1. Introduction

The propagation of high-power, ultrafast laser pulses in air is governed by a dynamic balance of nonlinear self-focusing and defocusing by plasma generation, leading to filaments with possible lengths up to kilometers [1]. This has triggered intensive research for remote sensing applications, such as laser-induced fluorescence (LIF) [2], laser-induced breakdown spectroscopy (LIBS) [3], or light detection and ranging (LIDAR) [4]. An additional feature of filamentation is a considerable broadening of the initial pulse spectrum, ranging from the UV to the NIR for initial pulse durations of ∼ 100 fs at 800 nm [5]. Among various applications, the large spectral width may be used for ultrabroadband coherent anti-Stokes Raman spectroscopy [6–9], where the same laser pulse provides the pump and the Stokes wavelengths. Using the 2^nd harmonic as probe, Odhner et al. [10] detected the ro-vibrational CARS signals of various gases using an initial 45 fs pulse at 800 nm. In a recent study, Mazza el. [11] used filamentation for flame diagnostics up to a Raman shift of 1600 cm^-1; Tian et al. [12] demonstrated vibrational N₂ thermometry up to a Raman shift of ∼ 3000 cm^-1. All of these studies relied on Ti:sapphire based laser systems, which are still the workhorse for the generation of multi-mJ, fs-lasers pulses. These systems provide initial pulse durations of ∼ 100 fs, that is, the initial pulse spectrum is already relatively broad. This is on one hand advantageous for ultrabroadband CARS, as additional spectral broadening may easily allow exciting Raman transitions with large shifts, e.g., H₂ with ∼4150 cm^-1 [13]. Yet, there are also considerable drawbacks. The large initial spectrum makes pulse propagation sensitive to dispersion. Moreover, combustion devices are often enclosed systems, requiring the laser beams to propagate through windows. In a recent study, Mazza et al. [14] performed CARS experiments behind a BK7 window, using an additional, external compressor. Furthermore, Ti:sapphire oscillators are sensitive to ambient conditions.

Thin-disk laser technology may provide an alternative approach. These laser systems are typically based on reliable fiber oscillators and can provide ∼ 1 ps pulses at a central wavelength of 1 µm with pulse energies of some tens of mJ after amplification. Recent studies indeed have shown evidence of filamentation in air at 1 µm and 1 ps [15,16]. However, for ultrabroadband CARS, the amount of the spectral broadening is crucial. Due to the long initial pulse duration, one expects less efficient spectral broadening, and the first results of Houard et al. [15] indeed showed spectral widths of only ∼ 40 nm after filamentation, in contrast to a few hundreds of nm for filamentation at 800 nm [5]. The ro-vibrational Raman excitation of N₂ (∼ 2330 cm^-1) [13], e.g., requires a spectral width of ∼ 200 nm at 1 µm.

In this paper, we study the spectral broadening induced by the propagation of 1 ps pulses in air with maximum pulse energies of 60 mJ. Furthermore, the spectrally broadened radiation is used as pump/Stokes beam in ultrabroadband, two-beam CARS. For possible applications in combustion diagnostics, the effect of optical windows within the beam path is studied.

2. Experimental setup

A thin-disk regenerative amplifier (Dira 500-10, Trumpf Scientific Lasers GmbH + Co. KG, Germany) is used as pump source, providing a maximum pulse energy of 60 mJ at a repetition rate of 1 kHz and a pulse duration of ∼1 ps at a central wavelength of 1030 nm. The beam is linearly polarized and has a Gaussian beam diameter of ∼12.7 mm (1/e²). The spectrum of the filamentary beam is investigated by the setup shown in Fig. 1(a). 90% of the pulse energy is reflected by beam splitter BS and then focused with different focal lengths (f = 2 m – 4 m) to induce filamentation. Approximately 1 m downstream from the filament, the beam is attenuated by a wedge and filters, before it is coupled into a spectrometer (Shamrock 500i with Marana CMOS camera, Oxford Instruments plc, UK). The signal is acquired over 100 ms (∼ 100 laser pulses). The filters block the central wavelength of the pump source (λ = 1030 nm). A short-pass filter (10SWF-1000-C, Newport) is used to detect the spectral range from 800 nm to 980 nm and a long-pass filter (1050 nm, high performance long-pass filter, OD 4.0, Edmund Optics) is used to detect the spectral range above 1050 nm. To evaluate the number of filaments, a digital camera records the beam incident on an optional metal plate, which is placed ∼ 4 m behind the focusing lens. The exposure time is 1/1250 s to allow single shot acquisition.

