Row decoding circuit for a semiconductor non-volatile electrically programmable memory and corresponding method

Abstract

The invention relates to a row decoding circuit for an electrically programmable and erasable semiconductor non-volatile storage device of the type which includes a matrix of memory cells laid out as cell rows and columns and is divided into sectors, said circuit being input row decode signals and supply voltages in order to drive an output stage incorporating a complementary pair of high-voltage MOS transistors of the pull-up and pull-down type, respectively, which are connected to form an output terminal connected to the rows of one sector of the matrix, characterized in that a MOS transistor of the P-channel depletion type with enhanced gate oxide is provided between the output terminal and the pull-down transistor. The control terminal of the depletion transistor forms a further input of the circuit.

Description

BACKGROUND OF THE INVENTION

1. Field of the Invention

The present invention relates to a row decoding circuit and method for an electrically programmable semiconductor non-volatile memory.

More particularly, the invention relates to a row decoding circuit for a semiconductor non-volatile storage device which is programmable and erasable electrically and of the type including a matrix of memory cells laid out as rows and columns of cells and divided into sectors, the circuit being input row decode signals and supply voltages in order to drive an output stage which incorporates a complementary pair of high-voltage MOS transistors, respectively of pull-up and pull-down, connected to form an output terminal for connection to the rows in a sector of the matrix.

2. Discussion of the Related Art

As is well known, read-only memories of the flash type have made substantial inroads as non-volatile memories because they can combine the advantages of high density that typify EPROMs with the capability of being programmed and erased electrically in the same way as EEPROMs.

Memories of this kind are constructed as matrices of cells, wherein each cell includes a floating gate transistor. Each transistor has a drain region and a source region which are formed on a semiconductor substrate and separated by a channel region.

A floating gate is formed over the substrate and isolated therefrom by a thin layer of gate oxide. A control gate is coupled capacitively to the floating gate by means of a dielectric layer. Metallic electrodes are arranged to contact the drain, source, and control gate such that predetermined voltage values can be applied to the memory cells.

The cell matrix is laid out into rows, or word lines, and columns, or bit lines. The cells of one word line share the electrical line that drives their respective control gates, while the cells of one bit line have their drain terminals in common.

By applying suitable voltage values to the cell terminals, the amount of the charge present in the floating gate can be varied, and since the floating gate has a high impedance toward any other terminals of the cell, the charge stored in it can be maintained indefinitely, even when the supply is cut off to the circuit in which it is included.

The operation whereby a charge is stored into the floating gate is called "programming" and consists of biasing the drain terminal and control gate at a predetermined higher value than the potential of the source terminal.

This programming of the individual cells is carried out by a hot electron injection process which allows electrons to become trapped inside the floating gate when the control gate is applied approximately 12 volts, with the source terminal being connected to ground and the drain terminal held at a drain voltage of about 5.5 volts.

In order to erase a flash memory cell, its source terminal is applied a voltage approximately equal to the supply voltage Vcc, the drain terminal is held floating, and the control gate is either led to ground or biased with a negative voltage.

In the field of application of the present invention, the negative voltage being applied to the decoding circuit during the steps of erasing the contents of the memory cells requires appropriate control.

A prior technical solution for properly controlling this negative voltage is described in U.S. Pat. No. 5,126,808, (hereinafter the "808 Patent") which discloses a structure of a flash EEPROM matrix and respective erasing circuitry by the sector.

While being in many ways advantageous, this prior solution is not devoid of drawbacks, as detailed herein below.

The row decoding circuit structure described in the '808 Patent provides for a final stage which includes a pull-up transistor and a pull-down transistor placed on opposite sides of an output terminal being connected to the word lines of the memory matrix.

Connected between the output terminal and the pull-down transistor is a P-channel transistor whose control terminal is applied a negative voltage during the step of erasing the memory cells.

This negative voltage must be supplied from a dedicated negative charge pump to allow the word lines to be driven to ground during the erasing step.

The provision of such a charge pump unavoidably involves some consumption of circuit area as well as of current, since the pump must be kept under power even while in the standby state to ensure a fast time of access to the memory upon exiting the standby state.

In addition, the input stage of the decoding circuit described in the '808 Patent includes a NAND portion incorporating P-channel transistors, which burdens the circuit construction.

Finally, no provision is made in this prior solution for an accurate regulation of the row decoding supply voltages to make the unselected sectors of the matrix fully independent of the operations being performed on a given selected sector.

