Investigation of optical fibers for coherent anti-Stokes Raman scattering (CARS) spectroscopy in reacting flows
- First Online:
- Cite this article as:
- Hsu, P.S., Patnaik, A.K., Gord, J.R. et al. Exp Fluids (2010) 49: 969. doi:10.1007/s00348-010-0961-6
The objective of this work is to investigate the feasibility of intense laser-beam propagation through optical fibers for temperature and species concentration measurements in gas-phase reacting flows using coherent anti-Stokes Raman scattering (CARS) spectroscopy. In particular, damage thresholds of fibers, nonlinear effects during beam propagation, and beam quality at the output of the fibers are studied for the propagation of nanosecond (ns) and picosecond (ps) laser beams. It is observed that ps pulses are better suited for fiber-based nonlinear optical spectroscopic techniques, which generally depend on laser irradiance rather than fluence. A ps fiber-coupled CARS system using multimode step-index fibers is developed. Temperature measurements using this system are demonstrated in an atmospheric pressure, near-adiabatic laboratory flame. Proof-of-concept measurements show significant promise for fiber-based CARS spectroscopy in harsh combustion environments. Furthermore, since ps-CARS spectroscopy allows the suppression of non-resonant background, this technique could be utilized for improving the sensitivity and accuracy of CARS thermometry in high-pressure hydrocarbon-fueled combustors.
Various nonlinear spectroscopic techniques exist for measuring temperature, velocity, and chemical-species concentrations in gas-phase reacting flows (Eckbreth 1996; Gord et al. 2008). Among those techniques, broadband multiplex coherent anti-Stokes Raman scattering (CARS) has been proved to be the most accurate and promising method for measuring temperature and major-species concentrations under high-temperature and high-pressure conditions (Eckbreth 1996; Roy et al. 2010) because of its ability to acquire single-shot spectra of transient phenomena under unsteady flow conditions (Snelling et al. 1994; Hahn et al. 1997; Kuehner et al. 2003; Roy et al. 2010). One of the major requirements for this nonlinear technique is precise spatial and temporal superposition of three laser beams—pump, Stokes, and probe—at the probe volume to generate the CARS signal. Hence, performing state-of-the-art CARS measurements based on free-standing optics in harsh environments such as combustors and gas-turbine-engine test facilities poses significant challenges due to vibration, thermal transients, and unconditioned humidity associated with these environments. Optical fibers, because of their capability to provide flexibility in transmitting light to a remote location, can provide access to such probe volumes, which will simplify the application of CARS-based spectroscopy in harsh environments. Fiber-based CARS spectroscopy has several advantages in such environments: (1) reduced need of free-standing optics in the test-cell environment, (2) ease of alignment of multiple laser beams with flexibility when needed and ability to access non-windowed test sections, (3) isolation of the high-power laser system from harsh environments, and (4) safe, guided, and confined laser delivery.
Recently, several fiber-based CARS systems have been investigated for microscopy in the condensed phase using single-mode fibers (SMFs) (Légaré et al. 2006) and large-mode-area photonic crystal fibers (LMA-PCFs) (Wang et al. 2006). In addition, continuously wavelength-tunable, fiber-based laser light sources have been used to help avoid coupling losses for the CARS input beams (Andersen et al. 2007; Murugkar et al. 2007; Marangoni et al. 2009). In these studies, it has been shown that the pulse energy required for CARS in the condensed phase is considerably below the damage threshold of the fiber (Légaré et al. 2006). In gas-phase reacting flows, because of the lower molecular densities, the effective optical depth is reduced by four to five orders of magnitude, and the pulse energy required for CARS signal generation is approximately four orders of magnitude higher than that required for the condensed phase (Légaré et al. 2006; Wang et al. 2006; Meyer et al. 2007). This high energy requirement for CARS in the gas phase imposes significant constraints on fiber-based CARS because of the intrinsic optical damage threshold of the fibers. The enhanced higher-order nonlinear processes during propagation of intense laser beams through the fiber can also cause spectral broadening of the input laser beam and complicate the spectral analysis of CARS signal.
