A key requirement in the field of ultrafast vibrational spectroscopy is to efficiently generate intense tunable narrowband picosecond laser pulses synchronized to a broadband femtosecond laser source. Current nonlinear methods for picosecond pulse generation suffer from complexities in both experimental implementation and pulse frequency tunability. We present here a straightforward method for spectral bandwidth compression that produces frequency tunable picosecond pulses with efficient power conversion. Broadband femtosecond laser pulses are compressed to narrowband picosecond pulses using frequency domain sum-frequency generation of spatially chirped pulses, achieving spectral bandwidths of <20 cm−1 and power conversion efficiency of ∼18%. The experimental design of the bandwidth compressor is presented and its application to stimulated Raman spectroscopy is demonstrated.
© 2018 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
Generation of intense narrowband picosecond laser pulses is of significant interest in ultrafast laser applications where high spectral resolution is required, such as femtosecond stimulated Raman spectroscopy (FSRS) [1–5] and sum-frequency generation (SFG) surface spectroscopy . More recently spectral compression of single photons for quantum information transfer has been demonstrated using nonlinear SFG in the time domain . However, spectrally compressing broadband femtosecond laser pulses to produce synchronized narrowband picosecond pulses introduces severe tradeoffs between achievable bandwidth compression, power conversion efficiency, frequency tunability and experimental simplicity.
Linear methods of bandwidth compression involve spectrally filtering a femtosecond broadband laser source, typically in a folded 4f grating filter configuration, to produce broadly tunable narrowband picosecond pulses. While experimentally straightforward, spectral filtering methods have inherently low power conversion, with throughput of about ∼0.3% for a typical spectroscopy configuration [8, 9]. Other spectral compression methods suffering limitations due to low efficiency include self-phase modulation in optical fiber [10, 11]. A more efficient method of spectral compression exploits the nonlinear phenomenon of sum-frequency generation (SFG) in the time-domain. In this technique two femtosecond pulses at a central frequency ωFF are equally and oppositely chirped in time and mixed in a suitably phase-matched nonlinear crystal to produce a narrowband picosecond pulse at the second harmonic frequency ωSH [12, 13]. Methods to generate temporally chirped pulses for non-linear mixing include grating-stretcher pairs [12–14], and chirped volume Bragg gratings . While successfully addressing conversion efficiency, time-domain nonlinear bandwidth compression techniques, including SFG and the related difference-frequency generation (DFG) [16,17], are generally experimentally complex to tune, optimize and characterize and require expensive optical components.
We present here a new, experimentally straightforward bandwidth compression technique which successfully addresses these challenges using frequency-domain SFG of spatially chirped pulses, generated using a single dispersive element and mirror imaging. The spatial chirp of each pulse is inherently equal and opposite, significantly reducing system optimization complexities, while frequency tuning of the signal pulse is readily achievable by simple system adjustments.
In this paper, we first discuss the method background, then present details of our experimental design and demonstrate efficient generation of pulses that are sufficiently narrow for vibrational spectroscopy applications and widely tunable. Finally, we present stimulated Raman spectroscopy measurements for solvents as a demonstration of one potential application for our new bandwidth compression technique.
Sum frequency generation (SFG) is a well described phenomenon whereby two photons of frequencies ω1 and ω2 and wave vectors k1 and k2 propagate together in a nonlinear medium producing a photon at the sum of their combined frequencies ω3 = ω1 + ω2 . Figure 1 shows a special case of SFG where photons of differing energies symmetric about a central frequency ωFF by amount ±Δω combine together to generate signal photons at a constant second harmonic frequency ωSH.
Our method takes advantage of this phenomenon, where spectral components of energies ω + Δω and ω − Δω mix together in the frequency domain to produce a compressed spectral bandwidth at the second harmonic. The method to generate the spatially chirped pulses uses simple mirror imaging following pulse dispersion, which naturally produces two pulses of perfectly opposite chirp symmetry. The signal bandwidth ΔλSH is proportional to the spectral bandwidths of the individual frequency components at the Fourier plane and can easily be deduced by the geometric configuration of the system. The compressed bandwidth can be predicted using Eq. (1),Eq. (2), Eq. (3), 19–21]. The use of a single dispersion element allows for a compact optical arrangement, and signal tuning over a broad frequency range is achieved with simple system adjustments.
