EEPROM memory cell with low-voltage read path and high-voltage erase / write path

Description

[0001] The invention relates to semiconductor based memory cells, e.g., EEPROM memory cells.

[0002] FIGURE 1 illustrates a prior art electrically erasable programmable read only memory cell, as illustrated and described in Fabrication and Characterization of a New EEPROM Cell With Spacer Select Transistor, by Junghwan Lee et al., IEEE Electron Device Letters, Vol. 26, No. 8, August 2005.

[0003] EP 1 069 619 discloses a semiconductor integrated circuit device, production and operation method thereof. US Patent US 6,101,131 discloses a flash EEPROM device employing polysilicon sidewall spacer as an erase gate. US Patent Application US 2011 / 070726 discloses a method for making a semiconductor device with a select gate and side wall spacers.

[0004] As described in Junghwan Lee et al., the cell has spacer select gates (labelled as "control gate" in FIGURE 1) on both side walls of floating gate, which helps result in a very small cell size as well as relief of topology during contact formation. The cell size is 0.95 µm 2< with 0.18 µm logic process. The cells are programmed and erased by Fowler-Nordheim tunneling. Programming requires 3 ms at 16 V, while erasing requires 2 ms at 14 V. The operating voltages for selected and unselected cells are shown in Table 1. Table 1: Bias condition during programming (writing), erasing, and reading, for both selected and unselected cells configured as shown in FIGURE 1. Program / WriteEraseReadControl GateSelected16V0V2.5VUnselected2.5V11V0VBit LineSelected0V14V1.0VUnselected12V11V0VP-WellSelected0V14V0VUnselected0V0V0VSourceSelectedFloatingFloating0VUnselectedFloatingFloating0V

[0005] As illustrated by Table 1, the junctions of this known cell operate at relatively high voltages (12V-16V). This means that the junctions underneath the control gate spacers (control gate) are relatively large and deep, and the gate oxide under the gate spacers (control gate) is relatively thick, in particular much thicker than the tunnel oxide underneath the floating gate. This relatively large thickness of the control gate oxide resists the flow of read current, such that the dimension of the cell into the page must be kept relatively large. Thus, the size of this memory cell cannot be easily scaled down to smaller geometries, while still providing sufficient read current. This and other objects can be achieved by an electrically erasable programmable read only memory (EEPROM) cell, method of operating such a cell, and memory cell array as defined in the independent claims. Further enhancements are characterized in the dependent claims.

[0006] One embodiment provides an electrically erasable programmable read only memory (EEPROM) cell that may include a substrate including at least one active region, a floating gate adjacent the substrate, a write / erase gate defining a write / erase path for performing high voltage write and erase operations of the cell, and a read gate defining a read path for performing low voltage read operations of the cell, wherein the read path is distinct from the write / erase path.

[0007] The write / erase gate is formed over the floating gate, and the read gate is formed laterally adjacent the floating gate and write / erase gate.

[0008] The EEPROM cell includes a floating gate oxide between the floating gate and the substrate, and a read gate oxide between the read gate and the substrate, wherein the read gate oxide is thinner than the floating gate oxide.

[0009] In a further embodiment, the write / erase path defined by the write / erase gate is configured for high voltage write and erase operations, and the read path defined by the read gate is configured for low voltage read operations.

[0010] In a further embodiment, an active region of the substrate is self-aligned with the read gate.

[0011] The read gate includes first and second portions formed on opposite sides of the floating gate, such that the floating gate is arranged between the first and second portions of the read gate.

[0012] In a further embodiment, the EEPROM cell includes first and second read gates that are independently-addressable.

[0013] In a further embodiment, the first read gate is formed adjacent a first lateral side of the floating gate, and the second read gate is formed adjacent a second lateral side of the floating gate.

