In femtosecond stimulated Raman microscopy (FSRM) a spectrally broad pulse (Raman probe) and a spectrally narrow pulse (Raman pump) interact in a sample and thereby generate a Raman spectrum of the focal volume. Here a novel light source for FSRM is presented. It consists of an 8-fs laser (repetition rate of 75 MHz) operating as Raman probe. A Yb3+ based fiber amplifier generates the Raman pump light at 980 nm. The amplifier is seeded by the spectral wing of the 8-fs laser output which ensures synchronisation of pump and probe pulses. Spectral and temporal characteristics of these pulses are reported and simultaneous recording of broadband Raman spectra relying on these pulses is demonstrated.
©2009 Optical Society of America
Raman microscopy and micro-spectroscopy [1, 2, 3] find applications in, e.g. histology [4, 5], microbial typing [6, 7], and polymer science [8, 9, 10]. These research areas profit from the good correlation between spectral signatures of Raman scattering and chemical structures. Raman microscopy thus carries the potential for label-free identification and quantification of chemical entities on the micrometer length scale.
Conventional Raman microscopy rests on the spontaneous Raman effect. It suffers from low Raman cross sections and concomitantly small signal levels and/or long acquisition times . This has motivated the introduction of non-linear techniques into Raman microscopy.
During the last decade the focus has hereby been laid on coherent anti-Stokes Raman scattering (CARS) [11, 12, 13, 14]. The CARS effect indeed increases signal levels tremendously —yet, at certain expenses. Due to a non-resonant background CARS spectra are distorted in comparison to conventional Raman spectra [14, 16, 17]. The CARS spectrum of a sample containing several components is not the (weighted) sum of the individual CARS spectra. CARS signals scale quadratically with sample concentrations . All three properties hamper the analysis of CARS data.
We have recently demonstrated the applicability of a different non-linear effect in Raman microscopy, namely femtosecond stimulated Raman scattering (FSRS) . In FSRS which originates from time resolved spectroscopy [19, 20] two synchronized laser pulses are employed (Fig. 1). An intense and spectrally narrow pulse (center frequency ω 0) is referred to as Raman pump. Its width Δω0 limits the spectral resolution. A spectrally broad and less intense pulse serves as Raman probe (frequency components ωc). Ideally, its spectral width ΔωPr should exceed 3000 cm-1, so that Raman resonances up to the largest frequency shifts can be detected. Raman pump and probe pulses overlap in space and time at the sample location. Due to stimulated Raman interactions the probe spectrum experiences modulations whenever the following condition is fulfilled: |ωc-ω 0|=ωR,i. Hereby, ωR,i denote the Raman resonances of the sample. On the anti-Stokes side (ωc-ω 0>0) stimulated Raman interactions lead to a signal reduction of the probe light (stimulated Raman loss, SRL), whereas on the Stokes side (ωc-ω 0<0) enhancements of the probe pulses (stimulated Raman gain, SRG) are observed. In the experiments described below the first condition applies. The Raman spectrum R(ωc-ω 0) is obtained from the modulated spectrum IRPpr (ωc-ω 0) — that is the spectrum in presence of the Raman pump - by dividing it by the spectrum in absence of the pump Ipr(ωc-ω 0):
Provided that the polarization planes of pump and probe beams are parallel, the spectrum R(ωc-ω 0) is equivalent to the polarized Raman spectrum in conventional Raman spectroscopy . The spectra obtained that way are undistorted in comparison to conventional ones, show additivity with respect to the contributions of individual components and scale linearly with the concentration.
In our first demonstration of femtosecond stimulated Raman microscopy (FSRM)  we have utilized a femtosecond laser amplifier system running at 1 kHz. Raman pump pulses were generated by spectrally narrowing the output pulses of the amplifier. Probe pulses were obtained by continuum generation in a sapphire plate. With this setup a simultaneous (multi-channel) detection of complete Raman spectra as a function of the position in the sample is possible. For a couple of reasons this proof of principle setup is not suited for “real world” applications in microscopy. Because of the low repetition rate of the amplifier (1 kHz) the energy of the Raman pump pulse needs to be of the order of 100 nJ. This translates into a peak intensity at the focal point of 1012 W/cm2 which is deemed to be too high for bio-imaging . Further, generating the Raman probe light via continuum generation is a substantial source of noise. The white light continuum has intensity variation of the order of 10-2 and its spectral shape shows fluctuations with time. Finally, laser amplifiers are complex and maintenance intensive instruments and are thus not very likely to enter microscopy labs.
