阅读说明：本技术 全耗尽型绝缘体上硅闪存存储器设计 (Fully depleted silicon-on-insulator flash memory design ) 是由 H·V·陈 A·李 T·武 H·Q·阮 于 2016-09-14 设计创作，主要内容包括：本发明涉及一种闪存存储器系统,其中一个或多个电路区块利用全耗尽型绝缘体上硅晶体管设计来使泄漏最小化。(The present invention relates to a flash memory system in which one or more circuit blocks utilize a fully depleted silicon-on-insulator transistor design to minimize leakage.)

1. A flash memory system, comprising:

an array of flash memory cells; and

a sense amplifier for reading a selected memory cell in the array of flash memory cells, the sense amplifier comprising one or more fully depleted silicon-on-insulator NMOS transistors and one or more fully depleted silicon-on-insulator PMOS transistors.

2. The flash memory system of claim 1, wherein the sense amplifier comprises a first input coupled to a first plane of the array and a second input coupled to a second plane of the array.

3. The flash memory system of claim 2, wherein the first plane provides a selected bit line with a selected memory cell and the second plane provides another selected bit line with an unselected memory cell.

4. The flash memory system of claim 2, wherein for the fully depleted silicon-on-insulator MOS, a forward body bias is used in an active mode and a reverse body bias is used in a standby mode.

5. The flash memory system of claim 4, wherein the body bias is used to reduce an offset of the sense amplifier.

6. The flash memory system of claim 1, wherein the sense amplifier comprises:

one or more fully depleted silicon-on-insulator NMOS transistors having an n-well under a buried oxide layer;

one or more fully depleted silicon-on-insulator NMOS transistors having a p-well under a buried oxide layer;

one or more fully depleted silicon-on-insulator PMOS transistors having an n-well under a buried oxide layer; and

one or more fully depleted silicon-on-insulator NPMOS transistors having a p-well under a buried oxide layer.

7. The flash memory system of claim 1, wherein the memory cells comprise source side injection flash memory cells, each source side injection flash memory cell comprising:

an erase gate to provide erase;

a coupling gate; and

a source line for providing a programming current.

8. A flash memory sensing system, comprising:

an array of flash memory cells; and

a sense amplifier for reading a selected memory cell in the array of flash memory cells, the sense amplifier comprising a first readout stage having bulk CMOS transistors and a second comparison stage having one or more fully depleted silicon-on-insulator NMOS transistors and one or more fully depleted silicon-on-insulator PMOS transistors.

9. The flash memory sensing system of claim 8, wherein the first readout stage is coupled to a power supply having a higher voltage than a power supply coupled to the second stage.

10. The flash memory sensing system of claim 8, wherein the first sensing stage comprises a PMOS transistor coupled to a reference current and an NMOS transistor coupled to a selected memory cell current.

11. The flash memory sensing system of claim 8, wherein the second stage comprises a latched inverter.

12. The flash memory sensing system of claim 11, wherein the latched inverter is coupled to a NMOS input pair coupled to an output of the first sense stage.

13. A flash memory system, comprising:

an array of flash memory cells arranged in rows and columns; and

a row decoder for selecting a row of flash memory cells in the array for a read or write operation, the row decoder comprising one or more fully depleted silicon-on-insulator NMOS transistors and one or more fully depleted silicon-on-insulator PMOS transistors;

a column decoder for selecting a column of flash memory cells in the array for a read or write operation, the column decoder comprising one or more fully depleted silicon-on-insulator NMOS transistors each having a p-well under a buried oxide layer, and one or more fully depleted silicon-on-insulator NMOS transistors each having an n-well under a buried oxide layer; and

a sense amplifier for reading a selected memory cell in the array of flash memory cells, the sense amplifier comprising one or more fully depleted silicon-on-insulator NMOS transistors and one or more fully depleted silicon-on-insulator PMOS transistors.

