Design of a Laser-Induced Breakdown Spectroscopy System for On-Line Quality Analysis of Pulverized Coal in Power Plants

Transcription

1 Design of a Laser-Induced Breakdown Spectroscopy System for On-Line Quality Analysis of Pulverized Coal in Power Plants WANGBAO YIN, LEI ZHANG,* LEI DONG, WEIGUANG MA, and SUOTANG JIA* State Key Laboratory of Quantum Optics and Quantum Optics Devices, College of Physics and Electronics Engineering, Shanxi University, Taiyuan , P.R. China It is vitally important for a power plant to determine the chemical composition of coal prior to combustion in order to obtain optimal boiler control. In this work, a fully software-controlled laser-induced breakdown spectroscopy (LIBS) system comprising a LIBS apparatus and sampling equipment has been designed for possible application to power plants for on-line quality analysis of pulverized coal. Special attention was given to the LIBS system, the data processing methods (especially the normalization with Bode Rule/DC Level) and the specific settings (the softwarecontrolled triggering source, high-pressure gas cleaning device, samplepreparation module, sampling module, etc.), which gave the best direct measurement for C, H, Si, Na, Mg, Fe, Al, and Ti with measurement errors less than 10% for pulverized coal. Therefore, the apparatus is accurate enough to be applied to industries for on-line monitoring of pulverized coal. The method of proximate analysis was also introduced and the experimental error of A ad (Ash, ad is an abbreviation for air dried ) was shown in the range of 2.29 to 13.47%. The programmable logic controller (PLC) controlled on-line coal sampling equipment, which is designed based upon aerodynamics, and is capable of performing multipoint sampling and sample-preparation operation. Index Headings: Laser-induced breakdown spectroscopy; LIBS; Pulverized coal; Sampling equipment; High-pressure gas cleaning device; Proximate analysis. INTRODUCTION In coal-fired power plants, the quality of coal not only impacts pollutant emissions, but also the boilers, affecting combustion stability, corrosion, and ash deposition and disposal. As properties and composition of coal vary widely, it is advisable that all coals to be fired in a boiler should be analyzed. Traditional techniques for coal analysis can be divided into chemical analysis and physical analysis. As for the former, one of those analyses is acid extraction, which is widely used but requires multi-reagents and complicated processes (2 3 hours); another technique is thermal sorption analysis (TDA), which takes less time (8 15 minutes) for analysis, but requires the consumption of oxygen and nitrogen gas. Physical analysis methods include X-ray fluorescence (XRF), 1 prompt-gamma neutron activation analysis (PGNAA), 2,3 inductively coupled plasma atomic emission spectroscopy (ICP-AES), 4 atomic absorption spectroscopy, 5 near-infrared absorption spectroscopy, 6 etc. Among these analytical techniques, XRF and PGNAA are the most preferable techniques for possible application to power plants and have already been commercially available. However, both of these two analytical techniques have difficulties in analyzing low atomic elements such as C and O. It is thus necessary to develop new techniques suitable for field monitoring. Laser-induced breakdown spectroscopy (LIBS) was chosen as an effective analysis tool that allows direct chemical analysis Received 26 January 2009; accepted 1 May * Authors to whom correspondence should be sent. k1226@sxu. edu.cn, tjia@sxu.edu.cn. of materials by its vaporization achieved by interaction with a laser beam of adequate frequency and intensity. Much has been published on this topic and there are several commercial LIBS systems available for sale. A few examples include application as a quality control technique for steel manufacturing, 7,8 quantitative characterization of gold interface in jewelry industry, 9 rapid analysis of multi-component pharmaceutical tablets, 10 diagnostics of pulsed laser-solid interactions to be applied for laser cleaning of historic buildings and statues, 11,12 investigation of the composition of archaeological ancient pottery, 13 and on board LIBS analysis of marine sediments. 