Published in Waste Management, Vol 17, No 1, pp.79-86, 1997

Transcription

1 Published in Waste Management, Vol 7, No, pp.79-86, 997 LOW-TEMPERATURE PYROLYSIS OF CCA TREATED WOOD WASTE: CHEMICAL DETERMINATION AND STATISTICAL ANALYSIS OF METAL IN- AND OUTPUT; MASS BALANCES L. Helsen* and E. Van den Bulck Department of Mechanical Engineering, Katholieke Universiteit Leuven, Celestijnenlaan 3A, 3 Heverlee, Belgium. K. Van den Broeck and C. Vandecasteele Department of Chemical Engineering, Katholieke Universiteit Leuven, de Croylaan 46, 3 Heverlee, Belgium. Abstract Low-temperature pyrolysis is proposed as an alternative method to dispose of CCA treated wood waste. In the frame of a study aiming at optimising the pyrolysis of CCA treated wood, an experimental facility has been built to examine the influence of important process parameters (pyrolysis temperature, residence time, heating rate, particle size,...) on the release of metals and on the resultant mass reduction. In order to perform a mass balance calculation for the total system, a method for metal analysis was developed. Two leaching procedures and one dissolution procedure were tested and compared with each other, resulting in an optimal procedure to bring the metals into solution: the BSI method to determine the total amount of Cr, Cu and As in the dried wood and the Reflux method to determine the total amount of Cr, Cu and As in the pyrolysis residue. These results illustrate that Cr is more strongly bound in the pyrolysis residue compared to the CCA treated wood. The analytical technique used was ICP-MS and the analytical problems like interferences and matrix effects were solved by using the appropriate isotope, an internal standard and mathematical corrections. The resulting optimal technique for CCA treated wood ( the BSI method ) was applied to wood samples with different particle sizes. A statistical analysis of the Cr, Cu and As content in the CCA treated wood shows the heterogeneous character of CCA treated wood samples. Heterogeneity becomes less important when using samples with a small range of particle sizes. The smaller wood particles have significantly higher metal concentrations than the larger particles. Realistic mass balances for the metals were obtained and showed that most of the Cr, Cu and As remained in the pyrolysis residue. * author to whom correspondence should be addressed tel: fax: lieve.helsen@mech.kuleuven.ac.be

2 Introduction Waterborne salts have been used to preserve wood from insects, fungi and water damage for many years. One of the more common formulations contains copper, chromium, and arsenic salts and is known as chromated copper arsenate or CCA. After impregnation of the wood with a CCA solution, the metal compounds will be fixed to the cell walls of the wood matrix. Substantial amounts of CCA remain in the wood for many years, and the disposal of scrap CCA wood is a growing problem in Europe. In Germany and in France e.g. the total amount of wood waste is around 3-4 million ton per year of which.-.4 million is hazardous. In France about 5 million poles treated with CCA (railway, electricity and telephone) are in service. Every year 5 poles (5 ton) get out of service, which means that the waste disposal problem will last for at least 5 years without putting new poles in service. Telephone poles, railway sleepers, timber from landscape and cooling towers, wooden silos, hop-poles, cable drums and wooden play-ground equipment generate wood waste for which environmentally benign disposal technologies need to be developed. The number of waste disposal sites is decreasing and redundant poles, piling and lumber, which constitute a large volume of material, may not be accepted at the limited number of sites in the future. Burning this wood waste emits highly toxic smoke and fumes in the environment whereas conventional pyrolysis systems (fixed bed, batch or grate; fluidized bed; rotary kiln,...) operate at a too high temperature to prevent the release of metal vapours, and often require that the wood is chopped before processing. Percentages of arsenic volatilized have been reported to range between 8 and 95 %. Amounts of copper and chromium volatilized are not well documented. Public concern has been raised over the possible formation of toxic smoke when CCA wood is burned in wood stoves, fireplaces, or boilers. It has been reported that a rural family of eight had been heating their home with a wood burning stove fuelled with leftover scraps of wood treated with CCA. The entire family came down with symptoms of copper, chromium, and arsenic poisoning. A study has been set up to design a low-temperature pyrolysis facility for the CCA treated wood waste such that at least 9 % of the metals are contained in a concentrated solid product stream, and the pyrolysis gases and liquid are used to their maximum potential with respect to energy recuperation. The resulting charcoal would subsequently be separated in a clean solid residue with a high carbon content and a concentrated metal residue, which both still have some market value and do not have to be landfilled. The use of low temperatures and no oxidising agent results in lower loss of metals, compared to combustion. In the frame of this study an experimental, labscale test rig for the pyrolysis of CCA treated wood has been built to examine the influence of some process parameters on the release of metals and on the product yields. To perform this parameter study an accurate determination of the metal content in the input and output streams of the system is necessary to calculate mass balances. This study focuses on the determination of Cr, Cu and As in the CCA treated wood and its pyrolysis residues. To determine the total amount of Cr, Cu and As in a solid stream, the metals have to be brought in solution. Several

