Issue



Taking the wet-developable route to applying BARC in implant layers


06/01/2004







Traditionally, bottom antireflective coatings (BARC) have mainly been used in critical layers for gates and contacts, but the application of BARC in implant layers also has become more desirable as device feature sizes shrink. Using BARC in implant layers promises greater tolerance for reflective notching and CD variations when wafer topography becomes smaller, but the feasibility of traditional dry-etch coatings is questionable due to process complexity, defectivity, and the potential for substrate damage. A wet-developable BARC tailored for implant layers promises to overcome these barriers.

A typical 130nm logic process flow may contain as many as 14 implant masking steps in which most of the layers need CD and overlay controls due to the budgetary limitation of line-edge placement (LEP) error. The table shows the implant layer types typical for today's CMOS technology, along with the major challenges.

Click here to enlarge image

In addition, several new challenges arise from a tighter LEP budget, such as requirements for more vertical resist profiles [1], greater substrate stability in DUV resist, less implantation shadowing [1, 2], and reduced reflective notching of resist patterns. In particular, reflective notching has been proven to play an important role in determining the device-leakage failure rate. For example, the experimental results of a CMOS device with 0.18µm technology (Fig. 1) clearly indicate that the device-leakage failure rate is strongly correlated to the resist linewidth of the source/drain (S/D) implant layer.


Figure 1. Device leakage failure vs. S/D feature size.
Click here to enlarge image

Because the severity of reflective notching is inversely proportional to the resist linewidth, it is not surprising to get a leakage failure curve like that in Fig. 1, although the CD and overlay are still in control. Reflective notching for implant resist patterns can be reduced or eliminated with a wet-developable BARC. While dry-etch organic BARCs are used extensively for critical lithography layers requiring plasma pattern transfer, they typically cannot be employed for implant layers for the following reasons:

  • The dry-etch process to clear the BARC is complex. Controlling the etch endpoint is very difficult, which makes the implant and diffusion process also difficult to control.
  • Adjusting implant energies to penetrate appropriately through the BARC layer is also difficult. BARC thickness may vary in relation to local topography.
  • Ions used for the dry-etch step are highly energetic and may participate in the implant and diffusion processes to decrease the efficiency of silicon performance.

Like any other BARC currently on the market, a wet-developable BARC is an organic liquid coating material used in conjunction with a photoresist during the photolithography step in semiconductor manufacturing. Unlike a dry-etch BARC, the wet DUV BARC is soluble in the developer so it can be removed during the resist development step. Figure 2 describes the differences between dry- and wet-patterning BARCs.


Figure 2. Comparison of a) dry-patterning and b) wet-patterning processes. PEB = post-exposure bake.
Click here to enlarge image

Semiconductor processing also places other requirements on a wet-developable BARC, such as a resolution ≤0.18µm, sidewalls as straight as those produced by a dry-etch ARC, spin-bowl compatibility, safe solvent systems, room-temperature stability, low wafer defects, broad bake process window, a development time ≤60 sec (with resist), 200mm and 300mm process conditions, the ability to be removed with industry-standard strippers, and broad resist compatibility.

Wet-developable BARCs were first introduced by Brewer Science in the early 1980s [4–6]. Traditionally, these coatings have been based on a polyamic acid platform in which the solubility of the BARC in the aqueous developer is controlled by the BARC bake process (cure temperature). During this stage, the developer-soluble polyamic acid is converted into a polyimide, which is not soluble in the base developer. For this type of BARC, the following factors contribute to the dissolution of the material in the developer: transport of the developer to the surface; adsorption of the developer to the surface; the development reaction; desorption of the reacted product from the surface; and transport of the product away from the surface [7].


Figure 3. Illustration of bake latitude for a wet-developable BARC.
Click here to enlarge image

The dissolution rate of the BARC then is dependent on the degree of BARC cure. A bake latitude is defined by the upper and lower temperatures that induce line collapse and scumming between lines or in open spaces (Fig. 3). It then follows that the best resolution is found in this bake "window" where the BARC undercutting and footing are controlled.

Reflective hole improvement

In the 130nm technology node, process engineers are often challenged by the task of defining a notching-free S/D implant layer. The difficulty arises mainly due to the reflective light from the sidewall of shallow-trench isolation (STI). Usually, after generating STI and then filling the trench with SiO2, chemical mechanical polishing (CMP) will be applied to the wafer to remove the main topography. However, because SiO2 is a transparent material, the reflective light still plays an important role in damaging resist patterns.


Figure 4. a) Top-down SEM photo shows not only four resist holes, but also the rough resist edge introduced by the reflective light from STI was observed. b) Applying a developable BARC provided a solution.
Click here to enlarge image

In order to understand the detailed mechanism that caused the reflective notching, a Solid-C simulation was performed on a special pattern of S/D implant layer. The pattern consists of a rectangular STI opening filled by SiO2 and then covered by a rectangular resist. The simulation predicts that the reflective light from the STI sidewall is mainly focused at the four corners of a rectangular resist pattern, and therefore four reflective holes on the resist pattern will be observed after the exposure and develop steps. The experiments proved the predictions from the Solid-C simulation. In Figure 4a, due to the reflective light from the STI, four resist holes were observed, accompanied by a rough edge near the corners. Fortunately, after applying the wet-developable BARC — called IMBARC by Brewer Science — all resist damage to the nearby four corners was removed (Fig. 4b).

Process window improvement

It is well known that reflective notching can reduce the process window (PW). We used a standard S/D implant layer as an example [1]. This layer was processed by using 0.65µm Shipley UV5 resist with illumination conditions NA = 0.57 and σ = 0.45.


