US20110121380A1

US20110121380A1 - Non-volatile electrically alterable memory cell for storing multiple data

Info

Publication number: US20110121380A1
Application number: US12/902,614
Authority: US
Inventors: Andy Yu; Ying W. Go
Original assignee: Nanostar Corp
Current assignee: Nanostar Corp
Priority date: 2004-03-16
Filing date: 2010-10-12
Publication date: 2011-05-26
Also published as: US20080074924A1; US20060113585A1; US7834388B2; US20060131640A1

Abstract

A memory cell that includes a control gate disposed laterally between two floating gates where each floating gate is capable of holding data. Each floating gate in a memory cell may be erased and programmed by applying a combination of voltages to diffusion regions, the control gate, and a well. A plurality of memory cells creates a memory string, and a memory array is formed from a plurality of memory strings arranged in rows and columns.

Description

CROSS REFERENCE TO RELATED APPLICATION

The present application is a divisional application of copending U.S. patent application Ser. No. 11/348,556, entitled “Memory Array of Non-Volatile Electrically Alterable Memory Cells for Storing Multiple Data,” filed on Feb. 6, 2006, which is a divisional application of copending U.S. patent application Ser. No. 10/801,789, entitled “Non-Volatile Electrically Alterable Memory Cell For Storing Multiple Data And An Array Thereof,” filed on Mar. 16, 2004, and claims the benefit thereof and are both incorporated herein by reference.

BACKGROUND OF THE INVENTION

1. Field of the Invention
The present invention relates to logic gate structures, and more particularly, to an electrically erasable and programmable read-only memory (EEPROM) and to Flash EEPROMs employing metal-oxide-semiconductor (MOS) floating gate structures.
2. Description of the Related Art
Electrically erasable and programmable non-volatile semiconductor devices, such Flash EEPROMs are well known in the art. One type of Flash EEPROM employs metal-oxide-semiconductor (MOS) floating gate devices. Typically, electrical charge is transferred into an electrically isolated (floating) gate to represent one binary state while an uncharged gate represents the other binary state. The floating gate is generally placed above and between two regions (source and drain) spaced-apart from each other and separated from those regions by a thin insulating layer, such as a thin oxide layer. An overlying gate is disposed above the floating gate provides capacitive coupling to the floating gate, allowing an electric field to be established across the thin insulating layer. “Carriers” from a channel region under the floating gate are tunneled through the thin insulating layer into the floating gate to charge the floating gate. The presence of the charge in the floating gate indicates the logic state of the floating gate, i.e., 0 or 1.
Several methods can be employed to erase the charge in a floating gate. One method applies ground potential to two regions and a high positive voltage to the overlying gate. The high positive voltage induces charge carriers, through the Fowler-Nordheim tunneling mechanism, on the floating gate to tunnel through an insulating layer that separates the overlying gate and the floating gate into the overlying gate. Another method applies a positive high voltage to a source region and grounds the overlying gate. The electric field across the layer that separates the source region and the floating gate is sufficient to cause the tunneling of electrons from the floating gate into the source region.
Typically, one control gate and one floating gate form a memory cell and store only one piece of data. Accordingly, to store a large number of data, a large number of memory cells are needed. Another problem faced with traditional memory cells is miniaturization. Shrinking the scale of transistors has made it more difficult to program the floating gate devices, and reduces the ability of the floating gate devices to hold a charge. When the overlaying gate cannot induce enough voltage onto the floating gate, the floating gate cannot retain enough charge for a meaningful read-out. Therefore, the traditional transistor layout is reaching a limitation in miniaturization.