3 leaching and dissolution procedures were tested and compared with each other to find the optimal procedure. The final solution obtained from this dissolution process was measured by Inductively Coupled Plasma Mass Spectrometry (ICP-MS). The analytical problems like interferences and matrix effects from the acids used to dissolve the samples will be discussed and solved. The resulting optimal technique for CCA treated wood has been applied to replicate wood samples with different particle sizes to perform a statistical analysis of the Cr, Cu and As content in the CCA treated wood, illustrating the heterogeneous character of CCA treated wood samples. To examine the influence of the particle size on this heterogeneity, samples of only large particles and samples of only small particles are also subjected to the statistical analysis. Histograms, mean values, standard deviations and confidence intervals are used to estimate the importance of sampling errors due to the non-uniform distribution of the metals in the wood samples. Using the resulting optimal procedures to bring the metals in solution and using a homogeneous wood sample as input, mass balances for Cr, Cu and As were determined. Methods and materials Feedstock The feedstock used in the experiments is pine wood ( Pinus Sylvestris ) impregnated with type C CCA salt : 3.5% CuSO 4.5H O, 4.% Na Cr O 7.7H O and 6.4% As O 5.H O. The wood chips measure between and 35 mm in length, between and 7 mm in width and between.5 and mm in thickness. The impregnation has been carried out specially for these experiments to simulate a level, higher than ever reached: the wood is treated twice for class 4 impregnation (wood may be in contact with groundwater and sweet water, requiring a salt retention of 9 kg CCA solution per m 3 of wood) and peeling is used because this is easy to impregnate. To determine the metal concentrations oven-dry wood is used. The drying of the wood is accomplished by putting the wood in a conventional oven with air circulation at a preset temperature of C, until the weight of the wood sample is constant (drying time was approximately 9 minutes). The percentage of moisture, calculated from the difference in mass before and after drying, is between and %. Pyrolysis process The experimental facility for the pyrolysis of non-treated and CCA-treated wood particles is shown in figure. The reactor that holds the wood is a stainless steel circular tube with a diameter of 5 cm and a length of cm. A flow of heated nitrogen is forced through the reactor. The reactor tube is thermally guarded with a heating tape with a controlled heating power such that the reactor outlet temperature as well as the mean reactor temperature equal the inlet temperature. 3

4 At the outlet of the reactor the nitrogen flow contains non-condensable pyrolysis gases as well as microscopic tar droplets. The gas flow that exits the reactor is immediately quenched in a water scrubber. The cold, wet gas flow at the outlet of the scrubber forms an aerosol of microscopic tar and water droplets. This aerosol is forced through a platecolumn to improve the contact between the water droplets and the condensing pyrolysis gas. In the bottom vessel the liquid (water and condensed pyrolysis vapours) and gaseous (nitrogen and non-condensable pyrolysis gas) streams are separated. The liquid is recirculated with a circulation pump and is re-used to feed the water scrubber. The gas is forced through an extra ice/water condenser in which the condensate is collected and finally through a tube filled with cotton wool that acts as a filter. The resulting product streams to be analysed for metals after each (batch) experiment are: the pyrolysis residue (charcoal), the liquid scrubber stream (water mixed with condensed pyrolysis vapours) and the loaded cotton wool filter. The metal content of the gas stream is determined by difference. The dried CCA treated wood as well as the pyrolysis residue, resulting from a pyrolysis process at 35 C for minutes, are the subject of this comparative study. Sample dissolution methods The dried wood (about 5 mm in length, between 5 and mm in width and mm in thickness) and the pyrolysis residues were cut into smaller (uniform) pieces to improve contact with the reagents and thoroughly mixed to obtain as homogeneous a sample as possible. The spread in particle size during leaching was very small to obtain as uniform a leaching process as possible. The pyrolysis residues resulted from a pyrolysis process applied to a CCA treated wood sample containing chips between and 5 mm long, between and mm wide and between and mm thick. Two leaching procedures, based on existing procedures in literature, ( the BSI method and the ASTM method ) and one dissolution procedure ( the Reflux method ) are described below. The existing procedures were somewhat altered for experimental or analytical reasons. All the reagents used were pro-analysis products and ultrapure Milli-Q water (Millipore) was used for the dilutions. The procedures were carried out on a minimum of three samples and on a blank sample. All the glassware was soaked for minimum one night in % nitric acid and washed with milli-q water before use. The BSI method This procedure is based on the British Standard Method (BSI) for wood preservatives and treated timber for the quantitative analysis of Cu, Cr and As. The weighted sample ( gram of dried wood or.5 gram of pyrolysis residue) was transferred into a 5 ml conical flask and 5 ml of H SO 4 (.5 mol dm -3, pro analysis) together with ml of H O (3 % v/v, pro analysis) was added. The flasks were covered and heated at 75 C in a water bath for hour with occasional swirling to mix the contents of the flask. The flask was then removed from the water bath and ml of milli-q water was added. After cooling to room temperature, the solution was filtered through a glass-fibre filter and the filtrate was transferred into a 5 ml volumetric flask and diluted with milli-q water to the mark. 4