The CARS process is a special type of four-wave mixing where the pump and Stokes beams generate Raman coherence that is scattered off by the probe beam to obtain the CARS signal. Thus, the signal strength is proportional to the product of the intensities I ~ E/τA of each input laser beam, with E being the pulse energy, τ the pulse length, and A the cross-sectional area at the focal point. Therefore, higher pulse energy, shorter pulse duration, and smaller cross-sectional area of the beam are favorable for increasing the signal-to-noise ratio (SNR). The accuracy and sensitivity of the measurements are determined by the spectral resolution of the CARS system (Eckbreth 1996). For a fiber-based CARS system, the spectral resolution depends on the bandwidth of the fiber-delivered beams (Gord et al. 2009; Hsu et al. 2010). Thus, for acquiring a high-quality CARS signal, retention of the bandwidth of the input laser pulse during propagation through the fibers is an important criterion. Furthermore, sufficient spatial resolution is required for the fiber-based CARS system to achieve accurate “point” measurements of temperature. The spatial resolution of fiber-based CARS depends on the spot size of the focused beam in the probe volume, which is dependent on the quality of the input laser beam, and the geometry of the phase-matching condition such as collinear CARS or a BOXCARS configuration (Eckbreth 1996; Roy et al. 2010). To achieve an accurate point measurement of temperature in a reacting flow, a high-quality beam at the target end of the fiber is required.
Hence, the design and performance of a fiber-based CARS system for gas-phase thermometry is dependent on three vital parameters: (1) delivery of high-energy/irradiance CARS beams for reacting flows, (2) retention of the bandwidth of the input pulse during propagation through the fiber, and (3) delivery of high-quality laser beams at the probe volume. The amount of energy/irradiance that can be delivered in such a system is limited by the damage threshold of the fiber for varying pulse duration, the physical structure of the fiber, and the wavelength of the input laser beam (Wood 1986).
The objective of this study was to investigate the feasibility of delivering intense ns and ps laser pulses through various fibers for CARS spectroscopy in reacting flows. In particular, the optical damage threshold, the output beam quality, the nonlinearities inside the fiber, and the temporal distortion of the input laser pulses were studied in detail. Based on the results of the fiber studies, a proof-of-principle, ps laser-based fiber-coupled collinear CARS system employing multimode step-index fibers (MSIFs) was designed and demonstrated for thermometry of nitrogen (N2) in an atmospheric pressure, nearly adiabatic H2-air flame. It is understood that the spatial resolution of the CARS system using collinear geometry will be significantly lower than the BOXCARS geometry, and the temperature accuracy of the collinear CARS will also suffer due to the spatial averaging of cold and hot region of the reacting flows. The collinear phase-matching geometry was chosen just to explore the feasibility of fiber-based gas-phase CARS spectroscopy.
In Sect. 2 of this paper, the experimental methods used for characterizing fibers for CARS operation are described. In Sect. 3, the investigation of suitable pulse duration for delivering the energy/irradiance required for nonlinear CARS spectroscopy in reacting flows is described. In Sect. 4, the damage threshold and transmission characteristics of various fibers are addressed. In Sect. 5, demonstration and application of fiber-based ps-CARS for temperature measurements in laboratory flames are discussed, followed by a summary in Sect. 6.
2 Experimental methods for testing fiber transmission
The nonlinear effect of a fiber is determined by the spectral broadening of the input pulses after propagating through the fiber. Spectral broadening of the fiber-delivered beam was measured using a 0.25-m spectrometer (Acton Research Corporation, Model Spectrapro 275), equipped with a 1,200-grove/mm grating, and the spectrum was recorded using a back-illuminated, unintensified, 2,048 × 512-pixel-array CCD camera (Andor Technologies, Model DU 440BU). The overall dispersion was estimated to be ~1.5 cm−1/pixel. The fine structure of the power spectra is resolved using a high-resolution 1.25-m spectrometer (Jobin Yvon, Model SPEX 1250M) with a resolution of ~0.174 cm−1/pixel (~0.005 nm).
