- Email: [email protected]
Unburned carbon in combustion residues from solid biofuels
Fuel 117 (2014) 890–899
Contents lists available at ScienceDirect
Fuel journal homepage: www.elsevier.com/locate/fuel
Unburned carbon in combustion residues from solid biofuels H. Bjurström a, B.B. Lind b,⇑, A. Lagerkvist c a
ÅF-Industry AB, SE-169 99 Stockholm, Sweden Swedish Geotechnical Institute, SE-412 96 Göteborg, Sweden c Luleå University of Technology, SE-971 87 Luleå, Sweden b
h i g h l i g h t s Different methods to determine unburned carbon in ash yield slightly different results. Unburned carbon is mostly elemental carbon, very little volatile or semi-volatile organic carbon. Raman results show that unburned carbon is similar in bottom ash and in fly ash.
a r t i c l e
i n f o
Article history: Received 20 November 2012 Received in revised form 7 August 2013 Accepted 10 October 2013 Available online 22 October 2013 Keywords: Combustion residue Ash Solid biofuels Unburned carbon Elemental carbon
a b s t r a c t Unburned carbon (UC) in 21 combustion residues from solid biofuels has been examined using several methods of analysis (including LOI and TOC) as well as micro-Raman spectroscopy. The concentration of unburned carbon in the residues varied over an order of magnitude and in several samples accounted for about 10% of the ash mass. It was observed that TOC had a poor correlation to organic carbon, especially for fly ashes. LOI at all tested temperatures showed a better correlation than TOC to the organic carbon content, whereas the TOC is better correlated to elemental carbon. LOI550 gave a larger variation and a less complete mobilisation of unburned carbon than LOI at 750 or 975 °C did, but at the highest temperature metal oxidation was notably affecting the mass balance to the extent that some samples gained mass. For this reason, and of the temperatures tested, LOI750 seem to be the most stable indicator for organic remains in the incineration residuals. Most of the unburned carbon is elemental, and only slowly degradable, so the potential emissions of organic compounds from ashes should not be assessed by using a TOC test. The structure of the detected elemental carbon in UC is similar to that of activated carbon, which indicates a potentially large specific surface. This should be borne in mind when assessing the environmental impact of using ash for different purposes, including use as a construction material. Field studies are needed to verify the actual impact as it may depend on environmental conditions. Ó 2013 Elsevier Ltd. All rights reserved.
1. Introduction The presence of unburned carbon (UC) in solid combustion residue not only hints at imperfections in the combustion process but also raises fears that the residue might release potentially noxious organic substances. However, examination of unburned carbon in ash from the combustion of different fuels has shown that unburned carbon is largely elemental carbon (EC) rather than organic carbon (OC) (among others [1–4]). Elemental carbon, or charcoal, is chemically stable and degrades very slowly in a natural environment – after several thousand years its structure is mostly unaffected (see e.g. Cohen-Ofri et al. [5] and Spokas [6]). It should therefore be noted that the unburned carbon detected in combus-
⇑ Corresponding author. Tel.: +46 317786566. E-mail address: [email protected] (B.B. Lind). 0016-2361/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.fuel.2013.10.020
tion residue is essentially something other than the presence of organic reactive carbon. There are several routine methods to determine the content of unburned carbon in combustion residue. These methods are also used for other materials, such as determining the content of organic matter in soil or water. While using the same method of analysis is desirable, some confusion may arise when interpreting the results, as different materials respond in different ways. Loss on ignition, or LOI, is a convenient method to determine what could have burnt provided the combustion residue has not absorbed water following extraction from the furnace, in which case LOI would consist of unburned carbon, carbonate carbon, hydroxide water or even hydration water [7,8]. LOI is carried out at a variety of temperatures depending on the fuel; 550 °C for biomass and 750 °C for coal [9] or the envisaged area of use, 950 °C for cement, [10]. The lowest temperature is used for biomass to ensure that potassium and chlorine are not counted as oxidisable carbon.
891
H. Bjurström et al. / Fuel 117 (2014) 890–899
LOI is not recommended for use with air pollution control (APC) residue from waste-to-energy plants because these materials are very hygroscopic and the adsorbed water will affect the LOI-value [11]. The suitability of the ash material as a silica source in concrete is checked by igniting at 950 °C which is supposed to reveal bound water and carbonates. On the other hand, TOC yields the carbon content. Although ASTM calls TOC ‘Total Oxidisable Carbon’ [12], the expression ‘Total Organic Carbon’ is often used for combustion residues, similar to the use of this method to determine organic content in soil or water [13]. It has been shown [3] that a large proportion of soil carbon may also be elemental. The ASTM definition will be used here. A factor that complicates the interpretation of thermal or chemical analyses is the complexity of combustion residues as they represent a matrix that may be involved in the processes in which carbon reacts. Although they are products of oxidation, i.e. combustion, their composition hints at reduced conditions when they are extracted. The particle size distribution may be quite broad and not all components in the larger particles may have time to react during the process. The present paper describes an examination of UC, determined using different analytical methods on solid residues from Swedish combustion plants, complemented by a study of the samples using a spectroscopic method. Three questions are addressed in the paper: What is the carbon content of combustion residues and how can it be assessed? How much of the unburned carbon is elemental and what is the remainder? What are the properties of EC in UC? As the fuels used in Swedish combustion plants are largely solid biofuels, the focus of this paper is on combustion residues from solid biofuels. Samples were chosen to provide a wide range of unburned carbon content.
air pollution control residue), different fuels (wood chips and pellets, bark, waste wood, municipal solid waste), large plants and small plants and different industrial sectors (cogeneration plants, heating plants, pulp and paper mills). These include efficient combustors as well as plants that are not as efficient as they should be, see Table 1 for a summary. Samples 1, 2, 5, 6, 9, 10 and 12 are bottom ash, which is usually handled separately from fly ash in large plants, whereas mixing fly ash and bottom ash (samples 13–20) is more common in smaller plants. The remaining samples are fly ash or APC residue. In the latter, samples 11 and 21, additives are also found, including activated carbon. With regard to the bottom ash samples, four of the plants were sampled for fresh ash and three were sampled for aged ash. In the micro-Raman investigation, a sample of activated carbon, DioxSorbBP2, was obtained from Jacobi Carbon Systems and used as reference material. 2.1. Sampling and sample treatment Some of the samples of combustion residues were gathered for this specific investigation and some originate from other studies. Sampling procedures vary widely and the samples should therefore not be considered representative in a regulatory context of the residues produced in these plants. However, collectively they should represent a cross-section of combustion residues that arise in real situations. All samples were first homogenised at the laboratory and subsamples were ground to <100 lm using a disc stainless steel grinder. 2.2. Analytical methods 2.2.1. Total solids Total solids were determined by weighting after drying at 105 °C for at least six hours.
2. Materials and methods The 21 samples from 18 combustion plants studied in this investigation represent various types of furnaces (grate, fluidised bed, pulverised fuel with burners). Also represented are different types of residue (bottom ash, various types of fly ash, mixed ash,
2.2.2. Loss on ignition There are numerous procedures for the determination of volatile substances, mainly aimed at quantifying oxidisable substances in a sample. Using different standards, there is considerable variation in the temperature at which ignition takes place.
Table 1 A summary of the ash types sampled. Sample number
Combustion plant
Type of furnace
Type of combustion residue
Fuel
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21
Händelö P11 Händelö P11 Händelö P11 Händelö P11 Dåva Braviken Skellefteå Iggesund Wargön Avesta Nynäshamn Öresundskraft Bureå Burträsk Jörn Kåge Byske PC Vindan Norsjö Boliden Händelö P14
Grate Grate Grate Grate Grate Grate CFB Grate Grate Grate BFB PF Grate Grate Grate Grate Grate Grate Grate Grate CFB
Fresh bottom ash Aged bottom ash Cyclone ash Baghouse filter ash Aged bottom ash Aged bottom ash ESP ash ESP ash Bottom ash Bottom ash Baghouse filter (APC residue) Bottom ash Mixed Mixed Mixed Mixed Mixed Mixed Mixed Mixed Baghouse filter (APC residue)
Waste wood and rubber Waste wood and rubber Waste wood and rubber Waste wood and rubber MSW Bark, de-inking sludge Wood chips Bark Bark MSW Waste wood Biomass Wood pellets Wood pellets Wood pellets Wood pellets Wood pellets Wood pellets Wood pellets Wood pellets Sorted waste
Nominal power MW fuel tires tires tires tires
70 70 70 70 77 66 92 70 24 15–18 24 220 <5 <5 <5 <5 <5 <5 <5 <5 75
892
H. Bjurström et al. / Fuel 117 (2014) 890–899
Here the combustion residues have been tested at 550 °C [9], 750 °C ([9] for mineral fuels) and 975 °C [14], in order to cover the normal spectrum. The samples, 1–2 g, were heated up slowly to the LOI temperature (60 min) and were then held at the final temperature for a further 60 min. The samples were left to cool in a desiccator before being reweighed.
2.2.3. Total oxidisable carbon The indirect method in EN 13137 [13] was used here: Inorganic carbonate was determined by means of emitted carbon dioxide when acid was added to part of the sample. Total carbon was determined by heating a 200 mg sample to 1000 °C and measuring the emitted carbon dioxide. Total oxidisable carbon was obtained by subtracting inorganic carbon from total carbon.
2.2.4. Colorimetry Chemical digestion is commonly used in the analysis of organic soil content. It was performed using dichromate sulphuric acid according to ISO 14235 [15]. A Cr(VI) solution is added to the solid sample and allowed to react with the sample for several hours. The Cr(III) content is then determined spectrophotometrically. The response is calibrated using glucose solutions. This method supposedly yields a concentration of organic carbon in the sample. It is reported as organic matter content by multiplying the carbon content by a factor of 1.724. It is thus assumed that carbon enters for 58 mass per cent of the organic matter in the sample. This is of course a rough estimate and different organic compounds may differ in their degradability. It has also been shown that iron and sulphur compounds, for example, may also contribute to the observed reduction in Cr(VI) [16].
2.2.5. Thermogravimetry Thermogravimetry, thermal analysis and mass spectrometry together provide information on weight loss over a temperature range (i.e. decomposition or combustion of a chemical compound), the heat balance and the species released from the sample during the process. All this information provides indirect evidence of the speciation of an element in the samples. All tests were performed with the sample exposed to a stream of air that acted as an oxidiser. A release of carbon dioxide indicates that a combustible, carbon-containing substance has burned or that a carbonate has been decomposed. When a release of carbon dioxide through combustion is not accompanied by a release of water vapour, it has been inferred that the carbon in the sample is predominantly elemental rather than organic; see also Ferrari et al. [2]. The thermobalance used in this investigation is Netzsch STA 409C, equipped with simultaneous thermogravimetric analysis (TGA), differential thermal analysis (DTA) and a quadrupole mass spectrometer (QMS) for analysing the evolved gases. A 50–100 mg sample is heated in a flow of air, 200 ml/min, from room temperature to approximately 1000 °C at a rate of 10 °C per minute. The following observations were recorded:
The weight of the sample. The set temperature. The DTA signal. QMS signals in the range 1–100 Da (g/mol).
The software accompanying the thermobalance was used to determine weight loss from TG measurements and the location (temperature) of step changes in weight.
