NONVOLATILE SEMICONDUCTOR MEMORY DEVICE AND METHOD OF OPERATING THE SAME

Abstract

A multi-level programmable nonvolatile semiconductor memory device comprises, a charge accumulation layer, a control gate which bias a potential to the charge accumulation layer, wherein the potential of the charge accumulation layer is controlled discretely according to the number of electrons accumulated in the charge accumulation layer.

Description

BACKGROUND OF THE INVENTION

1. Field of the Invention

This invention relates to an electrically rewritable and nonvolatile semiconductor memory device and the method of operating the same.

It is further related to a nonvolatile semiconductor memory device, such as NAND cell type, NOR cell type, DINOR cell type, and AND cell type EEPROM.

2. Description of the Related Art

An electrically erasable and programmable EEPROM, such as a NAND cell-type EEPROM that configures a NAND cell with a plurality of serially connected memory cells, has been known as one of semiconductor memory devices. A memory cell in the NAND cell-type EEPROM has a FETMOS structure with a charge storage layer (floating gate) and a control gate stacked on a semiconductor substrate. The memory cell stores data “0” or “1” depending on the amount of charge accumulated in the floating gate.

For market expansion of a semiconductor memory device, high integration of a semiconductor memory device and the reduction in cost are required.

Data programming of an NAND type flash memory is performed sequentially from the memory cell which is most separated from a bit line

Under the data programming, 0V (“0” data programming) or supply voltage Vcc (“1” data programming) is applied to a bit line according to program data.

Vcc is given to the selected bit-line side selection gate line.

When a bit line is 0V,at the connected selected NAND cell, the potential is fixed to 0V through a selection gate transistor.

When a bit line is biased at Vcc, selected NAND cell is charged to Vcc-Vtsg through a selection gate transistor. Here, Vtsg is the threshold voltage of a selection gate transistor.

Then, the control gate line of the selected NAND cell is set to Vpp (about 20V,program voltage) from 0V.

Further, the control gate line of non-selected memory cell in selected NAND cell is set to Vmg (about 10V: middle voltage) from 0V.

Here, in selected NAND cell when the bit line was 0V, channel part of NAND cell is fixed to 0V.

Potential difference (about 20V) occurs between the gate (Vpp potential) and channel part (0V) of selected NAND cell, and an electron injection arises from a channel part in floating gate.

Then the threshold of the selected memory cell shifts to positive direction. This state is data “0.”

On the other hand, in selected NAND cell, when the bit line is biased at Vcc, the channel part in NAND cell is floating state.

For this reason, with a voltage rise from 0 V to Vpp or Vmg of the control gate line in selected NAND cell and the control gate line under the influence of capacity coupling between channel parts, channel potential goes up from Vcc-Vtsg to Vmch (about 8V), while the potential of a channel part had maintained the floating state.

At this time, the potential difference between gate (Vpp potential) of selected NAND cell and channel parts (Vmch) are comparatively as small as about 12V. So an electron injection does not happen.

Therefore, the Vth (threshold voltage) of selected memory cell does not change but is maintained at the negative state. This state is data “1.”

Data erasure of an NAND type flash memory is simultaneously performed to all the memory cells in the selected NAND cell block.

First, all control gate in the selected NAND cell block is set to 0V.

The control gate, all the selection gates in non-selected NAND cell block, bit line, and source line, set floating. High voltage of about 20V is biased to p type well. Thereby, in all the memory cells under selected NAND cell block, the electron of floating gate is emitted to p type well, and threshold voltage is shifted in the negative direction. Thus, in an NAND cell type flash memory, data erasure will be performed per block.

SUMMARY OF THE INVENTION

A nonvolatile semiconductor memory device based on first aspect of the invention comprises,a charge accumulation layer; a control gate which bias a potential to the charge accumulation layer, wherein the potential of the charge accumulation layer is controlled discretely according to the number of electrons accumulated in the charge accumulation layer.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1A is a figure showing the structure of a memory cell which were used for the simulation.

FIG. 1B is a figure showing the equal circuit of a memory cell which were used for the simulation.

FIG. 2 is a figure showing the result of the simulation using the memory cell shown FIG. 1A and FIG. 1B.

FIG. 3 is a plane view showing change of the threshold Vth of the conventional memory cell.

FIG. 4A shows the image of electrons pouring to floating gate when a design rule is loose

FIG. 4B shows the image of electrons pouring to floating gate when a design rule is 1×nm.

