Body bias circuits and methods

Description

TECHNICAL FIELD

The present invention relates generally to integrated circuit (IC) devices, and more particularly to body bias circuits and methods for insulated gate field effect transistors of IC devices.

BACKGROUND

Integrated circuit (IC) devices can be subject to manufacturing variations that can impact performance. In a typical IC device, electrical signals can be transmitted via an interconnect structure formed by multiple metallization layers, separated from one another by interlayer dielectrics (ILDs). Signals can be generated by transistors driving metallization layers between different potentials.

Many fabrication processes utilize chemical-mechanical polishing (CMP) to planarize ILD surfaces. While CMP can improve planarization over other fabrication methods, it can still impart systematic and random thickness variations at the lot, wafer, die and pattern levels. Die level variation in such structures can result in performance and timing differences in the same type of devices. In particular, signal transmission paths can be affected by variations in metal-metal capacitance inherent in an interconnect structure. Differences in metal-metal capacitance can arise from variations in dielectric thickness between metallization lines and/or variations in metal thickness and width. Differences in resistance can also arise from such variations in metal thickness and width.

Active devices (e.g., transistors) may also be subject to some variation in performance. For example, uncontrollable manufacturing variations can result in “fast” transistors and “slow” transistors. Fast transistors can provide faster driving capability than slow transistors. Such fast vs. slow variations can be systematic, i.e., they can affect the mean value of performance parameters.

These and other variations can result in IC devices having differing performance limits. Conventionally, timing paths can be designed with sufficient margin to be adequate for a slowest case. This can require large signal driving devices and increased power consumption, as well as increased design effort. For some very high performance IC devices, such variations can present a limit to device speed and/or reliability.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a block schematic diagram of an integrated circuit device according to an embodiment.

FIG. 2 is a block schematic diagram of a combination logic section according to an embodiment.

FIG. 3 is a block schematic diagram of a monitoring section that can be included in embodiments.

FIGS. 4A and 4B are diagrams showing a monitoring section and operation that can be included in embodiments.

FIG. 5 is a block schematic diagram of another monitoring section that can be included in embodiments.

FIG. 6A is a block schematic diagram of a transistor under test (TUT) selection circuit that can be included in embodiments.

FIG. 6B is a block schematic diagram of a load capacitance selection circuit that can be included in embodiments.

FIG. 7 is a block schematic diagram of a drive monitor circuit according to another embodiment.

FIG. 8 is a diagram of a metal-metal capacitor that can be included in embodiments.

FIG. 9 is a diagram showing capacitance corresponding to metallization fabrication variations.

FIG. 10 is a block schematic diagram of an integrated circuit device according to an embodiment.

FIG. 11 is a block schematic diagram of a power gating circuit according to an embodiment.

FIG. 12 is a block schematic diagram of a core biasing circuit that can be used in an embodiment like that of FIG. 11.

FIG. 13 is a block schematic diagram of a regulator circuit according to an embodiment.

FIGS. 14A to 14D are side cross sectional views of devices that can be body biased according to embodiments.

FIG. 15A is a side cross sectional view of a deeply depleted channel (DDC) transistor that can be included in embodiments.

FIG. 15B is a side cross sectional view of FinFET transistor that can be included in embodiments.

FIG. 15C is a side cross sectional views of FinFET transistor having a screening region, and optionally, a threshold voltage set region, that can be included in embodiments.

FIGS. 16A-0 to 16B-1 are graphs showing how transistor performance can be measured with a circuit like that of FIG. 4A, and how a body bias can be adjusted to change transistor performance.

FIG. 17 shows one embodiment of a bias voltage distribution network using monitor circuits to implement a fine grained body biasing technique.

FIG. 18 shows one embodiment of the bias tuning circuit using a resistive divider to generate the local body bias voltage.

DETAILED DESCRIPTION

Various embodiments of the present invention will now be described in detail with reference to a number of drawings. The embodiments show integrated circuits and methods for biasing bodies of transistors in an integrated circuit device, where such biasing can be based on performance of actual circuit structures present on the integrated circuit device.

In the various embodiments below, like items are referred to by the same reference character but the leading digits corresponding to the figure number.

FIG. 1 is a block schematic diagram of an integrated circuit (IC) device 100 according to one embodiment. An IC device 100 can include a number of monitoring sections 102-0 to -n, a combination logic section 104, an n-type body bias circuit 106-0, a p-type body bias circuit 106-1, and an operational section 150. Monitoring sections (102-0 to -n) can include physical structures corresponding to those of the IC device 100.

Each monitoring section (102-0 to -n) can include physical circuit elements of the IC device 100 that can vary according to process variation. Monitoring sections (102-0 to -n) can output monitor values M0 to Mn that reflect variations in performance resulting from variations in process. In particular embodiments, each monitoring section (102-0 to -n) can provide a monitor value (M0 to Mn) that is dominated by one or more particular process features. In the particular embodiment shown, monitor sections (102-0 to -n) include those dominated by: n-type transistor performance, p-type transistor performance, wiring, and diffusion. It is understood that such features are but exemplary. In some embodiments, a monitoring section can include a ring oscillator circuit having an oscillating frequency reflecting the effect of one or more sources of process variations. However, in particular embodiments, a monitoring section can include a transistor drive monitoring section, embodiments of which are described in more detail below.

Monitor sections (102-0 to -n) can provide monitor values in various ways. For example, in some embodiments or modes of operation, monitor sections (102-0 to -n) can be operational during a test or characterization phase. Monitor values (M0 to Mn) generated in the test/characterization phase can be stored such that they remain fixed throughout the operation of the device. After the monitor values have been generated, the combination logic section 104 can access the stored monitor values to set the n-type and p-type transistor body bias voltages based on one or more of the stored monitor values during the operation of the device. The monitor values can be stored in internal storage that is part of the combination logic section 104, or in external storage that is located outside the combination logic section 104 and accessed by the combination logic section 104.

