materials having a specific average concentration. Alternatively, they can be understood as partial areas or sublayers having different densities or spatially varying densities. They may also exist as a small population of dopant atoms, a region of substantially similar dopant atoms or the like, or other physical form. The description of a region based on these characteristics is not intended to limit the shape or exact position or orientation. They also limit these regions or layers to a specific type or number of process steps, type or number of layers (eg, composite or single layer), semiconductor deposition techniques used, etching techniques or growth techniques. It is not intended to be. These processes include epitaxially formed region or atomic layer growth, dopant implantation methods, or specific longitudinal or lateral dopant profiles including linear, monotonically increasing, retrograde or other suitably spatially varying dopant concentrations. Can be included. The embodiments and examples included herein may illustrate the specific process techniques or materials used, such as, for example, the epitaxial processes and other processes shown in FIGS. 14A-14I and described below. These examples are intended as illustrative examples only and should not be construed as limiting. A dopant profile can have one or more regions or layers with different dopant concentrations, regardless of process, concentration variations, and how regions or layers are defined is optical including infrared spectroscopy. May be detectable by other dopant analysis tools using techniques, Rutherford backscattering (RBS), secondary ion mass spectrometry (SIMS), or various qualitative or quantitative dopant concentration determination methods It may not be possible.

FIG. 2A illustrates a field effect transistor (FET) 100 configured in accordance with one embodiment. The FET 100 includes a gate electrode 102, a source 104, a drain 106, and a gate stack 108 located on the channel 110. The channel 110 can be deeply depleted, that is, as will be described in more detail below, the channel depth measured from the gate stack to the screening (shielding) region is generally significantly greater than the conventional channel depth. . In operation, bias voltage V _SS 122 is applied to source 104 and P + terminal 126 is connected to P well 114 at connection 124 to close the circuit. The gate stack 108 includes a gate electrode 102, a gate contact 118, and a gate dielectric 128. A gate spacer 130 is included to separate the gate from the source and drain. A source / drain extension (SDE) 132 extends the source and drain below the dielectric 128.

  The FET 100 is shown as an N-channel transistor having a source and drain formed of an N-type dopant material, with the source and drain as a P-type doped silicon substrate that provides a P-well 114 formed on the substrate 116. Formed on the substrate. However, as will be appreciated, non-silicon P-type semiconductor transistors formed from other suitable substrates such as, for example, gallium arsenide based materials may be substituted with appropriate changes in the substrate or doping material.

  Source 104 and drain 106 can be formed using conventional dopant implantation processes and materials, and include, for example, stress-induced source / drain structures, raised or recessed source / drains, asymmetrically doped sources. Modifications such as / drain, counter-doped source / drain, crystal structure altered source / drain, or source / drain extension region implantation doping according to HDD (Highly Doped Drain) technology. The extension region 132 is generally formed in the substrate and facilitates absorption of a portion of the potential coupled to the drain. Various other techniques for altering the operating characteristics of the source / drain can also be used. Such a technique facilitates scaling of the device channel length by creating a local dopant distribution in the vicinity of the source / drain (S / D) region so that it can extend under the channel. Includes extensions (tips) or halo implantation. In certain embodiments, multiple dissimilar dopant materials can be used as compensating dopants that change electrical properties.

  The gate electrode 102 can be formed from conventional materials, including, but not limited to, certain metals, metal alloys, metal nitrides and metal silicides, and stacks and mixtures thereof. . The gate electrode 102 may also be formed from polysilicon (eg, including heavily doped polysilicon and polysilicon-germanium alloy). The metal or metal alloy may include aluminum, titanium, tantalum, and their nitrides including titanium-containing compounds such as titanium nitride. The formation of the gate electrode 102 may include a silicide method, a chemical vapor deposition method, and a physical vapor deposition method such as, but not limited to, a vapor deposition method and a sputtering method. Typically, the gate electrode 102 has a total thickness from about 1 nm to about 500 nm.

The gate dielectric 128 may include conventional dielectric materials such as oxides, nitrides, and oxynitrides. In other examples, the gate dielectric 128 can generally include higher dielectric constant dielectric materials, including, but not limited to, hafnium oxide, hafnium silicate, zirconium oxide, oxidized Includes lanthanum, titanium oxide, barium strontium titanate, lead zirconate titanate, metal-based dielectric materials, and other materials with dielectric properties. Suitable hafnium-containing _oxide, and the like _{HfO 2, HfZrO x, HfSiO x} , HfTiO x and HfAlO _x. Depending on the composition and available deposition processing equipment, the gate dielectric 128 may be thermally or plasma oxidized, nitrided, chemical vapor deposition (including atomic layer deposition), physical vapor deposition, etc. It can be formed by the method. In some embodiments, multiple layers or composite layers, laminates, and composition mixtures of dielectric materials may be used. For example, the gate dielectric is composed of a SiO ₂ based insulator having a thickness between about 0.3 nm and 1 nm and a hafnium oxide based insulator having a thickness between 0.5 nm and 4 nm. Can be formed. Typically, the gate dielectric has a total thickness from about 0.5 nm to about 5 nm. A channel region 110 is formed on the screening layer 112 under the gate dielectric 128. Channel region 110 contacts and extends between source 104 and drain 106. Preferably, the channel region comprises substantially undoped silicon, or advanced material such as material from the SiGe family, or very low level doped silicon. The channel thickness may typically be in the range of 5 nm to 50 nm.

The discussion that immediately follows focuses on bulk CMOS devices. In many nanoscale bulk CMOS FET devices, carrier mobility is adversely affected by the high concentration of channel dopant required to set the threshold voltage VT. High dopant concentration levels can prevent significant power leakage, but when dopants are present at high concentrations, they act as scattering centers that can significantly reduce the channel mobility of mobile carriers such as electrons. Can do. In such a case, electrons in the channel region are scattered and do not move efficiently in the channel between the source and drain. This effectively limits the maximum amount of current (I _dsat ) that the channel can carry. Also, very thin gates and the resulting high electric field at the gate dielectric / channel interface can cause severe quantum mechanical effects that reduce the inversion layer charge density at a given gate voltage. A decrease in inversion layer charge density is accompanied by a decrease in mobility and an increase in the threshold voltage V _T , which also degrades device performance. Due to these features, conventional scaling of bulk CMOS devices to smaller desired sizes has proved increasingly difficult.

  As a further benefit, the use of a substantially undoped channel region can enhance the effectiveness of certain prior art techniques that are often used to improve transistor performance. For example, the source 104 and drain 106 located on opposite sides of the channel region 110 can be structured to change the stress applied to the channel region. In another example, the channel region can be modified by a thin film lattice of lattice matched strained silicon germanium (SiGe) crystals arranged to create a compressive strain in the in-plane direction of the channel. This can cause a change in the band structure so that the hole mobility is increased compared to intrinsic Si. Stress conditions can be changed by changing the germanium (Ge) composition (the more Ge, the greater the strain and the higher the hole mobility). In the case of tensile strain, the Si in the channel region can be formed on lattice relaxed SiGe having a larger lattice constant. This will increase both electron mobility and hole mobility compared to an unstrained Si channel region. Again, as the germanium composition of the base SiGe increases, the amount of stress and carrier mobility in the strained Si channel region tend to increase. As will be appreciated, a continuous stress layer is not necessarily required to apply stress to the channel region. Discontinuous or separate stress layers, including multiple stress layers that are positioned above, below, or next to, or adjacent to each other, apply compressive or tensile forces at various locations along the channel region. And substantially allows greater control of the applied stress.

  In certain embodiments, a stress layer can mean any layer of material suitable for applying stress to a channel region when provided adjacent to or in contact with the channel. As an example, in certain embodiments, the stress layer may include materials having different coefficients of thermal expansion with respect to some or all of the other portions of the semiconductor substrate. In the manufacture of such an embodiment, as the temperature of the semiconductor substrate is lowered, certain subgroups contract differently, causing the channel region to stretch or compress. As a result, at least a part of the channel region is distorted to increase carrier mobility. In certain embodiments, the stress layer may comprise a material such as silicon nitride having a higher coefficient of thermal expansion than some or all of the semiconductor substrate. In addition or alternatively, a plurality of different stress layers may be provided in various portions of the FET 100 to selectively increase the mobility of either holes or electrons in the channel region. For example, in certain embodiments in which complementary pairs of n-type and p-type transistors are separated from each other by a suitable p-type and n-type well structure, n is applied to apply a tensile stress to the channel region of the n-type transistor. The type transistor may be provided with a stress layer. This tensile stress can induce strain in the channel region that increases the mobility of electrons in the channel region. Another stress layer may be provided on the p-type transistor to apply compressive stress to the channel region of the p-type transistor. This compressive stress can induce strain in the p-type channel region that increases hole mobility.

  Providing a transistor with a substantially undoped channel provides a further advantage when stress is applied. For example, the stress can be applied by compressive or tensile stress applied by source / drain or channel stress techniques. Compared to conventional nanoscale transistors with uniformly or heavily doped channels, strained channel region FET transistors have lower dopant concentrations (suppressed ionized impurity scattering) and lower electric fields (suppressed) near the gate dielectric. Enhanced surface mobility) provides greater strain enhanced mobility. Due to the suppression of scattering, the mobility enhanced by stress is significantly higher in conventional devices. The advantage of this mobility due to distortion is actually significant with transistor scaling.

  FIG. 2A is a schematic diagram of a transistor configured in accordance with one embodiment. 2B, 2C, and 2D are schematic diagrams further illustrating three different examples of DDC transistor channels that can be replaced with the channel 110 of FIG. 2A. Different regions may include a deep depletion region, a threshold voltage adjustment region, and a heavily doped screening region located in the vicinity of the gate dielectric (eg, dielectric 128 shown in FIG. 2A). FIG. 2B shows an example of a cross-section of a DDC transistor channel located adjacent to the gate dielectric, having two regions with different dopant concentrations. The channel cross-sectional profile includes a depletion region 202 located between the gate dielectric (not shown) and the screening region 204. The dopant atom 206 shows the dopant density in the screening region 204 in association with the relative dopant density in the depletion region 202 compared to the screening region 204.

  FIG. 2C shows a channel region 208 according to another example, and this channel region has three regions having different doping concentrations. In this example, the depleted dopant channel region 214 has the least amount of dopant 206, the threshold adjustment region 212 generally has a higher dopant atom concentration than the depleted dopant channel region 214, and the screening region 210 is It has the highest dopant atom concentration.

  FIG. 2D shows a further variation where the channel cross-section has a dopant atom concentration 224 that increases from the top channel region to the bottom. In various applications and embodiments, the dopant range at the top of the channel can vary, but is typically as low as possible towards the top of the channel as long as process and annealing conditions allow. The dopant range can be increased towards the center of the channel, with higher dopant concentration going through the bottom of the channel to the screening region.

In any of these configurations, the threshold voltage adjustment region can be formed as a separate epitaxially grown silicon layer, or it can be formed as part of a single silicon epitaxial layer that also includes a depletion channel region. The thickness of the threshold adjustment region may typically be in the range of 5 nm to 50 nm thickness. When substantially undoped, the threshold voltage is slightly adjusted by appropriate selection of the thickness of the region itself, but in more typical applications the threshold voltage adjustment region is 5 × 10 ¹⁷ atoms / cm. Doped to have an average concentration in the range between ³ and 2 × 10 ¹⁹ atoms / cm ³ . In certain embodiments, a dopant migration resistant layer, such as carbon, germanium, or the like, is used for threshold voltage adjustment to prevent dopant migration to the channel region or from the screening region to the threshold voltage adjustment region. It can be provided above and / or below the area.

The screening region is a heavily doped region embedded under the channel region and, if provided, the threshold voltage adjustment region. The screening layer is generally positioned at a distance so as to avoid direct contact with the source and drain. The screening region may be formed as a sheet extending under the plurality of source / drain / channel regions in other specific embodiments, and in other embodiments is self-aligned having the same extent as the channel region. It may be an injection or a layer. The thickness of the screening region can typically be in the range of 5 nm to 50 nm. The screening region is more heavily doped than the channel, threshold voltage adjustment region (if provided) and P-well. In practice, the screening region is doped to have a concentration between 1 × 10 ¹⁸ atoms / cm ³ and 1 × 10 ²⁰ atoms / cm ³ . In certain embodiments, a dopant migration resistant layer, such as carbon, germanium, or the like, may be provided over the screening region to prevent dopant migration to the threshold voltage adjustment region.

In operation, when a predetermined voltage higher than the threshold voltage is applied to the conductive gate, a deep depletion region is formed between the gate stack and the screening region. Under the conductive gate, the deeply depleted region typically extends down into the screening region, but in certain heavily doped embodiments, the deeply depleted region is the threshold voltage when provided. It can end within the adjustment area. As will be appreciated, the exact depth of the depletion region under the conductive gate is determined by a number of factors that can be adjusted by the design of the FET. For example, the depth of the depletion region is determined by the spatial location of other elements of the FET and the absolute or relative dopant concentration. For example, FET is arranged between the source region and the drain region, and under the gate having a gate length L _G, may have a channel made two strokes. The DDC depth (X _d ) can be set to be deeper than half of the gate length, possibly by a factor of ½ of the gate length, or a rate close thereto. In one example, this DDC depth may be set approximately equal to one half of the channel length, allowing an accurate setting of the threshold voltage during operation, even at low operating voltages below 1V. Depending on the requirements of a particular application, different depths can yield different advantageous results. According to this disclosure, as will be appreciated, different DDC depths are possible in different applications, different device configurations, and various parameters of a particular design. Depending on the parameters of a particular application, different region thicknesses, dopant concentrations and operating conditions used in forming a DDC transistor can yield different advantageous results.

