Method for forming high density nonvolatile memories with high capacitive-coupling ratio

[0002] This invention is a continuation-in-part application filed with a Ser. No. 09/046,343, filed Mar. 23, 1998, titled “High Density/Speed Nonvolatile Memories with a Textured Tunnel Oxide and a High Capacitive-Coupling Ratio”, and assigned to same assignee with the same inventor as the present application.

[0003] The present invention relates to a method for fabricating a nonvolatile memory cell, and more especially, to a method for fabricating rugged tunnel oxide with high electron injection efficiency and a large charge-to-breakdown for low power nonvolatile memory.

[0004] Nonvolatile memories, including mask read-only memories (Mask ROM), programmable ROM (PROM), erasable programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM or E²PROM) and flash memories, retain their memory data whenever the power is turned off, and have wide applications in the computer and electronic industry. In recent years, the markets of the portable computers and telecommunications develop rapidly and become a major driving force in semiconductor integrated circuit's design and technology. As stated by A. Bergemont, et al., in “Low Voltage NVG™: A New High Performance 3 V/5 V Flash Technology for Portable Computing and Telecommunications Application”, IEEE Trans. Electron Devices Vol. 43, p. 1510, (1996), it creates a great need for low power, high density, and electrically re-writable nonvolatile memories. That is, the memories programmable and erasable as EPROM, E²PROM or flash memories are required for aforementioned systems to store operating systems or application software.

[0005] The basic storage cell of these programmable and erasable memories contain a double polysilicon storage transistor with a floating gate isolated in silicon dioxide and capacitively coupled to a second control gate which is stacked above it. The E²PROM cell further comprises an access, or select, transistor. These memories execute the programming and erasure by charging or discharging their floating gates. For example, the EPROM is programmed by hot electron injection at the drain to selectively charge the floating gate and erased by discharging the floating gate with ultraviolet light or X-ray, which the latter has never been commercially applied for this purpose. The E²PROM and most of the flash memories are programmed by hot electron injection or cold electron tunneling named Fowler-Nordheim tunneling, and erased mostly by Fowler-Nordheim tunneling from the floating gate to the source, with the control gate ground.

[0006] Fowler-Nordheim tunneling, or cold electron tunneling, is a quantum-mechanical effect, which allows the electrons to pass through the energy barrier at the silicon-silicon dioxide interface at a lower energy than required to pass over it. H. Shirai, et al., stated in their paper “A 0.54 μm²Self-Aligned, HSG Floating Gate Cell for 256 Mbit Flash memories”, IEDM Tech. Dig. Vol. 95, p. 653, (1995) that, because of its low current consumption, the Fowler-Nordheim program/erase scheme becomes indispensable for low power operation of the E²PROM and flash memories. However, the Fowler-Nordheim program/erase scheme requires high voltage applied to control gate of the memory cell due to its need for a large reversible electric field to the thin oxide separating the floating gate from the substrate. Therefore, to lower the control gate bias, the memory cell must have a high capacitive-coupling ratio structure.

[0007] Y. S. Hisamune, et al., propose a method for fabricating a flash memory cell with contactless array and high capacitive-coupling ratio in “A High Capacitive-Coupling Ratio Cell for 3 V-Only 64 Mbit and Future Flash Memories”, IEDM Tech. Dig. Vol. 93, p. 19, (1993). However, this method achieves high capacitive-coupling ratio with four times of polysilicon deposition and has a complex fabrication. In addition, this cell structure makes it difficult to scale the size down and increase the integration of the memory due to its short tunnel oxides. Furthermore, as mentioned by C. J. Hegarty, et al., in “Enhanced Conductivity and Breakdown of Oxides Grown on Heavily Implanted Substrates”, Solid-State Electronics, Vol. 34, p. 1207, (1991), it is also difficult to fabricate a thin tunnel oxide on the heavily doped substrate with a high electron injection efficiency and a large charge-to-breakdown for low power nonvolatile memories. Thus, to reach high capacitive-coupling ratio, high electron injection efficiency and a large charge-to-breakdown with a simple manufacture is the subject of high density and low power nonvolatile memories today.

[0008] A method for fabricating a nonvolatile memory cell with rugged tunnel oxide is disclosed. First, the field oxide is formed, the active region is defined, and a semiconductor substrate is prepared. A stacked silicon oxide/silicon nitride layer is deposited on the substrate and then the tunnel oxide region is defined by a standard photolithography process followed by an anisotropic etching. A high temperature steam oxidation process is used to grow a thick thermal oxide on the non-tunnel region. After removing the masking silicon nitride layer, the phosphorus ions are implanted to form the doped regions and serve as source and drain, then a thermal annealing is performed to recover the implantation damage and to drive in the doped ions. Next, the pad oxide film is etched back, and a thin, undoped HSG-Si film is deposited. A wet etching process is performed to selectively etch back the HSG-Si film and the doped substrate regions by HNO₃/CH₃COOH/HF/DI, or phosphoric acid (H₃PO₄) at a temperature of about 140 to 170° C. The doped substrate regions are etched with a high etching rate than undoped HSG-Si, and rugged topography of the doped substrate region is then formed. Thereafter, a thin oxide is grown on the rugged doped substrate region to form a rugged tunnel oxide. Finally, the first n+ doped polysilicon film which serves as the floating gate, the interpoly dielectric such as NO or ONO, and the second n+ doped polysilicon film which serves as the control gate are sequentially formed, and the memory cell is finished.