To determine the proportion of OC and EC in the combustible carbon, the following method was used to distinguish between step changes in the TGA curves: Combustion of EC and OC is identified by means of exotherms in the DTA curve and decomposition of an inorganic carbonate (IC) by means of an endotherm (the temperature at which these processes take place varies depending on the type of ash). Combustion of OC is accompanied by an emission of water and carbon dioxide while combustion of EC is accompanied only by an emission of carbon dioxide. OC is generally found at lower temperatures than EC. When the transition between OC and EC steps in the TGA curve is too indistinct, cessation of water release and/or inflexion in the DTA curve are taken to mark the end of ‘OC combustion’. 2.2.6. Micro-Raman spectroscopy Micro-Raman is used routinely to study the structure of carbonaceous materials, e.g. carbon nanotubes, char and soot. A Renishaw Invia Raman Spectrometer was used. Its lower limit for spatial resolution is of the order of 1 lm. Two alternative laser sources were used to excite the sample: A Renishaw 300 mW HPNIR diode laser with a wavelength of 785 nm in combination with a 1200 lines/mm diffraction grating, which provides a linear image of the sample. A Laser Physics 20 mW argon ion laser with a wavelength of 514 nm and a 2400 lines/mm diffraction grating, which provides a specular image of the sample The spectrometer was calibrated daily against a reference silicon sample. A sample of about 2 g was mounted on a circular holder (diameter approximately 10 mm). The first step in the investigation was to produce a Raman spectrum in the range 800–1700 cm1 for five spots on the surface of the mounted sample, choosing areas with an average appearance under the microscope as well as areas differing from the average. The 785 nm laser was used (its wavelength is suited to transitions in organic substances) at 5% of its maximum power in order to avoid burning away any organic residues. To compensate for the low sensitivity, data were gathered over a comparatively long period. When peaks that were not elemental carbon were observed in a recorded spectrum, the spot on the sample was investigated further using lasers and spectra were recorded in an extended range of 200–3200 cm1. 2.2.7. GC/MS analysis of organic substances Semi-volatile compounds in extracts of the samples were screened semi-quantitatively using GC/MS. Two extracts were prepared, one for non-polar compounds by leaching with dichloromethane in a Soxhlet unit and the other for polar compounds by acidifying a suspension of ash with sulphuric acid, extracting with diethyl ether, drying the extract and derivatising with a silylating reagent (N,O-bis-trimethylsilyl-trifluoracetamide, trimethylchlorosilane and pyridine in the proportions 10:1:10). Compounds were identified using a database of mass spectra. 3. Results and discussion 3.1. The carbon content of the different combustion residues Results for the most commonly used variables, i.e. the loss on ignition, LOI or VS (volatile solids), the total oxidisable carbon, TOC, and colorimetrically determined organic carbon, are pre-
893
H. Bjurström et al. / Fuel 117 (2014) 890–899 Table 2 LOI, TOC and colorimetric OCcol (w/w %) of the sampled combustion residues. Temperature of LOI in °C. Residue stream
Sample
LOI550
LOI750
LOI975
TOC
OCcol
Bottom ash Bottom ash, aged Bottom ash, aged Bottom ash, aged Bottom ash Bottom ash Bottom ash Fly ash and APC-residues Fly ash and APC-residues Fly ash and APC-residues Fly ash and APC-residues Fly ash and APC-residues Fly ash and APC-residues Mixed fly and bottom ash Mixed fly and bottom ash Mixed fly and bottom ash Mixed fly and bottom ash Mixed fly and bottom ash Mixed fly and bottom ash Mixed fly and bottom ash Mixed fly and bottom ash
1 2 5 6 9 10 12 3 7 8 4 11 21 13 14 15 16 17 18 19 20
18.5 22.6 3 3.4 3.3 12.4 91.1 12.4 0.8 17.2 17.4 6.6 1.1 2.6 10 11.6 11.7 1 11.4 14 1.6
18.8 24.2 3.2 5.5 3.6 13.3 92 12.3 0.8 25.4 26.2 10.1 6.7 8.8 20.6 16.8 18.5 9.6 17.5 21.7 9.3
19.4 24.1 3.2 5.7 2.4 13.8 92 13.6 1.1 27.3 31.2 12.2 15.1 11.6 21.6 19.1 21.4 11.3 18.8 24.3 10.7
20.2 21.4 0.6 0.4 1 10.2 86.1 12 0.4 15.4 12.2 10.6 1 1.4 11.5 5.8 13 2.9 6.9 11.1 0.8
16.4 17.0 1.0 0.4 1.9 10.3 42.3 12.7 0.2 10.0 12.8 8.6 3.8 0.6 8.1 5.4 6.1 0.5 5.1 7.4 0.7
13.0 146 9, 13 74
17.4 108 13, 7 57
19.0 98 15, 4 55
11.7 157 7, 94 85
8.2 116 6, 45 84
Global average Global standard deviation Average excluding no 12 Std dev. excl. sample 12
% %
Table 3 Average, hxi and standard deviations, s, (w/w %) of the differences DLOI and (TOC–OCcol) depending on residue type. Differences and standard deviations given as % of LOI750, LOI975 and TOC respectively. Residue stream
(750–550°) hxi
s
(975–750°) hxi
s
TOC– OCcol hxi
s
Bottom ash Fly ash and APC residue Mixed fly and bottom ash
11.3 30.5 54.0
13.4 30.8 23.5
6.7 22.1 12.5
21.3 17.9 5.9
19.7 31.2 37.7
46.9 124 25.2
sented in Table 2, where the samples are listed according to the residue stream. LOI was determined at three temperatures, the lower temperature corresponding to the standard temperature used for biomass and the higher temperature being the standard temperature applied in TOC analysis. For each sample, the LOI value shown in Table 2 increase as the temperature rises although the increments between different temperatures vary. These increments are small for many samples, although samples 13, 17, 20 and 21 show a relatively minor mass loss at the lower temperature together with a much larger mass loss at the higher temperature. All the other samples show LOI values from 46% of the total LOI or more in the LOI550 step, but these four samples have 22% or as low as 7% of the weight loss at 550 °C. As also indicated by the large variation of LOI550 results as compared to those of higher temperature, this contrast may indicate that LOI550 is a less stable determination of LOI in combustion residues. At the highest tested temperature, 975 °C, a gain of mass was apparent for samples 2 and 9, which may be interpreted as a metal oxidation overshadowing the mobilisation of elemental and inorganic carbon. This is typical for bottom ashes, and all such samples show no or very low LOI at the highest temperature step, so using this elevated temperature will probably lead to an underestimation of total carbon emissions. The LOI should reflect as much as possible the amount of unburned material in the samples. But the obtained results of LOI550 are similar to, and sometimes even lower, than the different carbon content measurements. Thus the LOI550 is obviously not representing all the unburned material. The incomplete combustion at 550 °C is also indicated by comparison of standard deviations for all samples at the different
temperatures. This indicate that the LOI at 550 °C is more susceptible than the higher temperatures to random variations, e.g. of particle size distribution, sample size etc. However, the results from 975 °C are also problematic due to the oxidation of metals and the associated gain of mass. In conclusion the LOI750 seem to be the most stable indicator of carbon losses of the LOI tests used. The combustion plant from where sample 7 was taken is exceptionally efficient, yielding an almost totally mineralised residue, whereas sample 12 represents exceptionally poor combustion performance. This furnace is a pulverised fuel furnace, originally fired with coal but now converted to biomass. The high content of unburned carbon in its bottom ash is due to pre-treatment limitations, resulting in fuel particles that are too large. As a unique sample, it will be excluded from further comparisons but it serves as a reminder of the potentially large variation in bottom ash quality. The variation in data for different residue types is summarised in Table 3. As can be seen, a large change occurs between LOI550 and LOI750 while the change between LOI750 and LOI975 is smaller. For bottom ash, average mass for all samples increases above 750 °C rather than decreases. This is mainly due to sample 9. As can be noted in Table 4, the averages of LOI and TOC for the different groups of plants, fuels, furnaces and residue streams, vary considerably between the different types of combustion residues and for all analyses. The standard deviation of the averages is of the same order of magnitude as the average itself, which also indicate a wide spread of the results within each group. At the high end, the waste wood and tyres fuel fractions seem to generate samples with more unburned carbon. At the lower end, we find wood and bark fuel. There is also quite a large difference
894
H. Bjurström et al. / Fuel 117 (2014) 890–899
Table 4 Average and standard deviations (s) of LOI and TOC (w/w %D.S.) grouped by residue types, fuels, combustor types and capacity of the latter. Residue stream
LOI550
s
LOI750
s
LOI975
s
TOC
s
Bottom ash (n = 6) Fly ash and APC (n = 6) Wood and rubber tyres (n = 4) MSW (n = 3) Wood and bark (n = 4) Wood pellets (n = 8) Grate furnace (n = 17) Fluidized bed (n = 3) 66–220 MW plants (n = 9) 15–24 MW plants (n = 3)
10.5 9.3 17.7 5.5 3.5 8.0 10.2 2.8 10.7 7.4
8.6 7.5 4.2 6.1 2.4 5.3 6.7 3.3 8.6 4.6
11.4 13.6 20.4 7.7 5.0 15.4 15.0 5.9 13.7 9.0
8.8 10.2 6.2 5.1 3.9 5.3 7.5 4.7 10.2 4.9
11.4 16.8 22.1 10.7 5.4 17.4 16.4 9.5 15.6 9.5
9.1 10.9 7.4 6.5 5.0 5.4 8.3 7.4 10.8 6.2
9.0 8.6 16.5 3.9 3.1 6.7 8.6 4.0 9.3 7.3
9.9 6.3 5.0 5.4 5.0 4.8 6.8 5.7 8.8 5.4
Fig. 1. A principal component analysis of the OCcol, TOC, and LOI data for the samples. Colour coding according to the TOC values: dark blue is the lowest value and orange is the highest. Principal component 1 explains 88% of the data variation and component 2 explains 9.4%. The ellipses show the model’s 95% confidence area, and after removing sample no 12, there were no outliers. The principal components are projections on a plane of directions in a data cloud. Principal component 1 is in the direction of the largest variation of data, principal component 2 is perpendicular to the first one and oriented in the second largest direction of variation, and so on.
between furnace types. By far the most homogeneous group is the mixed fly and bottom ash, which has markedly less variation compared to the other groups. These samples are all derived from small plants with grate furnaces fuelled with wood pellets. Bivariate regression analysis between variables did not yield any satisfactory correlations. However, using multivariate data analysis, i.e. a simultaneous regression of all variables against each other, a principal component analysis shows some reasonably distinct clusters and quite a high degree of correlation, see Fig. 1. The corresponding loading plot, Fig. 2, indicates small differences for the variables, making for a higher value of t [1]. However, in the second principal component the spread is quite large so that the higher temperature LOI will distinguish the group, including sample 18, 8 and others from the group with samples 10, 3 and 11. Samples in the blue group (negative values of t [1]) have lower values for all variables compared to those in other groups, but less so for the higher LOI temperatures. The highest values of TOC and OCcol in relation to LOI at higher temperatures are found in samples 1 and 2. These clusters cut across groupings on physical grounds (fuel, grate or fluidised bed, bottom ash or fly ash, large or small furnace etc.). Knowing the type of furnace and type of fuel is therefore insufficient to determine the content of unburned carbon. In conclusion, both LOI and TOC vary over a wide range. The variations may to some extent be due to the uncertainty of the individual methods but also to variations in the sub-samples. In comparison to sample 7, the efficient combustor, the values of
Fig. 2. Loading plot corresponding to the principal component analysis of Fig. 1. The intensity of the factors indicated in the loading plot determines the placement of the different samples in the PCA of Fig. 1.
LOI and TOC in other samples are often an order of magnitude larger. The relationships between LOI at different temperatures, as well as between LOI and TOC, are also highly variable, even for the reasonably homogeneous group of mixed ash from grate furnaces. This raises questions regarding the consistency of such data in general and what is actually included in the results. The question to be discussed in the next section is how well these observations represent the amount of organic carbon remaining in the ash since this fraction will be biologically degradable and may give rise to redox gradients and greenhouse gas emissions. 3.2. The composition of UC Besides weight losses at a given temperature, the TGA experiments yield another type of result. The TGA was equipped with a quadrupole mass spectrometer, allowing the species released from the samples at different temperatures to be identified. The interpretation of the curves obtained from the TGA investigation can be illustrated with a few examples. The weight of the sample and the DTA signal recorded for sample 8 are shown in Fig. 3. The weight decreases continuously although there are three distinct steps: a first loss at 100–200 °C, a second loss from about 93% to 85% in the interval 300–400 °C, and a third loss from 82% to 75% at just above 700 °C. The DTA signal shows an exothermic phase (peak downwards) from 250 °C to 400 °C, corresponding to the second weight loss. The third step loss is accompanied by a small peak upwards, i.e. an endothermic phase. The QMS results for this test are shown in Fig. 4. There are two carbon dioxide emission peaks, one around 400 °C and the other between 700 °C and 800 °C. The main water emission peak is found below 200 °C and two small peaks are found on either side of the 400 °C mark.