In other embodiments or modes of operation, monitor sections (102-0 to -n) can be activated periodically and/or constantly, allowing monitor values to be updated while the device is in operation. In such embodiments, a combination logic section 104 can receive the periodically or continuously updated monitor values (102-0 to -n) and modify the n-type and p-type body bias voltages based on the updated monitor values.

In still other embodiments or modes of operation, monitor sections (102-0 to -n) can be enabled in response to a command input to the IC device 100, where the monitor sections generate and store updated monitor values during a test/characterization phase that is initiated in response to receiving the command input at the IC device. In other embodiments, the monitor sections (102-0 to -n) can be enabled in response to received status information indicative of particular conditions or changes to particular conditions (e.g., power-on, reset, temperature, power supply voltage, etc.). For example, the test/characterization phase can be initiated if temperature exceeds a predetermined threshold value, such that monitor sections (102-0 to -n) are enabled to generate and store updated monitor values that are used to modify the n-type and p-type body bias voltages.

A combination logic section 104 can receive monitor values (M0 to Mn), and from such values, generates n-type and p-type body bias control values (NBias, PBias). In particular embodiments, a combination logic section 104 can assign different weights to one or more of the monitor values (M0 to Mn) such that particular monitor values (M0 to Mn) have a greater weight over others, to reflect how one type of process variation can have a greater effect than other types of process variations on a particular performance characteristic of a circuit within the IC device 100. In this way, body biases can be tuned to optimize an IC device for a particular performance target. In some embodiments, bias control values (NBias, PBias) can be static during the operation of the IC device 100. However, in other embodiments, such values can be static, changing according to mode of operation, and/or operating conditions. Additionally, the weights can compensate for differing sensitivities of the monitoring circuits to the variations, as well as the IC's sensitivity to such variations.

It is noted that bias control values (NBias, PBias) generated by a combination logic section 104 can be digital or analog values.

An n-type body bias circuit 106-0 can receive a bias control value (NBias), and in response generate an n-channel transistor body bias voltage (Vbn). That is, a voltage Vbn can vary according to the bias control value NBias. In very particular embodiments, Vbn can vary from a “back bias” level (VBB) to a forward bias level (VFB). A back bias level can force a body region below a low power supply voltage (e.g., less than VSS) to increase a threshold voltage, but lower a switching speed of an n-type transistor (as compared to such a transistor having a body bias of VSS). In contrast, a forward bias level can force a body region above a low power supply voltage (e.g., greater than VSS) to lower a threshold voltage of a transistor, but increase a switching speed of an n-type transistor (as compared to such a transistor having a body bias of VSS).

In a similar fashion, a p-type body bias circuit 106-1 can receive a bias control value (PBias), and in response, generate a p-channel transistor body bias voltage (Vbp). In certain embodiments, Vbp can vary from a “back bias” level (e.g., greater than a high power supply voltage VDD) to a “forward bias” level (e.g., less than VDD).

An operational section 150 can include transistors formed with body regions that can receive the body bias voltages (Vbn, Vbp). Such transistors can be arranged to provide a predetermined function as part of different functional blocks having different predetermined target performance characteristics, where the performance of each functional block can be independently adjusted by selection of the n-type and p-type body bias voltages. Accordingly, an operational section 150 can include n-type transistors formed in p-type body regions 108-0, and p-type transistors formed in n-type body regions 108-1. Body regions (108-0/1) can vary according to transistor type. Various examples of transistors that can have a body bias will be described in more detail below.

Body bias voltages (Vbn, Vbp) can be applied to transistors of operational section 150 to compensate for process variations detected by monitoring sections (102-0 to -n). Thus, if a metallization structure presents a high capacitance, resulting in a slower signal transmission speed, a body bias voltage can be lowered, to increase the driving speed of the transistors in the device. Similarly, if such transistors are considered “slow” transistors, a body bias voltage can be reduced or increased, for p-type and n-type transistors respectively, to bring the performance of the transistors back into a desired speed range. The opposite type of adjustment is possible as well. If metallization presents a smaller capacitance or transistors are “fast”, a body bias can be increased or decreased to compensate for these variations by corresponding changes of the transistor speeds in response to the applied body bias.

FIG. 2 shows a combination logic section 204 according to an embodiment. In one particular embodiment, a combination logic section 204 can be one implementation of that shown as 104 in FIG. 1. A combination logic section 204 can include a monitor value store 210, weight value store 212, and a body bias value generator circuit 214. A monitor value store 210 can store monitor values (Mo to Mn) received from monitor sections (not shown). As noted above, monitor values can be static or dynamic. A monitor value store 210 can be formed with any suitable storage elements, including non-volatile elements, volatile storage elements, or combinations of thereof.

A weight value store 212 can provide weight values (W0 to Wi) to increase or decrease the effect that a monitor value can have on a resulting bias control value (PBias, NBias), typically achieved by multiplying the monitor value by the corresponding weight value. In some embodiments, weight values (W0 to Wi) can be determined through simulation or characterization of circuits within an integrated circuit device. In other embodiments, weight values (W0 to Wi) can be established by testing integrated devices of a same manufacturing group (e.g., lot), or by circuits on the device itself (such as monitor sections, described herein). Once such weight values are determined, the values (W0 to Wn) can remain fixed during the operation of the device, or alternatively, can be updated during the operation of the device. A monitor value store 210 can be formed with any suitable storage elements, including non-volatile elements, volatile storage elements (e.g., flash memory, fuses, SRAM, latches), or combinations of thereof. In the case of fixed weight values, a back end process, such as the opening of fusible links can be used to set the weight values.

A bias value generator circuit 214 can apply weighting values (W0 to Wn) to corresponding monitor values (M0 to Mn), and combine the results to generate bias control values (NBias, PBias) that are responsive to the weighting values (W0 to Wn) and the monitor values (M0 to Mn). A bias value generator circuit 214 can include any circuits suitable for combining/weighting monitor values. In some embodiments, a bias value generator circuit 214 can be an analog circuit, combining/multiplying voltages or currents. In other embodiments, a bias value generator circuit 214 can be a digital circuit, combining digital values through operations such as binary addition and multiplication. In the latter case, a bias value generator circuit 214 can be formed with custom logic, programmable logic, and/or can include a processor executing predetermined instructions.

