Nrom memory cell, memory array, related devices an methods

Description of equivalent WO 2004001802 A2

NROM Memory Cell, Memory Array, Related Devices and Methods

TECHNICAL FIELD

This invention relates to a NROM memory cells, arrays of such memory cells, electronic devices employing such memory cells and anays, and methods related to such memory cells.

BACKGROUND OF THE INVENTION

Various types of memory devices are used in electronic systems. Some types of memory device, such as DRAM (dynamic random access memory) provide large amounts of readable and writable data storage with modest power budget and in favorably small form factor, but are not as fast as other types of memory devices and provide volatile data storage capability. Volatile data storage means that the memory must be continuously powered in order to retain data, and the stored data are lost when the power is interrupted. Nonvolatile memories are capable of retaining data without requiring electrical power.

Other types of memory can provide read-only or read- write capabilities and nonvolatile data storage, but are much slower in operation. These include CD-ROM devices, CD-WORM devices, magnetic data storage devices (hard discs, floppy discs, tapes and so forth), magneto-optical devices and the like. Still other types of memory provide very high speed operation but also demand high power budgets. Static RAM or SRAM is an example of such memory devices.

In most computer systems, different memory types are blended to selectively gain the benefits that each technology can offer. For example, read-only memories or ROM, EEPROM and the like are typically used to store limited amounts of relatively infrequently-accessed data such as a basic input-output system. These memories are employed to store data that, in response to a power ON situation, configure a processor to be able to load larger amounts of software such as an operating system from a high capacity non-volatile memory device such as a hard drive. The operating system and application software are typically read from the high capacity memory and corresponding images are stored in DRAM. As the processor executes instructions, some types of data may be repeatedly fetched from memory. As a result, some SRAM or other high speed memory is typically provided as "cache" memory in conjunction with the processor and may be included on the processor integrated circuit or chip and/or very near it. Several different kinds of memory device are involved in most modem computing devices, and in many types of appliances that include automated and/or programmable features (home entertainment devices, telecommunications devices, automotive control systems etc.). As system and software complexity increase, need for additional memory increases. Desire for portability, computation power and/or practicality result in increased pressure to reduce both power consumption and circuit area per bit.

DRAMs have been developed to very high capacities in part because the memory cells can be manufactured to have a very small area, and the power draw per cell can also be made quite small. In turn, this allows memory integrated circuits to be made that incorporate millions of memory cells in each chip. Typical one-transistor, one- capacitor DRAM memory cells can be produced to have extremely small areal requirements.

Such areas are often equal to about 3F x 2F, or less, where "F" is defined as equal to one-half of minimum pitch (see Figure 4, infra). Minimum pitch (i.e., "P") is defined as equal to the smallest distance of a line width (i.e., "W") plus width of a space immediately adjacent the line on one side ofthe line between the line and a next adjacent line in a repeated pattern within the array (i.e., "S"). Thus, in many implementations, the consumed area of a given DRAM cell is no greater than about 8F².

However, because DRAMs are volatile memory devices, they require "refresh" operations. In a refresh operation, data are read out of each memory cell, amplified and written back into the DRAM. As a first result, the DRAM circuit is usually not available for other kinds of memory operations during the refresh operation. Additionally, refresh operations are carried out periodically, resulting in times during which data cannot be readily extracted from or written to DRAMs. As a second result, some amount of electrical power is always needed to store data in DRAM devices. As a third result, boot operations for computers such as personal computers involve a period during which the computer cannot be used following power ON initiation. During this period, operating system instructions and associated data, and application instructions and associated data, are read from relatively slow, non-volatile memory, such as a conventional disc drive, are decoded by the processing unit and the resultant instructions and associated data are loaded into modules incorporating relatively rapidly-accessible, but volatile, memory such as DRAM. Other consequences flow from the properties ofthe memory systems included in various electronic devices and the increasingly complex software employed with them, however, these examples serve to illustrate ongoing needs.

Needed are methods and apparatus relating to non- volatile memory providing high areal data storage capacity, reprogrammability, low power consumption and relatively high data access speed.

SUMMARY OF THE INVENTION

In a first aspect, the present invention includes a method for making an array of memory cells configured to store at least one bit per one F^~. The method includes doping a first region of a semiconductor substrate and incising the substrate to provide an array of substantially vertical edge surfaces. Pairs ofthe edge surfaces face one another and are spaced apart a distance equal to one half of a pitch ofthe anay of edges. The method also includes doping second regions between the pairs of edge surfaces and disposing respective structures each providing an electronic memory function on at least some respective ones of the edge surfaces. The method also includes establishing electrical contacts to the first and second regions.

In another aspect, the present invention includes a method for making an array of memory cells configured to store at least one bit per one F . The method includes disposing substantially vertical structures providing an electronic memory function spaced apart a distance equal to one half of a minimum pitch ofthe array and establishing electrical contacts to memory cells including the vertical structures.

In a further aspect, the present invention includes an array of memory cells configured to store at least one bit per one F² formed using vertical structures providing an electronic memory function spaced apart a distance equal to one half of a minimum pitch of the anay. The structures providing the electronic memory function are configured to store more than one bit per gate. The array also includes electrical contacts to the memory cells including the vertical structures.

In a still further aspect, the present invention includes a vertical metal oxide semiconductor field effect transistor (MOSFET) extending outwardly from a substrate, the MOSFET having a first source/drain region, a second source/drain region, a channel region between the first and the second source/drain regions, and a gate separated from the channel region by a gate insulator. A sourceline is formed in a trench adjacent to the vertical MOSFET, wherein the first source/drain region is coupled to the sourceline. A transmission line is coupled to the second source/drain region. The can be programmed MOSFET to have one of a number of charge levels trapped in the gate insulator adjacent to the first source/drain region such that the channel region has a first voltage threshold region (Vtl) and a second voltage threshold region (Vt2) and such that the programmed MOSFET operates at reduced drain source current.

These and other embodiments, aspects, advantages, and features ofthe present invention will be set forth in part in the description which follows, and in part will become apparent to those skilled in the art by reference to the following description ofthe invention and referenced drawings or by practice ofthe invention. The aspects, advantages, and features ofthe invention are realized and attained by means ofthe instrumentalities, procedures, and combinations particularly pointed out in the appended claims.

BRIEF DESCRIPTION OF THE DRAWINGS

Embodiments ofthe invention are described below with reference to the following accompanying drawings. Figure 1 is a simplified side view, in section, of a semiconductor substrate portion at one stage in processing, in accordance with an embodiment ofthe present invention.

Figure 2 is a simplified side view, in section, ofthe substrate portion of Figure 1 at a later stage in processing, in accordance with an embodiment ofthe present invention. Figure 3 is a simplified side view, in section, ofthe substrate portion of Figure 2 at a later stage in processing, in accordance with an embodiment ofthe present invention.

Figure 4 is a simplified plan view of a substrate portion showing a portion of a memory cell array, in accordance with an embodiment ofthe present invention.