895
H. Bjurström et al. / Fuel 117 (2014) 890–899 105
1,2
-0,4 -0,6
90
-0,8 85
-1,0
80
-1,2 -1,4
75
0,8 0,6 0,4 0,2
-1,6 -1,8 100 200 300 400 500 600 700 800 900 1000
0
0,0
100
-0,2 95
-0,4
90
-0,6
85
-0,8 -1,0
80 -1,2
0,8 0,6 0,4 0,2
75
-1,4
70
-1,6 100 200 300 400 500 600 700 800 900 10001100
0
Temperature (oC)
0,0
Temperature (oC)
Fig. 3. Thermogravimetry results for fly ash sample 8: TGA (black curve), DTA (red broken curve) and carbon dioxide emission from the sample (blue dotted curve).
Fig. 5. Thermogravimetry results for the aged bottom ash sample 2: TGA (black curve), DTA (red broken curve) and carbon dioxide (blue dotted curve).
1,2
1,2
1,2
1,0
1,0
1,0
0,8
0,8
0,6
0,6
0,4
0,4
0,2
0,2
ion current CO2 (10-9 A)
Ion current CO2 and H2O (10-9 A)
1,0
0,8
0,6
0,4
Ion current H2O (10-9 A)
70
Remaining weight (%)
95
1,0
Ion current CO2 (10-9 A)
-0,2
DTA signal (µV/mg)
Remaining weight (%)
100
0,0
Ion current CO2 (10-9 A)
1,2 0,0
DTA signal (µV/mg)
105
0,2 0,0 0
0,0 0
200
400
600
800
1000
100
200
300
400
500
600
700
800
0,0 900 1000
Temperature (oC)
Temperature (oC) Fig. 4. Thermogravimetry: emission of carbon dioxide (black curve) and water (blue broken curve) during the experiment on fly ash sample 8.
The first step at 100–200 °C is dehydration and the second step, in the interval 300–400 °C, is combustion producing large amounts of carbon dioxide, most probably elemental carbon as this step does produce only little water vapour. The third step, at 700– 800 °C, is decomposition of inorganic carbonate. A number of other processes are masked by these main weight losses. The two water emission peaks on either side of the 400 °C mark are probably not related to combustion only but also to the decomposition of, for example, hydroxides. If they had been caused only by combustion, the dip between the two peaks would also be seen in the carbon dioxide peak. However, these relatively small variations on the curves are difficult to interpret and one should also remember that EC in combustion residues might not be pure carbon but contain some hydrogen and oxygen, as described in Ferrari et al. [2] and van Zomeren and Comans [4]. The end of the exothermic phase in the DTA signal is also rather abrupt, indicating that an endothermic process is taking place at the upper end of the combustion range rather than it simply being the combustion tapering off. Furthermore, the small ‘wiggles’ on each curve could represent a minor process that is taking place at that temperature. The results for sample 1 are analogous; see Figs. 5 and 6. The first dehydration step is not noticeable. The second weight loss step is large, from 95% to 77%, is exothermic and occurs between 400 and 600 °C. The third is small but distinct, 77–74%, is endothermic and occurs between 650 °C and 700 °C. The transition between the two latter steps is abrupt in the DTA curve. Water is
Fig. 6. Thermogravimetry results for the aged bottom ash sample 2: emission of carbon dioxide (black curve) and water vapour (red broken curve).
released before the combustion step, around 100 °C and 300 °C. There is a small hump on the carbon dioxide emission curve at 650–700 °C. Smidt et al. [17] and more recently Rocca et al. [8] point out that between 450 °C and 600 °C weight loss in a TGA experiment may pause: the carbon dioxide produced during combustion is reabsorbed as carbonate by the ash [17]. This could also be the reason for the abrupt end in Figs. 3 and 5 of the exothermic process after combustion of residual carbon. The TGA/DTA/QMS results for all the samples have common features but there are some variations: The two carbon dioxide peaks may be distinct, as in Fig. 4, or merge into one broad peak, as in Fig. 6. For a sample with a very low TOC, only the carbonate peak may be visible, not the combustion peak. The opposite is also found: sample 21 only has a combustion peak, not a carbonate peak. The temperature in the sample at which carbon dioxide is detected in the QMS, i.e. the step weight loss due to combustion, is variable and depends on the type of residue. The combustion is completed by 400 °C for fly ash from large plants, but only starts at 400 °C for bottom ash from large plants. In the two APC residues that consist mainly of fly ash, with its UC and a minor quantity of activated carbon, the combustion step starts at higher temperatures than in fly ash: at 400 °C for sample 11, peaking at slightly below 450 °C and at 500 °C, and for sample
896
H. Bjurström et al. / Fuel 117 (2014) 890–899
Table 5 Interpretation of TGA results as organic carbon, elemental carbon and inorganic carbon in the combustion residues with TOC results for reference. Units: w/w %. Residue stream
Sample
OC
EC
IC
TOC
Bottom ash Bottom ash, aged Bottom ash, aged Bottom ash, aged Bottom ash Bottom ash Fly ash and APC residues Fly ash and APC residues Fly ash and APC residues Fly ash and APC residues Fly ash and APC residues Fly ash and APC residues Mixed fly and bottom ash Mixed fly and bottom ash Mixed fly and bottom ash Mixed fly and bottom ash Mixed fly and bottom ash Mixed fly and bottom ash Mixed fly and bottom ash Mixed fly and bottom ash
1 2 5 6 9 10 3 7 8 4 11 21 13 17 20 16 14 15 18 19
0.6 1.3 0.6 0.5 0 2.1 0.6 0.3 0.5 2.7 0.6 0.6 0.2 0.3 0.5 0.9 0.2 0.3 0.6 1.0
16.9 19 1.9 1.0 1.3 9.7 10.4 0.14 10 8.6 4.4 0 0.3 0.25 0 6 7.6 9.6 8 11.7
0.2 0.4 0.1 0.8 0.2 0.3 0.1 0.2 3.0 2.8 1.4 1.6 1.9 2.5 2.0 2.7 3.4 2.0 1.9 1.9
20.2 21.4 0.6 0.4 1.0 10.2 12 0.4 15.4 12.2 10.6 1 1.4 2.9 0.8 11.5 5.8 13.0 6.9 11.1
21 peaking at slightly above 650 °C. The samples from the small pellet furnaces, samples 13–20, behave similarly although the processes occur at another set of temperatures: a combustion peak with weight loss below 400 °C and a carbonate peak with a large weight loss, mostly above 700 °C. In the case of low TOC samples – samples 13, 17 and 20 – the combustion peak is distinct in carbon dioxide emission and DTA signal but not in weight loss (TGA). The difference in combustion temperatures could be the result of a matrix effect, as demonstrated by Müller and Belevi [18]. Interpretation of the TGA results as organic carbon (OC), elemental carbon (EC) and inorganic or carbonate carbon (IC) (following, for example, Ferrari et al. [2]) is summarised in Table 5. For the values to be commensurate, weight losses have been recalculated: Decomposition of carbonates has been recalculated to carbon, which is straightforward. Recalculating weight loss from organic substance to carbon requires an assumption of the carbon content of organic substances. A factor of 58% carbon in organic matter in the colorimetry standard ISO 14235 has been used rather than a factor of 40%, which is more representative of cellulosic material. Emissions of water and carbon dioxide, as registered in the QMS, are difficult to quantify.
What is the structure of elemental carbon in the combustion residues? How might the elemental carbon in combustion residues impact on their environment? In order to distinguish between different carbon structures, i.e. crystalline and amorphous forms, exploratory tests were carried out using micro-Raman spectroscopy, scanning in the range 200– 3000 cm1. In all spectra collected for samples, the double peak is found to be characteristic of C–C carbon bonds in elemental carbon: the ordered or graphitic structure (G) peaks at about 1600 cm1 and the disordered structure (D) peaks at about 1350 cm1. The disordered form is composed of randomly orientated microcrystals (or turbostratic layers, see e.g. [19,20]) while graphite is characterised by a crystalline structure with carbon atoms organised on specific planes. A comparison of spectrograms for a residue and for industrially activated carbon (DioxSorb BP2) shows that the carbon structures are very similar, see example in Fig. 7. Furthermore, the ratio of peak heights D/G, Table 6, was about 1.33 for all samples, which implies that the proportion of D and G forms of carbon is similar in all samples, including the activated carbon. The properties of elemental carbon determined by Raman spectroscopy are similar in both bottom ash and fly ash and are also similar to that of activated carbon. Samples 16 and 19 from pellet furnaces are outliers. There was not enough UC in samples 14 and 17 for the carbon double peak to be seen in the micro-Raman experiments. In sample 14, a stick or flake of charred material was retrieved from the larger sample and subjected to micro-Raman. Looking at the position of the Raman spectroscopy peaks for the samples, it is obvious (Fig. 8) that peak positions for most of the residues cluster together and are close to those of the activated carbon. There are, however, outliers on both sides of the cluster, i.e. sample 7, which is an exceptionally well-combusted ash, and the poorly burned piece of wood in sample 14. The temperature to which the carbon has been subjected affects the structure of the elemental carbon, see e.g. Franklin [21]. The higher the temperature, the more graphitic the structure becomes and the larger the microcrystals. The effect on the Raman spectra is a narrowing of the peak or bandwidth (see e.g. [22]). To test the data, the G peak width as a function of microcrystal size was calculated using the Tuinstra–Koenig empirical formula [23], as shown in Fig. 9. One would have expected to find bottom ash samples at
13000
The numbers in Table 5 demonstrate that elemental carbon dominates in the TOC, i.e. TOC is mainly EC.
12000
Results from the thermogravimetric analysis lead to the conclusion that elemental carbon (EC) is the dominant part of the unburned carbon (UC) in combustion residues and this is independent of the value of the content of UC, which within the set of samples varies from almost 0% to more than 90%. The weight loss interpreted as organic carbon (OC) represents a much smaller part of UC. The ratio of organic carbon to elemental carbon (OC/EC) did vary within broad limits from 0.05 to 1 but no specific relationship could be established for different types of ash. The variation within each group is far greater than between the different ashes. The dominance of the elemental species in the unburned carbon raises questions regarding the properties and the behaviour of the ash in different environments:
Counts (-)
11000
3.3. Properties of elemental carbon in UC of combustion residues
10000 9000 8000 7000 6000 800
1000
1200
1400
1600
1800
Raman shift (cm-1) Fig. 7. Raman spectrum for the aged bottom ash, sample 2 (black curve) and for the activated carbon reference sample (red broken curve).
897
H. Bjurström et al. / Fuel 117 (2014) 890–899 Table 6 Results of the analysis of the carbon double peaks in the Raman spectra. Residue stream
Sample no
D/G ratio (–)
D peak position (cm1)
D peak width (cm1)
G peak position (cm1)
G peak width (cm1)
La (nm)
Bottom ash, aged Bottom ash, aged Bottom ash Cyclone ash ESP ash ESP ash APC residue APC residue Mixed ash Mixed ash Charred stick DioxSorb BP2
2 10 12 3 7 8 11 21 16 19 From 14 –
1.41 0.97 1.23 1.40 1.51 0.86 1.46 1.38 1.91 3.91 1.55 1.57
1336 1345 1342 1339 1383 1336 1326 1354 1334 1337 1314 1328
183 226 246 161 319 144 179 246 163 131 236 178
1595 1586 1587 1585 1610 1591 1597 1603 1591 1589 1569 1593
103 216 117 201 260 236 102 252 139 186 99 106
3.1 4.5 3.6 3.1 2.9 5.1 3.0 3.2 2.3 1.1 2.8 2.8
350
1610 300
FWHM for D peak (cm-1)
Position of the G band (cm-1)
1620
1600
1590
1580
1570
1560 1310
250
200
150
1320
1330
1340
1350
1360
1370
1380
1390
Position of the D band (cm-1)
100 0
1
2
3
4
5
6
Microcrystallite size (nm) Fig. 8. Position of D and G bands from micro-Raman spectroscopy of combustion residue samples: j, bottom ash, s, fly ash and APC residue, d, mixed ash, }, activated carbon and h, charred wood stick.
the narrowest peak widths but there does not seem to be any general tendency in bandwidth. Most samples cluster at about 3 nm in microcrystallite size and samples 16 and 19 are again outliers. Activated carbon is well known for its large specific surface, from 500 to 1500 m2/g depending on the raw material and manufacturing process. This gives it a high adsorption capacity for both inorganic and organic substances. The values of the specific surface of 30–50 m2/g reported for various coal and wood fly ashes [24–26] are large in comparison to those of e.g. quartz sand (0.01 m2/g) and silt (0.5 m2/g). If the EC-content contributes to a large specific surface in the ashes, than this may enhance the adsorption capacities for both organic and inorganic substances. As a brief indication adsorbed organic substances was studied by screening samples 7, 8 and 2, representing low medium and high organic content. The number of semi-volatile substances detected was quite low and their concentrations were far lower than 8% of TOC. A similar absence of volatile organic substances was noted by Kuo et al. [27] for charcoal pyrolysed above 400 °C and by Spokas for ash from combustion of biomass [6]. The substances that could be identified with some certainty include the ubiquitous phthalates (DBP in the non-derivatised samples and DEHP in the derivatised samples), which are not residues of incomplete combustion but rather post-combustion contaminants. Other contaminant substances are (z)-9-octadecenamide (oleamide), a slip agent used in the plastic bags in which the samples were stored, and 2, 6-di-t-butyl-4-methyl phenol or DBP, an antioxidant in laboratory gloves. One possible interpretation is that the substances are adsorbed as contaminants in the
Fig. 9. Peak width of D peak as a function of micro-crystallite size from microRaman spectroscopy of combustion residue samples: j, bottom ash, s, fly ash and APC residue, d, mixed ash, }, activated carbon and h, charred wood stick.