It is noted that a combination logic section 204 can also modify a bias control value in response to additional control parameters. As but one example, the bias control values (e.g., NBias and/or PBias) can be adjusted according to a power supply voltage of the device. Such one or more of the bias control values can be adjusted by a predetermined amount if a change on the power supply voltage exceeds a threshold value. Such adjustments can be performed by directly modifying the bias control values, modifying the stored monitor values (M0 to Mn), or adjusting the weights (W0 to Wn) in response to received values of the control parameters.

FIG. 3 shows a drive monitoring circuit 302 according to one embodiment. A drive monitoring circuit 302 can be one of a type of monitor circuit, such as those shown as 102-0 to -n in FIG. 1). A drive monitor circuit 302 can include a transistor-under-test (TUT) section 316, a load section 318, and a monitor section 320. A TUT section 316 can include one or more transistors that can be tested for performance. Such transistors can include n-channel transistors, p-channel transistors, or combinations thereof.

A load section 318 can include an impedance that is driven by TUT section 316 in a monitor operation. In some embodiments, a load section 318 can include circuit elements (e.g., capacitors) formed with physical structures of an IC device, to reflect process variations of such structures.

A monitor section 320 can generate a monitor value Mx reflecting a TUT section 316 driving a load section 318. For example, in embodiments where the load section 318 is a capacitor, the monitor section 320 generates the monitor value Mx by first charging the capacitor to a predetermined first voltage. Typically, the TUT section is configured to be in a non-conductive state while the capacitor is being pre-charged to set a predetermined initial condition prior to the TUT measurement operation. (In certain embodiments, the capacitor can be pre-discharged to set the initial condition. For simplicity we use the terms pre-charged and dis-charged for the setting of the initial conditions and TUT measurement operations, respectively). After the capacitor has been charged to the predetermined first voltage, the monitor section 320 stops charging the capacitor and configures the TUT section 316 to a conductive state such that the capacitor starts discharging to a second predetermined voltage through the TUT section 316. Thus, a drive monitoring circuit 302 can be “dominated” by the performance of transistor(s) in TUT section 316 and/or physical features of an IC device reflected by load section 318. If different TUT sections 316 are used discharge a capacitor that is charged to the same predetermined first voltage, the rate at which the capacitor is discharged through the TUT section 316 represents the speed of the TUT section. Alternatively, if different load sections or different capacitors that are representative of different physical features of the IC device are charged to the same first predetermined voltage, and the same TUT section is used to discharge these load sections, then the rate at which the load section 318 is discharged represents the magnitude of the load resulting from the physical features of the IC device associated with the load section 318. Alternative embodiments of the drive monitoring circuit 302, can first discharge the load section 318 to a first predetermined voltage, and subsequently charge it to a second predetermined voltage through the TUT, where the rate of charging the load section 318 can represent the performance of the load section 318 or the TUT section 316.

FIG. 4A is a block schematic diagram of a drive monitor circuit 402 according to an embodiment. In one embodiment, a drive monitor circuit 402 can be one particular implementation of that shown in FIG. 3. A drive monitor circuit 402 can include a TUT section 416, composed of an n-channel TUT 428 having a source-drain path connected between a monitor node 426 and a low power supply voltage VSS, and a gate connected to a “Start” signal generated by a counter circuit 424. In a monitor operation, the TUT 416 can discharge monitor node 426 from a predetermined initial voltage to (or toward) VSS, where the time interval to discharge the monitor node to a reference voltage level Vref is representative of the TUT performance.

A load section 418 can be a capacitance, which in the embodiment shown, is connected between monitor node 426 and low power supply voltage VSS. Load section 418 can be charged to a predetermined voltage by monitor circuit 420 through switch element 421, and in a subsequent monitor operation, can be discharged by TUT section 416.

A monitor circuit 420 can include a switch element 421, a comparator 422, and a counter circuit 424. A switch element 421 can charge monitor node 426 to (or toward) VDD in response to a Start output from counter circuit 424. A comparator 422 can have a negative polarity input (labeled “−” in FIG. 4A) connected to monitor node 426, a positive polarity input (labeled “+” in FIG. 4A) that receives a reference voltage Vref, and an output connected to a Stop input of counter circuit 424.

A counter circuit 424 can generate a count value Mx(count) that corresponds (i.e., is inversely proportional) to a speed at which a load section 418 can be discharged by TUT section 416. In particular, counter circuit 424 can start a count operation while simultaneously activating a Start signal. The activated Start signal can start a monitor operation by disabling switch element 421 and enabling TUT section 416, thereby starting the discharge of the load section 418 through the TUT. When an output of comparator 422 goes high, i.e. after the monitor node has discharged to reference voltage Vref, the “Stop” signal is activated and the count operation stops.

FIG. 4B is a timing diagram showing one particular monitor operation for the drive monitor circuit 402 of FIG. 4A. At time t0, a Start output can be inactive (low). Switch element 421 can be enabled, and monitor node 426 can charge the capacitance of load section 418 to VDD. TUT section 416 can be disabled. With monitor node 426 above a Vref, an output of comparator 422 can be low.

At time t1, a Start output is activated (driven high). Switch element 421 is disabled and TUT section 416 is enabled. As a result, load section 418 starts to discharge toward VSS. At the same time, monitor circuit 420 initiates the count operation. The count operation continues as long as the voltage at the monitor node 426 exceeds the reference voltage Vref, and output of comparator 422 remains low.

At time t2, as the monitor node continues to discharge through TUT section 416 if the voltage at the monitor node 426 falls below Vref. As a result, the comparator 422 transitions to a high voltage level, thereby activating the “Stop” signal and stopping the count operation. The resulting count value after time count operation has stopped is the monitor value Mx(count).