3 Results—suitable pulse regimes for fiber-based CARS: picosecond vs. nanosecond
The main challenge involving fiber-based CARS in gas-phase reacting flows is the transmission of high-irradiance laser beam, required for signal generation, without optically damaging that fiber. Most of the commercially available fibers are made of silica-based materials because of their superior flexibility and higher damage threshold. However, because of the high pulse-energy requirement in realizing fiber-based CARS in gas-phase flows, understanding the damage mechanism of silica fiber is essential for designing the ideal fiber-based CARS system. The temporal duration of the laser pulse is one of the most critical parameters that determines the threshold for laser-induced damage. The threshold irradiance/fluence for bulk silica has been extensively investigated over the pulse duration from tens of ns to a few femtoseconds (fs) (Wood 1986; Stuart et al. 1995; Du et al. 1996; Pronko et al. 1998; Tien et al. 1999). For a pulse duration < 10 ps, the damage mechanism is dominated by an avalanche breakdown process that is determined by the peak irradiance of the laser rather than the fluence (Stuart et al. 1995). The typical characteristic irradiance required for avalanche breakdown of bulk silica is greater than 80 GW/cm2 (Stuart et al. 1996). However, since the typical irradiance required for both ns- and ps-CARS is significantly less than the above-mentioned breakdown irradiance, the effect of the avalanche process on fiber damage is negligible in the design of a fiber-based CARS system. Stuart et al. and Shen et al. reported that for a laser pulse with a duration of 10 ps–10 ns, the fiber damage is dominated primarily by fluence-based lattice heating and thermal processes that involve heating of the conduction-band electrons by incident radiation followed by the transfer of energy to the lattice via electron–phonon interaction (Shen et al. 1989; Stuart et al. 1995, 1996). Since the typical thermal conduction rate (electron–phonon interaction) is ~10 fs (Pronko et al. 1998), Stuart et al. observed that there is sufficient time for the interaction of the energetic electrons with the lattice leading to melting, boiling, or fracturing of the lattice for a laser pulse with τ > 10 ps. The model for lattice heating and thermal processes predicts a τ1/2 dependence of the threshold fluence of bulk silica on pulse duration (Wood 1986), in good agreement with numerous experimental measurements (Wood 1986; Stuart et al. 1995; Du et al. 1996; Tien et al. 1999). A few recent studies based on 8-ns and 14-ps pulses reported by Smith and collaborators showed that the threshold irradiance of bulk silica is higher than the characteristic avalanche irradiance and that the corresponding threshold fluences for ps and ns pulses do not fit a square-root pulse-duration scaling rule (Smith and Do 2008; Smith et al. 2008). They reported that the observed higher damage-threshold fluence and different pulse-duration effect, when compared to those of the previous damage studies (Wood 1986; Stuart et al. 1995; Du et al. 1996; Tien et al. 1999), could result from the use of a small focal spot size and single-longitudinal-mode pulses to suppress the stimulated Brillouin scattering (SBS) and self-focusing effects (Smith and Do 2008; Smith et al. 2008, 2009). Based on an experimental and theoretical investigation, Smith et al. concluded that the bulk damage mechanism for a pulse duration of 14 ps–8 ns pulses in fused silica is dominated by irradiance rather than fluence (Smith and Do 2008; Smith et al. 2008). See Table 4 in the “Appendix” that lists a review of the damage thresholds reported in the literature for fused silica corresponding to mechanisms related to bulk damage, surface damage, and core–clad interface damage in the fibers in ns and ps regime.
Maximum energy transmitted through MSIFs using ps and ns pulses
Fiber diameter (μm)
Damage fluence (J/cm2)
Damage irradiance (GW/cm2)
For the purpose of estimating the CARS signal, let us assume that all three input lasers are operating at the maximum threshold intensity Imax that can propagate through the fiber without damage. Since the CARS signal strength is proportional to the product of intensities of pump, Stokes and probe pulses, it is reasonable to estimate the relative strength of the ps and ns fiber-based CARS signals, Sps-CARS and Sns-CARS, respectively, as being proportional to (Imax)3. Since Imax ~ (Eout/τA), then the measured data in Table 1 show that Sps-CARS is ~500 times higher than Sns-CARS for the same cross-sectional area, A, of the laser beams at the focal point. Therefore, a substantially higher CARS signal can be expected using lasers with pulse widths of the order of 150 ps. Thus, based on the results shown in Table 1, the delivery of ps pulse energy not only sufficiently meets the requirements for generating robust CARS signals but also exceeds the optimal energy required for reacting-flow measurements. Therefore, from the damage-threshold point of view, ps laser-based fiber delivery is more favorable than ns laser-based delivery for obtaining a large signal for a fiber-based CARS system.
