Ammonia detection is highly relevant for combustion in boilers and furnaces since NH3 is able to suppress nitric oxide levels by catalytic as well as non-catalytic reduction. The mixing of ammonia with flue gases is an important parameter to obtain efficient non-catalytic reduction. In this paper picosecond DIAL was used for range-resolved, single ended, NH3 detection, utilizing a tunable picosecond laser source. The absorption spectrum of the A(ν2 = 1)←X(ν2 = 0) band was recorded and 212.2 and 214.5 nm was selected as the on- and off-resonance wavelength, respectively. One-dimensional concentration profiles with various NH3 concentration distributions are presented. The detection limit was found to be 40 ppm with a spatial resolution of 16 cm.
© 2012 OSA
Laser-based diagnostic methods for characterization of combustion processes enable non-invasive measurements of important parameters in situ with high spatial and temporal resolution . For example measurements of flow velocities, soot particle properties, and temperature in gas and on solid surfaces are all feasible. Of particular importance is the ability to monitor numerous species of interest, in many cases with imaging measurements. Applications of laser-based methods include studies in laboratory flames, internal combustion engines, and gas turbine burners. In many cases the measurement region typically has dimensions around 0.1 m and is located in a position providing proper optical access, allowing for detailed studies.
However, large-scale combustion devices with dimensions on the order of meters, such as power plant furnaces and boilers, require remote sensing, have limited optical access, and, thus, present a major challenge for laser-based techniques. Line-of-sight absorption measurements or methods with directional signal emission such as coherent anti-Stokes Raman Spectroscopy (CARS) can be carried out with relatively small optical access and have been applied for studies in power plant units [2–5]. Nevertheless, two optical ports are necessary for transmission of laser and signal, and little spatial information is obtained due to averaging along the optical path. Optical access through a single port only requires detection of signals back-scattered relative to the laser-beam propagation. This is utilized in the light detection and ranging (lidar) concept, in which pulsed lasers and time-resolved detection allow for measurements with spatial resolution along the laser beam path. Lidar is mainly used for remote sensing over very long distances (km), for example in studies of the atmosphere , and probing with nanosecond laser pulses provides sufficient spatial resolution, typically a few meters. This is however inadequate for measurements in combustion devices of size ~10 m and therefore a near-field lidar system based on a picosecond (ps) Nd:YAG laser has been developed  and applied for gas-phase thermometry [8, 9] as well as soot particle detection . Probing a specific species can be attained using differential absorption lidar (DIAL) based on absorption of the propagating laser beam and subsequent reduction in the back-scattered signal. Acetone detection using DIAL with the picosecond setup was demonstrated using the Nd:YAG laser harmonics at 266 and 532 nm . While the DIAL concept in principle worked, a tunable laser source allows more species to be probed and off-resonance measurement at wavelengths close to the probed absorption line.
In this work the ps-lidar setup has been extended with a tunable laser source and DIAL detection of ammonia (NH3) is demonstrated. Ammonia detection is highly relevant for combustion in boilers and furnaces since NH3 is able to suppress nitric oxide (NOx) levels by catalytic as well as non-catalytic reduction . The mixing of NH3 with flue gases has been identified as an important parameter to obtain efficient non-catalytic reduction [11, 12]. This is desirable to minimize NOx emissions as well as the amount of remaining NH3 since residual NH3 may be detrimental for the combustion device, for example by formation of ammonium sulfate deposits on surfaces . Clearly techniques for NH3 detection are of high interest for characterization of large-scale combustion units. Ammonia has been measured in power plant flue gases using differential optical absorption spectroscopy (DOAS) with NH3 excitation in the infrared regime around 10 µm [2–4]. However, measurements required dual ports for optical access and provided NH3 concentrations averaged over the optical path. Furthermore, single-ended NH3 detection has been demonstrated using a lidar system based on a CO2 laser, wavelength around 10 µm, for measurements of atmospheric NH3 levels averaged over a distance of kilometers .
