法国专利FR3021803A1 MEMORY CELLS TWINLY ACCESSIBLE INDIVIDUALLY READING

专利PDF首页>>法国专利

专利附录

专利说明

权利要求

类似技术

同族专利

引用文献

法律状态

优先权

专利摘要:
The invention relates to a non-volatile memory (MA2) on a semiconductor substrate, comprising: a first memory cell comprising a floating gate transistor (TRi, j) and a selection transistor (ST) having a buried vertical control gate ( CSG), a second memory cell (Ci, j + i) comprising a floating gate transistor (TRi, j + i) and a selection transistor (ST) having the same control gate (CSG) as the gate selection transistor. the first memory cell, a first bit line (RBLj) connected to the floating gate transistor (TRi, j) of the first memory cell, and a second bit line (RBLj + 1) connected to the floating gate transistor (TRi, j + i) of the second memory cell (Ci, j + i).
公开号:FR3021803A1
申请号:FR1454893
申请日:2014-05-28
公开日:2015-12-04
发明作者:Rosa Francesco La；Stephan Niel；Arnaud Regnier
申请人:STMicroelectronics Rousset SAS；
IPC主号:

专利说明:

[0001] The present invention relates to nonvolatile memories and in particular to a memory and memory cell structure of the type described in application US 2013/0228846. By way of reminder, FIG. 1 represents such a memory map structure MAO and shows memory cells Mij, Mi-ij, Mij + i, Mi_i j + 1 of the aforementioned type, here belonging to two adjacent physical pages Pi, Pi_1 of memory plane, of respective ranks "i" and "i-1". The memory cells Mij, Mi-ij, Mij + 1, Mi_ij + i are accessible in reading and writing via bit lines BLj, BLi + i, a word line WLi_i j and control lines of CGLi grid, CGLi_i. Each memory cell comprises a floating gate transistor, respectively Tij, Ti-ij, Tij + i, Ti_i j + 1. The drain terminals D of the transistors Tij, Ti_i j are connected to the bit line BLi and the drain terminals of the transistors Tij + i, Ti_ij + i are connected to the bit line BLi + 1. The control gates CG of the transistors Tij, Tij + 1 are connected to the gate control line CGLi and the control gates CG of the floating gate transistors Ti_ij, Ti_i j + 1 are connected to the gate control line CGLi-i. Each floating gate transistor T jj-pi also has its source terminal connected to a source line SL via a selection transistor ST. The selection transistors ST memory cells Mij and j share the same CSG control grid and the two memory cells are, therefore, called "binoculars". Likewise, memory cells Mij + i and j + 1 are twin memory cells and their selection transistors ST have a common control gate CSG. Each common control gate is preferably a vertical gate buried in a substrate receiving the MAO memory plane, the source line SL being also a buried line. These CSG common control grids, or twin memory cell selection grids, are connected to the word line WL1-1,1.
[0002] Such memory cells are erased or programmed by the channel, that is to say by carrying the substrate at a positive erase or negative programming voltage causing the extraction of electric charges from their floating gates or the injection of electric charges in their floating gates, by Fowler-Nordheim effect.
[0003] More particularly, the erasure of a memory cell is ensured by combining the positive voltage applied to the substrate with a negative voltage applied to the gate of its floating gate transistor while the gate of the gate transistor is controlled. The floating cell of the twin memory cell receives a positive erase inhibition voltage to prevent it from being simultaneously erased (Fig. 11 of the aforementioned application).
[0004] Likewise, the programming of a memory cell is ensured by combining a negative voltage applied to the bit line of the memory cell and the substrate, to a positive voltage applied to the control gate of its floating gate transistor, while that the control gate of the floating gate transistor of the twin memory cell receives a negative programming inhibit voltage to prevent it from being simultaneously programmed (Fig. 12 of the aforementioned application). Finally, the reading of a memory cell is ensured by applying a positive voltage to the control gate of its floating gate transistor, as well as a positive voltage to the corresponding bit line, while the twin memory cell, which is connected to the same bit line, receives on its control gate a negative reading inhibition voltage to prevent it being simultaneously read (Fig 9 of the aforementioned request). This memory plane structure having twin memory cells comprising a shared vertical selection grid 20 and buried in the substrate, has the advantage of being of a small size. The channel erasure method that they require is well suited to producing an erasable memory array per page, but is less suitable for producing a word-erasable memory array. This is apparent by comparing the word-erasable memory array shown in Fig. 24 of the above-mentioned application to the erasable memory map per page shown in Fig. 23 of this application, the former being more complex than the second. Thus, for the memory plane to be erasable by word, each CGL grid control line, instead of being connected to all the memory cells of a page, must be divided into a plurality of grid control lines. one grid check line per word. This results in significant complexity in the word and column line decoders, and requires the provision of various voltage switches for controlling, within each page, the grid control lines of the different words. It may therefore be desirable to provide an improvement of this memory and memory cell structure which is more suitable for the implementation of a word-erasable memory, and does not lead to an increase in complexity of the control elements. of the memory map. This conventional memory and memory cell structure also requires providing a word line decoder capable of applying a positive read voltage to a memory cell to be read while applying a negative read inhibit voltage to its memory cell. twin memory cell, as mentioned above. It may therefore also be desirable to provide a further improvement of this memory and memory array structure that enables a memory cell to be read without applying a negative voltage to the twin memory cell. Embodiments of the invention provide a non-volatile memory on a semiconductor substrate, comprising a first memory cell having a first floating gate transistor electrically connected to a selection transistor having a buried vertical control gate, the selection having a vertical channel region extending opposite a first face of the buried vertical control grid; a second memory cell having a first floating gate transistor electrically connected to a selection transistor having the same control gate as the selection transistor of the first memory cell and having a vertical channel region extending next to a second face of the buried vertical control gate, and opposite the channel region of the selection transistor of the first memory cell; a first bit line electrically connected to the first floating gate transistor of the first memory cell, and a second bit line electrically connected to the first floating gate transistor of the second memory cell. According to one embodiment, the memory comprises a first gate control line electrically connected to a control gate of the first floating gate transistor of the first memory cell, a second gate control line electrically connected to a control gate the first floating gate transistor of the second memory cell, and memory cell reading means, configured for, during reading of a memory cell, applying a positive selection voltage to the gate control line connected to the cell memory to be read, and read the first memory cell through the first bit line or read the second memory cell 35 through the second bit line.
[0005] According to one embodiment, the reading means of the two memory cells comprise a word line decoder configured to assign to the two grid control lines the same line address, and, during the performance of at least one operation selected from reading, programming or erasing operations of one of the memory cells, selecting the gate control line to which the memory cell is connected according to a part of the line address of the two Grid control lines and secondly the least significant bit of a column address of the memory cell.
[0006] According to one embodiment, each memory cell further comprises a second floating gate transistor whose floating gate is electrically connected to the floating gate of the first floating gate transistor, and comprising a conductive region electrically connected to the conduction terminal of the second floating gate transistor which extends opposite its floating gate via a tunnel dielectric layer.
[0007] According to one embodiment, the memory comprises a third bit line electrically connected to both the second floating gate transistor of the first memory cell and the second floating gate transistor of the twin memory cell.
[0008] According to one embodiment, the floating gates of the first and second floating gate transistors are formed by a single piece of a conductive material. According to one embodiment, the conductive region is a doped region of the substrate.
[0009] According to one embodiment, the common control gate of the selection transistors of the twin memory cells is a vertical grid buried in the substrate. Embodiments of the invention also relate to a semiconductor chip chip circuit comprising a nonvolatile memory according to the invention. Embodiments of the invention also relate to a method of manufacturing a memory according to the invention, comprising the steps of forming in the substrate insulating trenches delimiting at least a first substrate strip, forming in the substrate a conducting trench arranged transversely to the substrate strip, to form the common control gate to the selection transistors of the two twin memory cells, forming on the substrate a conductive floating gate arranged transversely to the substrate strip, with the interposition of a first dielectric layer, forming a control gate on the floating gate, with interposition of a second dielectric layer, to obtain a stack of grids, doping the substrate strip on each side of the stack of grids, forming in a first metal level two first sections of bit lines each electrically connected to a r drain region of one of the floating gate transistors, forming in a second metal level two second sections of bit lines each electrically connected to one of the first sections of bit lines, and forming in a third level of metal two bit lines each electrically connected to one of the second bit line sections of the second metal level. According to one embodiment, the method comprises a step of doping the second substrate strip before forming the stack of grids and before doping the two substrate strips on either side of the stack of grids to form a conductive region facing the floating gate of the second floating gate transistor. Embodiments of the invention also relate to a read / write method of a memory cell among a pair of twin memory cells comprising a first memory cell having a first floating gate transistor electrically connected to a selection transistor having a buried vertical control gate, the selection transistor of the first memory cell having a vertical channel region extending opposite a first face of the buried vertical control gate, and a second memory cell comprising a first gate transistor Floating vessel electrically connected to a selection transistor having the same control gate as the selection transistor of the first memory cell and having a vertical channel region extending opposite a second face of the buried vertical control gate , and in front of the channel region of the selection transistor of the first memory cell, the method comprising the steps of providing a first bit line electrically connected to the first floating gate transistor of the first memory cell, providing a second bit line electrically connected to the first floating gate transistor of the first memory cell; second memory cell, and, during reading of a memory cell, applying a positive selection voltage to the gate control line connected to the memory cell to be read, and reading the first memory cell through the memory cell. first bit line or read the second memory cell via the second bit line.
