WO1986006540A2

WO1986006540A2 - Memory cell for use in a read only memory

Info

Publication number: WO1986006540A2
Application number: PCT/US1986/000775
Authority: WO
Inventors: John Louis Janning
Original assignee: Ncr Corporation
Priority date: 1985-04-19
Filing date: 1986-04-16
Publication date: 1986-11-06
Also published as: JPS62502644A; EP0218697A1; JP2503219B2; WO1986006540A3; US4667217A

Abstract

A memory cell (30, 50) suitable for use in a read-only memory, includes a nonvolatile, alterable threshold transistor (31, 51) formed in a vertical, stacked relationship with a fixed threshold transistor (32, 52). The two transistors share a common channel structure (41) or a common gate (55). When incorporated in a ROM, selected ones of the transistors are programmed as inoperative by implanting or by the provision of thick gate dielectric, thereby providing two ROM codes constituted by first and second transistor arrays respectively. Selection between the ROM codes is effected by block write or block erase operations on the alterable threshold transistors (31, 51). In a modicifation, the transistors of each pair (31A, 31F; 51A, 51F) are arranged in side-by-side relationship in a common active region (65).

Description

MEMORY CELL FOR USE IN A READ ONLY MEMORY

Technical Field

This invention relates to memory devices of the kind including a memory cell comprising a first, alterable threshold memory device having a channel structure and a charge storage memory gate which includes a ga h electrode and a dielectric structure between the gate electrode and the channel structure, the memory gate structure having charge storage capability suitable for controlling the device threshold to a first, relatively low level and a second, relatively high level.

The invention has a particular application to memory devices configured as a read-only memory.

Background Art

Two of the critical continuing needs of the microelectronics industry are to increase device¹ packing densities and to enhance performance characteristics such as the speed of operation. The ongoing attempts to scale device sizes and associated structures has been successful to date in reducing the size and increasing the density of monolithic integrated circuits, albeit with some problems such as short channel effects.

An article by Colinge et al.. Stacked Transistors CMOS (ST-MOS), an NMOS Technology Modified to CMOS, IEEE Transactions on Electron Devices, Vol. ED-29, pp. 585-589 (1982) discloses a stacked CMOS inverter configuration. Referring to Fig. 1 of the drawings accompanying the present description, the Colinge et al. structure starts with a conventional NMOS structure formed in a silicon substrate 10, uses the NMOS gate 11 for both transistors, and adds a selectively implanted, laser-recrystallized polycrystalline silicon layer 12 which forms the PMOS source/drain (S/D) regions 13 and 15 together with intermediate channel region 14. The PMOS S/D 15 is formed in contact with NMOS S/D 16 to provide electrical interconnection for the stacked inverter. In connection with CMOS technology, the Colinge et al. structure eliminates isolation cells, decreases latchup and provides decreased wire routing complexity.

International Patent Application No. WO 84/04418 (European Publication No. 0 140 965) discloses a nonvolatile memory device formed by a gate electrode on an insulator substrate, a memory dielectric overlying the gate electrode and a recrystallized polysilicon layer having source- channel-drain regions overlying the memory dielectric. Thus, there is disclosed an inverted memory transistor which comprises from top to bottom, channel-oxide- nitride-silicoh gate electrode-insulator. The nitride is formed before the oxide, which permits forming the oxide by conversion of the nitride. The conversion process provides exceptional quality and precise, controllable reproducible growth rates for the very thin (approximately 20 Angstroms thick) memory silicon oxide. The possibility is also suggested in said international patent application of forming a common- gate vertically stacked nonvolatile memory device pair in a piggy-back configuration.

Disclosure of the Invention

It is an object of the present invention to provide a memory device including a memory cell whereby a high density of data storage may be achieved.

Therefore, according to the present invention, there is provided a memory device of the kind specified, characterized in that said memory cell further includes a second, fixed threshold device having a channel structure and a gate structure, in that said second device is formed in adjacent relationship to said first device and in that said first device and said second device have a common functional element.

It will be appreciated that a memory device according to the invention is suitable for implementation in a ROM structure.

According to one embodiment of the invention, the common functional element of the two devices is a common gate electrode.

