Method for manufacturing a power mos transistor.

BACKGROUND

This invention relates to MOS transistors and more specifically to methods for manufacturing MOS transistors with a minimum number of masking steps. The invention also relates to double diffused MOS ("DMOS") transistors.

A DMOS transistor is a MOS transistor having a channel length defined by the difference in diffusion of sequen­tially introduced impurities from a common edge or boundary. One example of a prior art process for manu­facturing a DMOS transistor is discussed in U.S. Patent 4,443,931. It is known in the art that it is desirable to manufacture DMOS transistors while minimizing the number of masks used. One reason for this is that by minimizing the numbers of masks, the number of alignment steps is also minimized and therefore the need to accommodate alignment tolerances is minimized. As is known in the art, if the need to accommodate alignment tolerances is minimized, the size and cost of the resulting transistor is minimized.

Another reason for minimizing the number of masking steps is that the complexity of the manufacturing process is correspondingly reduced and thus the cost of producing the transistor is also reduced.

SUMMARY

A method for manufacturing a DMOS transistor in accordance with the present invention includes the step of providing a gate insulation layer on a semiconductor substrate. The semiconductor substrate is typically N type silicon and the gate insulation layer typically comprises a silicon nitride layer formed on a silicon dioxide layer. A gate is then formed on the gate insulation layer. In one embodiment of the invention, the gate is polycrystalline silicon and is formed by depositing a polycrystalline silicon layer on the gate insulation layer and using a first photolithographic mask to protect portions of the polycrystalline silicon layer while the exposed portions of the polycrystalline silicon layer are removed.

The resulting gate is covered with a second insulation layer which in one embodiment comprises silicon dioxide. A second photolithographic mask is formed on the wafer. In accordance with one novel feature of this invention, the second mask includes both a first window region defining the deep body region of the DMOS transistor and a second window region defining the gate contact. The portion of the second insulation layer within the second window region is removed. Of importance, the portion of the silicon nitride layer within the first window region is also removed and the underlying portion of the semiconductor substrate is thermally doped or implanted with P type dopant, thereby forming the deep body region.

A silicon dioxide layer is then formed within the second window region, typically by thermal oxidation. Of importance, the thickness of the silicon dioxide within the first region also increases during this process. The exposed portion of the silicon nitride layer and the underlying silicon dioxide layer are then removed using a blanket etching process which does not require additional masking. At the conclusion of the blanket etching process, however, the silicon dioxide formed on the gate and the silicon dioxide formed above the deep body region remain. The silicon dioxide above the deep body region serves as an oxide mask which defines part of subsequently formed body and source regions.

P type dopants are then implanted into the semi­conductor substrate to form the body region of the DMOS transistor. Thereafter, N type dopants are implanted into the semiconductor substrate to form the source region. As mentioned above, the oxide mask grown over the deep body region prevents the underlying deep body region from being predoped or implanted with N type impurities. Of importance, the lateral extent of the source and body regions is defined by the edge of the gate and the oxide mask. Therefore, it is not necessary to use an extra masking step to define the source and body regions of the DMOS transistor.

The oxide mask above the deep body region, the silicon dioxide within the second window, and any silicon dioxide formed above the source region during source and body drive-­in are removed with a blanket etching process. A conductive layer (typically metal) is then formed on the surface of the wafer. A third photolithographic mask is applied to the wafer in order to pattern the conductive layer to form a gate lead and a source/body lead.

The wafer is then coated with a passivation layer of material such as silicon nitride or silicon dioxide at a low temperature (typically less than 450°C). A fourth photolithographic mask is used to define regions where the passivation layer is to be removed, e.g. bonding pad regions. Of importance, the above described process requires only four photolithographic masks. Accordingly, the process is simpler than prior art DMOS manufacturing processes. Further, the need to increase the transistor size in order to allow for misalignment of the various photolithographic masks is reduced.

