Preparation and characterization of reactively sputtered silicon nitride thin films

Critical evaluation of the state of the art of the analysis of light elements in thin films demonstrated using the examples of SiOXNY and AlOXNY films (IUPAC Technical Report)

“…Rather doubtful is the conclusion that the thin native oxide layer on the silicon substrates no longer influences the refractive index if the films are thicker than 20 nm. In [23], a rectilinear correlation between the refractive index and the oxygen content in at% was observed, but only six samples were measured. The refractive index at 634.8 nm determined by ellipsometry is given vs. x in the formula (SiO 2 ) X (Si 3 N 4 ) 1-X and a straightlined correlation was observed in [24].…”

“…Since the quantification seems to be straightforward, many authors do not give information on the problems they encountered or on the accuracy of the results. An example for this practice is [23], where the information was limited to the rudimentary data. A Physical Electronics model 560 spectrometer with a background pressure of less than 6.7 × 10 -7 Pa was applied and standard atomic sensitivity factors and instrumental corrections associated with the instrument were used.…”

“…The assignment of vibrations to wave numbers in several publications is summarized in Table 1 together with the references. Regardless of the interpretation of the structural background of the main IR absorption band of SiO X N Y films, many authors tried to find correlations between the peak position and the composition of the oxynitride films [10,14,23,[53][54][55], which by most authors is supposed to be rectilinear, between the peak position and the deposition parameters [27,56,57], or between the peak position and the physical properties [6,25,52,58,59]. In most publications, such a correlation is indicated, but not verified due to the low sample number or the lack in cross-checking.…”

“…One also has to take into consideration that the Si-O vibration has a higher extinction coefficient than the Si-N vibration. [23] no IR spectrum is given and the samples contain only approximately 6 at% oxygen, which is supposed to stem from surface oxidation. Also in [39], the IR spectra of oxynitride films in correlation to the composition of the samples are doubtful, because even films with an oxygen concentration given as approximately 8 at% show a sharp peak at 1040 cm -1 (Si-O vibration), but only a tiny feature at 830 cm -1 (Si-N vibration).…”

“…In most publications, such a correlation is indicated, but not verified due to the low sample number or the lack in cross-checking. The interpretation of peak frequency and peak width is complicated by the influence of other film properties such as density, stress and strain, hydrogen bonds, substoichiometry, or defects [23,53,55]. In some publications, the concentration of hydrogen is estimated by the absorption band intensity of hydrogen-related vibration bands [58,[60][61][62][63] and is even given in the unit atoms/cm 3 in [32].…”