Search Images Maps Play YouTube News Gmail Drive More »
Sign in
Screen reader users: click this link for accessible mode. Accessible mode has the same essential features but works better with your reader.

Patents

  1. Advanced Patent Search
Publication numberUS20020105834 A1
Publication typeApplication
Application numberUS 10/074,989
Publication dateAug 8, 2002
Filing dateFeb 13, 2002
Priority dateFeb 8, 2001
Also published asUS6504759, US6639841, US20020105828, US20020105830, US20020126538
Publication number074989, 10074989, US 2002/0105834 A1, US 2002/105834 A1, US 20020105834 A1, US 20020105834A1, US 2002105834 A1, US 2002105834A1, US-A1-20020105834, US-A1-2002105834, US2002/0105834A1, US2002/105834A1, US20020105834 A1, US20020105834A1, US2002105834 A1, US2002105834A1
InventorsChin-Yang Chen
Original AssigneeUnited Microelectronics Corp.
Export CitationBiBTeX, EndNote, RefMan
External Links: USPTO, USPTO Assignment, Espacenet
Doublel-bit non-voltatile memory unit and corresponding data read/write method
US 20020105834 A1
Abstract
A double-bit non-volatile memory cell structure and a method of programming the memory cell. The memory cell includes a pair of stacked gates above a substrate, a doped region in the substrate between the stacked gate pair and a source/drain region in the substrate on each side of the stacked gate pair. The source/drain regions and the doped region are doped identically. To write data into the memory cell, the channel underneath both stacked gates is opened simultaneously. Data is written into the desired floating gate by controlling current flow direction. To read data from a first floating gate of the memory cell, a read bias voltage is applied to the first control gate above the first floating gate. In the meantime, a transfer voltage is applied to the second control gate. The presence or the absence of a conductive channel between the source/drain regions indicates whether data has been written into the first floating gate or not. The read bias voltage is greater than the threshold voltage of the first/the second floating gate in the erased state but smaller than the threshold voltage in the written state. The transfer voltage is greater than the threshold voltage in the written state.
Images(3)
Previous page
Next page
Claims(16)
What is claimed is:
1. A double-bit non-volatile memory cell structure, comprising:
a substrate;
a pair of stacked gates above the substrate, wherein each stacked gate includes, from bottom to top, a tunnel layer, a floating gate, an inter-gate dielectric layer and a control gate;
a doped region, wherein the doped region is located in the substrate between the two stacked gates; and
a pair of source/drain regions, wherein the source/drain regions are in the substrate on each side of the stacked gate pair, and the two source/drain regions and the doped region are identically doped.
2. The structure of claim 1, wherein the double-bit non-volatile memory cell is applied to form a flash memory.
3. The structure of claim 1, wherein the tunnel layer includes a tunnel oxide layer.
4. The structure of claim 1, wherein material forming the floating gate includes polysilicon.
5. The structure of claim 1, wherein the inter-gate dielectric layer includes an oxide/nitride/oxide (ONO) composite layer.
6. The structure of claim 1, wherein the material forming the control gate includes polysilicon.
7. The structure of claim 1, wherein the width of the doped region is smaller than the width of any one of the pair of source/drain regions.
8. The structure of claim 1, wherein the doped region has a width identical to any one of the pair of source/drain regions.
9. The structure of claim 1, wherein the doped region and the two source/drain regions are all n-doped.
10. A method of programming a double-bit non-volatile memory cell, wherein the double-bit non-volatile memory cell comprises:
a substrate;
a first stacked gate and a second stacked gate above the substrate, wherein the first stacked gate includes a first floating gate and a first control gate and the second stacked gate includes a second floating gate and a second control gate;
a doped region located in the substrate between the first stacked gate and the second stacked gate; and
a pair of source/drain regions located in the substrate on each side of the stacked gate pair, wherein the two source/drain regions and the doped region are n-doped; and
the steps of writing data into the first floating gate includes:
applying a first voltage at the first control gate and applying a second voltage at the second control gate so that the channel underneath the first floating gate and the channel underneath the second floating gate are both opened; and
