Fabricating Method of an Ultra-Fast Pseudo-Dynamic Nonvolatile Flash Memory

1. Field of the Invention

The invention relates to the fabricating method of a flash memory, and more particular to the fabricating method of a flash memory which comprises a structure of a dynamic random access memory (DRAM).

2. Description of the Related Art

The development of nonvolatile devices has been intensively studied and progressed very quickly. Due to the inconvenience and low density, the conventional nonvolatile device is gradually substituted by a flash memory. The flash memory is therefore becomes more and more important.

Conventional flash memory arrays can be categorized into an NAND structure and an NOR structure, which are described as follows:

In 1990, a flash memory of Bi-polarity W/E NAND structure (of which the equivalent circuit is shown as FIG. 1A) was disclosed in a paper, "A NAND Structured Cell with a New Programming Technology for Highly Reliable 5V-only Flash EEPROM", by Toshiba. The paper is published in "VLSI Technology, pp. 129-130" by R. Kirisawa et al.

In 1992, a divided bit line NOR (DINOR) structure of a flash memory array (of which the equivalent circuit is shown as FIG. 1B) was disclosed in a paper, "A Novel Cell Structure Suitable for a 3V-Only Operation, Sector Era se Flash Memory", by Mitsubishi. The paper is published in "IEEE Tech. Dig. IEDM, pp.599-602", by H. Omoda et al.

In 1994, a NOR, AND structure of a flash memory array (of which the equivalent circuit is shown as FIG. 1C) was disclosed as the paper, "A 0.4-.mu.m.sup.2 Self-Aligned Contactless Memory Cell Technology Suitable for 256-Mbit flash memories" by Hitachi. The paper is published in "IEEE Tech. Dig. IEDM, pp.921-923", by M. Kato et al.

In 1995, a Dual String NOR (DuSNOR) structure of a flash memory array (of which the equivalent circuit is shown as FIG. 1D. It is similar to the disclosure presented by NEC in 1993) disclosed by Samsung. The paper, "A novel Dual String NOR Memory Cell Technology Scaleable to the 256Mbit and 1 Gbit Flash Memories" is published in "IEEE Tech. Dig. IEDM, pp.263-266", by K. S. Kim et al.

In 1995, an Asymmetrical Contactless Transistor (ACT) structure of a flash memory array (of which the equivalent circuit is shown as FIG. 1E) was disclosed by Hitachi. The paper, "A New Cell Structure for Sub-quarter Micron High Density Flash Memory", is published in "IEEE Tech. Dig. IEDM, pp.267-270", by Y. Yamauchi et al.

Referring to FIG. 2 and FIG. 1A to FIG. 1E, the common part of the above flash memory array structures can be represented by an equivalent circuit shown in FIG. 2. The conventional flash memory structure comprises a select transistor Q.sub.s2 and several memory cell transistors Q.sub.m2. The memory cell transistors Q.sub.m2 is connected to select transistors Q.sub.s2 via a local bit line (LBL) or a sub-bit line (SBL). A global bit lines GBL and the SBL are gated by the select transistor Q.sub.s2. Thus, only when the select transistor Q.sub.s2 is selected, the SBL can be connected to the GBL. Moreover, the memory cell transistors Q.sub.m2 are transistors with a floating gate which is, however, only a convenient way to describe a memory cell transistor. It does not mean that there is only one kind of memory cell transistor.

When the conventional flash memory is being coded or written, some select transistor Q.sub.s2 of the memory array must be in "on" status to write a signal into the memory cell transistors Q.sub.m2. Because the GBL voltage must be held until programming is completed, while writing a single memory cell, no signal can be written into any other memory cells sharing the same main bit line with this single memory cell before writing is completed. Namely, writing another signal of data into the main bit line has to wait until the previous coding is completed. Therefore, while coding or writing into a flash memory, there is only one coding process in act at a time in the memory cells which share the same main bit line. That is, there is only one signal allowed to write at a time. Thus, the operation speed of memory cells is limited and lowered.

It is therefore an object of the invention to provide a fabricating method of an ultra-fast pseudo-dynamic nonvolatile flash memory (UP flash memory) By coupling a capacitor to a local bit line, or serially connecting a capacitor to a controlling transistor and then to a local bit line, the operation speed of the flash memory is increased.