[0010] According to one embodiment, the method comprises the steps of assigning to the two grid control lines the same line address, and, during the performance of at least one operation selected from reading, programming or 5 of erasing one of the memory cells, select the grid control line to which the memory cell is connected according to a part of the line address of the two grid control lines and secondly the least significant bit of a column address of the memory cell.
[0011] Embodiments and methods for manufacturing a memory array structure and memory cells according to the invention, as well as methods for reading and writing memory cells according to the invention, will be described in the following in with reference to the accompanying drawings, in which: FIG. 1, previously described, is the electrical diagram of a conventional memory plane and memory cell structure; FIG. 2 is the electrical diagram of a first embodiment of FIG. embodiment of a memory array and memory cell structure according to a first improvement according to the invention; - FIG. 3 is a sectional view of a memory cell of FIG. 2; FIG. 4 is another sectional view. of the memory cell; FIG. 5 shows voltages applied to the memory array of FIG. 2 for erasing a memory cell; FIG. 6 is a sectional view of a memory cell of FIG. 5; and show The voltage applied to the memory cell, FIG. 7 is another sectional view of the memory cell, and shows the voltages applied to the memory cell, FIG. 8 shows the voltages applied to the memory array of FIG. for programming a memory cell by Fowler Nordheim effect; FIG. 9 is a sectional view of a memory cell of FIG. 8 and shows voltages applied to the memory cell; FIG. 10 is another view in section of the memory cell and shows the voltages applied to the memory cell, - Figure 11 shows voltages applied to the memory plane of Figure 2 for programming a memory cell by hot electron injection, - Figure 12 is a sectional view of a memory cell of FIG. 11 and shows voltages applied to the memory cell, FIG. 13 shows voltages applied to the memory array of FIG. 2 for reading a cell. 14 is a sectional view of a memory cell of FIG. 13 and shows voltages applied to the memory cell; FIGS. 15 to 24 show steps of a method of manufacturing a memory cell; FIG. memory cell shown in FIGS. 2 to 4; FIG. 25 is the electrical diagram of a memory comprising the memory plane of FIG. 2; FIG. 26 is the electrical diagram of a second embodiment of a structure. memory plane and memory cell according to the first improvement, - Figure 27 is the electrical diagram of a first embodiment of a memory plane structure and memory cell according to a second improvement according to the invention, - the FIGS. 28 to 32 show steps of a method of manufacturing a memory cell shown in FIG. 27; FIG. 33 shows voltages applied to the memory array of FIG. 27 for reading a memory cell; Figure 34 The sch Electrical emitting of a memory comprising the memory array of Fig. 27; Fig. 35 is the wiring diagram of a second embodiment of a memory array and memory cell structure according to the second enhancement. FIG. 2 is the electrical diagram of an embodiment of two memory cells C14.1 and of a memory plane MA1 according to a first improvement according to the invention of the memory plane and memory cell structure of FIG.
[0012] The memory cells are read and write accessible via a first bit line RBL, a second bit line EBL, a word line WL, _ ,, 'and two CGL gate control lines CGL, _, . The memory cell belongs to a physical page P, of the memory plane and the memory cell C, j belongs to an adjacent page 13, 4. Pages 131_1 may include various other memory cells and the memory plane MA1 may include various other pages. The memory cell comprises two floating gate transistors TR, j, TE, whose floating gates FGr, FGe are interconnected, the floating gate transistor TR being dedicated to reading the transistor memory cell and the gate transistor. floating TE, j dedicated to erasing the memory cell. In one embodiment, the interconnection 3021803 8 floating gates FGr, FGe is provided by manufacturing the two floating gates from the same conductive element CFG. The transistor TRIJ has a control gate CGr connected to the gate control line CGL1, a drain terminal D connected to the bit line RBLJ and a source terminal S connected to the drain terminal D of a transistor. ST selection whose source terminal S is connected to a source line SL. The transistor TEIJ has a control gate CGe connected to the gate control line CGL1, a drain terminal D connected to the bit line EBLJ and a source terminal S connected to the drain terminal D of a transistor of 10 ST selection whose source terminal S is connected to a source line SL. The memory cell Ci_i j has the same structure as the memory cell C, J and comprises two floating gate transistors TRi_i j, j whose floating gates FGr, FGe are interconnected and / or formed by the same conductive element CFG. The transistor TRi_i ja 15 has a control gate CGr connected to the gate control line CGLi_i, a drain terminal D connected to the bit line RBLJ and a source terminal S connected to the drain terminal D of a transistor. ST selection whose source terminal S is connected to a source line SL. The floating gate transistor TE1_i ja a control gate CGe connected to the gate control line CGLi_i, a drain terminal D connected to the bit line EBLJ and a source terminal S connected to the drain terminal D d ' a selection transistor ST whose source terminal S is connected to a source line SL. The selection transistors ST associated with the floating gate transistors TRIJ, j have a common control gate CSG connected to the word line WL14,1, which is preferably in the form of a vertical grid buried in a substrate receiving the memory map MAO. Similarly, the selection transistors ST associated with the floating gate transistors TEIJ, j have a buried vertical common control gate CSG which is connected to the word line WL14,1. The memory cell Cij is therefore, structurally, the equivalent of the combination of the memory cells Mij, MI, J + 1 shown in FIG. 1, the floating gates of which would have been electrically connected. Similarly, the memory cell j is, structurally, the equivalent of the combination of memory cells 1 41-1J, 1 41-1J + 1 shown in Figure 1 whose floating gates were electrically connected. The two memory cells can therefore be called "dual" memory cells. Since their selection transistors ST have the same control grid two by two, these memory cells can also be called "twin" memory cells.
[0013] The floating gate transistor TE, J, however, differs from the floating gate transistor T, J + of the memory cell M, J + 1 in that it comprises a conductive region IS that extends in front of its floating gate. FGe with interposition of a tunnel dielectric layer, and which is electrically connected to its drain terminal D. Similarly, the floating gate transistor TE, _, j differs from the floating gate transistor Ti_u_pi of the memory cell 1 4 , 4 j + i in that it comprises a conductive region IS which extends in front of its floating gate FGe via a tunnel dielectric layer, and which is electrically connected to its drain terminal D.
[0014] Fig. 3 is a sectional view of an embodiment of transistors TE, J, TE, _, j and their respective selection transistors ST. The floating gate transistors are made on a substrate PW, for example silicon, forming the upper region of a WF semiconductor chip. The WF chip is initially a wafer on which several integrated circuits are made, which is then cut into individual chips. The source lines SL of the selection transistors ST are formed by a deep doped region n0 which here forms a collective source plane for the entire memory plane. The common control gate CSG of the selection transistors ST is formed with a conductive material, for example polysilicon (polycrystalline silicon), deposited in a trench made in the substrate, and isolated therefrom by a dielectric layer OD. This "conducting trench" also forms the word line WL, 4 ,, along an axis perpendicular to the plane of the figure.
[0015] The floating gates FGe of the transistors TE, J, TE, 4 j are arranged on each side of the trench CSG, and rest on the substrate PW via a tunnel dielectric layer D1. They are formed here by a conductive part, for example made of polysilicon, which extends to the transistors TR, J, TR, 4 j (not visible in this sectional plane, see Fig. 4) to form also the grids floating of these transistors.
[0016] The control gates CGe of the transistors TE, J, TE, 4 j extend above the floating gates via a dielectric layer D2. They are formed here by conductive strips of polysilicon which also form the CGL 'CGL gate control lines, 4 along an axis perpendicular to the plane of the figure.
[0017] N2 and n3 doped regions implanted on each side of the gate stack FG / CGe respectively form the drain (D) and source (S) regions of the transistors TE, J, TE, 4 j, the regions n3 also forming the drain regions (D) of the selection transistors ST. The source regions (S) of the selection transistors ST are here formed by the nO layer, the CSG common vertical gate of the selection transistors extending here to the nO region. In an alternative embodiment, the lower end of the CSG conductive trench does not reach the nO region and a deep doped pocket is implanted between the trench and the nO layer to form the source region of the ST selection transistors.