5 Figure 5 is a simplified side view, in section, illustrating a relationship between the structures of Figures 1-3 and the plan view of Figure 4, in accordance with an embodiment ofthe present invention.

Figure 6 is a simplified plan view of a memory cell array illustrating an interconnection anangement for the memory cell anay of Figure 4, in accordance with an 0 embodiment ofthe present invention.

Figure 7 is a simplified side view, in section, taken along section lines 7-7 of Figure 6, illustrating part of an interconnection arrangement in accordance with an embodiment ofthe present invention.

Figure 8 is a simplified side view, in section, taken along section lines 8-8 of 5 Figure 6, illustrating part of an interconnection arrangement in accordance with an embodiment ofthe present invention.

Figure 9A is a block diagram of a metal oxide semiconductor field effect transistor (MOSFET) in a substrate according to the teachings ofthe prior art.

Figure 9B illustrates the MOSFET of Figure 9 A operated in the forward >0 direction showing some degree of device degradation due to electrons being trapped in the gate oxide near the drain region over gradual use.

Figure 9C is a graph showing the square root ofthe current signal (Ids) taken at the drain region ofthe conventional MOSFET versus the voltage potential (VGS) established between the gate and the source region.

.5 Figure 10A is a diagram of a programmed MOSFET which can be used as a multistate cell in accordance with an embodiment ofthe present invention.

Figure 10B is a diagram suitable for explaining the method by which the MOSFET ofthe multistate cell ofthe present invention can be programmed to achieve the embodiments ofthe present invention. Figure 1 OC is a graph plotting the current signal (Ids) detected at the drain region versus a voltage potential, or drain voltage, (VDS) set up between the drain region and the source region (Ids vs. VDS) in accordance with an embodiment ofthe present invention.

Figure 11 illustrates a portion of a memory anay in accordance with an embodiment ofthe present invention.

Figure 12 illustrates an electrical equivalent circuit for the portion ofthe memory anay shown in Figure 11.

Figure 13 is another electrical equivalent circuit useful in illustrating a read operation on the novel multistate cell in accordance with an embodiment ofthe present invention.

Figure 14 illustrates a portion of a memory array in accordance with an embodiment ofthe present invention.

Figure 15 A, illustrates one embodiment ofthe gate insulator for the present invention having a number of layers, e.g., an ONO stack, where the layer closest to the channel includes an oxide layer, and a nitride layer is formed thereon.

Figure 15B aids to further illustrate the conduction behavior ofthe novel multistate cell ofthe present invention.

Figure 16A illustrates the operation and programming the novel multistate cell in the reverse direction. Figure 16B illustrates the now programmed multistate cell's operation in the forward direction and differential read occurring in this differential cell embodiment, e.g., 2 transistors in each cell.

Figure 17 illustrates a memory device in accordance with an embodiment ofthe present invention. Figure 18 is a block diagram of an electrical system, or processor-based system, utilizing a multistate cell constructed in accordance with the present invention. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

In the following detailed description ofthe invention, reference is made to the accompanying drawings which form a part hereof, and in which is shown, by way of illustration, specific embodiments in which the invention may be practiced. In the 5 drawings, like numerals describe substantially similar components throughout the several views. These embodiments are described in sufficient detail to enable those skilled in the art to practice the invention. Other embodiments may be utilized and structural, logical, and electrical changes may be made without departing from the scope ofthe present invention.

The terms wafer and substrate used in the following description include any 0 structure having an exposed surface with which to form the integrated circuit (IC) structure ofthe invention. The term substrate is understood to include semiconductor wafers. The term substrate is also used to refer to semiconductor structures during processing, and may include other layers that have been fabricated thereupon. Both wafer and substrate include doped and undoped semiconductors, epitaxial semiconductor layers supported by a base 5 semiconductor or insulator, as well as other semiconductor structures well known to one skilled in the art. The term conductor is understood to include semiconductors, and the term insulator is defined to include any material that is less electrically conductive than the materials referred to as conductors. The following detailed description is, therefore, not to be taken in a limiting sense, and the scope ofthe present invention is defined only by the .0 appended claims, along with the full scope of equivalents to which such claims are entitled.

Figure 1 is a simplified side view, in section, of a semiconductor substrate portion 20 at one stage in processing, in accordance with an embodiment ofthe present invention. The portion 20 includes etched or incised recesses 22, doped regions 24 and 26 and caps 28. The etched recesses 22 form trenches extending along an axis into and out of !5 the page of Figure 1.

In one embodiment, the doped regions 24 are implanted n+ regions. In one embodiment, the doped regions 24 are formed by a blanket implant. In one embodiment, the caps 28 are dielectric caps and may be formed using conventional silicon nitride and conventional patterning techniques. In one embodiment, the etched recesses 22 are then )0 etched using conventional plasma etching techniques. In one embodiment, the doped regions 26 are then doped by implantation to form n+ regions. The etched or incised recesses 22 may be formed by plasma etching, laser-assisted techniques or any other method presently known or that may be developed. In one embodiment, the recesses 22 are formed to have substantially vertical sidewalls relative to a top surface ofthe substrate portion 20. In one embodiment, substantially vertical means at 90 degrees to the substrate 5 surface, plus or minus ten degrees.

Figure 2 provides a simplified side view, in section, ofthe substrate portion 20 of Figure 1 at a later stage in processing, in accordance with an embodiment ofthe present invention. The portion 20 of Figure 2 includes thick oxide regions 32, ONO regions 34 formed on sidewalls 36 ofthe recesses 22, gate material 38 and a conductive layer 40. In 0 one embodiment, the gate material 38 comprises conductively-doped polycrystalline silicon.

In one embodiment, conventional techniques are employed to oxidize the doped regions 24 and 26 preferentially with respect to sidewalls 36. As a result, the thick oxide regions 32 are formed at the same time as a thinner oxide 42 on the sidewalls 36. These oxides also serve to isolate the doped regions 24 and 26 from what will become transistor 5 channels along the sidewalls 36. Other techniques for isolation may be employed. For example, in one embodiment, high density plasma grown oxides may be employed. In one embodiment, spacers may be employed.

In one embodiment, conventional techniques are then employed to provide a nitride layer 44 and an oxide layer 46, as is described, for example, in "NROM: A Novel '.0 Localized Trapping, 2-Bit Nonvolatile Memory Cell", by Boaz Eitan et al., IEEE Electron Device Letters, Vol. 21, No. 11, November 2000, pp. 543-545, IEEE Catalogue No. 0741- 3106/00, or in "A True Single-Transistor Oxide-Nitride-Oxide EEPROM Device" by T.Y. Chan et al., IEEE Electron Device Letters, Vol. EDL-8, No. 3, March, 1987, pp. 93-95, IEEE Catalogue No. 0741-3106/87/0300-0093.