applying a different bias voltage at the two source/drain regions so that electrons in the channel underneath the second floating gate flow towards the channel underneath the first floating gate, and the electrons pick up sufficient energy from the channel underneath the first floating gate to produce hot electrons that can easily move into the first floating gate.
11. The programming method of claim 10, wherein after the step of writing data into the first floating gate, further includes:
applying a third voltage at the first control gate and applying a fourth voltage at the second control gate so that the channel underneath the first floating gate and the channel underneath the second floating gate are both opened; and
applying a different bias voltage to the two source/drain regions so that electrons in the channel underneath the second floating gate flow and acquire enough energy to generate hot electrons capable of moving into the second floating gate.
12. The programming method of claim 10, wherein the bias voltage applied to the source/drain region on one side of the second stacked gate includes a ground voltage.
13. A method of programming a double-bit non-volatile memory cell, wherein the double-bit non-volatile memory cell comprises:
a substrate;
a first stacked gate and a second stacked gate above the substrate, wherein the first stacked gate includes a first floating gate and a first control gate and the second stacked gate includes a second floating gate and a second control gate;
a doped region located in the substrate between the first stacked gate and the second stacked gate; and
a pair of source/drain regions located in the substrate on each side of the stacked gate pair, wherein the two source/drain regions and the doped region are n-doped; and
the steps of writing data into the first floating gate includes:
applying a higher bias voltage at the first control gate and applying a lower bias voltage at the source/drain region on one side of the first stacked gate so that electrons can tunnel into the first floating gate from the source/drain region on one side of the first stacked gate.
14. The programming method of claim 13, wherein after the step of writing data into the first floating gate, further includes:
applying a higher bias voltage at the second control gate;
applying a lower bias voltage at the source/drain region on one side of the second stacked gate so that electrons can tunnel into the second floating gate from the source/drain region on one side of the second stacked gate.
15. The programming method of claim 13, wherein the bias voltage applied to the source/drain region on one side of the first stacked gate includes a ground voltage.
16. A method of reading data from a double-bit non-volatile memory cell, wherein the double-bit non-volatile memory cell comprises of:
a substrate;
a first stacked gate and a second stacked gate on the substrate, wherein the first stacked gate includes a first floating gate and a first control gate and the second stacked gate includes a second floating gate and a second control gate, the channel underneath the first/the second floating gate is at a first threshold voltage when the first/the second floating gate is in the erased state, and the channel underneath the first/the second floating gate is at a second threshold voltage greater than the first threshold voltage when the first/second floating gate is in the written state;
a doped region in the substrate located between the two stacked gates; and
a pair of source/drain regions located on each side of the stacked gate pair, wherein the two source/drain regions and the doped region are n-doped; and
the steps of reading stored data from the first floating gate includes:
applying a read bias voltage at the first control gate, wherein the read bias voltage is greater than the first threshold voltage but smaller than the second threshold voltage;
applying a transfer bias voltage at the second control gate, wherein the transfer bias voltage is greater than the second threshold voltage so that the channel underneath the second floating gate is opened; and
determining the state of the first floating gate by checking the presence or absence of a conductive channel between the two source/drain regions, wherein conduction between the two source/drain regions indicates an unwritten state for the first floating gate while non-conduction indicates a written state.
Description
CROSS-REFERENCE TO RELATED APPLICATION