Several groups have recently shown that stimulated Raman microscopy is feasible with pulsed lasers operating in the 100 MHz range [23, 24, 25, 26]. In these experiments two synchronized picosecond laser pulses interact with each other. Stimulated Raman processes transfer intensity between these two pulses. Thanks to lock-in detection schemes these setups are very sensitive (intensity changes down to 10-7 can be recorded ). Yet, recording of complete Raman spectra requires the tuning of one of the pulses over a wide range and this tuning of course increases the acquisition times.
Here, we describe a laser light source, specifically tailored for the application in femtosecond stimulated Raman microscopy (FSRM). In contrast to a related light source described in ref. , we apply fiber amplification of picosecond laser pulses in the three level regime of a core pumped single mode fiber for the first time. The ~10x higher small signal gain coefficient at the three level transition around 975 nm enables the application of an extremely short length of amplification fiber , which dramatically reduces the effect of self phase modulation. This is the key to realize spectrally narrow pump pulses with low chirp at high efficiency and without external compression. Further, a careful characterisation of relative intensity noise, power, stability, wavelength and spectral shape is given. We demonstrate high speed Raman spectroscopy with full spectral coverage to show the suitability of the source for FSRM.
2. Description of the Setup
The light source relies on the large spectral band-width of sub-10-fs lasers. Because of this width the pulses emitted by such a laser can immediately serve as Raman probe pulses. No non-linear conversions are required. Raman pump pulses can be obtained by amplification of spectral components at the low or high frequency edge of the laser spectrum. In the actual setup (Fig. 2) pulses are supplied by a femtosecond laser (Fusion BB-300 from Femtolasers) emitting 8 fs pulses at a repetition rate of 75 MHz. This laser offers a good compromise between bandwidth and spectral smoothness (see Fig. 3). Lasers with even larger bandwidth often exhibit strong spectral modulations which renders them less suitable for spectroscopy. The average power of the laser amounts to 450 mW which translates into a pulse energy of 6 nJ.
The pulse train emitted by the laser passes a dielectric turning mirror (Layertec, type 107581) which transmits a spectrally narrow region around 980 nm. The reflected light serves as Raman probe. It passes a delay stage with which the temporal overlap between Raman pump and probe pulses at the sample can be achieved. Excursions of the delay stage in the micrometer range can be induced by a piezo actuator (Piezomechanik GmbH, type Pst 150/5×5/20). The transmitted light is coupled into a fiber amplifier. In the amplifier the pulses first pass a fiber based isolator (OFR, type IO-F-980APC) which decreases the amplified stimulated emission (ASE) and protects the femtosecond laser. The pulses then arrive at a wavelength division multiplexer (WDM, Optolink, type WDM-T-12-917-90-FA) which allows to feed the laser seed and the pump light into the amplifier. A diode laser (Axcel Photonics, type BF-915-0200-P5A) operating at 915 nm pumps the amplifier. Its maximum output power amounts to 200 mW. A Yb3+ doped single mode fiber (Liekki, type YB1200-4/125) of 5.2 cm length is used for amplification. Since the transition around 980 nm of the Yb3+ ions on which we rely here is of three-level nature [29, 30, 31] the length of the fiber needs to be well chosen. Otherwise, amplified light will be attenuated by absorption. Via a second WDM the pre-amplified seed is sent to a second amplifier stage and pump light is rejected. The second amplifier is identical in design to first except for the out-coupling. To minimize self-phase modulation of the amplified pulses after the second amplifier stage these pulses are coupled out immediately after the second Yb3+ doped fiber. Residual pump light is rejected by a bandpass filter (Lot-Oriel, type CH-980-20-24.2U). Pulses leaving the amplifier (Raman pump in the following) are re-collimated (C collimator, Thorlabs, type CFC-11-B-APC) and are combined with the Raman probe pulses by a glass wedge. The wedge transmits most of the Raman pump light and reflects ~0.5 % of the Raman probe power.
