A novel resist and post-etch residue removal process using ozonated chemistry
12/01/1998
A novel resist and post-etch residue removal process using ozonated chemistry
S. De Gendt, J. Wauters, M. Heyns, IMEC, Leuven, Belgium
A novel, environmentally friendly process has successfully removed photoresist and organic post-etch residues from silicon surfaces. The moist ozone gas phase process described here greatly increases organic removal efficiency and is expected to replace most sulfuric-based process steps in IC production.
Integrated circuit (IC) fabrication uses photoresist masking to transfer an image onto the desired circuit layer. As a result of these treatments, such as high energy/high dose implantation or reactive ion etching, the photoresist may be hardened, cross-linked, or even converted into polymer (organometallic) residues. Although critical for IC manufacturing, these masking layers must be removed afterwards.
Current organic photoresist stripping processes use combinations of dry and wet strip treatments. Dry strip techniques consist predominantly of reactive plasma ashing steps and suffer from drawbacks such as plasma damage, incomplete removal, and resist popping. A wet strip typically completes (or replaces) the dry stripping to avoid leaving residues behind. Wet stripping techniques either rely on organic solvent treatments or use aggressive chemistries such as H2SO4/H2O2 (sulfuric acid-peroxide mixture, or SPM) or H2SO4/O3 (sulfuric acid-ozone mixture, or SOM). The semiconductor industry uses large quantities of these wet chemicals, adding to the process cost and causing significant environmental problems. In addition, the chemicals require large quantities of DI-water for wafer rinsing. Sulfuric-based mixtures are especially notorious in this respect since they are very difficult to rinse, whereas solvent strippers tend to be very toxic.
Use of ozonated DI-water for removal of resist and post-etch residues is an alternative that reduces cost-of-ownership (less chemicals supply) and environmental impact. A new ozone technology, based on boundary-layer-controlled processing (moist ozone gas phase), was presented at a recent VLSI Symposium [1].
Ozone behavior in aqueous media
Ozone is frequently applied in wastewater treatment, and there have been extensive studies on its complex chemistry [2, 3]. Transfer of gaseous ozone (O3) into aqueous solution is limited by its solubility, thus leading to O3 loss through purging (Tech News, p. 28). The oxidative action of O3 relies on either direct or advanced oxidation (Fig. 1). Direct oxidation of solutes ("X") involves molecular O3 and is selective to carbon unsaturated bonds. In addition, O3 is likely to decompose via alternative reaction pathways that involve water (e.g. OH-), forming primary radicals (*OH). These radicals initiate chain formation of more radicals, react with the solutes X (advanced oxidation), or are removed from the reaction (scavenged). Advanced oxidation is more reactive and less sensitive compared with direct oxidation. This high reactivity, however, severely reduces the lifetime of *OH radicals.
Appropriate addition of selected chemicals allows interference in the complex ozone/water equilibria. Acidification inhibits the O3 decomposition into *OH radicals, thus enhancing the molecular O3 solubility up to 20%. Addition of *OH radical enhancers (e.g. H2O2) results in fast (up to 90%) depletion of molecular O3, whereas addition of *OH radical scavengers, such as acetic acid, stabilizes the O3/water mixture. The addition of *OH radical scavengers will improve resist stripping efficiency.
 |
Figure 1. Scheme of aqueous ozone reactions.
Chemical enhancement
Researchers at IMEC examined the effect of acetic acid (HAc) addition for various resist pre-treatments using wafers immersed in DI water at ambient temperature. Results showed the addition of 10-3 M (M = molar concentration, g-mol/liter) acetic acid led to as much as a 70% increase in process efficiency over traditional immersion-based ozone processes (Fig. 2).
Feasibility of ozone processing
A successful resist strip procedure must be able to remove organic contamination fully, within relatively short processing times. A photoresist film consists of long, repeating chains of a monomer, primarily composed of carbon and hydrogen. The reaction with ozone can be simplified as a complete oxidation of the carbon chain, leaving nothing but carbon dioxide and water.
 |
The photoresist is located in a confined layer near the wafer surface. Ozone and/or other reactive species have to migrate from the bulk of the solution, and the removal is thus transport-controlled.
Also, ozone-based resist stripping at ambient temperatures is only feasible in the presence of water, indicating the importance of the OH-radical-based (advanced) oxidation mechanism. The O3 solubility in DI water, however, and hence the concentration of oxidants (e.g. molecular O3 or *OH radicals), is very low (ppm range vs. % range for H2O2 in SPM). Results show that the solubility decreases with temperature, whereas reactivity of O3 toward organic contamination tends to increase with temperature [1]. Therefore, the cleaning efficiency/unit of ozone is higher at elevated temperatures. Because of the physical limitations of solubility and diffusion, typical immersion-based ozone/DI processes provide stripping rates that are too slow to fulfill the requirements of a production process.
