1 Introduction

Planar laser-induced fluorescence (PLIF) is one of the most widely used combustion diagnostic tools. The number of target species is extensive, and common ones include combustion intermediates. Hydroxyl (OH) is an attractive target species. It is relatively abundant in flames, and its concentration peaks near the flamefront. Furthermore, excitation wavelengths, λ ex ≈ 280–315 nm for the \(A^{ 2} \varSigma ^{ + } {-}X^{ 2} \varPi\)(v′ = 0, v″ = 0) and (1,0) bands, are conveniently generated with Nd:YAG-pumped dye lasers using efficient, long-lived red dyes. OH has thus been used as a flamefront marker, though with some ambiguity in complex flowfields. Methylidyne (CH) has also been a target species for PLIF-based combustion studies [1–5]. Its primary advantage over OH is that its distribution within a flame corresponds reasonably well to the region for peak heat release rate [6]. However, CH exists in low concentrations in hydrocarbon flames and thus is difficult to detect and susceptible to fluorescence interferences too. Initially, combustion researchers employed the \(A^{ 2} \varDelta {-}X^{ 2} \varPi\)(0,0) band near 431 nm for excitation and detection. Carter et al. [7] first demonstrated the advantages of \(B^{ 2} \varSigma ^{ - } {-}X^{ 2} \varPi\)(0,0) excitation (near 390 nm), particularly for coupling with particle image velocimetry. The approach commonly used involves detection of A–X(1,1) emission—and rejection of the B–X emission—where the A(v′ = 1) state is populated via electronic energy transfer from B(v′ = 0) [8]. However, the A–X fluorescence represents only a fraction of the total (estimated by Li et al. [5] at 35 %), thus exacerbating the problem of low concentration.

More recently, OH has been the target species in studies employing kHz-repetition-rate (aka high-speed) lasers [9–11]. Researchers have explored the potential for applying kHz probing to CH as well. Sutton et al. [12] and Miller et al. [13] employed a pump laser delivering a burst of pulses and an optical parametric oscillator that provides frequency conversion and wavelength tuning. With a continuously pulsing laser—as opposed to a system delivering a burst of pulses—exciting CH is more challenging, owing to (1) the low pulse energy delivered by this class of laser system and (2) the wavelengths involved with CH A–X and B–X excitation (that are not easily generated with efficient, long-lived red dyes). Nonetheless, Johchi et al. [14] demonstrated that CH PLIF using B–X(0,0) excitation is possible with a continuously pulsing laser operating at 10 kHz; here, they pumped an Exciton Exalite dye with the third-harmonic beam from an EdgeWave Nd:YAG laser and obtained ~0.8 mJ/pulse at 390 nm.

CH can also be excited employing transitions from the \(C^{ 2} \varSigma ^{ + } {-}X^{ 2} \varPi\)(0, 0) band near 314 nm [15–17]. While the C state is strongly predissociated [18], the fluorescence yield is not significantly affected at atmospheric pressures and above, as electronic quenching should be the dominant decay mechanism (Q e,C > 2.5 × 108 s−1 ≥ Q predis), as noted by Jeffries et al. [16]. Moreover, the C–X(0,0) emission coefficient is large, ~9 × 106 s−1, as are the absorption coefficients; for example, for most Q-branch transitions, B 12 > 2.5 × 1010 m2 J−1s−1 [19]. As a result, even with a 314-nm pulse energy of 0.1 mJ and a laser sheet size of 75 mm × 0.3 mm, the excitation rate with an 8-ns laser pulse at atmospheric flame conditions is ~109 s−1, which should be comparable to a transition saturation value. Of course, OH \(A^{ 2} \varSigma ^{ + } {-}X^{ 2} \varPi\)(0,0) and (1,1) lines exist in this spectral region, and OH concentrations are generally much higher than CH concentrations. Thus, interference from OH may be a limiting factor in CH PLIF employing the \(C^{ 2} \varSigma ^{ + } {-}X^{ 2} \varPi\)(0,0) band. Here, we show that this effect can be minimized with careful tuning of the laser wavelength. Furthermore, this coincidence of OH and CH bands might enable simultaneous OH and CH measurement [16] and/or easy tuning between OH and CH. Interference from polycyclic aromatic hydrocarbons (PAHs) can also compete with CH fluorescence in nonpremixed flames; however, PAH LIF is expected to be weaker than found with A–X or B–X excitation, owing to the strength of C–X transitions and the low laser excitation energy concomitant with continuously pulsing kHz-rate lasers.