[0014] Another embodiment provides a method of operating an electrically erasable programmable read only memory (EEPROM) cell having a substrate including at least one doped well, a floating gate formed over the substrate, a low voltage read path defined by a read gate, and a separate high voltage write / erase path defined by a write / erase gate distinct from the at least one read gate. The method may include performing a write operation to charge the floating gate by creating a high voltage differential between the write / erase gate and the at least one doped well, and performing a read operation to read the charge on the floating gate by creating a low voltage differential between the read gate and the at least one doped well. The method may also include performing an erase operation to discharge the floating gate by creating a high voltage differential between the at least one doped well and the write / erase gate.

[0015] In a further embodiment of the method, the EEPROM cell is an n-channel cell comprising at least one p-well, and the method includes performing the read operation to read the charge on the floating gate by applying a low read voltage bias to the write / erase gate while grounding the at least one p-well; performing the write operation to charge the floating gate by applying a high write voltage bias to the write / erase gate while grounding the at least one p-well; and performing an erase operation to discharge the floating gate by applying a high erase voltage bias to the at least one p-well while grounding the write / erase gate, wherein the high erase voltage bias may be the same or different voltage than the high write voltage bias.

[0016] In a further embodiment of the method, the EEPROM cell is a p-channel cell comprising at least one n-well, and the method includes performing the read operation to read the charge on the floating gate by applying a low read voltage bias to the at least one n-well while grounding the write / erase gate; performing the write operation to charge the floating gate by applying a high write voltage bias to the at least one n-well while grounding the write / erase gate; and performing an erase operation to discharge the floating gate by applying a high erase voltage bias to the write / erase gate while grounding the at least one n-well, wherein the high erase voltage bias may be the same or different voltage than the high write voltage bias.

[0017] In a further embodiment of the method, the EEPROM cell includes first and second read gates, and wherein the method comprises independently biasing the first and second read gates. In a further embodiment, the first and second read gates comprise poly spacers located on opposite lateral sides of the floating gate.

[0018] Another embodiment provides a memory cell array including a plurality of electrically erasable programmable read only memory (EEPROM) cells arranged in an array, each EEPROM cell comprising a substrate including at least one active region, a floating gate adjacent the substrate, a write / erase gate defining a write / erase path for performing write and erase operations of the cell, and a read gate defining a read path for performing read operations of the cell, wherein the read path is distinct from the write / erase path.

[0019] Example embodiments are discussed below with reference to the drawings, in which: FIGURE 1 shows a prior art EEPROM cell design; FIGURE 2 illustrates an example EEPROM cell, according to one embodiment; FIGURE 3 shows example biasing conditions for programming an n-channel EEPROM cell, according to one embodiment; FIGURE 4 shows example biasing conditions for erasing an n-channel EEPROM cell, according to one embodiment; FIGURE 5 shows example biasing conditions for programming a p-channel EEPROM cell, according to one embodiment; FIGURE 6 shows example biasing conditions for erasing a p-channel EEPROM cell, according to one embodiment; FIGURE 7 shows a top view of a portion of an example memory cell array including a dummy cell for connecting control gate contacts with respective poly spacer gates; FIGURE 8A shows a top view of a portion of an example memory cell array similar to FIGURE 7, but with a poly spacer end region etched away to isolate the two poly spacer gate regions from each another; and FIGURE 8B shows a three-dimensional view of the memory cell array portion shown in FIGURE 8A.

[0020] Embodiments of the present invention may include one or more key improvements to the known memory cell design shown in FIGURE 1 and discussed above. For example, some embodiments provide an EEPROM cell including a high voltage write / erase gate separate from a low voltage read gate, in addition to the floating gate. The write / erase gate may be formed over and insulated from the floating gate, while the read gate (or gates) may be formed as poly spacer on the lateral sides of the floating gate. A much thinner spacer oxide may be used underneath the read gate, as compared with the known design shown in FIGURE 1. As a result, the memory cell junctions may operate at much lower voltages (e.g., 3.3 V) instead of the higher voltages of the known design, as the thinner oxide has a better trans-conductance.