5 The ASTM Method The ASTM method 3 (American Society for Testing and Materials) covers the determination of Cr, Cu, Mn, Ni, Pb, V and Zn in coal ash and in coke ash, but is also useful for the determination of other elements. The samples ( gram of dried wood or.5 gram of pyrolysis residue) were placed in a PTFE beaker and ml of aqua regia was added. The beaker was then heated for hour on a heating plate. After heating, the beaker was cooled to room temperature and 5 ml of HF was added and heated to nearly dry. Again 5 ml of aqua regia was added and the beakers were heated to evaporate the HF. After cooling this mixture, ml of milli-q water was added and slightly heated. The mixture was then filtered through a glass-fibre filter and the filtrate was transferred into a ml volumetric flask and diluted with milli-q water to the mark. The Reflux Method This method was developed in this laboratory. The purpose of this method is to dissolve the sample completely in nitric acid without loss of elements. The sample (. gram of dried wood or pyrolysis residue) was transferred into a conical flask equipped with a water cooler to prevent loss by volatilisation during the dissolution process, and ml of HNO 3 (suprapur, 65%) was added. The flask was then heated on a heating plate until the sample was dissolved completely (approximately hours for the pyrolysis residue and 4 hours for the CCA treated wood). Statistical analysis applied to the metal concentrations in the CCA treated wood To assess the magnitude of sampling errors, due to the non-uniform distribution of the metal salts in the wood, the BSI method was applied to several replicate dried CCA treated wood samples to allow a statistical analysis. Eighteen mixtures of large and small particles (randomly chosen) are leached according to the BSI method. Six samples of only large particles and six samples of only small particles are also subjected to the BSI leaching procedure to examine the influence of the particle size. The dimensions and mass ranges of these treated wood particles are given in table. The metal content of the resulting leachates is determined with ICP-MS. Statistical analysis, using histograms, mean values, standard deviations, confidence intervals,... is performed on these resulting metal concentrations. In order to overcome any significant sampling error the samples have to be sufficiently large ( gram of dried CCA treated wood is used) such that the variation between samples due to the non-uniform distribution of the metal salts in the wood, is negligible compared to the standard deviation estimated for the leaching and analysis, i.e. between 3 and 4 %. ICP-MS measurements, spectral interferences and matrix effects. The ICP-MS used in this work is a PlasmaQuad PQ+ (Fisons Instruments) equipped with a conventional Meinhard nebuliser for sample introduction. The instrumental parameters are given in table. Because the dissolution of the dried wood and its 5