At time t3, a Start output can return to the inactive state. TUT section 416 can be disabled and switch element 421 can be enabled. As a result, monitor node 426 can be pre-charged once again to VDD, and the monitor operation can be repeated.

FIG. 5 is a block schematic diagram of a drive monitor circuit 502 according to another embodiment. The drive monitor circuit 502 can be one particular implementation of that shown in FIG. 3. A drive monitor circuit 502 includes sections corresponding to FIG. 4A, but with reversed transistor conductivity types. Thus, a count value Mx(count) can reflect the performance of a p-channel TUT section 516 and/or load section 518.

Thus, the TUT section 516 of the drive monitor circuit 502 is composed of a p-channel TUT having a source-drain path connected between a monitor node 526 and a high power supply voltage VDD, and a gate connected to a “StartN” signal generated by a counter circuit 524. In a monitor operation, the TUT section 516 can charge the monitor node 526 from a predetermined initial voltage (which is a negative voltage in this embodiment) to (or toward) VDD, where the time interval to charge the monitor node to a reference voltage level Vref is representative of the performance of the TUT section 516.

The load section 518 can be a capacitance, which in the embodiment shown is connected between the monitor node 526 and the high power supply voltage VDD. The monitor circuit 520 can include a switch element 521, a comparator 522, and a counter circuit 524. The “StartN” signal generated by the counter circuit 524 is an active-low signal that starts the monitor operation when it is at a low voltage level. When the “StartN” signal is at a high voltage level, the switch element 521 pre-discharges the monitor node 526 towards VSS, and therefore, the load section 518 is discharged until the voltage across the load section 518 is −VDD (or negative VDD). Subsequently, the counter circuit 524 starts a count operation and simultaneously activates the “StartN” signal by placing a low voltage level on that signal. The activated “StartN” signal starts the monitor operation by disabling switch element 521 and enabling the TUT section 516, thereby starting to charge the load section 518 through the TUT section 516. The charging of the load section continues either as long as the “StartN” signal is asserted, or until the monitor node voltage reaches VDD (i.e., the load section 518 has been charged such that the voltage across the load section 518 increases from −VSS to VDD). As the monitor node voltage increases from VSS towards VDD, the output of the comparator 522 goes high after the monitor node voltage exceeds the reference voltage Vref. This activates the “Stop” signal and stops the count operation of the counter circuit 524. The resulting count value after the count operation has stopped is the monitor value Mx(count).

An IC can have different transistor types, sizes, etc. comprising those on the critical timing paths, so it can be useful to monitor multiple devices by having corresponding TUTs. These multiple TUTs can have individual loads or may share loads. FIG. 6A is a block schematic diagram of a TUT section 616 that can be included in embodiments. TUT section 616 can be one particular implementation of that shown as 416 in FIG. 4). A TUT section 616 can select one of multiple transistors (628-0 to -i) for a monitoring operation.

A TUT section 616 can include multiple TUTs (628-0 to -i) and a multiplexer (MUX) 630. In response to select inputs TUT_SEL, one of TUTs (628-0 to -i) can be connected to monitor node 626. In the particular embodiment shown, TUTs can include a low threshold voltage transistor (LVT), a standard threshold voltage transistor (SVT) (which is slower than LVT transistors), and a high threshold voltage transistor (HVT) (which is slower than both SVT and LVT transistors). However, in other embodiments, TUTs can include transistors having other differences (e.g., doping profiles, width/length ratios, oxide thicknesses, etc.). Also, while FIG. 6A shows n-channel device, other embodiments can include p-channel transistors.

A MUX 630 can be formed from the same device types as corresponding TUTs (i.e., n-channel, p-channel). That is, a charge/discharge path flows through a transistor(s) having a same conductivity type. In other embodiments, a MUX 630 can include signal paths formed by complementary conductivity (e.g., CMOS) devices, which can introduce substantially no threshold voltage drop into the charge/discharge path. Regardless, the conductivity of the MUX should be much greater than that of the TUT so that the result is not significantly impacted. Alternatively, the weights and Vref on the comparator can be adjusted so that the measurement results properly represents the TUT and load combination.

Switching in different TUT types can enable a monitor value to represent the performance of such a transistor type.

FIG. 6B is a block schematic diagram of a load section 6718 that can be included in embodiments. Load section 618 can be one particular implementation of that shown as 418 in FIG. 4A or 518 in FIG. 5). A load section 618 can select one of multiple loads (634-0 to -j) for a monitoring operation.

A load section 618 can include multiple loads (634-0 to -j) and a multiplexer (MUX) 632. In response to select inputs Z_SEL, one of loads (634-0 to -j) can be connected to monitor node 626. In the particular embodiment shown, loads can include capacitors formed by various structures of an IC device, and hence can reflect process variations of the device. In FIG. 6, one capacitance 634-0 can be a metal-metal capacitor having one plate formed by a metallization layer(s), a second plate formed by metallization layer(s), and a dielectric formed by one or more interlayer dielectric(s). In contrast, another capacitance 634-j can be a transistor capacitance having one plate formed by a transistor gate, a second plate formed by a substrate (source/drain/channel), and a dielectric formed by a gate capacitance and diffusion capacitance.

However, in other embodiments, loads (634-0 to -j) can include other impedance types, or combinations of impedances, such as resistance, inductance, etc.

Switching in different load types can enable a monitor value to represent the delay caused by the capacitance presented by such a load type. Additionally, in the presence of systemic variation of the IC, the monitor value can also represent the performance impact resulting from such variations.

FIG. 7 is a block schematic diagram of a drive monitor circuit 702 according to another embodiment. In one embodiment, a drive monitor circuit 702 can be one particular implementation of that shown in FIG. 3. A drive monitor circuit 702 can include sections like those of FIG. 4A, including a TUT section 716 and a load section 718.

However, unlike FIG. 4A, a TUT section 716 can take a form like that of FIG. 6A, enable the selection of any of a number of different TUTs 728-0 to 728-i. While FIG. 7 shows a TUT section 716 with n-channel TUTs, TUT section 716 can include p-channel TUTs as well. A MUX 730 can selectively connect one or more TUTs (728-0 to 728-i) to monitor node 726. Similarly, a load section 718 can take the form of that shown in FIG. 6B, enabling the selection of any of a number of different loads 734-0 to 734-j.