EC phase after the generation of the residues. Contamination of the samples was a major issue in Pavasar’s screening of organic compounds in, among others, combustion residues [28]. The presence of semi-volatile substances that could be expected to disappear in the combustion process indicates that these substances can be adsorbed and ‘stored’ in the combustion residue. It has been shown that by applying activated carbon to contaminated sediments, the bio-uptake of PCBs in benthic worms was reduced by 69–99% [29]. From this and other studies it is known that activated carbon, due to its strong sorption capacity, reduces the bioavailability of organic contaminants in soil and sediments (see review by Hilber and Bucheli [30]). Most studies were performed in laboratories although recent field-scale studies confirm the potential of activated carbon as an effective amendment that adsorbs PAH [31]. Of equal interest is the documented substantial adsorption capacity of black carbon (e.g. [32]) and that black carbon may act as an effective adsorbent of PAH, reducing the bio-uptake of organic contaminants from soil and sediments by up to two orders of magnitude (see review article by Koelmans et al. [33]). The metal adsorption capacity of black carbon and activated carbon is well documented (e.g. [34–36]). The combustion residues contain UC including elemental carbon that may act as a potent adsorbent and it can be expected that the leaching behaviour of the ashes is influenced by the structure of the EC. One of the major obstacles to the use of ash in, for example, construction is the leaching of potentially hazardous substances and the basis of sustainable use of solid biofuel ash is a
898
H. Bjurström et al. / Fuel 117 (2014) 890–899
leaching process that can be retarded to fit the biogeochemical cycles of the recipients where it is used. The presence of EC is an important factor to understand. Another use of unburned carbon in ash may be as a cheap precursor to active carbon, as proposed by, among others, Maroto-Valer [24] and Bartonova et al. [26]. The role of the EC in the leaching process should be investigated further – as should the possibility of activating the elemental carbon by treating the ash in different ways, e.g. using steam or separation techniques.
4. Conclusions In a range of combustion residues from different fuels, furnaces and air pollution control processes, it was observed, through the use of TGA, that the unburned carbon is mainly elemental. At least 50% of the unburned carbon was found to be elemental carbon and for most samples considerably more. The content of elemental carbon was >10% of the total ash mass for several of the samples studied. The proportion and composition of the unburned carbon in the residues vary more with the combustion process than with the type of fuel. LOI at different temperatures, TOC and chemical oxidation all resulted in a variation of more than a magnitude between the best and most poorly incinerated sample. The LOI and TOC showed little correlation to organic carbon, whereas chemical oxidation showed some correlation to elemental carbon, especially in bottom ash. Since the waste is generated at temperatures of around 800 °C, it is not surprising that LOI at 550 °C or TOC at 900 °C would yield somewhat deviating results although no clear trends for the variation in temperature were found. From the magnitude of variation, it can be inferred that 550 °C is the least suitable temperature when testing ash using LOI. However, since loss of salts can occur at the highest temperature tested, the intermediate temperature, 750 °C, seems to be the safest alternative. The only property that was common to all samples was the relationship between graphitic and disordered forms of elemental carbon, which was about 1.33 on a mass-to-mass basis. This is also the case for commercial activated carbon. Observations indicated that organic compounds from sample bags and rubber gloves had been adsorbed onto the samples and the ash may thus be expected to act as a retainer as well as a releaser of contaminants. The organic compounds present in unburned carbon in bottom ash and fly ash are mainly non-volatile, macromolecular compounds and similar compounds may be formed slowly through the dissolution of graphitic carbon. Unfortunately, there is still no simple analytical procedure to assess the quality and even the amount of unburned carbon in residue. It is highly misleading to use TOC as measure of organic carbon in combustion residues. Unburned carbon in biofuel ash should not á priori be considered an environmental risk factor and in the process of balancing energy efficiency, economy and environmental impact of the residue, the EC content should be considered.
Acknowledgements The examination reported here took place within several projects with financial support mainly from Värmeforsk (the Swedish Thermal Engineering Research Institute). Additional support was obtained at the different stages from Åforsk (Research foundation of the ÅF Group), Skelleftekraft AB, Fortum Värme AB, Öresundskraft AB, Iggesund Paperboard AB, Holmen Paper AB, Svenska Energiaskor AB. Jacobi Carbon Systems AB graciously donated a sample of the activated carbon DioxSorb BP2.
Collaborators in the projects were Pascal Suér and Lennart Larsson (SGI), Menad Noureddine (Luleå University of Technology), Jacob Thyr and Lars Hälldahl (K-Analys AB).
References [1] Vosteen B, Beyer J. Obtainable residual carbon content in slags and ashes from incineration systems. VGB PowerTech 2000;9:48–53; Vosteen B, Beyer J. Obtainable residual carbon content in slags and ashes from incineration systems. VGB PowerTech 2000;10:73–5. [2] Ferrari S, Belevi H, Baccini P. Chemical speciation of carbon in municipal solid waste incinerator residues. Waste Manag 2002;22:303–14. [3] Kumpiene J, Robinson R, Brännvall E, Nordmark D, Bjurström H, Andreas L, et al. Carbon speciation in ash, residual waste and contaminated soil by thermal and chemical analyses. Waste Manag 2011;31:18–25. [4] van Zomeren A, Comans R. Carbon speciation in municipal solid waste incinerator (MSWI) bottom ash in relation to facilitated metal leaching. Waste Manag 2009;29:2059–64. [5] Cohen-Ofri I, Weiner L, Boaretto E, Mintz G, Weiner S. Modern and fossil charcoal: aspects of structure and diagenesis. J Archaeol Sci 2006;33:428–39. [6] Spokas K. Review of the stability of biochar in soils: predictability of O:C molar ratio. Carbon Manag 2010;1:289–303. [7] Brown RC, Dykstra J. Systematic errors in the use of loss-on-ignition to measure unburned carbon in fly ash. Fuel 1995;74:570–4. [8] Rocca S, van Zomeren A, Costa G, Dijkstra J, Comans R, Lombardi F. Mechanisms contributing to the thermal analysis of waste incineration bottom ash and quantification of different carbon species. Waste Manag 2013;33:373–81. [9] SS 18 71 87. Solid fuels – determination of the amount of unburned material in solid residues from combustion; 1984. [10] EN 196-2:2005. Cement – testing – Part 2: chemical analysis; 2005. [11] Chandler AJ, Eighmy TT, Hartlén J, Hjelmar O, Kosson DS, Sawell SE, et al. Municipal solid waste incinerator residues. Amsterdam: Elsevier; 1997. [12] ASTM D2974-00. Standard test methods for moisture, ash and organic matter of peat and other organic soils; 2000. [13] EN 13137:2001. Characterisation of waste. Determination of total organic carbon (TOC) in waste, sludges and sediments; 2001. [14] EN 1744-1:1998. Ballast – chemical properties – Part 1: chemical analysis, 1st ed.; 1998. [15] ISO 14235:1998; Soil quality – determination of organic carbon by sulfochromic oxidation; 1998. [16] Kylefors K, Ecke H, Lagerkvist A. Accuracy of COD test for landfill leachates. Water, Air, Soil Pollut 2003;146:153–69. [17] Smidt E, Meissl K, Tintner J, Ottner F. Interferences of carbonate quantification in municipal solid waste incinerator bottom ash: evaluation of different methods. Environ Chem Lett 2010;8:217–22. [18] Müller C, Belevi H. Differentiating between organic and elemental carbon compounds in residues from municipal solid waste incinerating plants. Müll und Abfall 2005;2005:352–7 [in German]. [19] Bourrat X, Langlais F, Chollon G, Vignoles G. Low temperature pyrocarbon: a review. J Braz Chem Soc 2006;17:1090–5. [20] Keiluweit M, Nico PS, Johnson MG, Kleber N. Dynamic molecular structure of plant biomass-derived Black Carbon (Biochar). Environ Sci Technol 2010;44:1247–53. [21] Franklin R. Crystallite growth in graphitizing and in non-graphitizing carbons. Proc Roy Soc A 1951;209:196–218. [22] Rhim Y-R, Zhang D, Fairbrother DH, Wespanick KA, Livi KJ, Bodnar RJ, et al. Changes in electrical and microstructural properties of microcrystalline cellulose as function of carbonization temperature. Carbon 2010;48:1012–4. [23] Tuinstra F, Koenig JL. Raman spectrum of graphite. J Chem Phys 1970;53:1126–30. [24] [24 Maroto-Valer MM, Taulbee DN, Schobert HH, Hower JC, Andrésen JM. Use of unburned carbon in fly ash as precursor for the development of activated carbons. In: International ash symposium. Centre for Applied Energy Research, Univ of Kentucky; 1999. Paper 19. [25] Hwang JY, Sun X, Li Z. Unburned carbon from fly ash for mercury adsorption. I. Separation and characterization of unburned carbon. J Miner Mater Charact Eng 2002;1:39–60. [26] Bartonova L, Juchelkova D, Klika Z, Cech B. On unburned carbon in coal ash from various combustion units. World Academy of Science, Engineering and Technology; 2011. p. 76. [27] Kuo L-J, Louchouarn P, Herbert BE. Fate of CuO-derived lignin oxidation products during plant combustion: Application to the evaluation of char input to soil organic matter. Org Geochem 2008;39:1522–36. [28] Pavasars I. Characterisation of organic substances in waste materials under alkaline conditions. Doctorate thesis presented at Linköping University, Linköping; 1999. [29] Beckingham B, Ghosh U. Field-scale reduction of PCB bioavailability with activated carbon amendment to river sediments. Environ Sci Technol 2011;45:10567–74. [30] Hilber TD, Bucheli. Activated carbon amendment to remediate contaminated sediments and soils: a review. Global NEST J 2010;12:305–17.
H. Bjurström et al. / Fuel 117 (2014) 890–899 [31] Hale EE, Elmquist M, Brändli R, Hartnik T, Jakob L, Henriksen T, et al. Activated carbon amendments to sequester PAHs in contaminated soil: a lysimeter field trial. Chemosphere 2012;87:177–84. [32] Brändli RC, Hartink T, Henriksen T, Cornelissen G. Sorption of native polyaromatic hydrocarbons (PAH) to black carbon and amended activated carbon in soil. Chemosphere 2008;73:1805–11. [33] Koelmans AA, Jonker MT, Cornelissen G, Bucheli TD, van Noort PC, Gustafsson Ö. Black carbon: the reverse of its dark side. Chemosphere 2006;63:365–77.
899
[34] Mohan D, Singh KP. Single- and multi-component adsorption of cadmium and zinc using activated carbon derived from bagasse – an agricultural waste. Water Res 2002;36:2304–18. [35] Qiu Y, Cheng H, Xu C, Sheng GD. Surface characteristics of crop-residuederived black carbon and lead (II). Water Res 2008;42:567–74. [36] Tay JH, Chen XG, Jeyaseelan S, Graham N. Optimising the preparation of activated carbon from digested sewage sludge and coconut husk. Chemosphere 2001;44:45–51.