[0021] As another example, in some embodiments, dummy cells may be provided in a row or column of memory cells to improve electrical contact between control gate contacts and poly spacers (read gates). As another example, in some embodiments, the two poly spacers of a particular cell may be configured for independent operation, e.g., by physically separating the two poly spacers using a suitable mask and etch process.

[0022] FIGURE 2 illustrates an example EEPROM cell 10, according to one embodiment. As shown, EEPROM cell 10 includes a substrate 12 including an active source region 14 separated from an active drain region 16, a floating gate 20 formed over a tunnel oxide 24 on substrate 12, a high-voltage write / erase gate 30 formed over and insulated from the floating gate 20, and low-voltage read gate regions 34A and 34B formed over a spacer oxide 38 on substrate 12. Floating gate 20, write / erase gate 30, and read gate regions 34A and 34B (poly spacers) are also referred to herein as "poly 1," "poly 2," and "poly 3," respectively, due to the order in which such regions would typically be formed. It should also be understood that cell 10 may be formed as either an n-channel or p-channel cell, using any suitable manufacturing / fabrication techniques.

[0023] As shown, read gate regions 34A and 34B may be formed as spacer select poly gates adjacent the lateral sides of the floating gate 20 and write / erase gate 30. Source and drain regions 14 and 16 may be formed self-aligned with read gate regions 34A and 34B. In some embodiments, read gate regions 34A and 34B are conductively connected (e.g., at the end of a cell row) and thus held at the same voltage, such that gates 34A and 34B act as a single control gate 36. In other embodiments, e.g., as discussed below with reference to FIGURES 8A-8B, read gate regions 34A and 34B are isolated from each other, and thus independently controllable, thereby defining two independent control gates 34. The following discussion thus refers to control gate(s) 36 to refer to both alternatives (i.e., one or two read gates).

[0024] The write / erase gate 30 defines a write / erase path for performing high-voltage write and erase operations of the cell 10, while each read gate 36 defines a read path for performing read operations of the cell 10, each read path being distinct from the write / erase path.

[0025] As used herein, the terms "high voltage" and "low voltage" are intended as relative terms. Thus, "high voltage" refers to a greater voltage than "low voltage." In some embodiments, high voltage operations of a cell (e.g., write / erase operations) involve a voltage bias at least 50% greater than low voltage operations of the cell (e.g., read operations). For example, in some embodiments, high voltage operations of a cell (e.g., write / erase operations) may involve a voltage bias at least 100% greater than low voltage operations of the cell (e.g., read operations). As another example, in some embodiments, embodiments, high voltage operations of a cell (e.g., write / erase operations) involve a voltage bias at least 300% greater than low voltage operations of the cell (e.g., read operations). As another example, in some embodiments, embodiments, high voltage operations of a cell (e.g., write / erase operations) involve a voltage bias at least 500% greater than low voltage operations of the cell (e.g., read operations).

[0026] In some embodiments, high voltage operations of a cell (e.g., write / erase operations) involve a voltage bias above 8V, while low voltage operations of the cell (e.g., read operations) involve a voltage bias below 8V. For example, in some embodiments, high voltage operations of a cell (e.g., write / erase operations) involve a voltage bias between about 10V and about 16V, while low voltage operations of the cell (e.g., read operations) involve a voltage bias below 5V (e.g., about 3.3V or about 1.8V).

[0027] Separating the high-voltage write / erase operations from the low-voltage read operations by adding the separate write / erase gate 30 over the floating gate 20 allows the thickness of the spacer oxide 38 underneath the read gate regions 34A and 34B to be reduced in the direction indicated as T SO , as well as in the direction into the page. In some embodiments, the spacer oxide thickness T SO may be less than the tunnel oxide thickness T TO . In various example embodiments, the spacer oxide thickness T SO may be less than 90%, less than 70%, less than 50%, or less than 30% of the tunnel oxide thickness T TO . For example, in an embodiment that uses a tunnel oxide 24 with a thickness T TO of 93Å underneath the floating gate 20, the spacer oxide thickness T SO can be about 73Å. This allows the memory cell junctions and read path to operate at much lower voltages (e.g., 3.3V) instead of the higher voltages of the previous design, as the thinner oxide has a better trans-conductance. As another example, a spacer oxide thickness T SO of about 35Å may be used, which allows the memory cell junctions and read path to operate at even lower voltages (e.g., 1.8V).

