The present invention relates generally to semiconductor devices, more particularly to dual-damascene processing in the fabrication of semiconductor devices, and still more particularly to hardmask materials for dual-damascene processing.
As integrated circuit density has increased, the former practice of using aluminum conductors for interconnections within integrated circuit devices have become a significant limiting factor. This is due, in large part, to aluminum's relatively poor performance as a conductor at the very small line widths associated with modern high-density integrated circuits. Similarly sized conductors formed of copper (Cu), which exhibits much lower resistivity than aluminum, are capable of performing reliably much higher current densities and are better suited to the newer fine-pitch design rules.
The use of copper interconnects, however, has necessitated new processing techniques. Direct patterning of copper conductors is generally impractical using modern processing techniques. Accordingly, copper conductors are typically formed using a dual-damascene process. In a typical dual-damascene process, trenches and vias are photolithographically created in a dielectric layer. Copper is then deposited into the trenches and vias, filling them. Any excess copper is then removed via a conventional planarization technique such as CMP (chemical-mechanical polishing).
In one dual-damascene processing scheme, tantalum nitride (TaN) is used as a hardmask (HM), which also serves as a line template. In this process, the etch scheme for defining trench patterns (Mx) utilizes the TaN HM. Critical dimension (CD) control for the lithographic process used to create these trenches (Mx Metallization level ‘x’) and vias (Vx Via level ‘x’) is heavily dependent on the thickness of the TaN hardmask. The patterns defined in the Mx lithography are etch-transferred to the TaN hardmask. This is followed by via-lithography and a subsequent dual-damascene etch. During the dual-damascene etch, the TaN hardmask is intended to preserve the etch patterns. However, for the etching processes necessitated by hybrid dielectric or inorganic dielectric materials, the TaN is eroded by the etch process, leading to loss of critical dimension (CD) control or “CD blowout”. CD control can be regained by increasing the thickness of the TaN hard-mask layer, but this increased thickness has the undesirable side-effect of decreasing the transparency of the TaN layer to a point where optical alignment of lithographic processes to underlying alignment features becomes difficult or impossible.
Where fine-line CD control is required, precise control of lithographic process alignment is also required. This presents two competing sets of requirements on the thickness of the TaN hardmask layer. Whereas precise photolithographic process alignment requires levels of optical transparency that can only be achieved with a thinner TaN hardmask layer, CD control considerations require a thicker TaN hardmask layer. As device geometry becomes smaller, the conflict between these competing requirements becomes greater, severely limiting the usefulness of TaN as a hardmask.
The present inventive technique solves the problem of TaN hardmask opacity with increasing thickness by oxidizing the TaN layer. Oxidation of the TaN hardmask produces two desirable results. First, it increases the thickness of the hardmask to two to four times its original thickness. This permits better CD control, especially when etching hybrid dielectric or inorganic dielectric materials. Second, it increases the transparency of the TaN hardmask, which facilitates precise optical alignment of the lithographic processes, further enhancing CD control. The transparency of oxidized TaN hardmask over TaN is improved by a factor of greater than ten times (as measured in terms of a wafer quality number). In combination, these two results produce a hardmask that is capable of simultaneously satisfying the competing requirements of a thicker hardmask and greater hardmask transparency.
According to the invention, two distinct process paths can be employed to create the oxidized tantalum nitride hardmask. In a first process, the tantalum nitride layer is subjected to an oxidation process in its entirety, converting the entire tantalum nitride (TaN) layer to tantalumoxy-nitride (TaOxNx). After oxidation, the oxidized tantalum nitride layer is lithographically etched to form trench openings therein, followed by normal dual-damascene via and trench formation. This process is referred to hereinafter as an “oxidize, then etch” methodology.
Alternatively, the tantalum nitride layer can be lithographically etched to form trench openings therein, prior to oxidation. After etching, the etched tantalum nitride layer is subjected to the oxidation process to form a patterned oxidized tantalum nitride layer. This process is referred to hereinafter as an “etch, then oxidize” methodology.
In dual-damascene processing, the tantalum nitride layer is a top-level hardmask layer on a “stack” comprising a base dielectric layer, a cap layer overlying the base dielectric, a dielectric layer overlying the cap layer, first and second hardmask layers (HM1 and HM2) overlying the dielectric layer, and the top-level TaN hardmask overlying the HM1 and HM2 layers. The dielectric layer can be a single layer organic or inorganic dielectric, or can be a multi-level hybrid dielectric. The base dielectric includes circuit elements (typically active silicon or conductors) to which electrical contact is to be made via the dual damascene process. The circuit elements are typically planarized with the base dielectric layer to produce a substantially flush surface.
According to an aspect of the invention, the oxidation process can be a combined thermal and plasma oxidation process. The oxidation environment is preferably provided in a chamber with a N2O flow rate between 500 and 5000 sccm (standard cubic centimeters per minute) at a pressure between 1 and 10 Torr. Preferably the oxidation process employs a substrate temperature of between 250 degrees C. and 400 degrees C. with a plasma power of between 250 and 1000 Watts.
According to one embodiment of the invention, the method comprises providing a semiconductor wafer having a base dielectric layer, said base dielectric layer having circuit elements embedded therein and planarized flush with the surface thereof to which a subsequent electrical connection is to be made. A cap layer is formed over the base dielectric layer and circuit elements. A dielectric layer is formed over the cap layer. This dielectric layer can be a single layer organic or inorganic dielectric or a multi-level hybrid dielectric. Hardmask layers are formed over the dielectric layer and a tantalum nitride hardmask layer is formed over the hardmask layers. The tantalum nitride layer is lithographically patterned and is then subjected to an oxidation process as described above.
