We demonstrate ultra-broadband Fourier-transform coherent anti-Stokes Raman scattering (FT-CARS) spectroscopy spanning over 3,000 cm−1 with a rapid-scan Michelson interferometer at a scan rate of 24,000 spectra/s. Using sub-10-fs optical pulses from a mode-locked laser, we measure broad CARS spectrum covering both the fingerprint region (500-1,800 cm−1) and the C-H, N-H, O-H stretching region (2,700-3,600 cm−1). To the best of our knowledge, this is the first demonstration of coherent Raman scattering spectroscopy covering over 3,000 cm−1 at a scan rate of more than 10,000 spectra/s. Our system holds the potential for high-speed or high-throughput label-free chemical analysis, such as investigating non-repetitive chemical dynamics, taking large area images of materials or biological specimens, or counting and sorting a large number of heterogeneous cells.
© 2018 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
Raman scattering spectroscopy provides information on the vibrational levels of molecules in a label-free and noninvasive manner, thus it is widely used for investigating materials, chemicals, and biological specimens. Using spontaneous Raman scattering spectroscopy, broadband vibrational spectra that provide rich chemical information of samples can be readily obtained. However, since the spontaneous Raman scattering process typically has a low scattering cross-section, measurements demand long data acquisition times, which is a limiting factor for many applications. Coherent Raman scattering (CRS) spectroscopy such as coherent anti-Stokes Raman scattering (CARS) or stimulated Raman scattering (SRS) spectroscopy has significantly improved the data acquisition time by virtue of the enhanced nonlinear signal [1–6]. When probing a single vibrational mode, i.e., a single spectral band, one is able to acquire CRS images at video frame rates with a scanning laser microscope [7,8]. However, this high-speed measurement sacrifices the multiplex (broadband) nature of Raman scattering spectroscopy, and requires a priori knowledge of Raman band locations to tune the excitation laser wavelength for imaging. It is desirable to observe the entire Raman spectrum over a range of 3,000 cm−1 including the fingerprint spectral region (500-1,800 cm−1) and the higher energy C-H, N-H and O-H stretching region (2,700-3,600 cm−1), especially when measuring complex molecules. For example, it is helpful to measure broadband spectrum at the fingerprint and the higher frequency region when quantifying the density ratio of DNA to lipid/protein in a cell or tissue, which is important for cancer diagnosis . Also, for characterizing carbon materials such as graphene or carbon nanotubes, measuring the entire graphene bands from D to 2D’ (1,350-3,240 cm−1) is required for identifying defects and dopants .
To improve the measurement speed, several broadband CRS spectroscopies have been developed  with a multichannel detector [9,11–14], frequency-swept lasers [15,16], dual frequency combs [17,18] or a Michelson interferometer [19–25]. Among these techniques, only a few have achieved ultra-broadband detection of CRS spectrum spanning over 3,000 cm−1 [9,18,21]. Point-scanning rates of these techniques are less than a few kHz, and, to the best of our knowledge, there is no technique that is able to acquire broadband CRS spectra spanning over 3,000 cm−1 at a rate higher than 10,000 spectra/s.
In this work, we demonstrate ultra-broadband rapid-scan Fourier-transform coherent anti-Stokes Raman scattering (FT-CARS) spectroscopy spanning over 3,000 cm−1, including both the fingerprint region (500-1,800 cm−1) and C-H/N-H/O-H stretching region (2,700-3,600 cm−1), with a spectral resolution of 13 cm−1 (without apodization) at a scan rate of 24,000 spectra/s. Combination of rapid-scan FT-CARS and a 6-fs Ti:Sapphire mode-locked oscillator enables us to measure the large spectral bandwidth by virtue of the excitation mechanism of impulsive stimulated Raman scattering. Although such a short-pulse can easily get stretched by transmitting through a medium, we carefully choose the optical components and keep the amount of dispersion of the system as low as possible so that we achieve the large spectral bandwidth by simply compensating the second-order dispersion only with a chirped mirror pair. This simple dispersion management drastically reduces the complexity of the system, otherwise one has to use an additional devise such as a spatial light modulator (SLM) in order to compensate higher-order dispersion. In addition, we evaluate the signal-intensity dependence on the material dispersion of a sample itself.