14 Laser-induced breakdown spectroscopy has been previously applied successfully to determine heavy metals in soils, and the detection limits for several metals (such as As, Cd, Cr, Cu, Hg, Ni, Pb, Tl, Zn, etc.) have been obtained For coal analysis, Ottesen 21 showed a preliminary application of LIBS for determining the composition (the inorganic constituents, such as Si, Al, Fe, Ti, Ca, Mg, etc.) of coal particulates in combustion environments; Zhang 22 performed LIBS in a large magnetohydrodynamics coal-fired flow facility, and a few elements including Ca, Fe, Al, Ti, and Sr were also quantified; Wallis 23 investigated the chemical components of 30 low-ash lignite samples by LIBS and the detection limits of Ca, Al, Na, Fe, Mg, and Si were obtained; Kurihara 24 applied an automated LIBS instrument to a 1000 MW pulverized coal-fired power plant for real-time monitoring of unburned carbon in fly ash, obtaining good agreement between LIBS and traditional methods for carbon measurement, with a standard deviation of 0.27%; Body 25 at the Cooperative Research Center for Clean Power developed an advanced LIBS instrument for analysis of pressed coal, and the measurement repeatability and accuracy for the inorganic components (such as Al, Si, Mg, Ca, Fe, Na, and K) was typically within 610%; Yu 26 analyzed pulverized coal with LIBS by employing a monochrometer and a photomultiplier (PMT), and the calibration curves of C, H, O, N, and S were obtained; Romero 27 performed LIBS at an 1150 MW coal-fired power plant, and the results demonstrated the feasibility of providing predicted fusion temperature information with enough resolution. However, there is no report on the design of a LIBS system suitable for on-line pulverized coal quality monitoring in coal-fired power plants. The present work is aimed at designing a system that attempts to capture the fundamental advantages of LIBS in a commercially viable form suitable for in situ quantitative analysis (including C, H, Si, Na, Mg, Fe, Ca, Al, Ti, and organic oxygen 28 ) and proximate analysis of pulverized coal in coal-fired power plants. EXPERIMENTAL Sample. In this work, eight coal samples and nine soil samples, which contain the substances of interest over a Volume 63, Number 8, 2009 APPLIED SPECTROSCOPY /09/ $2.00/0 Ó 2009 Society for Applied Spectroscopy

2 TABLE I. Coal sample specifications. The detailed names of the province in China are given in parentheses. No. Origination %C %Si %Al %Ca %Mg %Ti %Fe 1 Tieling (Liaoning) Xuanwei (Yunnan) Huozhou (Shanxi) Xinli (Zhengzhou) Datong (Shanxi) Yangquan (Shanxi) Zhangze (Shanxi) Yangcheng (Shanxi) wide area of concentration values, were dried prior to analysis. The standard powdery coal samples (including low-ash and high-ash coals) were finely crushed (mostly within lm in diameter). These samples were gathered from the pulverizers of different coal-fired power plants and certified by ICC (Institute of Coal Chemistry, China). A list of the samples is given in Table I, including the geographical origins, the element names, and the concentration values (in percentage by weight). For selecting the optimal data processing methods, samples with absolute concentration values were needed. Since values certified by ICC may not be precise enough due to the mechanical or artificial fault, a series of soil reference samples with Mn were prepared (see Table II). The standard soils were gathered from the suburban area in the proximity of Taiyuan, China. After being dried, ground to a grain size, and sieved through a 100 lm sieve, the soils were doped with varying concentrations of Mn (fine metal powder, ;80 lm). Finally, the confected samples were homogenized in vessels by continuous agitation for 12 hours. Instrumentation. The entire analysis system consists of a LIBS apparatus and sampling equipment. A schematic description of the LIBS apparatus is presented in Fig. 1. A portable Nd:YAG laser (90 mm 3 90 mm mm) operating at 1064 nm was employed as the ablation source, with a fixed energy of 120 mj/pulse. The ablation laser beam was