This paper presents single-ended, spatially resolved NH3 detection in the ultraviolet regime with excitation in the A←X band at around 212 nm using the picosecond lidar system. At detection distance of a few meters, NH3 levels down to 40 ppm have been monitored, at room temperature and pressure, with a spatial resolution of 16 cm. Thus, the detection scheme shows strong potential for combustion diagnostics and further application considerations are readily discussed.
2.1 Lidar setup
The experimental setup is schematically outlined in Fig. 1 . The fundamental 1064-nm radiation of a mode-locked Nd:YAG laser (Ekspla, PL 2143C) is used to pump an external amplifier (Ekspla, APL 70), in turn supplying an OPG/OPA system (Ekspla PG401-P80-SH) with the third harmonic radiation (355 nm). The OPG/OPA system is tunable in the range 210-2300 nm with a specified linewidth < 4 cm−1. Typical pulse energies are 0.3 to 1mJ in the ultraviolet (UV) and near-infrared regimes, whereas 3-5 mJ can be achieved in the visible regime. Laser pulses of 80 ps duration are delivered at a repetition rate of 10 Hz, and the laser beam diameter is approximately 12 mm.
The laser beam is directed towards the measurement volume using a planar UV-enhanced protected aluminum mirror, M1, with reflectance specified to 72% at 210 nm. The backscattered light was collected using a Newtonian-style telescope having a concave 10-cm diameter primary mirror, M2, with focal length 45 cm and UV-enhanced aluminum coating. The primary mirror is placed on a computer controlled 30-cm translation stage, allowing adjustable position of the focal plane. The collected light is directed towards the detector using a planar mirror, M3, having similar coating as M2.
The signal was detected with a micro-channel-plate photomultiplier tube (MCP-PMT, Hamamatsu R5916U-50), which allows time-gated detection for background suppression. The detector response is characterized by rise and fall times of 170 and 730 ps, respectively. The PMT signal was acquired with a 3-GHz bandwidth digital oscilloscope (Lecroy Wavemaster 8300)
An aluminum cell with 4 cm diameter quartz windows, separated by 11.0 cm, was used for measurements of NH3 absorption spectrum. The laser beam was directed through the cell and the transmission was measured, using an energy meter (Gentec-EO, QE25LP-S-MB) averaging the pulse energy over 250 shots, before and after the cell. The spectrum was recorded by scanning the OPG/OPA wavelength from 211.0 to 215.0 nm in steps of 0.1 nm having the cell filled with a mixture containing 0.5% NH3 and 99.5% methane (CH4) at atmospheric pressure, i.e. an NH3 number density of 2.0 × 1016 cm−3, yielding a minimum transmission of 0.52. In order to compensate for the transmission losses induced by the cell windows, the transmission was also measured for a wavelength scan with pure CH4 in the cell. Furthermore, the transmission at 212.2 nm was measured for seven different NH3/CH4 mixtures with NH3 concentrations ranging from 4.86 × 1014 cm−3 (20 ppm) to 3.11 × 1016 cm−3 (1250 ppm) for evaluation of the effective absorption cross section at this wavelength.
In order to demonstrate the ps-DIAL technique for range-resolved detection of NH3, a 77-cm tube, open in both ends, having an inner diameter of 7.0 cm and a gas inlet through the wall at the middle, was used. Mass-flow controllers (Bronkhorst) were used to control the proportions of NH3 and CH4 in a total inlet gas flow of 3 l/min. The gas mixture was varied in 20 steps from 25 ppm to 5000 ppm in randomized order. Before each measurement, the tube was purged with the gas mixture for 2 minutes to ensure that no residual gas from the previous measurement remained in the tube. The laser pulse energies were typically 250 μJ at the on-resonance wavelength, 212.2 nm, and 310 μJ at the off-resonance wavelength, 214.5 nm (further details are given in section 3.1). The on- and off-resonance measurements were carried out sequentially, i.e. on-resonance measurements for all gas mixtures were conducted first followed by all off-resonance measurements. Data were acquired as averages over 1500 laser shots. A reference measurement with 3 l/min pure CH4 flow was used to calibrate the acquired DIAL curves for different geometrical overlap functions at the two wavelengths.