To achieve the objects and advantages, and in accordance with the purpose of the invention, as embodied and broadly described therein, the invention is directed towards a fabricating method of an UP flash memory. The characteristic of the UP flash memory is that there is a capacitor coupling to a local bit line. The fabricating method of the UP flash memory comprises first that a silicon substrate comprising a P-well is provided. The substrate is defined into a memory cell region and a peripheral region. In addition, a field oxide layer is formed on the substrate to achieve the electrical isolation. A first gate oxide layer and a first polysilicon layer are formed on the memory cell region of the substrate. A first N-type dopant is implanted into the first polysilicon layer. A dielectric layer, such as an oxide/nitride/oxide (ONO) layer, is formed on the memory cell region of the substrate. On the peripheral region, a second gate oxide layer is formed. A second polysilicon layer is formed over the substrate. The second polysilicon layer on the peripheral region is etched to form a gate of the peripheral region. The second polysilicon layer, the ONO layer, and the first polysilicon layer are etched to form a stack gate layer of the memory cell region. A part of the field oxide layer is removed by self-aligned source etching. The step of self-aligned source etching to remove a part of the field oxide layer is to connect the first sources. The drain of the memory cell region is lightly doped by a second N-type dopant. The LDD of periphery CMOS is formed by a third N-type dopant for N-channel device and a first P-type dopant for P-channel device, respectively. An oxide spacer is formed on the substrate. A heavily doped source/drain region in the memory cell region is doped with a fourth N-type dopant for N-channel and a second P-type dopant for P-channel, respectively. A protecting first oxide layer is formed to cover the whole device with a first opening, so that the first drain is exposed. A first tungsten plug is formed to fill the opening. After the formation of the tungsten plug, a third poly-silicon layer is formed over the substrate. A fifth N-type dopant is implanted into the third polysilicon layer which is then etched and patterned. A oxy-nitride layer is formed over the substrate and covers the entire third polysilicon layer. A first metal layer is formed over the substrate and at least fills the second opening. The metal layer is then etched and patterned. The first metal layer, the oxy-nitride layer, and the third polysilicon layer form a capacitor, and the third polysilicon layer is the local bit line.

According to the object of the invention, another fabricating method of an UP flash memory is provided. The characteristic of the UP flash memory is that there is a capacitor coupling to a local bit line. The fabricating method of the UP flash memory comprises first providing a silicon substrate comprising a P-type well. The substrate is separated into a memory cell region and a peripheral region. In addition, a field oxide layer is formed on the substrate to achieve the electrical isolation. A first gate oxide layer and a first polysilicon layer are formed on the memory cell region of the substrate. A first N-type dopant is implanted into the first polysilicon layer. A pad oxide layer and a first nitride layer are formed on the memory cell region of the substrate. The first nitride layer, the pad oxide layer, the first polysilicon layer, and the first gate oxide layer are etched to form a floating gate. A second pad oxide layer, a second nitride layer, and a first oxide layer are formed on the memory cell region of the substrate. The first oxide layer is etched to form an oxide spacer. The second nitride layer and the second pad oxide layer which are not covered by the oxide spacer are etched away. A second N-type dopant is implanted to form a first drain (medium doped). A third N-type dopant is implanted into the memory cell region to form a source. By self-aligned thermal oxidation, a self-aligned oxide layer is formed. The first nitride layer and the first oxide layer are removed to expose the first polysilicon. A second polysilicon layer is formed on the memory cell region of the substrate. A third N-type dopant is implanted into the second polysilicon layer, and the second polysilicon layer is etched and patterned. An oxide/nitride/oxide layer is formed on the memory cell region of the substrate. On the peripheral region, a second gate oxide layer is formed. A third polysilicon layer is formed over the substrate and is implanted by a fourth N-type dopant. A titanium silicide layer is formed over the third polysilicon layer. The titanium silicide layer, the third polysilicon layer, and the second gate oxide layer on the peripheral region are etched and patterned. A fifth N-type dopant is implanted into the peripheral region of the substrate to form a third drain and a third source, and connect the third source with the first and the second drain. The third polysilicon layer, the ONO layer, and the second polysilicon layer on the memory cell region are etched. A first P-type dopant is implanted into the memory cell region of the substrate region to form a P-type doped region. A second oxide layer is formed over the substrate with an opening formed therein, so that the first and the second drain is exposed within the opening. A tungsten plug is formed to fill the opening. On the second oxide layer, a fourth polysilicon layer is formed and connects the tungsten plug. A sixth N-type dopant is implanted into the fourth polysilicon layer. The fourth polysilicon layer is then etched and patterned. An oxy-nitride layer is formed and covers the entire fourth polysilicon layer. The oxy-nitride layer is then etched and patterned to expose the tungsten plug. A metal layer is formed over the substrate. The metal layer is then etched and patterned.