Transmitted Raman pump and reflected Raman probe light enter a microscope objective (Leitz, 20×, NA 0.4) which focuses the light beam onto the sample. After re-collimation by a second identical objective the Raman pump light is rejected by a set of filters (Layertec type, 107581, two Laser Components, type HR1035) and the remaining probe light is focused onto a spectrograph (Princeton Instruments, Acton SP2358). It is equipped with a 300 lines/mm grating blazed at 775 nm. With this grating the spectrograph features a linear dispersion of 10 nm/mm. The dispersed light is detected with a 512 element diode array (Hamamatsu, type S3901-512Q). The pixel pitch of 50 µm of the diode array limits the spectral resolution of detection system to 0.5 nm or ~7 cm-1. The diode array is controlled by Tec5 electronics. The detection system is identical to the one described in ref. [18, 32] except that now an array with a larger fullwell capacity is used. The array is read-out at its maximal rate of 1 kHz. A
chopper placed in the Raman pump branch operates at 500 Hz. Thereby, alternatingly spectra with pump on and pump off are recorded. To trigger the chopper and the read-out electronics a small portion of the laser output is sent to a photo-diode (Thorlabs, type Det10A/M). The 75 MHz signal of this diode is then frequency divided to 1 kHz by a delay generator card (Bergmann Meßgeräte Entwicklung, type BME-SG05p).
3. Characterisation of the Light Source
The spectrum of the Raman probe pulse (Fig. 3(a)), i.e. the femtosecond laser pulse being reflected by the dielectric mirror (DM1 in Fig. 2), peaks at 12250 cm-1 and has a width of 2000 cm-1 (FWHM). Including the spectral wings this width suffices to address even the largest Raman shifts of organic compounds of around 3500 cm-1. This is possible provided that the Raman pump is located at the very edge of the laser spectrum—as for the Raman pump with a frequency ω0 of 10230 cm-1 used in the present setup. The width Δω0 of the Raman pump pulse equals to 49 cm-1 (FWHM). The average power of the light leaving the fiber amplifier amounts to 72 mW. To distinguish between the contribution of the pulses and the ASE to this power, the emission of the amplifier was analyzed using a fast photodiode (Thorlabs, type Det10A/M) and a digital oscilloscope. The recorded time trace (Fig. 3(b)) shows that the amplified pulses feature the repetition rate of the femtosecond laser. The oscilloscope trace shows in addition to the spikes of the pulses a small pedestal caused by ASE. The temporal integral over one pulse Epulse in relation to the total integral Etotal over one cycle yields a fraction of Epulse/Etotal of 0.74. This implies that the amplifier delivers 54 mW (or 0.7 nJ per pulse) of pulsed output. The power of the seed prior to amplification in the pertinent spectral range was 42 µW. Thus, the pulses are amplified by 31 dB. The amplification retains the linear polarisation of the light.
4. Recording of Raman Spectra
Spectra recorded for neat benzonitrile and water demonstrate that with this laser light source Raman data can be obtained (Fig. 5 and 6). The liquids were held in 1 mm fused silica cuvettes placed between the two microscope objectives. We stress that the interaction length for the stimulated Raman process is not 1mmbut is given by twice the Rayleigh length of the objective. We estimate that length to be 10 µm. When Raman pump (energy of 0.7 nJ) and probe (0.03 nJ) are made to overlap in space and time at the focal point of the objectives, signal reductions on the Raman probe spectrum can be observed. Referring to Eq. (1), these reductions yield the (stimulated) Raman spectrum of the sample. The diode array detection employed here puts a lower limit to the acquisition time of a Raman spectrum of 2 ms. It requires 1 ms to record one pump on and one pump off spectrum. For this acquisition time resonances of benzonitrile at 1011, 1194, 1601, 2235, and 3083 cm-1 are discernible matching published data  (spectrum (a) in Fig. 5 represents the raw data, spectrum (b) is obtained by applying Savitzky-Golay smoothing ). The stimulated Raman effect amounts to 10-3. The noise level as estimated from spectral regions free from Raman resonances is around 10-4. The line width of the Raman resonances is ~40 cm-1 (FWHM). This is smaller than the spectral width of the Raman pump pulse (50 cm-1). This is surprising at first sight since the width of the Raman pump should limit the spectral resolution . Yet, the pump pulses employed here are chirped. The Raman probe light, therefore, does not interact with all spectral components of the pump light. This chirp is also responsible for the small spikes at the high frequency edge of the Raman resonance. Simulations (data not shown) based on a coupled wave description of FSRS [35, 36] reproduce this effect.