^5 In one embodiment, the thin oxide 42, nitride layer 44 and oxide layer 46 combine to form the ONO layer 34, such as is employed in SONOS devices, while the polysilicon 38 forms a control gate. In operation, application of suitable electrical biases to the doped regions 24, 26 and the control gate 38 cause hot majority charge earners to be injected into the nitride layer 44 and become trapped, providing a threshold voltage shift

30 and thus providing multiple, alternative, measurable electrical states representing stored data. "Hot" charge caniers are not in thermal equilibrium with their environment. In other words, hot charge carriers represent a situation where a population of high kinetic energy charge caniers exist. Hot charge carriers may be electrons or holes.

SONOS devices are capable of storing more than one bit per gate 38. Typically, the hot caniers are injected into one side 47 or 47' ofthe ONO layer 34, adjacent a contact, such as the region 24 or the region 26, that provides a high electrical field.

By reversing the polarity ofthe potentials applied to the regions 24 and 26, charge may be injected into the other side 47' or 47 ofthe ONO layer 34. Thus, four electronically-discriminable and distinct states can be easily provided with a single gate 38. As a result, the structure shown in Figure 2 is capable of storing at least four bits per gate 38.

Figure 3 is a simplified side view, in section, ofthe substrate portion 20 of Figure 1 at an alternative stage in processing, in accordance with an embodiment ofthe present invention. The embodiment shown in Figure 3 includes the oxide regions 32 and 42, but a floating gate 48 is formed on the thin oxide region 42. A conventional oxide or nitride insulator 49 is formed on the floating gate 48, followed by deposition of gate material 38. Floating gate devices are known and operate by injecting hot charge earners, which may comprise electrons or holes, into the floating gate 48.

Floating gate devices can be programmed to different charge levels that can be electrically distinct and distinguishable. As a result, it is possible to program more data than one bit into each floating gate device, and each externally addressable gate 38 thus conesponds to more than one stored bit. Typically, charge levels of 0, Q, 2Q and 3Q might be employed, where Q represents some amount of charge conesponding to a reliably- distinguishable output signal.

Figure 4 is a simplified plan view of a substrate portion showing a portion of a memory cell array 50, in accordance with an embodiment ofthe present invention. Figure 4 also provides examples of pitch P, width W, space S and minimum feature size F, as described in the Background. An exemplary memory cell area 52, the physical area of a single transistor, can be seen to be about one F². Wordlines 54 are formed from the conductive layer 40, and bitlines 56 and 58 are formed. Figure 5 is a simplified side view, in section, illustrating a relationship between the structures of Figures 1-3 and the plan view of Figure 4, in accordance with an embodiment ofthe present invention. The trenches 22 conespond to bitlines 56 and 58, as is explained below in more detail with reference to Figures 6-8.

5 The density of memory anays such as that described with reference to Figures 1-

5 can require interconnection aπangements that differ from prior art memory anays. One embodiment of a new type of interconnection arrangement useful with such memory systems is described below with reference to Figures 6-8.

Figure 6 is a simplified plan view illustrating an interconnection arrangement 60 0 for the memory cell anay 50 of Figure 4, in accordance with an embodiment ofthe present invention. The interconnection arrangement 60 includes multiple patterned conductive layers 62 and 64, separated by conventional interlevel dielectric material 65 (Figures 7 and 8). The views in Figure 6-8 have been simplified to show conespondence with the other Figures and to avoid undue complexity. Shallow trench isolation regions 67 isolate selected 5 portions from one another.

Figure 7 is a simplified side view, in section, taken along section lines 7-7 of Figure 6, illustrating part of an interconnection arrangement in accordance with an embodiment ofthe present invention.

Figure 8 is a simplified side view, in section, taken along section lines 8-8 of :0 Figure 6, illustrating part of an interconnection anangement in accordance with an embodiment ofthe present invention.

With reference to Figures 6-8, the patterned conductive layer 62 extends upward to nodes 70, 70', 70" and establishes electrical communication between the conductive layers 62 and selected portions ofthe doped region 24. The patterned conductive layer 62 15 stops at the line denoted 72, 72'.

Similarly, other portions ofthe patterned conductive layer 62 extend from the line denoted 74, 74' and extend upward, providing electrical communication from nodes 76, 76', 76" to other circuit elements. The nodes 76, 76', 76" provide contact to selected portions ofthe doped region 24. In contrast, patterned conductive layers 64 extend from top to bottom of Figure 6 and electrically couple to nodes 78, 78' and thus to doped region 26.

Such is but on example of a simplified interconnection anangement suitable for use with the memory devices of Figures 1-5. Other arrangements are possible.

5 Figure 9A is useful in illustrating the conventional operation of a MOSFET such as can be used in a DRAM array. Figure 9A illustrates the normal hot electron injection and degradation of devices operated in the forward direction. As is explained below, since the electrons are trapped near the drain they are not very effective in changing the device characteristics.

0 Figure 9 A is a block diagram of a metal oxide semiconductor field effect transistor (MOSFET) 101 in a substrate 100. The MOSFET 101 includes a source region 102, a drain region 104, a channel region 106 in the substrate 100 between the source region 102 and the drain region 104. A gate 108 is separated from the channel region 108 by a gate oxide 110. A sourceline 112 is coupled to the source region 102. A bitline 114 is

[5 coupled to the drain region 104. A wordline 116 is coupled to the gate 108.

In conventional operation, a drain to source voltage potential (Vds) is set up between the drain region 104 and the source region 102. A voltage potential is then applied to the gate 108 via a wordline 116. Once the voltage potential applied to the gate 108 surpasses the characteristic voltage threshold (Vt) ofthe MOSFET a channel 106 forms in -0 the substrate 100 between the drain region 104 and the source region 102. Formation ofthe channel 106 permits conduction between the drain region 104 and the source region 102, and a current signal (Ids) can be detected at the drain region 104.

In operation ofthe conventional MOSFET of Figure 9A, some degree of device degradation does gradually occur for MOSFETs operated in the forward direction by 25 electrons 117 becoming trapped in the gate oxide 110 near the drain region 104. This effect is illustrated in Figure 9B. However, since the electrons 117 are trapped near the drain region 104 they are not very effective in changing the MOSFET characteristics.

Figure 9C illustrates this point. Figure 9C is a graph showing the square root of the cunent signal (Ids) taken at the drain region versus the voltage potential (VGS) 0 established between the gate 108 and the source region 102. The change in the slope ofthe plot of ^Jlds versus VGS represents the change in the charge carrier mobility in the channel 106.

In Figure 9C, ΔVT represents the minimal change in the MOSFET' s threshold voltage resulting from electrons gradually being trapped in the gate oxide 110 near the drain region 104, under normal operation, due to device degradation. This results in a fixed trapped charge in the gate oxide 110 near the drain region 104. Slope 1 represents the charge canier mobility in the channel 106 for Figure 9A having no electrons frapped in the gate oxide 110. Slope 2 represents the charge mobility in the channel 106 for the conventional MOSFET of Figure 9B having electrons 117 trapped in the gate oxide 110 near the drain region 104. As shown by a comparison of slope 1 and slope 2 in Figure 9C, the electrons 117 trapped in the gate oxide 110 near the drain region 104 ofthe conventional MOSFET do not significantly change the charge mobility in the channel 106.