[0028] As shown, high-voltage poly gate 30 is formed over and insulated from the floating gate 20 by an inter-poly dielectric regions 42, which poly gate 20 is used for write and erase functions. The write / erase gate handles the high voltage operation of cell 10 while leaving the read gate(s) 36 to handle the lower-voltage operation of the cell, thus allowing a thinner spacer oxide 38 as discussed above, as well as a smaller depth for the source / drain junctions 14 and 16. A dielectric hard mask 44 is added over the poly gate for providing an isolation metal contact landing on the WL-spacer (read gate) so that these contacts will not short to the poly 2 write / erase gate (see FIGURES 8A and 8B).

[0029] The features discussed above allow for the memory cell 10 to be scaled to much smaller dimensions as compared with the known cell design, thus greatly reducing the memory cell area while maintaining a high read current. For example, cell 100 transistor size width may be reduced from w = 0.4µm to w = about 0.25µm or less and length (including the floating gate) may be reduced from l = 0.9µm to l = about 0.75µm or less. Thus in some embodiments the transistor area portion of the cell may be reduced by at least 40% or by at least 50%, as compared with a conventional cell such as shown in FIGURE 1. In particular example embodiments the transistor area portion of the cell may be reduced by about 40% to 60%, and in some embodiments about 50%, as compared with a conventional cell. Other embodiments may provide different amounts of cell area reduction. Further, by having low junction voltages, the active overlap of contact can also be reduced as well as the contact to poly distance. These features also allow for scaling with finer geometry processes, because the active area and select transistor are now operating at low voltages. The memory cell design maintains the high endurance and high reliability of the memory cell. Finally, the memory cell design can eliminate the conventional requirement of separate source lines (i.e., columns in the array), as discussed below.Example n-channel cell operation

[0030] As mentioned above, cell 10 may be formed as either an n-channel or p-channel cell. Table 2 shows example operating voltages for an n-channel version of cell 10 formed with a single read gate 36, for selected and unselected states. Table 2: Example bias conditions for writing, erasing, and reading, for selected and unselected states of an example n-channel cell 10. Program / WriteEraseReadRead GateSelectedVdd11V-13VVddUnselected0V11V-13V0VWrite / Erase GateSelected15V0VVreadUnselected4V11VVreadBit LineSelected0VFloat = 15V1V-VddUnselectedVddFloat = 15V0VSource LineSelectedFloat = 0VFloat = 15V0VUnselectedVddFloat = 15V0VP-WellCommon0V15V0V

[0031] Thus, according to Table 2, the n-channel cell can be programmed (write) by applying a high voltage to the write / erase gate 30 while grounding the p-well, and erased by applying high voltage to the well and junctions simultaneously while grounding the write / erase gate 30. Supply voltage Vdd may be 3.3V or about 3.3V, for example. The read bias applied to the write / erase gate, Vread, can be centered in the cell program / erase window, and may be close to Vdd for the n-channel cell, as a read operation involves the write / erase gate 30 coupling the floating gate 20 high enough to turn on an erased memory cell into a conductive state (e.g., 1V), but not so high as to turn a written memory cell into a conductive state (e.g., 3.5V).

