In the fabrication of microelectronic chips, microlithography is used to transfer a pattern of circuit geometry from mask to semiconductor wafer. An important step in this process is the deposition of a thin and uniform layer of photoresist (often called resist) on which the lithographic image is exposed. Typical photoresist layers are less than 1 pum thick with a variation of 5 [angstroms] for advanced chips. Spin coating is the prevalent coating method to produce the required thickness and uniformity, but it typically wastes over 90% of the photoresist applied. A more efficient method needs to be developed for two reasons. The first is that 80% of the photoresist is an environmentally hazardous solvent. The second is the cost increase of photoresist. As the target of semiconductor industry moves toward the fabrication of smaller devices with larger capacity, the trend in photoresist shifts from i-line to deep UV resists, which allow for narrower linewidths on a chip. The price of this new resist is four to ten times higher than that of i-line resists. Reducing photoresist waste is desirable for both environmental and economical reasons. The current spin coating method has another problem in addition to low coating efficiency. Results from spin coating are unpredictable. The relationships between the inputs (process variables) and outputs (coating thickness and uniformity) can only be obtained by trial and error. Thus, a number of experiments have to be conducted to attain a certain coating thickness and uniformity. A more effective method would yield the predictable coating thicknesses and uniformities for given inputs.
展开▼
