Method of Fabricating a Flash Memory

1. Field of the Invention

The invention relates in general to the fabrication of an integrated circuit component, and more particularly to the fabrication of a flash memory designed to miniaturize the dimensions of a memory component.

2. Description of the Related Art

An electrical erasable programmable read only memory (EEPROM) is a memory component that is widely used in personal computers and electronic equipment. A conventional EEPROM memory cell has a floating gate transistor structure and has the ability to be programmed, to be erased, and to store data while connected in a circuit. These abilities give EEPROMs advantages over other types of memory. However, EEPROMs suffer from drawbacks, such as a slow storage and retrieval time (about 150 to 200 ns). Recently, a type of EEPROM with a faster storage and retrieval time (about 70 to 80 ns) has been developed, and has been coined "flash" EEPROM memory by Intel Corporation of the United States.

A conventional EEPROM has a floating gate transistor structure as shown in FIG. 1. The floating gate transistor includes a floating gate layer 10, a controlling gate layer 12, a tunnel oxide layer 14, a dielectric layer 15, a drain region 16, and a source region 18. Its operation is based on the so-called Fowler-Nordheim tunneling effect, in that electrons from the drain region 16 tunnel through the tunnel oxide layer 14 and eventually arrive and are stored in the floating gate layer 10. This has the effect of raising the threshold voltage in the floating gate transistor and enables data storage by the transistor. Alternatively, when the situation is reversed, such that the electrons in the floating gate layer are drawn out through the tunnel oxide layer into the source region 18, data is erased. The aforementioned EEPROM floating gate transistor structure has definite limitations when applied to a densely packed IC circuit. Furthermore, using the same tunnel oxide layer for both data programming and data erasure, as in this design, increases the wear rate of the transistor due to the bi-directional flow of electrons.

It is therefore an object of the invention to provide a method for fabricating flash memory such that its component dimensions can be shrunk, it is suitable for use in ultra-high density IC circuits, and it uses two separate tunneling layers, one especially for data programming and the other especially for data erasure, so that the level of wear in each component is greatly reduced.

To achieve the object stated above, the invention provides a method for fabricating flash memory which includes the following steps. First, a protruding plateau is formed on the surface of a silicon substrate. Then, an ion implantation process is performed to form a drain region on the top surface of the protruding plateau, as well as to form source regions in the substrate on either side of the base of the protruding plateau. A gate oxide layer is then formed on each side wall of the protruding plateau, so as to expose only part of the side wall of the drain region. A tunnel oxide layer, that is thinner than the gate oxide layer, is then formed over the source regions and drain region, and a floating gate terminal layer is formed on each side wall of the gate oxide layer and the tunnel oxide layer. A dielectric layer is then deposited on the substrate. Finally, a controlling gate terminal layer is formed on each side wall of the dielectric layer.

Other objects, features, and advantages of the invention will become apparent from the following detailed description of the preferred but non-limiting embodiment. The description is made with reference to the accompanying drawings in which:

FIG. 1 is a cross-sectional view showing the floating gate transistor of a flash memory formed by a conventional technique.

FIGS. 2A to 2H are cross-sectional views showing processing steps for fabricating a flash memory according to a preferred embodiment of the invention.

FIGS. 2A to 2H are cross-sectional views showing processing steps for fabricating a flash memory according to a preferred embodiment of the invention.

First, referring to FIG. 2A, a photoresist layer 32 is formed above a silicon substrate 30 using a photolithographic technique. The photoresist layer 32 exposes a portion of the silicon substrate 30 that will be etched. Then, referring to FIG. 2B, the portion of the silicon substrate 30 that was not covered by photoresist layer 32 is etched away, for example by a dry etching method having an etching depth between about 1000 .ANG. and about 10,000 .ANG.. After that, the photoresist layer 32 is removed, forming a protruding plateau 34.