Spectra (a) and (b) in Fig. 5 exhibit spectral modulations at wavenumbers larger than 3000 cm-1. These modulations are due to spectral interferences between the Raman probe pulse and the Raman pump pulse which has experienced non-linear frequency broadening at the sample location . Eq. (4) in ref.  states the effect is most pronounced when the ratio between the intensity of the broadened pump pulse and that of the probe pulse is large. Since for Raman shifts larger than 3000 cm-1 the probe spectrum decreases substantially (see Fig. 3) the modulations are most pronounced here. This effect shows interferometric sensitivity and can therefore be averaged out by changing the delay between pump and probe on the length scale of a wavelength . A piezo actuator (see Fig. 2) periodically varies the delay between pump and probe at a frequency of 80 Hz. Obviously, for an acquisition time of 2 ms this frequency does not suffice to average the modulations out. For an acquisition time of 20 ms, however, these modulations are virtually gone.
Benzonitrile features a largest Raman shift of ~3000 cm-1. Higher shifts of up to ~3500 cm-1 are observed for molecules containing OH or NH groups . At least for organic molecules these may be considered as the largest Raman shifts. The spectrum obtained for water (Fig. 6) exhibits the well known broad OH stretching resonance peaking around 3300 cm-1 . This demonstrates that with the present setup Raman resonances up to 3500 cm-1 can be covered.
These results clearly show that with the light source described here complete Raman spectra of substances in the focal point of a microscope can be recorded. By raster-scanning the sample as described in ref.  it will be possible to generate Raman images. Presently, the shortest pixel dwell time in this imaging process will be 2 ms. It is limited by the read-out rate of the detection system. This system was originally designed for a 1 kHz laser/amplifier light source. It remains to be seen whether with faster multi-channel detectors even shorter pixel dwell times are possible.
Our light source is also compatible with lock-in detection schemes described in ref. [23, 24, 25]. It only requires an exit slit in the spectrograph combined with a single channel detector and an acousto-optical modulator in the Raman pump branch.
Financial support of the Deutsche Forschungsgemeinschaft (project Gi349/2-1) is gratefully acknowledged. E.P. thanks the Boehringer Ingelheim Fonds for a PhD fellowship. T.K. and R.H. acknowledge support from the Emmy Noether program of the Deutsche Forschungsgemeinschaft (DFG - HU 1006/2-1) and the European Union project FUN OCT (FP7 HEALTH, contract no. 201880). We thank our bachelor student Mario Mangold for help in recording some of the data presented.