An object of the present invention is to provide a novel row decoding circuit and method which have such constructional and functional features as to overcome the aforementioned drawbacks of conventional solutions.

SUMMARY OF THE INVENTION

An embodiment of the invention is directed to a row decoding circuit for an electrically programmable and erasable semiconductor non-volatile storage device of a type having a matrix of memory cells laid out as rows and columns. The matrix is divided into sectors such that each sector has a corresponding plurality of rows. The row decoding circuit includes an input stage that receives input row decode signals and supply voltages, and an output stage that is driven by the input row decode signals and the supply voltages. The output stage has a complimentary pair of high-voltage MOS transistors including a pull up transistor and a pull down transistor. The pull up and pull down transistors are connected to form an output terminal. The output terminal is connected to a plurality of rows corresponding to a sector of the matrix. The output stage further includes a MOS transistors of the P channel depletion type with enhanced gate oxide. The MOS transistor is disposed between the output terminal and the pull down transistors.

Another embodiment of the invention is directed to a row decoding method for an electrically programmable and erasable semiconductor non-volatile memory of a type which includes a matrix of memory cells laid out as cell rows and columns and is divided into sectors. The memory receives as inputs row decode signals and supply voltages in order to drive an output stage incorporating a complimentary pair of high voltage moss transistors of the pull up and pull down type, respectively, which are connected to form an output terminal connected to rows of one sector of the matrix. The method includes a step of operating a plurality of local supply switches. Each local supply switch is associated with a decoding circuit of a matrix sector.

The features and advantages of the circuit and the method of the present invention will be apparent from the following detailed description of embodiments thereof, given by way of non-limitative examples with reference to the accompanying drawings.

BRIEF DESCRIPTION OF THE DRAWINGS

In the drawings:

FIG. 1 is a diagrammatic representation of the circuit structure of a row decoding circuit according to the invention;

FIG. 2 is another diagrammatic representation of an integrated storage device incorporating row decoding circuits according to the invention;

FIG. 3 is a diagram of a particular supply circuit according to the invention and intended for co-operation with the decoding circuit of FIG. 1;

FIGS. 4 and 5 are respective plots vs. time of voltage and current signals appearing in the decoding circuit according to the invention; and

FIG. 6 illustrates diagrammatically the connection of a row decoding circuit according to the invention to a row of the storage device with which the circuit is associated.

DETAILED DESCRIPTION

Referring in particular to the example of FIGS. 1 and 2, generally and schematically shown at 1 is a row decoding circuit for electrically programmable/erasable non-volatile memory cells, such as cells of the flash EPROM type.

Each cell includes a MOS transistor having a floating gate G and a further control gate CG. The cells are laid out as a matrix 2 having rows WL, or word lines, and columns, or bit lines. All the cells of one word line WL share the electrical line which drives their respective control gates CG, while all the cells of one bit line have their drain terminals D in common.

The circuit 1 is particularly, but not solely, intended for a flash memory with a single power supply at a voltage Vcc of 5 volts. The invention has been applied to a memory made with double-well technology for erasing by a negative voltage to the cell gates.

In one embodiment, the cell matrix 2 has a capacity of four Megabytes, and is divided into eight sectors of 512 kilobytes, each with memory words of eight bits. Thus, the matrix 2 has eight decoding circuits according to the invention associated therewith, one per sector of the matrix as shown in FIG. 2.

The circuit 1 of the invention has a plurality of inputs and a control output U for the word lines of the matrix 2.

A first input A coincides with the drain terminal of a high-voltage pull-up transistor M2 of the P-channel MOS type. This transistor M2 has its source terminal connected to a circuit node gate.sub.-- dr, its body terminal connected to the input A, and a control terminal B which coincides with a second input of the circuit 1.

Four N-channel MOS transistors are connected in series, between the circuit node gate.sub.-- dr and a signal ground GND of the circuit 1, and indicated at M12, M6, M17 and M8, in this order. They form essentially a row decoding portion 9 of the NAND type. The first, M12, of these transistors is of the high-voltage type.

Each of the transistors has a control terminal of its own which is coincident with an input of the circuit 1. The respective inputs C, D, E and F of the circuit 1 are shown which receive row decode signals p, ly, lx and ls, respectively.

