Simultaneous identification of multi-combustion-intermediates of alkanol-air flames by femtosecond filament excitation for combustion sensing

Laser filamentation produced by the propagation of intense laser pulses in flames is opening up new possibility in application to combustion diagnostics that can provide useful information on understanding combustion processes, enhancing combustion efficiency and reducing pollutant products. Here we present simultaneous identification of multiple combustion intermediates by femtosecond filament excitation for five alkanol-air flames fueled by methanol, ethanol, n-propanol, n-butanol, and n-pentanol. We experimentally demonstrate that the intensities of filament-induced photoemission signals from the combustion intermediates C, C2, CH, CN increase with the increasing number of carbons in the fuel molecules, and the signal ratios between the intermediates (CH/C, CH/C2, CN/C, CH/C2, CN/CH) are different for different alkanol combustion flames. Our observation provides a way for sensing multiple combustion components by femtosecond filament excitation in various combustion conditions that strongly depend on the fuel species.

atomic C and H in an ethanol-air flame could be simultaneously probed, which provides the possibility of simultaneous monitoring of multiple combustion intermediate species 14 . Since then, based on the FINS technique, new phenomena and new effects in the flame filament have been explored. For example, by comparing FINS spectrum with those obtained from ns-LIBS and the combustion emission itself in the ethanol-air flame, it was demonstrated that the fingerprint fluorescence produced in a flame filament mainly come from the excitation of intermediate species existing in the combustion flame, but not from the fragments generated by the dissociation of parent molecules by the intense femtosecond laser field 15 . It was also discovered that fingerprint emission from the specific species of CN in an ethanol-air flame array can be amplified through amplified spontaneous emission (ASE) by observing CN fluorescence in a backward direction of the laser propagation as a function of the plasma length 16 . This lasing action was suggested to be a potential method that can overcome fluorescence quenching effect and improve the signal-to-noise ratio especially for the high-temperature and high-pressure engine combustion environments. In particular, it was revealed recently that the critical power and clamping intensity in flames are much smaller than those in air 17 , which provides new insights into the understanding of interaction of combustion flames with femtosecond laser filamentation.
However, up to now all the investigations on combustion in the flame filament have been focused on the ethanol-air flame. Since in the combustion, the fuel undergoes complex decomposition process, produces plenty of rich radicals and induces a large number of combined reactions, reliable analysis of combustion processes becomes more difficult when the fuel molecules become larger 18,19 . In addition, the interaction processes of laser filamentation with combustion flames of different fuels may also be different. For example, the analysis of the LIBS spectrum confirms that the laser breakdown threshold in flames with fuel molecule consisting of more carbon atoms is noticeably different from that in flames with fuel molecule consisting of less carbon atoms 20 . In particular, femtosecond laser filamentation is a highly nonlinear process, and thus its properties such as the clamped laser intensity are strongly dependent on the working environments, that is, the fuels used in the combustion conditions in our current study. Therefore, the investigation regarding the fuel effect on the functionality of FINS in combustion diagnostics is anticipated. In the present study, we systematically investigate a series of fuel-air flames (i.e., methanol, ethanol, n-propanol, n-butanol, and n-pentanol) using the FINS technique. Analysis of the FINS spectra demonstrates that the fluorescence signals of the intermediates and their ratios depends strongly on the number of carbon atoms in the fuel molecule at different fuel-air flames. Even so, our results demonstrate that the FINS technique can be used for sensing combustion intermediates of different combustion conditions, which is of significance for rationalizing the combustion reaction dynamics, and shed more light on the understanding of the multi-component combustion diagnostics.