References and links
1. G. Turrell and J. Corset, eds., Raman microscopy developments and applications (Academic Press, San Diego, 1996).
2. M. Diem, M. Romeo, S. Boydston-White, M. Miljkovic, and C. Matthäus, “A decade of vibrational micro-spectroscopy of human cells and tissue (1994–2004),” Analyst 129(10), 880–885 (2004). [CrossRef]
3. I. Notingher and L. L. Hench, “Raman microspectroscopy: a noninvasive tool for studies of individual living cells in vitro,” Expert Rev. Med. Devices 3(2), 215–234 (2006). [CrossRef]
4. B. de Jong, T. Bakker, K. Maquelin, T. van der Kwast, C. Bangma, D. Kok, and G. Puppels, “Discrimination between nontumor bladder tissue and tumor by Raman spectroscopy,” Anal. Chem. 78(22), 7761–7769 (2006). [CrossRef]
5. C. Lieber, S. Majumder, D. Billheimer, D. Ellis, and A. Mahadevan-Jansen, “Raman microspectroscopy for skin cancer detection in vitro,” J. Biomed. Opt. 13(2), 024013 (2008). [CrossRef]
6. M. Ibelings, K. Maquelin, H. Endtz, H. Bruining, and G. Puppels, “Rapid identification of Candida spp. in peritonitis patients by Raman spectroscopy,” Clin. Microbiol. Infect. 11(5), 353–358 (2005). [CrossRef]
7. D. Willemse-Erix, M. Scholtes-Timmerman, J. Jachtenberg, W. van Leeuwen, D. Horst-Kreft, T. Schut, R. Deurenberg, G. Puppels, A. van Belkum, M. Vos, and K. Maquelin, “Optical Fingerprinting in Bacterial Epidemiology: Raman Spectroscopy as a Real-Time Typing Method,” J. Clin. Microbiol. 47(3), 652–659 (2009). [CrossRef]
8. J. Dong and Y. Ozaki, “FTIR and FT-Raman studies of partiallymiscible poly(methyl methacrylate)/poly(4-vinylphenol) blends in solid states,” Macromolecules 30(2), 286–292 (1997). [CrossRef]
9. B. von Vacano, L. Meyer, and M. Motzkus, “Rapid polymer blend imaging with quantitative broadband multiplex CARS microscopy,” J. Raman Spectrosc. 38(7), 916–926 (2007). [CrossRef]
10. J. Koenig and J. Bobiak, “Raman and infrared imaging of dynamic polymer systems,” Macromol. Mater. Eng. 292(7), 801–816 (2007). [CrossRef]
11. A. Zumbusch, G. Holtom, and X. Xie, “Three-dimensional vibrational imaging by coherent anti-Stokes Raman scattering,” Phys. Rev. Lett. 82(20), 4142–4145 (1999). [CrossRef]
12. J. Cheng and X. Xie, “Coherent anti-Stokes Raman scattering microscopy: Instrumentation, theory, and applications,” J. Phys. Chem. B 108(3), 827–840 (2004). [CrossRef]
13. A. Volkmer, “Vibrational imaging and microspectroscopies based on coherent anti-Stokes Raman scattering microscopy,” J. Phys. D-applied Phys. 38(5), R59–R81 (2005). [CrossRef]
14. M. Müller and A. Zumbusch, “Coherent anti-Stokes Raman scattering microscopy,” Chem. PhysChem. 8(15), 2157–2170 (2007).
15. C. Evans and X. Xie, “Coherent anti-Stokes Raman scattering microscopy: Chemical Imaging for Biology and Medicine,” Annu. Rev. Anal. Chem. 1, 883–909 (2008). [CrossRef]
16. A. Penzkofer, A. Laubereau, and W. Kaiser, “High-intensity Raman interactions,” Prog. Quantum Electron. 6(2), 56–140 (1979).
17. J. Cheng, A. Volkmer, L. Book, and X. Xie, “An epi-detected coherent anti-Stokes Raman scattering (E-CARS) microscope with high spectral resolution and high sensitivity,” J. Phys. Chem. B 105(7), 1277–1280 (2001). [CrossRef]
18. E. Ploetz, S. Laimgruber, S. Berner, W. Zinth, and P. Gilch, “Femtosecond stimulated Raman microscopy,” Appl. Phys. B - Lasers Opt. 87(3), 389–393 (2007). [CrossRef]
19. M. Yoshizawa and M. Kurosawa, “Femtosecond time-resolved Raman spectroscopy using stimulated Raman scattering,” Phys. Rev. A 61, 013808 (2000). [CrossRef]