There are two components to the effects of stress and hot electron injection. One component includes a threshold voltage shift due to the trapped electrons and a second component includes mobility degradation due to additional scattering of canier electrons caused by this trapped charge and additional surface states. When a conventional MOSFET degrades, or is "stressed," over operation in the forward direction, electrons do gradually get injected and become trapped in the gate oxide near the drain. In this portion ofthe conventional MOSFET there is virtually no channel underneath the gate oxide. Thus the trapped charge modulates the threshold voltage and charge mobility only slightly.

Applicant has previously described programmable memory devices and functions based on the reverse stressing of MOSFET' s in a conventional CMOS process and technology in order to form programmable address decode and conection. (See generally, L. Forbes, W.P. Noble and E.H. Cloud, "MOSFET technology for programmable address decode and correction," United States Patent Application Serial Number

09/383,804). That disclosure, however, did not describe multistate memory cell solutions, but rather address decode and conection issues.

According to the teachings ofthe present invention, normal MOSFETs can be programmed by operation in the reverse direction and utilizing avalanche hot electron injection to trap electrons in the gate oxide ofthe MOSFET. When the programmed MOSFET is subsequently operated in the forward direction the electrons trapped in the oxide are near the source and cause the channel to have two different threshold voltage regions. The novel programmed MOSFETs ofthe present invention conduct significantly less cunent than conventional MOSFETs, particularly at low drain voltages. These electrons will remain trapped in the gate oxide unless negative gate voltages are applied. The electrons will not be removed from the gate oxide when positive or zero gate voltages are applied. Erasure can be accomplished by applying negative gate voltages and/or increasing the temperature with negative gate bias applied to cause the trapped electrons to be re-emitted back into the silicon channel ofthe MOSFET. (See generally, L. Forbes, E. Sun, R. Alders and J. Moll, "Field induced re-emission of electrons trapped in SiO ," IEEE Trans. Electron Device, vol. ED-26, no. 11, pp. 1816-1818 (Nov. 1979); S.S.B. Or, N.

Hwang, and L. Forbes, "Tunneling and Thermal emission from a distribution of deep traps in SiO₂," IEEE Trans, on Electron Devices, vol. 40, no. 6, pp. 1100-1103 (June 1993); S.A. Abbas and R.C. Dockerty, "N-channel IGFET design limitations due to hot electron trapping," IEEE Int. Electron Devices Mtg., Washington D.C., Dec. 1975, pp. 35-38). Figures 10A-10C are useful in illustrating the present invention in which a much larger change in device characteristics is obtained by programming the device in the reverse direction and subsequently reading the device by operating it in the forward direction.

Figure 10A is a diagram of a programmed MOSFET which can be used as a multistate cell according to the teachings ofthe present invention. As shown in Figure 10A the multistate cell 201 includes a MOSFET in a substrate 200 which has a first source/drain region 202, a second source/drain region 204, and a channel region 206 between the first and second source/drain regions, 202 and 204. In one embodiment, the first source/drain region 202 includes a source region 202 for the MOSFET and the second source/drain region 204 includes a drain region 204 for the MOSFET. Figure 10A further illustrates a gate 208 separated from the channel region 206 by a gate oxide 210. A first transmission line 212 is coupled to the first source/drain region 202 and a second transmission line 214 is coupled to the second source/drain region 204. In one embodiment, the first transmission line includes a sourceline 212 and the second transmission line includes a bit line 214.

As stated above, multistate cell 201 is comprised of a programmed MOSFET. This programmed MOSFET has a charge 217 trapped in the gate oxide 210 adjacent to the first source/drain region 202 such that the channel region 206 has a first voltage threshold region (Vtl) and a second voltage threshold region (Vt2) in the channel 206. In one embodiment, the charge 217 trapped in the gate oxide 210 adjacent to the first source/drain region 202 includes a trapped electron charge 217. According to the teachings ofthe present invention and as described in more detail below, the multistate cell can be programmed to have one of a number of charge levels trapped in the gate insulator adjacent to the first source/drain region 202 such that the channel region 206 will have a first voltage threshold region (Ntl) and a second voltage threshold region (Nt2) and such that the programmed multistate cell operates at reduced drain source cunent.

Figure 10A illustrates the Nt2 in the channel 206 is adjacent the first source/drain region 202 and that the Vtl in the channel 206 is adjacent the second source/drain region 204. According to the teachings ofthe present invention, Vt2 has a higher voltage threshold than Vtl due to the charge 217 trapped in the gate oxide 217 adjacent to the first source/drain region 202. Multiple bits can be stored on the multistate cell 201. Figure 10B is a diagram suitable for explaining the method by which the

MOSFET ofthe multistate cell 201 ofthe present invention can be programmed to achieve the embodiments ofthe present invention. As shown in Figure 10B the method includes programming the MOSFET in a reverse direction. Programming the MOSFET in the reverse direction includes applying a first voltage potential VI to a drain region 204 ofthe MOSFET. In one embodiment, applying a first voltage potential VI to the drain region 204 ofthe MOSFET includes grounding the drain region 204 ofthe MOSFET as shown in Figure 10B. A second voltage potential V2 is applied to a source region 202 of the MOSFET. In one embodiment, applying a second voltage potential V2 to the source region 202 includes applying a high positive voltage potential (VDD) to the source region 202 of the MOSFET, as shown in Figure 10B. A gate potential VGS is applied to a gate 208 ofthe MOSFET. In one embodiment, the gate potential VGS includes a voltage potential which is less than the second voltage potential V2, but which is sufficient to establish conduction in the channel 206 ofthe MOSFET between the drain region 204 and the source region 202. As shown in Figure 10B, applying the first, second and gate potentials (VI, V2, and VGS respectively) to the MOSFET creates a hot electron injection into a gate oxide 210 ofthe MOSFET adjacent to the source region 202. In other words, applying the first, second and gate potentials (VI, V2, and VGS respectively) provides enough energy to the charge carriers, e.g. electrons, being conducted across the channel 206 that, once the charge caniers are near the source region 202, a number ofthe charge carriers get excited into the gate oxide 210 adjacent to the source region 202. Here the charge caniers become trapped. In one embodiment ofthe present invention, the method is continued by subsequently operating the MOSFET in the forward direction in its programmed state during a read operation. Accordingly, the read operation includes grounding the source region 202 and precharging the drain region a fractional voltage of VDD. If the device is addressed by a wordline coupled to the gate, then its conductivity will be determined by the presence or absence of stored charge in the gate insulator. That is, a gate potential can be applied to the gate 208 by a wordline 216 in an effort to form a conduction channel between the source and the drain regions as done with addressing and reading conventional DRAM cells.