Employing ps lasers for CARS also has the advantage of enabling non-resonant background suppression (NRB) with minimal loss in signal, which is important for fiber-based CARS systems that will be photon-limited. In a ns-laser-based CARS approach, the pump, Stokes, and probe beams overlap temporally to produce a significant NRB signal. The interference of the NRB signal with the resonant CARS signal is one of the major disadvantages that limit the applicability, sensitivity, and accuracy of ns-CARS at higher pressure–especially in hydrocarbon-rich environments (Meyer et al. 2007; Seeger et al. 2009; Roy et al. 2010). On the contrary, in the ps-CARS regime, it is possible to delay the probe beam temporally with respect to the pump and Stokes beams for suppressing the non-resonant contribution to the CARS signal, thereby improving the sensitivity and accuracy of CARS thermometry (Roy et al. 2005b, 2010; Meyer et al. 2007). Although beyond the scope of the current work, the ability to separate the probe pulse from the pump and Stokes pulses also allows ps-CARS to be used for studies of energy transfer processes by performing time-resolved measurements where the probe beam interacts with the coherence during various phases of its evolution (Roy et al. 2005b, 2010; Meyer et al. 2007; Seeger et al. 2009; Kulatilaka et al. 2010).
4 Results: characterization of various fibers for CARS
Ideally, a fiber-based CARS system needs to have sufficient energy/irradiance such that a CARS signal with reasonable SNR can be generated without fiber damage, with minimal beam profile distortion (i.e., with a smaller beam-quality factor M2). Therefore, the beams can be focused well at the probe volume, with minimal nonlinearity such that the bandwidth of the input laser does not change significantly, and with minimal dispersion of the beam so that the pulse duration is retained. Keeping these criteria in mind, the fiber characteristics for various ps and ns pulses through MSIFs were investigated in detail. For a 8-ns beam, the laser irradiance that could be delivered through a 1-mm-core-diameter silica fiber without damaging is ~0.2 GW/cm2, while it is ~3.3 GW/cm2 for a 100-ps laser beam. The corresponding energies are 12 and 4 mJ/pulse for ns and ps laser beams, respectively. Based on our experience and the results of other researchers involved in performing high-quality single-shot CARS measurement in a practical combustor, with a focal spot diameter of ~100 μm and an interaction length of ~1.5 mm, it was observed that the optimal energy for a 8- to 10-ns laser-based experiment is ~25 mJ/pulse and for a 100- to 120-ps laser-based experiment is ~0.5 mJ/pulse. Clearly, one needs to operate near the threshold of damage for fiber-based ns-CARS experiments, whereas the ps-CARS signal can be obtained well below the damage threshold of the fibers. Also, the advantages and disadvantages of using other fibers such as SMFs, fiber bundles of many SMFs, and state-of-the-art dispersion-compensated PCFs for a fiber-based CARS system were evaluated. Note that this study does not include comprehensive analysis of hollow-core fiber (HCF) because this type of fiber is very sensitive to bending losses; the bending losses of HCFs are ~2 dB/m at a bending radius of 30 cm (Robinson and Ilev 2004), which is approximately two to three orders of magnitude higher than those of solid-core fibers (Boechat et al. 1991). Since this investigation targeted production-level devices with limited optical access, these features were deemed problematic. However, with further advances in fiber-optic technology, evaluation of HCFs may be of significant future interest. Recently, Kriesel et al. investigated the feasibility of high-power laser-beam delivery through HCFs for CARS application (Kriesel et al. 2010). The transmission loss in HCF is ~1 dB/m (J. M. Kriesel, private communication) when compared to ~0.004 dB/m in our MSIF. However, the effect of bending on transmission loss in HCFs was not discussed in (Kriesel et al. 2010)
4.1 Investigation of damage threshold
Damage thresholds and beam quality M2 for various fibers (using 150-ps pulses at 532 nm)
Damage fluence (J/cm2)
Beam quality M2
340 ± 40
230 ± 15
190 ± 7
110 ± 5
60 ± 3
80 ± 2
30 ± 1
80 ± 2
100 ± 4
160 ± 10
4.2 Propagation-related characteristics of fibers
Since laser beam properties can be modified during propagation through a fiber, a few other parameters that are of importance for a fiber-based CARS system are as follows: quality of the output beam at the exit of the fiber, energy/irradiance stability of the delivered beam, spectral broadening due to possible nonlinearities experienced by the intense laser pulses, and possible temporal broadening of the pulses. All such propagation-related changes were studied for different types of fibers. However, as will be discussed in the next subsection, MSIFs showed the most potential and, hence, were chosen for the experiment to be discussed in Sect. 5.
