1. Introduction

Soot is mainly formed through incomplete combustion of fossil fuels, and has a typical dimension of a few to hundreds of nanometers [1]. When soot nanoparticles are released into the atmosphere, they are one of the most harmful pollutants having detrimental effects on air quality and human health [1–10]; while when they are used as a material in life, they are of special significance for many applications, such as serving as a high absorbance material and an indispensable component of black paints [11,12]. In addition, soot nanoparticles located in the combustion field can emit bright light, and thus are conducive to radiative heat transfer [6,13]. However, despite of the demand on precise control of soot yield in various combustion environments, a complete understanding of formation and growth mechanisms of soot has not yet been reached so far, which originates from the complex formation and growth process of soot in flames that is composed of a large number of combustion intermediate species and chemical reaction steps [1,4,8]. Indeed, detailed information on the dynamical evolution and reaction paths of soot nanoparticles is still lacking at the current stage, even though a variety of theoretical kinetic models have been developed [8,13]. Experimentally, the investigations on the combustion diagnostics by the conventional laser-based methods, such as laser-induced incandescence, absorption/extinction spectroscopy and light scattering, are mainly focused on characterizing the volume fractions, mean sizes and spatial distributions of soot in flames [3,6], and very rarely reported techniques enable to probe real-time reactions of the intermediate chemical species in the course of the soot formation and growth with high temporal resolution [14].

Interestingly, recent advance in ultrafast laser technologies has enriched the investigation tools in combustion science [15–20]. It was demonstrated that femtosecond (fs) laser can serve as an igniter source for ultralow-energy-threshold ignition of lean fuels with 100% success rate, which benefits from its unique nonlinear self-channeling propagation property resulting in a line flame kernel and thus breaking through the bottleneck of inferior energy deposition and low thermodynamic temperature in fs-laser-induced plasma [21]. Moreover, fs laser can serve as an excitation source for combustion diagnostics through single-photon resonance or nonlinear multiphoton excitations to induce fluorescence emissions from combustion intermediates such as O, H, OH, NH and CH [15,18,20,22]. As for soot nanoparticles, because of their inert and non-luminescent properties, an in situ fs laser Rayleigh scattering spectroscopy was recently developed to map the concentration distribution of soot nanoparticles in flames [16]. More recently, a fs near-infrared (NIR) strong laser pump and ultraviolet (UV) scattering probe approach was proposed, in which the dynamics of soot nanoparticles triggered by the strong pump laser were monitored by the UV probe light through Rayleigh scattering at a fs time resolution [14]. It was shown that soot nanoparticles undergo an ultrafast swelling at about 100 fs and a relative slow shrinking process at about 1 ps. However, since the Rayleigh scattering is sensitive to the size of soot nanoparticles, which depends on the fuel species and flame conditions varying from several nanometers to sub-micrometers from upstream to downstream of the flame [1], it is thus anticipated that a multi-color probe light scattering spectroscopy can be developed to detail the dynamical information on different sizes of soot nanoparticles.

In the present study, we employ multi-color femtosecond laser pulses at the wavelengths of 267, 400 nm and 600 nm as the probe lights to investigate the dynamics of soot nanoparticles induced in situ in a n-pentanol diffusion flame by a strong femtosecond pulse at 800 nm. We find that the pump-laser-induced swelling and shrinking process of soot nanoparticles is strongly dependent on the size of the nanoparticles. By comparing the scattered probe signals along the flame axis in both the presence and absence of the pump laser, we reveal the distribution and dynamics of different sizes of soot nanoparticles in the flame from the inception to the burnout regions, and demonstrate that the two-dimensional spectroscopy in the temporal and spectral domains provides a much higher sensitivity for mapping the distributions of different sizes of soot nanoparticles than those obtained only with the spectral intensities of the probe light.