Boundary-layer-controlled ozone processing
For a successful ozone application, sufficient oxidants are required in the vicinity of the confined layer near the wafer surface. Boundary-controlled processing (Fig. 3) overcomes the obvious contradiction between O3 solubility and reaction kinetics, and maximizes the O3 concentration at elevated temperatures. For an immersion-based process, solubility issues limit the availability of reactive O3 species and the diffusion limitation further reduces the wafer surface concentration. If, however, the wafer could be brought into direct contact with O3 gas, while maintaining a few monolayers of moisture on the surface, a significantly larger transfer of reactive O3 species toward the wafer surface would become possible.
 |
Figure 2. Resist strip rate for immersion process with and without acetic acid addition (10-3 M).
 |
Figure 3. Schematic illustration of the physical enhancement obtained by the boundary layer control process.
 |
Figure 4. Resist strip rate for immersion and boundary control process with and without acetic acid addition (immersion: 10-3 M; boundary control: 10-2 M).
 |
Figure 5. SEM graph of 0.4-?m via structures; (top) after O2 dry strip, and (bottom) after O3 moist gas phase process with HAc addition for Al overetched via structures.
The boundary control concept has been emulated by generating a moist ozone gas phase, but is generally applicable (e.g., spray processor). A sealed ambient is filled with water vapor while maintaining a constant O3 gas flow. By exposing the wafers to this moist O3 ambient, a thin condensation layer forms on the wafer, while the gas ambient maintains a continuous supply of O3. Also, this thin condensation layer reduces the diffusion limitation by allowing the short-lived reactive O3 components to reach the wafer surface, thus increasing process efficiency as much as fivefold over immersion-based O3 processes (Fig. 4).
The achieved enhancement in resist strip rates makes the boundary-layer-controlled ozone process competitive with SPM processes with respect to total process times. The long DI rinse process required after sulphuric acid treatments becomes obsolete after ozone processing. Additionally, chemical consumption and supply can be greatly reduced, thus lowering the overall cost of ownership of the resist removal process.
Application to via cleaning
Besides the replacement of most sulphuric-based process steps in front-end-of-line (FEOL) IC production, we also evaluated the process for possible use in back-end-of-line (BEOL) applications, using a moist ozone gas phase with acetic acid addition to clean via structures. CF4 plasma etching in oxide/SOG/oxide stacks prepared the via test structures. By overetching the holes in oxide on TiTiN/Al, we created many post-etch residues. Wafers, with resist and sidewall polymers, are exposed to the boundary control O3 clean directly. While we did not perform electrical testing, we monitored via cleanliness with secondary electron microscopy imaging, using a wafer dry-stripped for 45 min in an O2 plasma as a reference. After 45 min of O2 dry-strip, the typical post-etch polymer residues are still clearly visible (Fig. 5). A 10-min treatment using the moist O3 gas phase procedure, however, shows the dramatic enhancement in cleanliness. Both resist and post-etch polymer residues are stripped rapidly, without requiring any environmentally harmful solvent strippers.
Conclusion
IMEC`s new environmentally friendly process successfully removes photoresist and post-etch residues from silicon surfaces. Its improved performance over traditional processes is due to the enhanced reactive ozone availability near the wafer surface. Evidence indicates that this process can be used to eliminate all sulfuric acid-based process steps in IC production.n
Acknowledgments
The authors thank their colleagues Paul Snee, Ingrid Cornelissen, Marcel Lux, Rita Vos, Paul W. Mertens, Martin Knotter, and Marc Meuris for their valuable contribution to this research.
References
1. S. De Gendt, et al., "A novel resist and post-etch residue removal process using ozonated chemistries," Symposium on VLSI Technology Digest of Technical Papers, p. 168, 1998.
2. J. Staehelin, J.Hoigne, "Decomposition of ozone in water," Environ. Sci. Technol., 16, p. 676, 1982.
3. K. Sehested, et al., "The primary reaction in decomposition of ozone in acidic aqueous solutions," Environ. Sci. Technol., 25, 1589, 1991.
STEFAN DE GENDT received his PhD in chemistry in January 1996 from the University of Antwerp, Belgium. His current research topics include cleaning technology and analytical metrology for contamination control in CMOS processing. He has (co-)authored 40 technical papers in refereed journals and conferences and is the (co-)inventor of cleaning processes, resulting in several patent applications.
JAN WAUTERS received his PhD from the University of Leuven, Belgium. He is a scientific editor at IMEC with responsibility for authoring and editing the research organization`s numerous company technical documents and publications. Prior to joining IMEC, he was a nuclear research scientist at the University of Tennessee at Knoxville. IMEC, Kapeldreef 75, B-3001 Leuven, Belgium; e-mail [email protected].
MARC HEYNS received his PhD in 1986 from the Katholieke Universiteit Leuven, Belgium. From 1979 to 1985, he held a fellowship from the National Fund for Scientific Research (NFWO) in the Laboratory for Physics and Electronics of Semiconductors at K.U. Leuven. In 1986, he joined IMEC, where he is now the leader of a research group working on ultraclean processing technology, and environment, safety and health issues in IC production and CMP.