[0001] This application claims the priority benefit of Taiwan application serial no. 90102758, filed Feb. 8, 2001.

BACKGROUND OF THE INVENTION

[0002] 1. Field of Invention

[0003] The present invention relates to a semiconductor device and its method of operation. More particularly, the present invention relates to a double-bit non-volatile memory (NVM) unit and a data read/write method for the double-bit NVM unit.

[0004] 2. Description of Related Art

[0005] Non-volatile memory (NVM) is a type of fast access, miniature, power-saving, vibration-resistant and permanent storage media. Thus, the applications of NVM are wide. A prominent type of NVM is the flash memory. One major aspect of the flash memory is the capacity for block-by-block data erasure so that time is saved.

[0006]FIG. 1 is a schematic cross-sectional view showing the structure of a conventional non-volatile memory cell. As shown in FIG. 1, the non-volatile memory cell includes a stacked gate structure 110 over a substrate 100. A source/drain region 120 is formed in the substrate 100 on each side of the stacked gate structure 110. The stacked gate structure 110 further includes, from bottom to top, a tunnel oxide layer 112, a floating gate 114, an inter-gate dielectric layer 116 and a control gate 118. During programming, electrons are injected into the floating gate 114. To erase data, a high negative bias voltage is applied to the control gate 118 and hence electrons are channeled away from the floating gate 114.

[0007] However, in order to remove all electrons from the floating gate 114, over-erase of the aforementioned non-volatile memory cell often occurs. In other words, too many electrons may be forced out of the floating gate 114 during erasure, resulting in the accumulation of some positive charges in the floating gate 114. In the presence of excess positive charges, an inversion of the channel underneath the floating gate 114 may occur and ultimately this may lead to a permanent opening of the channel and possible data read errors.

[0008] To resolve the issue, a split gate structure is developed. FIG. 2 is a schematic cross-sectional view of a non-volatile cell having a conventional split-gate structure. As shown in FIG. 2, a split-gate structure 210 is formed over a substrate 200. A source/drain region 220 is formed in the substrate 200 on each side of the split gate structure 210. The split-gate structure 210 includes, from bottom to top, a tunnel oxide layer 212, a floating gate 214, an inter-gate dielectric layer 216, a control gate 218 and a transfer gate 218 a. The transfer gate 218 a extends from the control gate 218 to the side of the floating gate 214. In addition, both the transfer gate 218 a and the floating gate 214, as well as the transfer gate 218 a and the substrate 200, are separated by the inter-gate dielectric layer 216. In this type of design, the channel under the transfer gate 218 a is opened up only when a voltage is applied to the control gate 218/transfer gate 218 a. Hence, even if the channel underneath the floating gate 214 is permanently opened due to over-erase, the two source/drain regions 220 of the memory cell are still in a non-conductive state, thereby preventing data read errors.

[0009] Although the split-gate structure 210 is able to prevent errors due to over-erase, the design has an adverse effect on miniaturization. This is because the transfer gate 218 a needs to occupy extra area. In addition, the combined width of the control gate 218 and the transfer gate 218 a inside the split-gate structure 210 is different from the floating gate 214. The floating gate 214 and the control gate 218/transfer gate 218 a must be patterned in two separate photolithographic processes. Consequently, alignment problems between the floating gate 214 and the control gate 218/transfer gate 218 a may occur, leading to possible overlapping area errors between the control gate 218/transfer gate 218 a and the floating gate 214. Ultimately, electrical performance of each memory cell may be different and control of the memory cells may be difficult.

SUMMARY OF THE INVENTION

[0010] Accordingly, one object of the present invention is to provide a double-bit non-volatile memory cell capable of preventing over-erase problems. The cell includes a substrate, a pair of stacked gates, a doped region, a source region and a drain region. Each stacked gate includes, from bottom to top, a tunnel oxide layer, a floating gate, an inter-gate dielectric layer and a control gate. The doped region is located in the substrate between the two stacked gates. The source/drain regions are formed in the substrate on each side of the stacked gate pair. The source/drain regions and the doped region are doped identically.

[0011] This invention also provides a method of programming the said double-bit non-volatile memory cell. To write data into the first floating gate of the first stacked gate, a bias voltage is applied to the first and the second control gate of the first and the second stacked gate respectively. The channel underneath the first and the second floating gate is opened. In the meantime, a different bias voltage is applied to the respective source/drain regions so that electrons flow from the channel underneath the second floating gate to the channel underneath the first floating gate. The electrons obtain sufficient energy and produce hot electrons that are injected into the first floating gate.

[0012] This invention also provides an alternative method of programming the said double-bit non-volatile memory cell. To write data into the first floating gate of the first stacked gate, a higher bias voltage is applied to the first control gate of the first stacked gate. In the meantime, a lower bias voltage is also applied to the source/drain region on one side of the first stacked gate so that electrons move into the first floating gate via the source/drain region.