2. Experimental setup

The schematic diagram of the experimental setup used in this study is shown in Fig. 1. A fs Ti: Sapphire laser system (Spectra Physics, Spitfire ACE) was employed to produce 800-nm and 45-fs laser pulses at a repetition rate of 1 kHz. The laser output was split into two beams. One beam with a lower energy of ∼1 mJ was focused by a plano-convex fused silica lens with the focal length of f = 75 cm (Lens 1, Altechna) in a n-pentanol/air diffusion flame to generate a single plasma channel called filament (about 20 mm in length and 100 µm in diameter), which is a result of dynamical balance between the self-focusing and plasma defocusing effects [17,23]. In flames, the critical power for the self-focusing and the clamping intensity inside the filament core were determined to be ∼2.2 GW and ∼2.8 × 10¹³ W/cm², respectively [17]. The high-intensity filament beam was thus served as the pump to trigger the ultrafast dynamic of soot particles in flames. The other beam with a higher energy of ∼ 5 mJ was sent into an optical parametric amplifier (OPA, Light Conversion, TOPAS HE Prime) to generate fs probe pulses of different wavelengths at around 267 nm, 400 nm and 600 nm, respectively. Another plano-convex fused silica lens with the focal length of f = 50 cm (Lens 2, Altechna) was used to focus the probe beam. For the ∼267 nm and ∼400 nm laser pulses, a dichroic mirror (DM1, Altechna) with high reflectivity at ∼800 nm and high transmission at ∼267 and ∼400 nm was used; while for the ∼600 nm laser pulses, another dichroic mirror (DM2, Altechna) with high reflectivity at ∼800 nm and 70% transmission at ∼600 nm was used to combine the pump and probe pulses. The energy for all the three-wavelength probe pulses was set at ∼1 µJ. The temporal delay between the linearly polarized pump and probe pulses was controlled by a delay line inserted in the pump beam path. Since it is known that the intensity the scattering signal is proportional to sin²φ, where φ is the angle between the polarization direction of the incident laser and the collection direction [24], the polarization directions of the pump and probe laser pulses was set to be horizontal and vertical, respectively. In this case, the scattering signal from the self-generated third harmonic during the filamentation of the pump light can be minimized, but that from the probe light can be optimized in the collection direction [16].

 

Fig. 1. Schematic diagram of the experimental setup for the fs near-infrared laser pump and multi-color Rayleigh scattering probe measurements. HR: high reflective mirror; PMM: plano metallic mirror; NDF: neutral density filter; DM: dichroic mirror.

Download Full Size | PDF

A laminar diffusion n-pentanol/air flame with the height of 40 mm was produced using an ordinary alkanol burner, which was mounted on a one-dimensional translation stage to modify the interaction position of the laser pulses at different heights above the burner (HABs). The scattered probe light by soot nanoparticles in the flame was collected at a right angle to the laser propagation direction by a fused silica lens (Lens 3, 50.8 mm in diameter, f = 60 mm) with a 2f−2f imaging scheme, and then detected by a grating spectrometer (Andor Shamrock SR-303i) coupled with a gated-intensified charge-coupled device (ICCD, Andor iStar). The spectrum was dispersed by a 300 grooves/mm grating with the blazed wavelength at 500 nm. The entrance slit of the spectrometer was set at 100 µm. The gate width of the ICCD was opened for a period of Δt = 10 ns with a gate delay of t = −5 ns, where t =0 means the arrival time of the pump pulse at the interaction zone. For each measurement, the data was accumulated over 2000 laser shots to improve the signal-to-noise ratio.

3. Result and discussion

Figure 2 shows the scattering spectra of the probe light in the n-pentanol/air diffusion flame at HAB of 28 mm respectively for the three wavelengths at about (a) 267 nm, (b) 400 nm, and (c) 600 nm in the absence of the pump laser pulse. The scattering mechanism of the probe light is ascribed to the Rayleigh scattering of soot particles in the flame [16]. Based on the scattering spectra measured at different HABs, we plot in Fig. 2(d) the normalized scattering laser intensity as a function of HAB for 267 nm (rectangle), 400 nm (dot), and 600 nm (asterisk), from which it can be seen that the measured probe light signals at the three wavelengths show the same intensity variation trend, that is, they first increase and then decrease when the HAB becomes larger.