The spatial resolution of the ps-DIAL technique was investigated by measuring NH3 concentration profiles in a setup consisting of two 6-cm diameter porous-plug burners, placed in the area where the test tube was located for the aforementioned measurements, see Fig. 1. One burner was fixed whereas the other was placed on a translator, allowing for measurement at multiple locations along the x-axis. In order to determine the ultimate spatial resolution achievable with the present experimental system, measurements were made with two thin metallic wires placed along the x-axis, slightly separated in the y-direction. One of the wires was positioned on a translator to be movable along the x-axis.
3. Results and discussion
3.1 Absorption spectra of ammonia
The absorption spectrum of NH3 was measured in order to identify a suitable pair of wavelengths (λon and λoff), yielding a significant difference in absorption cross section and minimum background absorption in the dial measurements. Figure 2(a) shows a broadband absorption spectrum recorded using a Deuterium (D2) lamp and a spectrometer. The spectrum features parts of the v2’ progression in the A←X, v2” = 0 band, where the normal vibration mode v2 corresponds to a symmetric bend of the N-H bonds .
Since the shortest available laser wavelength in our setup is 210 nm, absorption bands appearing below 210 nm cannot be used, and, hence, the peak at around 212 nm, i.e. the A←X, ν2’ = 1 band , was found to be best suited for dial measurements. This absorption band was studied in more detail by scanning the laser from 211 to 215 nm and recording the absorption spectrum, from which the effective absorption cross sections using our setup was calculated utilizing the Beer–Lambert law, see Fig. 2(b). Based on this result the two DIAL wavelengths were selected: λon = 212.2 nm and λoff = 214.5 nm.
In order to improve the accuracy of the determined absorption cross section at 212.2 nm, the absorbance at this wavelength was measured for a set of NH3/CH4 mixtures in the cell. The measurement results are presented in Fig. 3 with absorbance, i.e. ln(I0/I), plotted versus column density, LN, where L is absorption path length and N is the absorber number density. A linear fit, weighted with the relative errors of the measured absorbance at each column density, results in a slope of 3.0 × 10−18 cm2, which thus corresponds to the effective cross section at 212.2 nm. The calculated absorption cross sections agree well with literature data [15–17].
3.2 Concentration measurements towards ammonia detection limit using ps-DIAL
Using the wavelengths 212.2 and 214.5 nm as λon and λoff, respectively, DIAL measurements were conducted, at room temperature and pressure, in a test tube purged with different NH3/CH4 mixes. Concentration profiles were evaluated using Eq. (1), which is a part of the full dial equation  and in the following referred to as the simplified dial equation.18] and the total contribution from this error source to the evaluated number density is 5 × 1011 cm−3, corresponding to 0.02 ppm at the prevailing conditions.
A further prerequisite for the simplified DIAL equation to be valid is that the ratio of the backscattering coefficients for the two wavelengths, rsca, is assumed to be range independent. Molecular scattering exhibits species-specific wavelength dependence, hence rsca is range dependent due to large variations in bulk gas composition. The impact of this can be estimated from the magnitude of the corresponding term in the full DIAL equation calculated for a change in gas composition from 100% N2 to 20% CH4 over a distance of 5 cm. Pure nitrogen represents a simplified air model and the CH4 level of 20% is chosen higher than the maximum concentration in the tube, as indicated from NH3 absorption and dial measurements (see Fig. 5). Extrapolating available scattering cross section data for N2 and CH4 , yielding σCH4(212.2) = 2.4141 × σN2(212.2) and σCH4(214.5) = 2.4100 × σN2(214.5), results in a positive number density offset of 1 × 1013 cm−3, corresponding to 0.4 ppm. Clearly, this contribution is negligible compared to the probed NH3 concentrations.
Data-processing of the acquired lidar curves included smoothing using a 5 data point (corresponding to a distance of 6 cm) running-average filter. Moreover, to account for differences in the overlap of the laser beam paths corresponding to λoff and λon, the signal ratio P(R, λoff)/ P(R, λon) in Eq. (1) was compensated with the reference measurement ratio, C(R), acquired with no NH3 present. The evaluation procedure is illustrated by the curves in Fig. 4 .