According to another embodiment of the invention, the common functional element of the two devices is a common channel.

The alterable threshold device and the fixed threshold device may be formed in stacked relationship or in side-by-side relationship in the same active area.

In brief summary of a preferred embodiment of the invention, a bit-selectable, two bit per cell memory device includes a nonvolatile, alterable threshold memory device which shares a gate electrode structure in common with a fixed threshold device. The word "device" as used here and throughout the present patent document includes conventional field effect transistors as well as gate structures such as gated capacitive devices which may be devoid of source and/or drain. Thus, the channel region of the nonvolatile, alterable threshold memory device may comprise a source-channel-drain structure or simply a channel/electrode. The memory device includes a charge storage memory gate structure which includes a gate electrode, and a gate dielectric formed between the gate electrode and the channel, such that the memory device can be written to first and second, low voltage and high voltage threshold states. The non- memory, fixed threshold device is formed either in stacked relationship with the memory device (in which case the two devices share a common recrystallized polychannel or a common gate electrode), or in side- by-side relationship to the memory device (in which case the two devices share the same active region and can have an electrically common gate electrode structure and common source/drain regions). In addition, the fixed threshold device is fabricated to have a threshold voltage which is intermediate the low (erased) and high (written) states of the NV, alterable threshold memory device.

With this configuration and choice of threshold voltages, the logic state of the NV alterable threshold memory device can be determined by applying an Erase sequence to the NV alterable threshold memory device, followed by the application of a Read voltage the magnitude of which is between the erased threshold voltage (VTO) of the NV alterable threshold memory device and the threshold voltage of the fixed threshold device.

Similarly, the logic state of the fixed threshold voltage device is read by applying a Write sequence to the NV alterable threshold memory device, followed by the application of a Read voltage the magnitude of which is between the threshold voltage of the fixed threshold device and the written threshold voltage (VTI) of the NV alterable threshold memory device.

Those with skill in the art will appreciate that in standard ROM programming, a logic "1" or "0" is provided at a particular ROM device address by manufacturing or programming the device to be inoperable/operable. Typically, devices are rendered inoperable by the use of very heavy implants or thick gate oxide to elevate the device threshold voltage well above the operating voltage range for the particular integrated circuit. As embodied, the ROM structure uses this standard inoperable/operable programming to provide logic 0/1 states. Thus, regardless of which logic 0/1 state the memory and non-memory transistors are in, application of the Erase plus Read sequence or the Write plus Read sequence selects for reading the memory or the non- memory device (or array), respectively, to the extent that 0/1 logic data has been programmed therein. An ROM array incorporating the cell structure of the invention has the advantage of permitting an ROM array to function electrically as two separate arrays.

In other aspects, the two bit per device construction of the present invention can be implemented in various combinations of SOI, MNOS (MNOS includes SNOS), and MOS (MOS includes SOS) configurations.

Brief Description of the Drawings

Embodiments of the present invention will now be described by way of example with reference to the accompanying drawings, in which:

Fig. 1 is a schematic cross-sectional representation through one portion of a prior art, stacked CMOS integrated circuit;

Figs. 2 and 3 are schematic cross-sectional representations through portions of respective stacked integrated circuit structures which are alternative embodiments of the present invention;

Figs. 4 and 5 are schematic plan view outlines of shared active region integrated circuit structures which are alternative embodiments of the present invention;

Figs. 6 and 7 are cross-sections taken along the lines 6-6 and 7-7 through devices in Fig. 5;

Figs. 8 and 9 are cross-sections taken along lines 8-8 and 9-9, respectively, through devices in Fig. 6; and Fig. 10 schematically illustrates a memory array circuit.

Best Mode for Carrying Out the Invention

Referring to Fig. 2, there is shown a first embodiment 30 of a twobit per cell structure in the form of an inverted, NVAT (non-volatile, alterable threshold) SOI memory transistor 31 and an overlying fixed threshold transistor 32. The transistors 31 and 32 have a common recrystallized polysilicon source- channel-drain structure. Specifically, the SOI NVAT memory transistor 31 is constructed in accordance with the SOI structure and fabrication sequence which are described in detail in the above-identified, international patent application No. WO 84/04418. The SOI structure includes a silicon substrate 33, an insulator dielectric layer 34, and the inverted trigate -memory configuratidn including^'polysilicon gate electrode 35 (poly I), nitride layer 36, oxide layer 37 and the recrystallized polysilicon layer 38 (poly II) which forms the source and drain 39-39 and channel region 41. The fixed threshold transistor 32 comprises the gate electrode 42 which is formed from a third layer of polysilicon (poly III), gate oxide layer 43 and the underlying source 39-channel 41-drain 39 which is common to device 31 as well.