In one embodiment, the transistor is laterally surrounded by an equipotential ring (EQR). In accordance with one novel feature of the invention, the EQR includes a P region (formed concurrently with the body region), an N+ region within the P region (formed concurrently with the source region) and a conductive ring formed on the N+ region (formed concurrently with the gate lead and the source/body lead). The conductive ring is electrically coupled to the substrate, the P region, and the N+ region, and is biased at the same voltage as the transistor drain. As is explained in greater detail below, the EQR enhances the reliability of the DMOS transistor. Of importance, the EQR is formed without any extra masking steps.

BRIEF DESCRIPTION OF THE DRAWINGS

Figures 1 to 9 illustrate a portion of a DMOS transistor during various steps of a manufacturing process in accordance with the present invention.

DETAILED DESCRIPTION

A process in accordance with the present invention begins with the step of forming a silicon dioxide layer 12 on a semiconductor substrate 10 (Figure 1). In one embodiment of the invention, water 10 consists of a layer of N type epitaxial silicon having a conductivity of about 1.5 ohm-­centimeters and about 12 micrometers thick on a heavily N+ doped substrate. However, other semiconductors materials and materials having other conductivities and conductivity types are also appro­priate. (As used herein, the term "wafer"" includes substrate 10 and all layers formed directly or indirectly on substrate 10.)

Silicon dioxide layer 12 is typically thermally grown to a thickness of about 50 nanometers (nm). A silicon nitride layer 14 is then formed on silicon dioxide layer 12. Silicon nitride layer 14 is also typically about 50 nm thick and typically is formed by chemical vapor deposition. As will be discussed in greater detail below, silicon dioxide layer 12 and silicon nitride layer 14 serve as gate insulation for a subsequently formed DMOS transistor.

A polycrystalline silicon layer 16 is then formed on silicon nitride layer 14. In one embodiment, polycrystal­line silicon layer 16 is formed to a thickness of about 500 nm by chemical vapor deposition. As described in greater detail below, polycrystalline silicon layer 16 serves as the gate of the DMOS transistor. Polycrystalline silicon layer 16 is then doped to a sheet resistance of about 15 ohms per square with an N type dopant such as phosphorous.

Polycrystalline silicon layer 16 is then covered with a photoresist layer 18 which is then patterned in a conven­tional manner, thereby exposing portions 16a to 16d of polycrystalline silicon layer 16. Exposed portions 16a to 16d of polycrystalline silicon layer 16 are then removed, thereby leaving polycrystalline silicon gates 16e and 16f and a field limiting ring 16g as illustrated in Figure 2. (In other embodiments, gates 16e and 16f and field limiting ring 16g are of a material other than polycrystalline silicon, e.g. a metal such as refractory metal or a silicide.) Al­though polycrystalline silicon gates 16e and 16f appear as separate structures, in one embodiment they are a single contiguous polycrystalline silicon gate 17 joined outside the cross section of Figure 2. (In another embodiment, gates 16e and 16f are not a single contiguous region. However, in such an embodiment, gates 16e and 16f can be electrically connected together with a subsequently formed conductive layer.) As described below, field limiting ring 16g surrounds the transistor and defines a subsequently formed EQR which surrounds ring 16g. In this way, photoresist layer 18 serves as a first mask for defining the gate of the DMOS transistor, a field limiting ring, and an EQR surrounding the transistor.

Referring to Figure 3, the remaining portions of photoresist layer 18 are removed and polycrystalline silicon gate 17 and ring 16g are covered with a silicon dioxide layer 20. In one embodiment, silicon dioxide layer 20 is thermally grown to a thickness of about 500 nm.

The wafer is then covered with a photoresist layer 22 which is then patterned in a known manner, thereby forming window regions 22a, 22b, and 22c in photoresist layer 22. As will hereinafter become apparent, window region 22a defines an electrical contact to polycrystalline silicon gate 17 while window regions 22b and 22c define a P+ deep body region of the DMOS transistor.