SUMMARY OF THE INVENTION

In one aspect, the invention is an electrically alterable memory device. The memory device includes a semiconductor substrate and a semiconductor well. The semiconductor substrate is doped with a first dopant in a first concentration, and a semiconductor well, adjacent the semiconductor substrate, is doped with a second dopant that has an opposite electrical characteristic than the first dopant. The semiconductor well having a top side on which two spaced-apart diffusion regions are embedded. Each diffusion region is doped with the first dopant in a second concentration greater than the first concentration. The two diffusion regions includes a first diffusion region and a second diffusion region, and a first channel region is defined between the first diffusion region and the second diffusion region. The memory device also includes a first floating gate, a second floating gate, and a control gate. The first floating gate is disposed adjacent the first diffusion region and above the first channel region and separated therefrom by a first insulator region. The first floating gate has a first height and is made from a conductive material and capable of storing electrical charge. The second floating gate is disposed adjacent the second diffusion region and above the first channel region and separated therefrom by a second insulator region. The second floating gate has a second height and is made from a conductive material and capable of storing electrical charge. The control gate is disposed laterally between the first floating gate and the second floating gate. The control gate is separated from the first floating gate by a first vertical insulator layer and separated from the second floating gate by a second vertical insulator layer. The control gate is disposed above the first channel region and separated therefrom by a third insulator region. The control gate has a third height and is made from a conductive material.
In another aspect, the invention is an electrically alterable memory string. The memory string includes a plurality of memory devices, each memory device having a control transistor capable of storing a plurality of data. The plurality of memory devices has a first end and a second end. A first select transistor connected to the first end, a second select transistor connected to the second end, and a connector connecting the first select transistor to a bit line.
In another aspect, the invention is an electrically alterable non-volatile memory array. The memory array includes a plurality of memory strings, each memory string having a fast connector connected to a drain of a first select transistor in the memory string, a second connector connected to a gate of the first select transistor, a third connector connected to a gate of a memory cell transistor in the memory string, and a fourth connector connected to a gate of a second select transistor in the memory string. The plurality of memory strings are arranged in such way that the drain of the first select transistor in a fast memory string is connected to a source of the second select transistor in an adjacent second memory string. The memory array also includes a plurality of bit lines, wherein each bit line being connected to the first connector of every memory string, a plurality of first select lines, wherein each first select line being connected to the second connector of every memory string, a plurality of control lines, where each control line being connected to the third connector of every memory string, and a plurality of second select lines, wherein each second select line being connected to the fourth connector of every memory string.
Other advantages and features of the present invention will become apparent after review of the hereinafter set forth Brief Description of the Drawings, Detailed Description of the Invention, and the Claims.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a top plan view of a plurality of memory strings according to one embodiment of the invention.

FIG. 2A is a cross sectional view of the memory string taken along line 2-2 of FIG. 1.

FIG. 2B is a cross sectional view of an alternative embodiment of the memory string taken along line 2-2 of FIG. 1.

FIG. 3 is a cross sectional view of the memory string taken along line 3-3 of FIG. 1.

FIG. 4 is a cross sectional view of the memory string taken along line 4-4 of FIG. 1.

FIG. 5 is a top plan view of a plurality of memory strings according to one alternative embodiment of the invention.

FIG. 6A is a cross sectional view of yet another alternative embodiment of the memory string taken along line 6-6 of FIG. 5.

FIG. 6B is a cross sectional view of yet another alternative embodiment of the memory string taken along line 6-6 of FIG. 5.

FIG. 6C is a cross sectional view of yet another alternative embodiment of the memory string taken along line 6-6 of FIG. 5.

FIG. 7 is a schematic of a top plan of a plurality of memory cells according to one embodiment of the invention.

FIG. 8 is a schematic of a top plan of a plurality of memory cells according to one alternative embodiment of the invention.

FIG. 9 lists several combinations of operational voltages according to one embodiment of the invention.