FIG. 5 is a figure showing change of the threshold Vth of the conventional memory cell.

FIG. 6 is a figure showing change of the threshold Vth of the memory cell in nonvolatile semiconductor memory device of this embodiment.

FIG. 7 is a figure showing the threshold voltage and its distribution of the memory cell concerning an embodiment,

FIG. 8 is a figure showing program operation of the data concerning to an embodiment.

FIG. 9 is a figure showing program operation of the data concerning to an embodiment.

FIG. 10 is a figure showing program operation of the data concerning to an embodiment.

DETAILED DESCRIPTION OF THE INVENTION

Hereafter, the embodiment of this invention is explained referring to drawings.

Introduction to the Invention

In NAND cell type EEPROM, program voltage Vpp at the time of program is set constant.

In early stages of program, when the quantity of the electrons in a charge-storage layer are comparatively little, change of the threshold voltage of a memory cell is quick.

In later stages of program, when the quantity of the electrons in a charge-storage layer are comparatively much, change of the threshold voltage of a memory cell is slow.

Moreover, in early stage of program, the electric field impressed to the insulated film, with which a tunnel current flows, is strong. And the electric field is weak in later stage of program.

For this reason, if the program voltage Vpp is raised in order to make program speed quick, the maximum threshold voltage after program will become high.

Moreover, the distribution width of threshold voltage after program becomes large.

And, the electric field impressed to the insulated film with which a tunnel current flows, becomes strong. This may result poor reliability.

On the contrary, program speed will become slow, if Vpp is lowered in order to obtain narrow threshold voltage distribution width after program.

In other words, a program voltage margin is narrow.

It is desirable to make form of a program pulse into “an ideal trapezoidal-shape wave” at the time of program. However, it is difficult to carry the program pulse generator which generates “an ideal trapezoidal-shape wave” on the same chip as a memory array on account of a program pulse generator.

Moreover, it is possible by making the interval of the program pulse of a stairs-like wave small to bring the form of a program pulse close to “an ideal trapezoidal-shape wave.”

As a result, the number of times of verification will increase and program and a blanking time will increase as the result.

The inventors disclosed the program method (henceforth “stepping-up method”) technology which can shorten program time. Which use the series of a program pulse, the potential of a program pulse gradually by step-up width delta Vpp. As a result, by performing program operation, program speed was improved.

By using the stepping-up method, in the series of the following program pulse after verification operation, it is possible to prevent a steep electric field impression to a memory cell, which result to prevent to cause a degradation, a breakdown, of a tunnel oxide film. And the reliability of nonvolatile semiconductor memory device can be improved.

On the other hand, the inventors examined what kind of influence appears in program operation as the design rule became small.

The semiconductor memory device and method of operating the same regarding to the first embodiment of this invention is explained using FIG. 1A to FIG. 10. The first embodiment discloses a nonvolatile semiconductor memory device and method of operating the same.

The structure of the memory cell used for examination is shown in FIG. 1A, and the equivalent circuit figure of the structure is shown in FIG. 1B.

As for the memory cell shown in FIG. 1A, laminating formation of floating gate (FG, charge-storage layer) 3 and control gate (CG) 1 are formed on P-type silicon substrate 7/n type well 6/p type well 5.

P type well 5 and floating gate 3 are insulated by a tunnel oxide film 4. Floating gate 3 and control gate 1 are insulated by an insulator layer (stacked layer of an silicon oxide film/silicon nitride film/silicon oxide film) 2.

N type diffusion layer 8 forms a source and a drain of a memory cell.

As shown in FIG. 1B, the capacity between floating gate 3 and control gate 1 is named as Cono, and the capacity between floating gate 3 and p type well 5 is named as Cox.

In addition, since it is negligibly small compared with Cono and Cox, the capacity between floating gate 3 and n type diffusion layer 8 is not taken into consideration here. A memory cell memorizes data with the electric charge stored in floating gate 3.

And the threshold voltage (Vth) of a memory cell is decided by the amount of electric charges stored in floating gate 3. The amount of electric charges which frequents floating gate at the time of data program/erasure is controlled by FN tunnel-current (Fowler-Nordheim current) Itunnel which passes through the tunnel oxide film 4.

If potential of control gate 1 is sufficiently high comparing to the potential of p type well 5 and n type diffusion layer 8, electrons will be poured into floating gate 3 through the tunnel oxide film 4, and the threshold voltage Vth of a memory cell will become high.

