Swelling and Char Structures from Density Fractions of Pulverized Coal

Swelling and Char Structures from Density Fractions ofPulverized Coal

Jianglong Yu,* John Lucas, Vladimir Strezov,† and Terry Wall

Cooperative Research Centre for Coal in Sustainable Development (CCSD) and Department ofChemical Engineering, University of Newcastle, Callaghan, NSW 2308, Australia

Received January 22, 2003. Revised Manuscript Received June 2, 2003

Char samples are prepared in a drop tube furnace (DTF) in a nitrogen atmosphere using density-separated coal samples of two Australian coals. The structure and morphology of the char samplesare analyzed under a scanning electron microscope (SEM). The porosities of individual chars aredetermined through image processing of the SEM images. The char particle size distribution(PSD) is analyzed using a Malvern Laser Sizer and through image analysis. The resultsdemonstrate the heterogeneity of the char structures and morphology among the density fractionsof the same coal. For both coals, porous chars with a cenospheric structure (Group I chars) aregenerated in the coal samples of low density (F1.25 and S1.25-F1.30). Solid char structures (GroupIII chars) are formed from the high density coal samples (S1.35-F1.50 and S1.50). The mediumdensity samples contain different types of chars with a moderate porosity and wall thickness.The weight loss, swelling ratio, and the porosity decrease with increasing parent coal density.Results of size fraction samples suggest that a larger swelling occurs in DTF when the particlesize decreases. Mechanisms for the formation of different char structures from different densityfractions are discussed, and influences of the presence of ash on char structure formation, inaddition to the dominant role of maceral compositions reported in previous works, are identified.

1. Introduction

Coal is well-known as a heterogeneous substance,with a mixture of organic material and mineral matter.The heterogeneity of the organic material in the coalarises from the fact that coal consists of complexmaceral constituents, classified in three groups, i.e.,liptinite (exinite) group, vitrinite group, and inertinitegroup.1 The heterogeneity of coal macerals causes widediversity in coal particle behavior, e.g., swelling, thermo-plasticity, burn-out propensity, etc., during the combus-tion and gasification process of pulverized coal.2-4

During the devolatilization stage, different types of charstructures are generated, namely cenospheric chars,foam structure (tenui-network), and dense chars (solidstructure). In recent work, the char structures havebeen classified into three major groups according toBenfell et al.5,6 In this classification system the group Ichars are porous (porosity > 80%) with a single void

and a thin wall in each char particle. The group II charhas a medium porosity (50%-80%) and wall thickness.The group III char has a relatively solid structure witha low porosity (<50%) and thick wall. The char structurehas been found to play an important role in the charcombustion and gasification processes.4,7 Fundamentalknowledge of char structural formation and swelling isessential to the development of advanced coal conversiontechnologies and in predicting coal combustion andgasification properties.

The fact that different structures are derived fromdifferent pf coal particles is attributed to the variationsof their maceral constituents.6-8 In general, particlescontaining liptinite or vitrinite generate porous charstructure, while those containing inertinite generaterelatively dense char structures. Therefore, samples ofselective maceral concentrations are quite useful forstudying char structure evolution during pyrolysis. Thiscan be achieved by selectively sampling from the dif-ferent coal faces.3-6 Empirical correlations between thechar structure and the vitrinite content have beenestablished by Liu et al.7 and Benfell et al.,6 as shownin the following equation:7

where nGrpI is the number percentage of group I char,

* Author to whom correspondence should be addressed at theDepartment of Chemical Engineering, P.O. Box 36, Monash University,Clayton, VIC 3800, Australia. Tel.: +61 3 9905 1961. Fax: +61 3 99055686. E-mail: [email protected].

† Newbolds Applied Research, The University of Newcastle, CNR,Frith & Gavey St., Mayfield, NSW 2304, Australia.

(1) Van Krevelen, D. W. Coal, typology-chemistry-physics-constitu-tion; Elsevier: Amsterdam-Oxford-New York, 1981; Vol. 3.

(2) Berkowitz, N. The chemistry of coal; Elsevier: Amsterdam, NewYork, 1985.