Connected between the circuit node gate.sub.-- dr and the output U of the circuit is an output stage 8 which includes a high-voltage P-channel MOS transistor M15 having its control terminal connected to the node gate.sub.-- dr, its drain and body terminals connected to the input A, and its source terminal connected to the output U. The transistor M15 acts essentially as a pull-up transistor between the input A and the output U. Advantageously, a pair of high-voltage transistors, as respectively indicated at M9 and M13, are placed between the output U and the ground GND. The transistor M13 acts as a pull-down transistor, and can be regarded as the complement of the transistor M15 in the output stage 8.

In essence, the transistor M9 is connected between the word line and the pull-down transistor of the decoding circuit. This transistor M9 is a P-channel depletion transistor having the gate oxide thickness increased to at least 330 Angstroms.

The control terminal of the transistor M9 forms an input H of the circuit 1. The body terminal of the transistor M9 is in common with the body terminal of the transistor M15.

The second transistor M13 in the complementary pair is an N-channel pull-down MOS, and has its control terminal connected to the node gate.sub.-- dr, its source terminal to ground, and its drain terminal to the source terminal of the transistor M9.

Referring now to the example of FIG. 6, it can be seen that the output stage 8 of the circuit 1 is connected to the rows or word lines of one sector of the matrix 2 through the output U being connected directly to the control gate CG terminals of the cells.

Advantageously, a row diode 25 is placed at the end of each word line which is connected, on the one side, to the row itself, and on the other, to a node X adapted to receive a negative voltage Vpump from a negative charge pump 26.

According to an embodiment of the invention, the diode 25 includes a high-voltage P-channel MOS transistor of the "nofield" type, in a diode configuration. This transistor has gate oxide of increased thickness compared to a standard transistor, and has a gate surrounding the active areas.

The diode 25 is located within a well which is biased by a signal W1, which is normally held at a signal Vpcxs and forced to ground at the matrix sectors being erased.

The use of a high-voltage diode 25 provides two advantages in that it has a higher breakdown voltage when reverse biased, and that it allows higher negative voltages to flow through the memory cell gates when operated at a low supply voltage Vcc.

The node X is driven at a negative voltage from the pump 26 during the erasing step only, and is otherwise held at the voltage Vpcxs by a local switch 16 to be described.

The operation of the circuit 1 according to the invention will now be described.

The decoding circuit 1 sends appropriate signals to the word lines of the matrix 2, according to row decode signals and other signals applied thereto.

The input A is applied a voltage signal Vpcxs, which is then transferred to the circuit node gate.sub.-- dr through the pull-up transistor M2. This transistor M2 is to keep the unselected rows grounded, both during the reading step and the step of modifying or verifying the contents of the cell. Under these conditions, in fact, the voltage Vpcxs is higher than the supply voltage Vcc.

The second input B receives a signal pgate from a regulator 30, to be described with reference to FIG. 3.

The technology currently in use to make integrated storage circuits requires that the aforementioned negative erase voltage be only applied to P-channel transistors formed in N wells biased at voltages equal to or higher than zero volts.

In fact, N-channel transistors are instead formed in P wells connected to ground, and would be forward biased if a negative voltage were applied to them.

By providing the P-channel transistor M9 of the depletion type, a voltage Vdep within the range of 0 volts to supply voltage Vcc can be applied to the input H.

Specifically, Vdep takes the value of Vcc during the erasing step, when a negative voltage must be applied to the gates of the memory cells.

Thus, the presence of the depletion P-channel transistor M9 has a major advantage in that it avoids the need for a negative charge pump which would take up circuit area and draw current even while the circuit is in the standby state.

Also, the wait time for bringing the input H to a negative voltage after a power-up is suppressed.

The supply voltage Vpcxs to the decoding circuit 1 advantageously assumes different values in the course of the steps for modifying or verifying the state of the memory cells. These values may be higher or lower than, or equal to, the supply voltage Vcc.

The different voltage values are generated by a voltage regulator having a charge pump associated therewith.

The regulator/charge pump assembly can be regarded as an ideal voltage generator having series resistance of about 15 kiloohms.

It matters that the switching from one voltage value to another should take place within the shortest possible time, e.g. within microseconds.

For this purpose, a decoding circuit according to the invention has been associated with each of the eight sectors of the matrix 2, which circuit is independently supplied a locally switched voltage.

In accordance with the invention, an arrangement for switching the row decode supply is provided. This arrangement includes a main switch 12 (Global Switch) and a set of eight local switches 16 (Sector <i> Switch).

The main switch 12 is connected to the supply voltage reference Vcc and the voltage generator 20. The main switch 12 is supplied a voltage Vx.sub.reg and a voltage value Vp.sub.cxh from the generator 20. Advantageously, the input node of the generator 20 which receives the voltage value Vx.sub.reg only sees some of the parasitic capacitance of the decoding circuit 1.

