Access Official Technical PDF

Share Summary:LinkedInEmail

Semiconductor wafer cleaning is a process-control challenge as much as a chemistry choice. A clean may need to address particles, metallic impurities, organic residue and contaminated native oxide, while also leaving a surface suitable for the next fabrication step. Findings presented in early work on wafer cleaning show why engineers must consider how a surface reacts, how contaminants move during rinsing and how energy is delivered to the wafer—not just which chemicals enter the process.

Semiconductor wafer cleaning begins at the surface

Silicon surfaces treated with hydrofluoric acid (HF) or buffered HF can be terminated primarily with hydrogen bonds. That termination can delay oxidation, but the surface’s behavior depends on its atomic-scale structure and on residual fluorine. Fluorine may remain at reactive locations such as atomic steps or surface defects, where it can help stabilize the surface against oxidation.

In the reported studies, the amount and location of surface fluorine influenced how quickly native oxide formed. A rough, hydrogen-terminated surface with sufficient fluorine at reactive sites could resist oxidation. A particularly flat silicon (111) surface prepared using pH-modified buffered HF also showed strong resistance to oxidation, despite having little detectable fluorine. By contrast, a rougher surface with less fluorine was more susceptible to oxidation.

These observations matter when a wafer must proceed from cleaning to a sensitive operation such as gate-oxide growth, contact formation or epitaxy. They show that a cleaning step can shape the condition of the surface left behind. Crystallographic orientation, roughness, termination and storage environment can all affect subsequent oxidation behavior. The reported layer-by-layer oxidation model further suggests that oxidation may begin at step edges and spread across the surface, rather than occurring uniformly at once.

Why megasonics can improve particle removal

Conventional spray processing can deliver chemicals effectively, but fluid motion near the wafer may limit the removal of very small particles. The work on megasonic cleaning describes a mechanism in which liquid clings to a hydrophilic wafer surface. During rinsing, insoluble contaminants in that liquid can be carried toward the wafer and remain attached. High-velocity sprays alone do not necessarily overcome this near-surface behavior.

Megasonic energy offers another way to address particle removal. In the reported evaluation, wafers were immersed in SC-1 or SC-2 chemistry while moved over bottom-mounted transducers. The tests examined process temperature, power, wafer movement and the number of passes over the transducers. Under the evaluated conditions, particle counts generally improved with higher power, more passes and slower movement. The reported optimum temperature for removing very small foreign material was 55°C; this is a result of that particular evaluation, not a universal recipe for every tool or wafer process.

The study also compared processing with megasonic energy on and off. For the tested wafer conditions, the megasonic-on processes achieved higher particle-removal efficiencies. This supports a practical engineering point: cleaning performance depends on how the chemistry is applied as well as on the chemistry itself. Process qualification should therefore account for equipment configuration and operating conditions, not simply the chemical sequence.

Aqueous ozone as an alternative oxidant

A separate approach used ozonated water in a three-step sequence: a sulfuric acid and ozonated water mixture (SOM), diluted HF (DHF), and an ozonated-water step. The sequence was evaluated against a modified RCA clean using sulfuric acid and hydrogen peroxide, DHF, ammonium hydroxide and hydrogen peroxide, and hydrochloric acid and hydrogen peroxide.

In this sequence, SOM serves to remove organic contamination. DHF removes the thin contaminated oxide layer. The final ozonated-water step oxidizes the exposed silicon, while also supporting the removal of light organics, metals and particles and forming a thin oxide. The approach was intended to reduce reliance on several chemicals used in the comparison sequence, including hydrogen peroxide, ammonium hydroxide and hydrochloric acid.

Ozone’s role is oxidation. The reported work notes that dissolved ozone decomposes and that its detailed reaction pathway in water is not fully established. Its practical use also depends on maintaining dissolved ozone: solubility is limited, decreases as water temperature rises, and can be affected by mass transfer, impurities and decomposition. Because ozone is unstable, the evaluated system generated ozonated water near the point of use. Under the test conditions described, the cleaning performance was reported as equal to or better than that of the modified RCA sequence for the measures assessed.

How to apply the findings to process development

These studies point to several considerations for engineers and process owners:

  • Define the contamination target. Particles, metals, organics and contaminated oxide require different removal mechanisms and may respond differently to a process sequence.
  • Evaluate the surface left after cleaning. Hydrogen termination, fluorine coverage, roughness and oxidation behavior can influence compatibility with the next operation.
  • Qualify tool settings alongside chemistry. Megasonic power, temperature, wafer movement and exposure can affect results in an immersed cleaning process.
  • Control point-of-use conditions. For aqueous ozone, generation location and water conditions matter because ozone concentration can decline through decomposition or other reactions.
  • Measure outcomes with appropriate methods. The studies used particle inspection and surface-sensitive analyses, among other methods, to distinguish cleaning effects and surface changes.

Conclusion

Effective semiconductor wafer cleaning is not a one-variable problem. Surface chemistry determines how silicon responds after treatment, fluid behavior can influence whether particles are removed or redeposited, and energy delivery can change cleaning efficiency. Megasonics and aqueous ozone represent approaches evaluated alongside established wet-cleaning methods, each with process-specific considerations. For manufacturing teams, the practical lesson is to qualify the complete sequence—including chemistry, tool conditions, rinsing and resulting surface state—against the requirements of the downstream process.

Access Official Technical PDF

About The Author