focused at the sample by a 90 mm diameter and 400 mm focal length quartz lens. Then, the plasma plume emission was guided to a three-channel spectrometer (AvaSpec-2048FT, nm, 0.3 nm spectral resolution) equipped with three gratings (3600, 1200, and 1200 grooves/mm) by means of a 2-m-long all-silica optical-fiber bundle (numerical aperture, 0.22; core, 600 lm) consisting of three fibers. In the front of the fiber bundle, a miniature focusing component was located 240 mm from the laser focal point on the sample at an angle of 458 from normal. Since the coal aerosol excited by the powerful laser beams can cause carbonaceous deposition and significantly change the transparencies of the two quartz lenses, a high-pressure gas cleaning device that could be remotely controlled by a computer through a solenoid valve was placed between the two lenses. It was notable that the two nozzles of the cleaning device should be installed properly to avoid any sample loss due to the turbulent flow. The output signal from the spectrometer was transported to the computer through a USB wire. Instead of a pulse generator, the spectrometer was triggered by a synchronous TTL pulse (10 Hz, 50 ms width) generated from the parallel port of the computer, which enabled the operator to carry out remote switch control of the apparatus. A simple circuit comprising a photocoupler, an inverter, and several resistors was employed to prevent the groundings of the computer and the spectrometer from being connected together. The programmable logic controlled (PLC) on-line coal sampling equipment, which was designed based upon aerodynamics, can be divided into two parts: a sampling module and a sample preparation module (see Fig. 1). The sampling module consists of an entrance pipe and an exit pipe. Since the coal is considered to be a heterogeneous medium, the entrance pipe is fed by several branch sampling pipes for multipoint sampling. In addition, a cyclone collector, two optical transmitter receivers, a vibrator, and an unloading valve are attached to the entrance pipe to supply a fixed amount of coal. In Fig. 1, both the side view and vertical view of the sample preparation module are shown in detail. The sample holder, which rotates at a speed of 2 rev/min by a small stepping motor, is attached at the end of a 0.25 m long rotation axis. The operation of the sampling equipment is described as follows. After switching on the negative pressure generator, the fine coal particles come into the cyclone collector with the airflow and fall into the bottom of the entrance pipe. The generator will work till a signal 1 is sent out from the upper optical transmitter receiver. The signal 1 means that the amount of coal has met the requirement for analysis. Then, the negative pressure generator is switched off, and the vibrator and the unloading valve are turned on. The vibrator and the uploading valve will not be turned off until a signal 0 is sent out from the lower optical transmitter receiver. The signal 0 means that all the coal in the entrance pipe has been loaded into the sample holder at the position of sample in. Here, another optical transmitter receiver is employed to check whether the holder has been packed. With the rotation of the axis, the scraper scrapes away the redundant coal particles and smoothes the coal surface in the holder. After that, the holder moves to the laser focus for LIBS analysis. Finally, at the position of sample out, both the residual coal in the sample holder and the generated aerosols in the proximity are sucked up and transported back to the coal chute through the exit pipe. It is advisable that the rotation axis should move steadily to prevent the sample surface from being uneven. A photograph of the sampling equipment is shown in Fig. 2. The left side of it is the control unit comprising an industrial computer, a spectrometer, and a control box. The right side of the equipment can be TABLE II. Soil sample specifications. The samples cover a wide range of Mn concentrations. No Mn (mg/kg) Volume 63, Number 8, 2009

