Laser-Induced Photofragmentation Fluorescence Imaging of Alkali Compounds in Flames

Laser-induced photofragmentation fluorescence has been investigated for the imaging of alkali compounds in premixed laminar methane–air flames. An ArF excimer laser, providing pulses of wavelength 193 nm, was used to photodissociate KCl, KOH, and NaCl molecules in the post-flame region and fluorescence from the excited atomic alkali fragment was detected. Fluorescence emission spectra showed distinct lines of the alkali atoms allowing for efficient background filtering. Temperature data from Rayleigh scattering measurements together with simulations of potassium chemistry presented in literature allowed for conclusions on the relative contributions of potassium species KOH and KCl to the detected signal. Experimental approaches for separate measurements of these components are discussed. Signal power dependence and calculated fractions of dissociated molecules indicate the saturation of the photolysis process, independent on absorption cross-section, under the experimental conditions. Quantitative KCl concentrations up to 30 parts per million (ppm) were evaluated from the fluorescence data and showed good agreement with results from ultraviolet absorption measurements. Detection limits for KCl photofragmentation fluorescence imaging of 0.5 and 1.0 ppm were determined for averaged and single-shot data, respectively. Moreover, simultaneous imaging of KCl and NaCl was demonstrated using a stereoscope with filters. The results indicate that the photofragmentation method can be employed for detailed studies of alkali chemistry in laboratory flames for validation of chemical kinetic mechanisms crucial for efficient biomass fuel utilization.


INTRODUCTION
Environmental issues and overall efforts to achieve a sustainable energy supply require power plant operators to increasingly consider utilization of a broader variety of biomass sources in the fuel mixture, such as herbaceous material (straw and grass), agricultural byproducts (pits, shells and hulls) and municipal waste. 1,2 However, many of these fuels result in the formation of alkali chlorides such as KCl and NaCl during combustion in power plant boilers. Alkali chlorides are key components for slagging and fouling and they also affect heatexchange surfaces. 3 In addition, chlorine exposure increases the risk of corrosion on furnace walls, super heaters, and economizers. 4 Problems related to alkali chlorides can be avoided using fuels with low content of alkali and chlorine, or suppressed by adding sulphur to the processeither by co-combustion with a sulphur-containing fuel or by injection of sulphur-containing additives. 5 Nevertheless, optimum reduction of alkali chlorides requires detailed understanding of the formation process and validation of mechanisms for alkali chemistry, which can be achieved by studies in laboratory flames under well-controlled conditions, see for example Li et al. 6 Potassium is the main alkali metal in many biomass fuels 7,8 and a major component in many of the problems outlined above, thus knowledge on potassium chemistry is highly relevant for proper combustion of biomass fuels.
Laser-based techniques provide non-intrusive probing, in many cases species-specific, with high temporal and spatial resolution for detailed studies of combustion processes. A comprehensive review by Monkhouse 9 summarizes the status up to 2011 of different techniques for online detection of alkali metals in flue gas. Alkali chlorides, such as KCl and NaCl, can be detected by means of photofragmentation fluorescence for which Oldenborg et al. 10 used highpower laser pulses in the ultraviolet regime for photodissociation of alkali compounds followed by detection of fluorescence from the excited atomic alkali fragment. Furthermore, Chadwick et al. 11 investigated photofragmentation fluorescence detection of NaCl and NaOH (sodium hydroxide) using an Excimer laser of wavelength 193 nm for photodissociation and subsequent measurement of Na fluorescence signals at 589 and 819 nm for detection of NaCl and NaOH, respectively. Further investigations of the photofragmentation fluorescence technique have included NaOH detection by multi-photon fragmentation using the 355 nm third-harmonic output from a Nd:YAG laser 12 and studies of collisional quenching of the alkali metal atom fragments. 13 The photofragmentation fluorescence technique has been applied for detection of alkali compounds in flue gas of Circulating Fluidized Bed (CFB) boilers. Investigations have been made for combustion of coal only [14][15][16] as well as for co-combustion with biomass with a high content of alkali. 17 Moreover, Erbel et al. used the technique in a study of biomass gasification. 18 Sorvajärvi  While previous studies include measurements in a single point or averaged along a line, also in practical combustion devices, spatially resolved imaging measurements are also of interest. Measurements of photofragmentation fluorescence emission spectra of KCl, KOH, and NaCl were made with the beam focused into a 50 mm sheet using a cylindrical lens of focal length f=300 mm. The fluorescence was collected into a spectrometer (Acton SP-150, grating 300 grooves/mm, Princeton Instruments) using an UV condenser, f=60 mm, and a long-pass filter (WG280, Schott) to suppress scattered laser light. The signals were detected with an intensified CCD camera (PI-MAX I, Princeton Instruments) connected to the spectrometer.
For imaging the detector was instead equipped with an f=50 mm objective (Nikkor f/1.4).
In measurements for quantitative analysis of the photofragmentation fluorescence signal, the 193 nm laser beam was focused using cylindrical lenses of focal lengths f=1000 mm and f=500 mm, which combined with an arrangement of razorblades resulted in a 20 mm vertical sheet.