[0032] FIGURE 3 shows the bias conditions for programming the n-channel cell, based on Table 2, while FIGURE 4 shows the bias conditions for erasing the n-channel cell, again based on Table 2. As shown in FIGURE 3, with the cell row being selected and cell column unselected, the inversion region is used to couple high enough (e.g., 8V) to avoid program disturb. That is, in some embodiments, if the electric field between the floating gate 20 (which will couple high when the 15V program voltage is applied to the poly2 write / erase gate 30) to the substrate 12 is more than about 5-6V for a 93Å tunnel oxide, then program disturb will occur because electrons will tunnel from the inversion region up to the floating gate increasing its threshold voltage. Therefore the inversion region needs to be coupled high (e.g., 8V) to reduce the electric field so that no program disturb occurs.Example p-channel cell operation

[0033] A p-channel version of cell 10 can be formed by changing the p-well of the cell array to n-well, with p+ doped source and drain regions 14 and 16. Table 3 shows example operating voltages for a p-channel version of cell 10 formed with a single read gate 36, for selected and unselected states. Table 3: Example bias conditions for writing, erasing, and reading, for selected and unselected states of an example p-channel cell 10. Program / WriteEraseReadRead GateSelected13V0-Vdd0VUnselected15V0-VddVddWrite / Erase GateSelected0V15VVreadUnselected13V0VVreadBit LineSelected15VFloat = 0V0V to Vdd-1VUnselected13VFloat = 0VVddSource LineSelectedFloat = 15VFloat = 0VVddUnselected13VFloat = 0VVddN-WellCommon15V0VVdd

[0034] Thus, according to Table 3, the p-channel cell can be programmed (write) by applying high voltage to the well and junctions simultaneously while grounding the write / erase gate 30, and erased by applying a high voltage to the write / erase gate 30 while grounding the N-well. Supply voltage Vdd may be 3.3V or about 3.3V, for example. As with the n-channel cell discussed above, the read bias applied to the write / erase gate, Vread, can be centered in the cell program / erase window, and may be close to 0V for the p-channel cell.

[0035] FIGURE 5 shows the bias conditions for programming the p-channel cell, based on Table 3, while FIGURE 6 shows the bias conditions for erasing the p-channel cell, again based on Table 3. As shown in FIGURE 5, with the cell row being selected and cell column unselected, counting on the inversion region to couple low enough (e.g., 7V) to prevent program disturb.Control gate contacts

[0036] Improved control gate contacts for contacting the poly spacers, i.e., read gate regions 34A and 34B, are also provided. In some embodiments, the poly spacers extend across all the columns of memory cells in the particular row of the cell array, with a control gate contact making electrical contact with the poly spacers at every 128 cells. As the size of the memory cells is reduced, including providing thinner poly spacers, it may be more difficult to reliably make the control gate contacts. Thus, the cell array includes one or more "dummy cells" at which the effective width of the poly spacers is enlarged to provide a better location with a greater tolerance for forming the control gate contacts.

[0037] FIGURE 7 shows a portion of an example cell array 100 including a dummy cell 102 in which the poly spacers 34A and 34B are formed with a bent, or winding, shape in order to provide an increased contact area for electrical contact with control gate contacts 62A and 62B formed over poly spacers 34A and 34B, respectively. In other embodiments, the poly spacers 34A and 34B may have any other bent, curved, or winding path to provide an increased contact area for control gate contacts, or may simply be formed with an increased width as compared with read gate regions 34A and 34B of usable memory cells in the array. Such dummy cells can be located at any suitable location, e.g., at the ends of the rows, and / or interspersed within the array.Independently controllable poly spacer gates (read gates)

[0038] In the embodiment shown in FIGURE 7, the read gate regions (poly spacers) 34A and 34B for each row of memory cells 10 is connected by poly spacer end region 34C at the end of the row, and thus read gate regions 34A and 34B are tied together such that are always held to the same voltage, thereby defining a single read control gate.

[0039] In other embodiments, the two read gate regions 34A and 34B for each row of memory cells may be physically separated and independently controlled, i.e., the two read gate regions 34A and 34B can be held at different voltages, which may provide advantageous functionality. For example, the poly spacers 34A and 34B can be formed in a single step whereby they are connected at the ends via a poly spacer end region 34C (see FIGURE 7), and then separated by a selective etch to remove the poly spacer end region 34C, thereby isolating the two poly spacers 34A and 34B from one another.