On the other hand, if potential of p type well 5 and n type diffusion layer 8 is made high comparing to the potential of control gate 1, electrons will be emitted from floating gate 3 through the tunnel oxide film 4, and the threshold Vth of a memory cell will become low.

Here, the result of the simulation of tunnel-current (Itunnel), the program time (Tprog), etc., when changing a design rule (generation) from 130 nm to 1×nm are shown in FIG. 2 using the memory cell shown in FIG. 1.

In addition, “4×”, “3×”, “2×”, and “1×” show the design rule of the level of 40 nm, the level of 30 nm, the level of 20 nm, and the level of 10 nm, respectively.

The capacitance Cono (where coupling ratio is 0.5) of the insulator layer between control gate and floating gate, the channel width/channel lengths of a memory cell (Cell W/L), the area (Cell Area Size) of the active layer of a memory cell, are shown in FIG. 2.

Moreover, in FIG. 2, program time (Tprog) is time taken to set change deltaVth (Vth shift) of a threshold voltage to 1V, when the program voltage VPGM is 20V.

Itunnel is a tunnel current in case the program voltage VPGM is 20V.

In FIG. 2, the number of electrons (Number of Electron) shows the number of the electrons poured into floating gate which needs a threshold voltage to change 1V.

Threshold voltage change/electron number (deltaVth/electron) shows change of the threshold voltage by one electron poured into floating gate.

In addition, in the upper part of FIG. 2, the shrink rate of between each design rule is shown, and the shrink rate was calculated as 0.7 after 55 nm generation.

According to the simulation result shown in FIG. 2, the number of the electrons, poured into floating gate which needs threshold voltage change of 1V decreases as a design rule becomes small.

It decreases from 342, to 237 pieces, 132 pieces, . . . , as the number of the electrons poured into floating gate which needs a threshold voltage change of 1V by design rule from 130 nm to 90 nm, 70 nm, . . . , respectively.

And when in design rule 1×nm, the number of the electrons poured into floating gate which needs a threshold voltage change 1V becomes only nine pieces.

By calculating reciprocal of electron number poured into floating gate which is necessary to change threshold voltage of 1V, on the other hand, change of deltaVth/electorn of the threshold by one electron being poured into floating gate can be estimated.

Threshold voltage change by one electron poured into floating gate is estimated from 2.9 mV, to 4.2 mV, 7.5 mV, . . . , according to the change of design rule from 130 nm, to 90 nm, 70 nm, . . . . And by 1×nm, threshold voltage change will be calculated to about 100 mV per electron.

In the case of design rule 70 nm, deltaVth/electron is 7.5 mV/electron. And change of the threshold voltage of delta Vth, which is equivalent to delta Vpp(Vpgm), as shown in FIG. 3, can be find from the above simulation result by setting step-up width deltaVpp as 7.5 mV.

In the case of design rule 1×nm, deltaVth/electron is about 100 mV, as compared with the case where a design rule is loose, change of the threshold per electron is large, and in using the stepping-up method, it is necessary to set the deltaVPGM>100 mV.

In other words, in design rule 1×nm, big change of threshold voltage can be produced by pouring one electron into floating gate.

In the case of design rule 1×nm, deltaVth/electron is about 100 mV, as compared with the case where a design rule is loose, change of the threshold per electron is large, and in using the stepping-up method, it is necessary to set the deltaVPGM>100 mV.

In other words, in design rule 1×nm, big change of threshold voltage can be produced by pouring one electron into floating gate.

When a design rule is loose, as this shows in FIG. 4A, threshold voltage change deltaVth=100 mV is produced by pouring a lot of electrons (being 13 pieces in the case of for example, at design rule 70 nm) into floating gate, although, when a design rule is 1×nm as shown in FIG. 5(b), threshold voltage change deltaVth=100 mV can be produced by pouring one electron.

In the simulation result shown in FIG. 2, in the case of design rule 70 nm, in order to fluctuate 0.1V threshold, it is necessary to pour 13 electrons into floating gate.

By one program pulse (VPGM), the threshold voltage Vth of a memory cell changes 0.1V, as it is shown in (a), (b), and (c) of FIG. 5, when program operation is performed by the stepping-up method in the case of design rule 70 nm, and the step-up width (deltaVpp) of a program pulse (VPGM) is 0.1V.