Accordingly, in operation, a drive monitor circuit 702 can evaluate circuit element performance by executing different monitor operations, as described herein or equivalents, that select different combinations of TUTs (728-0 to 728-i) with different loads (734-0 to 734-j). It is understood that such different monitor operations can select a single TUT and a single load, multiple TUTs and a single load, a single TUT and multiple loads, as multiple TUTs with multiple loads.

For example, in embodiments where the load section 718 includes a capacitor, the monitor section 720 can evaluate the performance of one or more transistors included in the TUT section 716 by performing a monitor operation for each of the transistors while setting the select input Z_SEL to connect the same capacitor of the load section 718 to the monitor node 726. During each monitor operation the monitor section 720 selects one of the TUTs of the TUT section 716 using the TUT_SEL input (e.g., TUT0, . . . , TUTi) and generates the monitor value Mx for the selected TUT by first charging the load section 718 to a predetermined first voltage. Typically, the TUT section 716 is configured to be in a non-conductive state while the load section 718 is being charged. After the load section 718 has been charged to VDD, the monitor operation is started by asserting the “Start” signal of the counter circuit 724 and simultaneously starting the count operation of the counter circuit 724. This stops the charging of the load section 718 and configures the selected TUT of the TUT section 716 to a conductive state such that the load section 718 starts discharging to VSS through the selected TUT. This discharging of the load section 718 continues either as long as the “Start” signal is asserted, or until the monitor node voltage reaches VSS (i.e., the load section 718 has been discharged such that the voltage across the load section 718 decreases from VDD to VSS). As the monitor node voltage decreases from VDD towards VSS, the output of the comparator 722 goes high after the monitor node 726 voltage goes below the reference voltage Vref. This activates the “Stop” signal and stops the count operation of the counter circuit 724. The resulting count value after the count operation has stopped is the monitor value Mx(count). After the monitor value for the selected TUT has been completed, the performance of additional TUTs in the TUT section are evaluated by charging the load section 718 to VDD, setting TUT_SEL input to a value that selects the next TUT of the TUT section 716 that is to be evaluated, and performing another monitor operation by activating the start signal to obtain the monitor value Mx for the next TUT. This process can be repeated to obtain the monitor value for additional TUTs in the TUT section 718.

Alternatively, the monitor circuit 702 can evaluate the performance impact resulting from different process variations and physical features of the IC device by performing the monitor operation for corresponding loads of the load section 718, e.g. 734-0 . . . j, under the control of the load select signal Z_SEL, while using the same TUT section to discharge the selected load section, then the rate at which the selected load section is discharged represents the performance impact of the load resulting from the process variation or physical feature of the IC device associated with the selected load. The value of the load select signal Z_SEL can be changed to select additional loads in the load section 718, and the steps for charging the monitor node 726 and subsequently discharging it using the same TUT section can be repeated to evaluate the performance impact of additional sources of process variation or physical features of the ID device.

FIG. 8 is a diagram showing one particular example of a metal-metal (M-M) capacitor 834 that can be included in embodiments. M-M capacitor 843 can include plate members formed by interdigitated metallization layers. Plate members 840-0/1 can be formed by a lower metallization layer. Plate members 838-0/1 formed by another metallization layer formed over members 840-0/1. Plate members 836-0/1 formed by another metallization layer formed over members 838-0/1. One plate of M-M capacitor 834 can be formed by members 836-0/838-0/840-0 conductively connected together by vertical interconnects 837-0/1. Another plate of M-M capacitor 834 can be formed by members 836-1/838-1/840-1 conductively connected together by vertical interconnects 837-0/1. Such M-M capacitors can be readily designed to match the IC timing paths, e.g., having the same line to line spacing on the same layers, as well as the requisite crossing widths and orientations so as to mimic signal lines on the IC.

Other M-M capacitors can be formed with single interdigitated metallization layer, to monitor a capacitance arising from lateral separation of wirings. Similarly, other M-M capacitors can be formed by vertically stacked metallization layers, to monitor a capacitance arising from vertical separation of wirings.

FIG. 9 is a diagram showing how differences in conductive layers can introduce variations in capacitance. FIG. 9 shows different conductive layers in relation to one another. Metallization layers can have different vertical thickness (e.g., tM4, tM5, tM6 . . . tMj, tMk), different horizontal widths (e.g., W4 . . . Wm), different vertical separation (dV4, dV5 . . . dVj, dVk) created by interlayer dielectrics, and different lateral separation (pitch) (e.g., S4 . . . Sm).

A capacitance between such layers (partially represented by capacitances C0 to C7) can vary according to process variation. Accordingly, by including capacitors formed by combinations of metallization layers as load elements in monitoring sections, the effect of such processes variation can be compensated for, to optimize circuit performance.

FIG. 10 is a block schematic diagram of an IC device 1000 according to one embodiment. An IC device 1000 can be one particular implementation of that shown as 100 in FIG. 1. An IC device 1000 can include a combination logic section 1004, an n-type body bias circuit 1006-0, a p-type body bias circuit 1006-1, and monitoring sections (not shown) that provide monitoring values to combination logic section 1004.

In the embodiment shown, an n-type body bias circuit 1006-0 can include a charge pump 1047 and a frequency adjust (frequency control) circuit 1046. A charge pump 1047 can drive its output Vbn to a potential based on a clock signal CLK_Q. Frequency adjust circuit 1046 can generate clock signal CLK_Q according to bias control values NBias, which can be generated as described for embodiments herein, or equivalents. Thus, to increase a magnitude of Vbn (i.e., drive Vbn to a lower potential), an NBias value can cause frequency adjust circuit 1046 to increase the frequency of CLK_Q. Conversely, to decrease a magnitude of Vbn (i.e., drive Vbn to a higher potential), an NBias value can cause frequency adjust circuit 1046 to decrease the frequency of CLK_Q.