According to another embodiment of the invention, the method comprises providing a semiconductor wafer having a base dielectric layer, said base dielectric layer having circuit elements embedded therein and planarized flush with the surface thereof to which a subsequent electrical connection is to be made. A cap layer is formed over the base dielectric layer and circuit elements. A dielectric layer is formed over the cap layer. This dielectric layer can be a single layer organic or inorganic dielectric or a multi-level hybrid dielectric. Hardmask layers are formed over the dielectric layer and a tantalum nitride hardmask layer is formed over the hardmask layers. The tantalum nitride layer is oxidized to form oxidized tantalum nitride. The oxidized tantalum nitride layer is then lithographically patterned.
These two embodiments produce substantially equivalent resulting structures which can be further processed via normal dual-damascene methodology to complete the formation of trench and via openings, followed by deposition of the conductor material (preferably copper).
These and further features of the present invention will be apparent with reference to the following description and drawing, wherein:
The present inventive technique employs oxidized tantalum nitride (TaN) as an improved hardmask for use in dual-damascene processing. By oxidizing a tantalum nitride hardmask (to produce TaOxNx tantalum oxy-nitride), the thickness of the hardmask is increased by a factor of two to four times over unoxidized TaN, while simultaneously increasing the transparency of the hardmask by a factor of greater than ten times. The thicker TaOxNx hardmask provides better critical dimension (CD) control against the etching processes used to etch hybrid or inorganic dielectrics. The increased transparency of the TaOxNx hardmask permits accurate optical alignment of lithographic processes to underlying alignment features (typically formed in the base dielectric layer well below the hardmask layer).
The TaN hardmask is oxidized by means of the combination of thermal oxidation and N2O plasma at low pressure. Preferably, a N2O flow rate between 1000 and 2000 sccm at a chamber pressure between 1 Torr and 6 Torr provides the oxidation ambient environment. A plasma power between 250W (watts) and 1000W in combination with a substrate temperature between 250° C. and 400° C. is preferably employed as the oxidation process.
Overlying the base dielectric 102 and circuit elements 114 and 116 is a cap layer 104. The cap layer 104 acts as a hermetic seal to protect the underlying structures (102, 114, 116) against damage and/or contamination (e.g., by moisture) in subsequent processing steps. Typically the cap layer 104 is SiCH, SiCOH, SiN, SiCNH, etc.
Overlying the cap layer 104 is a dielectric layer 106. The dielectric layer 106 can be a single-level organic or inorganic dielectric, or it can be a hybrid dielectric stack. In dual-damascene processes, it is common to use a hybrid dielectric stack to facilitate and control formation of trench and via openings.
Overlying the dielectric layer 106 is a first hardmask layer 108 (HM1). This HM1 layer 108 acts as a hermetic seal for the dielectric layer 106 and as a CMP (chem-mech polish) stop. It can be SiCOH, SiCNH, SiCH, SiN or other suitable material.
Overlying the HM1 layer 108 is a second hardmask layer 110 (HM2). This HM2 layer acts as a plasma rework barrier, and can be SiCOH, SiCNH, SiCH, SiN, SiO2 or other suitable material.
Overlying the HM2 layer 110 is a tantalum nitride (TaN) top hardmask layer 112, which preserves lithographic patterning during subsequent trench etching by RIE (reactive ion etch).
The aforementioned oxidation of the TaN hardmask can be accomplished by two different process paths.
1) Etch, then oxidize (Post Mx RIE oxidation); or
2) Oxidize, then etch (Pre Mx RIE oxidation).
Typically, the HM1 layer is 30-100 nm (nanometers) in thickness and is formed of a suitable hermetic-seal/polish-stop material as described hereinabove with respect to
In a next step 206, a TaN top level hardmask is disposed over the HM2 layer, typically to a thickness of 5-25 nm.
At this point, the process flow diagram splits to show two separate possible process flows. A leftmost process flow (as illustrated) comprising process steps 208A, 210A and 212A illustrates the “etch, then oxidize” methodology. A rightmost process flow (as illustrated) comprising process steps 208B, 210B and 212B illustrated the “oxidize, then etch) methodology. The two process flows re-converge onto a common process flow at a process step 214.
Directing attention to the “etch, then oxidize” process flow (the leftmost process path in
Now directing attention to the “oxidize then etch” process flow (the rightmost process path in
The “etch, then oxidize” process path ending in process step 212A and the “oxidize, then etch” process path ending in process step 212B produce essentially equivalent structures. At this point, the two process paths reconverge at a process step 214, wherein dual-damascene V‘x’ (via level ‘x’) lithography is performed, followed by a conventional dual-damascene RIE step 216 to form the vias (and complete the trenches).
Those of ordinary skill in the art will immediately understand that with the exception of the TaN top-level hardmask processes, the dual-damascene processes described herein are conventional dual-damascene processing steps, and that the present inventive technique can be adapted to any suitable dual-damascene process flow that employs a TaN top-level hardmask.
In
In
At this point in processing, the two methodologies (shown in
Although the invention has been shown and described with respect to a certain preferred embodiment or embodiments, certain equivalent alterations and modifications will occur to others skilled in the art upon the reading and understanding of this specification and the annexed drawings. In particular regard to the various functions performed by the above described components (assemblies, devices, circuits, etc.) the terms (including a reference to a“means”) used to describe such components are intended to correspond, unless otherwise indicated, to any component which performs the specified function of the described component (i.e., that is functionally equivalent), even though not structurally equivalent to the disclosed structure which performs the function in the herein illustrated exemplary embodiments of the invention. In addition, while a particular feature of the invention may have been disclosed with respect to only one of several embodiments, such feature may be combined with one or more features of the other embodiments as may be desired and advantageous for any given or particular application.