Búsqueda Imágenes Maps Play YouTube Noticias Gmail Drive Más »
Búsqueda avanzada de patentes | Imágenes de página | Historial web | Iniciar sesión

Patentes

  
[merged small][graphic][merged small]
[graphic][merged small][graphic][merged small][graphic][merged small]
[graphic]

1

METHOD FOR FABRICATING
SELF-ALIGNED DOUBLE LAYERED
SILICON-METAL NANOCRYSTAL MEMORY
ELEMENT

5

CROSS-REFERENCE TO RELATED
APPLICATION

This application claims the benefit of Taiwan Application No. 094138939, filed Nov. 7, 2005, the contents of which are 10 hereby incorporated herein by reference in their entirety.

FIELD OF THE INVENTION

The present invention relates to a memory device and a 15 method for fabricating the same, and more particularly, to a nanocrystal memory for accessing charges and a method for fabricating the same.

BACKGROUND OF THE INVENTION 20

Among existing non-volatile memory-based ICs, flash memory is the most advanced of its kind. A distinguishing feature of a conventional flash memory is that, in addition to an oxide insulating layer typically formed in a conventional 25 MOSFET, a floating gate is formed in between a gate and a channel in the MOS of the tradition flash memory, such that data are stored by introducing into or removing from the floating gate negative charges. However, as regards conventional flash memory, the gate is made of doped polysilicon. 30 With the conventional floating gate, electric conduction is achieved, using charges stored in doped polysilicon which forms the floating gate. Charges are stored in the doped polysilicon and thus they move freely within the doped polysilicon. Charges will hardly be maintained in the doped polysili- 35 con, if there is a current leakage from any point of a tunnel oxide layer beneath the polysilicon floating gate. Moreover, the tunnel oxide layer will have to be thinned down if the size of the memory is to be reduced. However, whatever thinningdown effort is subj ect to the physical limit of direct tunneling, 40 and thus a thinning down process does have its own limit.

In view of this, U.S. Pat. No. 5,714,766 proposes a nanostructure memory device. FIG. 1 (PRIOR ART) is a crosssectional view of a nanocrystal memory 100 known in the prior art. The nanocrystal memory 100, which has a basic 45 structure of a transistor comprising a substrate 101, a source 102, a drain 103, a tunnel oxide layer 104, and a gate 105, coupled with nanocrystals 106 embedded in a nanocrystal layer 107 between the gate layers, allows charges to be stored. The proposed nano-structure memory device not only solves 50 the drawbacks of flash memory, namely high operating voltage and slow reading speed, but also enhances memory retention. Unlike a conventional polysilicon floating gate, the proposed nano-structure memory device has charges stored in spaced-apart nanocrystals, and the spaced-apart nanocrystals 55 do not move in relation to each other. The proposed nanostructure memory device overcomes a drawback of the prior art, that is, it is difficult to store charges (wherein charges are stored in polysilicon which forms a floating gate to achieve electric conduction in accordance with the prior art) in the 60 presence of a current leakage from any point of a tunnel oxide layer underlying the conventional polysilicon floating gate. Moreover, the conventional polysilicon floating gate achieves charge blocking solely by being barrier-dependent, thus causing insurmountable difficulty in a thinning-down process per- 65 formed on the tunnel oxide layer. However, the nanocrystals provide insulation required for quantum confinement, so as to

2

store charges in quantum wells. In the absence of an applied voltage, charges are confined to the quantum wells and thereby are stored, allowing the tunnel oxide layer to be thinned down further.

Although the insurmountable difficulty in the foregoing design of the polysilicon floating gate is solved (by thinning down the tunnel oxide layer and thereby enabling charges to enter the quantum wells readily and thereby be stored), but charges stored in the silicon nanocrystal floating gate may readily tunnel out and hardly be retained, because disposed between the channel and the nanocrystals is a mere insulating layer, that is, the tunnel oxide layer.