[0009] The foregoing aspects and many of the attendant advantages of this invention will become more readily appreciated as the same becomes better understood by reference to the following detailed description, when taken in conjunction with the accompanying drawings, wherein:

[0010] FIG. 1 is a cross-sectional view of a semiconductor wafer illustrating the step of forming a silicon oxide layer and a silicon nitride layer on the substrate according to the present invention;

[0011] FIG. 2 is a cross-sectional view of a semiconductor wafer illustrating the step of defining the tunnel oxide regions on the substrate according to the present invention;

[0012] FIG. 3 is a cross-sectional view of a semiconductor wafer illustrating the step of growing a thick thermal oxide on the substrate according to the present invention;

[0013] FIG. 4 is a cross-sectional view of a semiconductor wafer illustrating the step of forming the source and drain regions on the substrate according to the present invention;

[0014] FIG. 5 is a cross-sectional view of a semiconductor wafer illustrating the step of performing a high temperature annealing and removing the pad oxide film according to the present invention;

[0015] FIG. 6 is a cross-sectional view of a semiconductor wafer illustrating the step of depositing a thin, undoped HSG-Si film on the substrate according to the present invention;

[0016] FIG. 7 is a cross-sectional view of a semiconductor wafer illustrating the step of performing a wet etching to etch back the undoped HSG-Si film and the doped substrate to form a rugged substrate surface according to the present invention;

[0017] FIG. 8 is a cross-sectional view of a semiconductor wafer illustrating the step of forming a thin tunnel oxide on the rugged substrate according to the present invention;

[0018] FIG. 9 is a cross-sectional view of a semiconductor wafer illustrating the step of depositing an n+ poly-Si film and then defining the floating gate according to the present invention;

[0019] FIG. 10 is a cross-sectional view of a semiconductor wafer illustrating the step of forming an ultra-thin interpoly dielectric on the floating gate according to the present invention;

[0020] FIG. 11 is a cross-sectional view of a semiconductor wafer illustrating the step of depositing another n+ poly-Si film and then defining the control gate according to the present invention;

[0021] The present invention proposes a simple method to fabricate a nonvolatile memory cell with high capacitive-coupling ratio. The method described here includes many process steps well known in the art like photolithography, etching or chemical vapor deposition (CVD) which are not discussed in detail. In addition, the present invention utilize a method for forming rugged tunnel oxide to attain high electron injection efficiency and a large charge-to-breakdown for low power nonvolatile memories.

[0022] Referring to FIG. 1, a single crystal silicon substrate 2 with a <100> crystallographic orientation is provided. A silicon oxide layer 4 is formed on the surface of the substrate 2. In addition to a pad oxide for the oxidation mask, the silicon oxide layer 4 can be used to act as a sacrificial oxide to prevent the channel effect during the later ion implantation. This pad oxide layer 4 has a thickness of about 40-300 angstroms, and can be grown by using thermal oxidation at a temperature of about 800-1100° C., or using low pressure chemical vapor deposition (LPCVD) at a temperature of about 400-750° C. Next, a thick silicon nitride layer 6 is deposited, for example, using a LPCVD process at a temperature of about 700-800° C., on the pad oxide layer 4 to serve as an oxidation mask.

[0023] The field oxide (FOX) pattern is now defined by using a conventional manner of photolithography including photoresist coating, exposure, and development processes, and then a dry etching is carried out to etch the thick silicon nitride layer 6 and the pad oxide layer 4. After photoresist is removed and wet cleaned, a thermal oxidation in an oxygen steam environment is performed, and the thick field oxide regions 8 are grown with a thickness of about 3000-8000 angstroms, to provide isolation between active regions on the substrate 2. Then, the silicon nitride layer 6 is optionally removed, and a new silicon nitride layer 10 is created over the substrate 2.

[0024] Turning next to FIG. 2, another photolithography process is used to define the tunnel oxide region. An etching step follows to selectively etch the silicon nitride layer 10 but not the pad oxide layer 4 and expose a portion of the pad oxide layer 4 which defines the non-tunnel region on the active region. This selectivity can be reached by a dry etching process using NF₃as the plasma source or by a wet etching process using hot phosphoric acid as the etching solution. It is noted that the wet etching will cause the undercut. A high temperature steam oxidation is then performed at a temperature of about 800-1100° C. to grow a thick thermal oxide 12 on the non-tunnel region, as shown in FIG. 3. This thermal oxide 12 has a thickness of about 300-2500 angstroms, and can raise the capacitive-coupling ratio of the memory cell.

