Electron-beam exposure mechanisms in hydrogen silsesquioxane investigated by vibrational spectroscopy and in situ electron-beam-induced desorption
(Color online) Baseline Raman spectra for silicon, fused silica, and glass substrates. Both silicon and fused silica have minimal baseline signal in the frequency range of interest , although silicon has considerable background signal at lower frequencies. The high broadband background signal of glass, which is likely from autofluorescence due to the presence of more impurities than in fused silica, makes it unsuitable for these experiments. The spectra for quartz was very similar to that of fused silica and is not shown here.
(Color online) (A) Raman spectrum of HSQ on a quartz substrate after 60 min of baking at various temperatures. The decrease in the peak at with increasing temperature is clearly visible, as is the presence of the peak at higher temperatures. (B) FTIR spectrum of HSQ on a silicon substrate (normalized to the Si background spectrum) after 60 min of baking at several temperatures. The decrease is visible, but even in the magnified inset it is difficult to see the peak.
(Color online) Schematic illustration of the redistribution reaction that may be partially responsible for cross-linking in HSQ. A and bond are broken on adjacent HSQ cages (1). The resulting oxygen radical bonds to the site formerly occupied by the hydrogen atom on the adjacent cage, while the now-free hydrogen atom bonds to the Si atom at the site formerly occupied by the oxygen atom, resulting in two slightly altered HSQ cages connected by an oxygen atom (2). Note that this reaction does not depend on the presence of any external reactant, such as water, in order to occur.
(Color online) Raman spectra in the band for HSQ film on silicon, exposed to various doses of electron-beam radiation. The background signal of the silicon substrate was subtracted from all measurements. As in the thermal-curing case, the peak shrinks steadily with increasing electron dose. Peaks in the region are broader and less well resolved than in the thermal case, however, and also appear to saturate at a low electron dose, suggesting that the behavior in this regime may be more complex than in the thermal case.
(Color online) Raman spectra of the exposure in Fig. 4, fitted to three different Gaussian–Lorentzian peaks. The peak at corresponds to the vibration, while the peak at corresponds to the vibration. As the dose increases, the peak decreases and the peak increases, suggesting that increasing hydrogen redistribution in the resist during electron exposure will eventually progress to the point of creating volatile silane .
(Color online) EBID data showing the partial pressures of and as a function of dose, obtained by exposing a 180-nm-thick HSQ film to a 5 keV electron beam and recording the partial pressures of the various evolved products with a mass spectrometer. The ambient chamber pressure was . When the beam is turned on, the partial pressures of both and immediately increased, and returned to baseline when the beam was switched off. The dose rate was , the energy deposition at 5 keV being equivalent to a dose rate of in a 100 keV e-beam lithography system. The presence of confirms the redistribution reaction suggested by the Raman results in the previous section. The solid lines are different experiments, and the broken line is a double-exponential fit to the data, with the same decay constants for and .
Vibrational modes in HSQ that can be observed as peaks in Raman and/or FTIR spectroscopy. The wave number of each peak is listed, as well as the response of each peak (increase, decrease, or broadening) to increased thermal excitation.
Article metrics loading...
Full text loading...