The value Vp.sub.cxh is a logic signal for enabling the generator 20, and a select signal of the main switch 12. When the value Vp.sub.cxh equals Vcc, the charge pump and regulator are powered up to bring the output from the main switch 12 to a value Vp.sub.cx =Vx.sub.reg.

The individual local switches 16 are input both the supply voltage Vcc and the output voltage signal Vp.sub.cx from the main switch 12.

The following Tables show the voltages supplied to the local switches 16 according to the operations to be performed on the memory cells, namely: reading, writing or erasing.

TABLE 1______________________________________GLOBAL SWITCH operation Vpcx______________________________________ read Vcc erase Vcc modify/verify Vxreg______________________________________ TABLE 2______________________________________SECTOR SWITCH Vpcxsoperation selected sector not selected sector______________________________________read Vpcx Vccerase Vpcx Vccmod/vfy Vpcx>Vcc+Vt Vpcx Vcc Vpcx<Vcc+Vt Vpcx Vpcx-Vt______________________________________

In Table 2, the voltage Vt represents the threshold for an N-channel transistor.

The selection of each switch 16 is effected through an eight-wire bus 14, through which an enable signal SEL.sub.-- SECT<0:7> is transmitted.

When the signal SEL.sub.-- SECT<i> has the same value as Vcc, the i-th switch 16 is selected. Two different switches 16 cannot be actuated simultaneously. Basically, they can only be actuated in a mutually exclusive manner.

Each local switch 16 outputs a voltage signal Vpcxs<i>, which is input directly to the corresponding row decoding circuit 1. As previously explained, the circuit 1 receives the voltage signal Vpcxs from the corresponding switch on the drain terminal of the pull-up transistors M2 and M15.

The solution adopted here has a major advantage in that the unselected sectors are left off, and are unaffected by operations being carried out on the selected sector.

In addition, the charge pump incorporated to the regulator 20 can be made small, and yet ensure a switching time on the order of microseconds.

The construction of the voltage regulator 30 producing the value pgate will now be described in detail with particular reference to the example of FIG. 3.

The regulator 30 includes a current generator 21 which is independent of the voltage Vpcxs being supplied to the circuit 1.

The generator 21 includes a set of transistors M52, M49, M54 and M40. These transistors are all of the high-voltage P-channel MOS type. The transistor M54 is of the depletion type and has its source terminal forced to a voltage value of 0 volts.

The transistors M52 and M49 have their respective control terminals connected together into a first current mirror. A similar connection is provided between the transistors M54 and M40.

The remaining transistors in the regulator 30 are basically enable transistors which receive actuation signals on their control terminals.

More particularly, the regulator 30 has an input terminal K which is applied the voltage signal Vpcxs, and an output terminal U1 for generating the signal pgate. Also connected to this input K are the body terminals of the transistors M52, M49 and M40.

A pair of transistors M38 and M59 in a diode configuration are cross-connected between the input terminal K and the drain terminals of the transistors M52 and M49 of the generator 21.

A series of two N-channel transistors M57 and M39 are placed between these drain terminals and the ground GND.

The control terminals of the transistors M57 and M39 are applied respective control signals sel.sub.-- sect and Vpcxh.

The control signals sel.sub.-- sect and Vpcxh also are applied respectively to the respective control terminals of a pair of N-channel transistors M46 and M51 connected in parallel with each other between the drain terminals of transistors M52 and M49 and the ground GND.

Likewise, the control signals sel.sub.-- sect and Vpcxh also are applied respectively to the respective control terminals of a pair of N-channel transistors M56 and M53 which are connected in parallel between the output terminal U1 of the regulator 30 and the ground GND.

With both control signals sel.sub.-- sect and Vpcxh active, the output terminal U1 is driven by the regulator 30. In all of the other cases possible, the terminal U1 is held at ground value.

The depletion transistor M54 has a null voltage drop across its gate and source, and sets a constant current in the first leg of the generator 21. This current is mirrored and multiplied by two by the transistor M49 in the other leg of the generator 21.

The output pgate is picked up by the control terminal of the first current mirror M52, M49 of the generator 21. Consequently, the generator 21 will present a negative feedback that makes it little responsive to variations in the voltage Vpcxs.

In fact, when the voltage Vpcxs is increased, the voltage value at the control terminal of the transistor M52 must also increase because the current in that leg is impressed by the depletion transistor M54.