 

Fig. 1. Sketch of the setup. BS: Beam splitter with reflectivity (R): transmission (T) = 90:10. ND: neutral density filter. (a) Scheme for the spectral analysis. The position of the wedge is ∼1 m after the filament. (b) Optical layout for filamentation based CARS. Pump: 90% of the laser source is focused by a lens to induce filamentation. Probe: 10% of the laser beam after frequency doubling to a wavelength of 515 nm. λ/2: half-wave plate. P: cubic polarizing beam splitter. BBO: 6*6*1.5 mm. Filter: short-pass filter. D: motorized delay stage. LDM: long-pass dichroic mirror with 490 nm cut-on wavelength. The inset shows a photograph of the filamentation.

Download Full Size | PDF

CARS signals in air are investigated by the pump-probe setup shown in Fig. 1(b). The initial laser pulse is split by the 90:10 beamsplitter (BS) mentioned above, and the transmitted 10% of the pulse energy is used for second harmonic generation (515 nm) to provide the probe pulse. A half-wave plate and a cubic polarizing beam splitter is used to control the power of the probe pulse in front of a 1.5 mm BBO crystal. The probe pulse passes through a color filter, which filters out the fundamental wavelength at 1030 nm, and is delayed by a motorized translation stage. The probe beam is focused with an f = 1 m lens at a small angle (θ ∼ 1°) with respect to the pump pulse. To optimize the spatial overlap of the pump and probe pulses, the focusing lens is placed on a 20 cm long translation stage. Downstream from the collimating lens (f = 1 m), the CARS signal is reflected by two long-pass dichroic mirrors (DMLP490, Thorlabs) and coupled into the spectrometer with an f = 300 mm cylindrical lens. To investigate the influence of dispersion on the CARS signal, we evaluate fused silica and BK7, from which typical windows of combustion cells are made. The windows have a thickness of 1 mm and 12 mm, respectively, and an anti-reflection coating in the spectral range of 650–1050 nm (“B-coating”, Thorlabs). The windows are placed 17 cm downstream from a 3 m focusing lens.

3. Results

3.1 Spectral analysis of filamentation

Figure 2(a) displays the spectra of the pump (λ = 1030 nm, red line) and probe (λ = 515 nm, green line) pulses with Gaussian fits (dashed lines), which show a spectral width of 1.8 nm and 1.3 nm, respectively. For an f = 3 m lens, the spectral broadening induced by filamentation at different pulse energies is shown Fig. 2(b). The largest spectral width of ∼280 nm with visible spectral intensities between ∼850 nm and 1130 nm occurs at a pulse energy of 34 mJ, corresponding to ∼2900 cm^-1. Similarly, these widths are determined for an f = 2 m and f = 4 m lens (Fig. 3). For f = 2 m, the spectral broadening is negligible even for the maximum pulse energy of 50 mJ. For f = 3 m and f = 4 m, the spectral broadening peaks at E_pulse = 34 mJ and is slightly larger for f = 4 m with spectral features between 840 nm and 1160 nm.

 

Fig. 2. (a) Initial spectra of the pump (red line) and probe (green line) pulses. The Gaussian fits (dashed lines) of the pump and probe spectra are centered at 1030 nm and 515 nm, respectively. (b) Spectra of the filaments generated with an f = 3 m focusing lens for different pulse energies. The spectral width peaks at a pulse energy of 34 mJ.

Download Full Size | PDF

 

Fig. 3. Spectra for pulse energies which show the largest spectral broadening at focal lengths of f = 2 m, 3 m, and 4 m.