2. Experimental setup
In rapid-scan FT-CARS, we use a mode-locked laser and a rapid-scan Michelson interferometer. An ultrashort pulse from the laser is split into a pair of pulses with a time delay introduced in the Michelson interferometer. After the pulse pairs are recombined, they are focused onto a sample. The first pulse excites molecular vibrations via impulsive stimulated Raman scattering (ISRS) and the second pulse probes them. Through the ISRS process, molecular vibrations which have oscillation periods longer than the excitation pulse duration can be excited by this single pulse. In other words, the spectral bandwidth of the FT-CARS is determined by the pulse duration of the excitation pulse. Depending on the delay between the first and second pulses, the second pulse generates either blue-shifted (anti-Stokes) or red-shifted (Stokes) scattered light due to the refractive index modulation of the sample induced by the first pulse. By scanning the delay, the intensity of the blue-shifted (or red-shifted) light is modulated at frequencies that reflect those of the molecular vibrations. By detecting the intensity modulation, i.e., interferogram, we observe the molecular vibrations in the time domain. The broadband CARS spectrum can then be retrieved from a Fourier-transform of the interferogram.
A schematic of the ultra-broadband rapid-scan FT-CARS system is shown in Fig. 1. We use a femtosecond Ti:Sapphire mode-locked laser at a repetition rate of 80 MHz with a pulse duration of 6 fs (venteon ultra, Laser Quantum) which covers a broad spectral range from 650 to 1,070 nm. The top and middle insets in Fig. 1 show an autocorrelation trace and a spectrum of the laser output. The pulses are guided into a Michelson interferometer with a rapid delay scanner that consists of a resonant scanner (CRS12kHz, Cambridge Technology), a curved mirror and a flat mirror aligned in a 4-f geometry as shown in the figure. After separation with a beamsplitter (UFBS5050, Thorlabs), the pulses travel into a scan- and a reference-arm. The rapid angle variation of the resonant scanner in the scan arm is translated into the change in optical-path-length. Since the delay scanner is constructed in the 4-f geometry, the pulses in the scan arm are reflected back along the same path as the incident pulses and re-combined with those from the reference arm at the beamsplitter [22, 26]. To make sure the pulses from the two arms have the same pulse duration, a dispersion-compensation glass (UDP10, Thorlabs) for the beamsplitter is inserted in one of the arms. The maximum optical-path-length difference between the arms is 0.75 mm, which corresponds to the spectral resolution of 13.3 cm−1. After passing through a 700 nm long-pass filter (FELH0700, Thorlabs), the pulses bounce off the surfaces of a pair of chirped mirrors (DCMP175, Thorlabs) several times, which compensates the second-order dispersion of the system. Then, the pulses are focused onto a sample in a cuvette (21-Q-2, internal thickness of 2 mm, Starna) with an aspheric lens at a focal length of 8 mm (C240TME-B, Thorlabs). The beam waist (radius) at the focus and Rayleigh length are estimated to be 1.0 µm and 4.1 µm, respectively. Note that the chromatic focal shift of this lens is to be 100 μm, leading to the inefficient CARS excitation. A reflective focusing element such as an off-axis parabolic mirror or reflective objective would avoid the chromatic aberration and improve the signal intensity of the CARS process. The total average power of the laser is 83.6 mW (corresponding to a pulse energy of 381 and 663 pJ for the pump and probe pulse, respectively). Anti-Stokes Raman scattered light generated at the focal point in the sample is collimated by a lens and detected by an avalanche photodetector (APD130A2/M, Thorlabs). A short-pass filter (FESH0700, Thorlabs) in front of the detector rejects the incident pump and probe pulses to improve signal-to-noise ratio of the CARS measurement. The difference in wavenumber (in our case 140 cm−1) between cut-offs of the long-pass and short-pass filters determines the lowest wavenumber of the CARS spectrum. The APD signal is sampled by a digitizer (ATS9440, AlazarTech). Since the rapid delay scanner increases the delay nonlinearly in time, it is necessary to make a time-grid correction on the CARS interferogram. For this purpose, we use interference signals (CW interferograms) of a collinearly introduced He-Ne laser (wavelength, 632.8nm) to the pulsed laser as an external clock for digitizing the CARS interferograms.