In the embodiment shown, a p-type body bias circuit 1006-1 can include a switched capacitor network 1045 and a switch control circuit 1044. A switched capacitor network 1045 can drive its output Vbp to a potential based on a switch control network 1044. Switch control network 1044 can generate control signals based on bias control values PBias, which can be generated as described for embodiments herein, or equivalents.

Body bias circuits as described herein, can generate forward body bias voltages VFB and reverse body bias voltages VBB. In particular embodiments, a forward body bias voltage can be generated by creating a voltage offset from a generated body bias voltage. One such embodiment is shown in FIG. 11.

FIG. 11 is a block schematic diagram of a power switching circuit 1100 according to an embodiment. A power switching circuit 1100 can include a p-type body bias circuit 1106-1, a voltage offset circuit 1148, a mode select MUX 1150, a power switching device 1152, and a load section 1156. A power switching circuit 1100 can provide a low impedance power supply path to a load 1156 that can vary according to mode of operation.

A p-type body bias circuit 1106-1 can generate a body bias voltage Vbp based on bias control values PBias. Bias control values PBias can be generated according to weighted monitor values, as described herein. That is, PBias can be a function of one or more monitor values (Mx) weighted by one or more weight values (Wx). In one particular embodiment, a body bias voltage Vbp can be a reverse body bias voltage VRB that can increase a threshold voltage and decrease leakage of a p-channel transistor.

A voltage offset circuit 1148 can generate a voltage having a predetermined offset with respect to a received voltage. In some embodiments, an offset can be a voltage lower than a received voltage. In the embodiment shown, voltage offset circuit 1148 can generate a voltage V1, where V1=Vbp−Voffset. V1 can be a forward body bias voltage VFB. In one particular embodiment, a voltage offset circuit can be a Widlar voltage source circuit, but any suitable offset circuit can be employed.

A mode select MUX 1150 can selectively apply either Vbp or V1 as a body bias voltage for power switching device 1152. In one mode (e.g., an active or high performance mode), mode select MUX 1150 can apply V1 as a body bias voltage, to lower a threshold voltage of a switching device 1152. In a second mode (e.g., an inactive or low leakage mode), mode select MUX 1150 can apply Vbp as a body bias voltage, to increase a threshold voltage of a switching device 1152.

A switching device 1152 can be one or more p-channel transistors having a body connected to an output of mode select MUX 1150, a source connected to a high power supply VDD, a drain connected to an internal power supply node 1158, and a gate connected to receive an enable signal EN. By operation of switching device 1152, a voltage VDD_PG can be generated at internal power supply node 1158, to provide power to load section 1156.

A load section 1156 can be connected between internal power supply node 1158 and a low power supply VSS. In some embodiments, a load section 1156 can include a circuit that provides some predetermined function for the IC device containing the power switching circuit 1100.

FIG. 12 is a block schematic diagram of a core biasing circuit 1256 that can be used in an embodiment like that of FIG. 11. In one embodiment, core biasing circuit 1256 can correspond to load section 1156 of FIG. 11. A core biasing circuit 1256 can include a “core” section 1262 and a core bias MUX 1260.

A core section 1262 can perform a predetermined function, and can include complementary (i.e., p-channel and n-channel) transistors, having bodies that receive adjustable body bias voltages (Vpcore, Vbn). A core section 1262 can be formed between an internal power supply node 1258 VDD_PG and a low power supply node VSS. In one embodiment, an internal power supply node 1258 VDD_PG can correspond to that shown as 1158 in FIG. 11 (i.e., it can be switched between a high body bias Vbp and a lower VFB of V1). In the embodiment shown, n-channel transistors of core section 1262 can have bodies biased to a voltage Vbn. A voltage Vbn can be a generated according to embodiments shown herein, or equivalents. Bodies of p-channel transistors of core section 1262 can be biased according to an output from core bias MUX 1260.

Core bias MUX 1260 can selectively apply either a body bias voltage Vbp or the voltage at internal power supply node VDD_PG (which can be less than Vbp), according to a mode value MODE_SEL. More particularly, in one mode (e.g., an active or high performance mode), core bias MUX 1260 can apply VDD_PG as a body bias voltage. In a second mode (e.g., an inactive or low leakage mode), core bias MUX 1260 can apply Vbp as a body bias voltage. Such a local switching of the p-channel wells of core section 1262 can reduce band to band tunneling junction current when a switching device (i.e., 1152 of FIG. 11) is off, and VDD_PG can be at a relatively low voltage.

In some embodiments, body biasing, as described for a power gating switch in FIG. 11, can also be applied to a device that supplies a regulated voltage. One such embodiment is shown in FIG. 13.

FIG. 13 is a block schematic diagram of a regulator circuit according to an embodiment. A regulator circuit 1300 can include sections like those of FIG. 11, including a p-type body bias circuit 1306-1, a voltage offset circuit 1348, and a mode select MUX 1350. Such like sections can operating in the same fashion as those shown in FIG. 11, or equivalents.

A voltage regulating circuit 1300 can also include a low dropout regulator (LDO) section 1368 and core section 1362. An LDO section 1368 can include a regulating device 1352, a mode control MUX 1364, and an amplifier 1366. A regulating device 1352 can be a p-channel transistor having a body connected to an output of mode select MUX 1350, a source connected to a high power supply VDD, a drain connected to a regulated node 1358, and a gate connected to receive a regulation control voltage Vreg, output from mode control MUX 1364.

A mode control MUX 1364 can selectively output either a high power supply voltage VDD or an output of amplifier 1366. In one mode (e.g., a regulated mode), mode control MUX 1364 can apply the output of amplifier 1366, which can drive a gate of regulating device 1352 to maintain regulated node 1358 at a predetermined voltage. In a second mode (e.g., an inactive or low leakage mode), mode control MUX 1364 can apply VDD as a body bias voltage, to turn off regulating device 1352.