Contents lists available at ScienceDirect
Fuel journal homepage: www.elsevier.com/locate/fuel
Unburned carbon in combustion residues from solid biofuels H. Bjurström a, B.B. Lind b,⇑, A. Lagerkvist c a
ÅF-Industry AB, SE-169 99 Stockholm, Sweden Swedish Geotechnical Institute, SE-412 96 Göteborg, Sweden c Luleå University of Technology, SE-971 87 Luleå, Sweden b
h i g h l i g h t s Different methods to determine unburned carbon in ash yield slightly different results. Unburned carbon is mostly elemental carbon, very little volatile or semi-volatile organic carbon. Raman results show that unburned carbon is similar in bottom ash and in fly ash.
a r t i c l e
i n f o
Article history: Received 20 November 2012 Received in revised form 7 August 2013 Accepted 10 October 2013 Available online 22 October 2013 Keywords: Combustion residue Ash Solid biofuels Unburned carbon Elemental carbon
a b s t r a c t Unburned carbon (UC) in 21 combustion residues from solid biofuels has been examined using several methods of analysis (including LOI and TOC) as well as micro-Raman spectroscopy. The concentration of unburned carbon in the residues varied over an order of magnitude and in several samples accounted for about 10% of the ash mass. It was observed that TOC had a poor correlation to organic carbon, especially for fly ashes. LOI at all tested temperatures showed a better correlation than TOC to the organic carbon content, whereas the TOC is better correlated to elemental carbon. LOI550 gave a larger variation and a less complete mobilisation of unburned carbon than LOI at 750 or 975 °C did, but at the highest temperature metal oxidation was notably affecting the mass balance to the extent that some samples gained mass. For this reason, and of the temperatures tested, LOI750 seem to be the most stable indicator for organic remains in the incineration residuals. Most of the unburned carbon is elemental, and only slowly degradable, so the potential emissions of organic compounds from ashes should not be assessed by using a TOC test. The structure of the detected elemental carbon in UC is similar to that of activated carbon, which indicates a potentially large specific surface. This should be borne in mind when assessing the environmental impact of using ash for different purposes, including use as a construction material. Field studies are needed to verify the actual impact as it may depend on environmental conditions. Ó 2013 Elsevier Ltd. All rights reserved.
1. Introduction The presence of unburned carbon (UC) in solid combustion residue not only hints at imperfections in the combustion process but also raises fears that the residue might release potentially noxious organic substances. However, examination of unburned carbon in ash from the combustion of different fuels has shown that unburned carbon is largely elemental carbon (EC) rather than organic carbon (OC) (among others [1–4]). Elemental carbon, or charcoal, is chemically stable and degrades very slowly in a natural environment – after several thousand years its structure is mostly unaffected (see e.g. Cohen-Ofri et al. [5] and Spokas [6]). It should therefore be noted that the unburned carbon detected in combus-
⇑ Corresponding author. Tel.: +46 317786566. E-mail address: [email protected] (B.B. Lind). 0016-2361/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.fuel.2013.10.020
tion residue is essentially something other than the presence of organic reactive carbon. There are several routine methods to determine the content of unburned carbon in combustion residue. These methods are also used for other materials, such as determining the content of organic matter in soil or water. While using the same method of analysis is desirable, some confusion may arise when interpreting the results, as different materials respond in different ways. Loss on ignition, or LOI, is a convenient method to determine what could have burnt provided the combustion residue has not absorbed water following extraction from the furnace, in which case LOI would consist of unburned carbon, carbonate carbon, hydroxide water or even hydration water [7,8]. LOI is carried out at a variety of temperatures depending on the fuel; 550 °C for biomass and 750 °C for coal [9] or the envisaged area of use, 950 °C for cement, [10]. The lowest temperature is used for biomass to ensure that potassium and chlorine are not counted as oxidisable carbon.
891
H. Bjurström et al. / Fuel 117 (2014) 890–899
LOI is not recommended for use with air pollution control (APC) residue from waste-to-energy plants because these materials are very hygroscopic and the adsorbed water will affect the LOI-value [11]. The suitability of the ash material as a silica source in concrete is checked by igniting at 950 °C which is supposed to reveal bound water and carbonates. On the other hand, TOC yields the carbon content. Although ASTM calls TOC ‘Total Oxidisable Carbon’ [12], the expression ‘Total Organic Carbon’ is often used for combustion residues, similar to the use of this method to determine organic content in soil or water [13]. It has been shown [3] that a large proportion of soil carbon may also be elemental. The ASTM definition will be used here. A factor that complicates the interpretation of thermal or chemical analyses is the complexity of combustion residues as they represent a matrix that may be involved in the processes in which carbon reacts. Although they are products of oxidation, i.e. combustion, their composition hints at reduced conditions when they are extracted. The particle size distribution may be quite broad and not all components in the larger particles may have time to react during the process. The present paper describes an examination of UC, determined using different analytical methods on solid residues from Swedish combustion plants, complemented by a study of the samples using a spectroscopic method. Three questions are addressed in the paper: What is the carbon content of combustion residues and how can it be assessed? How much of the unburned carbon is elemental and what is the remainder? What are the properties of EC in UC? As the fuels used in Swedish combustion plants are largely solid biofuels, the focus of this paper is on combustion residues from solid biofuels. Samples were chosen to provide a wide range of unburned carbon content.
air pollution control residue), different fuels (wood chips and pellets, bark, waste wood, municipal solid waste), large plants and small plants and different industrial sectors (cogeneration plants, heating plants, pulp and paper mills). These include efficient combustors as well as plants that are not as efficient as they should be, see Table 1 for a summary. Samples 1, 2, 5, 6, 9, 10 and 12 are bottom ash, which is usually handled separately from fly ash in large plants, whereas mixing fly ash and bottom ash (samples 13–20) is more common in smaller plants. The remaining samples are fly ash or APC residue. In the latter, samples 11 and 21, additives are also found, including activated carbon. With regard to the bottom ash samples, four of the plants were sampled for fresh ash and three were sampled for aged ash. In the micro-Raman investigation, a sample of activated carbon, DioxSorbBP2, was obtained from Jacobi Carbon Systems and used as reference material. 2.1. Sampling and sample treatment Some of the samples of combustion residues were gathered for this specific investigation and some originate from other studies. Sampling procedures vary widely and the samples should therefore not be considered representative in a regulatory context of the residues produced in these plants. However, collectively they should represent a cross-section of combustion residues that arise in real situations. All samples were first homogenised at the laboratory and subsamples were ground to <100 lm using a disc stainless steel grinder. 2.2. Analytical methods 2.2.1. Total solids Total solids were determined by weighting after drying at 105 °C for at least six hours.
2. Materials and methods The 21 samples from 18 combustion plants studied in this investigation represent various types of furnaces (grate, fluidised bed, pulverised fuel with burners). Also represented are different types of residue (bottom ash, various types of fly ash, mixed ash,
2.2.2. Loss on ignition There are numerous procedures for the determination of volatile substances, mainly aimed at quantifying oxidisable substances in a sample. Using different standards, there is considerable variation in the temperature at which ignition takes place.
Table 1 A summary of the ash types sampled. Sample number
Combustion plant
Type of furnace
Type of combustion residue
Fuel
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21
Händelö P11 Händelö P11 Händelö P11 Händelö P11 Dåva Braviken Skellefteå Iggesund Wargön Avesta Nynäshamn Öresundskraft Bureå Burträsk Jörn Kåge Byske PC Vindan Norsjö Boliden Händelö P14
Grate Grate Grate Grate Grate Grate CFB Grate Grate Grate BFB PF Grate Grate Grate Grate Grate Grate Grate Grate CFB
Fresh bottom ash Aged bottom ash Cyclone ash Baghouse filter ash Aged bottom ash Aged bottom ash ESP ash ESP ash Bottom ash Bottom ash Baghouse filter (APC residue) Bottom ash Mixed Mixed Mixed Mixed Mixed Mixed Mixed Mixed Baghouse filter (APC residue)
Waste wood and rubber Waste wood and rubber Waste wood and rubber Waste wood and rubber MSW Bark, de-inking sludge Wood chips Bark Bark MSW Waste wood Biomass Wood pellets Wood pellets Wood pellets Wood pellets Wood pellets Wood pellets Wood pellets Wood pellets Sorted waste
Nominal power MW fuel tires tires tires tires
70 70 70 70 77 66 92 70 24 15–18 24 220 <5 <5 <5 <5 <5 <5 <5 <5 75
892
H. Bjurström et al. / Fuel 117 (2014) 890–899
Here the combustion residues have been tested at 550 °C [9], 750 °C ([9] for mineral fuels) and 975 °C [14], in order to cover the normal spectrum. The samples, 1–2 g, were heated up slowly to the LOI temperature (60 min) and were then held at the final temperature for a further 60 min. The samples were left to cool in a desiccator before being reweighed.
2.2.3. Total oxidisable carbon The indirect method in EN 13137 [13] was used here: Inorganic carbonate was determined by means of emitted carbon dioxide when acid was added to part of the sample. Total carbon was determined by heating a 200 mg sample to 1000 °C and measuring the emitted carbon dioxide. Total oxidisable carbon was obtained by subtracting inorganic carbon from total carbon.
2.2.4. Colorimetry Chemical digestion is commonly used in the analysis of organic soil content. It was performed using dichromate sulphuric acid according to ISO 14235 [15]. A Cr(VI) solution is added to the solid sample and allowed to react with the sample for several hours. The Cr(III) content is then determined spectrophotometrically. The response is calibrated using glucose solutions. This method supposedly yields a concentration of organic carbon in the sample. It is reported as organic matter content by multiplying the carbon content by a factor of 1.724. It is thus assumed that carbon enters for 58 mass per cent of the organic matter in the sample. This is of course a rough estimate and different organic compounds may differ in their degradability. It has also been shown that iron and sulphur compounds, for example, may also contribute to the observed reduction in Cr(VI) [16].
2.2.5. Thermogravimetry Thermogravimetry, thermal analysis and mass spectrometry together provide information on weight loss over a temperature range (i.e. decomposition or combustion of a chemical compound), the heat balance and the species released from the sample during the process. All this information provides indirect evidence of the speciation of an element in the samples. All tests were performed with the sample exposed to a stream of air that acted as an oxidiser. A release of carbon dioxide indicates that a combustible, carbon-containing substance has burned or that a carbonate has been decomposed. When a release of carbon dioxide through combustion is not accompanied by a release of water vapour, it has been inferred that the carbon in the sample is predominantly elemental rather than organic; see also Ferrari et al. [2]. The thermobalance used in this investigation is Netzsch STA 409C, equipped with simultaneous thermogravimetric analysis (TGA), differential thermal analysis (DTA) and a quadrupole mass spectrometer (QMS) for analysing the evolved gases. A 50–100 mg sample is heated in a flow of air, 200 ml/min, from room temperature to approximately 1000 °C at a rate of 10 °C per minute. The following observations were recorded:
The weight of the sample. The set temperature. The DTA signal. QMS signals in the range 1–100 Da (g/mol).
The software accompanying the thermobalance was used to determine weight loss from TG measurements and the location (temperature) of step changes in weight.