The metal layer, the oxy-nitride layer, and the fourth polysilicon layer on the memory cell region form a capacitor, and the fourth polysilicon layer on the memory cell region is the local bit line.

It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the invention, as claimed.

The accompanying drawings, which are incorporated in and constitute a part of the specification, illustrate on embodiment of the invention and together with the description, serve to explain the principles of the invention. In the drawings,

FIGS. 1A to 1E shows various conventional structures of flash memory arrays;

FIG. 2 is an equivalent circuit drawing of the structure of a conventional flash memory;

FIGS. 3A to 3B show an equivalent circuit of the flash memory cell in a preferred embodiment of the invention;

FIGS. 3C and 3D illustrate processing time charts for both a conventional UP flash memory and one fabricated in accordance with the invention.

FIG. 4 shows the structure of a ultra-fast pseudo-dynamic nonvolatile flash memory array in a preferred embodiment of the invention;

FIGS. 4A to 4F are cross sectional views of the fabricating process of the memory structure in FIG. 4;

FIG. 5 shows another the structure of another ultra-fast pseudo-dynamic nonvolatile flash memory array in another preferred embodiment of the invention; and

FIGS. 5A to 5I are cross sectional views of the fabricating process of the memory structure in FIG. 5.

Referring to FIG. 3A and FIG. 3B, the equivalent circuits of ultra-fast flash memory arrays in a preferred embodiments according to the invention are shown. Comparing FIG. 3A and FIG. 3B to FIG. 2, it is known that the capacitor C.sub.3a in the dash-line block 30a, and the capacitor C.sub.3b and the controlling transistor Q.sub.c3b in the dash-line block 30b are the characteristic circuits of the invention. The memory cell transistors Q.sub.m3a, Q.sub.m3b are transistors with a floating gate, respectively. However, it is only a representative structure of a memory cell transistor. It does not mean that this kind of structure is the only kind of memory cell which can be used in the invention. In fact, all kinds of memory cell transistor structures can be applied in the memory array structure in the invention. For example, J. S. Lyu et al. disclosed an MFSFET in the paper "Metal-Ferroelectric-Semiconductor Field-Effect Transistor (MFSFET) for the single Transistor Memory by Using Poly-Si Source/Drain and BaMgF.sub.4 Dielectric, IEEE Tech. Dig. IEDM, pp.503-506". The MFSFET structure can be used as an memory cell transistor, and applied in the memory array structure in the invention.

In FIG. 3A, the select transistor Q.sub.s3a and the memory cell transistor Q.sub.m3a form the basic structure of a flash memory. On the other hand, the select transistor Q.sub.s3a and the capacitor C.sub.3a form the basic structure of the DRAM. Thus, the UP flash memory structure provided in the invention is to combine a DRAM structure with a flash memory structure. In the figure, memory cell transistor Q.sub.m3a is a metal-oxide semiconductor (MOS) with a floating gate used as a memory cell. The charge ejected out of the floating gate is controlled according to Fowler-Nordheim tunneling effect between gate and drain, and the threshold voltage is a reference for recording signals. While writing into memory, the capacitor C.sub.3a and the select transistor Q.sub.s3a are operated as a DRAM. Firstly, the select transistor Q.sub.s3a is charged to an "on" status to transfer the magnitude of the voltage which represents the magnitude of signal from the global bit line GBL.sub.3a, and store the transferred charge into the capacitor C.sub.3a at the terminal A. The select transistor Q.sub.s3a is then turned "off", and therefore, the global bit line GBL.sub.3a can be used to transfer next signal to other capacitors controlled by other select transistors. While writing or executing other action in other memory cell, the voltage difference between the drain (terminal A) and the gate (word line) of the selected cells programs or inhibits programming.

The difference between FIG. 3A and FIG. 3B is the extra controlling transistor Q.sub.c3b in FIG. 3B. The main function of controlling transistor Q.sub.c3b is to isolate the capacitor C.sub.3b. For example, while writing the memory cell, controlling transistor Q.sub.3cb is always in an "on" status. While erasing or reading (reading speed lowers with increasing capacitance), the controlling transistor is in an "off" status and thus, isolates the capacitor C.sub.3b to avoid the interference produced in the circuit.