14. The flash memory system of claim 13, wherein the flash memory system further comprises a high voltage decoder, the high voltage decoder comprising:

an erase gate decoder;

a coupling gate decoder; and

and a source line decoder.

15. The flash memory system of claim 14, wherein the high voltage decoder further comprises an enable latch comprising a fully depleted silicon-on-insulator transistor.

16. The flash memory system of claim 14, wherein the erase gate decoder, the coupling gate decoder, and the source line decoder each comprise bulk CMOS transistors.

17. The flash memory system of claim 13, wherein each flash memory cell comprises:

an erase gate to provide erase;

a coupling gate; and

a source line for providing a programming current.

18. The flash memory system of claim 13, wherein the array comprises a plurality of flash sub-arrays, each of the flash sub-arrays comprising an underlying individual p-well.

Technical Field

The present invention relates to flash non-volatile memory systems in which one or more circuit blocks utilize a fully depleted silicon-on-insulator transistor design to minimize leakage and optimize performance.

Background

A prior art non-volatile memory cell 110 is shown in fig. 1. The memory cell 110 includes a semiconductor substrate 112 of a first conductivity type, such as P-type. The substrate 112 has a surface on which a first region 114 (also referred to as a source line SL) of a second conductivity type, such as N-type, is formed. A second region 116, also of N-type (also referred to as a drain line), is formed on the surface of the substrate 112. Between the first region 114 and the second region 116 is a channel region 118. The bit line BL 120 is connected to the second region 116. Word line WL 122 is positioned over and insulated from a first portion of channel region 118. The word line 122 hardly overlaps or does not overlap the second region 116 at all. Over another portion of channel region 118 is a floating gate FG 124. The floating gate 124 is insulated from the other portion and is adjacent to the word line 122. The floating gate 124 is also adjacent to the first region 114. The floating gate 124 may overlap the first region 114 to provide coupling of the region 114 to the floating gate 124. A coupling gate CG (also referred to as a control gate) 126 is located over and insulated from the floating gate 124. An erase gate EG 128 is over the first region 114 and adjacent to and insulated from the floating gate 124 and the coupling gate 126. The top corner of the floating gate 124 may be directed to the inside corner of the T-shaped erase gate 128 to improve erase efficiency. The erase gate 128 is also insulated from the first region 114. Element 110 is described in more detail in USP 7,868,175, the disclosure of which is incorporated herein by reference in its entirety.

One exemplary operation of erasing and programming a prior art non-volatile memory cell 110 is as follows. The cell 110 is erased by a fowler-nordheim tunneling mechanism by applying a high voltage on the erase gate 128 while the other terminal is at zero volts. Electrons tunnel from the floating gate 124 into the erase gate 128, causing the floating gate 124 to be positively charged, thereby turning on the cell 110 under read conditions. The resulting cell erased state is referred to as the '1' state. The cell 110 is programmed by a source side hot electron programming mechanism by applying a high voltage on the coupling gate 126, a high voltage on the source line 114, a medium voltage on the erase gate 128, and a programming current on the bit line 120. A portion of the electrons flowing through the gap between the word line 122 and the floating gate 124 gain sufficient energy to be injected into the floating gate 124, causing the floating gate 124 to be negatively charged, thereby turning off the cell 110 under read conditions. The resulting programmed state of the cell is referred to as the '0' state.

Exemplary voltages that may be used for read, program, and erase operations in memory cell 110 are shown in table 1 below:

note that: "FLT" means floating.

For a program operation, the EG voltage may be applied much higher (e.g., 8V) than the SL voltage (e.g., 5V) to enhance the program operation. In this case, the unselected CG programming voltage is applied at a higher voltage (CG inhibit voltage), such as 6V, to reduce the undesirable erase effect of neighboring memory cells sharing the same EG gate of the selected memory cell.