Also, should the voltage pgate go up, the gate-source voltage drop of the transistor M40 would tend to increase, but this is prevented by the current in that leg of the generator 21 being constant and set to be twice that in the other leg. Accordingly, the voltage at the control terminal of the transistor M40 is bound to go up.

A similar reasoning can be made in the event of the voltage Vpcxs being decreased.

Shown in FIGS. 4 and 5 are the results of a simulation wherein the voltage Vpcxs undergoes changes which are typical of a step of modifying and/or verifying the contents of the memory cells.

The three currents plotted in FIG. 5 are, in this order:

Idepl--the current set by the depletion M54;

Imirr--the current of the first mirror M52, M49;

Ipu--the current of a decoding pull-up having the drain grounded and the source at Vpcxs.

The difference between the current Idepl and the current Ipu is only due to the difference between the voltage drop Vdg of the transistor M52 and the pull-up voltage.

This solution allows the P-channel portion of the selection NAND, as provided in conventional decoding circuits, to be omitted. In fact, only N-channel transistors M12, M6, M17 and M8 are present in the NAND portion 9 of the circuit 1.

When that portion 9 is made active by the concurrent appearance of the active signals p, ly, lx and ls, a current consumption toward ground occurs from the input A via the transistors M2, M12, M6, M17 and M8.

If the signal pgate were held at ground, this consumption would not be bearable in the presence of a voltage Vpcxs higher than the supply voltage Vcc.

Thus, by arranging a local regulator 30 to control the respective signal pgate of each sector, the consumption of the selected row can be limited without affecting the power-off speed of the unselected rows.

It can be stated that by using the signal pgate to control the decoding pull-up, the P-channel portion of the decoding NAND portion, which usually burdens conventional circuits, can be eliminated. Furthermore, the current draw from the pull-up can be kept under control.

Furthermore, it should be understood that the use of a high-voltage diode 25 at the row end enables the word line to withstand increased stresses during the erasing step, and hence to accept increased negative voltages.

Thus, the decoding circuit of the invention affords several advantages, and is an improvement on the prior art solutions that are still affected by problems of efficiency and reliability.

Modifications and changes may be made unto the circuit according to the invention within the scope of the following claims.

Having thus described at least one illustrative embodiment of the invention, various alterations, modifications, and improvements will readily occur to those skilled in the art. Such alterations, modifications, and improvements are intended to be within the spirit and scope of the invention. Accordingly, the foregoing description is by way of example only and is not intended as limiting. The invention is limited only as defined in the following claims and the equivalents thereto.

Claims

1. A method for operating an electrically programmable and erasable semiconductor non-volatile storage device having a matrix of memory cells arranged in rows and columns, the matrix including a sector of cells that are accessible through an associated plurality of rows, wherein an output terminal having a control signal thereon is directly connected to each of the plurality of rows, the method comprising an act of:

applying a non-negative voltage to a control terminal of a p-channel depletion transistor disposed between a ground reference terminal having a ground reference signal thereon and the output terminal to enable erasure of the cells of the sector of the matrix wherein the control terminal is a different terminal than the output terminal and the act of applying includes an act of:

providing a conduction path between the output terminal and the ground reference terminal such that, in response to the non-negative voltage, a voltage potential of the control signal is controlled to be substantially equal to a voltage potential of the ground reference signal.

2. The method of claim 1, further comprising an act of:

applying a voltage to a control terminal of a second transistor disposed between the output terminal and a first reference terminal having a first reference signal thereon.

3. The method of claim 2, wherein a body terminal of the p-channel depletion transistor is connected to a body of the second transistor.

4. The method of claim 3, wherein the p-channel depletion transistor has a gate oxide thickness of at least 330 Angstroms.

5. The method of claim 1, further comprising an act of:

applying a voltage to a control terminal of a first transistor disposed between the p-channel depletion transistor and the ground reference terminal.

6. The method of claim 5, further comprising an act of:

applying a voltage to a control terminal of a second transistor disposed between the output terminal and a first reference terminal having a first reference signal thereon.

7. The method of claim 5, wherein the p-channel depletion transistor has a gate oxide thickness of at least 330 Angstroms.

8. The method of claim 5, wherein a body terminal of the p-channel depletion transistor is connected to a body of the second transistor.

9. The method of claim 8, wherein the p-channel depletion transistor has a gate oxide thickness of at least 330 Angstroms.