3. Experimental setup
To demonstrate our technique, we used a Beta-Barium Borate (BBO) crystal of thickness L = 0.5 mm cut for second-harmonic generation (SHG) at λFF = 800 nm, generated from our Ti:Sapphire amplifier producing pulses with FWHM spectral bandwidth ΔλFF = 11 nm, at a 3 kHz repetition rate. Figure 2(a) shows the main components of our experimental layout, configured as an independent and portable system comprising a dispersive element (grating or prism), beam-splitter and mirror pair, focusing lens and BBO crystal arranged in a 2f configuration.
A flint H-ZF62 glass prism with angular dispersion of 6.7 nm/mrad (input at Brewster’s angle) was chosen as the dispersion element and, in a second set of experiments, a 1200 gr/mm diffraction grating blazed for 750 nm with a higher angular dispersion of 0.82 nm/mrad (as calculated for input angle 0.95 rad) replaced the prism. The input pulse energy was 13.3 μJ. Following dispersion, the pulse was split into two copies with a beam splitter (Thorlabs UFBS5050), and two mirrors inverted one pulse with respect to the other, naturally producing two pulses with perfectly symmetric equal and opposite spatial chirp. An achromatic doublet (Thorlabs AC508-250-A-ML) of focal length f = 250 mm focused the two chirped pulses to the Fourier plane as overlapping strips of length X = 2f tan Φ, where Φ is the dispersion cone angle generated by the dispersive element .
The two beams were mixed in a non-collinear geometry in the BBO crystal with an input angle of 0.07 rad between them, allowing separation of the signal and fundamental beams with an iris placed a distance after the BBO crystal. Optimization of the spatial and temporal overlap of the chirped pulses in the Fourier plane is achieved using the fine adjustment of a single optical element, mirror M1. Rotating the mirror spatially overlaps the focused strips in the perpendicular axis of the BBO crystal, while translating the same mirror along the optical axis allows path length matching to maximize temporal overlap. The signal beam is divergent perpendicular to the mixing axis and is collimated with a cylindrical lens.
4. Experimental results
4.1. Spectral compression
To characterize the spectra of the signal pulse, a fiber-coupled spectrograph (Brolight BIM-6602-02) with optical resolution (OR) of 0.2 nm recorded the pulse spectra. Figure 3(a) shows the spectra of the signal generated using an H-ZF62 prism and a 1200 gr/mm grating as the dispersive elements, demonstrating a significant compression factor of ∼9 of the input pulse, with a measured ΔλSH of 0.32 nm (20.0 cm−1) for the prism. This result is close to the predicted spectral bandwidth of ΔλSH ≈ 0.38 nm (23.8 cm−1) determined by Eq. (1).
The higher dispersing 1200 gr/mm diffraction grating resulted in the spectral compression improving beyond the OR limit of our spectrograph, 0.22 nm (13.8 cm−1), giving a compression factor of >12. The predicted bandwidth for this configuration is at ΔλSH ≈ 0.028 nm (1.75 cm−1)—a magnitude smaller than measured.
For comparison, a second harmonic pulse was generated using the same BBO, but without the use of the compressor producing a bandwidth of ΔλSH = 2.73 nm (170.6 cm−1). The temporal profile of the signal was resolved using a transient-grating method [22,23]. Results in Fig. 3(b) show a Gaussian profile with a temporal envelope of ΔτSH = 0.77 ps using the prism dispersive element and a temporal envelope of ΔτSH = 10.34 ps using the grating dispersive element. Assuming transform-limited pulses, predicted spectral bandwidths can be derived from the time bandwidth product for a Gaussian profile ΔνΔτ = 0.44, predicting a spectral bandwidth of 0.31 nm (0.32 nm measured) for the prism configuration, and 0.02 nm (1.25 cm−1) for the grating configuration. This is in good agreement with the spectral bandwidths predicted by Eq. (1). This bandwidth, and even the best measured bandwidth of 13.8 cm−1, at the OR limit of our spectrograph, is comparable or better than other systems with reported bandwidths of 1 cm−1 to 93 cm−1 [13,14,16,17].
The spatial profile of the signal beam is divergent in the vertical (y) axis and collimated in the horizontal (x) axis, orthogonal to the beam propagation (z) axis. Figure 4 shows the measured beam profile following collimation using a 50 mm focal length cylindrical lens. The beam has a Gaussian intensity profile and circular symmetry with a 1/e2 width of 1.68 mm ± 0.02 mm.