Another set of exemplary voltages that may be used for read, program, and erase operations in memory cell 310 (when negative voltages may be used for read and program operations) are shown in table 2 below:

another set of exemplary voltages that may be used for read, program, and erase operations in memory cell 310 (when negative voltages may be used for read, program, and erase operations) are shown in table 3 below:

for a program operation, the EG voltage is applied much higher (e.g., 8-9V) than the SL voltage (e.g., 5V) to enhance the program operation. In this case, the unselected CG programming voltage is applied at a higher voltage (CG inhibit voltage), e.g., 5V, to reduce the undesirable erase effect of neighboring memory cells sharing the same EG gate of the selected memory cell.

Also known in the art are fully depleted silicon-on-insulator ("FDSOI") transistor designs, as shown in fig. 2-4. FDSOI advantages include a back gate (with buried oxide as the gate oxide) to modulate the threshold voltage (forward or reverse body bias), an ultra-thin undoped channel resulting in higher mobility, and no random doping fluctuations. It has a ground plane on the back gate to adjust the implant and thus the threshold voltage. It also has a fully depleted channel to provide better electrostatic control, lower drain induced barrier reduction DIBL, and short channel effects. With minimal source drain junctions. The threshold voltage is also adjusted using the metal gate and channel length.

Fig. 2 shows an FDSOI CMOS circuit cross-section 210. The FDSOI CMOS circuit 210 includes a silicon substrate 211, a silicon-on-insulator 216, an FDSOI NMOS transistor 230, and an FDSOI PMOS transistor 240.

The FDSOI NMOS transistor 230 includes a gate 218 and source and drain 217. FDSOI NMOS transistor 230 also includes p-well 212, buried oxide layer 213 (which is an insulator), and channel 215. Channel 215 is an undoped fully depleted channel. During operation, the buried oxide layer 213 minimizes any leakage from the channel 214. FDSOI NMOS transistor 230 also includes a p-well back gate terminal 219 that may be used to apply a bias to p-well 212 in order to adjust the threshold voltage Vt of NMOS 230.

The FDSOI PMOS transistor 240 includes a gate 228 and source and drain 227. FDSOI PMOS transistor 240 also includes an n-well 222, a buried oxide layer 223 (which is an insulator), and a channel 225. Channel 225 is an undoped fully depleted channel. During operation, the buried oxide layer 223 minimizes any leakage from the channel 225. The FDSOI PMOS transistor 240 also includes an n-well back gate terminal 229 that may be used to apply a bias to the n-well 222 in order to adjust the threshold voltage Vt of the PMOS 240.

Fig. 3 shows an FDSOI CMOS circuit cross-section 310. The FDSOI CMOS 310 circuit includes a silicon substrate 311, a silicon-on-insulator 316, an FDSOI NMOS transistor 330, and an FDSOI PMOS transistor 340.

The FDSOI NMOS transistor 330 includes a gate 318 and source and drain 317. FDSOI NMOS transistor 330 also includes an n-well 312, a buried oxide layer 313 (which is an insulator), and a channel 315. Channel 315 is an undoped fully depleted channel. During operation, the buried oxide layer 313 minimizes any leakage from the channel 315. The FDSOI NMOS transistor 330 also includes an n-well back gate terminal 319 that may be used to apply a bias to the n-well 312 in order to adjust the threshold voltage Vt of the NMOS 330.

The FDSOI PMOS transistor 340 includes a gate 328 and source and drain 327. The FDSOI PMOS transistor 340 also includes a p-well 312, a buried oxide layer 323 (which is an insulator), and a channel 325. Channel 325 is an undoped fully depleted channel. During operation, the buried oxide layer 323 minimizes any leakage from the channel 325. The FDSOI PMOS transistor 340 also includes a p-well backgate terminal 329 that can be used to apply a bias to the p-well 322 in order to adjust the threshold voltage Vt of the PMOS 340.

Fig. 4 shows an FDSOI and bulk CMOS hybrid MOS circuit cross-section 410. Bulk CMOS refers to standard PMOS and NMOS transistors on bulk silicon. Hybrid MOS circuit 410 includes silicon substrate 411, silicon-on-insulator 416, FDSOI NMOS transistor 430, and NMOS transistor 440. The NMOS transistor 440 is a conventional NMOS transistor, not a FDSOI NMOS transistor.