For example, according to another embodiment, the depletion depth can be maintained from 1/3 of the gate length to a depth approximately equal to the gate length. However, as will be appreciated by those skilled in the art, if the transistor structure and operation is such that the depletion depth is less than half the gate length, the performance of the device with respect to power consumption will gradually degrade, The profit of DDC will decrease. Such as DDC transistor depletion depth under the gate is set to approximately 0.4 × L _G, when the depletion depth X _d is between 1/3 and 1/2 of the gate length, the device remains conventional A small improvement to the device can be achieved. In this example, a suitable thickness range for the screening region is between 5 nm and 50 nm, with dopant concentrations ranging from 1 × 10 ¹⁸ atoms / cm ³ to 1 × 10 ²⁰ atoms / cm ³ . The preferred thickness range of the threshold voltage adjustment region is between 5 nm and 50 nm, with the dopant concentration ranging from 5 × 10 ¹⁷ atoms / cm ³ to 2 × 10 ¹⁹ atoms / cm ³ . The channel region of the _undoped, X d> 1/2 × chosen to be deep enough to meet the _{L G} becomes constrained, and has a concentration of less than 5 × ^{10 17} atoms / cm ^3.

  In fact, providing a deep depletion region for a DDC transistor can significantly narrow the tolerance for setting a threshold voltage in a circuit having a plurality of transistors and related devices, and can further suppress variations due to RDF. As a result, a more predictable and reliable threshold voltage can be set across multiple devices of the integrated circuit. This benefit can be used to reduce power in the device or system and can result in better overall performance.

  Another benefit that can be realized by this embodiment is that of a device or system that can be set statically or configured using one or more of the transistor structures described herein. An adjustable threshold voltage that can be dynamically changed during operation. As also shown in FIG. 2A, a bias voltage can be applied between the source 104 of the transistor and a dopant material 126 connected to a P-type well 114 of opposite conductivity type. Conventional circuits are typically biased to the supply voltage so that current can flow from source to drain when an operating voltage is applied to the gate. Although the use of adjustable body biasing to dynamically set the threshold voltage has been previously proposed, it causes significant chip area penalty and thus hinders the level of on-chip integration Due to the tendency, it is generally not proven to be practical. According to this embodiment, a transistor (or a group of transistors) can be configured by changing the bias voltage applied to the well, whether configured in one integrated circuit or system or in separate circuits. If they share a common well, the circuit can be configured to change the threshold voltage of the transistor group). As described in more detail below, the ability to reliably control the threshold voltage within a tight range allows the threshold voltage to be reliably and dynamically changed during operation while reducing the chip area penalty. Combined with what can be done, it provides a device or system that can dynamically change the mode of operation of a transistor or group of transistors within the device or system.

FIG. 3 shows a graph 300 of dopant atom concentration versus channel depth below the gate dielectric according to one embodiment, illustrating the range of dopant concentration at various depth ranges within the channel. . Two curves, a more practical curve 308 and an ideal curve 310 are shown. As can be seen, three levels are represented: the first 5-20 nm channel region, the next 5-20 nm threshold voltage adjustment region after the channel region, and the next 5-20 nm screening region after the threshold voltage adjustment region. Has been. These different levels of concentration each reach a specific level 312, 314, 316. These specific levels are, but not necessarily, sometimes the inflection points of the graphs of the respective concentration levels, which have a channel dopant concentration “d” of less than 5 × 10 ¹⁷ atoms / cm ^3. A specific dopant concentration level 302, a specific dopant concentration level 304 with a threshold voltage adjustment region concentration “d” between 5 × 10 ¹⁸ atoms / cm ³ and less than 5 × 10 ¹⁷ atoms / cm ³ , and 5 × It corresponds to a specific dopant concentration level 306 with a screening region dopant concentration “d” higher than 10 ¹⁸ atoms / cm ³ . According to some embodiments, certain optimal benefits in nanoscale FETs that support deep depleted region operation can be realized within these dopant concentration ranges.

The dopant profile according to various embodiments is defined such that two regions occur. The three areas are defined in Table 1. Region 1 corresponds to a channel region located in the vicinity of the gate dielectric, region 2 corresponds to a threshold voltage adjustment region, and region 3 corresponds to a screening region. Where _LG is the gate length. As can be seen, the gate length is substantially equal to the channel length, and t ₁ , t ₂ and t ₃ are the thicknesses of each of the three regions. Each of these regions can be represented by a typical thickness and a dopant dose measured as the number of atoms per cubic centimeter. These thickness and dose values are presented in Table 1.

The layer thickness depends on the process node, and their respective thicknesses t ₁ , t ₂ and t ₃ are related to the gate length (L _G ) of the device of interest and the process node. Table 2 contains a representative value of the process nodes from 90nm to 15 nm, it illustrates the effect of L _G scaled to the thickness requirements of these regions.

FIG. 4 is a graph 400 of the change in boron concentration (atoms / cm ³ ) versus device depth in one implementation. In this example, the dopant concentration is lowest (less than 1 × 10 ¹⁷ ) in the low dopant region near the transistor gate at a depth from zero to approximately 20 nm, and slightly higher in the threshold voltage adjustment region from approximately 20 nm to 45 nm ( Approximately 5 × 10 ¹⁸ ). This example peaks at a higher value (approximately 5 × 10 ¹⁹ ) in the screening region from approximately 45 nm to 75 nm. This particular example shows three different simulated devices made using different processes, which are shown as an overlay graph. One uses a 15 second anneal at 975 ° C., one uses a 15 second anneal at 800 ° C., and the third uses no anneal at all. The graph results are substantially similar, indicating the reliability of the dopant concentration in these different process environments. As will be appreciated by those skilled in the art, different design parameters and applications may require different variations or numbers of regions with different doping concentrations.

In practice, designers and manufacturers collect statistical data from mathematical models and sample measurements from real circuits to determine the threshold voltage variance of the circuit design. Whether the voltage mismatch between transistors is due to manufacturing variation or RDF, it is determined as σV _T. As an example of such a statistical display, FIG. 5 shows a plot of different threshold voltages from various devices versus power supply voltage. In order for the circuit to operate as a whole, the operating voltage V _DD must be selected in consideration of σV _T. In general, the greater the variation, the greater the σV _T, so the operating voltage V _DD must be set high so that the transistor operates properly. If multiple devices are implemented in a circuit, V _DD may need to be set to the highest overall value in order for the circuit to operate properly.

A structure for reducing σV _T and reducing the variation range of threshold voltages of a plurality of transistors across an integrated circuit and a method for manufacturing the same are provided. With the reduced σV _T , the static value of V _T can be set more accurately and can be changed in response to changing the bias voltage. An example of an improved σV _{T according} to one embodiment is shown in FIG. FIG. 6 shows an improved threshold voltage range display evident from variations on the lower side of the threshold voltage obtained from different devices. The threshold voltage of nominally equal devices in the circuit can be set more accurately with a reduced σV _T , so that the device operates with a lower operating voltage V _DD and thus This makes it possible to consume less power. Also, there are more space for changing the V _T for a given transistor or transistor group, the device can be operated in different modes corresponding to different bias voltages for a particular mode. This adds functionality to many devices and systems, and can be particularly beneficial for devices where fine control of the device's power mode is useful.

FIG. 7A shows an example of a transistor 700 manufactured according to conventional processes and structures. This example is shown as an N-type FET having a source 702, a drain 704, and a gate stack including a conductive gate 706 and an insulating layer 708. Typically, the gate 706 is formed from heavily doped polysilicon and the insulating layer 708 is formed from a gate dielectric such as silicon oxide. The gate stack 706 electrically controls the current flowing between the source 702 and the drain 704. Channel 710 typically includes a dopant and extends down to P-well 712 and may surround both the source and drain. The channel depth X _d 714 is the distance from the gate dielectric 708 down to the bottom surface 720 of the channel. In operation, there are a plurality of lines of electric force, such as E716, that extend downward within the channel depth 714 and bend toward the source 702 and drain 704. These field lines are typically not linear as shown, but may be bent as a result of the structure and operation of the device. For example, mobile carriers such as electrons e ⁻ 718 travel between the source 702 and the drain 704 via the electric field E 716. A gate spacer 724 and SDE 722 are also shown.

In contrast, FIG. 7B illustrates one embodiment of a DDC transistor 700 ′ that operates with a significantly deeper depletion region compared to the conventional device 700 of FIG. 7A. This provides improved mobility characteristics and benefits without using a stress-inducing layer, and provides improved threshold voltage settings. This example is shown as an N-type FET having a source 702 ', a drain 704', and a gate 706 '. The transistor includes a gate 706 ′ formed on a gate dielectric 708 ′ that creates a depletion region 710 ′ when the gate-source voltage is biased above a threshold voltage, Control the current between 702 'and drain 704'. The depletion region 710 ′ may extend down to the screening layer 720 ′ ion implanted as a layer in the P-well 712 ′, as shown, and may surround both the source 702 ′ and the drain 704 ′. A gate spacer 724 'and SDE 722' are also shown. The depletion depth X _d '714' is the distance from the gate dielectric down to the screening region 720 'and is significantly deeper than the depletion region of the conventional device of FIG. 7A. Unlike the conventional device of FIG. 7A, the screening region 720 ′ of device 700 ′ provides a heavily doped termination for an electric field, such as E716 ′, that extends down to the screening layer. Deeper depletion X _d ′ makes these field lines generally longer and straighter than the electric field E 716 in the conventional structure 700. As with conventional devices, when biased, current flows from source 702 'to drain 704' and electrons e ^- 718 'travel between drain 704' and source 702 'via electric field E716'. However, in contrast to conventional devices, electrons flow more freely across these electric fields E716 ', providing improved current and better performance. This structure also improves σV _T by suppressing short channel effects and suppressing variations caused by random dopant fluctuations.

Referring to FIG. 8A, an FET 800 corresponding to the conventional structure shown in FIG. 7A is shown. Leaks that cause power loss occur at various points in the overall transistor structure even when the FET is not actively switched. FIG. 8A specifically illustrates the concept of leakage that occurs between the source 702 and the well 712. When cations 802 are present in well 712, they tend to move to holes 804 via leak path X _j 806. Due to the relatively short path 806, leaks are widespread in conventional nanoscale devices.

FIG. 8B shows an FET 800 ′ operating with a deep depletion region similar to that shown in FIG. 7B, and further illustrates the concept of leakage occurring between the source 702 ′ and the well 712 ′. ing. Cation 802 'is present in well 712'. However, due to the new structure with deeper wells, the path X _j 806 ′ is significantly longer and the cation 802 ′ is less likely to move to the holes 804 ′ via the leak path X _j 806 ′. In this case, the relatively long path 806 ′ does not spread the leak much compared to the conventional device. In addition, the low electric field E 716 ′ in the new structure greatly reduces the ability to excite electrons for leaks in the gate 706 ′ and the insulator 708 ′. As a result, leakage at the gate is substantially reduced. Thus, the new structure with DDC provides a significant reduction in leaks that have occurred in many places in the prior device.

DDC transistors also preferably provide increased carrier mobility, which is a major concern in industry. Mobility is a quantitative indicator of the ability of mobile carriers to move from source to drain across the channel of the transistor when a voltage higher than the threshold voltage V _T is applied to the gate. One goal of an optimization device is to move electrons or mobile carriers from source to drain with minimal obstruction, typically according to the relationship between the gate applied field and the measured mobility, known as the universal mobility curve. It is. This universal mobility curve is a well-established relationship between the carrier mobility in the inversion region of the channel and the electric field that induces the inversion region (or inversion charge) seen in MOSFET devices. FIG. 9 shows this universal curve (solid line) of the NMOS transistor, but a similar curve also exists in the case of the PMOS. In this figure, the universal mobility curve of an undoped channel is plotted. Region A corresponds to the mobility / field operation regime of MOSFET transistors according to typical current technology, and these devices have reduced mobility relative to mobility in the low field / low power region in the high power region. Shows that it works.

  The second mobility curve (dashed line) has a heavily doped channel (often necessary to compensate for scaling effects) and a proportionally scaled gate voltage and resulting low electric field. Suitable for nanoscale gate length transistors. These curves can be matched in operating conditions that support a high electric field in the channel. This is because the mobility is governed by the surface roughness associated with the interface between the gate dielectric and channel silicon. When operating a transistor with a lower gate voltage (and consequently lower electric field), these two curves show the presence of dopant atoms and channel dopant scattering (generally ionized impurities) that act to reduce electron mobility. Branches for dominating (called scattering). This is illustrated as region C. Although it is possible to build a low power device operating with an electric field entering region C, the required high concentration channel doping is due to dopant scattering in the region marked as region A in FIG. Cause a drop in

  The operating point of the DDC transistor is located along the universal mobility curve as shown as region B in FIG. DDC transistors not only operate in a low power regime using a low electric field, but can also benefit from being a deeply depleted device with substantially low dopant scattering that reduces its mobility. The DDC transistor can thus achieve a mobility increase of up to 120% over conventional high power devices in some preferred embodiments.

With these new structures and a manufacturing method thereof, it is possible to manufacture and configure the circuit to have the ability to dynamically change the V _T. This structure is preferably configured to have a small σV _T compared to a conventional device, and the device changes in response to a bias voltage as well as a lower nominal threshold voltage V _T and a lower operating voltage V _DD. possible precisely to be able to have also a adjustable V _T. In operation, it is possible to place a bias voltage to the transistor which operates to raise and lower the V _T of the device. This allows the circuit to specify the operating mode statically and / or dynamically change, efficiently and reliably, especially when the operating voltage V _DD is also dynamically controlled. . Moreover, the adjustment of V _T may be performed one or more transistors of the circuit, a group of transistors, and for different compartments or regions. This breakthrough allows the designer to use general purpose transistors that can be tuned to provide different functions in the circuit. There are also many circuit and system level innovations that result from the features and benefits of these integrated circuit structures.