In Fig. 4(a), six lidar curves, P(R, λ), are shown, corresponding to three different gas mixtures injected into the tube. The red and black curves represent P(R, λoff) and P(R, λon), respectively, and the injected NH3 number densities are indicated in the legend. The shape of the curves give a good indication of the spatial overlap function, which in near-field lidar is governed by the focus distance of the collection telescope mirror and the geometric overlap between the telescope optical axis and the laser beam paths. In these measurements the focus distance of the telescope is set to maximize the collected signal at distance 2 m, which in turn limits the available measurement range to approximately 1 m. In Fig. 4(b), the reference measurement ratio, C(R), is shown. The impact on data evaluation of C(R) is illustrated in Fig. 4(c) using the data presented in Fig. 4(a). The green curves show the ratios of P(R, λoff)/ P(R, λon) evaluated directly from acquired data whereas the black curves show the ratios compensated with the reference measurement ratio, C(R).
The three corresponding evaluated DIAL curves are shown in Fig. 5(a) . The full set of evaluated concentration profiles show that the detected NH3 response exceeds the noise at a number density of ~1 × 1015 cm−3, corresponding to ~40 ppm. The potential errors induced by using the simplified DIAL equation are thus negligible as they are much smaller than the estimated detection limit. The NH3 concentrations in the center of the tube were evaluated as the mean value of 4 data points separated by 1.5 cm, at a detection distance of 2 m, and are plotted versus the inlet concentrations in Fig. 5(b).
To validate the DIAL results, the absorption at 212.2 nm was obtained for the highest NH3 concentration by measuring the laser pulse energy before and after the tube. Using an estimation of the effective absorption path length, 36 cm, indicated by the profiles shown in Fig. 5(a), the concentration was calculated, using the Beer-Lambert law, to 1.1 × 1016 cm−3, which is in excellent agreement with the concentration found by the DIAL measurement. The discrepancy between the inlet NH3 concentration and the measured NH3 concentration is most likely due to the unknown mixing behavior of the injected gas with ambient air in the tube.
The obtained lower detection limit at 40 ppm corresponds to a 5% fraction of light absorbed and agrees well with results from other lidar and absorption measurements. Lidar-NH3 detection in the atmosphere, probing infrared transitions, showed a detection limit of 5 ppb over a distance of 2.7 km , which corresponds to ~2% absorption. A setup for laser-based NH3 absorption measurements applied in a power plant [2, 3] gave a detection limit of 1 ppm over a 10 m range, resulting in ~3% absorption. Furthermore, a detection limit of 2 ppb over a range of 350 m obtained for atmospheric NO lidar at 226 nm corresponds to ~1.4% absorption . It should also be noted that the lowest detectable differential absorption depends on signal and noise levels in the actual measurement.
Typical NH3 concentrations utilized for non-catalytic reduction of NOx are between 300 and 1200 ppm , which is readily detectable with ps-dial according to the result presented in Fig. 5(b). To improve the detection limit, which is inversely proportional to Δσ according to Eq. (1), it would, according to the spectrum displayed in Fig. 2(a), be possible to use the A←X, ν2’ = 2 band, located close to 208.5 nm, as λon, and 210.8 nm as λoff, using a suitable laser. At shorter wavelengths, absorption by atmospheric O2 inhibits DIAL measurements of NH3. If the NH3 concentration becomes high enough to potentially get into the non-linear regime of the Beer-Lambert law, the evaluated number density will not be correct. In such a case, λon with a lower absorption cross section might be chosen. If needed, absorption peaks with orders of magnitude lower cross sections are available between 220 and 230 nm .
Moreover, selection of wavelengths λon and λoff requires that interfering absorption from other species is considered. For example, both SO2 and NO have interfering absorption features and cross sections of the same order of magnitude as NH3 in the spectral range 200-230 nm , and needs to be treated carefully if they are simultaneously present.