The FDSOI NMOS transistor 430 includes a gate 418 and source and drain 417. The FDSOI NMOS transistor 430 also includes a p-well 412, a buried oxide layer 413 (which is an insulator), and a channel 415. Channel 415 is an undoped fully depleted channel. During operation, the buried oxide layer 413 minimizes any leakage from the channel 415. The FDSOI NMOS transistor 430 also includes a p-well back-gate terminal 419 that may be used to apply a bias to the p-well 412 in order to adjust the threshold voltage Vt of the NMOS 430.

NMOS transistor 440 includes a gate 428 and source and drain 427. The NMOS transistor 440 also includes a p-well body 422 and a doped channel 423. NMOS transistor 440 also includes a p-well body terminal 429 that may be used to apply bias to p-well body 422.

To date, fully depleted silicon-on-insulator transistor designs have not been used in flash memory systems. What is needed is a flash memory system that utilizes a fully depleted silicon-on-insulator transistor design. What is also needed is a partitioned flash memory chip that includes a body region and an FDSOI region to maximize area and minimize leakage.

Disclosure of Invention

In the embodiments described below, the flash memory device utilizes partitions that include body transistors, and the partitions include FDSOI transistors.

Drawings

FIG. 1 is a cross-sectional view of a prior art non-volatile memory cell.

Fig. 2 is a cross-sectional view of a prior art FDSOI CMOS circuit.

Fig. 3 is a cross-sectional view of a prior art FDSOI CMOS circuit.

Fig. 4 is a cross-sectional view of a prior art FDSOI CMOS circuit.

Fig. 5 illustrates various types of FDSOI NMOS and PMOS transistors used in embodiments.

Fig. 6 shows a die used in an embodiment.

Figure 7 shows the basic components of an array used in an embodiment.

FIG. 8 shows a decoder for generating different voltages for use with an embodiment.

Fig. 9 shows an embodiment of a row decoder.

Fig. 10 shows another embodiment of a row decoder.

Fig. 11 shows another embodiment of a row decoder.

Fig. 12 shows another embodiment of a row decoder.

Figure 13 shows an embodiment of an erase gate decoder.

FIG. 14 shows an embodiment of a source line decoder.

Fig. 15 illustrates an embodiment of a high voltage logic selector circuit.

Fig. 16 shows an embodiment of a coupled gate decoder.

Fig. 17 shows an embodiment of a low logic voltage circuit.

Fig. 18 shows a sensing system that can be used in embodiments.

Fig. 19 shows an embodiment of a sense amplifier.

FIG. 20 shows another embodiment of a sense amplifier.

FIG. 21 shows another embodiment of a sense amplifier.

FIG. 22 shows another embodiment of a sense amplifier.

FIG. 23 shows an embodiment of a column decoder.

Detailed Description

Fig. 5 shows eight FDSOI transistor types used in the embodiments described herein.

The standard fixed-bias FDSOI MOS transistor includes a PMOS transistor 510 and an NMOS transistor 550. The FDSOI PMOS transistor 510 includes an n-well biased to Vdd supply and optionally to ground potential, in which case the transistor channel length is modified to have a similar threshold voltage level. The FDSOI NMOS transistor 550 includes a p-well that is biased to ground potential. PMOS 510 and NMOS 550 are normal threshold voltage devices.

The flip-well fixed bias FDSOI MOS transistor includes a PMOS transistor 520 and an NMOS transistor 560. The FDSOI PMOS transistor 520 includes a p-well that is biased to ground potential. The FDSOI NMOS transistor 560 includes an n-well that is biased to ground potential. PMOS 520 and NMOS 560 are low threshold voltage devices, i.e., their threshold voltages are lower than the threshold voltages of PMOS 510 and NMOS 550.