 

Fig. 2. The scattering spectra of probe lights at about (a) 267 nm, (b) 400 nm and (c) 600 nm measured at HAB= 28 mm; (d) the normalized intensities of the scattered signals at 267 nm (rectangle line), 400 nm (dot line), and 600 nm (asterisk line) as a function of HAB; (e) the intensity ratios of 267 nm/400 nm (dot), 267 nm/600 nm (open pentagon) and 400 nm/600 nm (open asterisk) as a function of HAB.

Download Full Size | PDF

According to the Rayleigh scattering mechanism, the intensity of the scattering signal can be expressed as [24–26],

To understand the intensity variations of the scattered signals shown in Fig. 2(d), we first recall the formation and evolution mechanisms of soot in the flame. At the initiation stage of combustion, hydrocarbon fuels decompose into a variety of small radicals, such as OH, CH, and CH₂, which undergo polymerization reactions to generate larger sizes of hydrocarbon radicals, and under fuel-rich conditions, form polycyclic aromatic hydrocarbons (PAHs) known as the precursor of soot. The sizes of the radicals and the PAHs are normally at sub-nanometer scale [1,27,28]. PAHs further grow up through more reactions and nucleation into incipient soot particles with the characteristic size of a few nanometers [1,29–31]. Subsequent surface growth of soot through reactions involving many molecules such as acetylene and coagulation results in larger primary soot particles with the size in the range of ∼10–50 nm [5,30,32]. As soot matures, the aggregation of a collection of the primary particles finally results in the formation of the fractal soot with much larger sizes [1,33–35].

Clearly, the results in Fig. 2(d) show that at the inception phase of combustion (HABs ≤ 16 mm), the small radicals and PAHs dominate, which do not cause efficient scattering, resulting in the very weak scattering light intensities. As the HAB increases from ∼16 mm to 22 mm, the incipient soot particles begin to form, and weak scattered signals appear for the probe pulses at 267 nm and 400 nm; while these particles are still too small to efficiently scatter the probe light at 600 nm. With the HAB further increases, the primary soot is formed through the surface growth with the size being sensitive to all the three probe light wavelengths. In this process, both the volume fraction and the mean size of soot particles increase. In addition, during the soot evolution from formation to maturity, the refractive index of soot particles (n₁) increases [36,37], so that the scattering signals from all the three probe pulses are significantly enhanced. With soot maturing and aggregation in higher HABs, the soot particles increase to nearly 100 nm or even more [34,35], which exceed the Rayleigh-scattering sensitive region of the probe pulses, giving rise to a decrease in the scattering signals as observed in Fig. 2(d), where the intensity of the scattered probe light with shorter wavelength first decreases when HAB further increases. It should be pointed out that the oxidation or sublimation of soot particles can be neglected because the thermal temperature of the flame does not exceed 1100 K under our experimental conditions [38,39]. These results clearly indicate, according to Eq. (1), that the probe laser with longer wavelength is sensitive to the soot particle with larger size. Consequently, the ratios of the scattered light intensities of these three probe pulses at different HABs will reflect the relative concentration distribution of soot particles with different sizes. As a result, in Fig. 2(e), we plot the calculated ratios of the scattered signal intensities for 267 nm/400 nm (dot) and 267 nm/600 nm (open pentagon) and 400 nm/600 nm (open asterisk), which clearly reveal that at the initiation of soot formation in the flame, the small-size soot dominates, and with the increase of HABs, the proportion of large-size soot particles increases. Therefore, the results in Fig. 2(e) intuitively show that the relative concentration distribution of soot particles with different sizes along the flame axis.