To determine the proportion of OC and EC in the combustible carbon, the following method was used to distinguish between step changes in the TGA curves: Combustion of EC and OC is identified by means of exotherms in the DTA curve and decomposition of an inorganic carbonate (IC) by means of an endotherm (the temperature at which these processes take place varies depending on the type of ash). Combustion of OC is accompanied by an emission of water and carbon dioxide while combustion of EC is accompanied only by an emission of carbon dioxide. OC is generally found at lower temperatures than EC. When the transition between OC and EC steps in the TGA curve is too indistinct, cessation of water release and/or inflexion in the DTA curve are taken to mark the end of ‘OC combustion’. 2.2.6. Micro-Raman spectroscopy Micro-Raman is used routinely to study the structure of carbonaceous materials, e.g. carbon nanotubes, char and soot. A Renishaw Invia Raman Spectrometer was used. Its lower limit for spatial resolution is of the order of 1 lm. Two alternative laser sources were used to excite the sample: A Renishaw 300 mW HPNIR diode laser with a wavelength of 785 nm in combination with a 1200 lines/mm diffraction grating, which provides a linear image of the sample. A Laser Physics 20 mW argon ion laser with a wavelength of 514 nm and a 2400 lines/mm diffraction grating, which provides a specular image of the sample The spectrometer was calibrated daily against a reference silicon sample. A sample of about 2 g was mounted on a circular holder (diameter approximately 10 mm). The first step in the investigation was to produce a Raman spectrum in the range 800–1700 cm1 for five spots on the surface of the mounted sample, choosing areas with an average appearance under the microscope as well as areas differing from the average. The 785 nm laser was used (its wavelength is suited to transitions in organic substances) at 5% of its maximum power in order to avoid burning away any organic residues. To compensate for the low sensitivity, data were gathered over a comparatively long period. When peaks that were not elemental carbon were observed in a recorded spectrum, the spot on the sample was investigated further using lasers and spectra were recorded in an extended range of 200–3200 cm1. 2.2.7. GC/MS analysis of organic substances Semi-volatile compounds in extracts of the samples were screened semi-quantitatively using GC/MS. Two extracts were prepared, one for non-polar compounds by leaching with dichloromethane in a Soxhlet unit and the other for polar compounds by acidifying a suspension of ash with sulphuric acid, extracting with diethyl ether, drying the extract and derivatising with a silylating reagent (N,O-bis-trimethylsilyl-trifluoracetamide, trimethylchlorosilane and pyridine in the proportions 10:1:10). Compounds were identified using a database of mass spectra. 3. Results and discussion 3.1. The carbon content of the different combustion residues Results for the most commonly used variables, i.e. the loss on ignition, LOI or VS (volatile solids), the total oxidisable carbon, TOC, and colorimetrically determined organic carbon, are pre-
893
H. Bjurström et al. / Fuel 117 (2014) 890–899 Table 2 LOI, TOC and colorimetric OCcol (w/w %) of the sampled combustion residues. Temperature of LOI in °C. Residue stream
Sample
LOI550
LOI750
LOI975
TOC
OCcol
Bottom ash Bottom ash, aged Bottom ash, aged Bottom ash, aged Bottom ash Bottom ash Bottom ash Fly ash and APC-residues Fly ash and APC-residues Fly ash and APC-residues Fly ash and APC-residues Fly ash and APC-residues Fly ash and APC-residues Mixed fly and bottom ash Mixed fly and bottom ash Mixed fly and bottom ash Mixed fly and bottom ash Mixed fly and bottom ash Mixed fly and bottom ash Mixed fly and bottom ash Mixed fly and bottom ash
1 2 5 6 9 10 12 3 7 8 4 11 21 13 14 15 16 17 18 19 20
18.5 22.6 3 3.4 3.3 12.4 91.1 12.4 0.8 17.2 17.4 6.6 1.1 2.6 10 11.6 11.7 1 11.4 14 1.6
18.8 24.2 3.2 5.5 3.6 13.3 92 12.3 0.8 25.4 26.2 10.1 6.7 8.8 20.6 16.8 18.5 9.6 17.5 21.7 9.3
19.4 24.1 3.2 5.7 2.4 13.8 92 13.6 1.1 27.3 31.2 12.2 15.1 11.6 21.6 19.1 21.4 11.3 18.8 24.3 10.7
20.2 21.4 0.6 0.4 1 10.2 86.1 12 0.4 15.4 12.2 10.6 1 1.4 11.5 5.8 13 2.9 6.9 11.1 0.8
16.4 17.0 1.0 0.4 1.9 10.3 42.3 12.7 0.2 10.0 12.8 8.6 3.8 0.6 8.1 5.4 6.1 0.5 5.1 7.4 0.7
13.0 146 9, 13 74
17.4 108 13, 7 57
19.0 98 15, 4 55
11.7 157 7, 94 85
8.2 116 6, 45 84
Global average Global standard deviation Average excluding no 12 Std dev. excl. sample 12
% %
Table 3 Average, hxi and standard deviations, s, (w/w %) of the differences DLOI and (TOC–OCcol) depending on residue type. Differences and standard deviations given as % of LOI750, LOI975 and TOC respectively. Residue stream
(750–550°) hxi
s
(975–750°) hxi
s
TOC– OCcol hxi
s
Bottom ash Fly ash and APC residue Mixed fly and bottom ash
11.3 30.5 54.0
13.4 30.8 23.5
6.7 22.1 12.5
21.3 17.9 5.9
19.7 31.2 37.7
46.9 124 25.2
sented in Table 2, where the samples are listed according to the residue stream. LOI was determined at three temperatures, the lower temperature corresponding to the standard temperature used for biomass and the higher temperature being the standard temperature applied in TOC analysis. For each sample, the LOI value shown in Table 2 increase as the temperature rises although the increments between different temperatures vary. These increments are small for many samples, although samples 13, 17, 20 and 21 show a relatively minor mass loss at the lower temperature together with a much larger mass loss at the higher temperature. All the other samples show LOI values from 46% of the total LOI or more in the LOI550 step, but these four samples have 22% or as low as 7% of the weight loss at 550 °C. As also indicated by the large variation of LOI550 results as compared to those of higher temperature, this contrast may indicate that LOI550 is a less stable determination of LOI in combustion residues. At the highest tested temperature, 975 °C, a gain of mass was apparent for samples 2 and 9, which may be interpreted as a metal oxidation overshadowing the mobilisation of elemental and inorganic carbon. This is typical for bottom ashes, and all such samples show no or very low LOI at the highest temperature step, so using this elevated temperature will probably lead to an underestimation of total carbon emissions. The LOI should reflect as much as possible the amount of unburned material in the samples. But the obtained results of LOI550 are similar to, and sometimes even lower, than the different carbon content measurements. Thus the LOI550 is obviously not representing all the unburned material. The incomplete combustion at 550 °C is also indicated by comparison of standard deviations for all samples at the different
temperatures. This indicate that the LOI at 550 °C is more susceptible than the higher temperatures to random variations, e.g. of particle size distribution, sample size etc. However, the results from 975 °C are also problematic due to the oxidation of metals and the associated gain of mass. In conclusion the LOI750 seem to be the most stable indicator of carbon losses of the LOI tests used. The combustion plant from where sample 7 was taken is exceptionally efficient, yielding an almost totally mineralised residue, whereas sample 12 represents exceptionally poor combustion performance. This furnace is a pulverised fuel furnace, originally fired with coal but now converted to biomass. The high content of unburned carbon in its bottom ash is due to pre-treatment limitations, resulting in fuel particles that are too large. As a unique sample, it will be excluded from further comparisons but it serves as a reminder of the potentially large variation in bottom ash quality. The variation in data for different residue types is summarised in Table 3. As can be seen, a large change occurs between LOI550 and LOI750 while the change between LOI750 and LOI975 is smaller. For bottom ash, average mass for all samples increases above 750 °C rather than decreases. This is mainly due to sample 9. As can be noted in Table 4, the averages of LOI and TOC for the different groups of plants, fuels, furnaces and residue streams, vary considerably between the different types of combustion residues and for all analyses. The standard deviation of the averages is of the same order of magnitude as the average itself, which also indicate a wide spread of the results within each group. At the high end, the waste wood and tyres fuel fractions seem to generate samples with more unburned carbon. At the lower end, we find wood and bark fuel. There is also quite a large difference
894
H. Bjurström et al. / Fuel 117 (2014) 890–899
Table 4 Average and standard deviations (s) of LOI and TOC (w/w %D.S.) grouped by residue types, fuels, combustor types and capacity of the latter. Residue stream
LOI550
s
LOI750
s
LOI975
s
TOC
s
Bottom ash (n = 6) Fly ash and APC (n = 6) Wood and rubber tyres (n = 4) MSW (n = 3) Wood and bark (n = 4) Wood pellets (n = 8) Grate furnace (n = 17) Fluidized bed (n = 3) 66–220 MW plants (n = 9) 15–24 MW plants (n = 3)
10.5 9.3 17.7 5.5 3.5 8.0 10.2 2.8 10.7 7.4
8.6 7.5 4.2 6.1 2.4 5.3 6.7 3.3 8.6 4.6
11.4 13.6 20.4 7.7 5.0 15.4 15.0 5.9 13.7 9.0
8.8 10.2 6.2 5.1 3.9 5.3 7.5 4.7 10.2 4.9
11.4 16.8 22.1 10.7 5.4 17.4 16.4 9.5 15.6 9.5
9.1 10.9 7.4 6.5 5.0 5.4 8.3 7.4 10.8 6.2
9.0 8.6 16.5 3.9 3.1 6.7 8.6 4.0 9.3 7.3
9.9 6.3 5.0 5.4 5.0 4.8 6.8 5.7 8.8 5.4
Fig. 1. A principal component analysis of the OCcol, TOC, and LOI data for the samples. Colour coding according to the TOC values: dark blue is the lowest value and orange is the highest. Principal component 1 explains 88% of the data variation and component 2 explains 9.4%. The ellipses show the model’s 95% confidence area, and after removing sample no 12, there were no outliers. The principal components are projections on a plane of directions in a data cloud. Principal component 1 is in the direction of the largest variation of data, principal component 2 is perpendicular to the first one and oriented in the second largest direction of variation, and so on.
between furnace types. By far the most homogeneous group is the mixed fly and bottom ash, which has markedly less variation compared to the other groups. These samples are all derived from small plants with grate furnaces fuelled with wood pellets. Bivariate regression analysis between variables did not yield any satisfactory correlations. However, using multivariate data analysis, i.e. a simultaneous regression of all variables against each other, a principal component analysis shows some reasonably distinct clusters and quite a high degree of correlation, see Fig. 1. The corresponding loading plot, Fig. 2, indicates small differences for the variables, making for a higher value of t [1]. However, in the second principal component the spread is quite large so that the higher temperature LOI will distinguish the group, including sample 18, 8 and others from the group with samples 10, 3 and 11. Samples in the blue group (negative values of t [1]) have lower values for all variables compared to those in other groups, but less so for the higher LOI temperatures. The highest values of TOC and OCcol in relation to LOI at higher temperatures are found in samples 1 and 2. These clusters cut across groupings on physical grounds (fuel, grate or fluidised bed, bottom ash or fly ash, large or small furnace etc.). Knowing the type of furnace and type of fuel is therefore insufficient to determine the content of unburned carbon. In conclusion, both LOI and TOC vary over a wide range. The variations may to some extent be due to the uncertainty of the individual methods but also to variations in the sub-samples. In comparison to sample 7, the efficient combustor, the values of
Fig. 2. Loading plot corresponding to the principal component analysis of Fig. 1. The intensity of the factors indicated in the loading plot determines the placement of the different samples in the PCA of Fig. 1.
LOI and TOC in other samples are often an order of magnitude larger. The relationships between LOI at different temperatures, as well as between LOI and TOC, are also highly variable, even for the reasonably homogeneous group of mixed ash from grate furnaces. This raises questions regarding the consistency of such data in general and what is actually included in the results. The question to be discussed in the next section is how well these observations represent the amount of organic carbon remaining in the ash since this fraction will be biologically degradable and may give rise to redox gradients and greenhouse gas emissions. 3.2. The composition of UC Besides weight losses at a given temperature, the TGA experiments yield another type of result. The TGA was equipped with a quadrupole mass spectrometer, allowing the species released from the samples at different temperatures to be identified. The interpretation of the curves obtained from the TGA investigation can be illustrated with a few examples. The weight of the sample and the DTA signal recorded for sample 8 are shown in Fig. 3. The weight decreases continuously although there are three distinct steps: a first loss at 100–200 °C, a second loss from about 93% to 85% in the interval 300–400 °C, and a third loss from 82% to 75% at just above 700 °C. The DTA signal shows an exothermic phase (peak downwards) from 250 °C to 400 °C, corresponding to the second weight loss. The third step loss is accompanied by a small peak upwards, i.e. an endothermic phase. The QMS results for this test are shown in Fig. 4. There are two carbon dioxide emission peaks, one around 400 °C and the other between 700 °C and 800 °C. The main water emission peak is found below 200 °C and two small peaks are found on either side of the 400 °C mark.