Referring to Figure 4, portion 20a of silicon dioxide layer 20 within window region 22a is removed and the portions of silicon nitride layer 14 within window regions 22b and 22c are removed. In one embodiment, portion 20a is removed by placing the wafer in a buffered HF solution. Of importance, photoresist layer 22 protects the underlying portions of silicon dioxide layer 20. However, the portion of silicon dioxide layer 20 within window region 22a is unprotected and therefore is removed by the HF solution. In addition, an oxidised nitride skin, typically formed to a thickness of 5 to 10 nm during the preceding process steps within window regions 22b and 22c, is also removed by the HF solution. Thereafter, the wafer is removed from the buffered HF solution and placed, for example, in a phosphoric acid solution which removes the portions of silicon nitride layer 14 within window regions 22b and 22c, thereby forming window regions 14a and 14b. The wafer is then removed from the phosphoric acid solution and photoresist layer 22 is removed. The resulting structure is illustrated in Figure 5.

In other embodiments of the invention, instead of using an HF solution and phosphoric acid to etch silicon dioxide layer 20 and silicon nitride layer 14, respectively, plasma etching, reactive ion etching, or ion milling are used.

Referring to Figure 6, P+ deep body regions 24 are formed in silicon substrate 10, e.g., by implanting P type ions in the portion of substrate 10 within window regions 14a and 14b. In one embodiment, boron ions are implanted with a dosage of about 2 x 10¹⁴/cm² and an implant energy of about 40 Kev. (Of importance, the implant energy and dose are selected to guarantee that ions are not implanted into the portions of the wafer covered by both silicon dioxide layer 12 and silicon nitride layer 14.) Although during this process some ions are also implanted into poly­crystalline silicon gate 17 where silicon dioxide layer 20 has been removed, because of the high N type dopant concentration in polycrystalline silicon gate 17, the gate conductivity is not significantly changed during this process. It is also noted that the portions of silicon dioxide layer 12 within window regions 14a and 14b prevent unwanted introduction of impurities into the semiconductor material within window regions 14a and 14b and prevent unwanted loss of boron during subsequent high temperature processing steps.

The wafer is then subjected to a drive-in process. During this drive-in process, a thin silicon dioxide layer 20b (typically about 50 nm thick) is formed on poly­crystalline silicon 16e. In addition, portions 25 of silicon dioxide layer 12 above P+ deep body region 24 also increase in thickness. Of importance, during the diffusion of P+ regions 24, silicon dioxide layers 20b and 25 prevent impurities from diffusing out of gate 17 and into substrate 10.

Figure 6a illustrates in plan view a portion of the transistor at this point in the manufacturing process. Region 24 is not shown in Figure 6a to simplify the illustration.

The portions of silicon nitride layer 14 not covered by gate 17 or polycrystalline silicon ring 16g are then removed. In one embodiment this is done by placing the wafer in a phosphoric acid solution. However, the exposed portions of silicon nitride layer 14 can be removed in other ways as well, e.g., reactive ion etching or plasma etching. Silicon dioxide layer 20b and the exposed portions of silicon dioxide layer 12 are then removed by placing the wafer in a buffered HF solution. In other embodiments, layer 20b and the exposed portions of silicon dioxide layer 12 are removed with plasma etching or reactive ion etching. Of importance because of the thickness of silicon dioxide layers 20 (500 nm) and 25 (220 to 230 nm), silicon dioxide layers 20 to 25 remain largely intact during this etching process. In addition, the portions of silicon nitride layer 14 and silicon dioxide layer 12 underneath gate 17 and polycrystalline silicon ring 16g also remain. Of importance, up to this point in the process, only two photolithographic masks have been applied to the wafer.

Referring to Figure 7, P type body region 26 is formed, e.g. by ion implantation. In one embodiment, boron ions are implanted into substrate 10 with a dosage of about 5 x 10¹³/cm² and an implant energy of approximately 50 Kev. Of importance, the lateral extent of P type body region 26 is defined by edges 27 of gate 17. Therefore, no additional photolithographic masks or alignment steps are required during formation of body region 26. The wafer is then subjected to a drive-in process.

A P type region 26a is formed concurrently with body region 26. However, as described below, P region 26a does not serve as a body region, but rather as part of an EQR.

