Dual-beam FTIR spectroscopy with a mode-locked fiber laser source

In collaboration with the BNL Surface Dynamics Group, we have been exploring methods to combine fiber laser frequency comb sources with commercial FTIR instrumentation to provide sensitive and accurate measurements of weak absorption features, for example, in attenuated reflection spectroscopy of surface species.   We have developed a general dual-beam method, compatible with most common FTIR instrumentation that combines strong suppression of source noise with continuous correction for background drift. The use of a coherent broadband source enables additional signal enhancement through multi-pass geometries.   We replace the conventional lamp source in a commercial Bruker FTIR spectrometer with a polarization additive-pulse mode-locked Er fiber ring laser, centered at 1560 nm.  The modulated output beam is split into matched signal and reference beam paths and viewed with balanced InGaAs photodiode detectors.  In order to correct for intensity noise in the source,   the interferometer digitizes two channels simultaneously: Ref and Ref-Sig. This arrangement takes advantage of common mode noise suppression and increases the effective dynamic range of the acquired interferogram, while accounting for possible drift in the spectral power of the light source.

Baseline stability and small signal sensitivity comparison of single-beam conventional lamp FTIR (red) vs. dual beam laser FTIR spectra (blue).  The spectrum in the lower comparison is a portion of the second overtone of CO, measured in a single scan, using simultaneously recorded interferograms from reference and (reference signal) channels.

The noise suppression is demonstrated in the figure.  The upper panel shows the baseline noise, obtained as the log of the ratio of an empty cell blank spectrum and a (nominally identical) repeated scan.  With single-beam interferograms and an incandescent lamp source, uncompensated drift in the lamp intensity produces random baseline offsets, and higher frequency source noise is responsible for the spectral noise in the (red) baseline.  To implement the dual-beam laser variation, a (Ref-Sig) difference interferogram is algebraically subtracted from the simultaneously recorded Ref interferogram, and then inverse Fourier transformed to give a spectrum similar to a single-channel (empty cell) Sig spectrum, but computed with its broad-band background contribution derived primarily from the Ref channel.  Dividing this synthetic Sig spectrum by the inverse transform spectrum of the Ref interferogram, and taking the negative logarithm gives a dual-beam blank spectrum that depends primarily on small optical mismatch between the two paths, but not the source noise or spectral fluctuations.  The difference of two successive dual-beam blank spectra is plotted in the (blue) top panel of the Figure, and demonstrates excellent baseline stability.  Adding 160 Torr of CO to a 10 cm gas cell in the sample arm provides a weak test absorption in the second vibrational overtone.  The rovibrational spectrum is barely detectable with a single scan (40 kHz, 0.2 cm-1 resolution) using the conventional lamp source, filtered to a comparable optical bandwidth, (red center panel of the Figure).  Recorded with the dual-beam laser scheme under otherwise identical conditions (blue bottom panel) there is a greater than 10 times improvement in the data quality i.e. close to shot-noise limited sensitivity.

The intention in this work is to push FTS methods to low noise, rapid spectral acquisition, and high sensitivity, although not necessarily high spectral resolution, all without heroic laser technology that might discourage application by scientists whose interests and problems lie outside optical physics.  For gas phase samples, high sensitivity is perhaps more easily attained with cavity-enhanced methods, where the effective path length can be made large enough to accumulate more easily measurable fractional absorption signals.  For applications in interfacial spectroscopy, however, such as attenuated internal reflection or sub-monolayer film reflection spectroscopy, the effective path length cannot be made extremely large, and careful noise reduction methods like this seem the best approach for sensitive absorption spectroscopy.  The spatial coherence of broadband laser and supercontinuum sources does, however, support the prospect of at least 10- to 100-fold increases in effective path length through multiple reflections with potential applications to the near-infrared spectroscopy of surface species.

 V.V. Goncharov, G.E. Hall, Opt. Lett. 37 2406-2408 (2012).

Top of Page

Last Modified: September 6, 2013
Please forward all questions about this site to: Mahendra Kahanda