DETAILED DESCRIPTION OF THE INVENTION

Three electrically programmable and erasable non-volatile memory strings are shown in FIG. 1. Each memory string 100 includes an active region 106 running vertically and a plurality of control gates 102 running horizontally across multiple memory strings. The active region is heavily doped with a first dopant. The control gate is formed by polysilicon or other suitable material. A plurality of floating gates 104 are disposed adjacent to the control gate 102 and over the active region 106. Each control gate 102 is surrounded by two floating gates 104 on two sides.
The combination of two floating gates 104 surrounding one control gate 102 over one area of the active region 106 forms a memory cell 103. Each memory cell 103 stores two data, one on each floating gate 104. Each memory string 100 may have many memory cells 103.
The memory cells 103 on a memory string 100 are delimited by a first select gate 116 and a second select gate 120. The first select gate 116 and the second select gate 120 run horizontally over all memory strings 100 and over the active region 106. The area of the active region 106 not covered by the floating gates 104, the control gates 102, and the select gates 114, 116, 118, 120 are doped diffusion regions. A vertical connector 121 connects the active region 106 to a bit line 110 that runs vertically through multiple memory strings 100.
Each memory string 100 is connected to an adjacent memory string 100 through the active region 106. The separation of memory cells 103 in one memory string 100 from memory cells 103 of an adjacent memory string 100 may be accomplished through an isolation layer 122, such as localized oxidation (LOCOS), recessed LOCOS, shallow trench isolation (STI), or full oxide isolation. A plurality of memory strings 100 may form a high density memory array.
FIG. 2A is a cross section view 200 of a memory cell 103 taken along line 2-2 in FIG. 1. The memory cell 103 includes a semiconductor substrate 202 and a well 204 on the top of the substrate 202. The substrate is doped with a first dopant, which can be either N type or P type. The well 204 is a semiconductor doped with a second dopant with an electrical characteristic that is opposite of the first dopant. Two spaced-apart diffusion regions 106 a and 106 b, which are part of the active region 106, are placed on the top side of the well 204. The diffusion regions 106 a, 106 b are doped with the same dopant used for doping the substrate 202 but doped with a concentration that is higher than that of the substrate 202. A channel region 234 is defined between two diffusion regions 106 a, 106 b. An insulating layer 230 is placed on the top of the well 204 and the diffusion regions 106 a, 106 b. The insulating layer 230 may be formed by an insulating oxide material or other suitable insulating materials. Though FIG. 2A illustrates the diffusion regions 106 a, 106 b implemented in a single well, it is understood that other implementations, such as twin wells, triple wells, or oxide isolation well may also be used. The separation of active devices may be accomplished through localized oxidation (LOCOS), recessed LOCOS, shallow trench isolation (STI), or full oxide isolation. A first floating gate 104 a of polysilicon material is placed above the channel region 234 and adjacent diffusion region 106 a. The first floating gate 104 a may overlap slightly with the diffusion region 106 a; however, excessive overlapping may reduce the length of the channel region 234. The first floating gate 104 a is separated from the channel region 234 by a tunnel channel 214 a (also known as tunnel oxide) of the insulating layer 230. The thickness of the tunnel channel 214 a should be thin enough to allow removal of electrons from the first floating gate 104 a under the Fowler-Nordheim tunneling mechanism, but thick enough to prevent the occurrence of a leakage current between the first floating gate 104 a and the well 204. The length of the tunnel channel 214 a under the first floating gate 104 a can be smaller than one lambda, where the lambda is defined by the technology used. For example, if the technology uses 0.18 μm, then one lambda is defined as 0.18 μm.
A second floating gate 104 b of polysilicon material is placed above the channel region 234 and adjacent diffusion region 106 b. The second floating gate 104 b may overlap slightly with the diffusion region 106 b; however, excessive overlapping may reduce the length of the channel region 234. The second floating gate 104 b is separated from the channel region 234 by a tunnel channel 214 b (also known as tunnel oxide) of the insulating layer 230. The thickness of the tunnel channel 214 b should be thin enough to allow removal of electrons from the first floating gate 104 b under the Fowler-Nordheim tunneling mechanism, but thick enough to prevent the occurrence of a leakage current between the second floating gate 104 b and the well 204.
A control gate 102 is placed above the channel region 234, laterally between the first floating gate 104 a and the second floating gate 104 b. The control gate 102 is separated from the first floating gate 104 a by a first vertical insulating layer 212 a and from the second floating gate 104 b by a second vertical insulating layer 212 b. The insulating layers 212 a, 212 b can be oxide-nitride-oxide or other suitable material. The control gate 102 is separated from the channel region 234 by a separation channel 216 (also known as separation oxide) of the insulating layer 230. The thickness of the separation channel 216 should be thick enough to sustain the stress from the control gate's 102 voltage variation. The voltage at the control gate 102 may vary during operation of the memory cell 103 and cause stress on the separation channel 216, thus leading to the deterioration of the separation channel 216. The control gate 102 may be formed by a polysilicon grown at a different stage as the floating gates 104 a, 104 b. The control gate 102 is connected to control gates in other memory cells in different memory strings. The control gate 102 is surrounded by two floating gates 104 a, 104 b.