2 Experimental

The burner consisted of a central ~5-mm ID tube, from which issued chemically pure CH4 or a CH4–air mixture, and a 150-mm air coflow. For all measurements, the coflow velocity was \(\bar{V}_{\text{CF}} \, \approx \,0. 2 6 \,{\text{m}}/{\text{s}}\), based on a flowrate of 250 standard liters/min (SLPM, based on 273.15 K and 101.325 kPa). The source of both air streams was compressed air that was filtered for particles. For the stable laminar flame, CH4 and air flow rates were set to produce a flame with equivalence ratio ϕ = 1.4 (and respective fuel and air flow rates of 0.352 and 2.41 SLPM). For the turbulent flame, CH4 issued from the tube at 20 SLPM ( \(\bar{V}_{\text{Jet}} \, \approx \,1 9\,{\text{m}}/{\text{s}}\)), and the flame stabilized 40–50 mm downstream of the tube exit.

Initially, we conducted CH PLIF measurements using a 10-Hz-based laser system (Spectra-Physics GCR-170 Nd:YAG laser and Lumonics HD300 dye laser) operating with DCM laser dye. The 314-nm laser beam was expanded into a uniform 48-mm-high sheet that was focused over the flame. We detected fluorescence with an ICCD camera (Roper PIMAX-3) fitted with a 45-mm focal length, f/1.8 UV lens (Cerco); the field of view (FOV) was ~50 mm × 50 mm. Here, pixels were binned 4 × 4 before readout. For the excitation scan (described below), a bandpass filter, 311 nm < λ trans < 360 nm (Semrock FF01-334/40-50) was used. Because this filter rejects fluorescence below 311 nm, where many of the OH A–X(0,0) lines reside, it improves the CH–OH contrast; unfortunately, it also reduces the CH signal by about half due to “low” transmission for the 314–315 nm region where the strong CH C–X (0,0) Q-branch lines reside.

The diode-pumped, Q-switched Nd:YAG laser (EdgeWave Innoslab IS12II-E) operating at 10 kHz produced up to 62 W (6.2 mJ/pulse) of 532-nm radiation. This laser pumped a dye laser (Sirah CREDO) designed for high-frequency operation. With DCM dissolved in ethanol, the dye laser produced ~14 W (1.4 mJ/pulse) at 628 nm, near the peak of the efficiency curve. An integrated frequency-control unit houses a BBO crystal for frequency doubling to 314 nm; based on CREDO specifications, a linewidth of ~0.1 cm−1 is expected for the 314-nm beam. We measured the pulse duration to be ~7 ns (full width at half maximum height, FWHM). The pulse energy was as high as E p = 0.22 mJ at 10 kHz (with 6.2 mJ pump energy from the Nd:YAG laser), but for measurements shown here, E p was between 0.14 and 0.19 mJ. Dye lifetime was >300 Wh/L or equivalently 15 h based on 60 W of pump power and 3 L of dye solution (in both oscillator and amplifier reservoirs).