FIG. 3C shows the processing time chart while coding the UP flash memory provided in the preferred embodiment according to the invention. FIG. 3D shows the processing time chart while coding a conventional UP flash memory. The processing time t.sub.c of writing the capacitor C.sub.3a (FIG. 3A) in the memory structure in the invention is about 1 ps to 100 ns, while writing the flash memory cell needs a time t.sub.f of about 0.1 .mu.s to 10 .mu.s. Referring to FIG. 3C and FIG. 3D, assuming that t.sub.c is 10 ns and t.sub.f is 10 .mu.s. It is known that to write 1001 bits in the memory structure in the invention, a time duration of about 20.01 .mu.s is needed, while 20 .mu.s is needed to write 2 bit in a conventional flash memory structure. Thus, to write a bit in the memory in the invention, 20.01 .mu.s/1001 bits=20 ns/bit is needed, while writing a bit in a conventional flash memory needs 20.mu.s/2 bits=10 .mu.s/bit. Therefore, the array structure of the UP flash memory in the invention is operated in a speed of 500 times (The ratio could be different with different design) of a conventional flash memory.

Methods of fabricating the above UP flash memory structure in the preferred embodiments according to the invention are described as follows.

Method 1

Referring to FIG. 4 and FIG. 4A to FIG. 4F. FIG. 4 shows the array structure 40 of a UP flash memory in the second embodiment according to the invention. FIG. 4A to FIG. 4F are cross-sectional view of the fabricating process of the array 40.

FIG. 4A(a) is a cross section view of FIG. 4 along the intercepting line AA', and FIG. 4A(b) a cross section view of FIG. 4 along the intercepting line BB'. A silicon substrate 40a with a P-well 40a' formed therein is provided. The substrate 40a is divided into a memory cell region 400a and a peripheral region 400b. A field oxide layer 40b is formed on the substrate 40a for electrical isolation. On the memory cell region 400a of the substrate 40a, a first gate oxide 40c is formed. A first polysilicon layer 40d is formed and doped by a first N-type dopant on the memory cell region 400a of the substrate 40a. The thickness of the field oxide layer 40d is about 2000 .ANG. to 4000 .ANG., and it is about 60 .ANG. to 150 .ANG. for the first gate oxide 40c and 1500 .ANG. for the first polysilicon layer. The doping process including the implantation of phosphorus ion with a concentration of about 2E15 cm.sup.-2 to 5E15 cm.sup.-2 at an energy of about 30 KeV to 80 KeV.

FIG. 4B(a) is a cross section view of FIG. 4 along the intercepting line AA', and FIG. 4B(b) a cross section view of FIG. 4 along the intercepting line BB'. A dielectric layer, such as an ONO layer 40e is formed on the memory cell region 400a of the substrate 40a. A second gate oxide layer 40f is formed on the peripheral region 400b of the substrate 40a. A second polysilicon layer 40g is formed over the substrate 40a. The second polysilicon layer 40g on the peripheral region 400b is to be etched first and form a gate layer 40h1. The second polysilicon layer 40g on the memory cell region, the ONO layer 40e, and the first polysilicon layer 40d form a stacked gate layer 40h2.

FIG. 4C is a cross section view of FIG. 4 along the intercepting line AA'. A part of the field oxide layer 40b (not shown) is removed by self-aligned-source etching. A second N-type dopant is implanted into the memory cell region 400a by medium doped drain (MDD) ion implantation to form first doped regions 40i. A third N-type dopant is implanted into the peripheral region by light doped drain (LDD) ion implantation to form second doped regions 40j. For example, the MDD ion implantation is performed by arsenic ions with an energy at about 20 KeV to 60 KeV, and a concentration of about 3.times.10.sup.15 cm.sup.-2, while the LDD is performed by arsenic ions with an energy at about 20 KeV to 60 KeV, and a concentration of about 1.5.times.10.sup.13 cm.sup.-2.

FIG. 4D is a cross section view of FIG. 4 along the intercepting line AA'. An oxide spacer layer 40k is formed on the substrate 40a. A fourth N-type dopant is implanted into the memory cell region 400a of the substrate 40a to form first sources 401, first drains 40m, and a second source 40n. The step of self-aligned-source etching to remove a part of field oxide layer 40b is to ensure that first sources 401 are mutually connected. The fourth N-type dopant is implanted into source/drain region with arsenic ions at an energy of about 20 KeV to 60 KeV with a concentration of about 1E13 cm.sup.-2 to 4E13 cm.sup.-2.