The standard dynamic bias FDSOI MOS transistor includes a PMOS transistor 530 and an NMOS transistor 570. The FDSOI PMOS transistor 530 includes an n-well that is biased to the dynamic voltage source Vb _ PRW. FDSOI NMOS transistor 570 includes a p-well biased to a dynamic voltage source Vb _ NRW. The dynamic voltage source is used to forward body (well) bias the FBB or reverse body bias the RBB to optimize performance. For PMOS 530, the dynamic voltage source Vb _ PRW changes to a positive voltage of RBB (e.g., up to 3V) and to a negative voltage of FBB (e.g., up to-0.5V). For NMOS 570, the dynamic voltage source Vb NRW changes to a positive voltage (e.g., 0V to 3V) for FBB and to a negative voltage (e.g., 0V to-3V) for RBB. A deep n-well is required to isolate the p-well from the p-substrate, allowing the p-well to be biased at a high level (e.g., 3V or-3V).

The flip-well dynamically biased FDSOI MOS transistor includes a PMOS transistor 540 and an NMOS transistor 580. FDSOI PMOS transistor 540 includes a p-well that is biased to a dynamic voltage source Vb _ PLW. FDSOI NMOS transistor 580 includes an n-well that is biased to the dynamic voltage source Vb _ NLW. For PMOS 540, dynamic voltage source Vb _ PLW changes to a positive voltage of RBB (e.g., 0V to 3V) and to a negative voltage of FBB (e.g., 0V to-3V). For NMOS 580, the dynamic voltage source Vb NLW changes to a positive voltage (e.g., 0V to 3V) for FBB and to a negative voltage (e.g., 0V to-0.5V) for RBB. A deep n-well is required to isolate the p-well from the p-substrate, allowing the p-well to be biased at a high level (e.g., 3V or-3V).

In the following embodiments, one or more of the eight types of FDSOI transistors shown in fig. 5 are used in a flash memory system.

Fig. 6 illustrates an embodiment of an architecture of a flash memory system including a die 600. The die 600 includes: flash memory arrays 601 comprising rows and columns of memory cells of the type previously described in FIG. 1 as memory cells 110; row decoder circuits 602 for accessing rows in the flash memory array 601 for reading or writing; column decoder circuits 603 for accessing bytes in the flash memory array 601 for reading or writing; sense circuits 604 for reading data from the flash memory array 601; a High Voltage (HV) decoder 620 consisting of HV decode block 610 and HV transfer blocks 609 and 611 to deliver the voltages and biases required for non-volatile operation of the flash memory array 601; control logic 605 for providing various control functions such as redundancy and built-in self-test; an analog circuit 606; a body bias control 607 for controlling the voltage of the body (well) region of the transistor; a high voltage charge pump circuit 608 for providing elevated voltages for program and erase operations of the flash memory array 601. The chip partitioning of the blocks of FDSOI and bulk CMOS regions for best performance is as follows.

● Row decoder 602: standard Vt, flipped well Vt, dynamic Vt FDSOI

● column decoder 603: standard Vt, flipped well Vt, dynamic Vt FDSOI

● sense Circuit 604: standard Vt, flipped well Vt, dynamic Vt FDSOI

● control logic 605: standard Vt, flip well Vt FDSOI

● analog circuit 606: standard Vt, flipped well Vt, dynamic Vt FDSOI

● body bias control circuit 607: standard Vt, flipped well Vt, dynamic Vt FDSOI

● HV Charge Pump Circuit 608: bulk CMOS and FDSOI hybrid, FDSOI region comprising standard Vt, flip well Vt, dynamic Vt FDSOIHV decoder circuit 620: the bulk CMOS is mixed with the FDSOI region, which includes a standard Vt, a flip well Vt, a dynamic Vt FDSOI.