4.2.1 MSIFs with large core
Currently, two types of widely available large-core commercial fibers have the potential to deliver the intense laser beams necessary for CARS measurements in reacting flows—large HCFs and large solid-core MSIFs. HCFs are made of a glass capillary tube with an internal reflective coating and suffer higher loss due to Fresnel reflection (Parry et al. 2006). However, the absence of a solid core enables a higher damage threshold. Because of the significant transmission loss and degradation in beam quality during propagation through the fiber, the typical operational length of HCFs is limited to <2 m (Stephens et al. 2005). Furthermore, HCFs cannot withstand significant bend radii since this further degrades beam transmission and quality. For the state-of-the-art coated HCFs (Matsuura et al. 2002), the bending loss at 0.3 m bending radius was 2.3 dB, which is two to three orders of magnitude larger than that for commercial MSIF (~0.005 dB) (Boechat et al. 1991; Kovacevic and Nikezic 2006). Thus, such fibers are not suitable for applications involving high-power laser-beam delivery over a long distance and within enclosed test sections with complex geometries. On the contrary, solid-core MSIFs have a high transmission efficiency of greater than 70% and maintain the same beam quality over the entire length of the fiber (if the fiber is not bent) even though they have a lower damage threshold than HCFs because of stronger laser-material interaction in the solid core (Parry et al. 2006). Since large solid-core MSIFs are readily available commercially and are suitable for robust operations in harsh environments, they are the primary focus of the current study on laser transmission.
22.214.171.124 Output laser-beam quality
126.96.36.199 Stability measurements
188.8.131.52 Spectral broadening
184.108.40.206 Temporal broadening
To summarize Sect. 4.2.1, the advantages of using MSIFs for delivery of ps laser pulses include a top-hat output beam profile, commercially available large-core MSIFs for transmission of high-irradiance laser beams, reduced instrumental noise, and minimal dispersion over long fiber lengths. The disadvantages include reduced spatial resolution due to poor beam quality (M2 ≫ 1), which limits the beam-focusing ability, and significant nonlinear effects for longer length fibers.
Continuing with the feasibility study for a fiber-based CARS system, the characteristics of several other types of fibers for high-power laser-beam propagation were investigated, and the results are briefly summarized in the following section.
4.2.2 Single-mode graded-index fiber (SMF)
4.2.3 Fiber bundles
Fiber bundles, which can be constructed by assembling a large group of SMFs, have been widely used for high-power laser-beam transmission and for multichannel imaging transmission (Anderson et al. 1996; Estevadeordal et al. 2005). In the present research effort, damage thresholds for various sizes of fiber bundles were investigated using 150-ps pulses at a visible wavelength of 532 nm. The results are included in Table 2. The fiber bundles (Myriad Fiber Imaging Tech, Inc.) can deliver >2.5 mJ/pulse with the smallest core size (FIGH-10-500N, a bundle of 10,000 SMFs with an estimated core size of 450 μm). The largest fiber bundle tested (FIGH-30-850N, a bundle of 30,000 SMFs with an estimated core size of 780 μm) can deliver >4 mJ/pulse. The beam profiles after transmission through the fiber bundles are shown in Fig. 7b. The output beam exhibits a diffused Gaussian beam profile due to leakage of light through the relatively thin cladding layer. It was observed that the small fiber bundles provide the highest output beam quality.
Fiber bundles are known to have excellent heat and radiation resistance. Other advantages observed in characterizing fiber bundles for fiber-based CARS include a higher damage threshold (typically three times higher than that of MSIFs). However, fiber bundles are likely to be unsuitable for fiber-based CARS because of a diffused Gaussian beam profile, low coupling efficiency, difficulty in coupling light uniformly into each SMF (Hand et al. 1999), and possible reduction in spatial resolution because of poor beam quality (M2 ≫ 1).