However, now in its programmed state, the conduction channel 206 ofthe MOSFET will have a first voltage threshold region (Vtl) adjacent to the drain region 204 and a second voltage threshold region (Vt2) adjacent to the source region 202, as explained and described in detail in connection with Figure 10A. According to the teachings ofthe present invention, the Vt2 has a greater voltage threshold than the Vtl due to the hot electron injection 217 into a gate oxide 210 ofthe MOSFET adjacent to the source region 202.

Figure 10C is a graph plotting a cunent signal (Ids) detected at the second source/drain region 204 versus a voltage potential, or drain voltage, (VDS) set up between the second source/drain region 204 and the first source/drain region 202 (Ids vs. VDS). In one embodiment, VDS represents the voltage potential set up between the drain region 204 and the source region 202. In Figure 10C, the curve plotted as Dl represents the conduction behavior of a conventional MOSFET which is not programmed according to the teachings ofthe present invention. The curve D2 represents the conduction behavior ofthe programmed MOSFET, described above in connection with Figure 10A, according to the teachings ofthe present invention. As shown in Figure 10C, for a particular drain voltage, VDS, the cunent signal (IDS2) detected at the second source/drain region 204 for the programmed MOSFET (curve D2) is significantly lower than the cunent signal (IDS1) detected at the second source/drain region 204 for the conventional MOSFET which is not programmed according to the teachings ofthe present invention. Again, this is attributed to the fact that the channel 206 in the programmed MOSFET ofthe present invention has two voltage threshold regions and that the voltage threshold, Vt2, near the first source/drain region 202 has a higher voltage threshold than Vtl near the second source/drain region due to the charge 217 trapped in the gate oxide 217 adjacent to the first source/drain region 202.

Some of these effects have recently been described for use in a different device structure, called an NROM, for flash memories. This latter work in Israel and Germany is based on employing charge trapping in a silicon nitride layer in a non-conventional flash memory device structure. (See generally, B. Eitan et al., "Characterization of Channel Hot Electron Injection by the Subthreshold Slope of NROM device," IEEE Electron Device Lett., Vol. 22, No. 11, pp. 556-558, (Nov. 2001); B. Etian et al, "NROM: A novel localized Trapping, 2-Bit Nonvolatile Memory Cell," IEEE Electron Device Lett., Vol. 21, No. 11, pp. 543-545, (Nov. 2000)). Charge trapping in silicon nitride gate insulators was the basic mechanism used in MNOS memory devices (see generally, S. Sze, Physics of

Semiconductor Devices, Wiley, N.Y., 1981, pp. 504-506), charge trapping in aluminum oxide gates was the mechanism used in MIOS memory devices (see generally, S. Sze, Physics of Semiconductor Devices, Wiley, N.Y., 1981, pp. 504-506), and Applicant has previously disclosed charge trapping at isolated point defects in gate insulators (see generally, L. Forbes and J. Geusic, "Memory using insulator traps," US Patent Number 6,140,181, issued October 31, 2000).

In contrast to the above work, the present invention discloses programming a MOSFET in a reverse direction to trap one of a number of charge levels near the source region and reading the device in a forward direction to form a multistate memory cell based on a modification of DRAM technology.

Prior art DRAM technology generally employs silicon oxide as the gate insulator. Further the emphasis in conventional DRAM devices is placed on trying to minimize charge trapping in the silicon oxide gate insulator. According to the teachings of the present invention, a variety of insulators are used to trap electrons more efficiently than in silicon oxide. That is, in the present invention, the multistate memory cell employs charge trapping in gate insulators such as, wet silicon oxide, silicon nitride, silicon oxynitride SON, silicon rich oxide SRO, aluminum oxide Al O₃, composite layers of these insulators such as oxide and then silicon nitride, or oxide and then aluminum oxide, or multiple layers as oxide-nitride-oxide. While the charge trapping efficiency of silicon oxide may be low such is not the case for silicon nitride or composite layers of silicon oxide and nitride.

Figure 11 illustrates a portion of a memory anay 300 according to the teachings ofthe present invention. The memory in Figure 11 is shown illustrating a number of vertical pillars, or multistate cells, 301-1 and 301-2 formed according to the teachings ofthe present invention. As one of ordinary skill in the art will appreciate upon reading this disclosure, the number of vertical pillars are formed in rows and columns extending outwardly from a substrate 303. As shown in Figure 11, the number of vertical pillars, 301- 1 and 301-2 are separated by a number of trenches 340. According to the teachings ofthe present invention, the number of vertical pillars, 301-1 and 301-2, serve as transistors including a first source/drain region, 302-1 and 302-2, respectively. The first source/drain region, 302-1 and 302-2, is coupled to a sourceline 304. As shown in Figure 11, the sourceline 304 is formed in a bottom ofthe trenches 340 between rows ofthe vertical pillars, 301-1 and 301-2. In one embodiment, according to the teachings ofthe present invention, the sourceline 304 is formed from a doped region implanted in the bottom ofthe trench. A second source/drain region, 306-1 and 306-2 respectively, is coupled to a bitline (not shown). A channel region 305 is located between the first and the second source/drain regions.

As shown in Figure 11, a gate 307 is separated from the channel region 305 by a gate insulator 307 in the trenches 340 along rows ofthe vertical pillars, 301-1 and 301-2. In one embodiment, according to the teachings ofthe present invention, the gate insulator 307 includes a gate insulator 307 selected from the group of silicon dioxide (SiO₂) formed by wet oxidation, silicon oxynitride (SON), silicon rich oxide (SRO), and aluminum oxide (Al O₃). In another embodiment, according to the teachings ofthe present invention, the gate insulator 307 includes a gate insulator 307 selected from the group of silicon rich aluminum oxide insulators, silicon rich oxides with inclusions of nanoparticles of silicon, silicon oxide insulators with inclusions of nanoparticles of silicon carbide, and silicon oxycarbide insulators. In another embodiment, according to the teachings ofthe present invention, the gate insulator 307 includes a composite layer 307. In this embodiment, the composite layer 307 includes a composite layer 307 selected from the group of an oxide- aluminum oxide (Al₂O₃)-oxide composite layer, and oxide-silicon oxycarbide-oxide composite layer. In another embodiment, the composite layer 307 includes a composite layer 307, or a non-stoichiometric single layer, of two or more materials selected from the group of silicon (Si), titanium (Ti), and tantalum (Ta). In another embodiment, according to the teachings ofthe present invention, the gate insulator 307 includes an oxide-nitride-oxide (ONO) gate insulator 307.