We then measured the intensity variations of the probe light scattered by soot particles in the n-pentanol diffusion flames under the excitation of a near-infrared fs pump laser. In Figs. 3(a-c), we show the spectra for the probe pulses (solid line) at (a) 267 nm, (b) 400 nm and (c) 600 nm under the conditions of a delay time of Δt = 500 fs between the pump and probe light and HAB = 28 mm. For comparison, the spectra induced only by the pump pulses (dashed line) are also shown, in which the filament-induced nonlinear fluorescence emissions can be seen [15,20], but the scattering light from the self-generated third harmonic of the pump is almost unobservable. Note that the positive delay means that the pump pulse reaches the reaction zone in the flame earlier than the probe pulses. It can be seen from Figs. 2 and 3 that the scattering spectra of the probe lights are slightly modified when the pump light is introduced, which may result from the pump-laser-induced plasma in the flame that changes the spatial distribution of the medium’s refractive indices. Besides the strong scattered probe lights at the three wavelengths, two small spectral peaks at 248 nm and 468 nm can be observed, which are assigned to the 2s²2p³s $- $2s²2p² transition of atomic carbon C I [40] and the ${A^3}{\Pi _g} - {X^{\mathrm{^{\prime}}3}}{\Pi _u}$ transition of C₂ radical [41], respectively. These two peaks originate from the excitation of free carbon atoms and C₂ radicals in the flame by the pump laser [15,20]. Since the fluctuations of the pump-laser-induced emissions at 248 and 468 nm can reflect the instability of the flame, we remove the flame instability-induced fluctuation of the 267-nm scattering signals by dividing the emission signal intensity at 248 nm, and those of the 400-nm and 600-nm probe scattering signals by dividing the emission signal intensity at 468 nm, respectively, and then plot in Figs. 3(d–f) the intensities of the probe lights at (d) 267 nm, (e) 400 nm and (f) 600 nm as a function of the delay time Δt between the pump and probe pulses. Interestingly, it can be seen in Figs. 3(d–f) that for all the three cases the pump laser induces a sharp increase and then a slower decrease in the scattered signal intensity, in which the signals of the probe pulses at 267, 400 and 600 nm reach to their maximum at about 500 fs, 700 fs to 1.2 ps, and then return back to their original level at about 3, 5 and 8 ps, respectively.

 

Fig. 3. The measured spectra (solid line) with probe pulses at about (a) 267 nm, (b) 400 nm and (c) 600 nm under the conditions of Δt = 500 fs and HAB = 28 mm and the corresponding emission spectra (dashed line) induced only by the pump pulse; the intensities of the scattered (d) 267-nm, (e) 400-nm and (f) 600-nm probe pulses as a function of Δt.

Download Full Size | PDF

The sharp increase and subsequent decay of the probe light is ascribed to the pump-laser-induced ultrafast dynamics of soot particles that undergo an abrupt swelling and then a slow shrinking process in the flame by excluding all other possible mechanisms including soot fragmentation, the refractive index of soot particles (n₁), and the fractal structure that may change the scattering light signals [14]. For the swelling mechanism, we exclude the possibility of the laser heating contribution because the thermal temperature inside the filament is only about 1400 K [42], which is much smaller than that required for the heating-induced swelling of soot [43,44]. In this case, the swelling and shrinking process of soot can be explained as follows [14]: the positive charge generated by the multiphoton ionization process of the soot particles during the filamentation of the pump laser in the flame weakens the interlayer bonding between adjacent graphite layers in the crystalline shell structure of soot nanoparticles, thereby leading to the ultrafast stretching of the interlayer bonds and resulting in the abrupt swelling of soot nanoparticles; the subsequent electron recombination decreases the Coulomb repulsion inside the soot particles, and strengthens the inter-layer bonding of the graphite layers, giving rise to the shrinking of soot particles. In addition, the transfer of breathing vibration energy to other vibrational modes and the collisions in the electronic recombination process may promote the shrinking process. Therefore, for the strong dependence of the ultrafast dynamics of soot particles on the probe wavelengths, it can be ascribed to the fact that soot particles with larger size will experience more time to be swelled and shrunk. This is because when soot particles grow up, the maturation process is accompanied by the increase of the structural order, resulting in a longer graphite layer dimension and a decreased interlayer spacing [32,33,45,46], as well as more stretching vibration modes and interlayer C = C covalent bonds in soot particles.