FIG. 4E is a cross section view of FIG. 4 along the intercepting line AA'. A first oxide layer 40o is formed over the substrate 40a. An opening 40p is formed within the first oxide layer 40o, so that the first drains 40m and the second source 40n are exposed. A tungsten plug is formed and fills the opening 40p. A third polysilicon layer 40q is formed on the memory cell 400a of the substrate 40a, implanted by a fifth N-type dopant, and etched and patterned. An oxy-nitride layer 40r is formed and patterned over the substrate 40a to cover the entire third polysilicon layer 40q. The thickness of the first oxide layer 40o is about 0.5 .mu.m to 2 .mu.m, and it is about 2000 .ANG. to 4000 .ANG. for the third polysilicon layer 40q. The third polysilicon layer 40q is doped with arsenic ions of the concentration about 5E15 cm.sup.-2 at an energy of about 30 KeV to 80 KeV, and thickness of the oxy-nitride 40r is about 50 .ANG. to 200 .ANG..

FIG. 4F(a) is a cross section view of FIG. 4 along the intercepting line AA'. FIG. 4F(b) is a cross section view of FIG. 4 along the intercepting line BB'. A first metal layer 40t is formed and patterned.

The third polysilicon layer 40q is a local bit line of a flash memory array structure. The third polysilicon layer 40q, the oxy-nitride layer 40r, and the first metal layer 40t form a capacitor of a flash memory array structure.

Method 2

Referring to FIG. 5 and FIG. 5A to FIG. 5F, FIG. 5 shows the array structure 50 of a UP flash memory in the second embodiment according to the invention. FIG. 5A to FIG. 5F are cross-sectional view of the fabricating process of the array 50.

FIG. 5A is a cross section view of FIG. 5 along the intercepting line AA'. A silicon substrate 50a with a P-well 50a' formed therein is provided. The substrate 50a is separated into a memory cell region 500a and a peripheral region 500b. A field oxide layer 50b is formed on the substrate 50a for electrical isolation. On the memory cell region 500a of the substrate 50a, a first gate oxide 50c is formed. The thickness of the first field oxide layer 50b is about 2000 .ANG. to 4000 .ANG., and it is about 60 .ANG. to 150 .ANG. for the first gate oxide layer 50c.

FIG. 5B is a cross section view of FIG. 5 along the intercepting line AA'. A first polysilicon layer 50d is formed and doped by a first N-type dopant on the memory cell region 500a of the substrate 50a. A pad oxide layer 50e and a nitride layer 50f are formed on the first polysilicon layer 50d in sequence. The first nitride layer 50f, the first pad oxide layer 50e, the first polysilicon layer 50d, and the first gate oxide layer 50c are etched to form a floating gate structure. The first polysilicon layer 50d is about 1500 .ANG. thick and is doped by phosphorus ions with a concentration of about 1E20 cm.sup.-2 to 5E20 cm.sup.-2. The thickness of the first pad oxide layer 50e is about 200 .ANG. to 400 .ANG., and about 500 .ANG. to 1000 .ANG. for the first nitride layer 50f.

FIG. 5C is a cross section view of FIG. 5 along the intercepting line AA'. A second pad oxide layer 50g, a second nitride layer 50h, and a first oxide layer 50i (not shown) are formed on the memory cell region 500a of the substrate 50a. The first oxide layer 50i is etched back to form an oxide spacer layer 50j. The thickness of the second pad oxide layer 50g is about 200 .ANG. to 400 .ANG., and it is about 500 .ANG. to 1000 .ANG. for the second nitride layer 50h, and 1000 .ANG. to 2000 .ANG. for the first oxide layer 50i.

FIG. 5D is a cross section view of FIG. 5 along the intercepting line AA'. Referring to FIG. 5D and FIG. 5C, the second nitride layer 50h and the second pad oxide layer 50g which are not covered by the oxide spacer layer 50j are etched away. A second N-type dopant is implanted into the memory cell region 500a of the substrate 50a, and a first source 50k, a first drain 501.sub.1, and a second drain 501.sub.2 are formed. The second N-type dopant is arsenic ions with an energy of about 20 KeV to 40 KeV and a concentration of about 5.times.10.sup.15 cm.sup.-2.

FIG. 5E is a cross section view of FIG. 5 along the intercepting line AA'. A self-aligned oxide layer 50m is formed by self-aligned thermal oxidation (SATO) to isolate the word line from the buried diffusion region of the memory array structure. The buried region is the first source 50k, the first drain 501.sub.1, and the second drain 501.sub.2.

