Limitations in the Analysis of Vibrational Sum-Frequency Spectra Arising from the Nonresonant Contribution

Abstract

Sum frequency generation (SFG) spectroscopy is a powerful tool for probing the orientations of molecules at surfaces and interfaces, but oversimplification in the treatment of the nonresonant (NR) contribution has obscured some fundamental limitations in the analysis of SFG spectra. These difficulties are demonstrated for the case of polystyrene thin films. The NR signal invariably distorts the spectrum and can cause changes in the spectra even in the absence of actual structural changes. The NR signal originates not only from the substrate but from all materials in the system and should not be modeled as having a frequency-independent amplitude. Because of its complicated nature, NR signal must be isolated experimentally in order to obtain meaningful results. Suppression of NR signal, however, causes a different type of distortion, due to apodization of the resonant signal in the time-domain. Experimental methods need to be refined to take these limitations into account and to obtain unique spectral parameters to be used for orientation determination.

Use of Variable Time-Delay Sum-Frequency Generation for Improved Spectroscopic Analysis

“…6 We have also reported different onset times of the nonresonant signal, depending on the sample; we observed a lag in the maximum nonresonant signal on a Si/SiO 2 compared to a Au surface. 6 Thus, the assumption that the nonresonant signal has no phase lag should be checked experimentally for each new sample to be studied. It seems that the only way to be absolutely certain the nonresonant contribution is not affecting the results is to experimentally suppress it and properly account for the resulting apodization effects.…”

“…Furthermore, our work questions the appropriateness of modeling the nonresonant signal with only two parameters, B NR and Φ. 6 In short, the nonresonant signal is a complicated response that carries additional information about the sample, but it is unclear how to properly deal with it.…”

“…Unfortunately, the underlying nature of the nonresonant signal is not fully understood, and our recent work has called into question some of the commonly made assumptions regarding it. 4,6 We have shown that, in the case of polystyrene thin films, the intensity of the nonresonant signal changes with the history of the sample; that is, the nonresonant signal increases with aging or annealing of the polymer thin film. 4 We have also shown that the nonresonant signal does not arise solely from the substrate material.…”

“…6 Furthermore, MEM is known to suffer from problems such as overestimation of the number of peaks 11 and has been shown to not always be accurate in phase determination. 12 All numerical methods should be used with caution, and, where possible, additional constraints should be obtained experimentally.…”

“…6,15 In fact, any broadband VR-SFG measurement is subject to this apodization effect, and effects on the line shape with delay of the visible pulse have been observed. 6,18,19 Even disregarding the nonresonant signal, the upconversion process will preferentially sample either the beginning portion of the FID, which is primarily responsible for peak areas, or the latter portion, primarily responsible for resolution. Adjusting the visible time delay to provide a maximum signal will preferentially sample the leading portion of the FID but undersample the latter portion.…”