An embodiment of an array 601 is shown in FIG. 7. The array 601 includes a first plurality of sub-arrays 701 and a second plurality of sub-arrays 702. Here, the first plurality of subarrays 701 have bias voltages applied to their p-well and n-well regions (to achieve higher performance), and the second plurality of subarrays 702 do not have bias voltages applied to their p-well and n-well regions (to achieve less leakage). The array 601 also includes a row decoder 703, a high voltage subarray source 704, and a high voltage decoder 705.

Fig. 8 shows a decoder 800 for generating bias control voltages P1_ PW, P2_ PW, N1_ NW, and N2_ NW, which are used in the following embodiments. Decoder 800 includes NAND gate 801, inverter 802, and programmable voltage sources 803, 804, 805, and 806 as shown.

Fig. 9 shows a row decoder 900. The row decoder 900 includes a NAND gate 951, an inverter 952, and PMOS transistors 953, 954, 956, 958, 959, and 961 and NMOS transistors 955, 957, 960, and 962, as shown. The NAND gate 951 and the inverter 952 function as a row address decoder to decode the address signals XPA-D, thereby implementing row address decoding. PMOS 956 and NMOS 957 act as row drivers with stronger strength to drive the predetermined signal ZVDD into word line WL0-7 of the memory cell. PMOS 954, PMOS 953, and NMOS 955 serve a dual function as both a row pre-driver and as a decoder for address signal XPZB 0-7.

The NAND gate 951 includes a transistor of type FDSOI PMOS 520 with the P-well biased to P2_ PW and a transistor of type FDSOI NMOS 560 with the N-well biased to N2_ NW.

Inverter 952 includes transistors of type FDSOI PMOS 520 with P-well biased to P1_ PW, and transistors of type FDSOI NMOS 560 with N-well biased to N1_ NW.

PMOS transistors 953, 954, 958, and 959 are transistors of type FDSOI PMOS 520 with the P-well biased to P2_ PW. PMOS transistors 956 and 961 are transistors of type FDSOI PMOS 520 with the P-well biased to P1_ PW.

NMOS transistors 955 and 960 are transistors of type FDSOI NMOS 560 with the N-well biased to N2_ NW. NMOS transistors 957 and 962 are transistors of type FDSOI NMOS 560 with the N-well biased to N1_ NW. The well bias levels of P1_ PW/P2_ PW/N1_ NW/N2_ NW are such that speed performance is achieved using forward bias FBB and leakage is reduced using reverse bias RBB.

Fig. 10 shows a row decoder 1000. The row decoder 1000 is identical in structure to the row decoder 900 except that all transistors are of the type FDSOI PMOS 520 with the P-well biased to P1_ PW. The well bias level of P1_ PW is such that speed performance is achieved using forward biased FBB and leakage is reduced using reverse biased RBB.

Fig. 11 shows a row decoder 1100. The row decoder 1100 is identical in structure to the row decoder 900 except that all transistors are of the type FDSOI NMOS 560 with the n-well biased to P1_ NW. The well bias level of P1_ NW is such that speed performance is achieved using forward biased FBB and leakage is reduced using reverse biased RBB.

Fig. 12 shows a row decoder 1200. The row decoder 1200 is identical in structure to the row decoder 900, except that: the NAND gate 951 comprises a transistor of type FDSOI NMOS 550 with the P-well biased to P2_ PW; inverter 952 comprises a transistor of type FDSOI NMOS 560 with the n-well biased to P1_ NW; PMOS transistors 953, 956, 958, and 961 are transistors of type FDSOI PMOS 510 with the P-well biased to P1_ NW; PMOS transistors 954 and 959 are transistors of type FDSOI PMOS 520 with the P-well biased to P2_ PW; NMOS transistors 955 and 960 are transistors of type FDSOI NMOS 510 with the n-well biased to P2_ PW; and NMOS transistors 957 and 962 are transistors of type FDSOI NMOS 560 with the n-well biased to P1_ NW. The well bias levels of P2_ PW/P1_ NW are such that speed performance is achieved using forward biased FBBs and leakage is reduced using reverse biased RBBs.