FIG. 5F is a cross section view of FIG. 5 along the intercepting line AA'. Referring to FIG. 5F and 5E, the first nitride layer 50f and the first oxide layer 50e over the first polysilicon layer 50d are removed to expose the first polysilicon layer 50d. On the memory cell region 500a of the substrate 50a, a second polysilicon layer 50n is formed and implanted by a third N-type dopant. The second polysilcon layer 50n is then etched and patterned. The third N-type dopant is phosphorus ions with a concentration of about 10.sup.20 cm.sup.-2 to 5.times.10.sup.20 cm.sup.-2.

FIG. 5G(a) is a cross section view of FIG. 5 along the intercepting line AA'. FIG. 5G(b) is a cross section view of FIG. 5 along the intercepting line BB', or a cross section view of FIG. 5(a) along the intercepting line CC'. A dielectric layer, such as an ONO layer 50o is formed on the memory cell region 500a of the substrate 50a. A second gate oxide layer 50p is formed on the peripheral region 500b of the substrate 50a. A third polysilicon layer 50q is formed over the substrate 50a and implanted by a fourth N-type dopant. On the third polysilicon layer 50q, a titanium silicide layer 50r is formed. The titanium silicide layer 50r and the third polysilicon layer 50q are combined as a polycide layer. The thickness of the ONO layer 50o is about 60 .ANG.-80 .ANG./60 .ANG.-20 .ANG./60 .ANG.-80 .ANG., and the thickness of the third polysilicon layer is about 1500 .ANG. to 2000 .ANG.. The third polysilicon layer 50q is doped with arsenic ions at an energy of about 30 KeV to 60 KeV with a concentration of about 5.times.10.sup.15 cm.sup.-2. With the formation of the titanium silicide layer over the third polysilicon layer 50q, a low resistance of can be obtained.

FIG. 5H(a) shows the top view of FIG. 5. FIG. 5H(b) is a cross section view of FIG. 5 along the intercepting line BB', or a cross section view of FIG. 5(a) along the intercepting line CC'. The titanium silicide layer 50r, the third polysilicon layer 50q, and the second gate oxide layer 50p on the peripheral region 500b of the substrate 50a are etched. A fifth N-type dopant is implanted into the peripheral region 500b of the substrate 50a to form a third drain 50s, and a third source 50s.sub.2. The third source 50s.sub.2 is connected to the first drain 501.sub.1. The titanium silicide layer 50r, the third polysilicon layer 50q, the ONO layer 50o, the second polysilicon layer 50n, and the first polysilicon layer 50d on the memory cell region 500a of the substrate 50a are etched to form a word line WL.sub.50 as shown in the figure. A first P-type dopant is implanted into the memory cell region 500a of the substrate 50a to form a P-type doped region 50t for the purpose of isolating the word line WL.sub.50. The word line WL.sub.50 is formed by the third polysilicon layer 50q and the titanium silicide layer 50r. The P-type dopant is boron ions with an energy at about 20 KeV to 40 KeV and a concentration of about 5.times.10.sup.13 cm.sup.-2.

FIG. 5I is a cross section view of FIG. 5 along the intercepting line BB', or a cross section view of FIG. 5H(b) along the intercepting line CC'. A second oxide layer 50t is formed over the substrate 50a. A first opening 50u is formed within the second oxide layer 50t and exposes the first drain 501.sub.1 and the third drain 50S.sub.1. The first opening 50u is filled by a tungsten plug. On the second oxide layer 50t, a fourth polysilicon layer 50v is formed The fourth polysilicon layer 50v is implanted by a fourth N-dopant and etched and patterned to form the same pattern as the first drain 501.sub.1. An oxy-nitride layer 50w is formed and patterned on the entire surface of the fourth polysilicon layer 50v. Over the substrate 50a, a metal layer 50y is formed and patterned.

The metal layer 50y, the oxy-nitride layer 50w, and the fourth polysilicon layer 50v on the memory cell region 500a form the capacitor structure. The third polysilicon layer 50q and the titanium silicide layer 50r on the memory cell region 500a form the word line. The fourth polysilicon layer 50v is the local bit line.

In the two methods mentioned above, a fabricating method of a UP flash memory is provided and interpreted by the memory array structure 40 and 50 shown in FIG. 4 and FIG. 5, respectively. However, these fabricating methods can form the structure of the dash-line block 3b in FIG. 3B. The only difference is the pattern of the mask.

Other embodiments of the invention are apparent to those skilled in the art from consideration of the specification and practice of the invention disclosed herein. It is intended that the specification and examples to be considered as exemplary only, with a true scope and spirit of the invention being indicated by the following claims.