[0040] FIGURES 8A and 8B illustrate a portion of an example cell array 100 similar to that of FIGURE 7, but with the poly spacer end region 34C removed by a selective etch process at etch border EB shown in FIGURE 7, thereby isolating gate regions 34A and 34B from each another. This configuration essentially provides for two independent transistors which can be used to isolate the write voltages without the need for separate column source lines (as discussed below). FIGURES 8A and 8B also show a dummy cell 102 for improved connection to control gate contacts 62A and 62B.Common source line

[0041] Further, in some embodiment, a common source feature can further facilitate the scaling down of the memory cell size. The prior art memory cell described above (see FIGURE 1) requires that each column of memory cells in the memory array has a separate bit line and also a separate source line. This is required because when one cell in the selected row is programmed, its bit line is biased at 0V and unselected cells in the row have their bit lines biased at a higher voltage (e.g., 12V). Because the row is selected, both selected and unselected cells are in a conductive state, which without separate source lines would short the bit lines. This is apparent in the bias tables shown above where, during program (write), the source lines for selected and unselected cells perform different functions. In Table 1 above for the prior art memory cell, the source lines for both selected and unselected are floating independently and will go to different voltages depending on the bit line voltage during program (write). Thus, in the prior art design, two metal routes are required for each column of memory cells in the array, so that the minimum cell area is limited by the metal pitch of the technology.

[0042] In contrast, in some embodiments of the present disclosure, the conduction paths between the selected column (to program) and unselected column (to not program) are separated such that the two different voltage potentials applied to the bit lines are not shorted out. This isolation can be achieved by biasing the source-side poly spacer (e.g., read gate 34A shown in FIGURE 2) to keep the source-side transistor off during program. This allows the source side of the cell to share a common node with other bits in the same row and eliminates the need for column decoded source lines. Thus, similar to an EPROM or flash array, the source is shared between all bits common to two or more rows, and each column only needs a separate bit line (i.e., one metal route in the array).

[0043] To achieve this result, the n-channel biasing table shown above, Table 2, is modified to include two poly 3 rows of values, i.e., one for each of the two poly 3 gates for each column of memory cells, referred to as poly 3A and poly 3B. Poly 3A corresponds to the bit line junction and has the same values as shown for poly 3 in Table 2, while poly 3B corresponds to the source line junction and is common for both selected and unselected rows. An example of the resulting bias conditions for this cell are shown in Table 4. Table 4: Example bias conditions for writing, erasing, and reading, for selected and unselected states of an example n-channel array with shared source line. Program / WriteEraseReadPoly 3ASelectedVdd11V-13VVddUnselected0V11V-13V0VPoly 3BSelected0V11V-13VVddUnselected0V11V-13V0VWrite / Erase GateSelected15V0VVreadUnselected4V11VVreadBit LineSelected0VFloat = 15V1V-VddUnselectedVddFloat = 15V0VSource LineCommon0VFloat = 15V0VP-WellCommon0V15V0V

Claims

1. An electrically erasable programmable read only memory, EEPROM, comprising an EEPROM cell, comprising: a substrate (12) including at least one active region; a floating gate (20) above the substrate (12); a write / erase gate (30) arranged above the floating gate (20) defining a write / erase path for performing write and erase operations of the cell; and a floating gate oxide (24) between the floating gate (20) and the substrate (12); a read gate (34A, 34B) arranged above the substrate (12) and laterally adjacent to the floating gate (20) and the write / erase gate (30), the read gate (34A, 34B) defining a read path for performing read operations of the cell, wherein the read path is distinct from the write / erase path, and a read gate oxide (38) between the read gate (34A, 34B) and the substrate (12); wherein the read gate oxide (38) is thinner than the floating gate oxide (24) and thereby a distance (TSO) between the read gate (34A, 34B) and the substrate (12) is less than a distance (TTO) between the floating gate (20) and the substrate (12); characterized in that the read gate (34A, 34B) includes first and second portions (34A, 34B) formed on opposite sides of the floating gate (20), such that the floating gate (20) is arranged between the first and second portions (34A, 34B) of the read gate and the EEPROM further comprising a dummy cell (102) in which the first and second portions (34A, 34B) are formed with a bent, or winding, shape in order to provide an increased contact area for electrical contact with control gate contacts (62A, 62B) formed over the first and second portions, respectively.