Fig. 13 shows an erase gate decoder 1300. The FDSOI transistors are not used in the erase gate decoder 1300 in this example, but are of bulk CMOS type. The HV PMOS 1301 controls the current from the HV power supply VEGSUP, and the HV PMOS 1302 is used for address decoding. The HV NMOS 1303 functions as a pull-down device that pulls EG 1305 to a LOW level, or as a transfer transistor that transfers a BIAS level EG _ LOW _ BIAS 1304 into the EG terminal.

Fig. 14 shows a source line decoder 1400. The FDSOI transistors are not used in the source line decoder 1400 in this example, but are of a bulk CMOS type. NMOS 1401 is used to transmit SL power VSLSUP, NMOS 1402 is used to measure (monitor) the voltage on SL 1405, NMOS 1403 is used to transmit LOW BIAS level SLRD _ LOW _ BIAS at read or standby, and NMOS 1404 is used to transmit LOW BIAS level SLP _ LOW _ BIAS at programming.

Fig. 15 shows a high voltage circuit selector 1500 that, once enabled, outputs a positive high voltage level on ENHV and/or a negative high voltage level on ENHVNEG. FDSOI transistors are not used in the high voltage logic selector 1500 in this example.

Fig. 16 shows a coupled gate decoder 1600. FDSOI transistors are not used in the coupled gate decoder 16001400 in this example, but are of the bulk CMOS type. HV PMOS 1401 is used to transmit CG power, HV PMOS 1402 is address decoding, PMOS 1403 is used to control current from CG read power VCGRSUP, and HV PMOS 1404 is used to transmit CG read power. PMOS 1405 is used to isolate negative voltage levels. NMOS 1407 is used for address decoding, NMOS 1408 and 1409 are used for negative voltage isolation, and NMOS 1410 is used for transferring BIAS level CG _ LOW _ BIAS into CG 1406. NMOS 1411 is used to deliver the negative voltage supply VHVNEG and NMOS 1412 is used as the negative cascode arrangement.

Fig. 17 shows low voltage sector enable latch logic 1700. The low voltage logic 1700 includes latched inverters 1701 and 1702 and NMOS transistors 1703 (wordline enable), 1704 (sector enable), and 1705 (1701/1702 for reset latch), all constructed of transistors of the type that utilize a p-well. Alternatively, the inverter 1701 may be constructed of a transistor using an n-well.

Fig. 18 shows a sensing system 1800, which is similar to block 601/602/603/604 of die 600 of fig. 6. The sensing system 1800 includes sense amplifiers 1801, 1802, 1803, and 1804. Embodiments of the sense amplifiers 1801, 1802, 1803, and 1804 are shown in fig. 19-22. Reference sector 1810 is used to generate a reference bias from a reference memory cell for sensing. Two inputs of the sense amplifier are coupled to two bit lines of two array planes, e.g., sense amplifier 1801 is coupled to top array plane 1820 and bottom array plane 1821. One array plane provides selected bitlines (thus providing selected memory cells with one wordline enabled) and the other array plane provides unselected bitlines for sensing (for which all wordlines are disabled) for symmetric bitline sensing.