Furthermore, we show in Fig. 4 the pump-probe measurement results for the HABs at 13 mm ((Fig. 4(a), (d), (g)), 25 mm (Fig. 4(b), (e), (h)) and 31 mm (Fig. 4(c), (f), (i)) for the probe wavelengths at Fig. 4(a–c) 267 nm, Fig. 4(d–f) 400 nm and Fig. 4(g–i) 600 nm. It can be seen in Fig. 4 that when HAB = 13 mm, the variations in the intensity of the scattered probe pulses at the three wavelengths are very small and almost unresolvable, which may reflect the fact that the concentration of soot particles at HAB =13 mm is very small. While when HAB =25 and 31 mm, the pump-probe results are similar to those obtained when HAB = 28 mm, that is, the scattering signal intensities rapidly increase, and then slowly decrease.

 

Fig. 4. The scattering intensities of the probe laser pulses at about (a-c) 267 nm, (d-f) 400 nm and (g-i) 600 nm as a function of Δt for HABs at (a, d, g) 13 mm, (b, e, h) 25 mm and (c, f, i) 31 mm.

Download Full Size | PDF

To further analyze the size dependence of the dynamical swelling and shrinking processes of soot nanoparticles, we present the rise and decay times as a function of the wavelength of the probe light in Figs. 5(a) and 5(b), respectively. The rise time is characterized by the interval time between the time position where the scattering signal starts to rise and the time position where the maximum signal is reached. The decay time is characterized by fitting the pump-probe data in Figs. 3 and 4 to an exponential function (see the solid lines). It can be clearly seen in Figs. 5(a) and 5(b) that at the same HAB position, both the rise time and decay time increase as the wavelength becomes longer, which reflects that the larger soot particle, corresponding to the probe light with larger wavelength, needs more time to swell and shrink. On the other hand, it can be observed that for the probe light at a certain value, although the rise time and decay time have very similar values, they are slightly increased when the HAB increases. This may be due to the fact that the mean size of soot particles that is sensed by the same wavelength probe light slightly increases when the HAB increases, since the concentration of soot nanoparticles with the same size at the Rayleigh scattering sensitivity range of a ≤ λ/10 [26] is related to HAB.

 

Fig. 5. (a) The rise time, (b) the decay time, and (c) the size enhancement as a function of the wavelength of the probe light measured at the HABs of 25 mm (dot), 28 mm (open rectangle) and 31 mm (open triangle).

Download Full Size | PDF

Furthermore, according to Eq. (1), the Rayleigh scattering intensity is proportional to a⁶, and thus it is possible to estimate the size variation of soot particles based on the signal enhancement results obtained in Figs. 3 and 4. As a result, we find that under the three wavelength probe conditions the sizes of the soot nanoparticles are respectively increased by 48.3%, 65%, and 77.4% at HAB =25 mm, 50.5%, 61.4%, 72.8% at HAB = 28 mm and 62.2%, 77.2%, 80.2% at HAB = 31 mm, as shown in Fig. 5(c). Since the ultrafast swelling of soot particles is ascribed to the laser-induced bond weakening and subsequent stretching between the adjacent graphite crystalline shells of soot particles and it is also known that at a certain HAB position of a laminar diffusion flame, soot particles having different sizes will normally distribute with different particle densities [1], the results in Fig. 5(c) reflect that the variation in the soot size is mainly determined by the initial size of the soot particles characterized by the probe light wavelength, but not by the HAB positions.

4. Summary

In summary, we have proposed a fs time-resolution pump-probe approach to investigate the soot dynamics in situ in a n-pentanol/air laminar diffusion flame, in which a strong near-infrared laser pump and multi-color Rayleigh scattering probe pulses were employed. By monitoring the intensity variations of the scattered probe laser pulses at the three wavelengths of 267 nm, 400 nm and 600 nm, we have demonstrated that the strong pump laser can induce ultrafast swelling and shrinking dynamical processes that are strongly dependent on the initial size of soot nanoparticles. That is, the swelling rise time and the shrinking decay time of soot nanoparticles monotonically increase when the initial sizes of soot nanoparticles become larger. By comparing the scattered probe signals along the flame axis with and without the pump laser, we have found that the two-dimensional spectroscopy in the temporal and spectral domains can better reveal the distribution and dynamics of different sizes of soot nanoparticles in the flame from the inception to the burnout regions. Our results provide a viable method for online and in situ diagnostics of soot dynamics and may hold promising applications for revealing the temporal evolution of a variety of nanoparticles in combustion fields.