Most liquid are highly absorbing in the LWIR region because many of their fundamental vibrational frequencies lie in this region. Typical absorption coefficients are >10³ cm⁻¹, thus absorption depths are of the order of a few μm. Spectroscopic studies of such liquids require use of very thin absorption cells or other techniques that utilize μm absorption depths. Evanescent wave spectroscopy [1–3] is a widely used alternative to transmission studies of highly absorbing liquids using very thin cells. Attenuated total reflection (ATR) geometry cells are extensively used for such studies in conjunction with FTIR. Figure 1 shows the principle of ATR spectroscopy, where we have replaced the FTIR source with a tunable laser source.

 

Fig. 1 Principle of ATR spectroscopy.

Download Full Size | PDF

The ATR material is chosen to be transparent in the spectral region of interest and the injected radiation propagates through the material via one or more total internal reflections. At each reflection, evanescent wave penetrates the medium in which the ATR crystal is immersed and experiences attenuation arising from the medium. This results in a reduction in transmitted power from the incident beam, which can be monitored by measuring the exit radiation intensity. The depth to which the radiation propagates into the surrounding medium and therefore samples the thickness of the medium is typically of the order of the wavelength of light in the medium. Thus at ~10 μm the penetration depth is ~5 μm, if the refractive index of the medium is 2.0. The ATR cells, thus, provide transmission cells of equivalent thickness, i.e., of the order of $\lambda /n$, where $\lambda$ is the wavelength of light and $n$ is the refractive index of the medium. It should be noted, however, that the exact depth to which the radiation samples the liquid absorption depends on the refractive index of the ATR crystal, the refractive index of the liquid and the internal angle of incidence [3]. Commercially available ATR cells utilize single or multiple total internal reflections of the incident radiation. For the presently used setup, where we make measurements over a relatively narrow spectral region, the dispersion of the ATR material is negligible. However, as can be recognized, the total internal reflection of the incident radiation is effective only as long as the refractive index of the medium (in this case the liquid) is smaller than that of the ATR material.

To a large extent, ATR spectroscopy has been used with FTIR spectrometers. Laser based ATR spectroscopy is, however, attractive because of the higher powers that are available (in a given resolution bandwidth) and the laser light is already collimated, which is ideal for launching the light into the ATR crystal. In the present studies, we have used non-flow ATR arrangement manufactured by Specac. The ATR crystal material is ZnSe with a refractive index of ~2.4 in the spectral region of interest in this paper (~8.5 μm to ~9.5 μm), with relatively small dispersion in this range. The refractive index is isotropic. Moreover, ZnSe has very little absorption in the spectral range from ~2 μm to ~14 μm, making it ideal for ATR studies in the mid wave and long wave infrared regions. The input angle of the radiation into the ATR cell will determine the number of reflections encountered by the radiation before it exits the ATR cell, with the radiation sampling the liquid at each reflection. In the present case, where we are using a commercial cell, the input angle is fixed by design.

Figure 2 shows the experimental arrangement. The laser radiation is linearly polarized (vertical with extinction ratio >500:1) and the polarization is parallel to the internally reflecting surfaces of the ATR crystal. To explore potential anisotropic liquids (not done here), we can use a half wave plate in the path of the laser radiation to rotate its polarization from vertical to horizontal.

 

Fig. 2 Experimental setup for ATR measurements of liquids using tunable radiation from an AOM tuned QCL.

Download Full Size | PDF

We have used the recently developed AOM tuned QC laser (Pranalytica’s VeloXscan-90) [4–8] as the source of tunable radiation. The VeloXscan-90 provides tunable radiation from ~8.65 μm to ~9.45 μm. The VeloXscan-90 is an acousto-optically tuned (all electronic) QCL system, has no moving parts, and is capable of wavelength switching in less than ~700 ns [3] regardless of the size of the wavelength change. Laser wavelength tuning is accomplished via an acoustic phase grating created in an acousto-optic modulator (AOM). Laser wavelength is scanned by varying the RF frequency driving the AOM. In our system, the RF frequency is generated by a voltage controlled RF oscillator. With analog driven laser tuning, i.e., when the AOM RF frequency is changed using an analog voltage ramp to the voltage controlled oscillator (VCO), we have shown VeloXscan is capable of providing a complete scan from ~8.65 μm to ~9.45 μm in 10 μs – 20 μs [4]. The laser wavelength is directly determined by the AOM drive frequency, making the wavelength calibration convenient. In the present studies, a Vigo PVM-10.6 detector with VPDC-5H preamplifier, having a response time of ~30 ns, monitors the transmitted infrared radiation. The detector has a relatively flat spectral response (detectivity of >$1\times {10}^{7}cm\cdot \sqrt{Hz}\cdot W^{-1}$) from ~2.5 μm to >10.6 μm.

