Wet Etchant Diffusion through Photoresist during Gate Oxide Patterning

Gate oxide patterning during integrated circuits manufacturing is defined by a photolithography – etch sequence. Wet etchants are still preferred to plasma to keep smooth transistors channel. With the transistor nodes evolution, not only the lateral resist dimension has shrunk but also their thickness to get accurate optical performances: from microns thick down to few hundreds of nanometers. Therefore the resist protection of the underlying material is at risk due to wet etchant infiltration through the resist (figure 1) [1], hence the importance of controlling this phenomenon. Hereby, a wide screening is made among the wet etch conditions, polymer nature and plasma use on the resist before wet. Liquid diffusion through resist is in-situ monitored by ATR-FTIR (Attenuated Total Reflection Fourier Transform Infrared) spectroscopy coupled with a liquid cell in direct contact with wafer samples (Figure 1) [2][3]. First study deals with deionized water penetration through a commercial 248 nm Deep UV photoresist, 1,8 micron thick. The IR absorption spectrum of interest for water lays between 2900 and 3700 cm -1 (figure 2). Hence the δ(OH) absorbance band is followed along the contact duration between resist and water and its area is integrated. The absorbance ratio between value at time t and infinite time is drawn versus time (figure 3). The water diffusion follows a Carter and Kibler model [4] divided in three steps: a quick increase in OH band absorption in the first minutes, followed by a slower evolution and finally an even quicker absorption. The first phase corresponds to the free water penetration, the second one to the bound water penetration and finally irreversible polymer deterioration (swelling and hydrolysis). Although wet etch alone is mostly used to preserve smooth surfaces, etch combination can be found with a plasma followed by a wet etch. The plasma can significantly impact the wet etchant penetration through the resist. Indeed, the plasma decreases the resist thickness, changes the polymer surface and structure. The tested CF4 plasma increases both the water penetration speed and sorption in the resist (Figure 4). Furthermore four different 248 nm DUV resist are compared (figure 5). Their diffusion follows previously described model. However the first phase ends at different timing. Beyond this duration, the liquid has reached the underneath material and resist protection isn’t effective anymore. Conclusion Wet etchant penetration through p