The fast floating gate 104 a has a first height measured from its bottom edge to its top edge and the second floating gate 104 b has second height also measured from its bottom edge to its top edge. The control gate 102 has a third height measured from its bottom edge to its top edge. The first height, the second height, and the third height may be identical or may be different. The first height and the second height may be taller or shorter than the third height.
The cross section view of one alternative embodiments of the memory cell 103 taken along line 2-2 in FIG. 1 is shown in FIG. 2B. FIG. 2B illustrates a cross section view 300, where each floating gate 104 a and 104 b surrounds the control gate 102 on more than one lateral side. Because of greater exposure of the surface of a floating gate 104 a, 104 b to the control gate 102, greater the coupling ratio between the control gate's voltage and the floating gate voltage can be achieved.
Referring back to FIG. 2A, when a voltage is applied to the control gate 102, through a coupling effect, a voltage is induced on the floating gates 104 a, 104 b. The voltage induced depends on a coupling ratio between the control gate 102 and the floating gate 104 a. The coupling ratio is defined as the capacitance ratio between the capacitance between the control gate 102 and the floating gate 104 a and the capacitance between the floating gate 104 a and the substrate 204.
C(CG/FG)=capacitance between the control gate and the floating gate
C(FG/Substrate)=capacitance between the floating gate and the substrate
Gamma=coupling ratio
$Gamma = \frac{C (\frac{CG}{FG})}{C (\frac{CG}{FG}) + C (\frac{FG}{Substrate})}$
When a V_CGis applied to the control gate, the voltage at the floating gate is:
V _FG =V _CG×Gamma
The coupling effect depends on the thickness of the layers 212 a, 212 b separating the control gate 102 from the floating gates 104 a, 104 b and the area on each floating gate 104 a, 104 b exposed to the coupling effect. The coupling effect can be easily increased by increasing the area of the floating gate 210 exposed to the control gate 212, and the area of the floating gate 210 exposed to the control gate 2′12 may be increased by increasing the height 234 of the control gate 212 and the height 232 of the floating gate 210. A capacitor is formed between the control gate 102 and each floating gate 104 a, 104 b. When a floating gate 104 a, 104 b is surrounded by a control gate 102 in more than one lateral side, the coupling effect is increased and the capacitance between the floating gate 104 a, 104 b and the control gate 102 is increased. If the layer 212 a, 212 b separating the control gate 102 and the floating gate 104 a, 104 b is too thin, a leakage current may occur between the floating gate 104 a, 104 b and the control gate 102 when the floating gate 104 a, 104 b is charged with electrons. If the layer 212 a, 212 b is too thick, the coupling ratio may be low, resulting in a low voltage in the floating gate. One workable coupling ratio is between 50%-80%, i.e., 10 V applied to the control gate 102 results in 5 V to 8 V induced in the floating gate 104 a, 104 b. The combination of the control gate 102, the floating gates 104 a, 104 b, and the diffusion regions 106 a, 106 b forms a control transistor. The control transistor is capable of holding two data independently, one in each floating gate 104 a, 104 b. Each floating gate 104 a, 104 b may be independently programmed.
The induction of voltage on the floating gate 104 a, 104 b is important when erasing or programming a memory cell 103. When programming the floating gate 104 b of a memory cell 103 of N-type diffusion, a positive high voltage (Vpp) between 4V and 11V is applied to the control gate 102, and the diffusion region 106 a and the well 204 are left at 0 V. A positive high voltage between 4V and 11V is also applied to the diffusion region 106 b. The positive high voltage depends on the technology used. A voltage is induced to the floating gates 104 a, 104 b by the Vpp at the control gate 102 through the coupling effect. When the control gate 102 is at the Vpp and inducing voltages onto the floating gates 104 a, 104 b, the channel 234 between the diffusion regions 106 a and 106 b are conductive. With the channel 234 being conductive and the diffusion region 106 b at the Vpp, electrons flow between the diffusion regions 106 a, 106 b, and the phenomenon of impact ionization (several occurrences) occurs near the diffusion region 106 b. The impact ionization occurs when charge carriers moving toward the diffusion region 106 b generate electron-hole pairs from the lattice near the drain junction (diffusion region 106 b). The generated carriers look for high positive voltage and are injected into the floating gate 104 b. The carriers emitted from the source region 106 a experience lateral electrical field between the diffusion regions 106 a and 106 b. The average carder energy is higher near the drain junction of the diffusion region 106 b. The impact ionization tends to occur near the diffusion region 106 b. Of free electrons, only lucky few will be injected into the floating gate 104 b, and this is known as Lucky electron model. The amount of electrons injected into the floating gate 104 b depends on the positive high voltage applied to the control gate 102 and the duration of this positive high voltage. To program the floating gate 104 a, the similar process may be used but the voltages at the diffusion regions 106 a and 106 b are reversed, i.e., a positive high voltage is applied to the diffusion region 106 a while the diffusion region 106 b and the well 204 are at zero volt.