If variation in NH3 concentration or interfering sources is expected during acquisition of the corresponding on- and off-resonance signals, each single-shot recording should preferably include both signals. By overlapping two beams with wavelengths λon and λoff, respectively, with a proper delay between the two pulses, both the on- and off-resonance signals can be captured in the same measurement. Thus, the beam-overlap concept requires two tunable laser sources, which in practice must be pumped with the same laser to avoid temporal jitter.
3.3 Spatial resolution
To visualize the ability of the DIAL technique to perform spatially resolved NH3 concentration measurement along the optical axis, i.e. one-dimensional imaging, a measurement series was conducted with two porous plug burners supplied with a total flow of 7.5 l/min of the 0.5% NH3/CH4 mixture. The center-to-center distance of the burners was varied from 30 to 70 cm. For distances 30 and 70 cm, respectively, the lidar curves, P(R, λ), are shown in Fig. 6(a) , with the corresponding ratios, P(R, λoff)/ P(R, λon), shown in Fig. 6(b).
The resulting NH3 concentration profiles are shown in Fig. 7 . Data were evaluated as described in section 3.1, however the compensation, C(R), accounting for the geometric overlap function between the laser beams, was omitted. Consequently, the decrease in concentration baseline level observed in the left part of the figure is due to different geometrical overlap functions of the two lidar signals recorded on- and off-resonance, respectively. Moreover, the fluctuations in the peak concentration values are on the order of the detection limit.
Nevertheless, it is clearly seen that the two NH3 peaks are well resolved at all burner separations investigated. The measured distances between the peaks are in very good agreement with the actual distances, which demonstrate the high spatial accuracy of the system.
To determine the highest achievable spatial resolution, an experiment with scattering from two thin metallic wires was performed. The wires were moved closer up to the defined resolution limit, when the minimum intensity between the scattering peaks in the acquired lidar curve exceeds half the individual peak intensities, resulting in an optimum resolution of 16 cm. For further details on the resolution measurement procedure see Kaldvee et al. .
The spatial resolution, mainly limited by the temporal response of the MCP-PMT, may be increased using a streak camera. The spatial resolution would then be limited to ~6 cm, while providing a range interval of 1.5 m . However, signal losses due to lower quantum efficiency and reduced collection of light through the narrow streak camera slit, enabling the high spatial resolution, needs to be considered.
Single-ended remote detection of NH3 has been demonstrated using ps-DIAL in the ultraviolet regime. With an acquisition time of 2.5 minutes for λon and λoff, respectively, the detection limit was found to be 40 ppm with a spatial resolution of 16 cm and a spatial accuracy high enough to position NH3 peaks correctly on a centimeter scale. The ps-DIAL technique clearly allows for NH3 detection at levels relevant in combustion applications.
Moreover, the measurements can in principle be carried out with a laser source providing higher pulse energy, utilizing the NH3 A←X, ν2’ = 1 band, as it coincides fairly well with the fifth harmonic of the Nd:YAG laser at 212.9 nm. Such a configuration would in particular be attractive for practical applications, where the experimental conditions might be very demanding, requiring a robust and durable laser source. The extra pulse energy available using an Nd:YAG laser would also extend the detection range. However, without the possibility to tune the laser wavelength, off-resonance data need to be acquired either under conditions without NH3 in the measurement region or, alternatively, using the fourth harmonic radiation of the Nd:YAG laser at 266 nm.
As addressed in the previous discussion, the spectral region 210-230 nm contains absorption lines of SO2 and NO , which both are of relevance in combustion. Correct NH3 detection requires such interfering absorption to be considered when selecting on and off resonance wavelengths. However, there is also an opportunity to probe these species using the ps-DIAL concept with appropriate wavelengths. Thus, the technique permits single-ended, range resolved remote detection of several species of interest in combustion, and shows strong potential for further development and applications.
The authors would like to thank the Centre of Combustion Science and Technology (CECOST) and the European Research Council Advanced Grant DALDECS for financial support. We also very much appreciate the support by Odd Hole during the measurement of the NH3 absorption spectrum in Fig. 2(a).
