METHOD OF MAKING A DUAL-POLY NON-VOLATILE MEMORY DEVICE USING

THIRD POLYSILICON LAYER

Technical Field

The present invention pertains to semiconductor devices. Specifically, the present invention pertains to the integration of submicron CMOS devices with non-volatile memory devices.

Background Art

Erasable programmable read only memories, known as EPROMs, and electrically erasable programmable read only memories, known as EEPROMs, are well known "floating gate" devices of the art. Typically, these double layer polysilicon non-volatile memory devices are programmed and accessed using a separate device which is electrical- ly coupled to the memory device. In the past, such pro¬ gramming and accessing has been accomplished using a transistor formed during the formation of the memory de¬ vice. That is, the formation of the transistor was in¬ corporated into the manufacturing process flow of the memory device. Specifically, as the second layer of polysilicon was deposited to form the memory cell, the polysilicon was also deposited onto a separate region of the substrate. A transistor was then formed in that separate region having the second layer of polysilicon as one of the gates of the device. Incorporating the forma¬ tion of the transistor into the manufacturing process flow was considered to be advantageous in that it simpli¬ fied the manufacturing processes required in the forma¬ tion of the devices. Accessing the floating gate device using a high performance submicron CMOS transistor would be especially beneficial due to the high speed at which the submicron CMOS device operates. However, several incompatibilities

exist which inhibit integrating the formation of submi¬ cron CMOS devices, such as high performance N-channel and P-channel transistors, with the manufacturing processes used to form double layer polysilicon non-volatile memory devices such as EPROMs and EEPROMs.

Floating gate devices, such as EPROMs and EE¬ PROMs, require significant oxidation after the deposition of each of the polysilicon layers forming these devices. Multiple poly re-oxidations are necessary to achieve ade- quate charge retention characteristics. Unfortunately, submicron CMOS devices experience significant transcon- ductance and reliability degradation when exposed to ex¬ cessive poly re-oxidation. As a result, performance of submicron CMOS devices exposed to dual-poly formation processes is prohibitively reduced. Specifically, as submicron polysilicon gates are exposed to repeated oxi¬ dation, the edges of the gates tend to lift from the sub¬ strate due to oxidation of the gate edges. This decou¬ ples the gate from the channel region. As a result, gain degradation and hot electron reliability problems occur. Additionally, the re-oxidation thermal cycle causes do¬ pant diffusion of the channel's voltage adjust implant. Furthermore, the operation of dual-poly non¬ volatile memory devices is often incompatible with the use of high performance submicron CMOS devices. EPROMs and EEPROMs frequently require relatively high program¬ ming voltages of 12-18 volts. Such voltages are incom¬ patible with thin gate oxides and lower diode breakdowns found in submicron CMOS devices. Submicron CMOS devices typically have thin gate oxide thicknesses of less than 200 angstroms. A gate oxide of less than 200 angstroms, however, has an intrinsic breakdown of approximately 15 volts. Therefore, the programming voltages utilized in dual-poly non-volatile memory elements essentially de- stroy high performance submicron CMOS devices.

Therefore it is an object of the present invention to successfully integrate the formation and use of high performance submicron CMOS devices with the

manufacture and operation of dual-poly non-volatile memory devices.

Summary of the Invention This object has been achieved by depositing a third layer of polysilicon associated with a non-volatile memory device as one of the gates of a high performance submicron CMOS device. This is done in a manner which decouples the processing for the high performance CMOS device from the processing for the non-volatile memory device allowing for separate optimization of the two device types. We form a layer of thermal oxide over a dual-poly non-volatile memory device and over the portion of the surface of the semiconductor substrate on which the high performance CMOS device is to be formed. The thermal oxide is then removed from the active area on the substrate where the high performance submicron CMOS device is to be formed.

A thin gate oxide is formed over the active area, and a threshold voltage adjust implant is per¬ formed. A third layer of polysilicon is then deposited over the non-volatile memory device and the surface of the semiconductor substrate. The third layer of polysil¬ icon is doped and selectively removed from the surface of the semiconductor substrate such that doped layer of polysilicon is removed from everywhere on the substrate except for the active region where the submicron device gates are to be formed.

A high performance submicron CMOS device having a source, drain and gate, is then formed using the por¬ tion of the doped polysilicon layer remaining in the ac¬ tive region as a gate. Metallized contacts are made to the submicron CMOS device, and the device is covered with a protective coating. Because the devices are formed at separate times, separate optimization of dual-poly non-volatile memory devices and high performance submicron CMOS de¬ vices is possible. Additionally, the present invention

allows the submicron CMOS device to be decoupled from the source and drain diffusion cycles required to achieve higher junction breakthrough voltages in non-volatile memory devices. Furthermore, the separate optimization can be achieved without compromising the characteristics or reliability of either of the devices.

Brief Description of the Drawings

Figs. 1A-E are side sectional views of the steps used in the integration of the high performance submicron CMOS device and the dual-poly non-volatile memory device in accord with the present invention.

Fig. 2 is a circuit diagram of the integrated high performance submicron CMOS device and the dual-poly non-volatile memory device in accord with the present invention.