4.2. Power throughput
Power throughput was measured at the system input, immediately before the BBO crystal and immediately after the BBO crystal. The fundamental input beam was first attenuated to a pulse energy of 13.3 μJ using a variable neutral density filter. In the prism configuration, the combined energy of the two pulses was measured immediately before the BBO to be 10.3 μJ at the input to the BBO, with the narrowband signal output having a power of 1.8 μJ demonstrating an 17.5% conversion efficiency for the SFG process. Comparison with other bandwidth compressors shows our conversion efficiency is acceptable, with other compressors reporting efficiencies in the range of 0.7% to 40% [14,16,17]. SFG efficiency saturation was not achieved using the grating as the dispersive element, due to both higher power losses using a grating blazed for a Littrow geometry in a large input angle geometry, and lower power density at the BBO due to the wider strip length caused by higher angular dispersion.
4.3. Fine and broad tuning
Of significant interest to spectroscopic applications is the ability to produce tunable narrowband picosecond pulses across a wide spectral range. Temporal domain compressor systems typically report some broad tunability about the fundamental system wavelength, for example 720 nm to 890 nm  and 1000 nm to 1090 nm .
In our system, fine-tuning of the output signal frequency is easily achieved by translating one of the chirped strips with respect to the other in the plane of the BBO crystal using mirror M1. Figure 5(a) shows a normalized tuning range of ∼2.5 nm about the central frequency of 400 nm. Broad tuning was demonstrated using two additional input wavelengths generated in an optical parametric amplifier (Light Conversion TOPAS) that was pumped by our 800 nm fundamental laser. Figure 5(b) shows different SFG signal wavelengths generated by adjustment of the input angle to a diffraction grating, and rotation of the BBO crystal, generating a narrowband pulse at central wavelength 600 nm with ΔλSH of <0.41 nm (11.4 cm−1) from a λFF = 1200 nm input, recorded using a fiber-coupled spectrograph (Ocean Optics HR4000). Similarly a fundamental input pulse of λFF = 1400 nm generated a narrowband pulse at λSH = 700 nm with a bandwidth of ΔλSH <1.3 nm (26.5 cm−1) measured at the limit of our low-resolution USB spectrometer (OR 1.3 nm).
5. Application to stimulated Raman spectroscopy
To demonstrate an application of our bandwidth compressor, we used the technique of Femtosecond stimulated Raman spectroscopy (FSRS) in the steady state. FSRS is a powerful spectroscopic method developed in 2003 by Matthies and coworkers [1, 3, 4] for resolving both steady state and dynamic Raman spectra of molecules with high spectral resolution free from fluorescent backgrounds. A requirement of this method is an intense and spectrally narrow Raman pump laser pulse, which stimulates the emission of photons at Stokes-shifted (lower) energies. The spectral resolution of the FSRS system is determined by the spectral bandwidth of the Raman pump in the limit of the molecules vibrational dephasing time .
The principles behind FSRS have been well documented elsewhere . In brief, when two laser pulses of energies ωpump and ωprobe exist together in a Raman active medium, a vibrational coherence is induced when the difference in their energies Δω = ωpump − ωprobe matches a vibrational quanta of the system ωvib = Δω. Photons from the temporally long Raman pump then interact with this coherence to stimulate the the emission of photons at Stokes-shifted energies seen as a heterodyned Raman gain signal riding on the top of the probe pulse envelope. We present here steady state FSRS spectra for acetone and methanol, using the spectrally narrow 400 nm pulse generated by our prism bandwidth compressor as the Raman pump.
5.1. Experimental method
Figure 6 shows the main optical elements of the FSRS system. The generation of the Raman pump pulse is detailed in section 3 using the prism configuration. The supercontinuum (SC) pulse is generated by focusing a small potion of the fundamental laser pulse into a linearly translating CaF2 crystal to produce a probe pulse with spectral bandwidth of ∼350 nm to 800 nm. The probe and pump pulses are horizontally polarized, and focused to overlap at the center of a 1 mm cuvette in a non-collinear geometry.