  In one embodiment, a semiconductor structure with a DDC having a DDC depth is provided in which a channel is formed between a source region and a drain region. In one example, the DDC depth is at least half the size of the device channel length. These structures can operate at lower voltages than conventional devices and are not limited by the effects of RDF in the device channel. This new structure can also be fabricated using conventional bulk CMOS process tools and process steps.

  According to one embodiment, the channel region of the transistor can be configured to have multiple regions with different dopant concentrations. In one example, the DDC transistor is configured such that three distinct regions extend under the gate. From the gate dielectric deeper into the substrate, these regions include a channel, a threshold voltage adjustment region, and a screening region. As will be appreciated by those skilled in the art, there may be different combinations or substitutions of these regions.

  The channel region is a region where fractional carriers travel from the source to the drain during operation of the integrated circuit. This constitutes the current flowing through the device. The amount of dopant in this region affects device mobility via impurity scattering. Lower dopant concentration results in higher migration. Also, RDF decreases as the dopant concentration decreases. This undoped channel region may allow the DDC transistor to achieve both high mobility and low RDF.

The threshold voltage adjustment region allows complementary dopants such as N-type dopants in PMOS and P-type dopants in NMOS to be introduced below the channel region. The introduction of this _VT tuning region, combined with its proximity to the channel region and the level of dopant, preferably allows the threshold voltage tuning region to change the depletion region in the channel without directly doping the channel. Make it possible. The depletion control enables changing the V _T of the device to achieve the desired result. The _VT adjustment region can also help prevent subchannel punchthrough and leakage. In some embodiments, this provides improved short channel effects, DIBL and subthreshold slope.

  In conventional processes, those skilled in the art have addressed various transistor performance indicators by varying specific structures and concentrations. For example, a gate metal alloy or polysilicon may be used to adjust the doping concentration to improve short channel effects or other parameters. The gate dielectric located above the gate and above the channel can also be adjusted. There are other processes that can set the dopant concentration in or near the channel of the transistor. Unlike these conventional attempts to improve the short channel effect and other parameters of the device, some of the embodiments disclosed herein not only improve more parameters of the device, but also the threshold voltage of the device. It is also possible to improve the accuracy and reliability when setting. In addition, in some implementations, the improved device can also dynamically increase the threshold voltage of the device to enhance performance and to provide new functionality and operation of the apparatus and system when employed. Control may be possible.

  In one embodiment, a transistor device is provided that monotonically increases the dopant concentration from the top of the channel near the gate down into the channel. In one example, there is a linearly increased dopant starting from the gate dielectric side. This can be accomplished by forming a screening region at a distance from the gate and having a depletion region between the screening region and the gate. This depleted region may take different forms, including one or more regions with different dopant concentrations. These regions increase the reliability of setting a specific threshold voltage, increase the mobility in the transistor channel, and increase the dynamics of the threshold voltage to improve or extend different operating modes of the device. Addressing various improvements in transistor devices, including enabling such adjustments. The concentration of these dopants is a concentration graph starting from the top of the structure near the gate and going down through the different layers and into the screening layer, eg, as described above in FIG. 4 in relation to the channel depth of the device. It can be expressed as

The depleted channel region provides a region for electrons to move freely from the source to the drain of the transistor, thus improving mobility and overall performance. A threshold voltage adjustment region is used with the screening region to set a nominal unique threshold voltage for the device. The screening region is a heavily doped region that increases the body coefficient of the FET device. A higher body factor allows the body bias to have a greater effect in dynamically changing the threshold voltage of the FET. These three regions can be used in concert to achieve multiple specialized devices. Multiple combinations of two or three of these regions can be used to achieve various design benefits. For example, all of these regions are poly gates to achieve low power devices with various inherent _VT values (achieved by threshold voltage adjustment doping) and dynamic modes of operation (due to body effects). Or it can be used with a band edge metal gate.

  The channel region and screening region are used with a mid-gap metal gate stack to achieve an ultra low power device (the mid-gap metal functions to fully deplete the channel without the aid of a threshold voltage adjustment region). obtain. In other examples, the channel region and screening region can be used with a dual work function metal gate stack to achieve an ultra low power device. Also, the formation of these regions can be accomplished in multiple ways. In some implementations, a single epitaxial flow can be used so that in-situ doping that is controlled and modulated during growth yields the desired profile without further implantation. The profile can be achieved using multiple implants followed by an undoped epitaxial region. In other examples, a double epitaxial flow with implantation equivalent to the desired concentration can be used. Alternatively, a desired profile can be achieved using a plurality of epitaxial flows comprised of any number of epitaxial and ion implantation combinations. However, such variations do not depart from the spirit and scope of the appended claims.

In another example of the device, in addition to the DDC region formed on the substrate, an oxide region or other gate insulator may be formed on top of the substrate over the channel region. The device can include a metal gate region formed on the oxide region. The resulting device in this example is a transistor with a dynamically controllable threshold voltage while still insensitive to RDF in the channel region. In this example, in operation, the DDC region has a very low σV _T , and the low V _DD keeps the leakage in the deep depletion region low. Also, ion implantation may be used to enable legacy devices that require transistor operation above 1V.

  The following examples illustrate and describe various device configurations, systems incorporating such devices, and methods of manufacturing such devices and systems. These examples are described in a graphical manner that is well understood by those skilled in the art of such devices, systems, and methods of making them. These examples illustrate and illustrate the details of the device, along with a discussion of the feasibility of the underlying system and possible operating characteristics and performance.

Further comparison with the conventional structure is shown in FIGS. FIG. 10 shows the threshold voltage and body bias of a DDC transistor having a lightly doped (about 1 × 10 ¹⁷ atoms / cm ³ ) channel, and an equivalent sized conventional transistor having a uniformly doped channel without a screening region. An example of comparison is shown. As can be seen, although DDC transistors do not have significant channel dopants normally required for large body coefficients, the threshold voltage modulation due to body bias in DDC is still equivalent to a uniformly doped channel MOS. It is.

Thus, in certain embodiments, the DDC structure can provide benefits in short channel devices that are equivalent to those currently realized only in long channel devices that are not practical replacements for short channel devices. . Referring to FIG. 11, there is shown a comparison between a uniform channel MOS device and an example of a DDC device for σV _T versus body bias voltage. A significant degradation of the threshold voltage of the short channel device relative to the long channel device is shown. In this DDC device, even if the body bias voltage is increased, there is only a significantly small threshold voltage degradation. This reduction is driven by a heavily doped screening region that significantly suppresses the short channel effect.

  As described in the background art, certain transistors can be formed with a channel layer doped according to a super steep retrograde well (SSRW) profile. This technique uses a special doping profile to form a heavily doped region under a lightly doped channel. Referring to FIG. 12, a comparison between an example profile of a DDC structure and a conventional SSRW is shown. As can be seen, the SSRW has a very high dopant concentration adjacent to the channel, near the gate dielectric (not shown) that defines the top of the channel. Such a high doping concentration located near the channel and gate dielectric typically results in poor leakage performance in conventional devices and is serious for scaling this approach to nanoscale gate length transistors. There are many difficulties. Therefore, it does not provide a sufficient commercial solution to the overall power reduction and performance enhancement requirements in electronic devices. Embodiments of DDC transistors can include a channel that is deeply depleted and a screening layer that is heavily doped and spaced from the channel. Such a structure provides a significant improvement in circuit performance and can be simpler to manufacture than circuits that implement SSRW.

  A number of conventional CMOS fabrication processes can be used to fabricate the DDC transistor. FIG. 13 graphically illustrates a comparison 1300 of a conventional CMOS process for manufacturing a conventional device with a manufacturing process for a structure configured in accordance with embodiments disclosed herein. In one embodiment of the new CMOS device, shallow trench isolation (STI) 1302, 1302A, well channel implants 1304, 1304A, contacts 1308, 1308A, and metal interconnects 1310, 1310A are standard. obtain. Only the conventional CMOS gate stack process 1306 is different from the improved structure gate stack 1306A. This provides significant advantages for introducing new CMOS structures such as DDC devices. First, it eliminates the need to develop new processing steps with risk or cost to manufacture new devices. Hence, existing manufacturing processes and associated IP libraries can be used, thereby reducing costs and allowing manufacturers to bring these new advanced devices to market faster. The

  The DDC transistor process according to the example of FIG. 13 forms undoped epitaxial silicon regions on top of heavily doped N-type and P-type regions to create a DDC doping profile. The thickness of the undoped epitaxial silicon region is an important factor in device performance in some embodiments. In another example, double epitaxial silicon regions are used to achieve the final gate stack with high, medium and low doping (or no doping). Alternatively, an epitaxial silicon region is grown for the final stack with one heavily doped region close to the substrate level, and then in the epitaxial growth layer between the gate and the heavily doped screening region. A low to high concentration doping may be performed. To prevent dopant migration or diffusion between layers, in some implementations various anti-dopant migration techniques or layers may be used. For example, boron (B) diffusion can be suppressed by using carbon doping in P-type epitaxial silicon. However, in N-type epitaxial silicon, carbon can have an adverse effect on As doping. Carbon can be distributed throughout the silicon epitaxy or confined in a thin region at each interface. It is possible to use in situ doped carbon or ion implanted carbon. If in-situ doped carbon is used, the carbon can exist in both N-type and P-type. When carbon is implanted, in some embodiments it can only be used in P-type.

  The DDC transistor can be formed using available bulk CMOS process technology. Such techniques include techniques for depositing dopant resistant migration layers, advanced epitaxial layer growth, ALD, advanced CVD and PVD, or annealing. All of these are available in advanced integrated circuit process node technologies such as 65 nm, 45 nm, 32 nm and 22 nm. These process nodes generally have a low thermal budget for STI isolation, gate process and anneal, but are still suitable for forming DDC transistors.

  14A-14I show a process flow for manufacturing a device having a channel with a DDC doping profile. These figures show two devices to show how each of the NMOS and PMOS transistors with DDC and screening regions that provide advanced features and operation of the new DDC transistors and devices are configured. An example of the manufacture of is shown. A sample process for forming these two transistor devices will be described, progressively showing the structure in each step. In other examples, other process flows may be used to fabricate the DDC device, and this particular process and associated steps are shown for illustration. This process is described in relation to “regions” that are formed, deposited, or otherwise created to create a transistor structure, but regions of various shapes, sizes, depths, widths and heights. As well as layers of various forms or contours.

  First, referring to FIG. 14A, the structure 1400 starts with a substrate, for example a P-type substrate 1406. An NMOS device or a PMOS device can be created on this P-type substrate. In these and other figures, for simplicity and to illustrate possible embodiments and examples, this example of the process flow of a DDC device is shown as shallow wrench isolation to separate specific structures. And examples of NMOS and PMOS devices that combine partial trench isolation. Nevertheless, corresponding flows for other disclosed structures or devices will be readily understood. Although not shown, these processes are performed using various techniques known in the art, such as, for example, adjacent structures as different regions and masking used when forming regions stacked on each other. Can be executed.

  An N well implantation region 1402 and a P well implantation region 1404 are formed on the P substrate 1406 as necessary. Then, a shallow P well implantation region 1408 is formed on the N well 1402, and a shallow N well implantation region 1410 is formed on the P well 1404. These different regions can be formed by first forming a pad oxide film on P substrate 1406 and then performing a first N well implant of N well 1402 using a photoresist. Another photoresist may be used to implant the P-well 1404. Shallow N-well 1410 can be formed by ion implantation with another photoresist. A shallow PN well 1408 can then be implanted using another photoresist. This process can then be followed by an annealing process.

  Proceeding to FIG. 14B, the process continues with the formation of an NMOS RDF screening region 1412 on the shallow P-well 1408. According to this embodiment, the NMOS RDF region 1412 is a high dopant concentration screening region, for example, which provides many benefits of improved threshold voltage setting and reliability by suppressing RDF and transistor thresholds. Such as those described above with respect to enabling dynamic adjustment of the voltage. This screening region can be formed as an RDF screening implant region using another photoresist. A PMOS RDF screening region 1414 is formed on the shallow N-well 1410. This region can be formed as an RDF screening implant region for PMOS using another photoresist.

Subsequently, referring to FIG. 14C, after the initial oxide film is removed, an NMOS threshold voltage adjustment region 1416 is formed on the screening region 1412 using a photoresist. This threshold voltage adjustment region may be deposited using an epitaxial growth method or other similar technique. Similarly, a PMOS threshold voltage adjustment region 1418 is formed on the PMOS RDF screening region 1414 using a photoresist. Then, on each of the doped these threshold voltage adjustment regions on _{V T} adjustment area 1416 and PMOS _{V T} adjustment region 1418 of NMOS, region of undoped or lightly doped 1420, 1422 is deposited. Epitaxial growth or other similar techniques can be used to form these undoped or lightly doped regions. Through the above process group, a channel suitable for DDC is formed. In these examples, two epitaxial regions are used to create the desired DDC profile for each transistor, but instead a single epitaxial region may be used for each to create a DDC device.

  The above process flow prepares the device to be processed later to create two transistors or other more complex circuits by creating a channel. However, the following process probe discloses examples of the remaining steps to create n-channel and p-channel transistors as shown in FIGS. 14D-14E.