20. P. Kukura, D. W. McCamant, and R. A. Mathies, “Femtosecond stimulated Raman spectroscopy,” Annu. Rev. Phys. Chem. 58, 461–488 (2007). [CrossRef]
21. D. McCamant, P. Kukura, and R. Mathies, “Femtosecond broadband stimulated Raman: A new approach for high-performance vibrational spectroscopy,” Appl. Spectrosc. 57(11), 1317–1323 (2003). [CrossRef]
22. A. Vogel, J. Noack, G. Huttman, and G. Paltauf, “Mechanisms of femtosecond laser nanosurgery of cells and tissues,” Appl. Phys. B-Lasers Opt. 81(8), 1015–1047 (2005). [CrossRef]
23. C. Freudiger, W. Min, B. Saar, S. Lu, G. Holtom, C. He, J. Tsai, J. Kang, and X. Xie, “Label-Free biomedical imaging with high sensitivity by stimulated Raman scattering microscopy,” Science 322(5909), 1857–1861 (2008). [CrossRef]
24. P. Nandakumar, A. Kovalev, and A. Volkmer, “Vibrational imaging based on stimulated Raman scattering microscopy,” New J. Phys. 11(3), 033026 (2009). [CrossRef]
25. Y. Ozeki, F. Dake, S. Kajiyama, K. Fukui, and K. Itoh, “Analysis and experimental assessment of the sensitivity of stimulated Raman scattering microscopy,” Opt. Express 17(5), 3651–3658 (2009). [CrossRef]
26. K. Kieu, B. Saar, G. Holtom, X. Xie, and F. Wise, “High-power picosecond fiber source for coherent Raman microscopy,” Opt. Lett. 34(13), 2051–2053 (2009). [CrossRef]
27. Q. Hao, W. Li, and H. Zeng, “High-power Yb-doped fiber amplification system synchronized with a few-cycle Ti:sapphire laser,” Opt. Express 17(7), 5815–5821 (2009). [CrossRef]
28. R. Paschotta, J. Nilsson, A.C. Tropper, and D.C. Hanna, “Ytterbium-Doped Fiber Amplifiers,” IEEE J. Quantum Electron. 33(7), 1049–1056 (1997). [CrossRef]
29. H. Pask, R. Carman, D. Hanna, A. Tropper, C. Mackechnie, P. Barber, and J. Dawes, “Ytterbium-doped silica fiber lasers - versatile sources for the 1–1.2 µm region,” IEEE J. Sel. Top. Quantum Electron. 1(1), 2–13 (1995).
30. R. Selvas, J. K. Sahu, L. B. Fu, J. N. Jang, J. Nilsson, A. B. Grudinin, K. H. Yla-Jarkko, S. A. Alam, P.W. Turner, and J. Moore, “High-power, low-noise, Yb-doped, cladding-pumped, three-level fiber sources at 980 nm,” Opt. Lett. 28(13), 1093–1095 (2003). [CrossRef]
31. L. B. Fu, M. Ibsen, D. J. Richardson, and D. N. Payne, “977-nm all-fiber DFB laser,” IEEE Photon. Technol. Lett. 16(11), 2442–2444 (2004). [CrossRef]
32. S. Laimgruber, H. Schachenmayr, B. Schmidt, W. Zinth, and P. Gilch, “A femtosecond stimulated Raman spectrograph for the near ultraviolet,” Appl. Phys. B-Lasers Opt. 85(4), 557–564 (2006). [CrossRef]
33. I. Kozma, P. Baum, U. Schmidhammer, S. Lochbrunner, and E. Riedle, “Compact autocorrelator for the online measurement of tunable 10 femtosecond pulses,” Rev. Sci. Instrum. 75(7), 2323–2327 (2004). [CrossRef]
34. A. Savitzky and M. Golay, “Smoothing + differentiation of data by simplified least squares procedures,” Anal. Chem. 36(8), 1627 (1964). [CrossRef]
35. S. Lee, D. Zhang, D. McCamant, P. Kukura, and R. Mathies, “Theory of femtosecond stimulated Raman spectroscopy,” J. Chem. Phys. 121(8), 3632–3642 (2004). [CrossRef]
36. S. Yoon, D. McCamant, P. Kukura, R. Mathies, D. Zhang, and S. Lee, “Dependence of line shapes in femtosecond broadband stimulated Raman spectroscopy on pump-probe time delay,” J. Chem. Phys. 122(2), 024505 (2005). [CrossRef]
37. E. Ploetz, B. Marx, and P. Gilch, “Disturbing interference pattern in femtosecond stimulated Raman microscopy,” J. Raman Spectros. in press (2009). [CrossRef]
38. B. Schrader, ed., Infrared and Raman Spectroscopy (VCH, Weinheim, New York, Basel, Cambridge, Tokyo, 1995). [CrossRef]
39. B. Schrader and W. Meier, eds., Raman/IR Atlas of Organic Compounds (Verlag Chemie, Weinheim, 1975).