We formed the 314-nm beam from the kHz laser into a well-expanded sheet, about 44-mm (height, limited by the lens-holder aperture) by 0.3 mm (width, measured by traversing a sharp edge through the sheet). For some measurements, we employed a cylindrical mirror (coated for good reflectivity at 314 nm) for a simple retro-reflection of the excitation sheet. Like with the 10-Hz measurements, we maintained the polarization of the 314-nm beam at horizontal to minimize Rayleigh scattering interference. We detected fluorescence using a two-stage intensifier (LaVision HS-IRO, gated to 300 ns) that was equipped with the Cerco lens and coupled to a CMOS camera (Photron SA-5) having 8 GB of on-board memory. The image size was 896 pixels × 848 pixels and the corresponding FOV ~ 41.6 (H) mm × 39.4 (V) mm. We used Photron’s FASTCAM software to control the camera and send timing pulses to the laser and intensifier controller. Before the cylindrical lens, we placed a fused silica window, to reflect a small portion of the beam that was then sampled by a fast photodiode; this pulse and timing signals were monitored with an oscilloscope.

3 Results and discussion

In Fig. 1, we show an excitation scan (314.1–315.4 nm) from the rich laminar flame, acquired with the 10-Hz laser system with E p  ≈ 0.18 mJ/pulse, and the computed CH and OH excitation spectra from LIFBASE [19]. The ICCD camera imaged the fluorescence during the excitation scan; to construct the spectrum, we used the average signal from a region of interest (ROI) that lay along the CH layer, to reduce the interference from OH LIF. While this is effective, strong OH lines are still apparent, particularly those from the P-branch of the A–X(0,0) band. The spectrum includes the CH C–X(0,0) Q-branch bandhead; the dominant feature is that composed of blended lines near the excitation wavelength λ ex = 314.423 nm (air value), primarily from the Q1(7.5), Q2(2.5), Q2(3.5) and Q2(4.5) lines, as indicated by the LIFBASE simulation. To the red side also lies the OH A–X(1,1) Q1(6.5) transition. The level of interference from this line depends on the ratio of CH to OH concentration, the exact λ ex, and whether the Semrock filter was used. Here, to better isolate the CH fluorescence, we used the filter. The level of interference is not large within the ROI at λ ex = 314.423 nm, but the peak signal from the OH A–X(1,1) Q1(6.5) line at 314.429 nm is comparable to that from the CH line at 314.423 nm but spatially separated from the CH layer.

Fig. 1
figure 1

Experimental (offset) and LIFBASE calculated excitation spectra for CH C–X and OH A–X systems

Full size image

Using the 10-Hz laser system with λ ex set to the blue side of the Q1(7.5) line [to 314.415 nm near the peak of the blended Q2(5.5)–Q2(1.5) line], we investigated the effect of laser energy (varied with neutral-density filters). In Fig. 2, we show signal profiles from ensemble-averaged images with E p = 0.18, 0.62, or 1.7 mJ; inset is the image with E p = 1.7 mJ (with the line for the profile indicated). Here, the Semrock filter was not used. While the CH fluorescence is strong with E p = 1.7 mJ, OH fluorescence (from the A–X(1,1) Q1(6.5) transition) is reasonably strong too, from the region bounded by the CH along the inner flame cone to the outer cone boundary. Due to transition saturation effects, the CH-signal dependence on E p is sub-linear, whereas the OH-signal dependence is linear, as is evident from comparison of the signal profiles. This effect is more significant in the lifted jet flame (at these energies) due to the increase in the ratio of OH and CH concentrations. The CH–OH contrast can be improved (by ~2×) with use of the Semrock filter, but as noted, this improvement comes at the cost of CH signal.

Fig. 2
figure 2

PLIF signal across the rich, laminar CH4–air flame for 0.18, 0.62, and 1.7 mJ/pulse. Inset Average image with 1.7 mJ/pulse; the white line shows the sample location

Full size image

Figure 3 shows a sample single-shot image and an associated line plot, from the same laminar flame, acquired with the 10-kHz laser system with E p  = 0.15 mJ/pulse, the retro-reflector in place, and λ ex set to ~314.423 nm. Image processing consisted only of flat-field normalization (dividing the PLIF image by that from a Coherent cold cathode light panel). Image quality is excellent due to the high signal-to-noise ratio (SNR), estimated here for peak CH signals to be ≥17 based on comparable signals and gain from the Coherent light panel. The line plot shows a well-defined CH layer with a thickness (FWHM) of ~7 pixels = 0.32 mm, but due to the finite laser sheet thickness and other resolution limits, we expect the true CH layer thickness to be <0.32 mm. OH fluorescence is visible too but is relatively weak under these conditions.