  Referring to FIG. 14D, an STI transistor boundary 1424 is then formed by applying a shallow trench isolation (STI) process to isolate the transistor from adjacent transistors. Here, the depth of each STI 1424, 1426, and 1428 is appropriately set so that the STI enters the P-well. As can be seen, the STI trench extends below each of the shallow P well 1408 and the shallow N well 1410. This allows for improved isolation between transistors.

  Further, if necessary, partial trench isolation (PTI) 1430, 1434 may be applied to create a region where the well taps can be connected. The depth of the PTI 1430, 1434 is set so that the PTI partially enters the shallow P-well. Thereafter, as shown in FIG. 14E, an insulator such as oxide regions 1438 and 1442 is deposited in the region where the channel is to be formed. Here, silicon dioxide can be used as an insulator, but other types of insulators can also be used. Then, gate electrodes 1436 and 1440 are attached to the respective gate insulators, enabling the gate voltage to be supplied during operation.

  Referring to FIG. 14F, spacers 1446 are formed on the side surfaces of the NMOS and PMOS gate and insulator regions. Then, source and drain regions 1448 and 1450 are ion-implanted under the wafer surface across each gate region. At this time, the source and drain regions 1448 and the source and drain regions 1450 are subjected to N-type and P-type doping, respectively. As described below, in legacy mode devices, NMOS and PMOS halo processes may be performed as needed. Also, body contact regions 1444 and 1464 are subjected to p + type doping and n + type doping, respectively, to create a contact to the body of the transistor. Thus, NMOS and PMOS transistors can be created and contacts can be provided to supply the necessary voltages to the source and drain regions to operate the device, as shown in FIG. 14G. FIG. 14G also shows a second spacer 1452 and source / drain connections to NMOS and PMOS 1448, 1450 using photoresist. Source and drain regions 1454, 1456, 1458, 1460 are formed. Thereafter, contacts and metal are formed using a photoresist to achieve electrical contact with the device. Depending on where this process places the source and drain, the electric field can be greatly affected.

  Although a specific process for manufacturing a DDC device has been described, other optional processes may be included to further improve the performance of the device or to meet various application specifications. For example, as shown in FIG. 14G, a leak current can be reduced by applying a technique known in the art as a source / drain extension. As will be appreciated by those skilled in the art, many combinations of different regions are possible, and combinations of regions may be rearranged or replaced using different regions in accordance with the teachings herein.

The doping level of the threshold voltage adjustment region and the screening region is limited to the region between the spacer edges below the channel. In one approach, the silicon is etched outside the spacer 1452 using a mask defined by a spacer around each gate 1436 and 1440 and a hard mask on the gate. The depth of the etched silicon is made larger than the depth of the screen area. In this example, silicon is etched by both NMOS and PMOS in the same or different processes. After silicon etching, silicon 1446 is epitaxially grown to a height slightly above the gate dielectric, as shown in FIG. 14H. The epitaxially grown silicon may be doped in situ or using a source / drain implantation mask to form the source / drain regions 1468, 1470, 1472 and 1474 shown in FIG. 14I. Also good. A first gate dielectric 1438 and a second gate dielectric 1437 are overlaid in layers. Layers 1435 and 1436 are metal gate electrodes designed to have a suitable N + or P + work function. In FIG. 14I, polysilicon is replaced by a metal gate electrode integrated with the gate dielectric. Replacing poly with a metal gate requires two separate metals with appropriate work functions. To adjust the _{V T} of the NMOS device and PMOS device equal to the N + / P + doped poly which is traditionally used in CMOS process requires a metal work function of ~4.2eV and ～5.2EV. A spacer 1452 around the gate and a hard mask over the gate form a self-aligned source / drain region. This results in smaller source / drain-body capacitance. In another approach, compensated source / drain implantation may be performed. In this approach, a spacer around the gate and a hard mask on the gate allows the gate to self-align.

  As will be appreciated, it is desirable to be able to operate a circuit efficiently in multiple power modes. In addition, being able to switch quickly and efficiently between different power modes significantly affects the power saving capabilities and overall performance of transistors and chips made with such transistors, and systems that implement such chips. Can be improved. The ability to efficiently change the operating mode allows the device to deliver high performance when needed and save power by entering sleep mode when inactive. According to one embodiment, individual subcircuits and even individual devices can be controlled dynamically. By dynamically changing the threshold voltage of the device, the mode of the device can also be dynamically changed.

  Deep depleted channel devices can have a wide range of nominal threshold voltages and can be operated using a wide range of operating voltages. Some embodiments can be implemented within the current standard bulk CMOS operating voltage range of 1.0V to 1.1V, and even lower operating voltages such as 0.3V to 0.7V. It is also possible to work with. These provide a circuit configuration for low power operation. Also, the DDC device can be more responsive than conventional devices due to its strong body effect. In this regard, the strong body effect may allow the device to bring about changes in the circuit by a substantial direct connection with other devices via a common shared well. In one example, the shared well may include a common P-well or N-well located under the group of devices. In operation, these devices can change modes by changing the setting of their body bias voltage and / or their operating voltage. This may make switching between a single device or one or more groups of devices much faster and using less energy than conventional devices. Therefore, dynamic mode changes occur quickly and the system can better manage power savings and overall system performance.

  Also, in some applications, backward compatibility with existing environments may be required so that a DDC-based device can operate integrally with a conventional device. For example, a mixture of a new DDC-based device and a conventional device may operate at an operating voltage of 1.1V. There may be a need to perform level shifting in order to interface DDC-based devices with conventional devices. It is highly desirable that DDC-based devices work with legacy devices.

The screen region provides a high body effect that is utilized for multi-mode switching with good responsiveness in the transistor. The response of a transistor having a screen region can change within a wider range with respect to changes in body bias. More specifically, the heavily doped screening region allows the device on and off currents to vary more widely under various body biases, thereby supporting dynamic mode switching. be able to. This is because DDC devices can be configured to have a lower σV _T than conventional devices, ie, a small variation in the set threshold voltage. Therefore, the threshold voltage V _T can be set to a plurality of different values. Furthermore, the device or group of devices may be body bias to vary the threshold voltage, therefore, it is possible to change V _T itself in response to changing the body bias voltage. Thus, a lower σV _T results in a lower minimum operating voltage V _DD and a wider range of available V _T nominal eigenvalues. Increased body effect allows for dynamic control of V _T within such a wide range.

  Furthermore, it may be desirable to configure the device to maximize performance as needed, even if maximum performance can result in increased power consumption. In another embodiment, it may be desirable to place the device in a significantly lower power mode (sleep mode) when the device is not in high performance active operating conditions. When using a DDC transistor in the circuit, a sufficiently fast switching time can be given for mode switching so as not to affect the overall system response time.

There are several different types of modes that may be desirable for a transistor or group of transistors configured according to the various DDC embodiments and examples illustrated and described herein. One mode is a low power mode in which the bias voltage V _BS between the body and the source is zero. In this mode, the device operates with a lower operating voltage V _DD and lower active / passive power than non-DDC devices, but with performance comparable to conventional devices. Another mode is a turbo mode in which the device bias voltage V _BS is forward biased. In this mode, the device operates with low _VCC and passive power tailored for high performance. Another mode is a sleep mode in which the bias voltage _VBS is reverse-biased. In this mode, the device operates at a lower V _CC and a low passive power. In legacy mode, the process flow is modified to allow non-DDC type MOSFETs to operate in substantially the same way as legacy devices.

  Devices with a DDC structure also allow enhanced dynamic mode switching as a result of the strong body effect made available by the screen area while having significant performance advantages over conventional devices. The body tap allows the desired body bias to be applied to the device to achieve the desired mode. This can be accomplished using a DDC having a lightly doped channel and screening region as described above, or alternatively using a DDC having multiple regions or layers with different dopant concentrations. When multi-mode switching is used for a group of transistors, such as memory blocks or logic blocks, for example, individual transistor control using conventional bulk CMOS technology is impractical and is not substantial to the control circuit. Overhead can occur. There is a need to implement additional dedicated control circuitry and large dedicated wiring to control different devices or groups of devices, all of which are significant additions to the total cost of the integrated circuit. It is therefore desirable to develop a partial circuit or unit that can be used to create a group or more groups of transistors for dynamic mode switching. It is also desirable to provide a solution that can provide body bias control technology to legacy devices so that legacy devices can also benefit from dynamic control, either independently or in a mixed environment. .

  In addition, while conventional bulk CMOS devices may require physical design changes, the relatively high body effect of transistors with screen regions can be specific, whether static or dynamic, depending on the design. In this embodiment, the transistor is made suitable with respect to using body bias as a means to control the device to operate in various modes.

A basic multimode device having a heavily doped screen region and a mechanism for applying a body bias voltage to the body is shown in FIG. 15, which duplicates FIG. 2A with a corresponding table showing different modes. Show. As described in relation to FIG. 2A, to control the electric field of the device containing the electric field between the source and the device body, may apply a bias voltage V _BS between the well tap and the source. FIG. 15 shows a sample structure of an n-channel four-terminal MOSFET. Terminal 106 is designated as the drain and terminal 104 is designated as the source. During operation, current flows between these two terminals. Terminal 102 is called the gate electrode, and a voltage that controls the current between the drain and source is often applied to this terminal. Terminal 126 provides a connection to the body of the transistor, which in this example is a P-well. The voltage applied to the drain is a positive power supply voltage and is called V _DD, and the voltage applied to the source terminal is the lower power supply voltage. The electric field affects the device characteristics. According to various embodiments described herein, the device can be set to a plurality of distinct modes by appropriately selecting the bias voltage V _BS and the power supply voltage V _DD .

In conventional bulk CMOS devices, the substrate is often connected to the source to maintain the same source body voltage. Thus, the body bias is typically the same for all devices on the substrate. This is similar to the situation where the DDC device is used in the normal low power / low leakage mode described above where a normal operating voltage and zero bias voltage (V _BS = 0) are applied. However, multi-mode devices configured in accordance with various embodiments described herein may provide effective mode control means instead of body taps. This is especially the case when the device contains a heavily doped screen region at a distance from the gate as described above. Unlike silicon-on-insulator (SOI) -based devices with low body effects, DDC-based devices can be constructed on bulk silicon to create devices with high body effects. Therefore, a device having a DDC configuration can use a variable body bias as a means for realizing multi-mode operation. The multimode transistor shown in the example of FIG. 15 can have an n-channel on the P-well. A P + type region is formed on the P well. A body tap (not shown), described below, is coupled to the P + region to create a conductive contact to the P well that is the body of the n-channel device. Since the body tap is p + doped, the connection to the body tap will provide a connection to the P-well of the device (ie, the body of the device). A body bias voltage can be applied between the source and the body tap. The body bias voltage can efficiently control the operation mode of the n-channel device. As in n-channel devices, this dynamic mode switching technique can also be applied to p-channel devices on N-wells (where n + regions are formed to provide body taps). The new structure with strong body bias described herein is also applicable to CMOS devices where both n-channel and p-channel devices exist on the same substrate or well.

A body bias voltage applied between the source and the body can effectively change the behavior of the CMOS device. In the devices described above having body taps, the source-body voltage can be applied independently of the gate-source voltage and the drain-source voltage. One advantage of using body bias as a control means for multi-mode control is that the device is a conventional device, such as a device in which the gate-source and drain-source voltages are set the same. It can be connected. In this case, mode selection can be made in response to body bias. Hence, the device can be operated normally with zero bias, which is the same as a conventional device. When a higher performance mode (turbo mode) is desired, a forward bias voltage, ie V _BS > 0, can be applied between the well tap and the source. The turbo mode operating voltage may be the same as or slightly higher than the normal mode operating voltage. On the other hand, when the sleep mode is desired, a reverse bias voltage, ie, V _BS <0, can be applied between the well tap and the source. The sleep mode operating voltage may be the same as or slightly lower than the normal mode operating voltage.

  When a zero body bias is applied, the multimode device is operated in a normal low power mode. The body bias can be forward biased by applying a positive voltage between the body and the source as shown in the example of FIG. 15 to enhance device performance. This forward bias mode is referred to as “turbo mode” for enhanced performance in the form of high drive currents. However, enhanced performance comes at the expense of increased leakage current. In the deep sleep mode, in order to suppress the leakage current, a negative voltage is applied between the body and the source as shown in the example of FIG. 15, and the body is reverse-biased. This mode is desirable when the device is in an idle or inactive state.

FIG. 16 shows a comparison of threshold voltage V _T versus bias voltage V _BS between an example n-channel DDC device and a conventional n-channel device. Curve 1610 represents a DDC device and curve 1612 represents a conventional device. FIG. 16 shows that in some implementations, the threshold voltage of the DDC device is much more sensitive to the bias voltage than the conventional device. A DDC device can also provide a wide delay range in response to body bias. In the conventional device, the variation in the threshold voltage between the devices causes a large spread of the delay time as shown in FIG. 17A. Band 1702, 1704 and 1706, respectively, represent -0.5 V, the variations in delay for the bias voltage _{V BS} of 0.0V and + 0.5V. However, the delay time is shown in a relative scale in which the delay time of the conventional device at V _DD = 1.1V, V _BS = 0.0V, σV _T = 0.0V and temperature = 85 ° C. is normalized to 1. Yes. The horizontal axis corresponds to the 3σV _T value. The σV _T of conventional devices is typically around 15 mV, which results in 3σV _T = 45 mV. As shown in FIG. 17A, the three bands 1702, 1704, and 1706 have substantial overlap, which makes it difficult to distinguish modes according to delay time. FIG. 17B shows the improved delay time in the example of the DDC device. In FIG. 17B, the three bands not only have an overlap, but have a much smaller spread. At three different bias voltages -0.5V, 0.0V and + 0.5V (reverse bias, zero bias and forward bias), the DDC device exhibits three easily distinguishable bands 1708, 1710 and 1712. These distinguishable bands indicate that the DDC device is very effective for use in multiple modes of operation in some embodiments.