[0018] The gate stacks FG / CGe of the transistors TE, J, TE, 4 j are covered by a dielectric layer D3 on which the bit line EBL 1 extends. Cl contacts pass through the layer D3 to electrically connect the bit line EBL, to the drain regions n2 (D) of the transistors TE, J, TE, 4 ,,. The conductive regions IS of the transistors TE, 4 ,, are here n1 doped regions of the substrate which extend under the floating gates FG, between the drain regions n2 and source n3 of the transistors TE, J, TE, 4 , and are therefore covered by the tunnel dielectric layer D1. The dielectric layers D0, D1, D2 and D3 are, for example, silicon dioxide SiO2. FIG. 4 is a cross-sectional view of the floating gate transistors TR, J, TR, 4 of the memory cells C, i and their respective selection transistors ST. The structure of these transistors is essentially identical to that of transistors TE, J, TE, 4 and will not be described again. The control gates CGe of these transistors are formed by the polysilicon parts forming the CGL gate control lines CGL, 4 and their floating gates FG are formed by the same pieces of polysilicon as those forming the floating gates of the transistors. TE, J, TE, 4 (Fig. 3). Their drain regions n2 (D) are connected by contacts C1 to the RBL bit line, which extends over the dielectric layer D3. The transistors TR, J, TR, 4 j differ from the transistors TE, J, TE, 4 in that they do not include the conducting region IS extending under the floating gate FG. Thus, when these transistors are biased by a suitable gate voltage, a conductive channel CH1 or CH1 'may be formed between the drain region n2 (D) and the source region n3 (S). A vertical conductive channel CH2 or CH2 'can also be formed between the drain n3 (D) and source (S) regions of the selection transistor ST if the common vertical gate CSG of these transistors simultaneously receives a bias voltage. More particularly, the vertical channel region CH2 of the selection transistor ST of the memory cell extends opposite a first face of the buried vertical control gate CSG, and the vertical channel region CH2 'of the transistor selecting ST of the memory cell C j extends opposite a second face of the buried vertical control gate, and therefore opposite the channel region CH 2 of the selection transistor of the memory cell Ci, j.
[0019] The transistors TRIJ, j may also differ from the transistors TEIJ, TEI-tj by the thickness of their tunnel dielectric layer D1, which may be different from that which extends under the floating gates of the transistors TEIJ, j, this choice being provided to those skilled in the art depending on the method of programming the memory cells that will be retained, namely by Fowler-Nordheim effect by means of TEIJ transistors, or by injection of hot electrons using TRIJ transistors, j, these two options being described later. Unlike transistors TRIJ, j, transistors TEIJ can not have a conductive channel CH1 controlled by the voltage applied thereto, because the region extending between their drain regions n2 and source n3 is short. -circuited by the n1 doped region (Figure 3). These transistors are therefore always on regardless of their gate voltage, and can not be used to read memory cells Cij, j. However, they can be used to erase memory cells under conditions where no drain-source current passes through them, ie by Fowler-Nordheim effect. They can also be used to program the memory cells under conditions where no drain-source current passes through them, also by the Fowler-Nordheim effect, as will be seen later. As a result, the selection transistors ST associated with the TEIJ transistors are not used and are only present here to rationalize the fabrication of the memory cells according to an embodiment of a manufacturing method described below. It may indeed be simpler to make a useless transistor within a set of transistors used, when the non-realization of the useless transistor would involve additional masking and photolithography steps. As, on the one hand, these selection transistors are on when the word line WL14,1 to which they are connected receives a positive voltage, and, on the other hand, the floating gate transistors TEIJ, j are always on. Because of their region IS, it should be ensured, when designing the memory control devices, that the corresponding bit line EBLJ can not simultaneously receive a voltage other than zero.
[0020] In summary, the TEIJ transistor can be used as an erase transistor of the memory cell Cij by the Fowler-Nordheim effect, which consists of a static programming without programming current, while the transistor TRIJ can be used as a transistor. reading of the memory cell. Similarly, the transistor TE 4 j can be used as an erase transistor of the memory cell Ci_i j by the Fowler-Nordheim effect and the transistor TRi_i j can be used as the read transistor of the memory cell. The bit line RBLJ may be used as the read bit line and the EBLJ goal line as the erase bit line of the memory cell Cij or the memory cell Ci_u.
[0021] With regard to the programming of the memory cells CI, Ci_i j, embodiments of the invention provide two methods, at the choice of those skilled in the art, namely a Fowler-Nordheim effect programming method. by means of the erase transistor TEIJ or TEl_i j, or a method of programming by hot electron injection by means of the read transistor TRIJ or TE1-1, J.
[0022] Methods for erasing, programming and reading cells of the memory array MA1 will be described in the following, assuming by way of example that it is desired to erase, program and read the memory cell C1J.
[0023] Deletion of a memory cell by the Fowler-Nordheim effect via the erase transistor TEij A method of erasing the memory cell CI without erasing the memory cell j, via the erase transistor TEIJ, is described by the table 1 of the Annex, which is an integral part of the description. Figures 5 and 6 illustrate this erasure method. Figure 5 is the circuit diagram of Figure 2 in which the voltage values in Table 1 have been reported. FIG. 6 is a sectional view of transistors TEIJ, identical to that of FIG. 3, in which the voltage values shown in table 1 have been reported. The conductive region IS of the transistor TEIJ is brought to the positive voltage EBLV applied to the bit line EBLJ, here 6V, via the contact C1 and the drain region n1 (D) of the transistor. Since the control gate CGe of the transistor TEIJ is raised to the negative voltage CGV1, here -8V, a difference in voltage dV equal to -14V appears between this control gate and the conductive region IS. 1021803 13 electron extraction of the floating gate FGe by Fowler-Nordheim effect, which puts the transistor TE, J in the erased state. The bit line RBL, connected to the floating gate transistor TR, J being high impedance, this transistor plays no role in the erasing process of the memory cell. Since the floating gate of the transistor TR, J is however electrically connected to that of the transistor TE, J, the transfer of electric charges also causes the erasure of the transistor TR, J, the memory cell as a whole being thus erased via of transistor TE, J. The conductive region IS of the transistor TE, _, j of the twin memory cell C, _, j is also raised to the positive voltage EBLV applied to the bit line EBL, here 6V, via the contact Cl and of the drain region n1 (D) of the transistor. The control gate CGe of the transistor being brought to the positive voltage CGV, _ ,, here 3V, it appears between this control gate and the conductive region IS a voltage difference dV equal to -3V, which is insufficient to extract electrons of the floating gate of the transistor. The twin memory cell C, _, is thus not erased. This structure of memory array and memory cell thus allows an individual erasure of each memory cell, ie a deletion per bit. This possibility makes it possible to perform indifferently a memory erasable by bit, word or page without modifying the general structure of the memory plane or its control organs. FIG. 7 is a sectional view of two transistors TE, +1, TE, _, j +, of two memory cells C ,,, + 1, (not shown in FIGS. 2 or 5) which contiguously with the memory cells C, J, C, _ ,,,. The memory cells C ,,, + 1, are connected to the same word line WL 1 as the memory cells C, J, C, _ ,,, but are connected to a different bit line EBL, +, which receives the default voltage EBLV *, here OV. The transistors TE, J + 1, TE, _, j + have their control gates CGe connected to the same gate control lines CGL 'CGL, _, as the transistors TE, J, TE, _, j and thus receive the same voltages CGV 'here -8V, and CGV, _ ,, here 3V. Thus, the voltage difference dV between the control gate CGe of the transistor TE, J + and its conducting region IS is equal to -8V and this transistor undergoes an erase stress, that is to say a parasitic erase of low intensity which could, if the transistor was in the programmed state, and after many cycles of erasure of other memory cells connected to the gate control lines CGL 'CGL, _ ,, substantially alter its threshold voltage and therefore result in corruption of its state, and thus corruption of the data bit associated with the programmed state.
[0024] On the other hand, the voltage difference dV between the control gate CGe of the transistor TE, 4 j-pi and its conducting region IS is 3V and this transistor undergoes no erasure stress, the gate control line CGLi_i being raised to only 3V. Likewise, default voltages applied to memory cells connected to other word lines WL (not shown in the figures) do not cause erasure stress in these memory cells. Ultimately, the erasing method according to the invention not only allows individual erasure of each memory cell, but also limits the occurrence of erasure stress to memory cells connected to the same gate control line. , while various other known erasure methods, allowing only one deletion per word, if not per page, also cause erasure stress to the memory cells connected to other word lines. The management of the erasure stress, by methods known per se of refreshing the memory cells, is therefore simplified, given the reduced number of memory cells to be refreshed. For example, it will be possible to initiate a refresh sequence of the memory cells of a word line after N memory cell programming cycles of this word line, by providing an erase cycle counter associated with the word line. .
[0025] Programming of a memory cell by Fowler-Nordheim effect via the erase transistor TEij A method of programming the memory cell C, J without programming the memory cell Ci_i j, via the erase transistor TE, J, is described in Table 2 in the Annex.