6 pyrolysis residues requires large amounts of acid, this may result in polyatomic interferences. For HCl, for example, interferences of ClO with 53 Cr and of ArCl with 75 As occur. Many of these interferences can be avoided by choosing another isotope for quantification, e.g. 5 Cr instead of 53 Cr. In the case of As, which is mono-isotopic, this is not possible and only mathematical correction methods can give a solution to this problem. The mathematical correction for the interference of ArCl on As is based upon the fact that the ratio of the signal for 4 Ar 35 Cl to that for 4 Ar 37 Cl will be equal to the ratio of the abundance of the two isotopes 35 Cl and 37 Cl, this means 75.77/4.3. Measuring the signal of 4 Ar 37 Cl at mass 77 ( 77 Se) makes it possible to calculate the signal of 4 Ar 35 Cl present at mass 75 ( 75 As). Correction for 77 Se will be based on the 8 Se signal. The following equation is therefore used in this work: 75 As = 75 Signal ( 77 Signal - ( Signal)) Also matrix effects play an important role in ICP-MS. A first cause of matrix effect is the deposition of solids on the entrance orifice, which may cause instrumental drift. For this reason, solutions are diluted so that the concentration of dissolved solids is lower than g l -. Since the As, Cr and Cu concentrations are high enough after applying the dissolution procedures, it is possible to dilute so far that matrix effects are negligible and only an internal standard is added to correct for instrumental drift. Therefore the solutions resulting from the BSI method are diluted times and the solutions resulting from the Reflux method times. Indium, used as internal standard, is added in known and equal concentrations to the samples and standard solutions. Results and discussion Determination of the total concentration of Cu, As and Cr in CCA treated wood. The ICP-MS results for the total concentration of As, Cu and Cr in the dried wood obtained after subjecting the samples to the different dissolution and leaching procedures are shown in table 3. Because of the possible interference of 33 S 6 O 6 O from H SO 4, used in the BSI method, on 65 Cu, 63 Cu was chosen as the optimal isotope. The HCl used in the ASTM method can give interference of 37 Cl 6 O on 53 Cr and 4 Ar 35 Cl on 75 As, therefore the 5 Cr isotope was chosen and the results of As had to be corrected using the previously described equation for mathematical correction. The ASTM method and the BSI method were carried out on 3 samples, while the Reflux method was applied to only sample. The results were corrected with a blank and the standard deviations given in table 3 are the standard deviations between the 3 samples. The results for the ASTM, BSI and Reflux method procedures were in good agreement for all three elements. Since the BSI method is the easiest to handle, this method is chosen as the optimal method to bring the metals in solution starting from the dried CCA treated wood. Determination of the total concentration of Cu, As and Cr in the pyrolysis residues from CCA treated wood. The ICP-MS results for the total concentration of As, Cu and Cr in the pyrolysis residues from CCA treated wood obtained after subjecting the samples to the different dissolution 6

7 and leaching methods are also shown in table 3. The results for As and Cu obtained with the ASTM, BSI and Reflux methods are in satisfactory agreement. The results for Cr obtained with the BSI method are low compared to the other results, probably because Cr is not completely brought in solution when using the BSI method. This was verified by subjecting the filtered residues of the BSI method and the ASTM method to the Reflux method. Table 4 shows the results of this test and confirms that there is still a large amount of Cr left in the residue obtained after the BSI method, compared to the amount of Cr left in the residue obtained after the ASTM method. The BSI method is thus not suitable to bring the metals into solution starting from the pyrolysis residue, indicating a stronger bond between Cr and the pyrolysis residue compared to the CCA treated wood. A more aggressive leaching procedure or dissolution method (for example the Reflux method ) is necessary. Statistical analysis applied to the metal concentrations in the CCA treated wood Statistical analysis applied to the mixtures of large and small wood particles The histograms, presenting the ICP-MS results for the total concentration of Cr, Cu and As in the 8 mixed wood samples are shown in figure. The mean values, ranges and standard deviations are summarised in table 6. These results illustrate the wide distribution of the metal content for all three metals. The range related to the mean value varies between 6 and 9 %. The distribution is rather flat than normal, indicating that each of the values has almost the same probability to represent the total metal concentrations. The variation between the different samples is too large to neglect (standard deviations are between 5 and 6 %). To examine whether the particle size is an important parameter in obtaining homogeneous samples, replicate samples of only large and only small particles (table ) have been subjected to the BSI leaching procedure. Statistical analysis applied to large particles The histograms, presenting the ICP-MS results for the total concentration of Cr, Cu and As in the 6 large-particle-wood samples are shown in figure 3. The mean values, ranges and standard deviations are summarised in table 6. From figures and 3 it can be seen that the total metal concentrations of the large-particle-wood samples all are significantly lower than those for the mixed samples. The standard deviation obtained for the large particles is reduced to 3 %, similar to the standard deviation estimated for the leaching and analysis procedure. Statistical analysis applied to small particles The histograms, presenting the ICP-MS results for the total concentration of Cr, Cu and As in the 6 small-particle-wood samples are shown in figure 4. The mean values, ranges and standard deviations are summarised in table 6. From figures 3 and 4 it can be seen that the total metal concentrations of the small-particle-wood samples are significantly higher than those for the large-particle-wood samples. The difference between the small- 7