Since the 6 fs pulses are easily stretched especially when transmitting dispersive materials, we carefully design and build our system not to add excess material dispersion to the pulses. Therefore, we mostly use reflective optics to build the system. We also use a thin-plate non-polarizing beamsplitter in the interferometer instead of a polarizing beamsplitter cube and quarter-waveplates which have been used in rapid-scan FT-CARS systems for increasing the optical-power throughput and avoiding disruption of the mode-locking state of the laser caused by back-reflection. Although we lose a half of the optical power at the interferometer with the non-polarizing beamsplitter, the remaining average power is high enough at the sample for generating CARS signals. We avoid the disruption of the mode-locking state by slightly misaligning the interferometer so that the back-reflection beam does not come back to the laser cavity.
We evaluate the pulse duration at the sample position by interferometric autocorrelation via two-photon absorption in a GaP photodetector placed at the focal point of the focusing lens in front of the sample. To scan the pulse-delay for the autocorrelation measurement, we utilize the Michelson interferometer in the rapid-scan FT-CARS system. The measured autocorrelation trace is displayed in Fig. 1 (the bottom inset panel). To take into account the dispersion of a front-side wall of the cuvette, we deploy a glass plate made of the same material with the same thickness just in front of the lens while measuring the autocorrelation. By counting the number of fringes, we find the pulse duration is around 10 fs at the sample position. Based on this experimental pulse measurement, we estimate this FT-CARS system can achieve an observable spectral window of 140-3,600 cm−1. Note that we achieve such a short pulse condition at the sample position only by the second-order dispersion compensation with a pair of chirped mirrors without the need for additional higher-order dispersion compensation techniques such as pulse shaping with a SLM. The amount of the third-order dispersion of the optics in our system is expected to be as low as 400-500 fs3. The measurable bandwidth of the system could theoretically reach up to 5,000 cm−1 by compensating the higher-order dispersion, but most of the cases there is no need to see such a high frequency range because all the normal vibration modes lie below 3,600 cm−1, which is covered by our simple system.
3. Results and discussion
Figure 2(a) shows the continuous CARS interferograms of a liquid benzene measured by the developed system. The successive interferograms are obtained every 1,375 sampling points, which corresponds to 42 µs. Data sampling with the external clock at every zero-voltage-crossing of the CW interferograms ensures an equal grid spacing with respect to the optical-path-length difference. Since the pulses in the scan arm are scanned at the rate of 12 kHz, single-sided interferograms are measured at 24 kHz. Therefore, the large peaks due to the non-resonant four-wave-mixing process appear at every 83 µs. The CARS interferograms are segmented into units of single interferograms and each of them is apodized with the triangular function and Fourier-transformed separately. The non-resonant peaks are excluded from the segmentation window. Figure 2(b) shows sequential CARS spectra of benzene and a 500-averaged spectrum. Several CARS lines of benzene are observed in the fingerprint region (595, 992, 1,586 cm−1) and in the higher C-H stretching region (3,062 cm−1). The spectral line positions of the CARS signals do not fluctuate, thus the averaged spectrum appears with the sharp CARS lines without any line-broadening. This highly stable spectroscopic system promises capability for quantitative analysis. The signal to noise ratio of the line at 992 cm−1 is 44 in a non-averaged spectrum.