Fig. 19 shows a sense amplifier 1900. Sense amplifier 1900 includes PMOS transistors 1901, 1906, 1907, and 1903 (type FDSOI PMOS 520 with p-well coupled to ground potential), PMOS transistors 1905,1908,1909, and 1912 (with n-well coupled to V-well)_biasType FDSOI PMOS 510), NMOS transistors 1902, 1904, 1910, 1911, 1913, and 1914 (type FDSOI NMOS 560 with the n-well coupled to ground potential), and NMOS transistor 1915 (type FDSOI NMOS 550 with the p-well coupled to ground potential). PMOS 1901 and NMOS 1902 (and PMOS 1903 and NMOS 1904) are the first (readout) stage of the sense amplifier. PMOS 1901 mirrors a self-reference current Iref (such as from reference cells or resistors in reference sector 1810 of sensing system 1800). The NMOS 1902 is coupled to a cell current Icell through a bit line of the selected memory cell. NMOSThe drain of 1902 is a sense output node 1999 which is equal to the difference between Iref and Icell multiplied by the output impedance at node 1999, i.e., Vsensed = Ro (Icell-Iref). The drain of NMOS 1904 is the reference node 1998. PMOS 1903 is in the disabled state with Ileakpmos (replicating the off-state leakage of PMOS 1901), NMOS 1904 is coupled to the cell current leakage Icellleak through the unselected bit lines of the memory cell (the selected bit lines with all word lines disabled). The drain of NMOS 1904 is the sense output node 1999, which is equal to the difference between Ileakpmos and icelleak multiplied by the output impedance at node 1998, i.e., Vrefsen = Ro (icelleak-Ileakpmos). The sense node 1999 and the reference node 1998 are precharged to reference voltage levels 1920 and 1921, respectively, at the start of sensing. Transistors 1905-1915 are the second (comparison) stage of the sense amplifier. It is a dynamically latched differential amplifier with transistors NMOS 1913 and 1914 as the input pair, and sense output node 1999 and reference node 1998 as inputs. Transistors 1906, 1907, 1910, and 1911 are latched inverters with outputs ON and OP as the full voltage level (Vdd/gnd) sense output after sensing the difference between sense output node 1999 and reference node 1998. The PMOS transistor 1905,1908,1909,1912 is used to precharge the node of the latched inverter to a high power supply level. NMOS 1913 and 1914 are a footed input pair (meaning NMOS transistors connected in series to the latched inverter). NMOS 1915 is an enabled bias transistor for the input pair.

Fig. 20 shows a sense amplifier 2000. Sense amplifier 2000 is structurally identical to sense amplifier 1900 except that the n-well of NMOS transistor 1913 is coupled to variable voltage source NL5_ NWB, and the n-well of NMOS transistor 1914 is coupled to variable voltage source NL5_ NWB. A variable voltage source is used to dynamically bias the well to optimize speed when active (forward body bias) and reduce leakage when standby (reverse body bias). It can also be used to cancel the threshold voltage offset of the sense amplifier.

Fig. 21 shows a sense amplifier 2100. Sense amplifier 2100 is identical in structure to sense amplifier 1900 except that the p-wells of PMOS transistors 1901, 1903, 1906, and 1907 are coupled to a variable voltage source PL1_ PW, and the n-wells of NMOS transistors 1902, 1904, 1910, 1911, 1913, and 1914 are coupled to a variable voltage source NL1_ NW. A variable voltage source is used to optimize speed (forward biasing the well) when active and reduce leakage (reverse biasing the well) when standby.

Fig. 22 shows a sense amplifier 2200 with FDSOI and bulk CMOS mixed region regions. Sense amplifier 2200 is identical in structure to sense amplifier 1900 except that the p-wells of PMOS transistors 1906 and 1907 are coupled to variable voltage source PL1_ PW, and the n-wells of NMOS transistors 1910 and 1912 are coupled to variable voltage source NL1_ NW, and PMOS transistors 2201 and 2202 and NMOS transistors 2202 and 2204 are bulk CMOS transistors. PMOS 2201 and NMOS 2202 and PMOS 2203 and NMOS 2204 are the bulk cmos sense stages of the amplifier. The sense stage is coupled to a high power supply level (due to bulk CMOS transistors), e.g., 1.8v, instead of a logic power supply level, e.g., vdd1.2v, in order to achieve a wide sensing range.

Fig. 23 shows a column decoder 2300. The column decoder 2300 includes NMOS transistors 2301, 2303, 2305, 2307, and 2309 (type FDSOI NMOS 560 for N-well coupling to N1_ NW) for speed enhancement in column selection, and NMOS transistors 2302, 2304, 2306, 2308, and 2310 (type FDSOI NMOS 550 for p-well coupling to N1_ PW) for leakage reduction in column deselection.