3 FIG. 1. Schematic representation of the LIBS apparatus developed in this work. The numbers marked in the figure are defined as follows: (1) cyclone collector, (2) the upper optical transmitter receiver, (3) the lower optical transmitter receiver, (4) vibrator, and (5) unloading valve. divided into three equal portions vertically: on the top there are six branch sampling pipes and a negative pressure generator; in the middle portion there is a motor for driving the rotation axis, and the LIBS apparatus will be mounted here; the bottom portion comprises an exit pipe and an electrical distribution system. The equipment is fully software controlled, and the software is written in visual basic (VB) language. Operation. The operation of the LIBS system reported in this paper can be divided into three steps. First, by clicking the start button of the software (compiled in LabVIEW 8.0), a 10 Hz TTL level, which acts as the trigger source to the spectrometer, is generated. Then, the spectrometer responds by putting out a 10 Hz TTL output signal with a 10 ls width to fire the laser. It was found that acquiring the signal for 10 ms, after a delay of 200 ns, gave optimal signal-tobackground ratios for all the elements studied. The TTL pulses last until the pulse numbers come up to 100. Then, the results are displayed by processing the spectral data saved in the random access memory (RAM) of the computer. Second, the solenoid valve is kept open for 2 s, and the compressed air that flows through the two nozzles of the cleaning device blows away the aerosol deposited on the lenses. Third, the LIBS system pauses for a period of time till the temperature of the laser crystal falls to normal (293 K) or below. The length of the pause is varied according to the environmental temperature. Generally, this time interval is set to 2 min. The whole operation process usually consumes approximately 3 min. It should be noted that the operation described above is merely suitable for the present work performed in the laboratory. However, for industrial applications, the sampling and sample preparation processes should be further considered. DATA PROCESSING Various sophisticated methods for data processing have been fully demonstrated in previous articles In the current work, the data processing process for a given emission line could be summarized as: Subtracting the background level from the yielded 100 groups of spectra data, respectively. Normalizing the measured peak areas by ratio to the DC levels of the total plasma spectrums. Here, the peak area was integrated by using the Bode Rule, and the spectral lines for C (247.9 nm), Si (288.2 nm), Al (309.3 nm), Ca (422.7 nm), Mg (279.6 nm), Ti (334.9 nm), H (656.3 nm), and Fe (358.1 nm) were used. Discarding the negative or zero values and the highest 24% and lowest 17% of the data values. The selection of these levels was made by an empirical investigation of different data levels, such that minimal impact results on the data set s mean and median values. Averaging the remaining data and calculating the absolute concentration by substituting this mean value into the corresponding calibration equation. Converting the elemental compositions into proximate analysis results. Details of the calculation methods will be discussed in the next section. APPLIED SPECTROSCOPY 867

4 FIG. 2. Photograph of the programmable logic controlled (PLC) on-line coal sampling equipment. RESULTS AND DISCUSSION Since the experiments in the current investigation were carried out in the laboratory, the coal was loaded into the sample holder artificially, without the aid of the sampling equipment. Figure 3 depicts the typical averaged spectra obtained from the three-channel spectrometer. Figure 4 illustrates the relationships between peak intensities and the relevant plasma emission intensities for C (247.9 nm), Ca (422.7 nm), and H (656.3 nm) with 33 successive laser pulses. The linear ascending points of the three elements indicate the necessity for normalizing. In the current investigation, for comparison, each soil sample (Mn nm) was measured five times using six methods of normalization simultaneously. Various integral methods (Trapezoidal Rule, Simpson s Rule, Simpson s 3/8 