It should be noted that fiber based evanescent wave spectroscopy in the infrared region with QCLs [9] and other lasers [10,11] is an alternative to using the ATR prisms, described in the present paper.

In order to make the data more reproducible and amenable to integration, if necessary, we have modified the RF frequency change from being analog driven to being digitally driven. Instead of an analog voltage ramp driving the VCO, we generate a digital staircase whose step size is determined by the desired resolution. Application of the digital staircase voltage to the VCO, in turn, results in a wavelength scan that is discretely tuned in steps. At present, the speed of staircase generation is limited to ~1.25 μs per step (regardless of the size of the step) by the lap top computer that generates the voltage staircase. Nonetheless, a complete scan over the tuning range can be accomplished in ~500 μs if we have ~400 wavelength steps. The spectral resolution provided by 400 steps from ~8.65 μm to ~9.45 μm is ~2 nm, which is more than sufficient for spectroscopy of liquids, whose absorption features are generally 200 nm wide. We report herein liquid absorption spectra obtained with complete scan time of 500 μs.

The Specac cell ATR crystal samples liquid absorption on two of its vertical sides. The cell is filled to completely cover the two vertical sides, which takes about 2 cc of the liquid under test. Care is taken to avoid any air pockets between the liquid sample and the ATR crystal faces by gently vibrating the ATR cell prior to measurements. Actual liquid sample thickness in contact with the ATR crystal is about 2 mm, as determined by the Specac cell design. Having liquid on both sides of the crystal provides roughly twice the number of times the infrared radiation samples the liquid. In the present cell, there is no flexibility for filling the cell on one side, if required for liquids, which may have inordinately high absorption. The presently used ATR cell is of static design. However, Specac makes attachment for the cell that can convert it to a flowing cell, if necessary. This attachment was not used in the present studies.

Figure 3 shows the ATR spectrum of IPA. The predominant absorption peaks at 8.85 μm and 9.01 μm, expected from PNNL data, are well reproduced in the fast spectroscopic laser scan using the VeloXscan. There are some differences, however and these are in the relative intensities of the two absorption peaks compared to PNNL data. In the current preliminary studies, we have not addressed this issue. The absorption peak at ~8.61 μm is also seen. However, this wavelength is near the end of the scan and thus the data are noisy. The stability of the AOM tuned QCL system wavelength scan is reproducible over hours at a time.

 

Fig. 3 ATR absorption measurements of IPA. Single scan data with scan time of 500 μs.

Download Full Size | PDF

Figure 4 shows absorption measurements of water, which surprisingly show no distinctive absorption peaks. The absorption is small but essentially flat over the wavelength range from 8.65 μm to 9.4 μm. Thus, we anticipate that using water as a solvent for other chemicals should be possible for spectral measurements in the 8.65 μm to 9.4 μm region. The steep rise in absorption at either end of the scan is an artifact arising from the power instability of the laser operation at the edges.

 

Fig. 4 ATR absorption measurements of water. Single scan data with scan time of 500 μs.

Download Full Size | PDF

Figure 5 shows absorption measurements on a sample of ethanol, along with PNNL data obtained using an FTIR spectrometer. The expected absorption peak at 9.173 μm is clearly seen and can be used for calibrating ethanol concentration in a water/ethanol mix. This is possible because, as seen from Fig. 4, infrared absorption of water is flat in the spectral region of interest.

 

Fig. 5 ATR absorption measurements of absolute ethanol. Single scan data with scan time of 500 μs.

