The present invention relates to an electronic memory device comprising high density non-volatile memory cells with floating gates, and with a reduced capacitive coupling cell-to-cell interference.
More specifically, the non-volatile memory cells are integrated on a semiconductor substrate, and comprise a plurality of cells organized in a matrix of rows and columns. The memory cells may be of the Flash NAND type, and the following description is made with reference to this field of application by way of illustration only.
Non-volatile electronic memory devices are widely used in several apparatuses, such as cell phones, digital cameras, notebooks, etc., where storing a considerable amount of information in a compact memory with high capacity is required.
Semiconductor integrated non-volatile memory devices have been developed, which comprise a plurality of memory cells organized in a matrix, i.e., the cells are organized in rows called word lines WL and columns called bit lines BL, as shown in
Each cell comprises a floating gate transistor with a floating gate region FG above a channel region CH or tunnel oxide formed in the semiconductor substrate 1. The floating gate regions FG are separated from one another along the word lines, as highlighted in
A control gate region CG is then capacitively coupled to the floating gate region by an interposition of more dielectric layers overlapped to form a layer of the ONO (oxide/nitride/oxide), as shown in
Each cell has drain, source and body regions. In a matrix-like organization, via metal electrodes, only the drain of the select transistor is contacted near the last cell of the bit line, i.e., the DSL and the source of the transistor near the first cell of the bit line called SSL. The control gate regions are contacted so as to apply predetermined voltage values to the memory cells.
The charge stored in the floating gate region FG determines the logic state of the cell modifying its threshold voltage. The main feature of the memory cell is that of having two states, one with a low or erased threshold voltage, which corresponds to an erased cell, and one with a high or programmed threshold voltage which corresponds to a written cell. Different conventions can be provided. The voltage is externally applied through the control gate region CG, but the electrode effectively controlling the state of the channel is the floating gate region FG.
The need of storing, in a more reduced space, a greater amount of data has led to the development of devices with multilevel memory cells, wherein different logic states are stored in each cell.
For devices with multilevel memory cells, and in particular with cells of the Flash NAND type, the accurate control of the threshold distributions of the cells becomes fundamental so that they are correctly placed below the erase verify potential, for the erased cells, and between the program verify values and the Vpass programmed values for the written cells. The Vpass voltages are the threshold values relative to the voltages associated with the different logic states stored, and their value depends on the number of bits that can be stored in the same cell.
During the programming of a Flash NAND cell, a lot of attention has been paid to the presence of the effective disturbances which could vary the voltage stored in the cell and cause a variation of the voltages stored in the adjacent cells. This is a phenomenon called “widening of the distributions”.
In non-volatile memory devices the undesired capacitive cell-to-cell couplings are among the main contributors for the “widening of the distributions”. In multilevel devices the “widening of the distributions” is more damaging since the thresholds relative to the different values or logic states are separated within narrow voltage intervals.
In a memory of the Flash NAND type, with a high density, the main parasitic capacitive components are identified, with reference to
1. the direction Y of the bit lines;
2. the direction X of the word lines; and
3. direction XY diagonal couplings.
The three parasitic capacitances identified show different values from each other and, typically in the order indicated, they show decreasing values.
Problems of modification of the voltages in the cells and thus of “widening of the distributions” show up when the cells themselves are programmed. In fact, with particular reference to
From some experimental data detected and reported in
In the example shown in
On the same block of cells, moreover, two different measurements of the parasitic capacitances have been detected further to two different formulations and reported in
For the first detection the block of cells had been erased with an erase verify value of about 13V in a time t=1.5 ms, while for the second detection the erase verify erase value was of about 18V in a time t=1.5 ms. For both the detections a programming sequence of the cells has been followed corresponding to the increasing numbers indicated from 1 to 5 in
In
This confirms how the value of the parasitic capacitances significantly influences the threshold voltages of the single cells, and thus the correctness of the programmable and programmed values.
Several approaches are known to reduce the capacitances of type 2 along the direction X of the word lines in the devices of non-volatile memory cells.
A straightforward and efficient approach, for example, is shown in European Patent Application No. 02425805.5, which discloses filling in the space between two adjacent floating gate regions of cells belonging to the same word lines with a material having a low dielectric constant, as shown in
A further known approach is shown in U.S. Patent Application No. 2004/0012998 to Chien et al., which discloses reducing the capacitive effects of the floating gate regions along the direction of the word lines due to a particular conformation of the word lines themselves in correspondence with an insulating layer or gate oxide STI, interposed between two adjacent floating gate regions FG. This approach, shown in
All these approaches reduce the parasitic capacitances in the direction X. However, they do not teach how to reduce the parasitic capacitances along the axis Y.
