Low-frequency Raman spectroscopy is an indispensable analytical tool in multiple areas of scientific research. Single-wall (1) and multiwall (2) carbon nanotubes exhibit radial breathing mode vibrations in the range 150–200 cm^-1, which are used to characterize diameter distribution and overall quality of nanotubes as well as influence of external factors. Quality of semiconductor multilayered structures (superlattices) is assessed by observing folded acoustic modes in the range 0–100 cm^-1 (3). The relaxation modes in liquids, binary mixtures, and solutions, in the range 0–400 cm^-1, help to determine their dynamic structure (4).

While general-purpose Raman instruments incorporating a single-stage spectrograph and a notch filter to reject the laser line are sufficient for many applications, they have limitations for the aforementioned high-end research applications in terms of resolution and low-frequency data collection. With such systems, it is virtually impossible to collect Raman data at frequencies less than 50 cm^-1 from the laser line, and their resolution limits the mount of structural information that can be determined over a wider wavelength range. In this article, we introduce a triple-spectrograph system that provides the capability to obtain low-frequency Raman data with extremely high resolution.

Experimental Conditions

The experimental setup consisted of a triple-spectrograph system (TriVista 555 from Princeton Instruments/Acton, Trenton, New Jersey) with three stages of 500 mm focal length each and a triple spectrograph system (TriVista 557 from Princeton Instruments/Acton) with the first two stages of 500 mm and the third stage of 750 mm focal length. For the best stray light rejection, the system was used in subtractive mode where the first two stages were limiting the spectral bandpass and rejecting the excitation line, while the third stage was performing signal dispersion. The sample was excited by a 488- and 532-nm CW laser of 60–100 mW typical power. Ruled (900 g/mm) and holographic (1800, 2000, 2400 g/mm) gratings optimized for the visible range were used. Experiments were performed using back-illuminated CCD detectors (PIXIS:400B, SPEC-10:400B, and SPEC-10:2KB from Princeton Instruments/Acton) with different array configurations. The spectra were fully vertically binned to maximize the sensitivity. A 135° scattering scheme was used for all Raman measurements. Powder samples exhibiting low-frequency Raman modes; 99%+ molybdenum(VI) oxide (MoO₃), 99.999% elementary sulfur (S) and 99% L-cystine ([-SCH₂CH(NH₂)CO₂H]₂) were used. All sample substances were obtained from Fisher Scientific, Pittsburgh, Pennsylvania.

Results

Figure 1

Solid powder samples, such as molybdenum(VI) oxide (MoO₃), sulfur (S), and L-cystine are known for heavy dissipation of the excitation laser beam and producing strong stray light detrimental for low-frequency Raman measurements. These samples were chosen to demonstrate the capabilities of the triple spectrometer system. Figure 1 shows a 0–1000 cm^-1 spectrum of molybdenum(VI) oxide, which was taken with only 1-s integration time. Even with such a short integration time, the high-quality Raman spectrometer system obtained a signal-to-noise ratio (S/N) of about 30 in the low-frequency domain 80–200 cm^-1 and an S/N in excess of 150 for the 800–850 cm^-1 spectral region.