Another benefit that a transistor that can provide a reduced σV _T and thus a more accurately controllable V _T is that V _T can be dynamically controlled. In the conventional device, since σV _T is very large, V _T needs to be considered over a wide range. According to embodiments described herein, it is possible to dynamically change the V _T by adjusting a body bias voltage. The dynamic adjustment of V _T is achieved by an increased body effect and the range of dynamic control is given by the reduced σ V _T. Referring to FIG. 18, a V _T0 is a static V _T that is set for the device, the device is an example of a graph showing a plurality of V _T that can be adjusted is shown. Each has a corresponding ΔV _T , ie, a separate σV _T for each corresponding V _T value. According to embodiments described herein, the device is configured to have a dynamically adjustable V _T by adjusting a body bias voltage within and at a suitable voltage adjustment speed required voltage range be able to. In certain embodiments, the voltage adjustment can be made in predetermined increments or can be made continuously variable.

  According to another embodiment, FIG. 15 shows a sample of a multimode device that is likely to operate under various modes, but for the device to isolate the body for a group of transistors. It is also useful to include a structure. This allows the device to effectively operate independently under various modes. If the bodies of a group of multimode transistors are connected, the entire group will be switched simultaneously, limiting the ability to facilitate mode switching. On the other hand, if the bodies of the two groups of multimode transistors are not connected, the two groups can be controlled individually. Thus, the basic multi-mode transistor shown in FIG. 15 can further provide a group of transistors that can be divided into a number of blocks using individual body biases for each group. This will be described later.

  Thus, an improved system can be constructed utilizing a DDC structure, such as the transistor structure described above with reference to FIGS. 14A-14I, for example. Variations on such structures can be implemented in integrated circuits and systems that have strong performance advances. It has been shown how these structures can be configured to scale transistors. Here we show how these structures are used as building blocks for extending to wider integrated circuits and systems. For example, using DDC structures, STI, PTI, shallow wells and / or shared wells integrated into integrated circuits and systems can be configured for new and improved system performance. Aside from DDC, new innovations utilizing body taps and / or body access transistors can be used to realize new functions and benefits of integrated circuits and systems. Thus, using these innovations in bulk CMOS and other new structures and processes, newly scaled integrated circuit chips with greatly improved operation can be constructed.

The transistor embodiments described above may generally provide subsequent power scaling of bulk CMOS transistors and other devices, but fully exploit some of the benefits and features of the DDC structure, especially at the chip level. Those wishing to do so can do so by appropriate changes in the layout and routing of circuit blocks on the chip according to the transistor embodiments described herein. For example, as mentioned above, the concept of dynamically adjusting the transistor body bias voltage to adjust the threshold voltage of the transistor is known, but generally not practical for implementation in nanoscale devices. Not proven. This is because, in some implementations, (1) the large σV _T of conventional bulk CMOS nanoscale devices does not achieve sufficient differentiation between transistors for existing nanoscale devices; The relatively low body factor of the bulk CMOS nanoscale devices of SMC cannot switch between operating modes quickly enough to avoid impact on chip operation; and Routing to transistors or each circuit block includes significantly reducing the number of transistors that can be integrated on a chip, thus preventing scaling at the chip level. Some DDC transistor embodiments provide the first two problems: (1) significantly reduced σV _T so that the same transistor operates with different operating voltages as well as different threshold voltages And / or (2) providing a significantly increased body factor that allows transistors and circuit blocks to switch between operating modes quickly and efficiently; Can be solved by. A DDC transistor, in some embodiments, may be partly or entirely have the same structure and characteristics nominally but independently to operate as a transistor that had to be manufactured differently in conventional bulk CMOS. It can be treated as a field programmable transistor (FPT) such as a chameleon that is configurable. Improved routing of body bias wiring is another element in the following discussion, which also provides a further example of how multi-mode transistors can be used.

FIG. 19 shows a simplified concept of multi-mode operation of a group of transistors, where each block or circuit can operate in different modes based on the body bias voltage and operating voltage supplied to it. In some implementations, applying different body biases to individual blocks can dynamically adjust their threshold voltages so that commonly connected components operate in a common mode and are separately It may be possible to control the system to allow connected components or systems to operate in a separately controlled mode. In the exemplary situation shown in FIG. 19, the device 1900 is divided into five groups of transistor or circuit blocks 1910, 1920, 1930, 1940 and 1950 having separate body bias contacts. According to the embodiments described herein, the bodies of these five circuit blocks are separated from each other so that different body biases can be applied to each block independently. In this example, each of these circuit blocks has its own body separated from the other groups, which are connected to their respective body taps (1915, 1925, 1935, 1945 and 1955). These five blocks are intended to illustrate the need to provide isolation between groups of transistors in order to create isolated blocks. FIG. 19 also shows that each block is connected to a respective body bias V _B1 , V _B2 , V _B3 , V _B4 and V _B5 . As will be appreciated by those skilled in the art, each block also requires other supply voltages such as, for example, drain V _DD , source V _SS , gate V _G, and other signals. In addition, different operating voltages V _DD may be applied to each circuit block separately. The mode of each circuit block can be set statically by design (eg, by connecting different circuit blocks to different body bias voltages to set operating modes independently of each other) and / or operate The body bias and / or operating voltage of each circuit block can be set dynamically through a control circuit and algorithm that adjusts to set its operating mode. Due to the low σV _T and the ability to adjust the threshold voltage V _T over a relatively wide range of values, the operating mode of individual transistors or groups of transistors can be controlled separately.

  In the following examples, various transistors are described. These transistors are used as building blocks for forming one group of transistors into a plurality of blocks having isolated bodies. For example, referring again to FIG. 14G, one embodiment of a pair of CMOS transistors configured to have a new DDC structure is shown, these transistors having body taps, and these n-channel devices and The p-channel device is on the same substrate. These structures can be used to develop greatly improved performance circuits and systems, including the embodiments described below. Other transistors may be used in combination with a transistor with a new DDC structure, and some of the embodiments here may be configured without using a transistor with a DDC structure.

  FIG. 20 shows an example of an n-channel four-terminal transistor layout having a well structure with a single P well 2060 on a P substrate 2080. This four-terminal transistor layout 2000 shows source / drain pairs 2020 and 2030, a gate 2040 and a body tap 2050. A cross-sectional view at location 2010 is also shown, where the shallow trench isolation (STI) 2070 depth is less than the depth of the P-well. P well 2060 is common to all n-channel transistors on P substrate 2080. Therefore, this four-terminal transistor does not need to provide isolation between n-channel transistors. As shown in this example, the body tap is P + (P-plus; PP) -doped and arranged next to the transistor (based on the gate direction shown). In addition, the body tap is isolated from the transistor by STI 2070.

  FIG. 21 shows an example of an n-channel four-terminal transistor having a new shallow P-well (SPW), where the depth of SPW is made smaller than the depth of STI. This four-terminal n-channel transistor layout 2100 shows source and drain pairs 2020 and 2030, a gate 2040 and a body tap 2050. Cross-sectional view 2180 shows position 2110 and cross-sectional view 2190 shows position 2112. Shallow wells allow body isolation, and thus in certain implementations, for example, dynamic mode switching of groups of devices such as memory cells or other digital circuits, and thus the body that must be routed on an integrated circuit It is possible to reduce the number of bias voltage wirings. As shown in cross-sectional views 2180 and 2190, the transistor has a shallow P-well 2160 on a complementary N-well 2164. Due to the pn junction, the N well 2164 is not conductively connected to the shallow P well 2160, and the N well is not conductively connected to the P substrate 2080. Thus, this transistor can be isolated from other n-channel transistors having a shallow P well 2160 on an N well 2164 on the same substrate. The active area extends under the gate. A minimum active critical dimension (CD) is used for the active part extending under the gate. Extended active edges may be placed between the spacer edges to avoid shorting due to silicidation. The body contact can be formed on an active region that extends outside the gate. The edge of the N + implant region may be under the gate extension (end cap) region. Although this example illustrates one approach to making an n-channel four-terminal transistor, this layout can also be applied to create a p-channel four-terminal transistor. As shown in FIG. 21, in some implementations, the STI can be deeper than the SPW. In some embodiments, if two adjacent transistors do not have a common SPW, they can be biased independently of each other. In other examples, adjacent groups of transistors may have a common SPW and can be operated in the same mode by applying the same body bias.

  In yet another embodiment of the dynamic multimode transistor, a body access transistor may be formed between the actual transistor and the body tap, as shown in FIG. FIG. 22 shows an n-channel four-terminal transistor layout 2200 and associated cross-sectional view 2280, with a shallow P-well (SPW) 2160 isolated by STI 2070. The body access transistor can isolate the body tap from the transistor. The body access transistor can be created as if the gate 2041 functions as the gate of the body access transistor and the body tap is treated as a source / drain. This can simplify the process and reduce the area required to form the body tap connection. The use of body access transistors in combination with shallow wells is a useful building block to allow dynamic mode switching with fine granularity. With respect to a group of transistors or circuits that are switched together, they can be arranged to share the same shallow well. Also, one or more gate taps can be created by providing a connection to the body using a body access transistor and providing a body bias.

  As described above, partial trench isolation (PTI) is another suitable technique for isolating a body tap from a transistor. According to another embodiment shown in FIG. 23, an exemplary layout 2300 and cross-sectional view 2380 of an n-channel four terminal transistor includes a shallow P-well (SPW) and a partial trench isolation (PTI). Cross-sectional view 2380 corresponds to a cross-section at position 2310. The SPW depth can be smaller than the STI depth. The PTI oxide can prevent a silicide short circuit between the n-type source / drain and the p-type bulk tap. The depth of the PTI can be smaller than the depth of the shallow well so that the continuity of the shallow well in the transistor is maintained. The PTI approach can provide excellent protection against possible silicide shorts between the body tap and the source / drain in some implementations. However, PTI requires one or more additional process steps in the manufacture of the device. The depth of the PTI is preferably deeper than the source / drain junction in some embodiments so as to minimize N + / P + junction leakage by separating the P + bulk tap and the N + source / drain. .

  The relative planar positions of the source / drain active region and the well tap active region may be arranged differently to create a four-terminal transistor 2400 having a PTI as shown in the example of FIG. Cross-sectional views 2480 and 2490 correspond to locations 2410 and 2412, respectively. As shown, the shallow P-well is isolated by STI.

  The above example shows a four-terminal transistor that provides a body tap for applying a body bias voltage. However, there is a situation where the fourth terminal for body bias may be unnecessary. For example, when a CMOS transistor has a shallow P well and N well on a common N well, a p-channel transistor having a shallow N well on the N well will always have a common N well. In such an implementation, it may not be necessary to provide a separate fourth terminal that connects to the body. Accordingly, here are some examples of three-terminal transistors that can be used as building blocks to create a group of transistors having multiple blocks with isolated bodies. In another situation, a transistor may have a shallow well on a complementary well and the transistor is intended to operate with the body floating. In such an implementation, it may not be necessary to use the fourth terminal.

  As shown in FIG. 25, in the three-terminal structure 2500 according to an example, the local interconnect connects the gate and the body in order to reduce the number of terminals from four to three. Cross-sectional views 2580 and 2590 correspond to locations 2510 and 2512, respectively. In cross-sectional view 2580, local interconnect (LI) contact 2551 is used to connect the body contact to the extended gate. In this example, the gate-body contact is made on the extended active region using a metal contact. Rectangular contacts used in SRAM cells can also be used to connect the gate to the body.

  In yet another embodiment, a three terminal dynamic multimode transistor is formed by using a poly-underbody contact. The oxide film under the gate is removed using a GA (Gate to Active) contact mask. On this gate dielectric removal region, a polysilicon gate contact (PGC) implant having the same polarity as SPW may be performed. As shown in structure 2600 of FIG. 26, the use of PGC 2650 connects the body to the gate. Cross-sectional views 2680 and 2690 correspond to locations 2612 and 2614, respectively. This layout scheme has several potential advantages, including the ability to form self-aligned gate contacts to the body and / or the ability to perform self-aligned GC (gate contact) implants. Can do. Since the GC implant can have the same polarity as SPW (P + doping), in some embodiments, there can be no bending in the active region, which is a design-for-manufacturing; DFM) friendly. Using PGC for connection can result in higher contact resistance to the body. However, in static mode control in some embodiments, contact resistance is not critical. Therefore, the PGC may be used when static control is required.

  In another example, the body contact can be formed in an active region extending under the gate extension, similar to the three terminal single gate transistor 2700 shown in FIG. Cross-sectional views 2780 and 2790 correspond to locations 2712 and 2714, respectively. A minimum active critical dimension (CD) may be used for the extended active portion. The extended active edges can be placed between the spacer edges of the active area under the gate. The oxide under the gate can be removed using a GA contact mask. On the region where the gate is removed, a GC implant having the same polarity as SPW is performed, and the body can be coupled to the gate using the body. In some implementations, this approach includes the ability to use a self-aligned gate contact to the body or a self-aligned GC implant because the GC implant has the same polarity as SPW (P + doping). Similar benefits can be provided.

  As shown in the example of FIG. 27, the gate contact and the well tap contact can be at different locations along the poly, but they are placed at the same location as shown in structure 2800 of FIG. You may be. Cross-sectional views 2880 and 2890 correspond to locations 2812 and 2814, respectively.