The trigate SOI configuration of device 31 includes a very thin memory oxide section 37M (about 1-4 nanometers) formed central to the channel and the thicker sections 37 which flank memory oxide 37M at the source and drain sides thereof to eliminate the instabilities and threshold shifts described in U.S. Patent No. 3,719,866. Those with skill in the art will appreciate that while the trigate construction is preferred, dual gate and monogate constructions can be used in some applications. In the dual gate configuration, a single thick oxide region 37 is formed over the source or drain whereas the thick source/drain sections 37 are not used in the monogate configuration.

In fabricating the device 30, the starting material is a wafer or substrate 33 of material having a low coefficient of thermal expansion which is comparable to that of the dielectric material such as silicon dioxide or silicon nitride which is used in the device. Suitable materials include silicon, high temperature glass, and ceramics such as alumina (aluminum oxide) . Insulator 34 electrically isolates the memory device from the substrate 33 and the peripheral circuitry and typically comprises silicon dioxide (or silicon nitride) formed to a thickness of about 1,000 nanometers by chemical vapor deposition or, where the substrate 33 is silicon, by high temperature oxidation of the silicon. , - Next,, the first layer of polysilicon (poly I) is formed about 300 nanometers thick by conventional process techniques such as low pressure chemical vapor deposition (LPCVD) followed by delineation and patterning using conventional photolithographic and etching techniques to form the gate electrode 35. The polysilicon gate electrode can be doped in situ or by subsequent ion implantation.

Next, the nitride layer 36 is conformally deposited to a thickness of about 40 nanometers using a conventional LPCVD (silane/ammonia system; 750°C; 400 T) operation, followed by deposition of the oxide layer 37 to a thickness of about 70-80 nanometers using low temperature LPCVD (silane/oxygen system; 420°C; 300 mT) operation. Thereafter, the central memory oxide section is removed using conventional photolithographic and etching techniques, and a new oxide is thermally grown or deposited to ensure that section 37 M has thickness of about 1-4 nanometers. Suitable techniques include chemical vapor deposition. or conversion of the exposed region of nitride 36 by oxidation (wet oxygen; 1000°C; 30 minutes) into the oxide section 37 M.

Immediately after forming the memory oxide 37 M, and to avoid contamination of the memory oxide 37, a second polysilicon layer 38 (polysilicon II) is formed to a thickness of about 450-500 nanometers, again using conventional techniques such as LPCVD. The poly II layer 38 is then lightly doped by ion implantation (for example, with boron at 35keV and a dose of 1E12-2E13 ions/sq cm.) to provide the requisite channel conductivity. The poly II layer 38 is then capped with an anti-reflective coating of nitride (not shown) which is formed about 40-45 nanometers thick by conventional processing such as LPCVD. Next, the poly II layer 38 is exposed to a laser beam (for example, a continuous wave argon laser having an effective beam diameter of 45 micrometers, a beam power of 5 watts, and a scanning speed 200 cm per/sec.) to convert the polysilicon to device-quality material comprising a recrystallized matrix of crystallites having various crystal orientations. The nitride cap (not shown) is then removed using concentrated hydrofluoric acid and the recrystallized poly II layer is patterned to the source 39-channel 41-drain-39 configuration shown in Fig. 2.

Next, the fixed threshold transistor 32 is formed by depositing an oxide layer 43 to a typical thickness of 50 nanometers. A silane/oxygen LPCVD technique of the type described previously can be used here as well. Subsequently, the polysilicon gate 42 layer (poly III) is formed, typically to a thickness of approximately 300 nanometers using the same formation technique (LPCVD) and the same implantation doping techniques described relative to poly gate electrode 35. The structure is photolithographically processed to mask polysilicon gate 42 so that successive etching of gate 42 and oxide 43 provides for self-aligned implanting of source and drain regions 39-39 with n-type impurity.