[0026] Figures 8 and 9 illustrate this programming method. Figure 8 is the circuit diagram of Figure 2 in which the voltage values in Table 2 have been reported. FIG. 9 is a sectional view of transistors TE, J, TE 4 identical to that of FIG. 3, in which the voltage values in Table 2 have been reported. The conductive region IS of the transistor TEij is brought to the voltage EBLV applied to the bit line EBLi, here OV, via the contact C1 and the drain region n1 (D) of the transistor. Since the control gate CGe of the transistor TEij is raised to the positive voltage CGVi, here 14V, a positive voltage difference dV equal to 14V appears between this control gate and the conductive region IS (FIG. electron injection 3021803 15 in the floating gate FGe by Fowler-Nordheim effect, which puts the transistor TE, J in the programmed state. The bit line RBLJ connected to the floating gate transistor TR, J being high impedance, this transistor plays no role in the programming process of the memory cell. The floating gate of the transistor TR, J, however, being electrically connected to that of the transistor TE, J, the transfer of electric charges also causes the programming of the transistor TR, J, the memory cell Cij as a whole being thus programmed via of transistor TE, J. The conductive region IS of the transistor TE, _, j of the twin memory cell C 1, j is brought to the voltage EBLV applied to the bit line EBLJ, here OV, via the contact C1 and the region drain n1 (D) of the transistor. The control gate CGe of the transistor TE, _, j being brought to the positive voltage CGV, _ ,, here 3V, there appears between this control gate and the conductive region IS a voltage difference dV equal to 3V, which is insufficient to inject electrons from the floating gate of the transistor. The twin memory cell C, _, is therefore not programmed. FIG. 10 is a sectional view of transistors TE, J + 1, TE, _, j +, neighboring memory cells C, J + 1, C, _, j +, already described in relation to FIG. 7, connected to FIG. same word line WL, _ ,,, as the memory cells C, _ ,,, but connected to the neighboring bit line EBLJ +, which receives the default voltage EBLV *, here 6V. The transistors TE, J + 1, TE, _, j_p, have their control gates CGe connected to the same gate control lines CGL 'CGL, _, as the transistors TE, J, TE, _, j and therefore receive the same voltages CGV 'here 14V, and CGV, _ ,, here 3V. Thus, the voltage difference dV between the control gate CGe of the transistor TE, J +, and its conducting region IS is 8V and this transistor is under programming stress, ie a low intensity parasitic programming. . Moreover, the voltage difference dV between the control gate CGe of the transistor TE, _, j +, and its conducting region IS is 3V and this transistor undergoes no programming stress, the gate control line CGL, being raised to only 3V. Likewise, default voltages applied to memory cells connected to other word lines WL (not shown in the figures) do not cause any programming stress in these memory cells. Ultimately, this programming method, like the previously described deletion method, only causes electrical stress to memory cells connected to the same gate control line, the effects of which can be neutralized by a method of control. refreshment of the type mentioned above. Programming of a memory cell by injection of hot electrons via the read transistor TRij A programming method of the memory cell without programming of the memory cell C 1, j, via the read transistor TR, J, is described by Table 3 in the Appendix.
[0027] Figures 11 and 12 illustrate this programming method. Figure 11 is the circuit diagram of Figure 2 in which the voltage values in Table 3 have been reported. FIG. 12 is a sectional view of transistors TR, J, TR, 4 j identical to that of FIG. 4, in which the voltage values appearing in table 3 have been reported.
[0028] The transistor TR, J receives the positive voltage CGV 'here 10V, on its control gate and is in the on state, the conductive channel CH1 appearing in the substrate PW under the gate stack FG / CGr. The selection transistor ST associated with the transistor TR, J receives the positive selection voltage SV, here 1 to 2V, on its buried vertical grid CSG, and is in the on state, the vertical conductive channel CH2 appearing in front of the gate CSG. Since the RBL bit line is raised to the positive voltage RBLV, here 4V, and the source line SL is connected to the ground (0V), a current flows from the bit line to the source line through the transistor TR , J and the corresponding selection transistor ST. This current corresponds to an electron flow HE shown in FIG. 12, in the opposite direction of the current. This electron flux contains electrons with high kinetic energy (hot electrons) that bypass the n3 doped region ("cold" region). Some of these electrons are injected into the floating gate at an injection point HI, causing the programming of the transistor TR, J, as well as the programming of the erase transistor TE, J which plays no part in the programming process here. . The transistor TR, 4 j of the twin cell 30, on the other hand, receives the voltage CGV, 4 which is equal to OV, so that it undergoes no spurious programming process, nor any of the other reading transistors of the memory plane, which receive only zero voltages. Reading a memory cell via the read transistor TRij A method of reading the memory cell Cij via the read transistor TR, J is described in Table 4 in the Appendix. Figures 13 and 14 illustrate this reading method. Fig. 13 is the circuit diagram of Fig. 2 in which the voltage values in Table 4 have been reported. FIG. 14 is a sectional view of the transistors TR, J, TR, 4 j identical to that of FIG. 4, in which the voltage values appearing in table 4 have been reported. The transistor TR, J receives the positive voltage CGV 'here from 2 to 3V, which is lower than the threshold voltage of the programmed transistor but greater than the threshold voltage of the erased transistor. If the transistor TR, J is in the erased state, that is to say if it has a threshold voltage Vt lower than the voltage CGV 'the conductive channel CH1 appears in the substrate PW, under the stack of FG / CGr grid. The selection transistor ST associated with the transistor TR, J receives the positive selection voltage SV, here 3V, on its buried vertical grid CSG, and is in the on state, the vertical conductive channel CH2 appearing in front of the buried gate CSG . The bit line RBL being raised to the positive voltage RBLV, here 1V, and the source line SL being connected to ground (0V), the transistor TR, J is traversed by a read current Ir which flows from the line bit to the source line. This current Ir is on the other hand zero if the transistor TR, J is in the programmed state, that is to say if it has a threshold voltage greater than the voltage CGV. A current amplifier (not shown) connected to the RBL bit line, makes it possible to detect the presence or absence of the current Ir, and to deduce the erased or programmed state of the transistor TR, J, to which a logic value , 0 or 1, is assigned by convention.
[0029] The transistor TR, 4 j of the twin memory cell receives the negative voltage CGV, 4, here -2V. This transistor, if it is in the erased state, may have a threshold voltage close to zero. Applying a negative grid control voltage ensures that it remains in the off state. Indeed, since this transistor is connected to the same RBL bit line, as the transistor TR, J being read, its turning on could corrupt the reading of the transistor TR, J. Figures 15 to 22 show steps of a method of manufacturing memory cells Ci_i j previously described. FIG. 15 shows a preliminary stage of formation in the PW substrate of three STIO, ST1, ST2 insulation trenches of STI type ("Shallow Trench Insolation") which delimit two substrate strips 51, S2 in which the memory cells go to be realized. This step is preceded by an implantation step 3021803 18 in the substrate of the buried layer nO forming a source plane (not visible in the figure) or implantation of several source lines. A source plane is generally preferred over multiple source lines if it is intended to erase memory cells by hot electron injection.
[0030] During a step shown in FIG. 16, a conductive trench is formed transversely to the strips S1, S2 by etching the substrate, depositing the dielectric layer OD (not visible) and depositing a PO polysilicon layer and engraving of it. The trench is intended to form both the word line WL, 4 ,, and the buried vertical control grid 1 () of the selection transistors ST of the memory cells. During a step shown in FIG. 17, the substrate band S2 is doped by implantation of N-type dopants, the substrate band S1 being masked during this operation. This step makes it possible to make the conductive region IS which will extend under the floating gate of the erase transistors TEIJ, TE1-1, J. During a step shown in FIG. 18, the previously described tunnel dielectric layer D1 is deposited on the substrate PW, then a polysilicon band P1, intended to form floating gates, is deposited on the substrate strips S1 and S2. .
[0031] During a step shown in FIG. 19, the dielectric layer D2 is deposited on the substrate PW, then a layer of polysilicon P2 is deposited on the layer D2. The layer P2 is then etched with the layer D2 as well as with the layer P1 to form the gate control lines CGL1, CGLi_i, and, beneath it, the common floating gates CFG, resulting from the simultaneous etching of the During a step shown in FIG. 20, the substrate strips Si, S2 are doped by implantation of self-aligned dopants on the gate control lines CGL1, CGLI4 and on the word line WL14,1. substrate band S2 is therefore doped a second time. This step reveals the source S and drain D regions of the transistors TR j, j, TE1J, j, as well as the drain regions of the selection transistors ST. In a step shown in Fig. 21, the dielectric layer D3 is deposited on the substrate and holes are made in the layer D3, then metallized to form the contacts C1. Cl contacts extend above D drain regions of the TRIJ transistors, and others above the drain regions D of the TEIJ transistors. During a step shown in FIG. 22, a metal layer M1 ("metal 1") is deposited on the substrate and is etched to obtain two conductive strips which form the bit lines RBLJ and EBLJ, the first being arranged on the contacts C1 made above the drain regions D of the transistors TRIJ, j and the second arranged on the contacts made above the drain regions D of the TEIJ transistors, FIG. 23 is identical to FIG. 22 and shows the area occupied in width W and in length L by each memory cell C1J, j, the assembly forming a "basic brick" 10 of the memory plane, containing two twin dual cells, the repetition of which allows the design of a memory array MA1 of variable size chosen according to the intended application. The contacts C1 being in this case shared by memory cells made above and below memory cells Cij, Ci_u ("above" and "below" in the plane of the figure), which are not shown only half of the area occupied by the contacts C1 is considered to be part of the "base brick" 20. Although these memory cells C1j, j have a surface area double that shown in FIG. It will be appreciated by those skilled in the art that the semiconductor surface they occupy is not dissimilar to that occupied by conventional memory cells having "planar" type selection transistors that are not shared, because the grids Buried vertical selection methods greatly reduce their area and even more so that they are shared. Furthermore, in one embodiment of the invention, a memory array according to the invention may comprise a first memory zone made from memory cells as described in FIG. 1, forming an erasable mass memory per page. and a second memory zone made from memory cells according to the invention, forming a bit-wise or word-erasable data memory, offering a finer erasure granularity than the mass memory and more suitable for certain applications.