Following interaction in the sample, the probe is sent to a home built Czerny Turner transmission grating spectrometer for data collection. The spectrometer has an OR of 20 cm−1 across a spectral range of 3100 cm−1. Spectra are collected on a CMOS detector array (Lightwise LW-ELIS-1024a-1394), at a read out rate of 3 kHz to capture each Raman pump on-off shot. A sequence of shots with the Raman pump on and Raman pump off is generated using a mechanical chopper running at 1.5 kHz in the Raman pump line, allowing the Raman gain spectra to be extracted and normalized by division of the pump-off probe spectrum.
5.2. Experimental results
Stimulated Raman spectra were obtained for acetone and methanol, in a 1 mm cuvette with a Raman pump energy of 3.3 μJ. Data was averaged over 60,000 shots to obtain each spectrum. Figure 7 shows the averaged spectra normalized for each liquid showing good agreement with the expected Raman peak positions obtained from spontaneous Raman spectroscopy data (LabRAM HR800). Baseline artifacts due to cross phase modulation of the pump and probe beams were removed using a polynomial fit and subtraction computation. The FWHM bandwidth resolution of the measured spectra is ∼ 30 cm−1 demonstrating good performance of our Raman pump in comparison with other steady state FSRS systems [8,26].
6. Conclusions and outlook
Potential adaptations of our system to improve efficiency include the use of a transmissive dispersive element, such as a transmission grating. Further potential adaptations include using a non-collinear geometry in the vertical plane, rather than the horizontal plane, to allow collinear mixing of the two chirped strips and remove potential compression bandwidth limitations due to geometrical smearing caused by large non-collinear incident angles to the BBO.
In conclusion, we have demonstrated an experimentally straightforward method for generating intense narrowband picosecond pulses by nonlinear second harmonic conversion in the frequency domain of femtosecond pulses in the presence of equal and opposite spatial chirp. Our technique successfully addresses tradeoffs between achievable bandwidth compression, power conversion efficiency, frequency tunability and experimental simplicity in current bandwidth compression methods. The system generates frequency tunable picosecond pulses with a spectral bandwidth of 20.0 cm−1, power conversion efficiency of ∼18% and Gaussian temporal profile, operable across a broad spectral range. One application of our bandwidth compressor, femtosecond stimulated Raman spectroscopy, has been demonstrated on liquids. We anticipate that the fine and broad tunability of our system, and its simplistic and portable design will be highly appealing for use in spectroscopic applications where both high spectral and temporal resolution are required, such as time-resolved vibrational spectroscopy, and single photon compression for quantum information transfer applications.
Marsden Grant (VUW1620); Rutherford Foundation New Zealand Postdoctoral Fellowship (RFT-VUW1602-PD).
We gratefully acknowledge the Raman Laboratory Research Group in the School of Chemical and Physical Sciences at the Victoria University of Wellington for providing us spontaneous Raman spectra for comparison with our stimulated Raman spectra.
The authors declare that there are no conflicts of interest related to this article.
1. D. W. McCamant, P. Kukura, and R. A. Mathies, “Femtosecond broadband stimulated raman: A new approach for high-performance vibrational spectroscopy,” Appl. spectroscopy 57, 1317–1323 (2003). [CrossRef]
2. M. Yoshizawa, Y. Hattori, and T. Kobayashi, “Femtosecond time-resolved resonance raman gain spectroscopy in polydiacetylene,” Phys. Rev. B 49, 13259 (1994). [CrossRef]
4. L. Zhu, W. Liu, and C. Fang, “A versatile femtosecond stimulated raman spectroscopy setup with tunable pulses in the visible to near infrared,” Appl. Phys. Lett. 105, 041106 (2014). [CrossRef]