  In another embodiment, the layout allows for programmable four-terminal / three-terminal transistors. As shown in structure 2900 of FIG. 29, the gate and body are cut or connected using metal region 2950, resulting in four or three terminals, respectively. Cross-sectional views 2980 and 2990 correspond to locations 2912 and 2914, respectively. Thus, the metal area connection facilitates a programmable four-terminal / three-terminal transistor layout.

  Although various transistors have been described, in many instances the different structures described in the various embodiments and examples can be combined in different combinations and to create a useful system with improved performance over conventional systems. It may be used in the foundation structure. These transistor structures can also be used as building blocks to create transistor groups that are divided into multiple blocks and have individual body bias connections for dynamic mode switching. Some examples are described below.

One of the advantages of transistors configured in accordance with some of the embodiments described herein is that dynamic mode switching is possible. This can be achieved by applying a controlled body bias voltage to set or adjust the variable operating voltage. FIG. 30 shows an example of a circuit 3000 that can perform dynamic mode switching using a four-terminal transistor. In this figure, various bias voltages and operating voltages are shown. The circuit blocks a1-a4 correspond to a standard mode, a low leak mode, and two turbo modes, respectively. Each of these circuit blocks is a pair of four-terminal transistors, that is, p-channel four-terminal transistors 3010 and n with four terminals designated as S (source), D (drain), G (gate), and B (body). A channel four terminal transistor 3020 is used. In block a1, a four-terminal transistor having a body tap is used as a conventional transistor. n channel body of the device (the lower side of the transistor shown) is coupled to the source voltage V _SS. The body of the p-channel device (the upper transistor shown) is connected to the operating voltage V _DD . In block a2, the device is reverse biased to achieve low leakage when the device is not actively used. This reverse bias connects the body of the n-channel device to an n-channel reverse bias voltage V _BBN lower than V _SS and connects the body of the p-channel device to a p-channel reverse bias voltage V _BBP higher than V _DD. Can be achieved by connecting to If higher performance is desired, the device can be placed in a forward bias state as shown in blocks a3 and a4. In a3 (i), the p-channel body and the n-channel body are connected to dedicated forward bias voltages V _FBP and V _FBN , respectively. However, V _FBP is lower than V _DD and V _FBN is higher than V _SS . In another example, source and drain voltages can be used for forward bias to reduce system cost by eliminating the additional power supply required for forward bias voltage. As shown in a3 (ii), the p-channel body is coupled to V _SS and the n-channel body is coupled to V _DD . The circuits a4 (i) and a4 (ii) are similar to the circuits a3 (i) and a3 (ii) except that the high operating voltage V _DDH is connected.

As shown in FIG. 31, there are other variations in using a four-terminal device in a dynamic switching environment. In FIG. 31, a circuit block a1 shows a situation in which the body of the four-terminal device is left unconnected to make the body floating. There are two versions of the floating body 3100 shown in FIG. 31. The sub block a1 (i) uses V _DD as the operating voltage, and the sub block a1 (ii) uses V _DDH as the operating voltage. This will deliver intermediate performance. In the circuit block a2, the body and drain of the p-channel device and the n-channel device are all coupled together to achieve the turbo mode. According to one embodiment described herein, the same dynamic mode switching function is extended to a larger circuit having a larger number of transistors.

  FIG. 32A shows an implementation example of dynamic mode switching using a simplified case. FIG. 32A shows a circuit 3200 in which two circuit blocks 3220 and 3230 have an isolated body so that independent body biases can be applied. The body bias of the circuit block 3220 can be applied via the body contact 3225, and the body bias of the circuit block 3230 can be applied via the body tap 3235. A power supply rack for other voltages, such as that shown in FIG. 30, is not shown. However, the implementation of the system power rack in FIG. 32 will be readily understood by those skilled in the art. An exemplary cross section 3250 of such a circuit block is shown in FIG. 32B. FIG. 32B shows an n-channel device having shallow P-wells 3260 and 3261 over N-wells corresponding to circuit blocks 3220 and 3230. Shallow P-wells 3260 and 3261 are isolated between the two circuit blocks by STI 3263 to create separate shallow wells for the two circuit blocks. The two shallow P-wells 3260 and 3261 are not connected by an N-well 3264 located on the P-substrate 3266 below it due to the pn junction effect. Body access transistors are used to create taps and to isolate taps from active transistors that share an SPW well. A p-type contact region 3210 is used for a body contact that provides a connection to a shallow P-well. The example of FIG. 32B illustrates the use of a shallow channel STI 3262 along the body tap to create multiple isolated circuit blocks for dynamic mode switching. Although this example is shown for an n-channel device, it can easily be applied to a p-channel device.

  This can also be extended to the device 3300 shown in the example of FIG. 33A, which combines a p-channel device and an n-channel device in structure 3310. FIG. 33B represents a situation where the CMOS device has two shallow P-wells 3260, 3261 and a shallow N-well 3360. These shallow wells 3260, 3261 and 3360 include body contacts 3325, 3335 and 3345, respectively. These shallow wells are all on the N well 3264. Three circuit blocks are shown, with circuit block 3320 and circuit block 3330 being an n-channel device and circuit block 3340 being a p-channel device. Each of these circuit blocks can share the same N-well 3264. Due to the pn junction effect, the shallow P-wells of circuit blocks 3320 and 3330 can always be isolated from the p-channel device in some implementations. There may be more than one p-channel circuit block. However, since the shallow N-well is always connected to the underlying N-well, each of the p-channel devices can have the same body bias. Thus, in some applications, a shallow N-well, such as 3360, for p-channel devices cannot share a common N-well with other shallow N-well devices. In such applications, when a common well is used, the N-well device cannot be divided into a plurality of isolated shallow wells. Therefore, from the viewpoint of dynamic power mode switching, it may not be necessary to form a separate circuit block for the p-channel device. In some embodiments, in a single N-well situation, only n-channel devices can be controlled separately via a body bias mechanism. When the underlying transistor is configured to have a high body effect as described herein, using body bias can be an effective way to facilitate dynamic mode switching. For p-channel devices, the shallow N-well in the N-well is optional.

  The following figures illustrate a number of example circuits that can be used as building blocks of an integrated circuit according to embodiments described herein. These circuits can be formed using a number of methods and structures. The description will begin with an example using the process and structure of some building blocks currently used in the industry. The figures that follow illustrate examples using building block processes and structures that greatly improve the prior art.

  FIG. 34A shows an example of a circuit configured using different circuit elements that are generally used and will be used in the following diagram showing an implementation example of dynamic mode switching. FIG. 34A shows a coupling circuit 3410 having a NAND gate NAND2 3402, an inverter INV3404, and a body tap TAP3406. These useful structures can be used in accordance with various embodiments disclosed herein to provide a better configured and useful circuit with new enhancements.

  In FIG. 34B, layout 3420 shows a conventional approach for implementing a group of transistors, using dummy poly 3428 to create taps 3427 and 3429 in each well. This body tap provides a common well or substrate connection for all devices. FIG. 34B shows a body tap extending into the well. The lower part of the layout shows a device portion mounted in an n channel having a shallow P well on an N well. Shallow P-wells are isolated from adjacent devices by STI because their depth is less than that of STI. The upper part of this layout shows a device portion mounted in a p-channel having a shallow N-well on a P-well. Again, the shallow N-well is isolated from adjacent devices by STI. Since two separate wells (P-well and N-well) and each shallow well are used, a completely complementary device allows separate individual dynamic control for n-channel and p-channel devices. . In FIG. 34B, which includes NAND gate NAND2 3422, inverters INV 3424 and TAP 3426, the top and bottom of the device have their respective body connections 3427 and 3429. The lower part of the layout shows a device portion mounted in an n-channel having a shallow P well on the P well. The upper part of this layout shows a device portion mounted in a p-channel having a shallow N well on the N well. FIG. 34C, which includes NAND gate NAND2 3422, inverter INV 3424, and body access transistor TAP 3426, is similar to FIG. 34B, except that single body taps 3437 and 3439 are implemented based on the new body access transistor. These new body access transistors provide a new configuration that allows access to the body of the transistor. Unlike conventional device designs, these structures provide significant operational capabilities for devices and circuits.

  FIG. 34D shows an example of a circuit layout 3440 including a NAND gate NAND2 3422, an inverter INV3424, and a body access transistor TAP3446. Circuit layout 3440 uses body access transistor 3450 to create two body taps 3437 or 3439 separated by STI to achieve connection to the respective wells. In FIG. 34D, the connection to the body is realized using the body access poly. A body access transistor with two separate body taps is isolated by STI. That is, the left and right sides of the STI have isolated shallow wells, and individual body biases can be connected to the left and right sides. FIG. 34E shows cross-sectional views 3490 and 3495 corresponding to locations 3482 and 3484, respectively. In cross-sectional view 3490, there is an n-channel transistor (eg, 3460) on a shallow P-well 3462 that is isolated on both sides by STIs 3464 and 3465. Shallow P-well 3462 is on N-well 3466, which is on P-substrate 3468. A body tap 3439 is connected to the shallow P well 3462. The top of device 3440 includes a p-channel transistor (eg, 3470) on shallow N-well 3472 isolated by STI 3474 and STI 3475. Shallow N-well 3472 is on P-well 3476, which is on the same P-substrate 3468. A body tap 3437 provides a connection to the shallow N-well 3472. Device 3440 is implemented using fully complementary transistors with isolated shallow wells (3462 and 3472) with separate body taps (3439 and 3437) for dynamic mode control. An example of the form is shown.

  FIG. 34D shows a dynamic mode switching implementation based on a transistor configured to have a DDC, but dynamic mode switching is also applicable to a mixed environment with legacy and new devices. FIG. 35 illustrates one implementation using a mix of legacy and new devices for the same circuit configured with NAND gates NAND2 3502, INV3504 and TAP3506 including STIs 3524 and 3534 separating shallow wells. Again, both N-well and P-well are used. However, both NAND2 and TAP are implemented using a legacy approach where the shallow well is on the same doped well. NAND2 3502 and TAP 3506 always have a common well on either the N-well or P-well. Therefore, the shallow wells for NAND2 3502 and TAP3506 cannot be isolated by STI. This configuration only allows the shallow well for INV 3504 to be isolated. Depending on the design, the body of the INV 3504 may be floated (ie, no body taps are connected to each shallow well, or body taps are not connected) or connected to body bias. Also good. However, since two separate wells are used, two separate body bias voltages can be applied to the n-channel device on the P-well and the p-channel device on the N-well.

  FIG. 35 also shows cross-sectional views 3550 and 3560 at locations 3510 and 3512, respectively. Cross-sectional view 3550 shows both n-channel transistors and taps 3516 on shallow P-wells 3522 and 3521. Shallow P-wells 3522 and 3521 are both on P-well 3526 and P-well 3526 is on P-substrate 3528. Body tap 3516 provides a connection to the body of the n-channel transistor. The shallow N-well 3532 for the p-channel in the lower part is isolated and left floating. Cross-sectional view 3560 shows both p-channel transistors and taps 3514 on shallow N-wells 3533 and 3535. Both shallow N-wells 3533 and 3535 are on N-well 3536, which is on P-substrate 3538. Body tap 3514 provides a connection to the body of the p-channel transistor. The shallow P-well 3523 for the n-channel in the upper part is isolated and left floating. A body tap for a p-channel device in the shallow N-well 3532 and a body tap for an n-channel device in the shallow P-well 3523 may be added using the body access transistor described above.

  FIG. 36 shows an example implementation based on the legacy approach where two separate wells are used. The n-channel transistor is on a shallow P-well 3622 isolated by STI 3623 and 3624. Because this shallow P-well 3622 for all n-channel transistors is above P-well 3626, shallow P-well 3632 is to be isolated from the adjacent circuitry between STI 3624 and STI 3625. This is because the P-well provides a connection between n-channel transistors on other shallow P-wells. Both P well 3626 and N well 3636 are on deep N well 3628, and deep N well 3628 is on P substrate 3630. Body access contacts 3612 and 3614 are also shown.

  The above examples show implementation examples of various dynamic mode switching using bulk CMOS. However, the new body bond design is also applicable to semiconductor devices that use non-bulk CMOS devices. For example, the body tap can also be formed on a partially depleted (PD) SOI technology as shown in FIG. FIG. 37 includes NAND2 3722, INV3724, and TAP3746. Circuit 3700 is similar to FIG. 34D, where body access transistors are used to create separate body taps 3712 and 3714. FIG. 37 also shows cross-sectional views 3740 and 3760 corresponding to the layout along positions 3716 and 3718. The bottom of circuit 3700 is for an n-channel device on P-well 3744 isolated by STI 3743 and 3745. Therefore, it is possible to form a plurality of isolated P-wells on the SOI so that a body bias can be independently applied to each circuit block. The top of circuit 3700 is for a p-channel device on N-well 3764 isolated by STI 3747 and 3749. Therefore, it is possible to form a plurality of isolated N-wells on the SOI so that a body bias can be independently applied to each circuit block. Both the P well 3744 and the N well 3764 are on the buried oxide film (BOX) 3748. This structure facilitates separately biasing a group of transistors or associated switchable devices in accordance with various embodiments described herein.

  Inside or coupled to various digital processors, such as a central processing unit (CPU), microprocessor / microcontroller, digital signal processor (DSP), field programmable gate array (FPGA) and other devices, Static random access memory (SRAM) is widely used. Several device structures are widely used in industry. Among them, 6T-SRAM (6-transistor SRAM) cells are most often used. This is because the 6T-SRAM cell can be mounted using a general-purpose CMOS process. It can therefore be easily embedded in a digital processor. Utilizing the new structure described above, an improved SRAM with better performance and reduced circuit area can be constructed. By using new body taps, body access transistors, and / or new DDC structures, significantly improved SRAMs can be manufactured using well-known processing equipment and equipment. Also, some of these SRAM circuit embodiments may be fabricated using new DDC-structured transistors and other types of transistors in combination with new DDC-structured transistors. Some of the embodiments can now be configured without DDC-structured transistors while still benefiting from improved SRAM performance and functionality.