The parasitic capacitive coupling which is realized along the axis Y is due to the fact that between the word lines an insulating layer is interposed and that two opposite floating gate regions FG, arranged on two consecutive word lines, realize two plates of one capacitor, as highlighted in
This phenomenon has increased along with the increase in the density of the memory cells on the same device, and thus along with the decrease of the distance between the floating gates of cells placed on opposite word lines.
To limit the couplings in the direction Y techniques can be used by varying the geometric dimensions of the floating gate regions. For example, a technique reduces the height of the floating gate regions which reduces the opposite surface.
These approaches, however, show some drawbacks. In fact, the designation of the dimensions of the floating gate regions influences the programming voltage value and the threshold values which are, especially in the case of a multilevel memory cell device, important values for operation of the device itself. It is thus necessary not to excessively reduce the dimensions of the floating gate regions.
A further known approach to reduce the parasitic capacitances between the floating gate regions uses programming techniques of the non-volatile memory devices with ramp voltages. The main drawback of this approach is that high programming voltages are required.
A further known approach avoids the interposition, between the word lines, of materials with high dielectric constants, such as for example, silicon nitride, or reducing the amount of interposed dielectric.
Generally, as highlighted in
To reduce the amount of dielectric material interposed between the word lines, a known approach increases the length of the spacers, as highlighted in
This for example, however, shows the drawback that the spacers influence, in a fundamental way, the circuitry of the device. It is thus necessary to take into account the final length of the spacers in relation with the breakdown feature of the present NMOS and PMOS transistors.
An object of the present invention is to provide a non-volatile memory cell device of the Flash NAND type, organized in a matrix, and having structural and functional characteristics that allow a reduction in the capacitive coupling between opposite floating gate regions along consecutive word lines, thus overcoming the drawbacks of the devices formed according to the prior art.
This and other objects, advantages and features in accordance with the present invention are provided by a biased or floating conductive shield interposed between the rows or word lines of the matrix.
An electronic memory device may comprise a semiconductor substrate, and a plurality of non-volatile memory cells integrated on the semiconductor substrate. The memory cells may be organized in a matrix of rows and columns. Each memory cell may comprise at least one floating gate transistor comprising a floating gate projecting from the semiconductor substrate, and a control gate capacitively coupled to the floating gate. A plurality of word lines may be connected to the rows of the matrix, and a plurality of bit lines may be connected to the columns of the matrix. A coating comprising at least one floating conductive layer may be between memory cells of opposite word lines.
Another aspect is directed to a method for making an electronic memory device as defined above.
The characteristics and the advantages of the device according to the invention will be apparent from the following description of an embodiment thereof given by way of a non-limiting example with reference to the annexed drawings. In these drawings:
With reference to
The matrix comprises a plurality of rows or word lines 4 arranged as shown in
The device 1 comprises, as highlighted in
Each cell 2 comprises a transistor of the MOS type with a floating gate and has, along the direction X, as shown in
A control gate region 12 is capacitively coupled to the floating gate region 9, as highlighted in particular in
Each cell 2 also comprises a floating gate transistor, with an architecture of the NAND type, which has a body region not identified in the figures, and a source-drain junction 8 realized in correspondence with the direction Y of
Advantageously, according to the invention, a coating 15 is present above the structures 25, the control gates 12, the floating gate regions 9 and the source-drain junctions 8 in the direction Y of the bit lines 5 realized starting from the substrate 3. This coating 15 comprises at least one conductive layer 16 which can be floating or suitably biased.
Advantageously, according to the present embodiment, the coating 15 also comprises a thin insulating layer 17 interposed between the upper surface of the substrate 3 and the conductive layer 16 which serves as a shield between the two opposite structures 25 arranged in two consecutive word lines 4, as shown in
The thin insulating layer 17 can for example have a thickness of about 100 Angstroms. This value can vary according to the technology and to the techniques used. The thin insulating layer 17 laterally covers each structure 25 along the bit lines 5 allowing it to insulate in particular each floating gate region 9 laterally in the direction Y.
Moreover, the conductive layer 16 arranged above the thin insulating layer 17, suitably floating or biased, interposed between the opposite floating gate regions 9, along the direction Y, allows it to minimize or eliminate the capacitive couplings along the bit lines 5. The conductive layer 16 can be for example, but not exclusively, a polysilicon layer such as titanium, or a suitably doped polysilicon, or a refractory metal being thermally compatible with the process and the technology suitable to realize the device 1.
The thin insulating layer 17 can be a silicon nitride, a silicon oxide or other insulating material compatible with the realization process and the technology used. Moreover, the thin insulating layer 17 can be obtained by a deposition or growing process.