2. The EEPROM of Claim 1, wherein: the write / erase gate and the floating gate form a stack and wherein the read gate extends beyond the stack in vertical direction.

3. The EEPROM of Claim 1, comprising: a dielectric hard mask (44) covering the write / erase gate (30) wherein the read gate (34A, 34B) extends to a top surface of the dielectric hard mask (44).

4. The EEPROM of Claim 1, wherein: the write / erase path defined by the write / erase gate (30) for write and erase operations uses at least a first voltage; and the read path defined by the read gate (34A, 34B) for read operations uses a second voltage being lower than the first voltage.

5. The EEPROM of Claim 1, wherein: an active region of the substrate (12) is self-aligned with the read gate (34A, 34B).

6. The EEPROM of Claim 1, wherein the first and second portions (34A, 34B) are first and second read gates (34A, 34B) that are independently controllable.

7. The EEPROM of Claim 6, wherein: the first read gate (34A) is formed adjacent a first lateral side of the floating gate (20); and the second read gate (34B) is formed adjacent a second lateral side of the floating gate (20).

8. A method of operating an electrically erasable programmable read only memory, EEPROM, according to one of the preceding claims wherein the substrate (12) comprises at least one doped well comprising said at least one active region, the method comprising: performing a write operation to charge the floating gate (20) by creating a first voltage differential between the write / erase gate (30) and the at least one doped well; and performing a read operation to read the charge on the floating gate (20) by creating a second voltage differential between the read gate (34A, 34B) and the at least one doped well, wherein the second voltage differential is lower than the first voltage differential.

9. The method of Claim 8, further comprising performing an erase operation to discharge the floating gate (20) by creating a third voltage differential between the at least one doped well and the write / erase gate (30) being higher than the second voltage differential.

10. The method of Claim 8, wherein the EEPROM cell is an n-channel cell comprising at least one p-well, and the method comprises: performing the read operation to read the charge on the floating gate (20) by applying a low read voltage bias to the write / erase gate (30) while grounding the at least one p-well; performing the write operation to charge the floating gate (20) by applying a high write voltage bias to the write / erase gate (30) while grounding the at least one p-well; and performing an erase operation to discharge the floating gate (20) by applying a high erase voltage bias to the at least one p-well while grounding the write / erase gate (30), wherein the high erase voltage bias may be the same or different voltage than the high write voltage bias.

11. The method of Claim 8, wherein the EEPROM cell is a p-channel cell comprising at least one n-well, and the method comprises: performing the read operation to read the charge on the floating gate (20) by applying a low read voltage bias to the at least one n-well while grounding the write / erase gate (30); performing the write operation to charge the floating gate (20) by applying a high write voltage bias to the at least one n-well while grounding the write / erase gate (30); and performing an erase operation to discharge the floating gate (20) by applying a high erase voltage bias to the write / erase gate (30) while grounding the at least one n-well, wherein the high erase voltage bias may be the same or different voltage than the high write voltage bias.

12. The method of Claim 8, wherein the first and second portions (34A, 34B) are first and second read gates (34A, 34B), and wherein the method comprises independently biasing the first and second read gates (34A, 34B).

13. A memory cell array, comprising: the EEPROM of claims 1-7.

14. The memory cell array of Claim 13, wherein: the array of EEPROM cells comprises a plurality of cell rows; and at least two cell rows share a common source line.