Download Full Size | PDF

Furthermore, we investigated the number of filaments induced in a single shot. “Multi-filamentation” is caused by inhomogeneities of the beam profile, especially if the laser power is far above the critical power for self-focusing [17]. The laser peak power used in the present study (∼ 60 GW) is about one order of magnitude above the critical power in air (P_{crit, air} ∼ 5 - 10 GW) [18]. Figure 4 shows single-shot images of the white-light, which is induced by the filament on the metal plate. At low (E_pulse = 20 mJ, Fig. 4(a)) and maximum pulse energy (E_pulse = 40 mJ, Fig. 4(d)), a single filament is observed, whereas ∼ 3 filaments occur for pulse energies of 30–34 mJ (Fig. 4(b),(c)).

 

Fig. 4. Images of single shot filaments on a metal plate at various energies using an f = 3 m lens. E_pulse: (a) 20 mJ (b) 30 mJ (c) 34 mJ (d) 40 mJ.

Download Full Size | PDF

3.2 CARS-signals of N₂, O₂ and CH₄

Although the spectral broadening is maximum for an f = 4 m lens (Fig. 3), the CARS experiments described in the following were performed with f = 3 m due to the limited laboratory space, which is required for the additional optical components (Fig. 1(b)).

For pump-probe, two beam CARS, the spectrum induced by filamentation must cover the range of both the pump and the Stokes beam. Therefore, we investigated the CARS signals of air for different pump pulse energies, while the probe pulse at 515 nm was kept at E_pulse = 300 µJ. To avoid contributions from the non-resonant background, we adjusted the probe pulse delay to ∼ 1.5 ps. The CARS signals of O₂ and N₂ are starting to be detectable at comparable pulse energies (E_pulse ∼ 26 - 27 mJ, Fig. 5). Higher pulse energies increase the respective CARS signals, while both signals are blue shifted as shown in Fig. 5. The blue shift may be caused by amplified stimulated emission (ASE) and post-pulses [19], but a systematic analysis of this effect is beyond the scope of this study. The maximum CARS signal is observed at E_pulse = 34 mJ, which is consistent with the observed spectral broadening (Fig. 2(b)). The signal-to-noise ratio for N₂ is ∼30, which was determined by the ratio of the peak intensity and the average noise at negative probe pulse delay.

 

Fig. 5. CARS spectra of N₂ and O₂ in air at various energies with an f = 3 m lens.

Download Full Size | PDF

We further investigated whether the spectral broadening is sufficient to generate CARS signals in combustion relevant molecules with higher Raman shifts. Exemplarily, we chose CH₄ with a shift of ∼ 2917 cm^-1 compared to ∼ 2330 cm^-1 for N₂ [13]. The CH₄ flow was provided by a gas cell with open entrance and exit ports as described in a previous paper [20]. Figure 6 shows a clear CH₄ peak (E_pulse ∼ 34 mJ). The signal-to-noise ratio is lower than that for N₂ and O₂, as the Raman shift of CH₄ is at the edge of the broadened excitation spectrum (2900 cm^-1, Fig. 2(b)). Further theoretical analysis of the Raman excitation efficiency is required, especially for quantitative concentration measurements.

 

Fig. 6. CARS signal of CH₄ at ∼ 447 nm (∼ 2900 cm^-1) (E_pulse ∼ 34 mJ).

Download Full Size | PDF

The influence of optical windows on the CARS signals is shown in Fig. 7. For these experiments, the maximum pulse energy of the laser was used (E_pulse = 60 mJ). The delay between the pump and the probe pulse was zero. The black curve shows the CARS signals of air in the absence of windows. As expected, the CARS signals are less affected by the fused silica windows than by BK7. Yet even 12 mm of BK7 allow for the detection of N₂ and O₂, respectively.

 

Fig. 7. CARS signals in air without and with windows.

Download Full Size | PDF

4. Conclusion

In summary, we studied the spectral properties of filamentation induced by focusing ps-laser pulses at a central wavelength of 1 µm in air and their application for CARS. For an f = 3 m lens, the spectral broadening allowed detecting CARS signals of N₂, O₂ and CH₄. Maximum spectral broadening of 320 nm was observed for f = 4 m. The insertion of windows made of fused silica and BK7 does not suppress CARS signal generation in air. Future work will deal with a quantitative analysis of the CARS signal for thermometry and measurements of gas concentrations. Possible applications focus on diagnostics in combustion and gasification, but may also include atmospheric remote sensing; e.g., filamentation based on thin-disk laser technology has recently been applied to guide lightning in the atmosphere [21].