After removal of the implant mask, a low temperature oxide layer 44 about 900-1,000 nanometers thick is blanket deposited over the structure and is densified at a temperature of about 900°C in nitrogen. The densification step is also used to activate the ions implanted in the source and drain regions. Thereafter, the device can be completed, for example, by making contact cuts to the source and drain 39-39 and the gates 35, 42 through the oxide 44, followed by contact enhancement using phosphorus oxychloride (POCI3) deposition and thermal diffusion, followed by metallization and the formation of a passivation layer. The resulting metal contacts and passivation layer are not shown in Fig. 2, but electrode connec¬ tions are illustrated schematically by vertical lines connecting to the drain 39-39 and the gates 35 and 42.

The cell 30 shown in Fig. 2 is thus a non- alterable-over-an-alterable transistor configuration with a common source/channel/drain structure. One of the key aspects of this design and the associated operation is the dual function of the SOI nonvolatile alterable threshold device 31. That is, device 31 acts first as a switch to select between the alterable threshold device 31 and the fixed threshold device 32 for reading and, secondly, functions as a memory location to provide a logic 0/1 output.

For purposes of programming binary states into the nonvolatile alterable threshold device, the threshold voltages of the transistor 31 is conveniently adjusted to an inoperatively high threshold (logic 0) or to an operative threshold (logic 1) by adjusting the gate oxide thickness or by implantation. The same is true for programming the binary state represented by fixed threshold device 32. Also, the nonvolatile alterable threshold devices 31 selected to potentially represent the logic 1 state are adjusted by implant to provide an erased threshold voltage of -1 volt (VTO) and a written threshold voltage of +4 volts (VT1) . Using these examples of VT0/VT1 threshold voltage levels and an operative fixed threshold device with a +1 volt- threshold, two bits of information can be obtained from the single cell 30 by a simple addressing scheme, which is summarized in Table 1. The addressing approach assumes that logic 1/0 for each device 31 and 32 is determined by an operative/inoperative device at that location.

TABLE 1

(VTf = +1V)

ERASE/WRITE READ DEVICE PROGRAM VOLTAGE SELECTED VTO (-1V) 0V NV Alterable Threshold FET 31

VT1 (+4V) +2V Fixed Threshold FET 32

Referring to Table 1, the first sequence is designed to select the alterable threshold transistor 31 for reading and involves (a) an Erase operation which is designed to write the memory transistor 31 (if operative) to a low threshold, VTO state, followed by (b) a sense operation using a 0 volt Read voltage. (The Read voltage is keyed to the above-mentioned threshold voltages of VTO = -1 volt; fixed threshold FET Threshold of VTf = +1 volts; VT1 = +4 volts). Using this Erase plus 0 volt Read sequence, if the device 31 is (VTO = -1 volt), the device 31 conducts current, and this can be sensed as logic 1 by a standard sensing amplifier circuit. If, however, there is no operative device at that location, cell 30 does not conduct current, which is sensed as logic 0 at the cell 30 address. Note that with VTf at +l/volt, no conduction would be sensed even if gate electrodes 35 and 42 were made common.

Conversely, and referring further to Table 1, the fixed threshold transistor 32 location is selected by the application of (a) a Write program operation which is designed to write the alterable threshold memory transistor 31 to the high, VT1 threshold state, followed by (b) a Read voltage such as +2 volts, which voltage lies between VTf and VT1. Using this Write plus +2 volt Read sequence, the alterable threshold transistor 31 does not conduct current. (The binary state of the two transistor cell 30 is determined by whether or not transistor 31 is operative.) Logic 1/0 is then determined by the existence/nonexistence at the particular address location of an operative fixed threshold transistor 32 having a +4 volt threshold voltage.