Download Full Size | PDF

In order to evaluate the capability of the AOM tuned QCL system for quantitative measurements, we carried out studies of absorption caused by ethanol when it is progressively diluted from 100% to 0%, using water as the diluent. Figure 6 shows the measured absorption signal at ~9.173 μm plotted as a function of ethanol concentration in water. Since water has a relatively flat absorption in the spectral region of interest (Fig. 4), the measured absorption, when ethanol is diluted with water, has two contributions. The first is the absorption from ethanol, which goes down as the dilution with water increases. The second contribution is from water, whose absorption signal goes up as dilution increases. Model calculations indicate that the measured absorption should vary linearly with ethanol concentration. The measured data in Fig. 6 are fitted to a straight line (R² = 0.9975), in agreement with the expectations. With improved stability of the AOM tuned QCL, we expect to be able to determine the alcohol content of mixture to better than 1%.

 

Fig. 6 Ethanol absorption as a function of dilution with water.

Download Full Size | PDF

In order to assess the usefulness of the ethanol dilution data in Fig. 6, we also measured optical absorption of four commercial alcoholic beverages: Gin (Amsterdam, 80 proof), Gin (Bombay Sapphire, 94 proof), Vodka (Amsterdam, 80 proof) and Scotch (John Barr, 80 proof). As an example, Fig. 7 shows absorption spectrum of commercially available vodka (Amsterdam, 80 proof). Again the ethanol peak is clearly seen and comparing the peak size in Fig. 5 for absolute ethanol we estimate the alcohol content to be ~42%, close to the stated 80 proof of vodka. The alcohol beverage data are plotted on Fig. 6, showing a good fit with the ethanol dilution data. With improved version of the AOM tuned QCL system providing greater long term stability and simultaneous reference and transmitted intensity measurement setup, we anticipate that the measurement technique described here can be used for rapid assessment of alcohol content of liquids. With other alcoholic beverages, such as wines, there is a possibility that the presence of sugars (glucose) and other organic compounds may provide interference. Since glucose absorption does not overlap ethanol absorption, we do not anticipate problems. However, these tests will be carried out in the future with the next generation of measurement setup (see above) to confirm the expectation.

 

Fig. 7 ATR absorption measurements of vodka. Single scan data with scan time of 500 μs.

Download Full Size | PDF

Figure 8 shows data for 2,2,2-Trifluoroethanol (TFE), which shows very strong absorption at ~9.2 μm. (TFE is used as a solvent in organic chemistry and industrially trifluoroethanol is employed as a solvent for nylon as well as in applications of the pharmaceutical field.) Towards the shorter wavelength side, starting at ~8.9 μm, another strong absorption band starts, however, power output from the AOM tuned QCL is dropping steeply below ~8.8 μm, making quantitative assessment at shorter wavelengths difficult. Nonetheless, the expected main feature of liquid 2,2,2-Trifluroethanol at ~9.2 μm [12] is well reproduced in a single scan of 500 μs.

 

Fig. 8 ATR absorption measurements of 2,2,2-trifluoroethanol. Single scan data with scan time of 500 μs.

Download Full Size | PDF

Finally, Fig. 9 shows absorption data for MgSO₄ dissolved in water. The absorption peak at ~9.11 μm is identified as arising from the ${\nu }_3(b_1)$ mode of MgSO₄ [13].

 

Fig. 9 ATR absorption measurements of MgSO₄ solution in water. Single scan data with scan time of 500 μs.

Download Full Size | PDF

In conclusion, we have demonstrated the capability of AOM tuned QCL system for fast spectral absorption data collection for a broad variety of highly absorbing liquids. We have experimentally verified that the spectral scan can be speeded up for a scan time of 100 μs, without any loss of spectral resolution for faster studies. For studies of ethanol, we have shown data for dilution of ethanol with water. These measurements can be refined for use as a standard way for rapidly measuring proof of commercial liquors. The refinements will include using a reference detector (in addition to the currently used detector that measures the transmitted intensity) to simultaneously measure the incident power and the transmitted power and broader range of tuning from the AOM tuned QCL system. Finally, the speed with which data can be collected makes the AOM tuned QCL system ideal for real time process monitoring in chemical, biological and food and wine industry.