An amplifier 1366 can provide an output that can vary according to a feedback voltage Vfdbk provided from core section 1362. Thus, if a voltage at regulated node 1358 is too low, Vfdbk can fall below Vref, and an output of amplifier 1366 can be driven lower. Conversely, if a voltage at regulated node 1358 is too high, Vfdbk can rise above Vref, and an output of amplifier 1366 can be driven higher.

A core section 1362 is represented by load sections 1356-0 and 1356-1. It is understood that such sections (1356-0/1) can represent circuits for performing the various functions of an IC device.

Body biasing as described for embodiments herein can be applied to bodies of various transistor types. Examples of some transistor types which can receive a body bias voltage are shown in FIG. 14A to 15.

FIG. 14A shows a bulk substrate n-channel transistor 1470-A, in which a transistor body can be biased via a bulk substrate 1408-A. FIG. 14B shows a p-channel transistor 1470-B formed in an n-type well, formed in a p-type substrate. In such a transistor, a body can be biased via well 1408-B.

FIGS. 14C and 14D show a semiconductor-on-insulator type transistor (e.g., silicon-on-insulator, SOI) having an active semiconductor region 1475 separated in a vertical direction from a bulk substrate 1472 by an insulating layer 1474. Active region 1475 can be separated in a lateral direction from other active regions (not shown) by lateral insulating structures 1476.

FIG. 14C shows an n-channel transistor 1470-C having source and drain regions that do not extend to insulating layer 1474. A body region can be biased via body region 1408-C. FIG. 14D shows an n-channel transistor 1470-C having source and drain regions that extend to insulating layer 1474. A body region can be biased via body region 1408-D.

FIG. 15A shows a deeply depleted channel (DDC) type transistor 1570, which can be included in embodiments. A DDC transistor 1570 can be configured to have an enhanced body coefficient, along with the ability to set a threshold voltage (Vt) with enhanced precision. A DDC transistor 1570 can include a gate electrode 1582, source 1584, drain 1586, and a gate dielectric 1588 positioned over a substantially undoped channel 1511. Optional lightly doped source and drain extensions (SDE) 1590 can be positioned respectively adjacent to source 1584 and drain 1586. Such extensions 1590 can extend toward each other, reducing effective length of the substantially undoped channel 1511.

In FIG. 15A, the DDC transistor 1570 is shown as an n-channel transistor having a source 1584 and drain 1586 made of n-type dopant material, formed upon a substrate such as a p-type doped silicon substrate providing a p-well 1517. In addition, the n-channel DDC transistor 1570 in FIG. 15A can include a highly doped screening region 1515 made of p-type dopant material, and a threshold voltage set region 1513 made of p-type dopant material.

FIG. 15B shows FinFET type transistor 1570-B which can be included in embodiments. The FinFET transistor 1570-B can include a gate electrode 1582-B and gate dielectric 1588-B that surround a substantially undoped channel 1511-B on opposing sides. The view of FIG. 15B is taken along a channel length. Thus, it is understood that source and drain regions can extend into and out of the view shown. Further, such source and drain regions can be separated from screening region 1515-B by portions of undoped channel region 1511-B.

FIG. 15C shows a FinFET type transistor 1570-C having a screening region 1515-C which can be included in embodiments. As in the case of FIG. 15A, the FinFET transistor 1570-C having a screening region can be configured to have an enhanced body coefficient, along with the ability to set a Vt with enhanced precision. The transistor 1570-C includes a gate electrode 1582-C and gate dielectric 1588-C formed over a substantially undoped channel 1511-C on opposing sides. However, unlike FIG. 15B, a highly doped screening region 1515-C is formed in a substrate 1519 below substantially undoped channel 1511-C. Optionally, a Vt set region 1513-C is formed between the screening region 1515-C substantially undoped channel 1511-C.

As in the case of FIG. 15B, the view of FIG. 15C is taken along a channel length, and source and drain regions can extend into and out of the view, separated from screening region 1515-C by portions of undoped channel region 1511-C.

Embodiments of various structures and manufacturing processes suitable for use in DDC transistors are disclosed in U.S. patent application Ser. No. 12/708,497, filed on Feb. 18, 2010, titled Electronic Devices and Systems, and Methods for Making and Using the Same, by Scott E. Thompson et al.; U.S. Pat. No. 8,273,617, issued on Sep. 25, 2012, titled Low Power Semiconductor Transistor Structure and Method of Fabrication Thereof; U.S. patent application Ser. No. 12/971,955 filed on Dec. 17, 2010 titled Transistor with Threshold Voltage Set Notch and Method of Fabrication Thereof; U.S. patent application Ser. No. 12/895,785 filed on Sep. 30, 2010 titled “Advanced Transistors With Threshold Voltage Set Dopant Structures”; and U.S. patent application Ser. No. 13/071,399 filed on Mar. 24, 2011, titled “Analog Circuits Having Improved Transistors, and Method Therefor”, by Lawrence T. Clark et al; the disclosures of which are hereby incorporated by reference in their entirety. In addition, methods and structures for modifying the threshold voltage of DDC transistors are discussed in pending U.S. patent application Ser. No. 13/459,971 titled “Multiple Transistor Types Formed in a Common Epitaxial Layer by Differential Out-diffusion from a Doped Underlayer”, filed Apr. 30, 2012, the entirety of which disclosure is incorporated by reference herein.

It is understood that, with appropriate change to substrate or dopant material, conductivities of any of the transistors described above can be switched (i.e., from p-channel to n-channel and vice versa).

Having described various circuit features, methods of monitoring and adjusting transistor performance will now be described.