Results and Discussion
Fingerprint emissions induced by filamentation from the n-pentanol-air flame. Figure 1 shows a filament-induced spectrum of the n-pentanol (C 5 H 12 O)-air flame in the spectral range of 240-660 nm. The measurement was performed with the filament formed at a distance of 17 mm above the burner wick. The ICCD gate width and delay were set to Δ t = 210 ns and t = − 5 ns, respectively (note that the laser pulse arriving time at the interaction zone is t = 0 ns). All the results shown in this work were accumulated over 300 laser shots (for experimental details, see Methods). In addition, the measurements of the filament-induced spectra show good repetitiveness due to the laminar nature of the alcohol-air combustion flame. As shown in Fig. 1, the spectral bands can be assigned to the combustion intermediates of CN, CH, OH, NH, C and H. The spectral bands at 337 nm resulting from the nitrogen molecule N 2 in air can also be observed 8 . The main spectral feature shown in Fig. 1 is similar to that in the FINS spectrum of the ethanol-air flame obtained previously in ref. [14] although n-pentanol molecule (CH 3 (CH 2 ) 4 OH) is much larger than ethanol molecule (CH 3 CH 2 OH). This observation clearly indicates that the combustion reactions of the two different fuels produce identical combustion intermediate species, which are determined by the compositions (C, H and O) of the two fuels.
It should be pointed out that the FINS signals recorded with the employed ICCD gate width and time delay integrated all the fluorescence in the time domain because the decay times of the combustion intermediates are typically in the range of 10-30 ns, as shown in Fig. 2. On the other hand, it can be seen from Fig. 2 that the decays are different for different combustion intermediates. Therefore, if the ICCD gate width was set too small to collect all the fluorescence in the time domain, the measured FINS signals from different combustion intermediates could be varied differently. It should also be emphasized that the fluorescence signals obtained in the FINS technique could not be directly used, similarly to LIF or LIBS 2 , to evaluate the concentrations of combustion intermediates due to the calibration difficulty in complex combustion environments, but they can be used to show qualitatively the relative concentration distribution of the combustion intermediates in flames.
In addition, we performed a measurement of the FINS signals with different pulse durations of the pump laser. It was found that the shorter the pulse duration becomes, the stronger the FINS signal becomes (not shown). Since the clamping intensity inside a gas filament becomes lower when the pulse duration increases 13 , the stronger signals with shorter pulse duration can be ascribed to the higher clamping intensity inside the filament. However, it should be pointed out that for a pulse with a given set of laser parameters, the effect of intensity clamping inside filaments would maintain a homogeneous interaction zone for the reaction. That is to say, the FINS signals from the same chemical species are produced by the interaction with essentially the same intensity.
FINS spectra produced from different alkanol-air flames. Shown in Fig. 3 are the FINS spectra measured from the combustion flames of five alkanol fuels, i.e., methanol (CH 3 OH), ethanol (CH 3 CH 2 OH), n-propanol (CH 3 (CH 2 ) 2 OH), n-butanol (CH 3 (CH 2 ) 3 OH), and n-pentanol (CH 3 (CH 2 ) 4 OH). In this measurement, all the FINS spectra were carried out with the filament formed at a distance of 17 mm above the burner wick. The insets in Fig. 3 are the combustion flame photos of the five fuels taken by a digital camera. It can be seen from Fig. 3    The FINS signals appear to become stronger when the carbon-carbon bond chain of the molecule becomes longer. It should be pointed out that the FINS signals from all the alkanol-air flames are much stronger than the emissions from the flames themselves with the laser off. As shown in Fig. 4a, when all the experimental conditions were kept the same as those in the FINS measurement except for blocking the laser, the measured signals of free radicals from n-pentanol-air flame itself are too weak to be observed. When the ICCD gate width was increased from 210 ns to 21 μ s, the measured emissions (Fig. 4b) from the free radicals of OH, CH and C 2 can be seen. However, the signals are still about one order of magnitude smaller than that shown in Fig. 1, and the continuum emission from the flame is dominant in the spectrum. This indicates that the signal from the n-pentanol-air flame itself is about three orders of magnitude weaker than that obtained from the FINS measurement. The much weaker signals from the flames themselves show that the population in the electronically excited states of the combustion intermediates in the flame determined by the Boltzmann distribution at typical temperature of 700-1000 K 17 in the alkanol-air flames is many orders of magnitude smaller than their population in the electronic ground states.  On the other hand, chemical reactions of a variety of intermediates and molecules during combustion in the flames could produce the intermediate of CN (for example through C 2 + N 2 = 2CN) 23 giving rise to the higher concentration of CN for larger fuel molecules. Furthermore, it can be seen from Fig. 5 that the slope for C 2 is much steeper than those for other intermediates as the number of carbon atom(s) in the fuel molecules changes from 1 to 3. For methanol (CH 3 OH), it can be undoubtedly concluded that the generation of C 2 is not due to decomposition of parent molecule since the carbon-carbon bond is absent in the parent molecule. Therefore, when the number of carbon atoms in the fuel molecules becomes larger, the decomposition of the molecules might make additional contribution to the product of C 2 , leading to the steeper slope. It is worth stressing that the CN and C 2 signal intensities in the n-pentanol-air flame are about one order of magnitude stronger than those in the methanol-air flame.
Identification of the C1-C5 flames. Since the signal intensities of the intermediates in Fig. 5 increase at different slopes, we thus check the possibility to discriminate different fuels by comparing the fluorescence intensity ratios of different combustion intermediates in the FINS spectra, which is a common method to demonstrate the differences of various fuel-air flames 22 . As a result, the calculated ratios of the signal intensities between different intermediate species are shown in Fig. 6. For clarity, we plot respectively the ratios of C 2 /C, CN/C, CH/C in Fig. 6a and those of CH/C 2 , CN/C 2 , CN/CH in Fig. 6b as a function of the number of the carbon(s) since the C signal is much weaker than other intermediates. It can be clearly seen from Fig. 5 that except for the methanol-air flame (dashed circles), the ratio values for other four fuel-air flames show certain dependences on the number of the carbon atoms in the fuel molecules. That is, the ratios of CN/C, CN/CH, CN/C 2 increase, but those of CH/C and CH/C 2 decrease as the number of the carbon atoms of fuels increases. Therefore, based on the difference in these ratio values, the four fuel-air flames can be easily distinguished. In addition, the ratios of C 2 /C show a large fluctuation when the fuels change from ethanol to n-pentanol, but they are much larger than that in the