In one embodiment, a basic 6T-SRAM cell includes two pull-up (PU) transistors and two pull-down (PD) transistors that store one bit of data, and two pass gates that control the bit line and the inverted bit line. (PG) transistor. An example of this is shown in structure 3800 of FIG. Pass transistor switching can be controlled by the word line to allow the design of SRAMs with low operating power consumption and low leakage current. In the example of the 6T SRAM of FIG. 38, the PU transistor is mounted using a p-channel four-terminal transistor 3010, and the other is mounted using an n-channel four-terminal transistor 3020. FIG. 38 also shows various signals and power supplies for 6T-SRAM including word line (WL), bit line (BL), bit line negation (BLN), V _SS and V _DD . FIG. 38 also shows that a connection can be provided to the body of the n-channel transistor (shallow P-well, SPW) and the body of the p-channel transistor (N-well, NW).

Memory access can consume a significant amount of power in an electronic system. Efforts have been made to develop implementations and systems that reduce power consumption during memory access and data retention. SRAM is typically used for program and data storage in computer systems. During program execution or data access, some of the memory may be actively accessed and other parts may be idle. It would be advantageous if the SRAM operating mode could be dynamically switched with fine granularity. In one embodiment, the body of each cell can be structurally isolated so that the bias of the cell can be individually controlled. In practice, a group of cells in a row can be controlled together by connecting the source voltage of the row. In addition to the 6T SRAM mode switching control and Body tapped and the body access transistors based on the above V _SS, which is another method to produce a multimode SRAM. This approach can be implemented for use in SRAM, for example by dividing the shallow well diffusion of the block of cells using body access transistor technology. A desired body bias can be selectively applied to a block of SRAM cells via a body tap to determine a desired mode of operation.

  An example embodiment is provided that uses component building blocks to create a dynamic multi-mode SRAM array. These blocks include various four terminal transistors, three terminal transistors, and programmable three / four terminal transistors. By combining these building blocks together with various body connection structures, an improved SRAM circuit that operates more efficiently can be constructed. For example, a body access transistor can be formed by converting the poly on the STI to a transistor while treating the body tap as one of the source / drain pairs. Body access cells can be added to the surrounding region to isolate the shallow wells of the SRAM array so that a body bias can be individually applied to the SRAM array. In the following, an example of a 6T SRAM implementation and accompanying body access transistors is described along with the process of connecting SRAM cells and body access cells to create a dynamic multi-mode SRAM array.

  FIG. 39 shows a layout example of the 6T SRAM of FIG. The 6T SRAM cell includes six transistors, PG indicates the position of the pass gate transistor, PD indicates the position of the pull-down transistor, and PU indicates the position of the pull-up transistor. The PD transistor and the PG transistor are n-channel transistors and are formed in an N + (NP) implantation region 3910, and the PU transistor is a p-channel transistor and is formed in a P + (PP) implantation region 3920. N-channel transistors are formed on shallow P-well 3940 and p-channel transistors are formed on N-well 3950. The use of a shallow N well within N well 3950 is optional in this embodiment implementation. The signal wiring and the power supply wiring are shown in FIG. 38 and other drawings.

  A cross section of one preferred layout example of SRAM cell structure 3900 is shown in FIG. 40A. The sectional view 4010 corresponds to a straight line 4015 where the PG transistor and the PD transistor are located. Additional PG and PD transistors are disposed on the other end side of the SRAM cell and have a similar cross-sectional view. Cross-sectional view 4010 also shows that these transistors have a shallow P-well 3940 over N-well 4040. The N well is on a P-type substrate 4050. Cross-sectional view 4020 corresponds to straight line 4025 where the PU transistor is located. This cross-sectional view shows that the PU transistor has a shallow N well 3950 on the N well 4040. The shallow N-well 3950 of the p-channel transistor is on a well having the same dopant type (N-well). Therefore, the shallow N well and the N well can be conductively connected. The shallow N well in the N well is as needed. However, in an n-channel device, the shallow P-well 3940 can be isolated from the underlying N-well 4040. A 6T SRAM cell corresponding to FIG. 39 viewed in 3D is shown in FIG. 40B with a well structure and transistor type.

  FIG. 41A shows an example of a top view of one preferred well structure (the N well is not shown because it extends the entire cell region). In the 6T SRAM layout of FIG. 39, the shallow P-well extends from end to end in the y direction. Where x and y are arbitrary directions for characterizing the relative orientation of the 6T SRAM cell. FIG. 41B shows a 6T SRAM cell laid out to form a 2 × 2 array according to one embodiment, with one of the two cells adjacent in the y direction forming a mirror image of the cell. Inverted in the y direction. As shown in FIG. 41B, shallow P-well 3940 is continuous across cells in the y direction. Thus, if multiple cells are connected in the y direction, all of those cells will share the same shallow P-well. In order to increase the fineness of dynamic mode switching, it is necessary to use a shallow P-well 3940 continuity and breaking structure. The tap cell serves the purpose of isolating the shallow P-well and providing a connection to the shallow P-well.

  FIG. 42 illustrates an example tap cell layout that may be used with the embodiments described herein. This layout is designed to match the SRAM cell layout described below. The upper and lower portions of the tap cell have isolated shallow P-wells, and therefore each body bias source (shown as VSPW0 and VSPW1) can be individually connected. FIG. 43 shows an example of a cross-sectional view at two positions indicated by broken lines. This figure is rotated. A cross-sectional view 4210 corresponds to a cutaway view at position 4215. The left shallow P-well 3940 can be isolated from the shallow P-well 3940 on the right side of the STI. This shallow well isolation allows different body biases to be applied to the two shallow wells. A p-type implant is used in the source / drain region of the body access transistor to create a contact to the shallow P-well. Since this p-type source / drain region has the same doping type as the shallow P-well, conduction from the p-type source / drain (ie, body tap) to the shallow P-well is generated. Sectional view 4220 corresponds to position 4225. Since the body tap region is doped with the same doping type as the shallow N well, the body tap creates a connection to the shallow N well. The well structures in cross-sectional views 4210 and 4220 are similar to those in cross-sectional views 4010 and 4020, respectively.

  FIG. 44 shows an example of a top view of the tap cell of FIG. Unlike SRAM cells in which a shallow P-well 3940 extends end to end, the upper tap cell well structure 3940 can be isolated from that of the lower side by a dividing line 4480. As mentioned above, shallow N-well isolation is not critical. This is because the shallow N-well is conductively connected to the N-well extending below the entire cell. The tap cell also provides a connection to the shallow P-well 3940 via the SPW tap 4460 and a connection to the shallow N-well 3950 via the SNW tap 4470. FIG. 45 shows an example of forming a 2 × 2 SRAM array 4500 that utilizes a dynamic mode control function in accordance with the embodiments described herein. This SRAM array is composed of 2 × 2 SRAM cells and two tap cells forming SPW taps and isolation on each side of the y boundary. Again, the xy direction is a relative direction for indicating the orientation of the array. As shown in FIG. 45, two SRAM cells adjacent in the y direction have a continuous SPW. The SPW continuity across these two neighboring cells in the y direction is terminated by the STI in the tap cell 4200. Therefore, the body bias VSPWn is applied to this 2 × 2 SRAM array, the body bias VSPW (n−1) is applied to the adjacent array (not shown in the drawing) at the top, and the adjacent array (fully Can be applied with a body bias VSPW (n + 1). FIG. 46 shows an example of a 4 × 4 SRAM array 4600 that uses tap cells for SPW isolation. 45 and 46 illustrate the use of SRAM cells with SPW continuity in the y-direction and the use of body access cells (also referred to as tap cells) that terminate SPW continuity. Therefore, according to this, a dynamic mode switching type SRAM array having a desired size can be formed.

  45 and 46 focus on the SPW continuity and isolation example, but as noted above, numerous other signals and supply voltages are required to form a complete array. . The connection of these signals and power supply voltages to the SRAM array is well known in the art and will not be described in detail here. In a fully connected 4 × 4 SRAM array corresponding to FIG. 46, a word line (WL) signal is connected to each row (row) of the SRAM array, and a bit line (BL) is connected to each column (column) of the SRAM array. Signals can be connected.

  The body control signal (VSPWn) can run parallel to the word line. During operation of the SRAM array, the body bias of the selected word group can be switched positive when a word of the selected word line is selected. This helps to improve read performance and write performance. When reading or writing to a particular word group, all other word groups in the subarray can have a body that is reverse-biased (or zero-biased) to suppress leakage.

In some use cases of 6T SRAM using body tap / body access cells to facilitate mode switching, a shallow P-well body is used for dynamic switching and a p-channel body (N-well) is static. Can be used for bias. A word selected in the group can cause the shallow P-well body of all n-channel transistors in the selected word group to switch. The p-channel and n-channel biases can be set to zero and then forward or reverse biased according to the desired mode. The dynamic mode switching SRAM array based on the body access cell described above has advantages in fine granularity control that can be scaled. However, this approach requires a body access cell in addition to the SRAM cell. There are other approaches and systems that do not require additional body access cells. One such approach is in the method based on body access cell whereas all the cells in SRAM array share a common V _SS, using a _{V SS (V SS per-row} ) for each row . If V _SS can be individually controlled for each row, a unique V _SS can be applied to each row to generate a desired body bias for that row. In this situation, the body voltage may not be controllable. However, V _SS can be individually controlled to produce different V _BS voltages (voltage between body and source) to achieve dynamic mode switching.

Figure 47 shows an example of a multi-mode switch of 6T-SRAM circuit 4700 based on the _{V SS} for each row. In this case as well, the SRAM cell is composed of two pull-up (PU) transistors, two pull-down (PD) transistors, and two pass gate (PG) transistors. One difference between the example shown in FIG. 47 and the 6T SRAM cell of FIG. 38 is that the pass gate (PG) used in FIG. 47 is an n-channel three-terminal dual gate transistor 4710. The layout and corresponding cross-sectional view of a three terminal dual gate transistor are shown in FIGS. The dual gate transistor has a gate connected to the body. That is, the gate (that is, WL) of the PG transistor is connected to the body of the cell. The PU transistor and the PD transistor are the same type as in the example of FIG. FIG. 48 shows an example of the SRAM cell layout 4800 of FIG. 47, in which a cell boundary 4860 is shown. An n-channel device in a shallow P well is used for the PG transistor and the PD transistor, and a p-channel device is used for the PU transistor. Since the well structure of this SRAM cell is very similar to that of FIG. 39, a cross-sectional view is not shown. Both the SPW and SNW are on a common N-well used throughout the cell.

FIG. 49A shows a structure 4900 showing the SPW and SNW of the SRAM layout of FIG. In this _{layout, V SS} contact 4910 is shown clearly. When connecting multiple SRAM cells, the contacts are often connected using metal regions. FIG. 49B shows a 2 × 2 SRAM array 4920 using the SRAM cell of FIG. The SPW 3940 does not form continuity like the SRAM array of FIG. FIG. 49B also shows that V _SS (V _SS0 4921 and V _SS1 4922) is individually connected to each row. FIG. 49C shows a 4 × 4 SRAM array 4930 based on row-by-row V _SS technology, with each row using its own V _SS (V _SS0 4931, V _SS1 4932, V _SS2 4933 and V _SS3 4934). Has been.

In the complete layout of the 4 × 4 SRAM array corresponding to FIG. 49C, a word line (WL) is connected for each row, as in the dynamic mode switching type 4 × 4 SRAM array based on the body access cell technology, and for each column. A bit line (BL) can be connected. Each row word line may be connected to an SPW (ie, the body of the respective device). _VSS can also be connected for each row. Therefore, individual body bias can be achieved for each row. N-well body taps may be provided for every 16 (or 32) word lines.

Of another implementation of a _{V SS} based mode switching 6T SRAM5000 shown in FIG. 50. A three terminal dual gate transistor epibody contact is formed on PG and a cell boundary 5060 is shown. FIG. 51A shows SPW and SNW of the SRAM layout of FIG. In this _{layout, V SS} contact 4910 is shown clearly. FIG. 51B shows a 2 × 2 SRAM array 5120 using the SRAM cell of FIG. The SPW 3940 does not form continuity like the SRAM array of FIG. FIG. 51B also shows that in structure 5100 V _SS (V _SS0 4921 and V _SS1 4922) is individually connected to each row. FIG. 51C shows a 4 × 4 SRAM array 5130 based on row-by-row V _SS technology, with each row using its own V _SS (V _SS0 4931, V _SS1 4932, V _SS2 4933 and V _SS3 4934). Has been. The characteristics and area of this cell are the same as in the example of FIG.

The operation mode of the cell is determined according to a plurality of conditions including V _SS , n channel bias, word line (WL) state, bit line (BL) state, V _DD and p channel body bias. V _SS , n-channel bias, word line (WL) state, bit line (BL) state may be used for dynamic mode switching, and V _DD and p-channel body bias may be used for static mode control. In this SRAM array, dedicated V _SS (V _SS0 -V _SS2 , V _SS3 ) is used for each row. Similarly, WLs connected to shallow P-wells to dynamically control n-channel body bias are also organized to have one WL (WL0-WL3) per row. The BL and V _DD lines are used to connect a plurality of cells in the vertical direction. As shown, both BL and V _DD are organized to provide one BL and one V _DD per column. A typical SRAM may include Read / Write, NOP (no operation), and deep sleep modes. In the following, further details of these modes will be described.