As will be known to those familiar with the nonvolatile alterable threshold device technology, the transistor 31 is erased to VTO, for example, by applying a large (20-25 volts; 1-100 ms pulse duration) negative program voltage to the gate electrode 35 with respect to the channel to drive to the channel electrons which are trapped in the gate dielectric at the oxide-nitride interface and in the nitride bulk. Conversely, the transistor 31 is written to the high threshold, VT1 state, by applying a large (20-25 volts; 1-100 ms pulse width) positive program voltage between the gate 35 and the channel (with the source and drain being maintained at ground potential) to tunnel electrons into the gate dielectric for storage. Electrode connections for writing and erasing the transistor 31 are illustrated schematically at 35P and 39P, Fig. 2. ^"Of course, the other NV alterable threshold devices described herein are written/erased similarly. Another version 50 of the alterable threshold transistor-fixed threshold transistor cell is shown in Fig. 3, in the form of a trigate SNOS transistor 51 having an upside down SOI fixed threshold transistor 52 formed thereon. This SNOS/SOI cell 50 is configured with a single gate electrode 55 and separate source-channel-drain structures.

Typically the SNOS alterable threshold device 51 is formed using a <100> silicon substrate 53. The trigate dielectric comprises trigate memory oxide 57 and silicon nitride layer 56. The silicon dioxide layer can be formed to a thickness of about 70-80 nanometers using, for example, the same process which is used to form the memory silicon dioxide 37, Fig. 2. The oxide is then selectively removed from the central, memory region, for example, by etching with buffered hydrofluoric acid, and the memory oxide 57M is grown or deposited to a thickness of about 1-4 nanometers. This memory oxide layer can be grown by dry thermal oxidation of the silicon substrate, or can be formed on the substrate 53 by chemical vapor deposition using the same process that was used for the oxide layer 57. After forming the gate oxide 57M, the silicon nitride layer 56 is deposited to a thickness of about 40 nanometers by conventional LPCVD. The polysilicon gate electrode 55 is then formed to a thickness of about 300 nanometers, typically using the same process which was used to deposit, delineate, pattern, and implant the gate 42 of Fig. 3.

As shown in Fig. 3, the oxide 57 and nitride 56 are removed from the regions outside the gate region. This can be done prior to the gate implant, using the same etching mask which is used to delineate and pattern the gate 55, or separately after the gate implant. Next, the polysilicon electrode 55 is used as an implant mask and the source and drain 54-54 are formed self-aligned with the gate 55. This completes the SNOS device 51.

Next, the gate silicon dioxide layer 58 for the SOI fixed threshold device 52 can be formed to a thickness of about 50-55 nanometers using the same process which was used to form the fixed threshold oxide 43 of device 32, Fig. 2. A polysilicon layer 59 (polysilicon II) is formed over the existing structure to contact the source and drain 54-54, and is recrystallized using a laser beam as described^' previously for poly layer 38 (Fig. 2).

Next, the source 60-channel 61-drain 60 for the fixed threshold device 52 is selectively implanted, the source-drain 60.-60 being implanted with same conductivity type impurity as the source-drain 54-54. The same formation, patterning,^* recrystallization and implant technique that was used for the recrystallized poly layer 38 of Fig. 2, can be used. The cell 50 is then completed, for example, by forming and densifying an intermediate silicon dioxide isolation layer, making contact cuts, and defining metallization, followed by the application of a passivation layer. Here, the contacts (shown schematically extending downwards for reasons of clarity) are made to the single polysilicon gate electrode 55 and to source/drain electrodes 54.

Two additional alternative versions 70 and 90 of the two bit/cell structure are shown in schematic top views, in Figs. 4 and 5, respectively. The two bit/cell 70 of Fig. 4 contains NV alterable threshold device 31A and fixed threshold device 31F. The MOS/MNOS two bit/cell 90 of Fig. 5 contains NV alterable threshold device 51A and fixed threshold device 51F. It should be noted that devices 31A and 31F are similar to device 31 of Fig. 2 and devices 51A and 51F are similar to device 51 of Fig. 3, except that, preferably, devices 31F and 51F do not use a nitride layer such as 36 (Fig. 2) or 56 (Fig. 3).