[0025] Turning next to FIG. 4, the masking silicon nitride film 10 is removed by a wet etching with hot phosphoric acid. An n+ ion implantation is performed to implant phosphorus ions through the silicon oxide layer 4, but not the thick oxide 12, into the substrate 2 to form doped substrate region 14 and serve as the source and drain. The implantation energy and dosage are respectively about 0.5 to 150 keV and about 5×10¹⁴to 5×10¹⁶atoms/cm². During the ion implantation, the silicon oxide layer 4 act as a buffer to prevent the substrate 2 from damage and to prevent the doped ions from channel effect. The substrate 2 is then thermal annealed to recover the implantation damage by a preferable method as rapid thermal processing (RTP) at a temperature of about 800 to 1150° C. The dopants are activated and driven in to form the best distribution profile at this step, as shown in FIG. 5. The silicon oxide 4 is now removed, with a suitable etchant such as buffered oxide-etching (BOE) solution or diluted solution of hydrofluoric acid (HF).

[0026] Referring to FIG. 6, a thin, undoped hemispherical grained silicon (HSG-Si) film 16 is deposited over the substrate 2, with a thickness of about 20-300 angstroms. In one embodiment, the deposition rate of the undoped hemispherical grained silicon can be controlled at about 20 angstroms/min.

[0027] Thereafter, a wet etching process is performed to etch back the undoped hemispherical grained silicon (HSG-Si) film 16 and the doped substrate region 14 in the substrate. In one embodiment, the recipe of the etchant can be selected to be HNO₃/CH₃COOH/HF/DI. As disclosed by S. H. Woo et al. in their work “Selective Etching Technology of in-situ Doped Poly-Si (SEDOP) for High Density DRAM Capacitors” (Symposium on VLSI Tech., p.25, 1994), a high selectivity to phosphorus-doped silicon over undoped HSG-Si of about 60 to 80, will be exhibited. The exposed portion of the doped substrate region 14 is then etched much faster and thus deeper than undoped HSG-Si 16. Alternatively, hot phosphoric acid (H₃PO₄) at a temperature of about 140 to 170° C. can also be applied as the etchant in another embodiment. According to the paper “A Novel Stacked Capacitor with Porous-Si Electrodes for High Density DRAMs” by Hirohito Watanabe et al. in Symposium on VLSI Tech., p.17, 1993, a porous-Si structure and micro cavity surface will be formed on the doped substrate region 14. After etching back the HSG-Si film 16 by above wet solution, the topography of the phosphorus-doped substrate region 14 in the substrate is then become rugged as shown in FIG. 7.

[0028] Referring to FIG. 8, a thin oxide layer 18 is now formed on the rugged doped region 14. The thin oxide layer 18 can be formed by thermal oxidation performed in a dry oxygen ambience at a temperature of about 750-1050° C., or by a chemical vapor deposition (CVD). Alternatively, a nitridation followed by a re-oxidation can be applied to form this oxide layer 18. With underlying rugged surface of the doped substrate region, a rugged tunnel oxide 18 is then formed, and a rugged Si/SiO₂interface is obtained. As mentioned by S. L. Wu et al. in “Characterization of Thin Textured Tunnel Oxide Prepared by Thermal Oxidation of Thin Polysilicon Film on Silicon”, IEEE Trans. Electron Devices, Vol. 43, p. 287, (1996), this rugged interface will result in localized high electric field and subsequently enhance the electron injection from the substrate 2 into oxide. Thus, the rugged tunnel oxide exhibits higher electron-injection efficiency, a significantly lower charge trapping rate, and a large charge-to-breakdown in comparison with the conventional tunnel oxide.

[0029] Next, referring to FIG. 9, the conductive layer 20 is deposited on the substrate 2 preferably with a material of doped or in-situ doped n+ polysilicon by using a conventional LPCVD. A standard photolithography process is used to define the floating gate pattern. An anisotropic etching with Cl₂, HCl₂or SiCl₄as the plasma source is then carried out to etch the conductive layer, thereby the floating gate 20 is formed on the active region and a portion of the field oxide region.

[0030] The ultra-thin interpoly dielectric (IPD) layer 22 deposited on the surface of the floating gate 20 is shown in FIG. 10. This interpoly dielectric layer 22 can be a material of a double film of silicon nitride and silicon oxide (NO), a triple film of silicon oxide, silicon nitride and silicon oxide (ONO), or any other high dielectric constant film such as tantalum pentoxide (Ta₂O₅) or barium strontium titanate (BST). Finally, referring to FIG. 11, another conductive layer formed of doped or in-situ doped n+ polysilicon is deposited and patterned on the interpoly dielectric layer 22 to serve as the control gate 24. Thus, the nonvolatile memory cell with rugged tunnel oxide is finished according to the present invention.

[0031] As is understood by a person skilled in the art, the foregoing preferred embodiments of the present invention are illustrated of the present invention rather than limiting of the present invention. It is intended to cover various modifications and similar arrangements included within the spirit and scope of the appended claims, the scope of which should be accorded the broadest interpretation so as to encompass all such modifications and similar structure.