PCFs have attracted widespread attention because of their configurable dispersion properties when compared to those of standard fibers (Russell 2003, 2006; Zolla et al. 2005). In particular, high peak-power transmission (Borghesi et al. 1998; Shephard et al. 2004; Konorov et al. 2004), low optical loss (Smith et al. 2003; Roberts et al. 2005; Kristensen et al. 2008), and high-quality beam profiles (Shephard et al. 2006) are achievable with PCFs (Knight et al. 1998; Cregan et al. 1999; Russell 2003, 2006).
However, if hollow-core PCFs become more readily available, their advantages over standard MSIFs for fiber-based CARS would include higher damage threshold due to the photonic band-gap guiding mechanism, negligible nonlinear effects, manipulation of the zero-dispersion point, good spatial-mode filter, and low transmission loss over long distances (~2 dB/Km) (Roberts et al. 2005).
4.3 Summary of fiber characterizations
Multimode step-index fiber (MSIF)
Large core-size fiber commercially available
Top-hat beam profile
High coupling efficiency
Low transmission loss
Large nonlinear effect in fiber during high laser power delivery
Limited focusing ability with large core
Single-mode fiber (SMF)
Excellent spatial mode
Low transmission loss
Low pulse-energy delivery
Same nonlinear effect as MSIF
Higher damage threshold (3× MSIF)
Low transmission loss
Diffused beam profile
Difficult to couple uniformly to each SMF
Same nonlinear effect as MSIF
Limited focusing ability with large core
Photonic crystal fiber (PCF)
High damage threshold (5× MSIF)
Excellent spatial mode
Low transmission loss
Limited commercial availability
PCFs need to be specially designed for operation wavelength
Low pulse-energy delivery
5 Fiber-based ps-CARS experiments
The feasibility of delivering intense laser pulses through optical fibers for CARS spectroscopy was investigated. It was demonstrated that the propagation of ps laser pulses through large-core MSIFs allows transmission of sufficient laser energy for performing CARS thermometry in reacting flows. It was also determined experimentally that the use of ps pulses would provide significantly larger CARS signal without damaging the fiber. The transmission characteristics of ps pulses through MSIFs were investigated in great detail. Other fibers such as SMFs, fiber bundles, and state-of-the-art dispersion-compensated PCFs were also studied as potential candidates for a fiber-based CARS system. Based on the fiber characteristics studies, it was concluded that all-silica MSIFs are currently the most suitable fibers for fiber-based CARS. Although hollow-core PCFs appeared to have high potential for fiber-based nonlinear spectroscopy techniques, their major shortcomings are the significant cost associated with the production of each fiber for a specific wavelength.
A proof-of-principle, fiber-based ps N2 CARS system employing MSIFs has been developed and demonstrated for gas-phase thermometry in flames. It has been demonstrated experimentally that the temperatures measured using fiber-based CARS agree well with those from direct-beam CARS measurements, and the difference between the two cases is within ±2%. The proof-of-concept measurements show significant promise for extending the application of fiber-based CARS measurements to the harsh environments of combustors and engine test facilities. The use of ps-CARS also enables reduction in non-resonant background without significant loss in signal, which is important for maintaining sufficient SNR in the hot-band spectra for combustion thermometry. Future work includes measurements using the BOXCARS phase-matching geometry for improving spatial resolution and tests with longer fibers for characterizing the effects of spectral broadening.
The authors gratefully acknowledge useful discussions with Prof. Thomas Seeger of the University of Erlangen-Nuremberg, Prof. Margaret M. Murnane of the University of Colorado/JILA, and Dr. Hans Stauffer and Ms. Amy Lynch of the Air Force Research Laboratory. The authors thank Mr. Shuvro Roy for the help with the experimental setup. Funding for this research was provided by the Air Force Research Laboratory (AFRL) under Contract Nos. FA9101-09-C-0031 and FA8650-09-C-2001 and by the Air Force Office of Scientific Research (Dr. Julian Tishkoff and Dr. Tatjana Curcic, Program Managers).
This article is distributed under the terms of the Creative Commons Attribution Noncommercial License which permits any noncommercial use, distribution, and reproduction in any medium, provided the original author(s) and source are credited.