Next, we evaluate the CARS signal intensity dependence on the material dispersion of the sample. Since the laser pulses have a broadband spectrum, material dispersion of the sample itself gives rise to decrease the signal intensity. In order to evaluate this, CARS spectra are measured at several sample positions shifted along the optical axis with respect to the focal point of the laser. Shifting the sample position adds/reduces the amount of material dispersion on the ultrashort pulses. Figure 3 shows the normalized CARS intensity of benzene at 992 cm−1 with respect to the relative sample position against the position where the largest signal is observed. Estimating the amount of the second-order dispersion added by the chirped mirror pair, the zero position is expected to be around the center of the cuvette. We learn from the result that we expect limited effect of the sample’s material dispersion when measuring a sample with a thickness of less than 100 µm. When measuring thicker samples than this, dispersion effects must be taken into account. Note that the dispersion discussed here is highly dependent on the material. In our evaluation, we use highly dispersive benzene liquid, which has a second-order dispersion of 105 fs2/mm at 800 nm .
Finally, we measure broadband CARS spectra of various chemical species. Figure 4 shows the obtained CARS spectra of 4 chemicals such as chloroform, dimethyl sulfoxide (DMSO), toluene, and benzene. All the spectra are averaged by 500. We clearly observe the vibrational modes in the fingerprint region such as C-Cl3 stretching at 667 cm−1 in chloroform, C-S stretching at 670 cm−1 in DMSO, ring-CH3 stretching at 787 and 1,210 cm−1, ring stretching at 1,003, 1,028, 1,584 and 1,600 cm−1 in toluene, ring deformation at 595 cm−1, ring stretching at 992 and 1,586 cm−1 in benzene. Also, in the higher wavenumber region, we observe the modes such as C-H stretching at 3,017 cm−1 in chloroform, C-H stretching at 2,913 cm−1 and 2,999 cm−1 in DMSO, C-H3 stretching at 2,920 cm−1, C-H stretching at 3,053 cm−1 in toluene, and C-H stretching at 3,062 cm−1 in benzene. These results clearly show that our system is able to measure ultra-broadband CARS spectrum covering entire fingerprint and high wavenumber region around 3,000 cm−1.
As demonstrated above, we can obtain broadband CARS spectra covering over 3,000 cm−1, which includes the Raman silent region (1,800-2,700 cm−1). It has been shown that detecting vibrational modes of C-C triple bonds in the silent region is useful for biological imaging [28,29]. These alkyne-tagged molecules can be synthesized so that different conjugated molecules have specific vibrational frequencies in the broad silent region enabling multi-color fluorescence-like images. Our broadband FT-CARS spectroscopy with high spectral resolution could be an ideal system for measuring these multiple tags simultaneously.
The system can be improved with some further modifications, for example the sensitivity of the developed system can be increased by installing a heterodyne detection technique . This involves utilizing a part of the laser pulse as a local oscillator to increase the signal-to-noise ratio without increasing the excitation laser power. Increasing the signal-to-noise ratio is important for detecting lower concentrations of chemicals, such as those typically found in biological samples. Also, reducing the excitation power (pulse energy) by the heterodyne detection technique would be crucial when one measures a biological specimen which could be damaged via nonlinear optical processes. The scan rate can also be improved by using a polygonal mirror scanner, which has been demonstrated with a scan rate of 50,000 spectra/s . Furthermore, spectral resolution can be improved by elongating the maximum optical-path-length difference. A resolution of 4.2 cm−1, which is near the physical linewidth of the vibrational modes of liquids, has been demonstrated so far .
In conclusion, we developed ultra-broadband rapid-scan FT-CARS spectroscopy covering broad spectral bandwidth over 3,000 cm−1 with a spectral resolution of 13.3 cm−1 at a scan rate of 24,000 spectra/s. This is, to the best of our knowledge, the first demonstration of broadband FT-CARS spectroscopy covering over 3,000 cm−1 at the scan rate of above 10,000 spectra/s. We evaluated the effect of sample dispersion on the CARS signal intensity and showed that our system is robust for measuring highly dispersive samples with a thickness of less than 100 µm. For thicker samples, dispersion effects must be taken into account. Finally, we showed ultra-broadband FT-CARS spectra of various chemical liquids in which the characteristic Raman lines clearly appear. Our ultra-broadband rapid-scan FT-CARS has a great promise for applications such as a high-speed spectroscopy for tracking fast kinetics of chemical phenomena, a hyperspectral Raman microscope for visualizing a large area of biological or material samples, or a high-throughput flow cytometer for statistically analyzing biological cells.