[0032] FIG. 24 is a sectional view of the memory cell Cij according to a section plane AA 'shown in FIG. 23 and perpendicular to the sectional plane of FIGS. 3 and 4, together showing the transistors TRIJ, TEIJ and their common floating gate FGT. This figure also shows that it is possible to further reduce the width W of the memory cell by reducing the width of the central isolating trench STIO which separates the transistors TRIJ, TEIJ, this insulating trench does not need to present the insulation width usually used for trenches STI1, STI2 which separate neighboring memory cells, since transistors TR, J, TE, J are electrically coupled. FIG. 25 is the electrical diagram of a memory MEM1 comprising the memory plane MA1 according to the invention, only the cells Ci_i j being represented. The memory comprises a control circuit CCT1, a word line decoder RD1, a column decoder CD1, reading amplifiers SA in number equal to the number of bits of a word DTR to be read in the memory, for example a word of eight bits BO-B7, and BLT1 programming latches for applying voltages to the RBL or EBL bit lines, depending on a DTW word to be written to the memory, for example an eight bit word BO-B7 . The RD1 word line decoder controls the voltages applied to the CGL gate control lines' CGL, 4 and to the word line WL, 4 ,, as a function of a most significant address A (n-1) -A (x) a word, or line address. The decoder CD1, in combination with the latches BLT1, controls the voltages applied to the bit lines RBL, EBL, as a function of a low-order address A (x-1) -A (0) of the word, or address column, the row and column addresses together forming the address A (n-1) -A0 of a word to read or write in the memory plane. In read mode, the decoder CD1 connects the SA amplifiers to the RBL bit lines, connected to the memory cells to be read, and the read amplifiers provide the DTR word. The circuit CCT1 comprises, for example, a CPU, a VGEN voltage generator, and address and data registers. It executes read or write commands, controls the decoders, provides the voltages necessary for the read or write operations (erasure-programming), the supply of the high and low order addresses to the decoders , and if necessary runs a program for refreshing memory cells. Although the improvement just described was originally designed to be applied to a memory cell structure of the type shown in FIG. 1, it will be apparent to those skilled in the art that embodiments of this enhancements may apply to other types of memory cells. By way of example, FIG. 26 shows an embodiment of this improvement applied to memory cells without selection transistor. The memory plane MA1 'represented comprises memory cells C, 4 ,, each comprising a floating gate transistor TRIJ, respectively TRi_i j and a floating gate transistor TEIJ, respectively TEl_i j, of the same structure as those described above. The transistor TRIJ has a drain terminal connected to the RBLJ bit line, a control gate CGr connected to the word line WL1, and a source terminal connected directly to the source line SL.
[0033] The TEIJ transistor comprises a drain terminal connected to the EBLJ bit line, a CGr control gate connected to the word line WL1, and an unconnected source terminal. As previously, the floating gate FGr of the transistor TRIJ is electrically connected to the floating gate FGe of the transistor TEIJ and the latter comprises the conducting region IS facing its floating gate, allowing the memory cell to be erased. The memory cell Cl_i j has an identical structure and the description above applies by replacing the index i by the index i-1. Still other variants could be provided, for example by removing the source terminal of the TEIJ transistors, in the embodiment of FIG. 26 or in the embodiment of FIG. 2, or by omitting the selection transistors. ST associated with transistors TEIJ, j in the embodiment of FIG. 2. FIG. 27 is the electrical diagram of a memory plane MA2 comprising two memory cells Cij, Cij-pi according to a second improvement of the memory plane structure and 1, this improvement being implemented here in combination with the improvement previously described in relation with FIG. 2. The memory cells are accessible for reading and writing via a first bit line RBLJ, a second line RBLi-pi bit, third EBLJ bit line j + i, one WLI word line and two CGL11, CGL21 gate control lines. The memory cell CI belongs to a physical page P1 of the memory plane and the memory cell Ci_u belongs to an adjacent physical page P1_1. The pages PI, 1314 may include various other memory cells and the memory plane MA1 may comprise various other pages.
[0034] According to the preceding improvement, the memory cell Cij comprises two floating gate transistors TRIJ, TEIJ whose floating gates FGr, FGe are interconnected, the floating gate transistor TRIJ being dedicated to the reading of the transistor memory cell and the gate transistor floating TEIJ dedicated to erasing the memory cell. As before, the floating gates FGr, FGe can be formed by the same conductive element CFG and each transistor TRIJ, TEIJ comprises a conductive region IS which extends opposite its floating gate via a layer of tunnel oxide. The transistor TRIJ has a control gate CGr connected to the gate control line CGL11, a drain terminal D connected to the bit line RBLJ and a source terminal S connected to the drain terminal D of a transistor. ST selection whose source terminal S is connected to a source line SL. The floating gate transistor TEIJ has a control gate CGe connected to the gate control line CGL11, a drain terminal D connected to the EBLJ j-pi bit line and a source terminal S connected to the drain terminal 10 D of a selection transistor ST whose source terminal S is connected to a source line SL. The memory cell CI J + i has the same structure as the memory cell CI, and comprises two floating gate transistors TRIJ-pi, TE1J-pi whose floating gates FGr, FGe are interconnected or formed by the same conductive element CFG. The transistor TR1J + 1 has a control gate CGr connected to the gate control line CGL21, a drain terminal D connected to the bit line RBLJ + 1 and a source terminal S connected to the drain terminal D d a selection transistor ST whose source terminal S is connected to a source line SL. The floating gate transistor TE1J + 1 has a control gate CGe connected to the gate control line CGL11, a drain terminal D connected to the bit line EBLJJ-pi and a source terminal S connected to the gate terminal. drain D of a selection transistor ST whose source terminal S is connected to a source line SL. The selection transistors ST associated with the floating gate transistors TRIJ, TR1J + 1 have a buried vertical common control gate CSG. Likewise, the selection transistors ST associated with the floating gate transistors TEIJ, TE1J + 1 have a buried vertical common control gate CSG. The memory plane MA2 thus differs from the memory plane MAI of FIG. 2 in that the transistors TRIJ, TR1J + 1 of the twin memory cells CI J, J + i are connected to different bit lines RBLJ, RBLJ + 1. As will be seen below, these two bit lines make it possible to implement a method of reading memory cells that does not require, during the reading of a memory cell, to apply a negative voltage to the control line of the memory. CGL11 or CGL2 grid of the twin memory cell.
[0035] It will be noted in advance that the addition of an additional bit line in a structure of two twin memory cells as shown in FIG. 1 implies an increase in the area of the pair of memory cells, considered as "basic brick" of a memory plane, because it does not allow to receive two lines of bit. In contrast, a structure of two twin dual memory cells of the type shown in FIG. 2 makes it possible to provide the two bit lines RBLJ, RBLJ + 1 dedicated to reading the memory cells in addition to the bit line EBLJ j + i dedicated to erasing memory cells, without increasing their area. This will be shown in relation with FIGS. 28 to 32, which show a variant of the manufacturing method previously described of a pair of memory cells forming a "basic brick" of the memory plane, this variant making it possible to produce the two lines of RBLJ bit, RBLJ + 1 instead of the RBLJ bit line without increasing the area of the pair of memory cells. The initial steps of the manufacturing process, previously described in connection with FIGS. 15 to 21, are not modified. The step of FIG. 22, of making the bit lines RBLJ, EBLJ, is replaced by the step shown in FIG. 28. During the step of FIG. 28, a metal layer M1 ("metal 1" ) is deposited on the substrate and is etched to obtain on the one hand a conductive strip which here forms the EBLJ j-pi bit line (previously designated EBLJ) and on the other hand two conductive rectangles parallel to the EBLJ bit line. J + i, which form two sections of bit lines P1 (RBLJ), P1 (RBLJ + 1). These two sections of bit lines extend above the contacts C1 which have been made above the drain regions D of the transistors TR, J, TR, J + 1 during the step shown in FIG. .
[0036] In a step shown in Fig. 29, a dielectric layer D4 is deposited on the substrate, and orifices are made in the layer D4 above the sections of bit lines P1 (RBLJ), P1 (RBLJ + 1) and then metallized to form the C2 contacts.
[0037] During a step shown in FIG. 30, a metallic layer M2 ("metal 2") is deposited on the substrate and then etched to obtain two other conducting rectangles forming two sections of P2 bit lines (RBLJ), P2 (RBLi + i) perpendicular to the sections Pl (RBLJ), Pl (RBLi + i) and which extend above the contacts C2.
[0038] During a step shown in FIG. 31, a dielectric layer D5 is deposited on the substrate, and orifices are made in the layer D5 above the bit line sections P2 (RBLJ), P2 (RBLJ +1), then metallized to form the contacts C3.
