The essentials of single-wafer wet processing: selectivity and partnering
04/01/2006
Chipmakers are migrating to single-wafer tools, driven by the need for highly reliable processes that can meet the stringent performance and throughput requirements of state-of-the-art fabs. The strategy is not just a function of the toolset, but rather of the overall process used. This means that the right chemistry, or combination of chemistries, is necessary to avoid the materials and yield losses that increasingly accompany batch processing, particularly wet-bench tools.
One thing is clear: the keyword for future chemistry implementation is selectivity. In front-end-of-line (FEOL) processing, impending challenges will likely require a radical shift in how cleans are done. This could mean a re-evaluation of solvent use in FEOL-something traditionally avoided because of concerns about organic contamination and other issues. We could also see semi-aqueous solvents used in back-end-of-line (BEOL) being introduced into FEOL. At present, however, the overall trend is toward dilute chemistries, and interesting challenges will certainly arise when we get to 32nm and below. Similarly, alternatives may be needed for film removals traditionally done with sulfuric peroxide or ammonium peroxide mixtures.
Chemistries used in past years may not work in the future, at least not in the traditional sense; they will likely become ever more dilute in the short term, though long-term solutions are not yet clear. Chemistries will need to have higher and higher selectivity while still effectively performing cleans.
Dielectric material integration, for example, has had major issues in both FEOL and BEOL. Take the example of BEOL cleans. The effectiveness of a post-etch residue (polymer) clean is directly related to the effectiveness of the previous process steps. Everything from the choice of gases used in etch and plasma-ash steps to the duration and power settings of the plasma strip have a direct influence on the wet clean. For ultra-low-k materials, changing plasma-ash characteristics has been shown to significantly modulate cleaning efficiency and the amount of damage observed after the wet step.
Clearly, identifying a chemistry strategy is no small task and will only grow more complex with coming device generations. End-users want complementary solutions, and thus will go to their equipment vendors, ask for evaluations of different chemicals from different suppliers, and then assess the results to determine which combinations work best on different tools. They want to know that an equipment maker has multiple partnerships with chemistry vendors because it means the supplier has a vested interest in refining its chemical formations for optimal performance on the end-user’s equipment.
The biggest challenge, of course, is anticipating the end-users’ actual needs. At 45nm and below, both FEOL and BEOL will see the introduction of new materials that will require the development of new innovative cleans solutions. That’s where the third part of the relationship equation comes in. Close equipment and chemistry supplier partnerships are not enough-end-user involvement is paramount.
The IDMs need to be willing to discuss their integration schemes and materials choices as much as possible and to collaborate more closely with the vendors. Because most supplier companies cannot generate their own test-wafer structures, end-user interaction will be even more important, especially to help ensure that a fully developed solution will be available when manufacturing demand ramps up. Integration schemes are driven by the end-user and are critical to understand, particularly for wet-clean development. Whether via-first, trench-first, or hard masks, etc., are used, all influence the effectiveness of the clean step.
Whether wet cleans will become truly “value added” is the subject for some debate. However, they will increasingly become more enabling as technology nodes shrink. Furthermore, it will become increasingly apparent that cleans steps can no longer be viewed in isolation. Thus, a modular view of processes might be needed to account for the impact of upstream and downstream process steps.
Challenges being faced in cleans processing will help to determine whether Moore’s Law remains a viable standard for measuring technology progress. The development of new selective cleaning processes will require partnering between chipmakers, equipment companies, and chemical suppliers.
 |
Contact Leo Archer, VP, emerging technologies worldwide, at The SEZ Group, 4829 South 38th St., Phoenix, AZ 85040; ph 602/437-5050, e-mail [email protected].