Figure 12 illustrates an electrical equivalent circuit 400 for the portion ofthe memory anay shown in Figure 11. As shown in Figure 12, a number of vertical multistate cells, 401-1 and 401-2, are provided. Each vertical multistate cell, 401-1 and 401-2, includes a first source/drain region, 402-1 and 402-2, a second source/drain region 406-1 and 406-2, a chamiel region 405 between the first and the second source/drain regions, and a gate 409 separated from the channel region by a gate insulator 407. Figure 12 further illustrates a number of bit lines, 411-1 and 411-2, coupled to the second source/drain region 406-1 and 406-2 of each multistate cell. In one embodiment, as shown in Figure 12, the number of bit lines, 411-1 and 411-2, are coupled to the second source/drain region 406-1 and 406-2 along rows ofthe memory array. A number of word lines, such as wordline 413 in Figure 12, are coupled to the gate 409 of each multistate cell along columns ofthe memory array. And, a number of sourcelines, such as common sourceline 415, are coupled to the first source/drain regions, e.g. 402-1 and 402-2, along columns ofthe vertical multistate cells, 401-1 and 401-2, such that adjacent pillars containing these transistors share the common sourceline 415. In one embodiment, column adjacent pillars include a transistor which operates as a vertical multistate cell, e.g. 401-1, on one side of a shared trench, the shared trench separating rows ofthe pillars as described in connection with Figure 11, and a transistor which operates as a reference cell, e.g. 401-2, having a programmed conductivity state on the opposite side ofthe shared trench. In this manner, according to the teachings ofthe present invention and as described in more detail below, at least one of multistate cells can be programmed to have one of a number of charge levels trapped in the gate insulator, shown generally as 417, adjacent to the first source/drain region, e.g. 402-1, such that the channel region 405 will have a first voltage threshold region (Vtl) and a second voltage threshold region (Vt2) and such that the programmed multistate cell operates at reduced drain source current.

Figure 13 is another electrical equivalent circuit useful in illustrating a read operation on the novel multistate cell 500 according to the teachings ofthe present invention. The electrical equivalent circuit in Figure 13 represents a programmed vertical multistate cell. As explained in detail in connection with Figure 11, the programmed vertical multistate cell 500 includes a vertical metal oxide semiconductor field effect transistor (MOSFET) 500 extending outwardly from a substrate. The MOSFET has a source region 502, a drain region 506, a channel region 505 between the source region 502 and the drain region 506, and a gate 507 separated from the channel region 505 by a gate insulator, shown generally as 517.

As shown in Figure 13 a wordline 513 is coupled to the gate 507. A sourceline 504, formed in a trench adjacent to the vertical MOSFET as described in connection with Figure 11, is coupled to the source region 502. A bit line, or data line 511 is coupled to the drain region 506. The multistate cell 500 shown in Figure 13 is an example of a programmed multistate cell 500 having one of a number of charge levels trapped in the gate insulator, shown generally as 517, adjacent to the first source/drain region, 502, such that the channel region 505 will have a first voltage threshold region (Vtl) and a second voltage threshold region (Vt2) and such that the programmed multistate cell 500 operates at reduced drain source cunent. According to the teachings ofthe present invention, the second voltage threshold region (Vt2) is now a high voltage threshold region which is greater than the first voltage threshold region (Vtl).

Figure 14 illustrates a portion of a memory array 600 according to the teachings ofthe present invention. The memory in Figure 14 is shown illustrating a pair of multistate cells 601-1 and 601-2 formed according to the teachings ofthe present invention. As one of ordinary skill in the art will understand upon reading this disclosure, any number of multistate cells can be organized in an anay, but for ease of illustration only two are displayed in Figure 14. As shown in Figure 14, a first source/drain region, 602-1 and 602-2 respectively, is coupled to a sourceline 604. A second source/drain region, 606-1 and 606-2 respectively, is coupled to a bitline, 608-1 and 608-2 respectively. Each ofthe bitlines, 608- 1 and 608-2, couple to a sense amplifier, shown generally at 610. A wordline, 612-1 and 612-2 respectively, is couple to a gate, 614-1 and 614-2 respectively, for each ofthe multistate cells, 601-1 and 601-2. According to the teachings ofthe present invention, the wordlines, 612-1 and 612-2, run across or are perpendicular to the rows ofthe memory array 600. Finally, a write data/precharge circuit is shown at 624 for coupling a first or a second potential to bitline 608-1. As one of ordinary skill in the art will understand upon reading this disclosure, the write data/precharge circuit 624 is adapted to couple either a ground to the bitline 608-1 during a write operation in the reverse direction, or alternatively to precharge the bitline 608-1 to fractional voltage of VDD during a read operation in the forward direction. As one of ordinary skill in the art will understand upon reading this disclosure, the sourceline 604 can be biased to a voltage higher than VDD during a write operation in the reverse direction, or alternatively grounded during a read operation in the forward direction.

As shown in Figure 14, the array structure 600, including multistate cells 601-1 and 601-2, has no capacitors. Instead, according to the teachings ofthe present invention, the first source/drain region or source region, 602-1 and 602-2, are coupled directly to the sourceline 604. In order to write, the sourceline 604 is biased to voltage higher than VDD and the devices stressed in the reverse direction by grounding the data or bit line, 608-1 or 608-2. If the multistate cell, 601-1 or 601-2, is selected by a word line address, 612-1 or 612-2, then the multistate cell, 601-1 or 601-2, will conduct and be stressed with accompanying hot electron injection into the cells gate insulator adjacent to the source region, 602-1 or 602-2. As one of ordinary skill in the art will understand upon reading this disclosure, a number of different charge levels can be programmed into the gate insulator adjacent to source region such that the cells is used as a differential cell and/or the cell is compared to a reference or dummy cell, as shown in Figure 14, and multiple bits can be stored on the multistate cell.

During read the multistate cell, 601-1 or 601-2, is operated in the forward direction with the sourceline 604 grounded and the bit line, 608-1 or 608-2, and respective second source/drain region or drain region, 606-1 and 606-2, ofthe cells precharged to some fractional voltage of Vdd. If the device is addressed by the word line, 612-1 or 612-2, then its conductivity will be determined by the presence or absence ofthe amount of stored charge trapped in the gate insulator as measured or compared to the reference or dummy cell and so detected using the sense amplifier 610. The operation of DRAM sense amplifiers is described, for example, in U.S. Pat. Nos. 5,627,785; 5,280,205; and 5,042,011, all assigned to Micron Technology Inc., and incorporated by reference herein. The anay would thus be addressed and read in the conventional manner used in DRAM's, but programmed as multistate cells in a novel fashion.

In operation the devices would be subjected to hot electron stress in the reverse direction by biasing the sourceline 604, and read while grounding the sourceline 604 to compare a stressed multistate cell, e.g. cell 601-1, to an unstressed dummy device/cell, e.g. 601-2, as shown in Figure 14. The write and possible erase feature could be used during manufacture and test to initially program all cells or devices to have similar or matching conductivity before use in the field. Likewise, the transistors in the reference or dummy cells, e.g. 601-2, can all initially be programmed to have the same conductivity states. According to the teachings ofthe present invention, the sense amplifier 610 can then detect small differences in cell or device characteristics due to stress induced changes in device characteristics during the write operation.