5. B. G. Oscar, C. Chen, W. Liu, L. Zhu, and C. Fang, “Dynamic raman line shapes on an evolving excited-state landscape: Insights from tunable femtosecond stimulated raman spectroscopy,” The J. Phys. Chem. A 121, 5428–5441 (2017). [CrossRef] [PubMed]
6. S. Nihonyanagi, J. A. Mondal, S. Yamaguchi, and T. Tahara, “Structure and dynamics of interfacial water studied by heterodyne-detected vibrational sum-frequency generation,” Annu. review physical chemistry 64, 579–603 (2013). [CrossRef]
7. J. Lavoie, J. M. Donohue, L. G. Wright, A. Fedrizzi, and K. J. Resch, “Spectral compression of single photons,” Nat. Photonics 7, 363 (2013). [CrossRef]
8. D. W. McCamant, P. Kukura, S. Yoon, and R. A. Mathies, “Femtosecond broadband stimulated raman spectroscopy: Apparatus and methods,” Rev. scientific instruments 75, 4971–4980 (2004). [CrossRef]
9. L. J. Richter, T. P. Petralli-Mallow, and J. C. Stephenson, “Vibrationally resolved sum-frequency generation with broad-bandwidth infrared pulses,” Opt. letters 23, 1594–1596 (1998). [CrossRef]
10. H.-P. Chuang and C.-B. Huang, “Wavelength-tunable spectral compression in a dispersion-increasing fiber,” Opt. letters 36, 2848–2850 (2011). [CrossRef]
12. F. Raoult, A. Boscheron, D. Husson, C. Sauteret, A. Modena, V. Malka, F. Dorchies, and A. Migus, “Efficient generation of narrow-bandwidth picosecond pulses by frequency doubling of femtosecond chirped pulses,” Opt. letters 23, 1117–1119 (1998). [CrossRef]
13. S. Laimgruber, H. Schachenmayr, B. Schmidt, W. Zinth, and P. Gilch, “A femtosecond stimulated raman spectrograph for the near ultraviolet,” Appl. Phys. B 85, 557–564 (2006). [CrossRef]
14. S. Kovalenko, A. Dobryakov, and N. Ernsting, “An efficient setup for femtosecond stimulated raman spectroscopy,” Rev. Sci. Instruments 82, 063102 (2011). [CrossRef]
15. M. Nejbauer, T. M. Kardaś, Y. Stepanenko, and C. Radzewicz, “Spectral compression of femtosecond pulses using chirped volume bragg gratings,” Opt. letters 41, 2394–2397 (2016). [CrossRef]
16. H. Luo, L. Qian, P. Yuan, and H. Zhu, “Generation of tunable narrowband pulses initiating from a femtosecond optical parametric amplifier,” Opt. Express 14, 10631–10635 (2006). [CrossRef] [PubMed]
17. G. Xu, L. Qian, T. Wang, H. Zhu, C. Zhu, and D. Fan, “Spectral narrowing and temporal expanding of femtosecond pulses by use of quadratic nonlinear processes,” IEEE J. selected topics quantum electronics 10, 174–180 (2004). [CrossRef]
18. R. W. Boyd, Nonlinear optics (Elsevier, 2003).
19. C. Radzewicz, Y. Band, G. Pearson, and J. Krasinski, “Short pulse nonlinear frequency conversion without group-velocity-mismatch broadening,” Opt. communications 117, 295–302 (1995). [CrossRef]
20. A. Monmayrant, S. Weber, and B. Chatel, “A newcomer’s guide to ultrashort pulse shaping and characterization,” J. Phys. B: At. Mol. Opt. Phys. 43, 103001 (2010). [CrossRef]
21. M. Born and E. Wolf, Principles of optics: electromagnetic theory of propagation, interference and diffraction of light (Elsevier, 2013).
22. R. Trebino, K. W. DeLong, D. N. Fittinghoff, J. N. Sweetser, M. A. Krumbügel, B. A. Richman, and D. J. Kane, “Measuring ultrashort laser pulses in the time-frequency domain using frequency-resolved optical gating,” Rev. Sci. Instruments 68, 3277–3295 (1997). [CrossRef]
23. K. Chen, J. K. Gallaher, A. J. Barker, and J. M. Hodgkiss, “Transient grating photoluminescence spectroscopy: an ultrafast method of gating broadband spectra,” The journal physical chemistry letters 5, 1732–1737 (2014). [CrossRef]
24. M. Marangoni, D. Brida, M. Quintavalle, G. Cirmi, F. Pigozzo, C. Manzoni, F. Baronio, A. Capobianco, and G. Cerullo, “Narrow-bandwidth picosecond pulses by spectral compression of femtosecond pulses in a second-order nonlinear crystal,” Opt. express 15, 8884–8891 (2007). [CrossRef] [PubMed]
25. S.-Y. Lee, D. Zhang, D. W. McCamant, P. Kukura, and R. A. Mathies, “Theory of femtosecond stimulated raman spectroscopy,” The J. chemical physics 121, 3632–3642 (2004). [CrossRef]
26. M. Plewicki and R. Levis, “Femtosecond stimulated raman spectroscopy of methanol and acetone in a noncollinear geometry using a supercontinuum probe,” JOSA B 25, 1714–1719 (2008). [CrossRef]