8 particle-wood samples and the mixed samples (figures and 4) is not as pronounced as with the large-particle-wood samples because there is some overlap but the concentrations of the small-particle-wood samples are clearly located at the upper part of the mixed samples. The standard deviation obtained for the small particles is reduced to 3 % or even % for As, similar to the standard deviation estimated for the leaching and analysis procedure. Discussion From the results of the statistical analysis it can be seen that the smaller particles have significantly higher Cr, Cu and As concentrations than the larger particles and a mixture of large and small particles has metal concentrations falling in between. The particle size has thus certainly an influence on the metal content (expressed as mass of metal per mass of dry wood) of CCA treated wood. This phenomenon could be explained by assuming that the CCA treatment is a surface treatment. The smaller particles have more external surface area than the larger particles, resulting in higher metal concentrations. However, the producers of CCA treated wood products claim that the treatment with an impregnation solution (accomplished by a vacuum-pressure-vacuum procedure) is an indepth-treatment (at least 6 mm beneath the surface) instead of a surface treatment. Another explanation can be found in the wood structure and the nature of metal compounds within the wood 5. Low-aged wood contains a lot of sugars and has a very porous structure, while older wood contains less sugars and is less porous. Since the metals in the wood are bound to these sugar compounds 6 (lignin and cellulose compounds in the cell wall), young wood would contain more metals than old wood. When a size reduction method is applied to the wood, the weakest bonds (mechanical and chemical) will be broken first. Low-aged wood is less strong than older wood and will deliver the largest proportion of small particles, resulting in the phenomenon that the smaller particles contain more metals per unit of weight than the larger particles. The spread of metal concentration in different CCA treated wood samples is small enough to perform pyrolysis parameter studies if samples containing particles of almost the same size are used. Then the sampling error (including analysis error) is of the same magnitude as the standard deviation caused by the leaching and analysis method, i.e. approximately 3 %. A less homogeneous sample can not be used for the pyrolysis parameter study in order to obtain accurate mass balance calculations. The problem of different metal concentrations in samples of different particle sizes will be carried through the pyrolysis residues. Their particle size will depend on the particle size of the starting CCA treated wood material. The high spread in the metal concentrations may be due to the heterogeneous character of peeling. Treated sapwood may deliver more homogeneous wood samples, and will also be subjected to the statistical analysis procedure. Metal mass balances 8

9 Realistic mass balances for the metals can be obtained using the resultant optimal procedures to bring the metals into solution (the BSI method for the dried wood and the Reflux method for the pyrolysis residue) and using homogeneous wood samples as input. The percentages Cr, Cu and As found in the CCA treated wood, the pyrolysis residue, the liquid product, the filter and the gas stream in a typical pyrolysis experiment are represented in table 5. The liquid product can directly be analysed by ICP-MS after dilution and addition of the internal standard, while the filters have been subjected to the BSI leaching procedure. The metal content of the gaseous product is calculated by difference. Pyrolysis experiments carried out at different temperatures and residence times are described elsewhere 4, together with the metal mass balances. From table 5 it can be seen that low-temperature pyrolysis of CCA treated wood is a promising technique, although not yet optimised. The percentages Cu and Cr volatilized are low enough to conclude that they will pose no problems under the process conditions used. As, on the other hand, is released for about 5 %, which is still too high. The percentage mass reduction obtained under these process conditions is 63 % relative to the non-dried CCA treated wood. Further optimisation of the process conditions to obtain a higher retention of As is being carried out. Conclusions To determine the total amount of Cr, Cu and As in the dried CCA treated wood, the BSI leaching method seems to be the most suitable method: reproducible, easy to handle and it guarantees that all metals are brought into solution. Since Cr is not completely dissolved when applying the BSI method to the pyrolysis residue, a more aggressive leaching procedure or a dissolution method (for example the Reflux method ) is necessary. The statistical analysis applied to the total amounts of Cr, Cu and As in several replicate dried wood samples illustrates the heterogeneous character of CCA treated wood. The spread in metal content can be reduced (from 6 to 3 %) when using samples of the same size, the smaller particles having a significant higher metal content for all of the three metals compared to the larger particles. Realistic mass balances for the metals can be obtained using the resulting optimal procedures to bring the metals into solution. These mass balances illustrate that the lowtemperature pyrolysis is a promising alternative to dispose of CCA treated wood waste. Acknowledgements The authors would like to thank Herman Cooreman, Catherine Celens and Stijn Vermeulen for their collaboration in this research, especially for the dissolution procedures. We also thank Beaumartin S.A. and Mr. Hery in particular for the financial support and the wood samples. 9