As one of ordinary skill in the art will understand upon reading this disclosure such anays of multistate cells are conveniently realized by a modification of DRAM technology. According to the teachings ofthe present invention a gate insulator ofthe multistate cell includes gate insulators selected from the group of thicker layers of SiO₂ formed by wet oxidation, SON silicon oxynitride, SRO silicon rich oxide, Al₂O₃ aluminum oxide, composite layers and implanted oxides with traps (L. Forbes and J. Geusic, "Memory using insulator traps," US Patent Number 6,140,181, issued October 31, 2000). Conventional transistors for address decode and sense amplifiers can be fabricated after this step with noπnal thin gate insulators of silicon oxide. Figures 15 A- 15B and 16A- 16B are useful in illustrating the use of charge storage in the gate insulator to modulate the conductivity ofthe multistate cell according to the teachings ofthe present invention. That is, Figures 15A-16B illustrate the operation of the novel multistate cell 701 formed according to the teachings ofthe present invention. As shown in Figure 15 A, the gate insulator 707 has a number of layers, e.g. an ONO stack, where layer 707A is the oxide layer closest to the channel 705 and a nitride layer 707B is formed thereon. In the embodiment shown in Figure 15A the oxide layer 707A is illustrated having a thickness of approximately 6.7 nm or 67 A (roughly IO^"6 cm). In the embodiment shown in Figure 15A a multistate cell is illustrated having dimensions of 0.1 μm (10^-5 cm) by 0.1 μm. For purposes of illustration, the charge storage region near the source can reasonably have dimensions of 0.1 micron (1000 A) by 0.02 micron (200 A) in a 0.1 micron technology. If the gate oxide 707A nearest the channel 705 is 67 A then a charge of 100 electrons will cause a threshold voltage shift in this region of 1.6 Volts since the oxide capacitance is about 0.5 micro-Farad (μF) per square centimeter. If the transistor has a total effective oxide thickness of 200 A then a change in the threshold voltage of only 0.16 Volts near the source, conesponding to 10 electrons, is estimated to change the transistor current by 4 micro Amperes (μA). The sense amplifier described in connection with Figure 14, which is similar to a DRAM sense amplifier, can easily sense this charge difference on the data or bitlines. In this embodiment, the sensed charge difference on the data or bitlines will be 40 femto Coulombs (fC) over a sense period of 10 nano seconds (nS).

To illustrate these numbers, the capacitance, Ci, ofthe structure depends on the dielectric constant, e j, (which for silicon dioxide SiO₂ equates to 1.06/3 x 10^" F/cm), and the thickness ofthe insulating layers, t, (given here as 6.7 x 10^"7 cm), such that Ci = e i/t = ((1.06 x IO^"12 F/cm/(3 x 6.7 x IO^"7 cm)) = 0.5 xlO^"6 Farads/cm² (F/cm²). This value taken i i <y over the charge storage region near the source, e.g. 20 nm x 100 nm or 2 x 10^" cm , results in a capacitance value of Ci = IO^"17 Farads. Thus, for a change in the threshold voltage of ΔV = 1.6 Volts the stored charge must be Q = C x ΔV = (IO^"17 Farads x 1.6 Volts) = 1.6 x 10^"17 Coulombs. Since Q = Nq, the number of electrons stored is approximately Q/q = (1.6 x IO^"17 Coulombs/1.6 x IO^"19 Coulombs) or 100 electrons. In effect, the programmed multistate cell, or modified MOSFET is a programmed MOSFET having a charge trapped in the gate insulator adjacent to a first source/drain region, or source region, such that the channel region has a first voltage threshold region (Vtl) and a second voltage threshold region (Vt2), where Vt2 is greater than Vtl, and Vt2 is adjacent the source region such that the programmed MOSFET operates at reduced drain source current. For Δ Q = 100 electrons in the dimensions given above, if the transistor has a total effective oxide thickness of 200 A then a change in the threshold voltage of only 0.16 Volts near the source, conesponding to 10 electrons, is estimated to change the transistor cunent by 4 micro

Amperes (μA). As stated above, the sense amplifier described in connection with Figure 14, which is similar to a DRAM sense amplifier, can easily sense this charge difference on the data or bitlines. And, the sensed charge difference on the data or bitlines will be 40 femto Coulombs (fC) over a sense period of 10 nano seconds (nS) for this representative one of a number of stored charge levels according to the teachings ofthe present invention. Again, a number of different charge levels can be programmed into the gate insulator adjacent to source region such that the cell is used as a differential cell and/or the cell is compared to a reference or dummy cell, as shown in Figure 14, and multiple bits can be stored on the multistate cell ofthe present invention.

Figure 15B aids to further illustrate the conduction behavior ofthe novel multistate cell ofthe present invention. The electrical equivalent circuit shown in Figure 15B illustrates a multistate cell 701 having an equivalent oxide thickness of 200 A. The charge storage region near the source 702 can reasonably have a length dimension of 0.02 micron (20 nm) in a 0.1 micron technology with a width dimension of 0.1 micron (100 nm). Therefore, for a change in the drain source voltage (Δ V_DS) in this region an electric field of E = (0.1 V/2 xlO^"6 cm) = 0.5 x IO⁵ V/cm or 5 x IO⁴ V/cm is provided. The drain cunent is calculated using the formula I_D = μC_ox x (W/L) x (Vgs - Vt) x Δ V_DS- In this example, μC_0X = μCj is taken as 50 μA/N and W/L = 5. Appropriate substitution into the drain cunent provides I_D = (50 μA/V² x 5 x 0.16 Volts x 0.1 Volts) = 2.5 x 1.6 μA = 4 μA. As noted above this drain cunent I_D conesponds to 10 electrons trapped in the gate insulator, or charge storage region 707 near the source 702. Sensed over a period of 10 nanoSeconds (nS) produces a cunent on the bitline of 40 fC (e.g. 4 μA x 10 nS = 40 x 10^"15 Coulombs).

Figure 16A and 16B, illustrate the operation and programming the novel multistate cell as described above. However, Figures 16A and 16B also help illustrate an alternative anay configuration where adjacent devices are compared and one ofthe devices on the opposite side of a shared trench is used as a dummy cell transistor or reference device. Again, the reference devices can all be programmed to have the same initial conductivity state. Figure 16A illustrates the operation and programming the novel multistate cell in the reverse direction. As shown in Figure 16 A, a transistor 801-1 on one side ofthe trench (as described in connection with Figure 11) is stressed by grounding its respective drain line, e.g. 811-1. As shown in Figure 16A, the drain line 811-2 for the transistor 801-2 on the opposite side ofthe trench is left floating. A voltage is applied to the shared sourceline 804 located at the bottom ofthe trench (as described in connection with Figure 11) which now acts as a drain. As shown in this electrical equivalent circuit, the neighboring (shared trench)/column adjacent transistors, 801-1 and 801-2, share a gate 807 and the wordline 813, e.g. polysilicon gate lines, coupling thereto run across or are peφendicular to the rows containing the bit and source lines, e.g. 811-1, 811-2, and 804. A gate voltage is applied to the gates 807. Here the multistate cell 801-1 will conduct and be stressed with accompanying hot electron injection into the cells gate insulator 817 adjacent to the source region 802-1.