Rule, and Bode Rule) were adopted to integrate the peak area and the baseline. The Bode Rule/Smooth Section method was used to normalize the peak intensity by ratioing the integral value of the peak area with the Bode Rule to a short and smooth spectral section ( nm of Channel 1), which was selected to represent the matrix effect. The so-called matrix effect means the produced plasma composition can be affected by the physical and chemical properties of the sample. For the Bode Rule/DC Level method, the DC Level could be estimated FIG. 3. (a, b, and c) Averaged spectra of the coal sample obtained from the three-channel spectrometer. directly through the specific function block of LabVIEW. The criterion to gauge the effectiveness of these methods was to compare the relative standard deviation (RSD) values. The results are shown in Fig. 5. It can be seen that the RSD values of the first three methods are larger than the fourth one, probably because the Bode Rule has the largest order of approximation 868 Volume 63, Number 8, 2009

5 FIG. 4. Peak intensity as a function of total plasma emission intensity integrated with the trapezoidal rule for the coal sample from Xinli (Shanxi). Three linear relationships are obtained for C (247.9 nm), Ca (422.7 nm), and H (656.3 nm). FIG. 6. Normalized intensity ratios of C with 100 consecutive laser shots for the coal sample from Tieling. Shown in open circles are those shots ascribed to the uncertain events, as defined in the text. within these integral methods. It is also shown that the application of the Bode Rule/DC Level method yields the most even distribution of values around the mean (the RSD is of the order of 8.2% or less). In other words, the Bode Rule/DC level normalization method can enhance the accuracy of quantitative analysis and yield the most accurate measurements. In Fig. 6, the data points that satisfy the discard criterion as described above, and therefore are rejected, are shown as open circles. This is to minimize the matrix effects caused by various uncertain events, such as coal particle sizes, undesired aerosol events, rough surface of coal powders, etc., during the plasma formation. Following the data processing methods, each coal sample listed in Table I was analyzed five times. The calibration curves and error bars for some of the elements (C, Ca, Mg, and Ti) present in coal are displayed in Fig. 7. The error bars indicate the experimental uncertainties and the heterogeneous distribution of the relevant composition in the same sample. Since the intensity ratio is approaching zero if the relevant concentration is quite small in coal, the (0, 0) point of the drawing was employed to obtain the best calibration curve. The experimental points for Mg, Ti, and Fe were fitted to the straight line y ¼ a þ bx. For some of the elements analyzed, the measured values grew nonlinearly with increasing certified concentrations, thus indicating the presence of line self-absorption. The nonlinearity is particularly pronounced for C, Si, Al, and Ca owing to either the major components (C, Si, and Al) present in coal or the large transition strength (Ca, nm). In this study, we performed a nonlinear fitting 33 of the experimental points to the expression: y ¼ a þ bcð1 e x=c Þ where x represents the concentrations and y is the intensity ratios. This expression, which is approximated by the straight line y ¼ a þ bx at low concentrations, is found to be useful to describe the saturation of the calibration curves. The corresponding values of a, b, c, and the correlation coefficients R are listed in Table III. As can be seen, high correlations are obtained in most cases, with values of R above To check the reproducibility of the LIBS apparatus, the No. 8 coal sample shown in Table I and three other national certified reference pulverized coal samples were analyzed. Table IV shows the excellent accuracy and the comparison of the limits of detection (LOD) with other reports 23,25 for some of the elements studied. The statistical measurement error is about 10%, and the LOD values in the measurements are 2 3 times those of the references. Note that the coal samples ð1þ TABLE III. Values of a, b, c, and R obtained for the laboratory nonlinear and linear calibration curves. Element a b c R FIG. 5. The RSD is defined as the ratio of the standard deviation in five measurements to the mean value of Mn for each soil sample with different methods of normalization. C Si Al Ca Mg Ti Fe APPLIED SPECTROSCOPY 869