Accordingly, device components in transistors 31A or 31F (Figs. 4, 6, 7) and transistors 51A or 51F (Figs. 5, 8, 9) which are identical to those for devices 31 (Fig. 2) and 51 (Fig. 3) are designated by the same reference numerals used for devices 31 and 51. Modified components are indicated by the respective suffix A or F in transistors 31A or 31F and 51A or 51F. In each cell, the fixed threshold transistors 31F or 51F and the alterable threshold transistor 31A or 51A are formed side-by-side in a common active area 65 of the integrated circuit. In each cell 70 or 90, the source (and the drain) 39 or 54, respectively, of both devices is part of the same diffusion region or is connected in common. Also, the same gate line (or an electrically common gate line) serves both devices of the cell. The operation of the cells 70 and 90 is identical to that of devices 30 and 50, described above.

Referring now to Fig. 6 in addition to Fig. 4, the NV alterable threshold transistor 31A of cell 70 has the same construction as the NV alterable transistor 31 of Fig. 2. Similarly, and referring to Fig. 7 in addition to Fig. 4, the fixed threshold transistor 31F preferably has the same construction as transistor 31 of Fig. 2, with the exception that the gate oxide 37F of fixed threshold device 31F is of a uniform non-memory thickness and the gate nitride is omitted.

The basic fabrication sequence for the upside down device 31A was described for the corresponding device 31 of Fig. 2. The fixed threshold device 31F and the memory device 31A can be formed using a sequence in which the nitride 36 is formed everywhere then removed from the non-memory device areas 31F; the non-memory oxide 37F is formed everywhere; then the non-memory oxide is etched away in the memory regions of devices 31A, and memory oxide 37M is grown.

Considering now the side-by-side cell 90 of Fig. 5, the alterable threshold device 51A shown in Figs. 5 and 8 can have the same SNOS construction as the alterable threshold device 51 of Fig. 3. Although the fixed threshold device 51F shown in Figs. 5 and 9 can have essentially the same SNOS construction as the transistors 51 (with the exception that the threshold of device 51F is non-alterable).

The basic fabrication sequence for the alterable threshold device 51 of Fig. 3 can be used to form the alterable threshold device 31A of Fig. 8 and the fixed threshold device 51F of Fig. 9. The dielectric sequence is basically the reverse of that used for side-by-side devices 31A and 31F, described above. First,- the non-memory oxide 57F is formed everywhere; then the non-memory oxide is etched away in the memory regions of the devices 51A and the memory oxide 57M is grown there; and the nitride 56 is deposited everywhere, then removed from the non-memory device areas 51F.

The cells 30, 50, 70, and 90 can be formed into a memory pattern such as a ROM (read-only memory) array, as alluded to previously. Consider now such a dual ROM array built with the piggyback devices 30 (Fig. 2), or 50 (Fig. 3), or the side-by-side devices 70 (Fig. 4), or 90 (Fig. 5), in which each ROM array forms a separate pattern, such as a ROM pattern used to effect a different ROM code. For example, the channels of the alterable threshold transistors can be implanted selectively in the particular 1/0 pattern of a code 1 in which the operable logic 1 devices can be electrically written to either VTO or VTl, whereas the logic 0 device sites are programmed to inoperatively high threshold voltages. Likewise, the fixed threshold voltage transistors are programmed to the particular 1/0 pattern of a code 2, in which the fixed threshold voltage is +4 volts/inoperative. Then, by applying a block ERASE sequence to the array of alterable threshold devices and addressing the array with 0 volts the ROM is selected to code pattern 1. When code pattern 2 is desired, a WRITE sequence is applied to the alterable threshold array and the ROM is addressed with +2 volts. Since the alterable threshold devices cannot respond to the +2 volts in the written state, VTl = +4 volts, the upper fixed threshold transistors are selected. Thus, simply by writing or erasing all alterable threshold devices simultaneously, either memory bank can be addressed.

The versatility of the present bit-selectable concept is further exemplified by the program selectable AND-OR.ROM circuit 80 which is shown in Fig. 10.. The^" figure illustrates the location of the NV alterable threshold devices, device location A, and the fixed threshold devices, device location B, as well as various examples of threshold voltage combinations for the devices.