To address the foregoing problems, academics suggest that charge loss should be prevented by an improved floating gate made of self-aligned double layered silicon (R. Ohba, IEDM, December 2000). Referring to FIG. 2B (PRIOR ART), a nanocrystal structure known in the prior art is replaced with a double layered nanocrystal structure 200. Referring to FIG. 2A (PRIOR ART), the fabricating method comprises the steps of: growing a tunnel oxide layer 204 on a silicon substrate 201; depositing a silicon layer 208 on the tunnel oxide layer 204; forming an oxide layer 209 on the silicon layer 208; depositing a silicon dot 210 (Si dot) on the oxide layer 209; and oxidizing the silicon dot 210 before finalizing the selfaligned double layered silicon structure as shown in FIG. 2B (PRIOR ART). Silicon nanocrystals 211 are formed by a portion of the silicon layer 208 covered by the silicon dot 210 and thereby not oxidized. With the double layered quantum well structure, after charges are stored in the nanocrystals, it is difficult for subsequent charges to enter the nanocrystals because of Coulomb blockade. Accordingly, charges are confined to the quantum wells and are unlikely to tunnel out due to Coulomb blockade; therefore, charge retention is enhanced.

However, this method still has practical drawbacks. Although its structural theory is better than the foregoing single layered nanocrystal, but insurmountable difficulty still exists in practice when fabrication parameter is considered. If the previous level of the operating voltage is to be maintained or even lowered, the oxide layer will have to be thinned down, but the thinning down process will cause a memory retention problem. Although memory retention can be enhanced if the thickness of the oxide layer is increased, the operating voltage is compromised.

In addition to the aforesaid problems, a nanocrystal fabrication process has to address an issue, that is, the control of the formation of nanocrystals during the fabrication process. For instance, it will be impossible to store enough charges in a nanocrystal layer disposed with nanocrystals, if the nanocrystals disposed in the nanocrystal layer are too small or over-dispersed; as a result, the charges present in a channel beneath an oxidation layer decrease, causing difficulty in interpretation. In other words, in the event that few charges are stored in the nanocrystal layer, there will be a negligible difference between a threshold voltage associated with the presence of nanocrystals in the nanocrystal layer and a threshold voltage associate with the absence of nanocrystals from the nanocrystal layer, and in consequence it is impossible to determine whether charges are stored in the nanocrystal layer and perform effective interpretation. Accordingly, a nanocrystal fabrication process should be preferably provided with sufficient nanocrystals to store charges so as to create a significant difference between the threshold voltage associated with the presence of charge storage and the threshold voltage associated with the lack of charge storage, thereby enabling the memory to perform effective interpretation.

3

There are presently two methods for storing more charges, namely enhancing nanocrystal density, and increasing the number of charges stored in each nanocrystal. According to current technology, it is difficult to increase nanocrystal density, and thus considerations are given to attempts to store 5 more charges in each nanocrystal. However, after a charge is stored in a nanocrystal, it is difficult for the next charge to enter the well and be stored in the nanocrystal because of an increase in potential energy. Even if the charge is forcibly stored in the well, the barrier will be insufficient to stop the 10 charge from tunneling out; as a result, the charge tunnels into an oxide layer and fails to be stored in the nanocrystal.

Therefore, metal nanocrystal is used, which has higher charge storage density due to a metal has more states in conduction band compared to the silicon nanocrystal. More- 15 over, a metal with a higher work function can be chosen to be the nanocyrstal to enhance the potential barrier of a quantum well, so as to enhance charge retention.

The urgent problem to be solved is to maintain the foregoing advantages of silicon nanocrystal when operating voltage 20 and charge retention are considered, and keep the foregoing advantages of metal crystals such as containing more charges.

SUMMARY OF THE INVENTION

25

In light of the above drawback of the prior art, a primary objective of the present invention is to provide a method for fabricating a nanocrystal memory element which stores more charges in a nanocrystal charge storage layer.

Another objective of the present invention is to provide a 30 method for fabricating an insulating layer to maintain the quantum confinement of the foregoing nanocrystal to enhance charge retention with combination of the advantage of higher data storing density of a metal nanocrystal.

Still another objective of the present invention is to provide 35 a memory having double layered silicon-metal nanocrystals.