[0013] This invention also provides an alternative method of reading data from the said double-bit non-volatile memory cell. In this memory cell, the threshold voltage (the voltage at the control gate when the channel of the stacked gate is opened) of the channel underneath any stacked gate in the erased state is referred to as a first threshold voltage. Threshold voltage of the channel underneath any stacked gate in the written state is referred to as a second threshold voltage. The second threshold voltage is higher than the first threshold voltage. To read data stored in the first floating gate of the first stacked gate, a read bias voltage is applied to the first control gate of the first stacked gate. The read bias voltage is higher than the first threshold voltage but lower than the second threshold voltage. In the meantime, a transfer bias voltage is applied to the second control gate of the second stacked gate. The transfer bias voltage is higher than the second threshold voltage so that the channel underneath the second floating gate is forced open. Whether the first floating gate is in the written state or not is determined by the channel between the respective source/drain regions. If the channel between the source/drain region is conductive, data has not been written into the first floating gate. On the contrary, if the channel is non-conductive, data has been written into the floating gate.

[0014] The double-bit non-volatile memory of this invention uses two stacked gates but only a pair of source/drain regions. Therefore, a conductive channel is formed joining the two source/drain regions only when the channel underneath both the first and the second floating gates is opened. Since the probability of having both floating gates over-erased is small, the chance of maintaining a permanent channel between the source/drain terminals is slim. Thus, errors in data determination are greatly reduced. Moreover, one of the stacked gates can be regarded as a transfer gate because the transfer gate actually prevents the other transfer gate from being over-erased.

[0015] In addition, the double-bit memory cell is capable of holding two binary bits with one stacked gate serving as the transfer gate of the other stacked gate. Unlike the cell of a conventional split-gate having an additional transfer gate on one side of the control gate, the area for holding each bit is smaller. Furthermore, unlike a conventional split-gate design that demands a patterning of the floating gate before patterning the control gate and the transfer gate together, one stacked gate serves as the transfer gate of the other stacked gate in this invention. Hence, the floating gate and the control gate may be produced by a self-alignment method so that any non-conformity between different electrical devices is minimized.

[0016] It is to be understood that both the foregoing general description and the following detailed description are exemplary, and are intended to provide further explanation of the invention as claimed.

BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The accompanying drawings are included to provide a further understanding of the invention, and are incorporated in and constitute a part of this specification. The drawings illustrate embodiments of the invention and, together with the description, serve to explain the principles of the invention. In the drawings,

[0018]FIG. 1 is a schematic cross-sectional view showing the structure of a conventional non-volatile memory cell;

[0019]FIG. 2 is a schematic cross-sectional view of a non-volatile cell having a conventional split-gate structure;

[0020]FIG. 3 is a schematic cross-sectional view of a double-bit non-volatile memory cell according one preferred embodiment of this invention;

[0021]FIG. 4 is a schematic cross-sectional view illustrating a first method for programming the double-bit non-volatile memory cell according to this invention;

[0022]FIG. 5 is a schematic cross-sectional view illustrating a second method for programming the double-bit non-volatile memory cell according to this invention; and

[0023]FIG. 6 is a schematic cross-sectional view illustrating a method for reading data from the double-bit non-volatile memory cell according to this invention.

DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0024] Reference will now be made in detail to the present preferred embodiments of the invention, examples of which are illustrated in the accompanying drawings. Wherever possible, the same reference numbers are used in the drawings and the description to refer to the same or like parts.

[0025] The following is a description of a double-bit non-volatile memory cell structure, two programming methods and a reading method according to this invention. The double-bit non-volatile memory cell can be applied to a flash memory, for example.

[0026]FIG. 3 is a schematic cross-sectional view of a double-bit non-volatile memory cell according to one preferred embodiment of this invention. As shown in FIG. 3, the memory cell includes a substrate 300, a pair of stacked gates 310 a and 310 b, a pair of source/drain regions 320 a and 320 b and a doped region 333. The stacked gate 310 a/b includes, from bottom to top, a tunnel oxide layer 312 a/b, a floating gate 314 a/b, an inter-gate dielectric layer 316 a/b and a control gate 318 a/b. The doped region 333 is located in the substrate 300 on each side of the stacked gate pair 310 a and 310 b. Both the source/drain region 320 a/b and the doped region 333 are similarly doped, for example, n-doped. In addition, both the floating gate 314 a/b and the control gate 318 a/b are made from a material such as polysilicon while the inter-gate dielectric layer 316 a/b is an oxide/nitride/oxide (ONO) composite layer, for example. Furthermore, the doped region 333 is used only for connecting the channels underneath the tunnel oxide layers 312 a and 312 b. Hence, width a of the doped region 333 can be smaller than width b of the source/drain region 320 a/b.

