Methods of fabricating lateral double-diffused metal oxide semiconductor devices

The present disclosure relates generally to lateral double-diffused metal oxide semiconductor devices (hereinafter referred to as “LDMOS”) and, more particularly, to methods of fabricating LDMOS transistors.

Because MOS field effect transistors (hereinafter referred to as “MOSFETs”) have higher input impedance than bipolar transistors, their power gains are high and their gate driving circuits are very simple. Generally, when devices are turned off, minority carrier storage or minority carrier recombination causes time delay. However, because the MOSFET is a unipolar device, it does not experience any such time delay. Thus, applications (e.g., switching mode power supplies, lamp ballast and motor driving circuits), for MOSFETs are expanding. MOSFETs usually adopt the DMOSFET(double diffused MOSFET) structure embodied by planar diffusion technology. A typical LDMOS transistor is described in Sel Cloak, U.S. Pat. No. 4,300,150. Also, an LDMOS transistor integrated with a CMOS transistor and a bipolar transistor, was described on pages 322–327 of the “ISPSD 1992” in a paper entitled “A 1200 BiCMOS Technology and Its Application”, by Vladimir Rumennik and on pages 343–348 of the “ISPSD 1994” in a paper entitled “Recent Advances in Power Integrated Circuits with High Level Integration,” by Stephen P, Robb.

It is important for DMOS transistors to be employed with power devices which can handle high voltage. Such power devices have good current handling capacity per unit area or a good ON-resistance per unit area. Because the voltage ratio is fixed, the ON-resistance per unit area can be reduced by decreasing a cell area of the MOS device.

In the field of power transistors, a cell pitch of a device is determined by the combined width of a polysilicon region and a contact region, which form a gate electrode and a source electrode, respectively. Reducing a P-type well junction depth is a well-known technique for diminishing the width of a polysilicon region in a DMOS power transistor. However, the breakdown voltage restricts the junction depth.

A conventional LDMOS device is well suited to VLSI processes due to its simple structure. Nevertheless, these LDMOS devices have been regarded as less attractive than VDMOS (vertical DMOS) devices. Recently, RESURF (reduced surface field) LDMOS devices have been shown to have a good ON-resistance characteristic. However, their structure is very complex, they can only be applied for devices whose sources are earthed, and they are difficult to use in other applications.

Particularly, in the past, DMOS transistors were used as discontinuous power transistors or elements of monolithic integrated circuits. Because DMOS transistors are fabricated using a self-aligned manufacturing procedure, they basically comprise a semiconductor substrate.

To form a self-aligned channel region with a gate electrode, a channel body region is generally formed by implanting either p-type dopants or n-type dopants through apertures within a mask, which is made of materials for the gate electrode. A source region is formed by implanting conductive dopants opposite to what are used for the channel body region. The source region is then self-aligned to both the gate electrode and the channel body region. This is a reason why the DMOS transistor has a compact structure.

Referring to FIG. 1 , an LDMOS transistor device 10 actually has two LDMOS transistors 10 a and 10 b.

The transistor device 10 a is formed on a SOI (silicon on insulator) substrate comprising a silicon substrate 11 , a buffer oxide layer 12 and a semiconductor layer 14 . In the illustrated example, the semiconductor layer 14 is formed over the silicon substrate 11 . This prior art FET (field effect transistor) comprises a source region 16 a and a drain region 18 a . The N-type doped source region 16 a is formed within a P-type doped well region 20 . The well region 20 is often called a P-type body. The P-type body 20 may extend to the upper surface of the buffer oxide layer 12 or be only within the semiconductor layer 14 .

The drain region 18 a is in contact with one end of a field insulation region 23 a . The field insulation region 23 a includes a field oxide layer such as a thermally grown silicon oxide layer.

A gate electrode 26 a is formed on the surface of the semiconductor layer. The gate electrode 26 a extends from the upper part of the source region 16 a to the upper part of the field insulation region 23 a . It is made of polysilicon doped with impurities. The gate electrode 26 a is isolated from the semiconductor layer 14 by a gate dielectric 28 a . The gate dielectric 28 a may comprise oxide, nitride or a combination thereof (e.g., a stacked NO or ONO layer)

Sidewall insulation regions (not shown) may be formed on the sidewalls of the gate electrode 26 a . The sidewall insulation regions commonly comprise oxide such as silicon oxide or nitride such as silicon nitride.