A ramping positive high voltage (Vppr) may be applied to the control gate 102 to program a floating gate 104 b in a memory cell of P-type diffusion. A positive high voltage between 4V and 11V is initially applied to the control gate 102, and this positive high voltage is gradually ramped down to 0V and then ramped up back to 4V-11V. A positive high voltage is applied to the diffusion region 106 a and 0V is applied to the diffusion region 106 b. When the control gate 102 is at the positive high voltage of 4V-11V, a voltage is induced onto the floating gate 104 b and the channel 234 between the diffusion regions 106 a and 106 b is turned off. Although the floating gate 104 b is at a positive voltage level, no electrons are injected into the floating gate 104 b because the channel 234 is off and there is no flow of electrons between the diffusion regions 106 a and 106 b. As the voltage at the control gate 102 ramps down, the potential difference between the control gate 102 and the well 204 turns on the channel between the diffusion regions 106 a and 106 b, and electrons start to flow in the channel 234. The voltage at the floating gate 104 b also drops as the voltage at the control gate 102 ramps down, but the voltage at the floating gate 104 b is still sufficient to cause some high energy electrons (also known as hot electrons) to be injected into the floating gate 104 b. When the control gate 102 reaches zero voltage, the channel 234 is turned on, but no electrons are injected into the floating gate 104 b because the floating gate 104 b is also at zero voltage. When the voltage at the control gate 102 starts to ramp up back to 4V-11V, the voltage at the floating gate 104 b also ramps up, and high energy electrons from the channel 234 start to be injected into the floating gate 104 b again. When the control gate 102 is at positive high voltage of 4V-11V, the channel 234 is turned off, electrons stop flowing, and no more electrons are injected into the floating gate 104 b. The number of electrons injected into the floating gate 104 b depends on the duration of the ramp down/up process and the concentration of dopants in the channel region. This voltage ramping process may be repeated for the floating gate to retain enough charge to represent a logic state properly. Once charges of electrons are inside of the floating gate 104 b, the floating gate 104 b may hold the charges for years. The voltage ramping may also be used to program memory cells of N-type diffusion.
The amount of charge injected into a floating gate 104 a, determines the threshold voltage for the control transistor formed by the control gate 102, the floating gates 104 a, 104 b, and the diffusion regions 106 a, 106 b. The floating gate 104 a may hold different amount of charges, thus having different threshold voltages. In one embodiment of the invention, through repeating the voltage ramping process, the floating gate 104 a may have four different levels of threshold voltages and capable of representing four logic states. The four logic states may be read and distinguished by measuring the current flowing between the diffusion regions 106 a, 106 b.
A P-type diffusion memory cell may also be programmed with a different mechanism. Applying a negative voltage between −1V and −10V to diffusion region 106 b, a positive high voltage to the control gate 102, and a voltage between 0V and Vcc to the well 204, charges can be programmed into floating gate 104 b. The high positive voltage of the control gate 102 induces a voltage into the floating gate 104 b. The difference of potential between the well 204 and the diffusion region 106 b causes a soft avalanche breakdown between the diffusion region 106 b (P-type) and the well 204 (N-type). Some of the electrons from this soft breakdown are injected into the floating gate 104 b because the floating gate 104 b is at higher voltage.
A negative voltage is applied between −4.5V and −10V to the control gate 102, a positive high voltage is applied to the well 204 when it is desired to erase charges in the memory cell 103 of N-type diffusion. The negative voltage at the control gate 103 is induced to the floating gates 104 a, 104 b. The combination of an induced negative voltage at the floating gates 104 a, 104 b and positive high voltages at the well 204 forces electrons out of the floating gates 104 a, 104 b and into the well 204, thus removing the electrons from the floating gates 104 a, 104 b. The electrons are removed through the Fowler-Nordheim tunneling mechanism.
A negative voltage is applied between 4.5V and −10V to the control gate 102, a positive high voltage is applied to the well 204 and the diffusion regions 106 a, 106 b when it is desired to erase charges in the memory cell 103 of P-type diffusion. The mechanism to remove the electrons is similar to what has been described above for the N-type diffusion except that the positive high voltage is needed at the diffusion regions 106 a and 106 b because otherwise the channel 234 may be floating at an unknown voltage and impeding the exit of electrons from the floating gates 104 a, 104 b.