Best Mode for Carrying Out the Invention

With reference to Fig. 1A, a cross-sectional view of the starting step in the formation of the present invention is shown. A p-doped silicon substrate 20 con¬ taining a p-doped well 22 and having an EPROM 24 formed thereon is shown. Although the semiconductor substrate 20 is formed of silicon in the preferred embodiment, any other suitable semiconductor material may be used. Addi¬ tionally, the substrate 20 may also have a different con¬ ductivity type if desired. Further, although an EPROM 24 is used in the preferred embodiment, an EEPROM is also compatible with the methods of the present invention. The EPROM 24 is formed of two stacked and aligned layers of polysilicon, 26 and 28, formed over a high voltage source 30 and drain 32. The two layers of polysilicon, 26 and 28, are separated by an insulating dielectric layer 34, and are subjected to re-oxidation. After re-oxidation of the two polysilicon layers, 26 and 28, a layer of thermal oxide 36 is formed over the EPROM 24 and the silicon substrate 20. In the preferred

embodiment of the present invention, the thermal oxide 36 is formed to a depth of approximately 200 angstroms.

As shown in Fig. IB, the thermal oxide layer 36 is then removed from a region 38 of the silicon substrate 20 above the p-doped well 22. In so doing, the thermal oxide 36 is cleared from the active region 38 of the sil¬ icon substrate 20 where the thin gate oxide layer of the high performance submicron CMOS transistor is to be formed. In the preferred embodiment, the thermal oxide 36 is removed using a wet HF etch, however, any of the numerous etching techniques well known in the art are suitable.

Referring now to Fig. 1C, a thin gate oxide layer 40 is formed in the active region 38 of the silicon substrate 20. The gate oxide 40 is typically formed to a thickness of approximately 100 to 150 angstroms. In forming the thin gate oxide 40, additional re-oxidation also occurs in the two polysilicon layers, 26 and 28, of the EPROM 24. As a result the polysilicon layers, 26 and 28, are oxidized to a final thickness of about 500 ang¬ stroms. After the formation of the thin gate oxide 40, an enhancement implant 42 is made into the p-doped region 22 of the silicon substrate 20. Enhancement implant 42 is a light dose implant of BF ₂ , or any other well known dopant, which is used to adjust the threshold voltage of the high performance submicron CMOS transistor.

With reference to Fig. ID, a third layer of polysilicon 44 is deposited over the surface of the sili¬ con substrate. As a result, both the thin gate oxide 40 and the EPROM 24 are covered by the layer of polysilicon 44. The polysilicon 44 is typically deposited to a thickness of about 2000 to 5000 angstroms. The third layer of polysilicon 44 is then doped with an n-type do¬ pant such as phosphorous, producing an n ⁺ type conductiv- ity in the third polysilicon layer 44.

As shown in Fig. IE, the third doped layer of polysilicon 44 is then removed from everywhere on the silicon substrate 20 except for the area above the thin

gate oxide 40. In so doing, a gate region 46 for the high performance submicron CMOS transistor is formed. In forming gate 46 of the high performance submicron CMOS transistor from third polysilicon layer 44, the submicron CMOS device is effectively decoupled from the EPROM de¬ vice 24, allowing for separate optimization of the two devices. As a result, the transistor is not adversely affected by the high programming voltages, 12-20 volts, necessary for the EPROM 24. An additional etch step is then performed in order to remove any residual polysili¬ con that may have been deposited onto the sidewalls of the first two polysilicon layers, 26 and 28, during the deposition of the submicron third polysilicon layer 44. The formation of the submicron CMOS transistor is completed by implanting a low voltage source 48 and drain 50, and forming metallized contacts, not shown, to low voltage source 48 and drain 50, and gate 46. The device is then covered with a protective coating. In the preferred embodiment of the present invention, an N-chan- nel type high performance submicron CMOS transistor is formed. However, the methods of the present invention would also apply to the formation of a P-channel type high performance submicron transistor, by forming the transistor in an n-doped well containing a p-doped source and drain region.

Referring now to Fig. 2, a circuit diagram il¬ lustrating the integration of the submicron CMOS device and EPROM 24 of the present invention is shown. EPROM 24 and high performance sub-micron CMOS transistor 60, used to access EPROM 24, are coupled in series between column line 62 and ground line 64. The drain terminal of submi¬ cron CMOS transistor 60 is connected to column line 62 through metal contact 66. The gate of transistor 60 is coupled to access line 68. Additionally, the gate of EPROM 24 is coupled to a read line 70. In so doing, high speed submicron CMOS transistor 60, may be used to access EPROM 24.

Referring again to Fig. 1, the present inven¬ tion as described above has several advantages over the prior art. The third polysilicon layer 44 allows the high performance submicron CMOS transistor to be formed without having to be subjected to the significant re-oxi¬ dations required in the formation of the EPROM 24.

By using a submicron CMOS transistor, the EPROM can be accessed and read at higher speeds than were pos¬ sible with the standard transistors of the prior art. Additionally, the third layer of polysilicon 44 decouples the submicron CMOS transistor and the EPROM 24 such that they may be separately optimized. As a result, both of the devices can be utilized without compromising the characteristics or reliability of the other. Additionally, the two devices can be manufac¬ tured in the same process flow, thereby reducing the man¬ ufacturing cost of the system, by eliminating manufactur¬ ing steps, while simultaneously improving the yield and reliability of the manufacturing processes.