[0039] In a step shown in Fig. 32, a metal layer M3 ("metal 3") is deposited on the substrate and is etched to obtain two conductive strips which form the RBLJ and RBLJ + 1 bit lines, the first being arranged on the contacts C3 connected to the drain region D of the transistor TR, J, and the second arranged on the contacts C3 connected to the drain region D of the transistor TR, J + 1. As shown in Fig. 32, the bit lines RBLJ and RBLJ + 1 are parallel to the bit line EBLJ j + i, which is two metal levels below them, and the set goes into the template cells C, J, C, J + i without requiring an increase in their area. The memory cells are programmed and erased as previously described. They are read by means of bit lines RBLJ, RBLJ + 1 in a manner to be described. Reading a memory cell via a bit line RBLi or RBLi-Ei A method of reading the memory cell C, J of FIG. 27 via the read transistor TR, j and the bit line RBLJ is described by FIG. Table 5 in the Annex. Fig. 33 is an electrical diagram similar to that of Fig. 27 in which the voltage values in Table 5 have been reported. The transistor TRij receives on its control gate the positive voltage CGV1i, here from 2 to 3V, and turns on if it is in the erased state. The selection transistor ST associated with the transistor TRij receives the positive selection voltage SV on its control gate, here 3V, and is in the on state. Since the bit line RBLi is raised to the positive voltage RBLV 1, here 1 V, and the source line SL is connected to the ground (0 V), the transistor TRij is traversed by a read current Ir flows from the bit line to the source line. The transistor TRI & + i of the twin memory cell receives the zero voltage CGV2i. This transistor, if it is in the erased state, may have a threshold voltage close to zero or even negative and switch into the on state. Indeed, the bit line RBLi-pi is not connected to any sense amplifier during the reading of the transistor TRij and the transistor TRI & + i is not read. Thus, even in the absence of a negative voltage on the gate control line CGL2i, the transistor TRI & + i can not corrupt the reading of transistor TRij, and vice versa when reading transistor TRi 3021803 FIG. 34 is the electrical diagram of a memory MEM2 comprising a memory plane MA2 according to the invention, only cells C, J + 1 being represented. The memory includes a CCT2 control circuit, a RD2 word line decoder, a CD2 column decoder, SA sense amplifiers, and BLT2 program locks. The word line decoder RD2 controls the voltages applied to the gate control lines CGL1 'CGL2, and to the word line WL, as a function of a most significant address A (n-1) -A (x) d a word (line address). The decoder CD2, in combination with the 10 locks BLT1, controls the voltages applied to the bit lines RBL, RBL, + 1, EBL, J +, as a function of the column address A (x-1) -A (0 ) the word. In read mode, the decoder CD2 connects the sense amplifiers SA to the RBL bit lines, connected to the memory cells to be read, and the sense amplifiers provide a read word DTR in the memory, for example 8 bits BO-B7. The circuit CCT2 comprises, as the circuit CCT1 previously described, a CPU, a VGEN voltage generator, and address and data registers. It executes read or write commands, controls decoders, provides the necessary voltages for read or write operations (erasure-programming), the supply of high and low-weight addresses, and if necessary executes a program for refreshing the memory cells. In one embodiment, the word line decoder RD2 is configured to separately control the voltages applied to the twin gate control lines CGL1 'CGL2' which here have the same high-order address A (n-1) -A (x). This separate control of the voltages can be reserved for the erase operations, to apply a positive voltage to these memory cells located on a twin page of that containing the memory cell (s) being erased (see Figs 6 and 7). ). In read mode, however, the decoder can apply the same voltage to the twin grid control lines CGL1 'CGL2, or even to all the grid control lines of the memory plane to limit logic gate switching and thus limit power consumption. of the memory, because the selection of memory cells for reading is ensured by means of word lines WL. In such an embodiment, the decoder RD2 receives, in addition to the most significant address A (n-1) -A (x) of a word, the least significant bit A (0) of the low-order address A (x-1) - A (0) of the word. The decoder RD2 also receives from the circuit CCT2 an information signal EPR which indicates to it whether the address decoding to be performed is involved in the context of a reading, a deletion or a programming of memory cells. If decoding occurs as part of an erasure, the decoder RD2 differentiates the two gate control lines CGL1 'CGL2, as a function of the bit A (0). In other words, the decoder RD2 selects the gate control line CGL1, if the bit line RBL, is designated by the complete address received by the memory, or selects the gate control line CGL2, if the RBL bit line, +, is designated by the full address received by the memory. In an equivalent variant, the decoder may receive a signal from the column decoder CD2, indicating which of the two grid control lines must be selected. 1 () One skilled in the art can naturally provide other embodiments of the decoder, for example to control separately the voltages applied to the CGL1 CGL2 twin grid control lines, reading, programming and erasure. Although the second improvement which has just been described was initially designed to be applied to a memory cell structure according to the first improvement, as shown in FIG. 2, it will be apparent to one skilled in the art that embodiments of this second improvement can be applied to other types of memory cells.
[0040] By way of example, FIG. 35 shows an example of application of the second improvement to the memory cell structure shown in FIG. 1, and shows a memory plane MA2 'comprising four memory cells D, J + 1, D, J + 1, D, J + 3 of the same structure and layout as the cells Mij, M, -, j, M, j + 1, j +, of FIG. 1, but considered as being part of the same logical page that is, having the same high-order address. The memory cells Dij, are thus twin memory cells and share the same selection grid CSG. The memory cells Di j + 2, Di j + 3 are twin memory cells and share the same CSG selection grid. The control gates of the floating gate transistors T, J, T, J + 2 of the memory cells Di j, Di j + 2 are connected to the gate control line CGL1, and the control gates 30 of the floating gate transistors T, J + 1, T, J + 3 memory cells D, J + 1, D, J + 3 are connected to the gate control line CGL2 ,. The drain terminals of the twin memory cells Dij, D, J + 1, instead of being connected to the same bit line, are connected to different bit lines BL ,, BL, + ,. Similarly, the drain terminals of the twin memory cells D, J + 2, D, J + 3, instead of being connected to the same bit line, are connected to different bit lines BL, + 2, BL + 3.
[0041] 3021803 27 APPENDIX Table 1: Fowler-Nordheim erasure of Cià via TE ,,,, Figs. 5 and 6 Ref. Description Sign Example GTC; Erase voltage applied to the CGL control grid; transistors TRjj, TEjj of the memory cell Cij (memory cell selected by deletion) via the gate control line CGL; negative -8V CGV.-1 Suppression inhibition voltage applied to the control gate CGLi_I transistors TRi_I j, TEi_I j of the memory cell Ci_j j (twin memory cell not selected in erasure) via the gate control line CGL; positive 3V EBLV Erase voltage applied to the EBL bit line; positive 6V RBLV Voltage applied to the RBL bit line; - HZ (*) SV Selection voltage applied to the word line WL._1 ,, common to the twin memory cells C 1 j, Cij - OV SPV Source line voltage applied to all SL source lines (or to the plane of source) - OV VB Electrical potential of PW-OV substrate CGV * Default voltage applied to all other 3V EBLV positive CGL grid control lines * Default voltage applied to non-selected EBL bit lines - OV RBLV * Voltage per fault applied to an unselected RBL bit line (eg "RBL1 + 1") - OV SV * Voltage applied to non-selected WL word lines - OV (*) High impedance, ie line disconnected from the rest of the circuit 5 3021803 28 Table 2: Fowler-Nordheim programming of Cu via TE ,,, Figs. 8 and 9 Ref. Description Sign Example GTC; Voltage applied to the CGL control grid; transistors 14V positive TRjj, TEm of the memory cell Cm (memory cell selected in programming) via the CGL grid control line; CGV.-1 Voltage applied to the CGL control gate; _ I transistors TRi_j j, TEi_j j of the memory cell Ci_i j (twin memory cell not selected in programming) via the gate control line CGL; positive 3V EBLV Voltage applied to the EBL bit line; - OV RBLV Voltage applied to the RBL bit line; - HZ SV Selection voltage applied to word line WL.4 ,, common to twin memory cells C 1 j, Cm - OV SPV Source line voltage applied to all SL source lines (or source plane) - OV VB PW-OV substrate electrical potential CGV * Voltage applied to all other 3L EBLV positive CGL grid control lines * Voltage applied to non-selected EBL bit lines - 6V RBLV * Voltage applied to unselected RBL bit lines - HZ SV * Voltage applied to unselected WL word lines - OV 3021803 29 Table 3: Programming of Cià by injection, via TRià, Figs. 11 and 12 Ref. Description Sign Example GTC; Voltage applied to the CGL control grid; positive transistors 10V TRjj, TEm of the memory cell Cm (memory cell selected in programming) via the gate control line CGL; CGV.-1 Voltage applied to the CGL control gate; _ I transistors TRi_j j, TEi_j j of the memory cell Ci_j j (twin memory cell not selected in programming) via the gate control line CGL; - OV EBLV Voltage applied to bit line EBL - OV or higher RBLV Voltage applied to RBL bit line; positive 4V SV Selection voltage applied to the word line WLi_I j common to the twin memory cells C jj, Cm positive 1-2V SPV Source line voltage applied to all source lines SL (or source plane) - OV VB Electrical potential of PW-OV substrate CGV * Voltage applied to all other gate control lines CGL - OV EBLV * Voltage applied to unselected EBL bit lines - OV RBLV * Voltage applied to unselected RBL bit lines - OV SV * Voltage applied to unselected WL word lines - OV 3021803 Table 4: Reading Ci via TRià, Fig. 13 and 14 Ref. Description Sign Example GTC; Voltage applied to the CGL control grid; transistors TRjj, TEjj of the memory cell Ci j (memory cell selected for reading) via the gate control line CGL; positive 2-3V CGVi_I Voltage applied to the control gate CGLi_I of the transistors TRi_j j, TEi_j j of the memory cell Cil_ j (twin memory cell not selected in reading) via the gate control line CGL; negative -2V EBLV Voltage applied to the EBL bit line; - OV RBLV Voltage applied to the RBL bit line; positive 1V SV Selection voltage applied to the word line WLi_i j common to twin memory cells Cil_ 1, positive Cm 3V SPV Source line voltage applied to all source lines SL (or source plane) - OV VB Potential voltage PW - OV CGV * Voltage applied to all other gate control lines CGL - OV EBLV * Voltage applied to unselected EBL bit lines - OV RBLV * Voltage applied to unselected RBL bit lines - OV SV * Voltage applied to unselected WL word lines - OV 3021803 31 Table 5: Cià reading via TRià and RBLi, figure 33 Ref. Description Sign Example CGV1i Voltage applied to the CGL1 control gate of the positive 2-3V transistors TRjj, TEjj of the memory cell Ci j (memory cell selected for reading) via the gate control line CGLli CGV2i Voltage applied to the control gate CGL2 of the OV transistors TR41 + 1, TEjj + 1 of the sky memory cell (memory cell not selected for reading) via the grid control line CGL2i EBLV Voltage applied to the bit line EBL; ; +1 - OV RBLV, Voltage applied to the RBL bit line, selected from the 1V positive memory cell pair RBLVj + 1 Voltage applied to the unselected RBLj + bit line of the OV memory cell pair SV Applied selection voltage at the word line WL; common to twin memory cells Ci1 + 1, positive Cile 3V SPV Source line voltage applied to all source lines SL (or source plane) - OV VB Electrical potential of PW-OV substrate CGV * Voltage applied to all other gate control lines CGL - OV EBLV * Voltage applied to non-selected EBL bit lines - OV RBLV * Voltage applied to non-selected RBL bit lines - OV SV * Voltage applied to non-selected WL word lines - OV