FIG. 16A-0 is a graph showing performance variations of n-channel TUTs in a circuit like that of FIG. 4A. Curve 1651 shows transistor performance for a “slow” n-channel TUT. Curve 1653 shows transistor performance for a “fast” n-channel TUT. Curve 1655 shows a nominal transistor performance for an n-channel TUT. The n-channel TUTs of FIG. 16A-0 are DDC transistors like that shown in FIG. 15, having a reverse body bias voltage of 0.3V, and source-drain voltage (VDS) of 0.9V. A monitor circuit can determine the performance of the TUT by determining the duration of time taken by a monitor node voltage to dis-charge from an initial voltage level to a predetermined voltage level. For example, in the embodiment shown in the figure, the monitor circuit determines three different performance levels for the TUT (shown as t_fast, t_nom, and t_slow) by determining the time (e.g. count) at which the voltage at the monitor node falls below 0.44V.

FIG. 16A-1 is a graph showing how an adjustment of a body bias can compensate for variations in transistor performance. Curve 1651′ shows a “slow” n-channel device performance after the body bias has been adjusted in the forward body bias direction to 0.04V. As shown, after such an adjustment, the slow n-channel transistor can closely follow the nominal case 1655. Similarly, curve 1653′ shows a “fast” n-channel device performance after the body bias has been adjusted in the reverse body bias direction to 0.46V. As shown, after such an adjustment, the fast n-channel transistor now closely follows the nominal case 1655.

FIG. 16B-0 is a graph showing performance n-channel TUTs like that of FIG. 16A-0, but at a lower source-drain voltage (VDS) of 0.6V, and a capacitive load that is 20% of that used for FIG. 16A-0. Like FIG. 16A-0, curve 1671 shows transistor performance for a slow n-channel TUT, curve 1673 shows transistor performance for a fast n-channel TUT, curve 1675 shows a nominal transistor performance for an n-channel TUT. Speeds of transistors can are based on the time (e.g., count) when a monitor node falls below 0.3V (shown again as t_fast, t_nom, and t_slow).

FIG. 16B-1 shows how adjustment of a body bias can compensate for variations in transistor performance in the same manner as FIG. 16A-1 (but with different body bias adjustments).

Accordingly, according to monitor results, a body bias voltage can be adjusted to bring a transistor performance within a desired limit. Thus, if devices are manufactured having different transistor performance, such difference can be detected by monitoring circuits, and a body bias adjusted to enable the devices to have similar or substantially identical performance to the nominal case. Thus the IC can have its systematic variability compensated towards the nominal IC performance, thereby limiting IC non-uniformity due to manufacturing variations.

The monitor circuit embodiments described above can be used to implement a fine grained body biasing scheme, where the body bias voltages used for individual functional blocks or different regions of the integrated circuit can be independently adjusted to compensate for local variations (e.g., process, temperature, or voltage variations) across the integrated circuit. FIG. 17 shows one embodiment of a bias voltage distribution network that uses monitor circuits to implement a fine grained body biasing technique. FIG. 17 shows an integrated circuit 1700, with four functional blocks 1702, 1704, 1706, and 1708. Functional block 1702 includes a monitor circuit 1710 and a body bias tuning circuit 1712, functional block 1704 includes a monitor circuit 1714 and a body bias tuning circuit 1716, functional block 1706 includes a monitor circuit 1718 and a body bias tuning circuit 1720, and functional block 1708 includes a monitor circuit 1722 and body bias tuning circuit 1724. The functional blocks 1702-1708 can be SRAM arrays having different sizes or different performance targets, graphics processors, general purpose processors, memory controllers, etc. In addition, the design of one or more of the functional blocks 1702-1708 can be the same as one of the other functional blocks on the integrated circuit, where the different identically designed functional blocks are configured to operate at different performance targets (e.g., speed, power consumption, etc.) by the application of different body bias voltages. In alternative embodiments, the integrated circuit 1700 can be divided into several regions, instead of functional blocks where each region has a monitor circuit and a body bias tuning circuit.

The integrated circuit 1700 uses one body bias generator 1750 that generates a pmos body bias voltage and an nmos body bias voltage sufficient to provide the highest body bias voltage that is to be supplied to the p-channel and n-channel transistors, respectively, of the functional blocks in the integrated circuit 1700. The nmos and pmos body bias voltages generated by the body bias generator 1750 are distributed to the functional blocks 1702-1708 using wires 1752 and 1754, respectively. The body bias tuning circuit at each functional block, receives the body bias voltages on the wires 1752 and 1754, and generates local pmos and nmos body bias voltages for the corresponding functional block in accordance with the body bias control signal received from the monitor circuit of that functional block. In alternative embodiments, where only one body bias voltage is required (e.g., only the nmos or pmos transistors are back biased), the body bias generator 1750 can generate only one body bias voltage, and only one wire can be used to distribute that body bias voltage to the functional blocks in the integrated circuit.

FIG. 18 shows one embodiment of the bias tuning circuit using a resistive divider 1805 to generate the local body bias voltage 1815. In the embodiment shown, the voltage 1810 is received from the body bias generator, and different body bias voltages are generated using the resistive divider 1805. The number of resistors (R) and the values of the resistors in the resistive divider are determined in accordance with the different body bias voltages to be provided to the corresponding functional block. A switch network 1825 receives the voltages generated by the resistive divider and connects the generated voltage selected by the bias voltage select signal 1820 to the local body bias voltage terminal 1815. The body bias voltage select signal is received from the monitor circuit of the functional block.

It should be appreciated that in the foregoing description of exemplary embodiments of the invention, various features of the invention are sometimes grouped together in a single embodiment, figure, or description thereof for the purpose of streamlining the disclosure aiding in the understanding of one or more of the various inventive aspects. This method of disclosure, however, is not to be interpreted as reflecting an intention that the claimed invention requires more features than are expressly recited in each claim. Rather, as the following claims reflect, inventive aspects lie in less than all features of a single foregoing disclosed embodiment. Thus, the claims following the detailed description are hereby expressly incorporated into this detailed description, with each claim standing on its own as a separate embodiment of this invention.

It is also understood that the embodiments of the invention may be practiced in the absence of an element and/or step not specifically disclosed. That is, an inventive feature of the invention may be elimination of an element.

Accordingly, while the various aspects of the particular embodiments set forth herein have been described in detail, the present invention could be subject to various changes, substitutions, and alterations without departing from the spirit and scope of the invention.