For device locations A, the NV alterable threshold device location, the notation "-1, +4" indicates that an operable memory device exists at that location which is switchable between VTO and VTl. That location is binary "1" in the NV alterable threshold code pattern. See, for example, device location A at address gate line 1-sense line 1 (i.e., address 1-1). In contrast, the notation "+4" in the A location at address 1-2 indicates there is no operable memory device at that location. Instead, device A has a fixed threshold of +4 volts. That location is binary "0" in the NV alterable threshold code pattern; the A pattern.

In fixed threshold device location B, the notation "+1" denotes the existence of an operable fixed threshold device (operable at the selected operating voltage conditions) which has an exemplary threshold of +1 volts, and binary "1". See, for example, addresses 1-1 and 2-2. Conversely, the notation "+4" at device locations B indicates an inoperable/non-existent fixed threshold device, and binary "0" in the B code pattern.

Program selection and reading are also accomplished as described previously. For example, to read location 1-1 (gate line 1, sense line 1), device location -A, an erase sequence is applied to erase an operable device to VTO (-1 volt), then the address is read by grounding the gate line 1 (0 volts) with gate lines 2-4 at +5 volts and sense line 1 at the appropriate system voltage such as +5 volts. As described previously, the READ sequence turns on an operable memory device,^* which is at VTO, (binary 1), but not an inoperable one (binary 0).

To illustrate the READ sequence for device location B, consider again address 1-1. Initially, a WRITE sequence is applied to device location A so that this location will not conduct. Then, a READ voltage of +2 volts, rather than 0 volts, is applied to selected gate line 1, again with +5 volts on non- selected gate lines 2-4 and with the appropriate voltage on sense line 1. For an operative fixed threshold device, VTf = +1 volt, and device B conducts current, which can be sensed as binary 1. For an inoperative device B location, location B does not conduct current, representing binary "0". Of course, the same type of procedure applies for reading the A or B locations at other addresses.

Thus, there has been described preferred and alternative constructions for the present invention along with process descriptions for each structure. For instance, it is foreseeable that the alterable threshold device could be a volatile device which has been coded by the transfer of charge to the threshold altering element of the device in a cell. In this regard, it is also contemplated that the volatile alterable threshold device so coded is subject to change by selective transfers of charge.

Claims

1. A memory device, including a memory cell (30, 50) comprising a first, alterable threshold memory device (31, 51) having a channel structure and a charge storage memory gate (35, 36, 37M; 55, 56, 57M) which includes a gate electrode (35, 55) and a dielectric structure (36, 37M; 56, 57M) between the -) gate electrode (35, 55) and the channel structure, the memory gate structure (35, 36, 37M; 55, 56, 57M) having charge storage capability suitable for controlling the device threshold level to a first, relatively low level and a second, relatively high level, characterized in that said memory cell (30, 50) further includes a second, fixed threshold device (32, 52) having a channel structure (41, 61) and a gate structure (42, 43; 55, 58), in that said second device (32, 52) is formed in adjacent relationship to said first device (31, 51), and in that said first device (31, 51) and said second device (32, 52) have a common functional element (41, 55).

2. A memory device according to claim 1, characterized in that said common functional element is a common gate electrode (55).

3. A memory device according to claim 1, characterized in that said common functional element is a common channel structure (41).

4. A memory device according to any one of claims 1, 2 or 3, characterized in that said memory cell (30, 50) has a vertically integrated structure, with said first device (31, 51) and said second device (32, 52) being formed one on the other.

5. A memory device according to any one of claims 1, 2 or 3, characterized in that said memory cell (30, 50) has a horizontally integrated structure, with said first device (31A, 51A) and said second device (31F, 51F) being formed in the same active area (65) of an integrated circuit.

6. A memory device according to claim 1, characterized in that first alterable threshold memory device (31, 51) is nonvolatile.

7. A memory device according to any one of the preceding claims, characterized in that said device is configured in the form of a read-only memory and including a plurality of said memory cells (30, 50) wherein selected ones of said first and second devices (31, 51, 32, 52) are configured as operable devices, the remaining ones of said first and second - devices being configured as inoperable devices, whereby a first read-only memory code is stored in said first devices (31, 51) and a second read-only memory code is stored in said second devices (32, 52).

8. A memory device according to claim 7, characterized in that selection between said first and second read-only memory codes is effected by block erase and block write operations effected on said first, alterable threshold devices (31, 51).