Broadband Coherent anti-Stokes Raman Spectroscopy Characterization of Polymer Thin Films

Coherent anti-Stokes Raman Scattering (CARS) is a high sensitivity alternative to conventional Raman spectroscopy. Ultrafast lasers are used to enable multiplex CARS, with demonstrated advantages for the study of ultra-thin films.

The advent of robust, ultrafast lasers has enabled the application of coherent nonlinear spectroscopies to routine sample characterization. Coherent anti-Stokes Raman scattering, CARS, is the nonlinear equivalent of conventional spontaneous Raman scattering. Conventional Raman is a scattering process, and the signal is highly dependent upon collection efficiency. The coherent nature of the CARS process gives rise to a directional signal simplifying collection and improving the sensitivity. Surprisingly, there has been little work exploring the utility of CARS as a thin film diagnostic. Surface enhanced Raman scattering, SERS, has been used to probe the interfacial region of thin polymer films on metal surfaces. However, limitations inherent in SERS leave much unknown. For example, SERS enhancement often requires roughened coinage metal (Au, Ag, Cu) substrates. It is difficult to characterize film surface interactions when enhancement only comes from particular surface sites. Additionally, SERS enhancement only extends a few nm into the film, making it sensitive to the interface, but blind to contributions from further into the bulk. CARS alleviates both of these restrictive conditions as nonlinear mixing occurs throughout the focal overlap region of the incident beams and is not dependent upon substrate enhancement. We have demonstrated a simple modification to an existing nonlinear spectrometer that enables CARS characterization of thin polymeric films. We find that this methodology permits rapid acquisition of high signal to noise (S/N) spectra.

A schematic of a typical, degenerate CARS experiment is given in Figure 1. Two laser pulses are incident on the sample, with frequencies ν1 and ν2. They mix nonlinearly to give an output pulse at the frequency ν3=2ν1-ν2. If the difference frequency δ=ν1-ν2 is resonant with a Raman active transition in the sample, the mixing process is enhanced and a vibrational spectrum can be recorded. In the NIST experiments, a 100 fs pulse duration laser is used to generate a broadband packet of frequencies at ν2; this provides high peak fields, improving the nonlinear mixing, and simplifies the data collection by reducing the need to tune the laser providing ν2.

Raman Spectroscopy Spectrum Illustration

Experiments were performed with a variety of thin polymer film samples to establish the potential of broad-bandwidth CARS in comparison with conventional Raman spectroscopy. Shown in the lower panel of Figure 2 is the Raman spectrum of a 138 nm polystyrene, -[CH2CHC6H5]n-, film, recorded with a 633 nm laser and a commercial Raman collection system. The weak feature at 3060 cm-1 is the Raman active fully symmetric stretching vibration of the 5 C-H bonds on the phenyl ring (C6H5: see schematic in Fig. 1). The broad background is due to fluorescence from both the polymer and the Si substrate. Shown in the upper panel of Figure 2 is the CARS spectrum of the same film. For the CARS measurement, the polystyrene (PS) vibrational features appear as ‘anti resonances’ on a non-resonant background. Both the conventional and CARS spectra in Figure 1 were taken with identical acquisition times (60 s) and comparable total power onto the sample. The CARS spectrum is clearly of much higher quality. Quantitative analysis of the data indicates that CARS measurement displays a ~25 times improvement in the Signal-to-Noise ratio, SNR.

Additional experiments for films of various polymers have been performed on diverse substrates (Si, Au, glass). The film thickness was varied from ~ 20 nm to ~300 nm. In all cases, high quality vibrational spectra were acquired, with demonstrated improved SNR compared to conventional Raman spectra. A mathematical framework for the quantitation of the CARS spectra, accounting for the influence of the substrate dielectric response, was developed and shown to be consistent with the measurements.