Summary
We have systematically investigated the effect of fuels on the feasibility of FINS for sensing multiple intermediates of combustion flames with five types of fuels including methanol, ethanol, n-propanol, n-butanol, and n-pentanol. Comparison in the FINS spectra of different fuel-air flames demonstrates that the fluorescence intensities of the intermediates strongly depend on the number of carbons at different fuel-air flames. The fluorescence signals of all the four intermediates of C, C 2 , CH, CN increase as the number of carbons in the fuel molecules increases, but they show different slopes. The latter provides a way for the discrimination of the C1-C5 alkanol-air flames by comparing the differences of the signal ratios of the intermediates in different flames. Since the availability of high-power femtosecond laser system with high repetition rate of up to 10 kHz, our results reveal the possibility for high-speed monitoring of multiple combustion intermediates by means of femtosecond laser filament excitation.

Methods
The experiments were carried out with a 0.6 mJ/100 fs, 1 kHz Ti:sapphire laser system. The laser pulses were focused by a fused-silica lens of 200 mm focal length into the fuel-air flames on an alcohol burner to generate a single filament with the length of ∼ 1 cm. The burner was fixed on an X-Y-Z translation stage, which could control the interaction positions between femtosecond laser filament and the flames. The flames were surrounded by a top-open black box to avoid the wind from the laboratory air conditioner that may cause a strong swing of the flame. Characteristic fingerprint emissions from the flame filament were collected using a fused-silica lens (50.8 mm in diameter, 60 mm focal length) from the side of the laser propagation direction and then focused on the entrance slit of a spectrometer (Andor Shamrock SR-303i) coupled with a gated intensified charge coupled device (ICCD, Andor iStar) in a 2f-2f imaging scheme. The entrance slit width for the spectrometer was set to 100 μ m. The fluorescence was dispersed by a grating of 1200 grooves/mm (blazed wavelength at 500 nm) and detected by the ICCD camera.