Thereafter, N+ source regions 28 are formed, e.g. by ion implantation. In one embodiment, arsenic ions are implanted into substrate 10 with a dosage of about 5 x 10¹⁵/cm² and an implant energy of about 40 Kev. The lateral extent of N+ source region 28 is also defined by edges 27 of gate 17 and the edge of silicon dioxide 25. The wafer is then subjected to another drive-in process. Thereafter, silicon dioxide 25 and any silicon dioxide formed above source regions 28 or within window 22a during the same drive-in process is removed in a blanket etching process, e.g. by placing the wafer in a buffered HF solution.

An N+ region 28a is formed concurrently with N+ source region 28. However, as described below, N+ region 28a does not serve as a source region but rather as part of the EQR.

Referring to Figure 8, the wafer is covered with a conductive layer 30. Conductive layer 30 is typically a metal layer such as aluminum or an alloy of aluminum. In other embodiments, conductive layer 30 is another metal. Conductive layer 30 serves as a source/body lead for the transistor as well as a gate lead. The wafer is then covered with a photoresist layer 32 which is patterned in a conventional manner, thereby exposing portions of conductive layer 30. The exposed portions of conductive layer 30 are then removed as illustrated in Figure 8.

Photoresist layer 32 is then removed and the wafer is covered with a passivating layer of insulation such as plasma deposited silicon nitride layer 34. In other embodiments, layer 34 is silicon dioxide. Portions of conductive layer 30 extend to an area of the wafer where bonding pads are formed (not shown). Passivating layer 34 is patterned by apply a fourth photolithographic mask to the wafer. The fourth photolithographic mask includes window regions exposing the portions of the passivating layer 34 above the bonding pads. The exposed portions of passivating layer 34 are then removed, thereby exposing the portions of conductive layer 30 at the bonding pad.

After passivation layer 34 is patterned, a conductive layer 36 (also typically a metal such as aluminum) is formed on the bottom of the wafer (Figure 9). Conductive layer 36 serves as a drain contact, and is typically not patterned.

A portion 30a of conductive layer 30 (the EQR) is then shorted to substrate 10 and P region 26a. In one embodi­ment, this is accomplished by applying a high voltage to portion 30a of conductive layer 30 relative to substrate 10. Portion 30a is typically shorted to substrate 10 and P region 26a during wafer sort. (Wafer sort is the electrical testing of devices formed in the wafer before the wafer is cut into separate devices.) As mentioned above, the EQR is formed at the periphery of the transistor. As is known in the art, EQRs enhance reliability of a transistor. EQRs are described in "Surface Breakdown In Silicon Planar Diodes Equipped With Field Plate" by Conti et al., published in Solid State Electronics in 1972, Vol. 15.

As mentioned above, polycrystalline ring 16g serves as a field limiting ring. Ring 16g is typically electrically insulated from the other structures formed on the wafer. In an alternative embodiment, a plurality of polycrystalline silicon field limiting rings surround the transistor to further enhance transistor reliability.

It will be appreciated that a process for manufacturing a DMOS transistor requiring only four photolithographic masks has been described in detail. The process is simple and minimizes the requirement of alignment tolerances. Accordingly, a transistor manufactured with the present process can be manufactured on a smaller surface area than prior art transistors. For example, in a typical prior art transistor, the gate to gate distance (e.g. from poly­crystalline silicon 16e to 16f) is about 25 to 30 µm. By using the process of the present invention, this distance can be reduced to about 22 to 28 µm. This represents a savings of about 20% of the surface area of a transistor using a plurality of square cells.

In an alternative embodiment, the above-described process is used to form an insulated gate bipolar transistor. In such an embodiment, source 28, body 26 and deep body region 24 are formed in an N type epitaxial layer which in turn is formed on a P+ substrate. The operation of insulated gate transistor is described in detail in "The Insulated Gate Transistor: A New Three-Terminal MOS-­Controlled Bipolar Device," by Baliga et al., published in IEEE Transactions on Electron Devices, Vol. ED-31., No. 6, in June 1984.

While the invention has been described with reference to specific details, those skilled in the art will recognise that changes can be made in form and detail without depart­ing from the spirit and scope of the invention. For example, both N and P channel transistors can be formed using this process. In addition, the transistor can be formed either in a substrate or in an epitaxial layer formed on a substrate. Accordingly, all such modifications come within the present invention.