895
H. Bjurström et al. / Fuel 117 (2014) 890–899 105
1,2
-0,4 -0,6
90
-0,8 85
-1,0
80
-1,2 -1,4
75
0,8 0,6 0,4 0,2
-1,6 -1,8 100 200 300 400 500 600 700 800 900 1000
0
0,0
100
-0,2 95
-0,4
90
-0,6
85
-0,8 -1,0
80 -1,2
0,8 0,6 0,4 0,2
75
-1,4
70
-1,6 100 200 300 400 500 600 700 800 900 10001100
0
Temperature (oC)
0,0
Temperature (oC)
Fig. 3. Thermogravimetry results for fly ash sample 8: TGA (black curve), DTA (red broken curve) and carbon dioxide emission from the sample (blue dotted curve).
Fig. 5. Thermogravimetry results for the aged bottom ash sample 2: TGA (black curve), DTA (red broken curve) and carbon dioxide (blue dotted curve).
1,2
1,2
1,2
1,0
1,0
1,0
0,8
0,8
0,6
0,6
0,4
0,4
0,2
0,2
ion current CO2 (10-9 A)
Ion current CO2 and H2O (10-9 A)
1,0
0,8
0,6
0,4
Ion current H2O (10-9 A)
70
Remaining weight (%)
95
1,0
Ion current CO2 (10-9 A)
-0,2
DTA signal (µV/mg)
Remaining weight (%)
100
0,0
Ion current CO2 (10-9 A)
1,2 0,0
DTA signal (µV/mg)
105
0,2 0,0 0
0,0 0
200
400
600
800
1000
100
200
300
400
500
600
700
800
0,0 900 1000
Temperature (oC)
Temperature (oC) Fig. 4. Thermogravimetry: emission of carbon dioxide (black curve) and water (blue broken curve) during the experiment on fly ash sample 8.
The first step at 100–200 °C is dehydration and the second step, in the interval 300–400 °C, is combustion producing large amounts of carbon dioxide, most probably elemental carbon as this step does produce only little water vapour. The third step, at 700– 800 °C, is decomposition of inorganic carbonate. A number of other processes are masked by these main weight losses. The two water emission peaks on either side of the 400 °C mark are probably not related to combustion only but also to the decomposition of, for example, hydroxides. If they had been caused only by combustion, the dip between the two peaks would also be seen in the carbon dioxide peak. However, these relatively small variations on the curves are difficult to interpret and one should also remember that EC in combustion residues might not be pure carbon but contain some hydrogen and oxygen, as described in Ferrari et al. [2] and van Zomeren and Comans [4]. The end of the exothermic phase in the DTA signal is also rather abrupt, indicating that an endothermic process is taking place at the upper end of the combustion range rather than it simply being the combustion tapering off. Furthermore, the small ‘wiggles’ on each curve could represent a minor process that is taking place at that temperature. The results for sample 1 are analogous; see Figs. 5 and 6. The first dehydration step is not noticeable. The second weight loss step is large, from 95% to 77%, is exothermic and occurs between 400 and 600 °C. The third is small but distinct, 77–74%, is endothermic and occurs between 650 °C and 700 °C. The transition between the two latter steps is abrupt in the DTA curve. Water is
Fig. 6. Thermogravimetry results for the aged bottom ash sample 2: emission of carbon dioxide (black curve) and water vapour (red broken curve).
released before the combustion step, around 100 °C and 300 °C. There is a small hump on the carbon dioxide emission curve at 650–700 °C. Smidt et al. [17] and more recently Rocca et al. [8] point out that between 450 °C and 600 °C weight loss in a TGA experiment may pause: the carbon dioxide produced during combustion is reabsorbed as carbonate by the ash [17]. This could also be the reason for the abrupt end in Figs. 3 and 5 of the exothermic process after combustion of residual carbon. The TGA/DTA/QMS results for all the samples have common features but there are some variations: The two carbon dioxide peaks may be distinct, as in Fig. 4, or merge into one broad peak, as in Fig. 6. For a sample with a very low TOC, only the carbonate peak may be visible, not the combustion peak. The opposite is also found: sample 21 only has a combustion peak, not a carbonate peak. The temperature in the sample at which carbon dioxide is detected in the QMS, i.e. the step weight loss due to combustion, is variable and depends on the type of residue. The combustion is completed by 400 °C for fly ash from large plants, but only starts at 400 °C for bottom ash from large plants. In the two APC residues that consist mainly of fly ash, with its UC and a minor quantity of activated carbon, the combustion step starts at higher temperatures than in fly ash: at 400 °C for sample 11, peaking at slightly below 450 °C and at 500 °C, and for sample
896
H. Bjurström et al. / Fuel 117 (2014) 890–899
Table 5 Interpretation of TGA results as organic carbon, elemental carbon and inorganic carbon in the combustion residues with TOC results for reference. Units: w/w %. Residue stream
Sample
OC
EC
IC
TOC
Bottom ash Bottom ash, aged Bottom ash, aged Bottom ash, aged Bottom ash Bottom ash Fly ash and APC residues Fly ash and APC residues Fly ash and APC residues Fly ash and APC residues Fly ash and APC residues Fly ash and APC residues Mixed fly and bottom ash Mixed fly and bottom ash Mixed fly and bottom ash Mixed fly and bottom ash Mixed fly and bottom ash Mixed fly and bottom ash Mixed fly and bottom ash Mixed fly and bottom ash
1 2 5 6 9 10 3 7 8 4 11 21 13 17 20 16 14 15 18 19
0.6 1.3 0.6 0.5 0 2.1 0.6 0.3 0.5 2.7 0.6 0.6 0.2 0.3 0.5 0.9 0.2 0.3 0.6 1.0
16.9 19 1.9 1.0 1.3 9.7 10.4 0.14 10 8.6 4.4 0 0.3 0.25 0 6 7.6 9.6 8 11.7
0.2 0.4 0.1 0.8 0.2 0.3 0.1 0.2 3.0 2.8 1.4 1.6 1.9 2.5 2.0 2.7 3.4 2.0 1.9 1.9
20.2 21.4 0.6 0.4 1.0 10.2 12 0.4 15.4 12.2 10.6 1 1.4 2.9 0.8 11.5 5.8 13.0 6.9 11.1
21 peaking at slightly above 650 °C. The samples from the small pellet furnaces, samples 13–20, behave similarly although the processes occur at another set of temperatures: a combustion peak with weight loss below 400 °C and a carbonate peak with a large weight loss, mostly above 700 °C. In the case of low TOC samples – samples 13, 17 and 20 – the combustion peak is distinct in carbon dioxide emission and DTA signal but not in weight loss (TGA). The difference in combustion temperatures could be the result of a matrix effect, as demonstrated by Müller and Belevi [18]. Interpretation of the TGA results as organic carbon (OC), elemental carbon (EC) and inorganic or carbonate carbon (IC) (following, for example, Ferrari et al. [2]) is summarised in Table 5. For the values to be commensurate, weight losses have been recalculated: Decomposition of carbonates has been recalculated to carbon, which is straightforward. Recalculating weight loss from organic substance to carbon requires an assumption of the carbon content of organic substances. A factor of 58% carbon in organic matter in the colorimetry standard ISO 14235 has been used rather than a factor of 40%, which is more representative of cellulosic material. Emissions of water and carbon dioxide, as registered in the QMS, are difficult to quantify.
What is the structure of elemental carbon in the combustion residues? How might the elemental carbon in combustion residues impact on their environment? In order to distinguish between different carbon structures, i.e. crystalline and amorphous forms, exploratory tests were carried out using micro-Raman spectroscopy, scanning in the range 200– 3000 cm1. In all spectra collected for samples, the double peak is found to be characteristic of C–C carbon bonds in elemental carbon: the ordered or graphitic structure (G) peaks at about 1600 cm1 and the disordered structure (D) peaks at about 1350 cm1. The disordered form is composed of randomly orientated microcrystals (or turbostratic layers, see e.g. [19,20]) while graphite is characterised by a crystalline structure with carbon atoms organised on specific planes. A comparison of spectrograms for a residue and for industrially activated carbon (DioxSorb BP2) shows that the carbon structures are very similar, see example in Fig. 7. Furthermore, the ratio of peak heights D/G, Table 6, was about 1.33 for all samples, which implies that the proportion of D and G forms of carbon is similar in all samples, including the activated carbon. The properties of elemental carbon determined by Raman spectroscopy are similar in both bottom ash and fly ash and are also similar to that of activated carbon. Samples 16 and 19 from pellet furnaces are outliers. There was not enough UC in samples 14 and 17 for the carbon double peak to be seen in the micro-Raman experiments. In sample 14, a stick or flake of charred material was retrieved from the larger sample and subjected to micro-Raman. Looking at the position of the Raman spectroscopy peaks for the samples, it is obvious (Fig. 8) that peak positions for most of the residues cluster together and are close to those of the activated carbon. There are, however, outliers on both sides of the cluster, i.e. sample 7, which is an exceptionally well-combusted ash, and the poorly burned piece of wood in sample 14. The temperature to which the carbon has been subjected affects the structure of the elemental carbon, see e.g. Franklin [21]. The higher the temperature, the more graphitic the structure becomes and the larger the microcrystals. The effect on the Raman spectra is a narrowing of the peak or bandwidth (see e.g. [22]). To test the data, the G peak width as a function of microcrystal size was calculated using the Tuinstra–Koenig empirical formula [23], as shown in Fig. 9. One would have expected to find bottom ash samples at
13000
The numbers in Table 5 demonstrate that elemental carbon dominates in the TOC, i.e. TOC is mainly EC.
12000
Results from the thermogravimetric analysis lead to the conclusion that elemental carbon (EC) is the dominant part of the unburned carbon (UC) in combustion residues and this is independent of the value of the content of UC, which within the set of samples varies from almost 0% to more than 90%. The weight loss interpreted as organic carbon (OC) represents a much smaller part of UC. The ratio of organic carbon to elemental carbon (OC/EC) did vary within broad limits from 0.05 to 1 but no specific relationship could be established for different types of ash. The variation within each group is far greater than between the different ashes. The dominance of the elemental species in the unburned carbon raises questions regarding the properties and the behaviour of the ash in different environments:
Counts (-)
11000
3.3. Properties of elemental carbon in UC of combustion residues
10000 9000 8000 7000 6000 800
1000
1200
1400
1600
1800
Raman shift (cm-1) Fig. 7. Raman spectrum for the aged bottom ash, sample 2 (black curve) and for the activated carbon reference sample (red broken curve).
897
H. Bjurström et al. / Fuel 117 (2014) 890–899 Table 6 Results of the analysis of the carbon double peaks in the Raman spectra. Residue stream
Sample no
D/G ratio (–)
D peak position (cm1)
D peak width (cm1)
G peak position (cm1)
G peak width (cm1)
La (nm)
Bottom ash, aged Bottom ash, aged Bottom ash Cyclone ash ESP ash ESP ash APC residue APC residue Mixed ash Mixed ash Charred stick DioxSorb BP2
2 10 12 3 7 8 11 21 16 19 From 14 –
1.41 0.97 1.23 1.40 1.51 0.86 1.46 1.38 1.91 3.91 1.55 1.57
1336 1345 1342 1339 1383 1336 1326 1354 1334 1337 1314 1328
183 226 246 161 319 144 179 246 163 131 236 178
1595 1586 1587 1585 1610 1591 1597 1603 1591 1589 1569 1593
103 216 117 201 260 236 102 252 139 186 99 106
3.1 4.5 3.6 3.1 2.9 5.1 3.0 3.2 2.3 1.1 2.8 2.8
350
1610 300
FWHM for D peak (cm-1)
Position of the G band (cm-1)
1620
1600
1590
1580
1570
1560 1310
250
200
150
1320
1330
1340
1350
1360
1370
1380
1390
Position of the D band (cm-1)
100 0
1
2
3
4
5
6
Microcrystallite size (nm) Fig. 8. Position of D and G bands from micro-Raman spectroscopy of combustion residue samples: j, bottom ash, s, fly ash and APC residue, d, mixed ash, }, activated carbon and h, charred wood stick.
the narrowest peak widths but there does not seem to be any general tendency in bandwidth. Most samples cluster at about 3 nm in microcrystallite size and samples 16 and 19 are again outliers. Activated carbon is well known for its large specific surface, from 500 to 1500 m2/g depending on the raw material and manufacturing process. This gives it a high adsorption capacity for both inorganic and organic substances. The values of the specific surface of 30–50 m2/g reported for various coal and wood fly ashes [24–26] are large in comparison to those of e.g. quartz sand (0.01 m2/g) and silt (0.5 m2/g). If the EC-content contributes to a large specific surface in the ashes, than this may enhance the adsorption capacities for both organic and inorganic substances. As a brief indication adsorbed organic substances was studied by screening samples 7, 8 and 2, representing low medium and high organic content. The number of semi-volatile substances detected was quite low and their concentrations were far lower than 8% of TOC. A similar absence of volatile organic substances was noted by Kuo et al. [27] for charcoal pyrolysed above 400 °C and by Spokas for ash from combustion of biomass [6]. The substances that could be identified with some certainty include the ubiquitous phthalates (DBP in the non-derivatised samples and DEHP in the derivatised samples), which are not residues of incomplete combustion but rather post-combustion contaminants. Other contaminant substances are (z)-9-octadecenamide (oleamide), a slip agent used in the plastic bags in which the samples were stored, and 2, 6-di-t-butyl-4-methyl phenol or DBP, an antioxidant in laboratory gloves. One possible interpretation is that the substances are adsorbed as contaminants in the
Fig. 9. Peak width of D peak as a function of micro-crystallite size from microRaman spectroscopy of combustion residue samples: j, bottom ash, s, fly ash and APC residue, d, mixed ash, }, activated carbon and h, charred wood stick.