Next, referring, to FIG. 2C, an ion implantation process is performed to form a drain region 36 on the top surface of the protruding plateau 34, as well as to form source regions 38 in the substrate on either side of the base of the protruding plateau 34. For example, the ion implantation process can be performed at an energy level of about 50 KeV and provide a dosage of about 1.times.10.sup.15 ions/cm.sup.2 using phosphorus or arsenic ions as a doping impurity.

Thereafter, referring to FIG. 2D, a layer of oxide 40 is deposited over the substrate 30, covering the drain region 36 and the source regions 38. For example, a layer of silicon dioxide having a thickness of between about 100 .ANG. and about 300.ANG. may be deposited in an atmosphere of oxygen and nitrogen, with reaction temperature set at about 800.degree. C. and a reaction time set at about 20 minutes. Next, referring to FIG. 2E, part of the oxide layer 40 is selectively etched by employing a plasma dry etching method, exposing part of the drain region 36 and source regions 38, and leaving the remaining oxide layer 40 portions flanking the side walls of the protruding plateau 34 to act as an oxide layer for the gate terminal.

Next, referring to FIG. 2F, tunnel oxide layers 42a and 42b having a thickness between about 50 .ANG. and about 100 .ANG. and thinner than the oxide layer 40 are formed covering exposed portions of source regions 38 and drain region 36, respectively. Then, referring to FIG. 2G, a polysilicon floating gate terminal layer 44 is formed, for example, by performing deposition, ion implantation, and dry etching of a polysilicon layer having a thickness between about 500 .ANG. and about 1500 .ANG..

Referring to FIG. 2H, a stacked polysilicon dielectric layer 46 is formed. For example, this layer can be an oxide/nitride/oxide (ONO) three-layered structure with a deposition thickness of between about 100 .ANG. and about 400 .ANG.. Finally, a polysilicon controlling gate terminal layer 48 is formed on the side walls of the stacked polysilicon dielectric layer 46, for example, by performing deposition, ion implantation, and dry etching of a polysilicon layer.

In summary, a vertical design for a floating gate terminal layer and controlling gate terminal layer structure is fabricated according to the preferred embodiment of the invention, resulting in an overall dimensional reduction of the transistor component. Furthermore, by using separate tunnel oxide layers to carry out data programming and erasure, wear of the component is greatly minimized, resulting in better operational reliability.

Examples of the voltages to be applied at the different device terminals when the flash memory fabricated according to the invention is undergoing data erasure, programming, and read operations are listed in Table 1.

Thus, when the EEPROM is programmed, a negative voltage of -10V is applied to the controlling gate terminal layer 48, a positive voltage of +3V is applied to the drain region 36, and both the source region 38 and substrate 30 are grounded. Powered in this manner, electrons flow from the floating gate terminal layer 44 and pass through the tunnel oxide layer 42b to enter the drain region 36, and therefore electrons in the floating gate terminal layer 44 diminish, resulting in a lowering of the threshold voltage. During a read operation, a positive voltage of +3V is applied to the controlling gate terminal layer 48, a positive voltage of +1V is applied to the drain region, with both the source region and substrate 30 still grounded. For those memory units that have been previously programmed, the threshold voltage is lowered and the memory cell is switched to an ON state. Those memory units that have not been previously programmed remain in an OFF state. When data must be erased, a positive voltage of +12V is applied to the controlling gate terminal layer 48, there is no voltage connection to the drain region 36, and both the source region 38 and substrate 30 are grounded. Powered in this manner, electrons will flow into the floating gate terminal layer 44 from the source region 38 via the tunnel oxide layer 42a. The number of electrons in the floating gate terminal layer 44 will increase, and after awhile will regain the former high threshold voltage level.

While the invention has been described by way of example and in terms of a preferred embodiment, it is to be understood that the invention is not limited thereto. To the contrary, it is intended to cover various modifications and similar arrangements. The appended claims, therefore, should be accorded the broadest interpretation so as to encompass all such modifications and similar structures.