权利要求:
Claims (13)
[0001]
REVENDICATIONS1. Non-volatile memory (MEM2, MA2) on a semiconductor substrate (PW), comprising: a first memory cell (C, J) having a first floating gate transistor (TR, J) electrically connected to a selection transistor (ST) having a buried vertical control gate (CSG), the selection transistor having a vertical channel region (CH2) extending opposite a first face of the buried vertical control gate 1 (), a second memory cell ( C, J + 1) comprising a first floating gate transistor (TR, J + 1) electrically connected to a selection transistor (ST) having the same control gate (CSG) as the selection transistor of the first memory cell ( C, J) and having a vertical channel region (CH2 ') extending opposite a second face of the buried vertical control gate, and facing the channel region (CH1) of the selection transistor of the first memory cell (Ci, j), characterized e in that it comprises: - a first bit line (RBLi) electrically connected to the first floating gate transistor (TR, J) of the first memory cell (C, J), and 20 - a second bit line ( RBLi +,) electrically connected to the first floating gate transistor (TR, j + i) of the second memory cell (Ci, i + i).
[0002]
The memory of claim 1, comprising: - a first gate control line (CGL1,) electrically connected to a control gate (CGr, CG) of the first floating gate transistor (TR, J) of the first cell memory (C, J), - a second gate control line (CGL2,) electrically connected to a control gate (CGr, CG) of the first floating gate transistor (TR, +1) of the second memory cell (C j + 1), and means (CCT2, RD2, BLT2, CD2, SA) for reading the memory cells, configured for: - during the reading of a memory cell, applying a positive selection voltage (CGV) to the gate control line (CGL1 CGL2,) connected to the memory cell to be read, and read the first memory cell via the first bit line or read the second memory cell via the second bit line. 3021803 33
[0003]
3. Memory according to claim 2, wherein the reading means of the two memory cells comprise a word line decoder (RD2) configured to: - assign to the two grid control lines (CGL1 CGL2,) the same line address 5 (A (n-1) -A (x)), and - during the performance of at least one operation selected from reading, programming or erasing operations of one of the memory cells, select the line gate control device (CGL1 'CGL2,) to which the memory cell is connected according to a part of the line address (A (n-1) -A (x)) of the two grid control lines and on the other hand, the least significant bit (A (0)) of a column address (A (x-1) -A (0)) of the memory cell.
[0004]
4. Memory according to one of claims 1 to 3, wherein each memory cell further comprises a second floating gate transistor (TE ,,, TE, j + i) whose floating gate (FGe) is electrically connected to the floating gate (FGr) of the first floating gate transistor, and comprising a conductive region (IS, n1) electrically connected to the conduction terminal (D) of the second floating gate transistor which extends opposite its floating gate ( FGe) via a tunnel dielectric layer (D1).
[0005]
The memory according to claim 4, comprising a third bit line (EBLi, i +,) electrically connected both to the second floating gate transistor (TE, J) of the first memory cell (C, J) and to the second transistor floating gate (TE, j + i) of the twin memory cell (Ci, i + i).
[0006]
6. Memory according to one of claims 4 and 5, wherein the floating gates (FGr, FGe) of the first and second floating gate transistors are formed by a single piece (CFG) of a conductive material.
[0007]
7. Memory according to one of claims 4 to 6, wherein the conductive region (IS) is a doped region (n1) of the substrate (PW).
[0008]
8. Memory according to one of claims 1 to 7, wherein the common control gate of the selection transistors of the twin memory cells is a vertical grid (CSG) buried in the substrate. 3021803 34
[0009]
9. Integrated circuit on a semiconductor chip, comprising a non-volatile memory (MEM2) according to one of claims 1 to 8.
[0010]
10. A method of manufacturing a memory according to one of claims 1 to 8, comprising the steps of: - forming in the substrate (PW) insulating trenches (STI) delimiting at least a first (S1) band of substrate, - forming in the substrate a conductive trench (P0, CGC) arranged transversely to the substrate strip, to form the control gate common to the selection transistors 10 of the two twin memory cells, - forming on the substrate a conductive floating gate (FG, P1) arranged transversely to the substrate strip, with the interposition of a first dielectric layer (D1), - forming a control gate (CG, CGL,) on the floating gate, with the interposition of a second dielectric layer (D2), to obtain a stack of grids, - doping the substrate strip (Si) on each side of the stack of grids, - forming in a first metal level (M1) two first sections (Pl (RBL, ), 131 (RBL, +,)) from bit lines each electrically connected to a drain region of one of the floating gate transistors; forming in a second metal level (M2) two second sections (P2 (RBL,), P2 (RBL, +,) ) bit lines each electrically connected to one of the first bit line sections, and forming in a third metal level (M3) two bit lines (RBL ,, RBL, +,) electrically connected to each other. one of the second sections of bit lines of the second level of metal. 25
[0011]
11. The manufacturing method according to claim 10, comprising a step of doping the second substrate strip (S2) before forming the stack of grids (FG, CG) and before doping the two substrate strips (S1, S2). ) on each side of the gate stack, to form a conductive region (IS, n1) facing the floating gate of the second floating gate transistor (FG).
[0012]
12. A method of reading / writing a memory cell from a pair of twin memory cells C ,,, +,) comprising: - a first memory cell (C 1, C) including a first floating gate transistor 35 (TR, J) electrically connected to a selection transistor (ST) having a buried vertical control gate (CSG), the selection transistor of the first memory cell 3021803 having a vertical channel region (CH2) extending next to a first face of the buried vertical control gate, - a second memory cell (C, J + 1) comprising a first floating gate transistor (TR, J + 1) electrically connected to a selection transistor (ST) having the same control gate (CSG) that the selection transistor of the first memory cell (C ,,,) and having a vertical channel region (CH2 ') which extends opposite a second face of the control gate vertical buried, and in front of the region of ca nal (CH1) of the selection transistor of the first memory cell (Ci, j), characterized in that it comprises the steps of: - providing a first bit line (RBL) electrically connected to the first transistor floating gate (TR, J) of the first memory cell (C, J); - providing a second bit line (RBL, +,) electrically connected to the first floating gate transistor (TR, j + i) of the second cell memory (C, J +,), and - during reading of a memory cell: - applying a positive selection voltage (CGV) to the gate control line (CGL1, CGL2,) connected to the memory cell to be read, and read the first memory cell through the first bit line or read the second memory cell through the second bit line. 20
[0013]
The method according to claim 12, comprising the steps of: - assigning to the two grid control lines (CGL1 CGL2,) the same line address (A (n-1) -A (x)), - during the performing at least one operation selected from reading, programming or erasing operations of one of the memory cells, selecting the gate control line (CGL1 'CGL2,) to which the memory cell is connected in function on the one hand of the line address (A (n-1) -A (x)) of the two grid control lines and on the other hand of the least significant bit (A (0)) of a column address (A (x-1) -A (0)) of the memory cell. 30