When it is desired to read the content from a floating gate 104 a of a memory cell 103, a voltage between 0V and Vcc is applied to the control gate 102, a voltage between 0 V to Vcc is applied to diffusion region 106 b, and 0V to −2V is applied to two select gates (not shown in FIG. 2A) in the memory string, one at each end of the memory string. The voltage at the control gate 102 turns on the portion of the channel 234 under it. The threshold voltage (Vt) for the floating gate 104 b is lowered because of drain-induced barrier lowering (DIBL) and a depletion region is created under the floating gate 104 b. If the floating gate 104 a has charge, the portion of the channel 234 under it will be on and a current flows from diffusion region 106 b to diffusion region 106 a. The channel 234 under the floating gate 104 a and the control gate 102 is on, the current passes under the floating gate 104 a and the control gate 102, and then enter the depletion region under the floating gate 104 b. The current will continue to flow through the depletion region under the floating gate 104 a toward the diffusion region 106 a. Because the select gates are at 0V to −2V, the current resulting from the electron flow is sensed by a bit line and a sense-amplifier connected to the bit line. The data stored in the floating gate 104 a comes out from the drain of the control transistor. When the floating gate 104 a is programmed to store different levels of charge and thus with different levels of threshold voltage, the intensity of the current flowing between diffusion region 106 a and diffusion region 106 b depends on the threshold voltage of the floating gate 104 a. The intensity of this current can be sensed by the sense-amplifier, thus the logic level of the floating gate 104 a determined.
If the floating gate 104 a is without charge, then the portion of the channel 234 under the floating gate 104 a will not be turned on and there will be no current or small leakage current flowing between diffusion region 106 b and diffusion region 106 a. The leakage current should be different from the current flowing when the floating gate 104 a is charged. If the floating gate 104 a is not charged, then no channel is established between the diffusion regions 106 a and 106 b and the sense-amplifier will not be able to detect any current. The absence of a current between the diffusion regions 106 a and 106 b indicates the floating gate 104 a is without electrons. A floating gate 104 a with electrons is assigned to a first logic state while a floating gate 104 a without electrons or with too few electrons is assigned to an opposite second logic state. Other operations not described here can be easily understood based on voltages listed in FIG. 9 and operations described above by those skilled in the art. When reading the content of a floating gate 104 a in an N-type diffusion device, 0V is applied to 106 a, and 1V to 2.5V is applied to 106 b. The voltage in the diffusion region 106 b will lower the threshold voltage of the floating gate 104 b because of DIBL effect.
FIG. 3 is a cross section view 400 taken along line 3-3 in FIG. 1. FIG. 3 illustrates a cross section view of a memory string 100. The memory string 100 includes a substrate 202 doped with a first dopant, which can be either N type or P type, and a well 204 doped with a second dopant with an electrical characteristic that is opposite of the first dopant. A plurality of diffusion regions 106, which are part of the active region 106 in FIG. 1, are placed on the top side of the well 204. The diffusion regions 106 are doped with the same dopant used for doping the substrate 202 but doped with a concentration that is higher than that of the substrate 202. A plurality of control transistors are placed adjacent each other. Each control transistor includes a control gate 102, two floating gates 104, and two diffusion regions 106, one diffusion region being the drain of the control transistor while the other diffusion region is the source. Two adjacent control transistors share one common diffusion region 106. Each control transistor is a memory cell. There is one select transistor 402 at one end of this “string” of memory cells and another select transistor 404 at the other end of the string of memory cells. There is a vertical contact 406 connecting one diffusion region 106 at the end of the memory string to a bit line 108 in FIG. 1. A diffusion region 106 from one memory string 100 is connected to a diffusion region 106 of an adjacent memory string 100 (shown in FIG. 1). FIG. 4 is a cross section view 500 taken along line 44 in FIG. 1. It is shown that memory strings represented by a floating gate 104 are separated from each other by isolation layers 122.