JSPS KAKENHI (17H04852, 17K19071); JST PRESTO (JPMJPR17G2); Advanced Photon Science Alliance (APSA); Research Foundation for Opto-Science and Technology; The Murata Science Foundation.
We thank Laser Quantum for the use of their laser, and Makoto Kuwata-Gonokami and Junji Yumoto for letting us use their equipment. We are grateful to Faris Sinjab and Venkata Ramaiah Badarla for their critical reading of the manuscript. Junko Omachi was supported by the Advanced Integration Science Innovation Education and Research Consortium Program of the Ministry of Education, Culture, Sports, Science and Technology in Japan (MEXT).
References and links
2. J. X. Cheng and X. S. Xie, Coherent Raman Scattering Microscopy (CRC, 2012).
4. C. H. Camp Jr and M. T. Cicerone, “Chemically sensitive bioimaging with coherent Raman scattering,” Nat. Photonics 9(5), 295–305 (2015). [CrossRef]
5. C. Krafft, I. W. Schie, T. Meyer, M. Schmitt, and J. Popp, “Developments in spontaneous and coherent Raman scattering microscopic imaging for biomedical applications,” Chem. Soc. Rev. 45(7), 1819–1849 (2016). [CrossRef] [PubMed]
6. D. W. Shipp, F. Sinjab, and L. Notingher, “Raman spectroscopy: Techniques and applications in the life sciences,” Adv. Opt. Photonics 9(2), 315–428 (2017). [CrossRef]
7. C. L. Evans, E. O. Potma, M. Puoris’haag, D. Côté, C. P. Lin, and X. S. Xie, “Chemical imaging of tissue in vivo with video-rate coherent anti-Stokes Raman scattering microscopy,” Proc. Natl. Acad. Sci. U.S.A. 102(46), 16807–16812 (2005). [CrossRef] [PubMed]
8. B. G. Saar, C. W. Freudiger, J. Reichman, C. M. Stanley, G. R. Holtom, and X. S. Xie, “Video-rate molecular imaging in vivo with stimulated Raman scattering,” Science 330(6009), 1368–1370 (2010). [CrossRef] [PubMed]
9. C. H. Camp Jr, Y. J. Lee, J. M. Heddleston, C. M. Hartshorn, A. R. Hight Walker, J. N. Rich, J. D. Lathia, and M. T. Cicerone, “High-speed coherent Raman fingerprint imaging of biological tissues,” Nat. Photonics 8(8), 627–634 (2014). [CrossRef] [PubMed]
11. H. Kano and H. Hamaguchi, “Ultrabroadband (>2500cm−1) multiplex coherent anti-Stokes Raman scattering microspectroscopy using a supercontinuum generated from a photonic crystal fiber,” Appl. Phys. Lett. 86(12), 121113 (2005). [CrossRef]
12. R. Arora, G. I. Petrov, V. V. Yakovlev, and M. O. Scully, “Detecting anthrax in the mail by coherent Raman microspectroscopy,” Proc. Natl. Acad. Sci. U.S.A. 109(4), 1151–1153 (2012). [CrossRef] [PubMed]
13. W. Rock, M. Bonn, and S. H. Parekh, “Near shot-noise limited hyperspectral stimulated Raman scattering spectroscopy using low energy lasers and a fast CMOS array,” Opt. Express 21(13), 15113–15120 (2013). [CrossRef] [PubMed]
14. C. S. Liao, M. N. Slipchenko, P. Wang, J. Li, S. Y. Lee, R. A. Oglesbee, and J. X. Cheng, “Microsecond scale vibrational spectroscopic imaging by multiplex stimulated Raman scattering microscopy,” Light Sci. Appl. 4(3), e265 (2015). [CrossRef] [PubMed]
15. Y. Ozeki, W. Umemura, Y. Otsuka, S. Satoh, H. Hashimoto, K. Sumimura, N. Nishizawa, K. Fukui, and K. Itoh, “High-speed molecular spectral imaging of tissue with stimulated Raman scattering,” Nat. Photonics 6(12), 845–851 (2012). [CrossRef]
16. S. Karpf, M. Eibl, W. Wieser, T. Klein, and R. Huber, “A Time-Encoded Technique for fibre-based hyperspectral broadband stimulated Raman microscopy,” Nat. Commun. 6(1), 6784 (2015). [CrossRef] [PubMed]
17. T. Ideguchi, S. Holzner, B. Bernhardt, G. Guelachvili, N. Picqué, and T. W. Hänsch, “Coherent Raman spectro-imaging with laser frequency combs,” Nature 502(7471), 355–358 (2013). [CrossRef] [PubMed]
18. P. L. Luo, M. Yan, T. W. Hänsch, and N. Picqué, “Ultra-broadband dual comb coherent anti-Stokes Raman spectroscopy,” in Fourier Transform Spectroscopy 2016, OSA Technical Digest (online) (Optical Society of America, 2016), paper FW2E.2.