Fig. 3
figure 3

Single-shot CH PLIF image from the rich, laminar CH4–air flame and relative signal profile (sample location at indicated white line)

Full size image

Lifted jet flames have been studied extensively [20], but only a few studies have employed kHz-rate PLIF [21, 22]. The lifted jet flame exists in a relatively simple flowfield but is affected by important phenomena such as local extinction, finite-rate chemistry, and turbulence-flame interactions. With regard to CH PLIF, the CH4-air lifted jet flame is challenging due to the wide range of observed signals along the flame surface (whereas in a stoichiometric CH4–air flame, the CH PLIF signals are much more uniform and reasonably strong too, as shown in Ref. [4]). Indeed, in the lifted jet flame, signals range from near the detection limit to values with SNR ≈ 5–10. In Fig. 4, we show a single-shot image from a recording at 10 kHz. Here, we employed an E p of 0.19 mJ and the retro-reflector; in addition, we averaged signal across 2 × 2 pixel regions (after readout) to improve the image quality and subtracted an average background. We acquired images for a few λ ex settings to explore the issue of CH–OH contrast. The image shown in Fig. 4 was from a sequence judged to have the optimum λ ex that was near the peak of the Q2(5.5)–Q2(1.5) line. Weak OH LIF is visible on the air side of the CH layer, but details of the flamefront location via the CH distribution are clearly evident. In this frame, flame-base “hooks” can be seen (where the CH layer bends outward toward the air side) that have been associated with edge flame propagation [23]. In Fig. 5, we show a montage from a sub-region (11 mm × 16 mm), illustrating the apparent formation and closure of a flame hole (Δt = 2.9 ms), while the entire 500-frame sequence is provided as supplementary material. We note that the structure in the Fig. 5 sequence is similar to that previously associated with flame holes [24].

Fig. 4
figure 4

Single-shot CH PLIF image from a lifted jet flame. y is the distance from the fuel tube

Full size image
Fig. 5
figure 5

Sub-region montage of CH PLIF images from a lifted jet flame, near the stabilization point (left-hand side). The downstream location (in mm) is as shown

Full size image

Finally, we note that image quality could be improved by (1) reducing the intensifier gate to 100 ns and employing a UG-5 Schott-glass filter to reduce chemiluminescence interference; (2) optimizing λ ex to increase the CH LIF and CH–OH contrast; and (3) designing/employing a filter with a sharp leading edge near λ = 314.0 nm and high transmission for λ > 314.3 nm.

4 Summary and conclusions

We investigated the potential for kHz-rate planar laser-induced fluorescence (PLIF) of the CH radical using the \(C^{ 2} \varSigma ^{ + } {-}X^{ 2} \varPi\)(v′ = 0,v″ = 0) band and have shown that due to its strength and accessible wavelength (~314 nm), the C–X band is a good choice for CH LIF studies wherein suppression of background scattering is not required. The problem posed by interference from polycyclic aromatic hydrocarbons (PAHs) is reduced with C–X excitation and detection, due to the strength of the band; interference from OH A–X(0,0) and (1,1) LIF can be significant but is mitigated with judicious choice of excitation wavelength. Strong transition strength can lead to saturation and thus limit the optimum excitation energy, as the potential for OH and/or PAH interference increases once strong CH C–X saturation is achieved. Nonetheless, this work demonstrates the efficacy of the approach, using a Nd:YAG-dye laser system operating continuously at 10 kHz and producing <0.2 mJ/pulse, with measurements in a rich, laminar CH4–air flame (that show excellent image quality) and in a lifted jet flame. Measurements in the latter flowfield show the complexity of the flame’s lead edge and the dynamics of the flamefront, including instances of hole formation and closure.