Measurements of potassium species were made using a bandpass filter centred at 766 nm (50 mm dia., OD 4, FWHM 10 nm, Edmund Optics) for detection of K-atom fluorescence and suppression of scattered laser radiation. Simultaneous measurements of KCl and NaCl were made using a stereoscope (Lavision) mounted in front of the objective. Band-pass filters centred at wavelengths 766 and 589 nm (50 mm dia., OD 4, FWHM 10 nm, Edmund Optics) were inserted in the stereoscope for detection of K-and Na-atom fluorescence signals, respectively.
A burner originally made for atomic absorption spectroscopy (Perkin-Elmer), shown in the photo of Fig. 1b, was used for measurements in alkali-seeded methane-air flames. The burner consists of a spray chamber and a water-cooled head with a central compartment for the premixed fuel-air blend and an outer channel for a co-flow shielding the flame. The diameter of the inner compartment is 23 mm and the head is topped by a circular mesh plate (pore size ~1 mm and length 20 mm). The burner allows for stabilization of flat laminar premixed flames (cf. Fig. 1b). Alkali-seeding is achieved using a nebulizer fed with part of the supplied air, which extracts liquid KCl solution via a sample tube. The KCl solution from the nebulizer is pre-treated in order to have only finer aerosol droplets to pass through the chamber to the burner head whereas larger droplets are removed via a drain tube. Methane and auxiliary air are mixed together with the nebulizer air in the burner's spray chamber to get the total correct fuel-air equivalence ratio (Φ) of the mix.
A nitrogen co-flow of 10 l/min was supplied to the burner head to shield the flame. A steel cylinder (cf. Fig. 1b) was mounted 30 mm above the burner for flame stabilization required for quantitative signal analysis and to be able to make sequential photofragmentation and Rayleigh scattering measurements under steady-state conditions. The total gas flow of air and fuel to the Rayleigh scattering measurements. The 532 nm beam was aligned into the beam path using a dichroic mirror, cf. Fig. 1a, and further shaped into a laser sheet of 10 mm height using cylindrical lenses of focal lengths f=-40 mm, f=200 mm, and f=500 mm. A half-wave plate positioned in the beam path was adjusted to achieve vertical polarization for optimal Rayleigh scattering.
Absorption measurements. KCl absorption measurements were made in the flames using the experimental set-up shown in Fig. 1c. UV light from a high-intensity (150 W) UV light source (L1314, Hamamatsu) was radiated through an aperture and collimated using a 90° offaxis parabolic mirror coated with UV-enhanced aluminium and a reflective focal length of f=150 mm (diameter 50 mm, Thorlabs). The collimated UV light beam subsequently passed through another aperture and a plano-convex focusing quartz lens of focal length f=150 mm.
After the lens the UV beam passed over the burner top and was reflected five times using UVenhanced aluminium mirrors (diameter 25.4 mm, Thorlabs) before it was collected in an UVenhanced collimator (250 -450 nm, diameter 12 mm beam, SMA, Thorlabs). This construction resulted in a total absorption path length of 138 mm. The collected UV light was then transferred through an optical fibre (FC-UV600-0.5-SR, Azpect Photonics) and subsequently dispersed in a spectrometer (grating 2400 grooves/mm, slit width 50 µm, AVABENCH-75-2048, Azpect Photonics). KCl concentrations were evaluated from the collected spectra by a least-squares fit to a calibration spectrum measured at 860 °C following the procedure presented by Forsberg et al. 22 Fluorescence data evaluation. Alkali species concentrations have been evaluated from the photofragmentation fluorescence signal, F, which can be expressed in emitted photons by Eq. (1) where Ω is the detection solid angle, l the probe volume length, εF the detection efficiency for the fluorescence signal, and A the probe volume cross section area. The ratio Q A A fi fi  represents the fluorescence quantum yield where Afi is the Einstein coefficient for spontaneous emission, Q the collisional quenching rate, and T represents a factor accounting for fluorescence losses due to absorption, so-called trapping. The quantity N is the concentration of alkali atoms generated in the excited K-atom 4 2 P states by photofragmentation. The fluorescence is obtained from the 4 2 P transitions at 766 and 769 nm and the frequencies are ν=3.91·10 14 s -1 and ν=3.89·10 14 s -1 , respectively. Assuming that K-atom photofragments are distributed between both the 4 2 P states, the average of the coefficient for spontaneous emission for the two transitions can be employed in the evaluation, using values presented by Nandy et al. 23 give Afi=3.8·10 7 s -1 . Collisional quenching data has been presented by Jenkins 24 In Eq. (2) ERayleigh is the laser pulse energy for Rayleigh scattering measurements, h Planck's constant, νRayleigh the Rayleigh photon frequency,    the differential Rayleigh scattering cross section, εR the detection efficiency for the Rayleigh scattering signal, and Ntot the total gas number density in the probe volume. The cross section for air at ambient pressure and temperature for wavelength 532 nm is 6.25·10 -32 m 2 /sr. 25 Using the Rayleigh scattering signal for calibration provides the detection solid angle, Ω, and the probe volume length, l, in the concentration evaluation using Eq. (1). The detector quantum efficiencies are specified to 10% and 3.3% at 532 and 766 nm, respectively. These values together with the transmission, 85%, of the interference filter used for fluorescence detection have been employed for determination of εR and εF.