Here, when 13 electrons should be poured into floating gate by one program pulse, for example, if 12 electrons, or 14 electrons, might be poured into floating gate, ±7.5 mV variation will arise in the threshold Vth of a memory cell, and the threshold distribution of a memory cell will spread. If the program pulse VPGM is impressed one after another, by adding step-up width deltaVpp, the threshold distribution of a memory cell will spread having 7.5 mV variation.

In the simulation result shown in FIG. 2, when design rule is 1×nm, in order to fluctuate 0.1V threshold voltage, one electron should be poured into floating gate.

By one program pulse (VPGM), the threshold voltage Vth of a memory cell changes 0.1V, and as i is shown in FIG. 6(a), (b), and (c), when program operation is performed by the stepping-up method, and the step-up width (deltaVpp) of a program pulse (VPGM) is 0.1V.

Since there are few electrons poured in floating gate, where one electron should be poured into floating gate by one program pulse, for example, 0 or 2 electrons has been poured into floating gate, variation does not occur in the threshold Vth of a memory cell, and the threshold distribution of a memory cell does not spread.

And even if it impresses the program pulse VPGM one after another, adding step-up width deltaVpp, a threshold distribution spreads and the threshold Vth of a memory cell does not.

As a result of examining in detail the simulation result shown in FIG. 2 as above, by controlling the number of the electrons poured in to floating gate, the inventors thought that threshold voltage change of the request in a memory cell could be produced.

In this embodiment, using design rule of 1×nm, programming by the stepping-up method is performed with step-up width deltaVpp=0.1V, and nonvolatile semiconductor memory device of this invention which can memorize the information on 32 values to one memory cell is explained.

FIG. 7 is the figure showing the threshold voltage Vth and threshold voltage distribution of a memory cell in nonvolatile semiconductor memory device of this embodiment. In order to make one memory cell memorize the 32 values in this embodiment, it is necessary to change a threshold voltage to 3.1V at 0V to 0.1V interval, as shown in FIG. 7 (c).

In addition, in order to program four values or 8 values, FIG. 7(a) and (b) shows an example. One memory cell which vary the threshold voltage 0V to 0.3V at 0.1V interval in order to make one memory cell memorize the information on four values or 8 values, the example to which the threshold was changed to 0.7V at 0V to 0.1V interval is also shown. If the simulation result shown in FIG. 2 mentioned above is taken into consideration, in order to fluctuate a threshold voltage of 0.1V, what is necessary will be just to pour one electron into floating gate here.

That means that, in this embodiment, in order to make one memory cell memorize the information on 32 values and to control the threshold of a memory cell by 0V to 0.1V interval to 3.1V, pouring of one electron to floating gate should be controlled.

In other words, the program pulse VPGM for pouring one electron into floating gate of a memory cell will be impressed.

If the simulation result of the above-mentioned FIG. 2 is taken into consideration, in order to pour one electron into floating gate of a memory cell, it becomes controllable by impressing the program pulse VPGM by step-up width deltaVpp as 0.1V.

FIG. 8 shows how to write data in the memory cells M0-M31 connected to the word line WL0 of memory block of nonvolatile semiconductor memory device in this embodiment.

Upper part of FIG. 8 shows the memory cells M0-M31 with control gate (CG) connected to the word line WL0, and by which the end of a source or a drain was connected to bit lines BL0-BL31, respectively.

Middle part of FIG. 8 shows the timing chart which impresses the program voltage VPGM to memory cells M0, M1-M31, respectively, in order to write “1”, “2”, . . . “31”, making it step up by deltaVpp=0.1V.

It is referred to as program voltage VPGM of 20.0V, and is made to step up by deltaVpp of 0.1V, and program voltage is made to rise gradually here at the beginning of program until it is set to VPGM=3.1V.

That is, since the upper limit Vthu of change of a threshold is 3.1V , deltaVth, calculated as 3.1/(32−1)=0.1V, corresponding to step-up width deltaVpp=0.1V will be impressed.

If n data should be memorized in each memory cell, and the maximum value of a threshold is to Vthu, it will be impressed two or more threshold change pulses which have high potential gradually with the aperture of the voltage which produces change of the threshold of (Vthu/(n−1)) to said control gate of said memory cell.

Bottom part of FIG. 8 shows the timing chart of the signal voltage impressed to each bit lines BL0-BL31 when programming “1”, “2”, and . . . “31” to these memory cells M0, M1-M31, respectively.