FIG. 5 is a top plan view of an alternative embodiment of the invention. In this embodiment the memory cells 603 in one memory string 600 are between two diffusion regions 106 a, 106 b. Each control gate 102 runs horizontally in FIG. 5 and across different memory strings 600. Each memory string 600 is delimited by two select gate transistors SG0 a, SG1 a in a manner similar as that depicted in FIG. 1. One STI 612 separates one diffusion region of one memory string 600 from a diffusion region for an adjacent memory string. The diffusion region 106 a is connected to a bit line 602 through a buried contact and the diffusion region 106 b is connected to a bit line 604 also through a buried contact. When it is desirable to read a data from a floating gate 104 b of a memory cell 603, the control gate 102 for the selected memory cell is turned on. A bit line 604 is connected to a source voltage. The voltage at the bit line 604 is propagated through the diffusion region 106 b to the memory cell 603. The read operation at the memory cell 603 is similar to that described for FIG. 2. The data is read from the drain of the control gate transistor, bit line 602. The program and ease operations for the embodiment of FIG. 5 are same as those previously described for FIG. 2. For the embodiment of FIG. 5, there is no need to turn on the control transistors of unselected memory cells. As matter of fact, the control transistors of unselected memory cells are turned off to prevent short between diffusion regions 106 a and 106 b.
FIG. 6A is a cross section view 700 taken along line 6-6 in FIG. 5. The floating gates 104 a, 104 b are surrounded by the control gate 102 from top and two lateral sides. FIG. 6B is a cross section view 800 of an alternative embodiment taken along line 6-6 in FIG. 5. FIG. 6C is a cross section view 900 of yet another alternative embodiment taken along line 6-6 in FIG. 5. It is also illustrated in FIG. 6C two oxidation sections 612 a, 612 b. Each oxidation section 612 a disposed on the top of a diffusion region 106 a. The oxidation section 612 does not divide a diffusion region 106 into two, but it does lessen the capacitance of the diffusion region 106. When the capacitance of a diffusion region 106 is smaller, the fast is the speed a data can be read out from a memory cell 603.
FIG. 7 is a schematic representation of part of a memory array 1000 made from the memory strings 100. Memory cells 1002 in one memory string (running vertically in FIG. 7) are interconnected. The drain of one memory cell is connected to the source of an adjacent memory cell. Each memory string includes two select transistors 1003, 1005 one at each end of the memory string. One end of the memory string is connected to a bit line 1018 and also connected to an adjacent memory string. The memory strings are disposed parallel to each other and the resulting memory array are organized in rows and columns. The select transistors 1003 of odd columns are controlled by one select line 1004, while the select transistors 1003 of even columns are controlled by another select line 1006. Similarly, the select transistors 1005 of odd columns are controlled by one select line 1016, while the select transistors 1005 of even columns are controlled by another select line 1014. The control transistors in one memory row are interconnected together and controlled by a control line. Data operations to one floating of a memory cell in one memory string is controlled by activating proper control line, select lines, and bit lines as described above for FIG. 2A. The activation of control lines and select lines depends on an X-address decoder (not shown) and the activation of bit lines depends on a Y-address decoder (not shown). Each bit line may be connected to a charging transistor and a discharging transistor (not shown) that are also controlled by the Y-address decoder.
FIG. 8 is a schematic presentation of a memory array 1100 made from memory strings 600. One side of a control transistor of the memory cell 1112 is connected to a bit line 1102, other side of the control transistor is connected to another bit line 1104, and the gate of the control transistor is connected to a control gate line 1106. Other select logics for enabling and selecting each memory cell are not shown in FIG. 8 but are easily understood by those skilled in the art.
The thickness of each gate (control gate, and floating gate) depends on the manufacturing process; currently most common thickness is about 3000 Angstroms or 0.3 micron. The thickness of the tunnel channels 214 a, 214 b depend also on manufacturing process. However, a preferred thickness for the tunnel channels 214 a, 214 b is between 70 Angstroms and 110 Angstroms. Similarly, the thickness of the insulating layer separating the control gate 102 from the well 204 is between 150 Angstroms and 250 Angstroms. The materials and measurements mentioned heretofore are for illustration purposes and not intended to limit the scope of the present invention. It is recognized that as technology evolves, other suitable materials and manufacturing processes may be employed to realize the present invention. It is also understood that the structures disclosed heretofore can be easily implemented by any of existing semiconductor manufacturing processes known to those skilled in the art. It is also understood that the voltages illustrated in FIG. 9 is for illustration purposes, and other voltage combinations may be used. For example, voltages may be reduced for embodiments that have a large coupling ratio, and small voltages make manufacturing easier and enhance reliability.
Although, the present application is described for Flash EEPROMs, it is understood that the invention is equally applicable for one-time-programmable (OTP) memories, multiple-time-programmable (MTP) memories, and other non-volatile memories.
While the invention has been particularly shown and described with reference to a preferred embodiment thereof, it will be understood by those skilled in the art that various changes in form and detail may be made without departing from the spirit and scope of the present invention as set forth in the following claims. Furthermore, although elements of the invention may be described or claimed in the singular, the plural is contemplated unless limitation to the singular is explicitly stated.