类似技术:

公开号 | 公开日 | 专利标题

FR3021803A1|2015-12-04|MEMORY CELLS TWINLY ACCESSIBLE INDIVIDUALLY READING

FR3021804A1|2015-12-04|DUAL NON-VOLATILE MEMORY CELL COMPRISING AN ERASING TRANSISTOR

EP1495496B1|2009-11-18|Non-volatile, programmable, electrically erasable memory semiconductor device having a single grid material layer and corresponding magnetic core plane

EP1014447A1|2000-06-28|One-time-programmable memory cell manufactured in CMOS tecnology

FR3025353A1|2016-03-04|NON-VOLATILE COMPOSITE MEMORY WITH ERASE BY PAGE OR BY WORD

FR2996680A1|2014-04-11|NON-VOLATILE MEMORY HAVING VERTICAL SELECTION TRANSISTORS

FR2692720A1|1993-12-24|EPROM device with single layer of polycrystalline silicon with fast erasure.

FR2987696A1|2013-09-06|METHOD FOR READING WRITTEN NON-VOLATILE MEMORY CELLS

FR3021806A1|2015-12-04|METHOD FOR PROGRAMMING NONVOLATILE MEMORY CELL COMPRISING A SHARED SELECTION TRANSISTOR GRID

FR3030883A1|2016-06-24|VERTICAL SELECTION GRID MEMORY CELL FORMED IN A FDSOI TYPE SUBSTRATE

FR2987700A1|2013-09-06|NON-VOLATILE MEMORY COMPRISING MINI FLOATING POTENTIAL HOUSINGS

EP0896370B1|2012-04-11|Floating-gate memory device on SOI and corresponding manufacturing process

FR2871940A1|2005-12-23|TRANSISTOR MOS WITH FLOATING GRID, WITH DOUBLE CONTROL GRID

EP1863035B1|2011-07-27|EEPROM memory with an improved resistor against transistors breakdown

EP0543703A1|1993-05-26|Single-transistor electrically programmable integrated memory

FR3036221A1|2016-11-18|INTERCONNECTION STRUCTURE OF BINARY MEMORY CELLS

EP0750313A1|1996-12-27|EEPROM memory, programmed and erased by Fowler-Nordheim effect

EP2977988B1|2019-06-19|Non-volatile memory with programmable resistance

FR3054723A1|2018-02-02|COMPACT EEPROM CELL-MEMORY WITH REDUCED TUNNEL INJECTION AREA

FR2776820A1|1999-10-01|Electrically erasable floating grid memory circuit

FR3059458A1|2018-06-01|COMPACTED NON-VOLATILE MEMORY DEVICE OF THE CHARGE TRAP TYPE IN A DIELECTRIC INTERFACE

FR2616953A1|1988-12-23|MEMORY EPROM ERASABLE BY PULSES

FR3070537A1|2019-03-01|NON-VOLATILE MEMORY WITH RESTRICTED SIZE

FR2769747A1|1999-04-16|Hot carrier programmable memory used in nonvolatile and programmable integrated circuit memories

JP2006236424A|2006-09-07|Nonvolatile memory device, and its electron charges injection method

同族专利:

公开号 | 公开日

FR3021803B1|2017-10-13|

US20150348981A1|2015-12-03|

CN204966056U|2016-01-13|

CN105280229A|2016-01-27|

CN110265077A|2019-09-20|

US9653470B2|2017-05-16|

CN105280229B|2019-07-16|

引用文献:

公开号 | 申请日 | 公开日 | 申请人 | 专利标题

US6433382B1|1995-04-06|2002-08-13|Motorola, Inc.|Split-gate vertically oriented EEPROM device and process|

US20020153546A1|2001-04-20|2002-10-24|Koninklijke Philips Electronics N.V.|Two-transistor flash cell|

FR2844090A1|2002-08-27|2004-03-05|St Microelectronics Sa|MEMORY CELL FOR QUICK READ NONVOLATILE REGISTER|

US20050275002A1|2004-05-27|2005-12-15|Skymedi Corporation|Vertical split gate memory cell and manufacturing method thereof|US9627068B2|2015-05-11|2017-04-18|StmicroelectronicsSas|Twin memory cell interconnection structure|

FR3049380A1|2016-03-22|2017-09-29|StmicroelectronicsSas|IMPROVING PERFORMANCE IN READING A NON-VOLATILE MEMORY DEVICE, PARTICULARLY A NON-VOLATILE MEMORY DEVICE WITH BITTER SELECTION TRANSISTOR|US5182725A|1987-11-20|1993-01-26|Mitsubishi Denki Kabushiki Kaisha|Nonvolatile semiconductor memory device with reduced variation in source potential of floating gate type memory transistor and operating method therefor|

US6788576B2|2002-10-28|2004-09-07|Tower Semiconductor Ltd.|Complementary non-volatile memory cell|

US6828618B2|2002-10-30|2004-12-07|Freescale Semiconductor, Inc.|Split-gate thin-film storage NVM cell|

US6894339B2|2003-01-02|2005-05-17|Actrans System Inc.|Flash memory with trench select gate and fabrication process|

US7358134B2|2003-09-15|2008-04-15|Powerchip Semiconductor Corp.|Split gate flash memory cell and manufacturing method thereof|

US8139408B2|2006-09-05|2012-03-20|Semiconductor Components Industries, L.L.C.|Scalable electrically eraseable and programmable memory|

KR100752192B1|2006-09-06|2007-08-27|동부일렉트로닉스 주식회사|Single-poly structure of flash memory device and manufacturing method thereof|

US7696044B2|2006-09-19|2010-04-13|Sandisk Corporation|Method of making an array of non-volatile memory cells with floating gates formed of spacers in substrate trenches|

US7723774B2|2007-07-10|2010-05-25|Silicon Storage Technology, Inc.|Non-diffusion junction split-gate nonvolatile memory cells and arrays, methods of programming, erasing, and reading thereof, and methods of manufacture|

US7800159B2|2007-10-24|2010-09-21|Silicon Storage Technology, Inc.|Array of contactless non-volatile memory cells|

US8072811B2|2008-05-07|2011-12-06|Aplus Flash Technology, Inc,|NAND based NMOS NOR flash memory cell, a NAND based NMOS NOR flash memory array, and a method of forming a NAND based NMOS NOR flash memory array|

JP5193830B2|2008-12-03|2013-05-08|株式会社東芝|Nonvolatile semiconductor memory|

US8203187B2|2009-03-03|2012-06-19|Macronix International Co., Ltd.|3D memory array arranged for FN tunneling program and erase|

KR20100115612A|2009-04-20|2010-10-28|삼성전자주식회사|Non-volatile semiconductor memory device capable of preventing program disturb and programing method thereof|

US8355287B2|2009-08-25|2013-01-15|Aplus Flash Technology, Inc.|Method and apparatus for operation of a NAND-like dual charge retaining transistor NOR flash memory device|

EP2393115A1|2010-06-03|2011-12-07|Nxp B.V.|Memory cell|

US8958245B2|2010-06-17|2015-02-17|Ememory Technology Inc.|Logic-based multiple time programming memory cell compatible with generic CMOS processes|

US8582363B2|2011-01-31|2013-11-12|Aplus Flash Technology, Inc.|Method and apparatus for management of over-erasure in NAND-based NOR-type flash memory|

WO2013079020A1|2011-12-02|2013-06-06|Tsinghua University|Nor flash memory array structure, mixed nonvolatile flash memory and memory system comprising the same|

FR2987696B1|2012-03-05|2014-11-21|St Microelectronics Rousset|METHOD FOR READING WRITTEN NON-VOLATILE MEMORY CELLS|

US8901634B2|2012-03-05|2014-12-02|Stmicroelectronics Sas|Nonvolatile memory cells with a vertical selection gate of variable depth|

JP5972700B2|2012-07-31|2016-08-17|ルネサスエレクトロニクス株式会社|Memory device|

US20140198583A1|2013-01-17|2014-07-17|Infineon Technologies Ag|Method and System for Reducing the Size of Nonvolatile Memories|

TW201508753A|2013-08-29|2015-03-01|Chrong-Jung Lin|Memory cell, memory array and operation method thereof|

FR3021804B1|2014-05-28|2017-09-01|Stmicroelectronics Sas|DUAL NON-VOLATILE MEMORY CELL COMPRISING AN ERASING TRANSISTOR|

FR3021803B1|2014-05-28|2017-10-13|Stmicroelectronics Sas|MEMORY CELLS TWINLY ACCESSIBLE INDIVIDUALLY READING|FR3021803B1|2014-05-28|2017-10-13|StmicroelectronicsSas|MEMORY CELLS TWINLY ACCESSIBLE INDIVIDUALLY READING|

FR3021804B1|2014-05-28|2017-09-01|StmicroelectronicsSas|DUAL NON-VOLATILE MEMORY CELL COMPRISING AN ERASING TRANSISTOR|

FR3025353B1|2014-09-03|2016-09-09|StmicroelectronicsSas|NON-VOLATILE COMPOSITE MEMORY WITH ERASE BY PAGE OR BY WORD|

JP6739327B2|2016-12-27|2020-08-12|ルネサスエレクトロニクス株式会社|Semiconductor device|

US11011533B2|2018-01-10|2021-05-18|Ememory Technology Inc.|Memory structure and programing and reading methods thereof|

JP2019179799A|2018-03-30|2019-10-17|ルネサスエレクトロニクス株式会社|Semiconductor storage device|

法律状态:
2015-04-22| PLFP| Fee payment|Year of fee payment: 2 |

2015-12-04| PLSC| Publication of the preliminary search report|Effective date: 20151204 |

2016-04-22| PLFP| Fee payment|Year of fee payment: 3 |

2017-04-21| PLFP| Fee payment|Year of fee payment: 4 |

2018-04-23| PLFP| Fee payment|Year of fee payment: 5 |

2019-04-19| PLFP| Fee payment|Year of fee payment: 6 |

2021-02-12| ST| Notification of lapse|Effective date: 20210105 |

优先权:

申请号 | 申请日 | 专利标题

FR1454893A|FR3021803B1|2014-05-28|2014-05-28|MEMORY CELLS TWINLY ACCESSIBLE INDIVIDUALLY READING|FR1454893A| FR3021803B1|2014-05-28|2014-05-28|MEMORY CELLS TWINLY ACCESSIBLE INDIVIDUALLY READING|

US14/671,606| US9653470B2|2014-05-28|2015-03-27|Individually read-accessible twin memory cells|

CN201520251293.0U| CN204966056U|2014-05-28|2015-04-23|Nonvolatile memory and integrated circuit on semiconductor chip|

CN201510196923.3A| CN105280229B|2014-05-28|2015-04-23|Individually read out addressable pairing memory cell|

CN201910456430.7A| CN110265077A|2014-05-28|2015-04-23|Individually read out addressable pairing memory cell|

[返回顶部]