0V are impressed to the bit line which performs program, and Vdd (program prohibition voltage: here about 4V) is impressed to the bit line to which program is inhibited. While the program voltage VPGM rises 0.1 v at a time as shown in FIG. 8, it is impressed by control gate of each memory cells M0-M31 through a word line WL0.

And to the memory cell by which program voltage (0V) is impressed to a bit line, one electron is poured into floating gate by impression of each program voltage VPGM, and a threshold is changed by it.

Moreover, electrons are not poured into the memory cell by which program prohibition voltage (Vdd) is impressed to a bit line by impression of each program voltage VPGM, and a threshold is not changed.

Repeating the sequence which is shown in this embodiments in bottom part of FIG. 8, at memory cells M0, M1, and M2 . . . M30, M31, floating gate of M30 and M31 an electron, —one piece, two pieces, three pieces, and . . . 31 pieces, will be poured in and 30 thresholds of each memory cell can be controlled as shown in FIG. 7.

This program operation can be performed to all word lines. In FIG. 8, each memory cells M0, M1-M31, will be “1”, “2”, . . . “31” programmed. Although shown taking the case of the sequence concerning program operation when writing in the value of “31”, it cannot be overemphasized by making a timing change that the data written in each memory cells M0-M31 can be changed easily.

Thus, in nonvolatile semiconductor memory device of this embodiment, by controlling the electrons poured into floating gate of a memory cell per one piece, it becomes controllable, and the data storage of 32 values becomes possible at each memory cell.

In addition, in the embodiments above, each memory cell which memorize the data of 32 values was explained. What is necessary is to control change of a threshold, according to the amount of data, when it is not necessarily limited to this and makes the data of multiple values, such as binary, four values, and an octal.

The timing chart of the signal voltage impressed to each bit lines BL0-BL7 is shown in FIG. 9 about the case where the multiple-value data of an octal is stored in FIG. 10 for the timing chart of the signal voltage impressed to each bit lines BL0-BL3 in the case of making the data of the multiple value of four values memorize again.

Here, each of the parameters disclosed in FIG. 2 may differs about 20% to 40%, because the size variation of the products.

For example, delta Vth/electron for 2×nm generation may differs 40 mV-59 mV around the center value of 48.5. And the cell size of the 1×nm generation may differs about 0.00046-0.00110 square micrometer around the center value of 0.000784 square micrometer.

Claims

1. A multi-level programmable nonvolatile semiconductor memory device comprising: a charge accumulation layer; a control gate which bias a potential to the charge accumulation layer, wherein the potential of the charge accumulation layer is controlled discretely according to the number of electrons accumulated in the charge accumulation layer.

2. The multi-level programmable nonvolatile semiconductor memory device according to claim 1, wherein the potential of the charge accumulation layer vary discretely by the interval between 85 mV and 125 mV

3. The multi-level programmable nonvolatile semiconductor memory device according to claim 2, wherein the discrete change of the potential of the charge accumulation layer depends on a electron poured in the charge accumulation layer.

4. The multi-level programmable nonvolatile semiconductor memory device according to claim 1, wherein the potential of the charge accumulation layer vary discretely by the interval between 40 mV and 58 mV.

5. The multi-level programmable nonvolatile semiconductor memory device according to claim 1, wherein the multi-level programmable nonvolatile semiconductor memory device comprises the NAND type memory cell array.

6. The multi-level programmable nonvolatile semiconductor memory device according to claim 4, wherein the discrete change of the potential of the charge accumulation layer depends on a electron poured in the charge accumulation layer.

7. A method of operating a multi-level programmable nonvolatile semiconductor memory device comprising: bias a charge accumulation layer at first voltage; bias a charge accumulation layer at second voltage bigger than the first voltage, wherein the difference of the first voltage and the second voltage is determined by a number of electrons accumulated in the charge accumulation layer.

8. The method of operating a multi-level programmable nonvolatile semiconductor memory device according to claim 7, wherein the difference of the first voltage and the second voltage is between 85 mV and 125 mV.

9. The method of operating a multi-level programmable nonvolatile semiconductor memory device according to claim 7, wherein t the difference of the first voltage and the second voltage is between 40 mV and 58 mV.

10. The method of operating a multi-level programmable nonvolatile semiconductor memory device according to claim 7, wherein the multi-level programmable nonvolatile semiconductor memory device comprises the NAND type memory cell array.