Claims

1. An electrically alterable memory device, comprising:

(a) a semiconductor substrate having a surface, comprising:

a first semiconductor layer of a first conductivity type at the surface of the semiconductor substrate, the first semiconductor layer having a first dopant concentration; and

a first diffusion region and a second diffusion region also provided at the surface of the semiconductor substrate spaced apart by a portion of the first semiconductor layer, thereby forming a channel region in the first semiconductor layer between the diffusion regions, each diffusion region being of a second conductivity type opposite to the first conductivity type and each diffusion region having a second dopant concentration greater than the first dopant concentration such that the first semiconductor layer and each diffusion region form a junction which breaks down according to a soft avalanche breakdown mechanism;

(b) a gate dielectric layer provided over the surface of the semiconductor substrate;

(c) a first floating gate and a second floating gate each comprising a conductive material, the floating gates respectively disposed at least in part overlapping the junctions, separated therefrom by the gate dielectric layer; and

(d) a control gate also comprising a conductive material, the control gate being disposed laterally between the first floating gate and the second floating gate and separated from each floating gate by an insulator layer having a thickness such that, when an electrical potential is imposed on the control gate, a portion of the electrical potential is capacitively coupled to the floating gates, and wherein the gate dielectric layer is of a thickness such that, when a soft avalanche break down occurs in a junction, electrical charge is injected by the capacitively coupled electrical potential into the corresponding overlapping floating gate.

2. The memory device of claim 1, wherein the gate dielectric layer allows tunneling of electrical charge between one of the floating gates and the channel region.

3. The memory device of claim 1, wherein the gate dielectric layer is between 70 Angstroms and 110 Angstroms thick.

4. The memory device of claim 1, wherein the insulator layer comprises a silicon dioxide having a thickness that is sufficiently thick to prevent current leakage between the control gate and each of the floating gates.

5. The memory device of claim 1, wherein the insulator layer comprises an oxide-nitride-oxide layer.

6. The memory device of claim 1, wherein each floating gate stores a predetermined amount of electrical charge controlled by the capacitively coupled electrical potential.

7. The memory device of claim 1, further comprising contacts to the first diffusion region and the second diffusion region, each contact being selectably coupled to a reference voltage or a variable voltage, so as to allow independent charging and discharging of the first floating gate and the second floating gate.

8. The memory device of claim 1, wherein the first and second diffusion regions are defined by a self-aligned process using the first and second floating gates, respectively.

9. The memory device of claim 1, wherein the first diffusion region and the second diffusion region are P-type.

10. The memory device of claim 9 wherein, when the soft avalanche breakdown mechanism occurs, the channel region is rendered non-conducting.