6 FIG. 7. Calibration curves for the elements (a) C,(b) Ca, (c) Mg, and (d) Ti in pulverized coal samples. Nonlinear and linear fittings are applied to the selfsaturation and low-concentration samples, respectively. measured here were in powdery form rather than hard pellets, which has introduced additional plasma instability. The results for C are presented in Table V, where the measured concentrations are slightly larger than the certified standards. This is probably due to the accuracy limits of the calibration curve. The relative error for C is of the order of %. Similar investigations were also performed on other elements such as Na, Zn, Mn, and H recently. TABLE IV. Quantitative analysis results of the No. 8 (Yangcheng) coal sample and the relevant detection limits compared with others. Sample Ca Mg Fe Al Si Our work (%) Stat Standard analysis LOD WL (nm) 423 nm 280 nm 358 nm 309 nm 288 nm Body s work (%) LOD WL (nm) 393 nm 285 nm 358 nm 309 nm 288 nm Wallis s work (%) LOD WL (nm) 396 nm 285 nm 238 nm 309 nm 288 nm 870 Volume 63, Number 8, 2009

7 TABLE V. Example of the LIBS analysis of three standard national pulverized coal samples that have been accurately certified. GBW11108e GBW11105b GBW11101h LIBS measurement with 100 laser pulses (%) Stat Standard analysis Certified For power plants, it is worth transforming the elemental composition analyzed by LIBS into proximate analysis forms, including M t,ad (total moisture), FC ad (fixed carbon), A ad (ash), and Q net,ad (net calorific value at constant volume). The M t,ad value is determined by a near-infrared reflection (NIR) spectroscopic analyzer installed in the front of the conveying belt; the F Cad value is expressed by the measured concentration of C; the A ad value is calculated by summing up the proportions of the inorganic components (including SiO 2,Al 2 O 3,Fe 2 O 3, CaO, TiO 2, and MgO). The corresponding formula can be expressed as A ad ¼ 2:14C Si þ 1:89C Al þ 1:43C Fe þ 1:40C Ca þ 1:67C Ti þ 1:66C Mg ð2þ where the coefficients such as 2.14 and 1.89 are obtained from the atomic weight ratios (Si þ 2O)/Si and (2Al þ 3O)/2Al, respectively; the Q net,ad value is calculated by using the empirical formula 34,35 Q ad ¼ 0:3491C C þ 1:1783C H þ 0:1005C S 0:0151C N 0:1034C O 0:0211A ad where C S and C N with little contribution to the calorific value are supposed to be zero. In the present work, a comparison of the certified and the measured A ad values is shown in Fig. 8. However, since the sulfur trioxide, potassium monoxide, and other unknown impurities were not taken into account, most of ð3þ the A ad values obtained here are slightly lower than the certified standards, with the relative error ranges from 2.29% to 13.47%. CONCLUSION In this investigation, a LIBS system comprising a LIBS apparatus and sampling equipment developed for application to power plants has been described. We have demonstrated that it is feasible to implement the technique of LIBS for in situ characterization of pulverized coal. Specific details of some embodiments of the structure and operation of the LIBS system are shown. The LIBS apparatus can be controlled by a PC through a special triggering source (the parallel port of the computer), and the optical components in the apparatus can be periodically cleaned by the highpressure gas cleaning device to prevent the optical lens from accumulating aerosol deposition. The sampling equipment is capable of performing multipoint sampling and sample preparation operation. By using the Bode Rule/DC Level normalization and the proposed data conversion methods, both elemental (C, Ca, Mg, Ti, Si, H, Al, Fe, etc.) and proximate analysis (FC ad, A ad, and Q ad ) of pulverized coal can be accomplished. The relative measurement errors presented here for elemental analysis are within 10%, while those of A ad are in the range of %. The newly designed LIBS system described in the current investigation is capable of providing the operators with reliable information on the coal quality, according to which the boiler efficiency could be improved in a timely manner by control of the rotation speed of the mill rotary separator. 