A body region 30 doped at a high concentration exists within the P-type body 20 . This body region 39 is in good contact with the P-type body 20 . It is doped at a higher concentration than the P-type body 20 .

A source contact plug 34 and a drain contact plug 32 a exist within the transistor device 10 a . These contact plugs ( 34 and 32 a ) are provided to electrically connect the source region 16 a and the drain region 18 a to other elements of the circuit. Referring to FIG. 1 , the single contact plug 34 is used for the source regions, 16 a and 16 b , of the two transistors, 10 a and 10 b . The prior art technology described above is further described in Ng et al., U.S. Pat. No. 5,369,045.

However, during a diffusion process for forming channels, the prior art requires a thermal treatment process at a high temperature, which affects devices badly.

FIG. 1 is a cross-sectional view illustrating a prior art LDMOS device.

FIGS. 2 a through 2 d are cross-sectional views illustrating an example process of fabricating an LDMOS device performed in accordance with the teachings of the present invention.

Referring to FIG. 2 a , plural device isolation structures are formed by a shallow trench isolation (hereinafter referred to as “STI”) process. More specifically, a pad oxide layer is first formed on a semiconductor substrate 100 . A nitride layer is formed on the pad oxide layer. Next, a photoresist layer is provided on the nitride layer. The photoresist layer is then patterned. A trench for device isolation is formed by etching the substrate using the photoresist pattern as a mask. The photoresist pattern is then removed. Finally, an insulation layer is formed on the substrate. Device isolation structures 101 are then formed by performing a planarization process. During the planarization process, the nitride layer is used as an etching stop layer. Thus, when the nitride layer is exposed, the planarization process is stopped and the nitride layer is removed.

Referring to FIG. 2 b , a photoresist pattern 102 is formed on the substrate where the device isolation structures 101 are formed. In the illustrated example, the photoresist pattern is patterned to expose the region on which an LDMOS structure is to be formed. Next, ions are implanted into the exposed region 103 by using both the photoresist pattern and the device isolation structure as masks. In the illustrated example, Ga ions or B ions, (a Group 13 element), are implanted to form a P-type ion-implanted layer 103 a . Then, P ions or As ions, (a fifth family element), are implanted to form an N-type ion-implanted layer 103 b.

Referring to FIG. 2 c , a self-aligned double diffused well is formed and the device isolation structure of the open region is removed. Specifically, the ion-implanted layers are diffused to form a P-type well 105 and an N-type well 106 by performing rapid thermal processing (hereinafter referred to as “RTP”) on the ion-implanted substrate. A short channel may be automatically formed by adjusting the implant energy of the Group B element and the Group 15 element. The substrate in which the P-type well 105 and the N-type well 106 are formed is then wet etched to remove the device isolation structure of the open region. After a gate oxide layer 107 and a polysilicon layer are deposited on the substrate, a gate electrode 104 is formed by patterning the polysilicon layer and the gate oxide layer 107 . Finally, the photoresist pattern is removed.

Referring to FIG. 2 d , an LDMOS is subsequently completed by performing well-known processes.

In view of the foregoing, persons of ordinary skill in the art will appreciate that the disclosed methods can form self-aligned double diffused wells by using the insulation layers of STI regions as masks and by adjusting ion implant energies. As a result, a high temperature thermal treatment process can be skipped. Consequently, the illustrated method fabricates a semiconductor with a narrow linewidth.

It is noted that this patent claims priority from Korean Patent Application Serial Number 10-2003-0101107, which was filed on Dec. 31, 2004, and is hereby incorporated by reference in its entirety.

Although certain example methods, apparatus and articles of manufacture have been described herein, the scope of coverage of this patent is not limited thereto. On the contrary, this patent covers all methods, apparatus and articles of manufacture fairly within the scope of the appended claims either literally or under the doctrine of equivalents.