In addition, Rayleigh scattering can be utilized for flame temperature measurements by comparison of signals measured in flame and at ambient conditions, taking differences in cross sections into account. For the investigated stoichiometric flame, using Rayleigh scattering cross section data compiled by Zetterberg 26 and major species concentrations determined from chemical equilibrium calculations, the Rayleigh scattering cross section of the product gas in the post-flame region was found to be a factor of 1.1 higher than that of ambient air. Including this factor in the ratio between Rayleigh signals measured in flame and ambient air allowed for temperature measurements in the flame.
The relation between the concentrations of K atoms, N, and parent species, NKCl, is given by The factor Φ is the yield of photofragments generated from the parent species and is equal to 1 for the investigated case since dissociation of one KCl molecule results in creation of one K atom. Furthermore, σ is the absorption cross section, E the laser pulse energy, and νlaser the laser frequency. Equation (1)

RESULTS AND DISCUSSION
Prior to imaging experiments, fluorescence emission spectra were measured to identify potential interferences. Figure 2a shows photofragmentation fluorescence emission spectra measured in stoichiometric flames seeded with KCl (black) and NaCl (grey). Distinct spectral lines from atomic Na and K can be observed at wavelengths 589 and 766 nm, respectively. Katom fluorescence at 766 nm is also obtained for KOH-seeding as shown in a higher resolution in the spectrum of Fig. 2b, although the line is approximately eight times weaker compared with the case of KCl-seeding. In Fig. 2b the spectrum measured for NaCl seeding also shows a line from Na at 820 nm, also observed previously by Chadwick et al. 11 and mainly attributed to photofragmentation of NaOH formed during combustion. All spectra show fluorescence peaks at wavelengths 250 -350 nm, shown in Fig. 2c, mainly attributed to OH radicals generated from photodissociation of H2O. This could potentially interfere in imaging experiments, but can be removed with suitable filters transmitting the strong contributions of atomic Na and K.  Fig. 3, covering 13-20 mm above the burner, are plotted in Fig. 4 together with temperature profiles determined by Rayleigh scattering. The fluorescence signal is proportional to gas number density and the profiles have been compensated by multiplication with the temperature profiles in order to obtain profiles representing relative concentrations.
Profiles for the two KCl-seeding concentrations show apparent differences. The 0.5 M KClseeding result in maximum signal at the flame edges, cf. Fig. 3a and 4a, where two distinct peaks can be observed. In contrast the image and profile measured for 0.01 M KCl-seeding, cf.  This suggests KCl to be the major signal contribution in the image measured in the KCl-seeded flame for which the temperature is around 1500 K. In the KOH-seeded flame the lack of chlorine results in more K atoms available for formation of KOH, which becomes the major post-flame potassium compound.
The formation and distribution of KOH and KCl are temperature-dependent which also could affect the shapes of the profiles in Fig. 4. Formation of KCl is reportedly promoted at lower temperature while the formation of KOH is promoted at higher temperature. 6  Chadwick et al. 11 have shown that it is possible to discriminate between NaCl and NaOH in photofragmentation fluorescence measurements. For both species the photofragmentation process produces Na atoms in the excited 3 2 P state, resulting in emission at 589 nm. However, the dissociation energy of NaOH is lower and excess energy is therefore available for further excitation of the Na-atom fragment resulting in additional fluorescence emission for Na, for example at 820 nm as observed in Fig. 2b. As mentioned previously the situation is analogous for the potassium compounds where additional energy available after dissociation of KOH allows for excitation into the 5 2 S and 3 2 D states. 23  to a multiphoton process as the excess energy after photofragmentation is insufficient for OH excitation. 11 In our investigations of KOH, the fluorescence spectra do show OH lines (cf. Fig.   2c), however no difference were observed between spectra measured in KOH-seeded or unseeded flames. A contribution to the OH signal from a KOH multiphoton process is therefore probably much lower than the OH signal obtained from photodissociation of water in the postflame region.
The photofragmentation process is influenced by the laser fluence according to Eq. (3) and Davidovits and Brodhead. 28 Corresponding data for KOH is, however, not available in literature but following the discussion by Sorvajärvi et al. 21 it can be approximated with that of NaOH reported by Self and Plane 29 which gives a value of 5.6·10 -18 cm 2 at 300 K. The higher cross section of KCl results in a complete dissociation at lower fluence levels than for KOH, which shows a slower increase in the fraction of dissociated molecules. While these calculations indicate that KOH does not fully reach complete dissociation for the experimental conditions with laser pulse energies of around 40 mJ in a focused laser sheet resulting in a fluence of 800 mJ/cm 2 , the degree of dissociation is above 90% and both species are considered measured under saturated conditions. Detection limit determined for a signal-to-background ratio of 3 in KCl-seeded flames for images averaged over 300 pulses (circles) and single-shot images (squares).