Striking a Compromise Between Good and Perfect Silicon
-- Semiconductor International, 10/1/1999
Wafer cleaning is the most prevalent process in IC manufacturing. While wet chemical cleans are still the mainstay technology, the use of 'dry' gas-phase techniques is being driven by contamination control and ~0.2 µm resolution requirements for gigabit-level devices. Thus new dry techniques are emerging to meet process needs.
Most dry UV cleaning methods use excimer lasers, often causing surface damage to the wafers, and remove contamination through photothermal ablation, resulting in carbon deposits and airborne ablative debris. One new technique, UV-photoreactive cleaning, is wafer-safe and leaves behind no residue. Developed by researchers at UVTech Systems Inc. (Wayland, Mass.), photoreactive cleaning is a process in which the 'photo' component of reactions dominates the thermal component. By using UV radiation, contaminants are removed through photochemical erosion instead of ablation, resulting in a cleaner clean process.
UVTech's Laser Broom dry cleaning system combines reactive gases and UV light to form a gas reaction zone. A different gas mixture is formulated for each type of contaminant. The Broom sweeps a 248 nm KrF excimer laser beam across the surface while infusing the gas mixture into the reaction zone (Figure). With temperatures kept below 200°C, the reactive gas mixture assists the conversion of contaminants into volatile byproducts, enabling the removal of organic layers and particulate contamination.
 | |
|
Figs. 1 Laser Broom removal of photoresist from
SiO2. |
Resist Strip Process Developed
for Low-k Materials
A plasma strip process designed for highly selective removal of photoresist and residues over HSQ low-k material has been jointly developed by Eaton Semiconductor Equipment Operations' Fusion Systems Division (Rockville, Md.) and Dow Corning (Midland, Mich.) for 0.18 µm processes. The ultimate goal of the program is to develop a full-dry plasma strip process that will have high photoresist to HSQ ash selectivity, low k-value change, and be free of 'via poisoning' for the HSQ-based FOx low-k films which are targeted to replace SiO2 for 0.18 µm processes.
The study looked at k-value shifts and strip selectivity during resist strip using four gas mixtures with differing amounts of hydrogen, fluorine and oxygen gas flows; ashing pressure and temperature. Using Eaton's FusionGemini ES (enhanced strip) plasma asher, an ion-depleted downstream microwave plasma source limited radiation and ion production to achieve a high ash selectivity (as high as 100:1) and very small change in k-value of the films (as low as 0) on blanket-coated FOx film wafers. Overall results indicate low loss of FOx material, high ash selectivity (>20:1 for all acceptable throughput processes), high Si-H bond retention and correspondingly small changes in k-value (~0.1). The hydrogen-bearing gas mix provided encouraging results, especially at low pressure, producing less FOx loss, higher Si-H bond retention and lower k-value with higher ash selectivity. A high-temperature process maintained FOx film thickness and film properties for higher ash selectivity when a fluorine-bearing gas mix was used. This high-temperature process also enabled a faster process time. Results also showed that uncontrolled addition of O2 causes a reduction in film thickness and formation of Si-OH bonds in FOx , thereby increasing the k-value.
This study demonstrated highly selective dry strip over FOx with high strip rates while maintaining
k-values, on blanket coated wafers. The next stage of the study will optimize
the strip parameters further and demonstrate the process, including strip and
residue removal, on patterned wafers.