[0027] The following is a description of two methods for programming the said double-bit non-volatile memory cell of this invention when the source/drain region 320 a/b and the doped region 333 are n-doped.

[0028]FIG. 4 is a schematic cross-sectional view illustrating a first method for programming the double-bit non-volatile memory cell according to this invention. The first programming method is actually a channel hot electron (CHE) injection method. Bias voltages V1 and V2 are applied to the control gate 318 a and 318 b respectively so that the channel underneath both the floating gates 314 a and 314 b is simultaneously opened. If data need to be written into the floating gate 314 b, a bias voltage V3 such as a ground voltage is applied to the source/drain region 320 a on one side of the control gate 318 a. In the meantime, a bias voltage V4 greater than the bias voltage V3 is applied to the source/drain region 320 b on one side of the control gate 318 b. Ultimately, electrons flow from the source/drain region 320 a to the source/drain region 320 b shown by the arrow in FIG. 4. In here, voltage difference between V4 and V3 must be sufficiently large for electrons flowing into region underneath the floating gate 314 b to acquire enough energy to produce hot electrons. These hot electrons have sufficient energy to cross into the floating gate 314 b but not enough to generate hot electrons underneath the floating gate 314 a.

[0029] Similarly, to write into the floating gate 314 a, the channel underneath both the floating gates 314 a and 314 b must be opened simultaneously and a reverse voltage applied to the source/drain region 320 a and 320 b respectively. This time hot electrons are generated only in the channel underneath the floating gate 314 a and these hot electrons are injected into the floating gate 314 a. In addition, writing into the floating gate 314 a is unaffected by written data in the floating gate 314 b. Data can be written into the floating gate 314 a as long as a sufficiently large bias voltage is applied to the floating gate 314 b so that the channel underneath the floating gate 314 b is opened. In other words, writing data into the floating gate 314 b can be carried out after data is written into the floating gate 314 a.

[0030]FIG. 5 is a schematic cross-sectional view illustrating a second method for programming the double-bit non-volatile memory cell according to this invention. A bias voltage V5 greater than zero is applied to the control gate 318 b. In the meantime, a bias voltage such as a ground voltage V6 smaller than V5 is applied to the source/drain region 320 b on one side of the floating gate 318 b. Here, the difference between the voltages V5 and V6 must be sufficiently large to permit electrons to flow from the source/drain region 320 b to the floating gate 314 b by Fowler-Nordheim tunneling.

[0031] Similarly, to write data into the floating gate 314 a, a bias voltage sufficiently higher than the voltage at the source/drain region 320 a is applied to the control gate 318 a. The writing of data into the floating gate 314 a is unrelated to the floating gate 314 b and the control gate 318 b. Hence, the writing of data into the floating gate 314 a is unaffected by the written state of the floating gate 314 b. In other words, data can be written into the floating gate 314 a after data is written into the floating gate 314 b.

[0032] The following is a description of a method of reading data from the said double-bit non-volatile memory cell of this invention when the source/drain region 320 a/b and the doped region 333 are n-doped. Because of the presence of electric charges, the threshold voltage (bias voltage at the control gate 318 a (318 b) when the channel opens) of the channel underneath the floating gate 314 a (314 b) when the gate is in a written state is greater than the threshold voltage in the erased state. Here, the threshold voltage at the written state is denoted by VTwrite and the threshold voltage at the erased state is denoted by VTerase, wherein VTwrite>VTerase.

[0033]FIG. 6 is a schematic cross-sectional view illustrating a method for reading data from the double-bit non-volatile memory cell according to this invention. In FIG. 6, the process of reading data from the floating gate 314 a is chosen as an example. A positive bias voltage V8 greater than VTwrite is applied to the control gate 318 b so that the opening of the channel underneath the floating gate 314 b is ensured. In the meantime, a positive bias voltage V7 is applied to the control gate 318 a. In general, VTwrite>V7>VTerase. Different bias voltages V9 and V10 are applied to the source/drain regions 320 a and 320 b respectively. The state of the floating gate 314 a is determined by whether the channel between the source/drain regions 320 a and 320 b are conductive or not.