Thus, advantageously, the coating 15 creates a conductive shield which allows a capacitive partition of the parasitic capacitances to be formed, which could be formed between the opposite floating gate regions 9 arranged on consecutive word lines 4. This allows the total capacity in the direction Y to be significantly reduced.
To realize the suitable circuitry elements for example, as shown in
The coating 15 is realized on the matrix, and in particular, preferably on the word lines 4 along the bit lines 5 and on part of the select transistors 20 along the direction Y except from the contact points which, as known in non-volatile memory devices 1 organized in a matrix, are arranged in correspondence with the peripheral edge 21 as shown in
A non-volatile memory device 1 organized in a matrix of cells, has word lines and bit lines connected to the cells, has a plurality of structures 25 described above and of the type comprising floating gate regions 9, realized above a channel region 10. Each structure 25 comprises a control gate region 12 capacitively coupled to the floating gate region 9 through an overlapping of one or more dielectric layers 13, such as ONO (oxide/nitride/oxide), for example. Each structure 25 comprises, above the control gate region 12, a silicide layer 11.
Laterally, each structure 25 comprises a plurality of spacers 22 which are arranged parallel to the direction X of the word lines. The direction X in
Advantageously, a coating 15 laterally covers along the direction Y, and possibly in the upper part of each structure 25. This coating 15 comprises a conductive layer 16, which occupies all the space interposed between the spacers 22 of the opposite structures 25 along the direction Y, and placed on consecutive word lines.
Advantageously, between the opposite floating gate regions 9 along the direction Y, the conductive layer 16 is interposed between insulating spacers 22. In particular, in this embodiment the conductive layer 16 of the coating 15 creates, together with the spacers 22, a conductive shield which allows a partition of the capacitances to be formed, thus reducing the total parasitic capacitance in the direction Y.
To complete the device 1 by a known technique, the conductive layer 16 together with the circuitry structure are planarized by a CMP process which allows the device 1 to be compatible with the successive process steps.
A further embodiment is shown in
The device 1 along the direction Y comprises a plurality of structures 25, of the above described type, each one comprising a floating gate region 9 capacitively coupled to an upper control gate 12 by one or more dielectric layers 13 and a silicide layer 11 arranged on top.
Each cell 2 also comprises a floating gate transistor, with an architecture of the NAND type, which has a body region, not identified in the figures, and a source-drain junction 8 realized in correspondence with the direction Y interposed between a structure 25 and the successive one.
Advantageously, along the direction Y a continuous coating 15 is present, arranged on the matrix above the substrate 3 at least along the bit lines and on part of the selectors realized for the contact, as shown in
The coating 15 comprises at least one continuous conductive layer 16 which covers, in a shaped way, along the direction Y, the plurality of structures 25 and the interposed source-drain junctions 8. The shaped conductive layer 16 laterally covers, along the direction Y, the plurality of structures 25, and in particular the floating gates 9.
The conductive layer 16 can be for example a titanium layer, but it could be realized also with different conductive materials according to the technologies used. Suitably, the coating 15 also comprises at least one thin insulating layer 17 interposed between the upper surface of the substrate 3 and the shaped conductive layer 16 to avoid possible contacts.
According to this embodiment, a suitable continuous spacer layer 26 is present above the coating 15. In particular, the spacer layer 26 suitably covers the gap between each structure 25 and the successive one along the direction Y above the source-drain junctions 8, and possibly suitably covering the structures 25.
The coating 15 defines, also in this embodiment, a conductive shield between the opposite floating gate regions 9 and thus realizes a physical separation between the floating gate regions reducing the parasitic capacitances in the direction Y along the bit lines.
In the presence of select transistors 20 it is possible, by a removal process, to remove the conductive layer 16 for realizing the suitable circuit elements, as shown in
The non-volatile memory cell device realized by way of the present invention has a coating 15 or conductive or floating shield which allows the parasitic capacitances to be limited which would have to be created between the floating gate regions 9 arranged on consecutive word lines 4 along the direction of the bit lines 5, i.e., along the direction Y. It also allows a reduction in the interaction and disturbance problems both in the storing and erasing or reading step of the values inserted in the cells 2.
Number | Date | Country | Kind |
---|---|---|---|
MI2004A2373 | Dec 2004 | IT | national |
Number | Name | Date | Kind |
---|---|---|---|
5455793 | Amin et al. | Oct 1995 | A |
20030201487 | Yang et al. | Oct 2003 | A1 |
20030210582 | Kinoshita | Nov 2003 | A1 |
20040232496 | Chen et al. | Nov 2004 | A1 |
Number | Date | Country |
---|---|---|
1435657 | Jul 2004 | EP |
Number | Date | Country | |
---|---|---|---|
20060158931 A1 | Jul 2006 | US |