21. K. Isobe, A. Suda, M. Tanaka, H. Hashimoto, F. Kannari, H. Kawano, H. Mizuno, A. Miyawaki, and K. Midorikawa, “Single-pulse coherent anti-Stokes Raman scattering microscopy employing an octave spanning pulse,” Opt. Express 17(14), 11259–11266 (2009). [CrossRef] [PubMed]
22. K. Hashimoto, M. Takahashi, T. Ideguchi, and K. Goda, “Broadband coherent Raman spectroscopy running at 24,000 spectra per second,” Sci. Rep. 6, 21036 (2016).
23. C. S. Liao, K. C. Huang, W. Hong, A. J. Chen, C. Karanja, P. Wang, G. Eakins, and J. X. Cheng, “Stimulated Raman spectroscopic imaging by microsecond delay-line tuning,” Optica 3(12), 1377–1380 (2016). [CrossRef]
24. M. Tamamitsu, Y. Sakaki, T. Nakamura, G. K. Podagatlapalli, T. Ideguchi, and K. Goda, “Ultrafast broadband Fourier-transform CARS spectroscopy at 50,000 spectra/s enabled by a scanning Fourier-domain delay line,” Vib. Spectrosc. 91(1), 163–169 (2017). [CrossRef]
25. M. S. Alshaykh, C. S. Liao, O. E. Sandoval, G. Gitzinger, N. Forget, D. E. Leaird, J. X. Cheng, and A. M. Weiner, “High-speed stimulated hyperspectral Raman imaging using rapid acousto-optic delay lines,” Opt. Lett. 42(8), 1548–1551 (2017). [CrossRef] [PubMed]
27. K. Moutzouris, M. Papamichael, S. C. Betsis, I. Stavrakas, G. Hloupis, and D. Triantis, “Refractive, dispersive and thermo-optic properties of twelve organic solvents in the visible and near-infrared,” Appl. Phys. B 116(3), 617–622 (2014). [CrossRef]
28. H. Yamakoshi, K. Dodo, A. Palonpon, J. Ando, K. Fujita, S. Kawata, and M. Sodeoka, “Alkyne-Tag Raman imaging for visualization of mobile small molecules in live cells,” J. Am. Chem. Soc. 134(51), 20681–20689 (2012). [CrossRef] [PubMed]
29. L. Wei, F. Hu, Y. Shen, Z. Chen, Y. Yu, C. C. Lin, M. C. Wang, and W. Min, “Live-cell imaging of alkyne-tagged small biomolecules by stimulated Raman scattering,” Nat. Methods 11(4), 410–412 (2014). [CrossRef] [PubMed]
30. K. Hiramatsu, Y. Luo, T. Ideguchi, and K. Goda, “Rapid-scan Fourier-transform coherent anti-Stokes Raman scattering spectroscopy with heterodyne detection,” Opt. Lett. 42(21), 4335–4338 (2017). [CrossRef] [PubMed]