Figure 5. A resist pattern for an S/D implant layer in 0.15µm technology was used in generating a process window. Due to the topography from STI and poly gate, a) severe notching can be seen. After using BARC, b) reflective notching is removed.
Click here to enlarge image

If the resist features without the notching were measured, the PW showed a rectangular shape with 0.6µm DOF and 40.45% exposure latitude. However, the effective PW was severely damaged by reflective notching (Fig. 5a). Although the PW can be improved by using more incoherent illumination [3], this compromise can sometimes fail to provide good pattern definition, particularly in cases where the resist hole and line coexist in layout. Meanwhile, the use of wet-developable BARC provides possible removal of all reflective notching without compromising the illumination conditions (Fig. 5b).


Figure 6. A resist line sitting on the trench oxide after CMP was lifted by gradually increasing the exposure dose from a) to c) due to the reflective light from STI. The tilted lines in the substrate represent the profile of STI.
Click here to enlarge image

In an S/D implant layer, reflective light from the STI profile may also introduce resist lifting. The severity of resist lifting is dependent upon the exposure dose (Fig. 6). To avoid resist lifting, a lower exposure dose may be needed. However, getting nominal CD size then becomes impossible without resist lifting. Meanwhile, the IMBARC application eliminated all resist lifting (Fig. 7). The corresponding PW, without resist lifting, was increased significantly. The process can be easily defined either at the nominal CD or at an even lower CD, but the IMBARC process needs to be optimized further because the SEM cross-section showed some slight undercutting.

Wet-developable BARC challenges

Using a wet-developable BARC requires selecting an appropriate film thickness and optimizing both the bake and develop processes. The BARC thickness not only affects the reflectivity control for a given substrate and film stack, but it can also shift the optimum bake window.


Figure 7. An IMBARC application totally removed the resist lifting, as shown, but further optimization of the process is needed because the cross-section indicates some undercutting occurred.
Click here to enlarge image

The dissolution rate — and hence the degree of undercutting or footing — that occurs with the wet-developable BARC is most easily modified through the bake process. In the case of IMBARC, the bake temperature has the greatest impact on develop rate with bake time a secondary parameter, as seen in Fig. 3. Once the optimum bake temperature is determined, the bake time can be used to fine-tune the desired profile.

The removal of resist and BARC between fine features or over deep topography is a distinct challenge for this generation of wet-developable, organic antireflective coatings. The clearing of BARC and resists out of a contact hole or between poly lines, or clearing resist patterned over trench topography, must be evaluated for critical areas. Smaller feature sizes limit developer transport to localized regions on the substrate. Depending on the application, developer and rinse processes may be used to improve removal of the residue. Reduced bake temperature (which enhances the develop rate) could also be used, but there is a trade-off with potential degradation of the resist/BARC lithography profile.

Conclusion

Using a wet-developable BARC in an implant layer application is a new approach for achieving tight CD control without damaging the substrate. It has been demonstrated that wet-developable BARC can eliminate resist notching/holes and improve the PW for exposure latitude and depth of focus. Additionally, these bottom antireflective coatings completely eliminated resist lifting and significantly improved the PW.

The main challenge for applying the current wet-developable BARC in an implant layer is the residue that remains after resist/BARC development, which is caused by incomplete development. This challenge is particularly apparent when the application has a relatively high topography (>150nm) and the space is small (<100nm). While a somewhat effective remedy is achieved by applying process variations — such as double rinse, double puddle development, etc. — work is progressing to solve this problem by adjusting the BARC development rate through chemical modification in BARC systems.

References

  1. Y. Gu, J. Sturtevant, "Implant layers: 'Non-critical' Lithography?" Microlithography World, pp. 18–20, 22, Nov. 2002.
  2. Y. Gu, D. Chou, S.Y. Lee, W.R. Roche, J. Sturtevant, "Reduction of Implantation Shadowing Effect by Dual-wavelength Exposure Photo Process," Proc. SPIE, Vol. 5039, pp. 1312–1318, 2003.
  3. Y. Gu, D. Chou, C. Lom, J. Sturtevant, "Implant Layer Lithography: Who Says It's 'Non-critical?'" ARC Symposium, Oct. 2003.
  4. US Patent No. 4,910,122, "Antireflective Coating," J. Arnold, T.L. Brewer, S. Punyakumleard, March 20, 1990.
  5. US Patent No. 5,057,399, "Method for Making Polyimide Microlithographic Compositions Soluble in Alkaline Media," T. Flaim, J. Lamb, G. Barnes, T.L. Brewer, Oct. 15, 1991.
  6. US Patent No. 5,688,987 and US Patent No. 5,892,096, "Non-subliming Mid-UV Dyes and Ultra-thin Organic Arcs Having Differential Solubility," J. Meador, X. Shao, V. Krishnamurthy, E. Murphy, T. Flaim, et al., Nov. 18, 1997, and April 6, 1999.
  7. P. Williams, X. Shao, "Process Consideration for Organic Bottom Antireflective Coating Optimization for Front- and Back-End of Line Integration," Semicon China, pp. 229–238, 2003.

Xie Shao is an R&D division manager at Brewer Science Inc., 2401 Brewer Dr., Rolla, MO 65401; ph 573/364-0300, fax 573/364-0650, e-mail [email protected].
Alice Guerrero is a product engineer for DUV wet-developable BARC in the ARC global technical support operation at Brewer Science.
Yiming Gu is R&D photolithography manager of the R&D Center at Integrated Device Technology Inc., 3131 NE Brookwood Pkwy., Hillsboro, OR 97124.