References and links
1. K. Kohse-Höinghaus and J. B. Jeffries, eds., Applied combustion diagnostics (Taylor&Francis, 2002).
2. W. Meienburg, H. Neckel, and J. Wolfrum, “In situ measurement of ammonia with a 13CO2-waveguide laser system,” Appl. Phys. B 51(2), 94–98 (1990). [CrossRef]
3. W. Meienburg, J. Wolfrum, and H. Neckel, “In situ measurement of ammonia concentration in industrial combustion systems,” Proc. Combust. Inst. 23, 231–236 (1990).
4. A. Hinz and S. Horler, “CO2-laser sensor system for in-situ measurement of ammonia in flue gas,” Tech. Mess. 63, 282–287 (1996).
6. C. Weitkamp, ed., Lidar Range-resolved Optical Remote Sensing of the Atmosphere (Springer, 2005).
8. B. Kaldvee, J. Bood, and M. Alden, “Picosecond-lidar thermometry in a measurement volume surrounded by highly scattering media,” Meas. Sci. Technol. 22(12), 125302 (2011). [CrossRef]
9. B. Kaldvee, Division of combustion physics, Lund University, Box 118, 221 00 Lund, Sweden, J. Wahlqvist, M. Jonsson, C. Brackmann, B. Andersson, P. van Hees, J. Bood, and M. Aldén are preparing a manuscript to be called “Room fire characterization using lidar diagnostics and CFD.”
10. L. J. Muzio and G. C. Quartucy, “Implementing NOx control: research to application,” Pror. Energy Combust. Sci. 23(3), 233–266 (1997). [CrossRef]
11. M. Østberg, K. Dam-Johansen, and J. E. Johnsson, “Influence of mixing on the SNCR process,” Chem. Eng. Sci. 52(15), 2511–2525 (1997). [CrossRef]
12. G.-W. Lee, B.-H. Shon, J.-G. Yoo, J.-H. Jung, and K.-J. Oh, “The influence of mixing between NH3 and NO for a DeNOx reaction in the SNCR process,” J. Ind. Eng. Chem. (Amsterdam Neth.) 14, 457–467 (2008).
14. A. Duncan, “The ultraviolet absorption spectrum of ammonia,” Phys. Rev. 47(11), 822–827 (1935). [CrossRef]
15. B.-M. Cheng, H.-C. Lu, H.-K. Chen, M. Bahou, Y.-P. Lee, A. M. Mebel, L. C. Lee, M.-C. Liang, and Y. L. Yung, “Absorption cross sections of NH3, NH2D, NHD2, and ND3 in the spectral range 140–220 nm and implications for planetary isotopic fractionation,” Astrophys. J. 647(2), 1535–1542 (2006). [CrossRef]
16. R. Gall, D. Perner, and A. Ladstätter-Weissenmayer, “Simultaneous determination of NH3, SO2, NO and NO2 by direct UV-absorption in ambient air,” Fresenius J. Anal. Chem. 340, 646–649 (1991). [CrossRef]
17. G. H. Mount, B. Rumburg, J. Havig, B. Lamb, H. Westberg, D. Yonge, K. Johnson, and R. Kincaid, “Measurement of atmospheric ammonia at a dairy using differential optical absorption spectroscopy in the mid-ultraviolet,” Atmos. Environ. 36(11), 1799–1810 (2002). [CrossRef]
18. R. B. Miles, W. R. Lempert, and J. N. Forkey, “Laser Rayleigh scattering,” Meas. Sci. Technol. 12(5), R33–R51 (2001). [CrossRef]
19. J. A. Sutton and J. F. Driscoll, “Rayleigh scattering cross sections of combustion species at 266, 355, and 532 nm for thermometry applications,” Opt. Lett. 29(22), 2620–2622 (2004). [CrossRef] [PubMed]
21. H. Volten, J. B. Bergwerff, M. Haaima, D. E. Lolkema, A. J. C. Berkhout, G. R. van der Hoff, C. J. M. Potma, R. J. W. Kruit, W. A. J. van Pul, and D. P. J. Swart, “Two instruments based on differential optical absorption spectroscopy (DOAS) to measure accurate ammonia concentrations in the atmosphere,” Atmos. Meas. Technol. 5(2), 413–427 (2012). [CrossRef]