Figure 16B illustrates the now programmed multistate cell's operation in the forward direction and differential read occurring in a this differential cell embodiment, e.g. 2 transistors in each cell. To read this state the drain and source (or ground) have the noniial connections and the conductivity ofthe multistate cell is determined. That is, the drain line, 811-1 and 811-2, have the normal forward direction potential applied thereto. The shared sourceline 804 located at the bottom ofthe trench (as described in com ection with Figure 11) is grounded and once again acts as a source. And, a gate voltage is applied to the gates 807. As one of ordinary skill in the art will understand upon reading this disclosure, a number of different charge levels can be programmed into the gate insulator 817 adjacent to source region 802-1 and compared to the reference or dummy cell, 802-2. Thus, according to the teachings of present invention multiple bits can be stored on the multistate cell.

As stated above, these novel multistate cells can be used in a DRAM like anay. Two transistors can occupy an area of 4F squared (F = the minimum lithographic feature size) when viewed from above, or each memory cell consisting of one transistor utilizing an area of 2F squared. Each transistor can now, however, store many bits so the data storage density is much higher than one bit for each IF squared unit area. Using a reference or dummy cell for each memory transistor where the reference transistor is in close proximity, e.g. the embodiment shown in Figures 16A and 16B vs. that shown in Figure 12, results in better matching characteristics of transistors, but a lower memory density.

In Figure 17 a memory device is illustrated according to the teachings ofthe present invention. The memory device 940 contains a memory array 942, row and column decoders 944, 948 and a sense amplifier circuit 946. The memory anay 942 consists of a plurality of multistate cells 900, formed according to the teachings ofthe present invention whose word lines 980 and bit lines 960 are commonly ananged into rows and columns, respectively. The bit lines 960 ofthe memory anay 942 are connected to the sense amplifier circuit 946, while its word lines 980 are connected to the row decoder 944. Address and control signals are input on address/control lines 961 into the memory device 940 and connected to the column decoder 948, sense amplifier circuit 946 and row decoder 944 and are used to gain read and write access, among other things, to the memory anay 942.

The column decoder 948 is connected to the sense amplifier circuit 946 via control and column select signals on column select lines 962. The sense amplifier circuit 946 receives input data destined for the memory anay 942 and outputs data read from the memory anay 942 over input/output (I/O) data lines 963. Data is read from the cells ofthe memory array 942 by activating a word line 980 (via the row decoder 944), which couples all ofthe memory cells conesponding to that word line to respective bit lines 960, which define the columns of the anay. One or more bit lines 960 are also activated. When a particular word line 980 and bit lines 960 are activated, the sense amplifier circuit 946 connected to a bit line column detects and amplifies the conduction sensed through a given multistate cell, where in the read operation the source region of a given cell is couple to a grounded array plate (not shown), and transfened its bit line 960 by measuring the potential difference between the activated bit line 960 and a reference line which may be an inactive bit line. The operation of Memory device sense amplifiers is described, for example, in US Patent Numbers 5,627,785; 5,280,205; and 5,042,011, all assigned to Micron Technology Inc., and incoφorated by reference herein.

Figure 18 is a block diagram of an electrical system, or processor-based system, 1000 utilizing multistate memory cells 1012 constructed in accordance with the present invention. That is, the multistate memory cells 1012 utilizes the modified DRAM cell as explained and described in detail in connection with Figures 2-4. The processor-based system 1000 may be a computer system, a process control system or any other system employing a processor and associated memory. The system 1000 includes a central processing unit (CPU) 1002, e.g., a microprocessor, that communicates with the multistate memory 1012 and an I/O device 1008 over a bus 1020. It must be noted that the bus 1020 may be a series of buses and bridges commonly used in a processor-based system, but for convenience puφoses only, the bus 1020 has been illustrated as a single bus. A second I/O device 1010 is illustrated, but is not necessary to practice the invention. The processor- based system 1000 can also includes read-only memory (ROM) 1014 and may include peripheral devices such as a floppy disk drive 1004 and a compact disk (CD) ROM drive 1006 that also communicates with the CPU 1002 over the bus 1020 as is well known in the art.

It will be appreciated by those skilled in the art that additional circuitry and control signals can be provided, and that the memory device 1000 has been simplified to help focus on the invention. At least one ofthe multistate cell in NROM 1012 includes a programmed MOSFET having a charge trapped in the gate insulator adjacent to a first source/drain region, or source region, such that the channel region has a first voltage threshold region (Vtl) and a second voltage threshold region (Nt2), where Nt2 is greater than Vtl, and Vt2 is adjacent the source region such that the programmed MOSFET operates at reduced drain source cunent.

It will be understood that the embodiment shown in Figure 18 illustrates an embodiment for electronic system circuitry in which the novel memory cells ofthe present invention are used. The illustration of system 1000, as shown in Figure 18, is intended to provide a general understanding of one application for the structure and circuitry ofthe present invention, and is not intended to serve as a complete description of all the elements and features of an electronic system using the novel memory cell structures. Further, the invention is equally applicable to any size and type of memory device 1000 using the novel memory cells ofthe present invention and is not intended to be limited to that described above. As one of ordinary skill in the art will understand, such an electronic system can be fabricated in single-package processing units, or even on a single semiconductor chip, in order to reduce the communication time between the processor and the memory device.

Applications containing the novel memory cell ofthe present invention as described in this disclosure include electronic systems for use in memory modules, device drivers, power modules, communication modems, processor modules, and application- specific modules, and may include multilayer, multi chip modules. Such circuitry can further be a subcomponent of a variety of electronic systems, such as a clock, a television, a cell phone, a personal computer, an automobile, an industrial control system, an aircraft, and others.

Conclusion Utilization of a modification of well established DRAM technology and arrays will serve to afford an inexpensive memory device which can be regarded as disposable if the information is later transferred to another medium, for instance CDROM's. The high density of DRAM array structures will afford the storage of a large volume of digital data or images at a very low cost per bit. There are many applications where the data need only be written a limited number of times, the low cost of these memories will make it more efficient to just utilize a new memory anay, and dispose ofthe old memory array, rather than trying to erase and reuse these arrays as is done with cunent flash memories. The novel multistate cells can be used in a DRAM like array. Two transistors can occupy an area of 4F squared (F = the minimum lithographic feature size) when viewed from above, or each memory cell consisting of one transistor utilizing an area of 2F squared. Each such transistor can now, however, store many bits so the data storage density is much higher than one bit for each IF squared unit area. Using a reference or dummy cell for each memory transistor where the reference transistor is in close proximity, e.g., the embodiment shown in Figures 16A and 16B vs. that shown in Figure 12, results in better matching characteristics of transistors, but a lower memory density.

It is to be understood that the above description is intended to be illustrative, and not restrictive. Many other embodiments will be apparent to those of skill in the art upon reviewing the above description. The scope ofthe invention should, therefore, be determined with reference to the appended claims, along with the full scope of equivalents to which such claims are entitled.