In standby and data retention modes (corresponding to deep sleep mode), V _SS is positively biased to reverse bias the body of the n-channel device to reduce the effective V _DS . This setting reduces standby leaks. For example, V _SS can be set to 0.3 V and V _DD can be set to less than 0.6 V so that V _DS ≦ 0.3 V. Under this condition, both the PG transistor and the PD transistor are reverse-biased. The p-channel device is zero-biased or reverse-biased to maintain a 1000x PU transistor current with PD off-current. In the NOP mode, both PG and PD n-channel devices have a reverse-biased body, and the body of the PU p-channel device is biased with zero or reverse bias. As an example, V _DD is set to 1.0 V and V _SS and BL are set to 0.6 V so that V _DS ≦ 0.4 V and a low standby current is achieved.

In Read mode, both PG and PD n-channel devices can have forward bias. Dynamic _VSS switching may be limited to a selected word (or row). In the PG device, V _GS = V _BS ≦ 0.6 V and V _DS ≦ 0.6 V are set. In the PD device, V _GS = 1.0V and V _BS ≦ 0.6V. With a larger PD _VDS , a suitable PD / PG beta ratio can be achieved. The width of the PG device can be the same as the width of the PD device. This can achieve a favorable static read noise margin and low read cell current.

In Write mode, both PG and PD n-channel devices can have forward bias. Dynamic _VSS switching may be limited to a selected word (or row). In the PG device, V _GS = V _BS ≦ 0.6V. In this example, n-channel PG and PD transistors and p-channel PU transistors in a shallow P-well are used, but to achieve the same design goals, p-channel PG and PD transistors in a shallow N-well are used. Transistors as well as n-channel PU transistors can also be used.

The row-by-row _VSS technology does not require body access cells for shallow well isolation, and each SRAM cell is larger than a technology based SRAM cell based on body access cells. To the cell from adjacent cells facilitates body bias control for each row based on the isolate to V _SS, inactive areas around the cells may be added. Thus, in this example, the cell height can be increased by 130 nm. This corresponds to an increase of about 38% in cell area. All transistors are oriented in the same direction. As one design example, transistor dimensions may be as follows:
Pass gate (PG): W / L = 70 nm / 40 nm
Pull-down (PD): W / L = 85nm / 35nm
Pull-up (PU): W / L = 65nm / 35nm
This example results in an area of x × y = 0.72 μm × 0.475 μm = 0.342 μm ² at 45 nm process node.

  FIG. 52 illustrates a system 5200 that includes a number of functional units that are interconnected using an interconnect 5210 as needed. For example, in some cases, interconnect 5210 provides a common path for communication among all of functional units 5204-1, 5204-2, 5204-3 through 5204-n. In other cases, the interconnect provides a common communication path between other sets of functional units while providing point-to-point communication between one set of functional units. Thus, the interconnect 5210 is suitable for satisfying the purpose of the system designer using functional units available in the target system using conventional communication technologies including, for example, wired, wireless, broadcast and point-to-point. Any method may be used. “N” of 0n is intended to convey that there may be as many functional units as the system designer thinks necessary, and suggests that there are a maximum of nine functional blocks. It is not a thing.

  According to some embodiments, system 5200 is an electronic system having a plurality of independently packaged components and / or subassemblies. Examples of such systems today are personal computers, mobile phones, digital music players, e-book readers, game consoles, portable game systems, cable set top boxes, televisions, stereo equipment, and disclosed herein. Includes other electronically similar electronic systems that can benefit from the enhanced power consumption control provided by the technology. In such a system, the functional units 5201, 5202, 5203, 5204-1 through 5204-n are typical system components of such a system, and the interconnect 5210 is typically a printed wiring board or backplane ( (Not shown). For example, in the case of a personal computer, the functional components include a CPU, system memory, and mass storage devices such as hard disk drives or solid state disk drives, all of which are mounted on the motherboard as required. Interconnected by system interconnect. Similarly, mobile phones include, for example, a variety of one or more chips and display panels, all of which are typically interconnected using one or more printed wiring boards (PWB) that may include flexible connectors. Is done.

  According to another embodiment, system 5200 is a system in package (SIP), each of the functional units being an integrated circuit, all of which are packaged together in a single multi-chip package. . In a SIP system, the interconnect 5210 can be a direct chip-to-chip interconnect, such as a wire bond, lead bond, solder ball or gold stud bump, and a common bus interconnect, point-to-point interconnect, power plane and ground It can be realized by an interconnection provided by a package substrate that can include a plane.

  According to yet another embodiment, the system 5200 is a single chip, such as a system on chip (SOC), for example, and the functional unit may be a common semiconductor substrate or a semiconductor on insulator substrate (eg, SOI). It is mounted as a group of transistors above (when bulk CMOS and SOI structures are mounted on the substrate). In such an embodiment, interconnect 5210 may be implemented using any technique available for interconnecting multiple circuit blocks within an integrated circuit.

  As described above, the described transistor and integrated circuit techniques can be implemented independently on a common semiconductor substrate, either statically by design and / or dynamically by adjusting body bias and / or operating voltage. Enables the manufacture and use of multimode transistors that can be specified. These same technologies may also provide similar benefits at the system level even if only one of the functional units implements the technology. For example, the functional unit 5202 may include logic (not shown) that dynamically adjusts the operating mode of its DDC transistor to reduce power consumption. This can be done, for example, by digital or analog technology implemented on the functional unit 5202. In another example, functional unit 5202 may control power consumption in response to an external control signal from another functional unit, such as functional unit 5201, for example. Whether the power consumption in each functional unit is controlled locally by that functional unit, centrally controlled by a controller functional unit, or controlled in a complex manner, typically more power Consumption control can be achieved.

  Control of power consumption at the system level is already known from time to time, especially in computing systems. For example, the Advanced Configuration and Power Interface (ACPI) specification is an open standard for power management of system components by an operating system. The deep depleted channel transistor and integrated circuit technology described above complements the potential of such a power management approach by allowing system control of individual circuit blocks within each functional unit in the system. Expand. For example, the lowest level of control provided by ACPI is the device level, which corresponds to a functional block (eg, chip or hard drive) of a multi-component system such as a personal computer. By providing fine-grained individual control over the power consumption of individual circuit blocks within a device, more power states of devices and systems are possible.

System level power management can be particularly beneficial in SOC systems using the DDC concept. As mentioned above, the DDC structure allows a high level of programmability in nanoscale transistors. Of DDC structure, the nominal threshold voltage V _T is available over a relatively wide range, and relatively low [sigma] v _T, by a relatively high body factor, all and the same operation have the same intrinsic V _T voltage V _DD Transistors manufactured to be operated at can then be reinforced to operate in different operating modes using different actual _VTs and possibly different actual operating voltages V _DD for each circuit block. . This type of flexibility allows the same chip to be designed to be used in a variety of target systems and operating conditions, and to be dynamically set according to in-situ operation. This can be particularly useful for systems that are sometimes connected to an AC power source, regardless of whether they are SOC, but otherwise use batteries.

  FIG. 53 shows a network 5300 that includes a number of systems 5301, 5302, and 5303 that are interconnected using an interconnect 5310 as needed. For example, in some cases, interconnect 5310 provides a common path for communication between all of systems 5304-1 through 5304-n. In other cases, the interconnect provides a common communication path between other sets of systems while providing point-to-point communication between one set of systems. Thus, the interconnect 5310 is a network designer using a system that can be connected to a target network using conventional communication technologies including, for example, wired, wireless, broadcast, point-to-point and peer-to-peer. It may be configured by any method suitable for satisfying the purpose. The “n” in 5304-n is intended to convey that there can be as many systems as the network allows, and does not imply that there are a maximum of nine systems.

  The above-described deeply depleted channel transistor, integrated circuit and system technologies provide fine-grained control capabilities for systems coupled to a network. Having such a high level of control over networked systems can be particularly useful in enterprise networks to reduce the energy costs incurred by devices that are turned on but not used. . Such control also controls power consumption, switches system capabilities on or off depending on application requirements, and selectively selects specific functional units or parts thereof for high performance operating modes ( For example, it may be of a subscriber-based wireless network including, for example, a cellular telephone network, regardless of whether it is assisted to put in “turbo mode”).

  FIG. 54 illustrates an exemplary method of using a system, such as that described with reference to FIG. 52, whether used with a network, such as that described with reference to FIG. 53, or used alone. Is shown. After the system is turned on in step 5410, the system. Sets the power mode of a system component (eg, functional unit). This can be an external signal provided on the network, a central mode control signal provided by a functional unit in the system, or a local mode control signal generated separately at each functional unit capable of multi-mode operation, In response to any of the above, using transistors, transistor groups and / or integrated circuits of the type described herein. As mentioned above, a single component can have different parts that are set to operate in different modes. For example, some components may be set to operate in legacy mode and other portions of the same component may be set to operate in low power / low leak mode. At step 5430, the system monitors its usage and decides whether to change its power mode. This monitoring function may be performed centrally by one functional unit, or may be distributed over multiple functional units, each of which may make local decisions regarding the mode based on monitoring specific conditions. Both of these (for example, one functional unit may sleep itself based on its own criteria, even though the central supervisor has not decided to place the entire system in deep sleep mode. Similarly, after a single component has decided to place itself in turbo mode for better performance after the initial mode setup, the central monitor May decide to put it in deep sleep mode.) Step 5430 is repeated until the state of the system or functional unit changes and a new power mode is required. If a new power mode is required, step 5440 is executed. As shown, if a power down of the system is requested at step 5440, the system is shut down at step 5450. Otherwise, step 5420 is repeated for one or more functional units depending on what state change is requested. Thus, users of systems or chips manufactured using the techniques described herein can benefit from the benefits.

  While particular exemplary embodiments have been illustrated and described in the drawings, it will be understood that various other modifications may be apparent to those skilled in the art, and these embodiments are merely exemplary. It is not intended to limit the broad scope of the invention, and the invention is not limited to the specific structure and construction shown and described. The specification and drawings are accordingly to be regarded in an illustrative rather than restrictive sense.

  This application is based on US Provisional Application No. 61 / 247,300 filed on September 30, 2009, US Provisional Application No. 61 / 262,122 filed on November 17, 2009, February 2010. And claims the benefit of US patent application Ser. No. 12 / 708,497, filed on the 18th, the disclosure of which is incorporated herein by reference.

Claims (12)

A gate having a gate insulator and a conductive electrode;
A shielding region implanted in the transistor body and doped to a thickness of at least 5 nm and extending below the gate and above the transistor body, the transistor body being doped lower than the shielding region; A shielding area;
An undoped channel region having a dopant concentration of less than 5 × 10 ¹⁷ atoms / cm ³ between the shielding region and the gate insulator and having a thickness of at least 5 nm;
A threshold voltage adjustment region formed between the undoped channel region and the shielding region, the threshold voltage adjustment region being in contact with the shielding region and separated from the gate insulator by the undoped channel region; The threshold voltage adjustment region is doped lower than the shielding region; and
A source and drain between which the undoped channel region extends; and
A bias voltage is selectively applied to the transistor body that is implanted into the transistor body laterally outside one of the source and the drain and is in electrical contact with the shield region below the shield region. A body tap to apply,
Have
The shielding region, the transistor body, the undoped channel region, and the body tap have a conductivity type opposite to that of the source and the drain;
Field effect transistor.   The field effect transistor according to claim 1, wherein the transistor is a four-terminal device that can be switched on or off by applying a voltage to the conductive electrode of the gate or the transistor body.   The field effect transistor of claim 1, wherein the transistor is a three-terminal device that can be switched on or off by applying a voltage to the conductive electrode of the gate.   The field effect transistor according to claim 1, wherein the bias voltage is statically set to increase or decrease a threshold voltage.   The field effect transistor of claim 1, wherein the bias voltage is dynamically controlled to adjust a threshold voltage.   The field effect transistor of claim 1, wherein the shielding region is spaced from both the source and the drain.   The field effect transistor of claim 1, wherein the source and drain are raised.   The field effect transistor of claim 1, wherein the channel region is undoped epitaxially grown silicon. Implanting into the transistor body a shielding region doped over a thickness of at least 5 nm and extending below the gate and above the transistor body, the transistor body being doped lower than the shielding region; Process,
Forming a threshold voltage adjustment region on the shielding region prior to growing an undoped channel region, the threshold voltage adjustment region contacting the shielding region; and
Epitaxially growing an undoped channel region having a dopant concentration lower than 5 × 10 ¹⁷ atoms / cm ³ on the threshold voltage adjustment region, the undoped channel region having a thickness of at least 5 nm; ,
Forming a source and drain between which the undoped channel region extends; and
Forming a gate having a gate insulator and a conductive electrode on the undoped channel region between the source and the drain;
A body tap for selectively applying a bias voltage to the transistor body under the shielding region and in electrical contact with the shielding region, on the laterally outer side of one of the source and the drain. Forming in the transistor body;
Have
The shielding region, the transistor body, the undoped channel region, and the body tap have a conductivity type opposite to that of the source and the drain;
A method of manufacturing a field effect transistor.   The method of manufacturing a field effect transistor according to claim 9, wherein the source and drain are formed to be separated from the shielding region.   The method for manufacturing a field effect transistor according to claim 9, further comprising performing shallow trench isolation after epitaxial growth of the undoped channel region.   The method of manufacturing a field effect transistor according to claim 9, further comprising performing partial trench isolation after epitaxial growth of the undoped channel region.

Images