10 References. Mc Mahon, C.K., Bush, P.B. and Woolson, E.A., Release of copper, chromium, and arsenic from burning wood treated with preservatives. Paper presented at the 78th annual meeting of the air pollution control association, Detroit, Michigan, 6- June, British Standard Methods of analysis of wood preservatives and treated timber, part 3: quantitative analysis of preservatives and treated timber containing copper/chromium/arsenic formulations, BS 5666 Part 3, British Standard Institution, London, American Standard Test Method Designation D (reapproved 989), Standard Test Method for Trace Elements in Coal and Coke Ash by Atomic Absorption, Philadelphia, PA. 4. Helsen, L. and Van den Bulck, E., Release of metals during the pyrolysis of preservative impregnated wood. In Developments in Thermochemical Biomass Conversion, Vol., ed. A.V. Bridgwater and D.G.B. Boocock. Chapman & Hall, London, 997, pp Hery, J.S., Beaumartin S.A., Personal Communication, Pizzi, A., The Chemistry and Kinetic Behaviour of Cu-Cr-As / B Wood Preservatives. IV Fixation of CCA to Wood. Journal of Polymer Science: Polymer Chemistry Edition, 98,,

11 Tables and figures Table : Dimensions and mass ranges of the wood particles used for the statistical analysis: large particles, small particles and mixtures of both. large particles mass / g length / mm width / mm thickness / mm small particles mass / g length / mm width / mm thickness / mm mixture of small and large mass / g length / mm width / mm thickness / mm minimum maximum Table : Instrumental parameters for the PQ+ with the Meinhard nebuliser. instrumental parameter Meinhard nebuliser nebuliser gas flow / l min -.8 nebuliser pressure / 5 Pa.3 cool gas flow / l min - 3 auxiliairy gas flow / l min -.9 sample uptake rate / ml min - wash time / uptake time (s/s) 6/9 data acquisition time / s 6 Table 3 : Total concentration (µg g - ) of As, 63 Cu and 5 Cr in CCA treated wood and its pyrolysis residue obtained using ICP-MS after the different leaching ( BSI and ASTM ) and dissolution ( Reflux ) methods. As Cu Cr Method Wood Pyrolysis residue Wood Pyrolysis residue Wood Pyrolysis residue ASTM 844 ± 4 ± ± 6 85 ± 9 9 ± 3 7 ± 58 BSI 859 ± ± ± 3 86 ± 69 9 ± 4 87 ± 3 Reflux

12 Table 4 : Results (µg g - ) for 5 Cr obtained with the Reflux method on the residue obtained with the BSI method and the ASTM method. Method Cr BSI 87 ± 3 Reflux on residue of BSI 549 ± 5 ASTM 7 ± 58 Reflux on residue of ASTM 3 ± Table 5 : Percentages 5 Cr, 63 Cu and As in the CCA treated wood, the pyrolysis residue, the liquid product, the filter and the gas stream. Cr Cu As CCA treated wood / % pyrolysis residue / % liquid product / % filter / %.6.4. gas / % (*) (*) calculated by difference Table 6 : Mean values, ranges, and standard deviations obtained from the statistical analysis applied to a mixture of small and large particles, only large particles and only small particles. Cr mean value / µg g - range / µg g - standard deviation absolute / µg g - relative / % Cu mean value / µg g - range / µg g - standard deviation absolute / µg g - relative / % As mean value / µg g - range / µg g - standard deviation absolute / µg g - relative / % small and large large small

13 Figure : The experimental labscale facility for the pyrolysis of CCA-treated wood particles. 3

14 Figure : Histogram presenting the ICP-MS results for the total concentration of Cr, Cu and As in the 8 mixed wood samples chromium concentration (µg chromium / g dry wood) copper concentration (µg copper / g dry wood) arsenic concentration (µg arsenic / g dry wood) 4

15 Figure 3 : Histogram presenting the ICP-MS results for the total concentration of Cr, Cu and As in the 6 large-particle-wood samples chromium concentration (µg chromium / g dry wood) copper concentration (µg copper / g dry wood) arsenic concentration (µg arsenic / g dry wood) 5

16 Figure 4 : Histogram presenting the ICP-MS results for the total concentration of Cr, Cu and As in the 6 small-particle-wood samples chromium concentration (µg chromium / g dry wood) copper concentration (µg copper / g dry wood) arsenic concentration (µg arsenic / g dry wood) 6