In accordance with the foregoing and other objectives, the present invention proposes a method for fabricating a selfaligned, double layered silicon-metal nanocrystal memory element. The fabricating method comprises the steps of: pro- 40 viding a substrate; growing a tunnel oxide layer on the substrate; depositing a polysilicon layer on the tunnel oxide layer; depositing a dielectric layer on the polysilicon layer; implanting metal ions on the dielectric layer, wherein the metal ions are implanted as deep as the deposited polysilicon 45 layer top; performing thermal oxidation like a rapid thermal annealing process whereby the implanted metal ions are crystallized to form metal nanocrystals; performing on the dielectric layer and the polysilicon layer a selective oxidation process, for example, a rapid thermal annealing process 50 performed on tungsten in the presence of hydrogen-rich steam, such that silicon nanocrystals are formed by the polysilicon layer and the metal nanocrystals thereon by means of self-alignment; forming an integration layer by combining the selectively oxidized polysilicon layer with the dielectric 55 layer; and forming a gate on the integration layer.

The present invention further provides a method for fabricating a self-aligned double layered silicon-metal nanocrystal memory element. The fabricating method comprises the steps of: providing a substrate; growing a tunnel oxide layer on the 60 substrate; depositing a polysilicon layer on the tunnel oxide layer; depositing a dielectric layer on the polysilicon layer; depositing a metal layer on the dielectric layer; depositing another dielectric layer on the metal layer; performing thermal oxidation like a rapid thermal annealing process whereby 65 the metal ions in the metal layer are crystallized to form metal nanocrystals; performing on the dielectric layer and the poly

4

silicon layer a selective oxidation process, for example, a rapid thermal annealing process performed on tungsten in the presence of hydrogen-rich steam, such that silicon nanocrystals are formed by the polysilicon layer and the metal nanocrystals thereon by means of self-alignment; forming an integration layer by combining the selectively oxidized polysilicon layer with the dielectric layer; and forming a gate on the integration layer.

As regards a method for fabricating a self-aligned double layered silicon-metal nanocrystal memory element of the present invention, with selective oxidation, the polysilicon disposed beneath and thereby covered with the metal nanocrystals are not oxidized, whereas the polysilicon not disposed beneath and thereby not covered with the metal nanocrystals are oxidized. With a self-aligned silicon-metal nanocrystal structure formed in the aforesaid manner, the metal nanocrystals and the silicon nanocrystals can be formed by selfalignment.

The present invention further provides a self-aligned double layered silicon-metal nanocrystal memory element comprising a substrate, a tunnel oxide layer formed on the substrate, an integration layer formed on the tunnel oxide layer, and a gate formed on the integration layer. The integration layer comprises a dielectric material, and a plurality of metal nanocrystals and silicon nanocrystals disposed in the dielectric material. The metal nanocrystals and the silicon nanocrystals are formed by self-alignment.

As regards a self-aligned double layered silicon-metal nanocrystal memory element of the present invention, the silicon nanocrystals underlying the metal nanocrystals act as an insulating layer for quantum confinement, such that charges are confined to quantum wells and therefore unlikely to get lost. Moreover, the metal nanocrystals have relatively high work function, and thus potential barrier of the nanocrystals increases. The metal nanocrystals have more states in conduction band to store more charges. Thus potential energy does not increase and potential barrier does not decrease, after a charge has entered. Accordingly, charge retention of a nanocrystal memory is enhanced dramatically.

The present invention further proposes a nanocrystal memory with a self-aligned double layered silicon-metal nanocrystal memory element. The proposed nanocrystal memory comprises a substrate, a source and a drain both formed on the substrate and spaced apart by an appropriate distance, a tunnel oxide layer formed on the substrate and disposed between the source and the drain, an integration layer formed on the tunnel oxide layer, and a gate formed on the integration layer. The integration layer comprises a dielectric material, and a plurality of metal nanocrystals and silicon nanocrystals disposed in the dielectric material. The metal nanocrystals and the silicon nanocrystals are formed by selfalignment.

BRIEF DESCRIPTION OF THE DRAWINGS

The present invention can be more fully understood by reading the following detailed description of the preferred embodiments, with reference made to the accompanying drawings, wherein:

FIG. 1 (PRIOR ART) is a cross-sectional view of a conventional nanocrystal memory;

FIG. 2A (PRIOR ART) is a cross-sectional view of a conventional double layered silicon nanocrystal memory element before oxidation;

FIG. 2B (PRIOR ART) is a cross-sectional view of a conventional double layered silicon nanocrystal memory element;

« AnteriorContinuar »