[0034] Since the control gate 318 b has a voltage V8 such that V8>VTwrite>VTerase, the channel underneath the floating gate 314 b is opened independent of any data written into the floating gate 314 b. On the other hand, the control gate 318 a has a voltage V7 such that VTwrite>V7>VTerase. The channel is opened when the floating gate 314 a is in an erased state while the channel is closed when the floating gate 314 a is in the written state. Because both the doped region 333 and the source/drain region 320 a/b are identically doped, conduction between the source/drain regions 320 a and 320 b indicates the floating gate 314 a is in the erased state. Conversely, if no conductive channel is formed between the source/drain regions 320 a and 320 b, the floating gate 314 a is in the written state.

[0035] Similarly, to read stored data from the floating gate 314 b, a bias voltage V7 greater than VTwrite is applied to the control gate 318 a. In the meantime, a bias voltage V8 greater than VTerase but smaller than VTwrite is applied to the control gate 318 b. Data value contained within the floating gate 314 b is determined by the presence or absence of a conductive channel between the source/drain regions 320 a and 320 b.

[0036] The double-bit non-volatile memory of this invention uses two stacked gates 310 a and 310 b but uses just a pair of source/drain regions 320 a and 320 b. Therefore, a through channel is formed between the two source/drain regions 320 a and 320 b only when the channel underneath both the first floating gate 314 a and the second floating gate 314 b are opened. Since the probability of having both floating gates 314 a and 314 b over-erased is small, the chance of maintaining a permanent conductive channel between the source/drain terminals 320 a and 320 b is slim. Thus, errors in data determination are greatly reduced. Moreover, one of the stacked gates 310 a (310 b) can be regarded as a transfer gate because the transfer gate actually prevents the other transfer gate from being over-erased.

[0037] In addition, the double-bit memory cell is capable of holding two binary bits with one stacked gate 310 a (310 b) serving as the transfer gate of the other stacked gate 310 b (310 a). Unlike the cell of a conventional split-gate having an additional transfer gate on one side of the control gate, the area for holding each bit is smaller.

[0038] Furthermore, unlike a conventional split-gate design that demands a patterning of the floating gate before patterning the control gate and the transfer gate together, one stacked gate 310 a (310 b) serves as the transfer gate of the other stacked gate 310 b (310 a). Hence, the stacked gate 310 a/b may be produced by a self-alignment method so that any non-conformity between different electrical devices is minimized.

[0039] Furthermore, the doped region 333 serves as a current path only. Hence, the doped region 333 can be doped identically as the source/drain region 320 a/b. In other words, width a of the doped region 333 can be smaller than width b of the source/drain region 320 a/b. Ultimately, the area occupied by each double-bit non-volatile memory cell is smaller than the conventional split gate cell design as well as the conventional stacked gate design.

[0040] It will be apparent to those skilled in the art that various modifications and variations can be made to the structure of the present invention without departing from the scope or spirit of the invention. In view of the foregoing, it is intended that the present invention cover modifications and variations of this invention provided they fall within the scope of the following claims and their equivalents.

Referenced by
Citing PatentFiling datePublication dateApplicantTitle
US6950348Jun 20, 2003Sep 27, 2005Sandisk CorporationSource controlled operation of non-volatile memories
US7173860Aug 9, 2005Feb 6, 2007Sandisk CorporationSource controlled operation of non-volatile memories
Classifications
U.S. Classification365/185.28
International ClassificationG11C11/56, G11C16/04, G11C16/10
Cooperative ClassificationG11C16/0475, G11C16/10, G11C2211/5612, G11C16/0458, G11C11/5621
European ClassificationG11C16/04M2, G11C11/56D, G11C16/04F4P
Legal Events
DateCodeEventDescription
Feb 13, 2002ASAssignment
Owner name: UNITED MICROELECTRONICS CORP., TAIWAN
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:CHEN, CHIN-YANG;REEL/FRAME:012602/0691
Effective date: 20010213