EC phase after the generation of the residues. Contamination of the samples was a major issue in Pavasar’s screening of organic compounds in, among others, combustion residues [28]. The presence of semi-volatile substances that could be expected to disappear in the combustion process indicates that these substances can be adsorbed and ‘stored’ in the combustion residue. It has been shown that by applying activated carbon to contaminated sediments, the bio-uptake of PCBs in benthic worms was reduced by 69–99% [29]. From this and other studies it is known that activated carbon, due to its strong sorption capacity, reduces the bioavailability of organic contaminants in soil and sediments (see review by Hilber and Bucheli [30]). Most studies were performed in laboratories although recent field-scale studies confirm the potential of activated carbon as an effective amendment that adsorbs PAH [31]. Of equal interest is the documented substantial adsorption capacity of black carbon (e.g. [32]) and that black carbon may act as an effective adsorbent of PAH, reducing the bio-uptake of organic contaminants from soil and sediments by up to two orders of magnitude (see review article by Koelmans et al. [33]). The metal adsorption capacity of black carbon and activated carbon is well documented (e.g. [34–36]). The combustion residues contain UC including elemental carbon that may act as a potent adsorbent and it can be expected that the leaching behaviour of the ashes is influenced by the structure of the EC. One of the major obstacles to the use of ash in, for example, construction is the leaching of potentially hazardous substances and the basis of sustainable use of solid biofuel ash is a
898
H. Bjurström et al. / Fuel 117 (2014) 890–899
leaching process that can be retarded to fit the biogeochemical cycles of the recipients where it is used. The presence of EC is an important factor to understand. Another use of unburned carbon in ash may be as a cheap precursor to active carbon, as proposed by, among others, Maroto-Valer [24] and Bartonova et al. [26]. The role of the EC in the leaching process should be investigated further – as should the possibility of activating the elemental carbon by treating the ash in different ways, e.g. using steam or separation techniques.
4. Conclusions In a range of combustion residues from different fuels, furnaces and air pollution control processes, it was observed, through the use of TGA, that the unburned carbon is mainly elemental. At least 50% of the unburned carbon was found to be elemental carbon and for most samples considerably more. The content of elemental carbon was >10% of the total ash mass for several of the samples studied. The proportion and composition of the unburned carbon in the residues vary more with the combustion process than with the type of fuel. LOI at different temperatures, TOC and chemical oxidation all resulted in a variation of more than a magnitude between the best and most poorly incinerated sample. The LOI and TOC showed little correlation to organic carbon, whereas chemical oxidation showed some correlation to elemental carbon, especially in bottom ash. Since the waste is generated at temperatures of around 800 °C, it is not surprising that LOI at 550 °C or TOC at 900 °C would yield somewhat deviating results although no clear trends for the variation in temperature were found. From the magnitude of variation, it can be inferred that 550 °C is the least suitable temperature when testing ash using LOI. However, since loss of salts can occur at the highest temperature tested, the intermediate temperature, 750 °C, seems to be the safest alternative. The only property that was common to all samples was the relationship between graphitic and disordered forms of elemental carbon, which was about 1.33 on a mass-to-mass basis. This is also the case for commercial activated carbon. Observations indicated that organic compounds from sample bags and rubber gloves had been adsorbed onto the samples and the ash may thus be expected to act as a retainer as well as a releaser of contaminants. The organic compounds present in unburned carbon in bottom ash and fly ash are mainly non-volatile, macromolecular compounds and similar compounds may be formed slowly through the dissolution of graphitic carbon. Unfortunately, there is still no simple analytical procedure to assess the quality and even the amount of unburned carbon in residue. It is highly misleading to use TOC as measure of organic carbon in combustion residues. Unburned carbon in biofuel ash should not á priori be considered an environmental risk factor and in the process of balancing energy efficiency, economy and environmental impact of the residue, the EC content should be considered.
Acknowledgements The examination reported here took place within several projects with financial support mainly from Värmeforsk (the Swedish Thermal Engineering Research Institute). Additional support was obtained at the different stages from Åforsk (Research foundation of the ÅF Group), Skelleftekraft AB, Fortum Värme AB, Öresundskraft AB, Iggesund Paperboard AB, Holmen Paper AB, Svenska Energiaskor AB. Jacobi Carbon Systems AB graciously donated a sample of the activated carbon DioxSorb BP2.
Collaborators in the projects were Pascal Suér and Lennart Larsson (SGI), Menad Noureddine (Luleå University of Technology), Jacob Thyr and Lars Hälldahl (K-Analys AB).
References [1] Vosteen B, Beyer J. Obtainable residual carbon content in slags and ashes from incineration systems. VGB PowerTech 2000;9:48–53; Vosteen B, Beyer J. Obtainable residual carbon content in slags and ashes from incineration systems. VGB PowerTech 2000;10:73–5. [2] Ferrari S, Belevi H, Baccini P. Chemical speciation of carbon in municipal solid waste incinerator residues. Waste Manag 2002;22:303–14. [3] Kumpiene J, Robinson R, Brännvall E, Nordmark D, Bjurström H, Andreas L, et al. Carbon speciation in ash, residual waste and contaminated soil by thermal and chemical analyses. Waste Manag 2011;31:18–25. [4] van Zomeren A, Comans R. Carbon speciation in municipal solid waste incinerator (MSWI) bottom ash in relation to facilitated metal leaching. Waste Manag 2009;29:2059–64. [5] Cohen-Ofri I, Weiner L, Boaretto E, Mintz G, Weiner S. Modern and fossil charcoal: aspects of structure and diagenesis. J Archaeol Sci 2006;33:428–39. [6] Spokas K. Review of the stability of biochar in soils: predictability of O:C molar ratio. Carbon Manag 2010;1:289–303. [7] Brown RC, Dykstra J. Systematic errors in the use of loss-on-ignition to measure unburned carbon in fly ash. Fuel 1995;74:570–4. [8] Rocca S, van Zomeren A, Costa G, Dijkstra J, Comans R, Lombardi F. Mechanisms contributing to the thermal analysis of waste incineration bottom ash and quantification of different carbon species. Waste Manag 2013;33:373–81. [9] SS 18 71 87. Solid fuels – determination of the amount of unburned material in solid residues from combustion; 1984. [10] EN 196-2:2005. Cement – testing – Part 2: chemical analysis; 2005. [11] Chandler AJ, Eighmy TT, Hartlén J, Hjelmar O, Kosson DS, Sawell SE, et al. Municipal solid waste incinerator residues. Amsterdam: Elsevier; 1997. [12] ASTM D2974-00. Standard test methods for moisture, ash and organic matter of peat and other organic soils; 2000. [13] EN 13137:2001. Characterisation of waste. Determination of total organic carbon (TOC) in waste, sludges and sediments; 2001. [14] EN 1744-1:1998. Ballast – chemical properties – Part 1: chemical analysis, 1st ed.; 1998. [15] ISO 14235:1998; Soil quality – determination of organic carbon by sulfochromic oxidation; 1998. [16] Kylefors K, Ecke H, Lagerkvist A. Accuracy of COD test for landfill leachates. Water, Air, Soil Pollut 2003;146:153–69. [17] Smidt E, Meissl K, Tintner J, Ottner F. Interferences of carbonate quantification in municipal solid waste incinerator bottom ash: evaluation of different methods. Environ Chem Lett 2010;8:217–22. [18] Müller C, Belevi H. Differentiating between organic and elemental carbon compounds in residues from municipal solid waste incinerating plants. Müll und Abfall 2005;2005:352–7 [in German]. [19] Bourrat X, Langlais F, Chollon G, Vignoles G. Low temperature pyrocarbon: a review. J Braz Chem Soc 2006;17:1090–5. [20] Keiluweit M, Nico PS, Johnson MG, Kleber N. Dynamic molecular structure of plant biomass-derived Black Carbon (Biochar). Environ Sci Technol 2010;44:1247–53. [21] Franklin R. Crystallite growth in graphitizing and in non-graphitizing carbons. Proc Roy Soc A 1951;209:196–218. [22] Rhim Y-R, Zhang D, Fairbrother DH, Wespanick KA, Livi KJ, Bodnar RJ, et al. Changes in electrical and microstructural properties of microcrystalline cellulose as function of carbonization temperature. Carbon 2010;48:1012–4. [23] Tuinstra F, Koenig JL. Raman spectrum of graphite. J Chem Phys 1970;53:1126–30. [24] [24 Maroto-Valer MM, Taulbee DN, Schobert HH, Hower JC, Andrésen JM. Use of unburned carbon in fly ash as precursor for the development of activated carbons. In: International ash symposium. Centre for Applied Energy Research, Univ of Kentucky; 1999. Paper 19. [25] Hwang JY, Sun X, Li Z. Unburned carbon from fly ash for mercury adsorption. I. Separation and characterization of unburned carbon. J Miner Mater Charact Eng 2002;1:39–60. [26] Bartonova L, Juchelkova D, Klika Z, Cech B. On unburned carbon in coal ash from various combustion units. World Academy of Science, Engineering and Technology; 2011. p. 76. [27] Kuo L-J, Louchouarn P, Herbert BE. Fate of CuO-derived lignin oxidation products during plant combustion: Application to the evaluation of char input to soil organic matter. Org Geochem 2008;39:1522–36. [28] Pavasars I. Characterisation of organic substances in waste materials under alkaline conditions. Doctorate thesis presented at Linköping University, Linköping; 1999. [29] Beckingham B, Ghosh U. Field-scale reduction of PCB bioavailability with activated carbon amendment to river sediments. Environ Sci Technol 2011;45:10567–74. [30] Hilber TD, Bucheli. Activated carbon amendment to remediate contaminated sediments and soils: a review. Global NEST J 2010;12:305–17.
H. Bjurström et al. / Fuel 117 (2014) 890–899 [31] Hale EE, Elmquist M, Brändli R, Hartnik T, Jakob L, Henriksen T, et al. Activated carbon amendments to sequester PAHs in contaminated soil: a lysimeter field trial. Chemosphere 2012;87:177–84. [32] Brändli RC, Hartink T, Henriksen T, Cornelissen G. Sorption of native polyaromatic hydrocarbons (PAH) to black carbon and amended activated carbon in soil. Chemosphere 2008;73:1805–11. [33] Koelmans AA, Jonker MT, Cornelissen G, Bucheli TD, van Noort PC, Gustafsson Ö. Black carbon: the reverse of its dark side. Chemosphere 2006;63:365–77.
899
[34] Mohan D, Singh KP. Single- and multi-component adsorption of cadmium and zinc using activated carbon derived from bagasse – an agricultural waste. Water Res 2002;36:2304–18. [35] Qiu Y, Cheng H, Xu C, Sheng GD. Surface characteristics of crop-residuederived black carbon and lead (II). Water Res 2008;42:567–74. [36] Tay JH, Chen XG, Jeyaseelan S, Graham N. Optimising the preparation of activated carbon from digested sewage sludge and coconut husk. Chemosphere 2001;44:45–51.