24 Future studies will be devoted to quantitative analysis of other elements such as K and S, to which industries are attaching great importance. ACKNOWLEDGMENTS This work was supported by 973 Program (Grant. No. 2006CB921603), National Natural Science Foundation of China (Grant No and ), Science and Technology Project of Taiyuan (Grant. No ), and Shanxi Province Foundation for Returned Overseas Scholars. The authors are grateful to Haitong Automation Technique Ltd. and the Institute of Coal Chemistry (ICC) for their financial and technical support. FIG. 8. The calculated A ad (ash) values of the eight coal samples compared to the certified standards. 1. O. M. Kalfa, Z. Üstündağ, İ. Özkırım, and Y. K. Kadıoğlu, J. Quant. Spectrosc. Radiat. Trans. 103, 424 (2007). 2. Y. Toh, M. Oshima, M. Koizumi, A. Osa, A. Kimura, J. Goto, and Y. Hatsukawa, Appl. Radiat. Isot. 64, 751 (2006). 3. G. Steinhauser, J. H. Sterba, M. Bichler, and H. Huber, Appl. Geochem. 21, 1362 (2006). 4. Y. Toyama-Kato, K. Yoshida, E. Fujimori, H. Haraguchi, Y. Shimizu, and T. Kondo, Biochem. Eng. J. 14, 237 (2003). 5. G. A. Petrucelli, R. J. Poppi, R. L. Mincato, and E. R. Pereira-Filho, Talanta 71, 620 (2007). 6. A. M. Mouazen, M. R. Maleki, J. De Baerdemaeker, and H. Ramon, Soil Tillage Res. 93, 13 (2007). APPLIED SPECTROSCOPY 871

8 7. R. Noll, H. Bette, and A. Brysch, Spectrochim. Acta, Part B 56, 637 (2001). 8. J. Grubera, J. Heitza, and H. Strasser, Spectrochim. Acta, Part B 56, 685 (2001). 9. A. Jurado-López and M. D. Luque de Castro, Talanta 59, 409 (2003). 10. L. St-Onge, E. Kwong, and M. Sabsabi, Spectrochim. Acta, Part B 57, 1131 (2002). 11. P. Maravelaki-Kalaitzaki, D. Anglos, and V. Kilikoglou, Spectrochim. Acta, Part B 56, 887 (2002). 12. S. Klein, F. Fekrsanati, and J. Hildenhagen, Appl. Surf. Sci. 171, 242 (2001). 13. V. Lazic, F. Colao, and R. Fantoni, J. Cul. Heritage 4, 303 (2003). 14. R. Barbini, F. Colao, and V. Lazic, Spectrochim. Acta, Part B 57, 1203 (2002). 15. R. Barbini, F. Colao, and R. Fantoni, Appl. Phys. A 69, 175 (1999). 16. F. Capitelli, F. Colao, and M. R. Provenzano, Geoderma 106, 45 (2002). 17. F. Hilbk-Kortenbruck, R. Noll, and P. Wintjens, Spectrochim. Acta, Part B 56, 933 (2001). 18. R. T. Wainner, R. S. Harmon, and A. W. Miziolek, Spectrochim. Acta, Part B 56, 777 (2001). 19. J. H. Choi, C. J. Damm, and N. J. O Donovan, Appl. Spectrosc. 59, 258 (2005). 20. V. Lazic, R. Barbini, and F. Colao, Spectrochim. Acta, Part B 56, 807 (2001). 21. D. K. Ottesen and L. L. Baxter, Energy Fuels 5, 304 (1991). 22. H. S. Zhang, J. P. Singh, F. Y. Yueh, and R. L. Cook, Appl. Spectrosc. 49, 1617 (1995). 23. F. J. Wallis, B. L. Chadwick, and R. J. S. Morrison, Appl. Spectrosc. 54, 1231 (2000). 24. M. Kurihara, K. Ikeda, Y. Izawa, Y. Deguchi, and H. Tarui, Appl. Opt. 42, 6159 (2003). 25. D. Body and B. L. Chadwick, Rev. Sci. Instrum. 72, 1625 (2001). 26. L. Y. Yu, J. D. Lu, W. Chen, G. Wu, K. Shen, and W. Feng, Plasma Sci. Technol. 7, 3041 (2005). 27. C. E. Romero and D. Saro, Lehigh Energy Update 26 (2008). 28. L. Zhang, L. Dong, H. P. Dou, W. B. Yin, and S. T. Jia, Appl. Spectrosc. 62, 458 (2008). 29. D. Body and B. L. Chadwick, Spectrochim. Acta, Part B 56, 725 (2001). 30. A. Ciucci and M. Corsi, Appl. Spectrosc. 53, 960 (1999). 31. A. M. El Sherbini, T. M. El Sherbini, and H. Hegazy, Spectrochim. Acta, Part B 60, 1573 (2005). 32. H. H. Telle, D. C. S. Beddows, and G. W. Morris, Spectrochim. Acta, Part B 56, 947 (2001). 33. C. Argón, J. A. Aguilera, and F. Peñalba, Appl. Spectrosc. 53, 1259 (1999). 34. S. V. Loo and J. Koppejan, Eds., Handbook of Biomass Combustion and Co-Firing (Twente University Press, Enschede, Netherlands, 2002). 35. I. Obernberger and G. Thek, Proceedings of the 1st World Conference on Pallets, 115 (2002). 872 Volume 63, Number 8, 2009