(3) Wall, T. F.; Tate, A. G.; Bailey, J. G.; Jenness, L. G.; Mitchell,R. E.; Hurt, R. H. Symp. (Int.) Combust., [Proc.] 1992, 24, 1207-1215.

(4) Benfell, K. E.; Liu, G.-S.; Roberts, D. G.; Harris, D. J.; Lucas, J.A.; Bailey, J. G.; Wall, T. F. Proc. Combust. Inst. 2000, 28 (Part 2), pp2233-2241.

(5) Benfell, K. E.; Bailey, J. G. 8th Australian Coal Science Confer-ence, Proceedings, 1998, Sydney, Australia; The Australian Instituteof Energy; pp 157-162.

(6) Benfell, K. E. Assessment of Char Morphology in High-PressurePyrolysis and Combustion. Ph.D. Thesis, Dept. of Geology, Universityof Newcastle, Newcastle, Australia, 2001.

(7) Liu, G.-S. Mathematical modelling of coal char reactivity in apressurised entrained flow gasifier. Ph.D. Thesis, Dept. of ChemicalEngineering, University of Newcastle (N.S.W.), Newcastle, Australia,xxii, 237 leaves, 1999.

nGrpI(%) ) 0.6 × Pt + 0.53 × vitr + 37

1160 Energy & Fuels 2003, 17, 1160-1174

10.1021/ef030022h CCC: $25.00 © 2003 American Chemical SocietyPublished on Web 06/28/2003

Pt is the total pyrolysis pressure (atm), and vitr is thevitrinite (including liptinite) content (%). Reactivity ofmaceral concentrations was also investigated by Cai etal.9,10

Density separation techniques, such as the sink-floattechnique and centrifugation, provide a detailed insightinto the nature of coal and coal macerals.11,12 The sink-float technique has been described in the literature1, andhas been extensively employed in recent years by coalresearchers.1,8,13-16 The principle of the density-separa-tion is due to the difference in the density of differentmacerals. Within the same coal, the density of maceralsfollows the order liptinite < vitrinite < inertinite.1Density separation can also be achieved using thecentrifugation method.11 During density separation, coalis usually crushed to a particle size below 38 µm in orderto effectively liberate maceral components. Minerals arealso chemically removed before the density separationoperation.8,11,12 Thus, a very high vitrinite concentration(above 95%) can be achieved in the light density fraction,while relatively high inertinite concentration results inthe heavy density-fraction sample.

Cloke et al.14 characterized density separated coalfraction samples using FTIR and petrographic analysis.It was concluded that Aar/Aal ratios increased for higherdensity samples, indicating a change in aromaticity asthe density of the fraction was increased. Kawashimaet al.17 used NMR technique to characterize the density-separated components during pyrolysis. They found thatthe lighter components underwent greater structuralchanges than the heavier components. The eliminationof aliphatic side chains occurred to a greater extent inlower rank coals, and the content of aliphatic moietiesin each maceral group determines its reactivity. Gilfillanet al.8,18 examined the structure and reactivity ofdensity-separated coal fractions of six world coals. Themorphological analysis of the char clearly indicated thatthin-walled chars were produced from the light densityfraction, and the proportion decreased with increasingdensity. The chemical structure investigations usingFTIR and 13C NMR showed that aromaticity increased

with coal density, with the inertinite-rich fractionshaving the highest value. In the performance of thedensity separation process, the particle size is usuallyreduced to below 38 µm.8 The raw coal particle sizesare even reduced to several micrometers to achieveeffective liberation of the maceral components, andminerals are removed prior to the density separation.12

In investigating the heterogeneity of large pf coalparticles during the devolatilization process, Gibbinsand his colleagues19 found some synergic effects duringdevolatilization, and suggested that the ability of usingrelatively pure macerals (from hand-picking, or micron-izing, demineralization, and density separation) torepresent actual pf combustion behavior may be limited.In their study, a wide range of particle mineral contentwas also observed, and the minerals were generally welldispersed and frequently associated with the inertinitemaceral. By measuring the overall char intrinsic reac-tivity, they suggested that minerals may act as acatalyst for char oxidation, but the effect is very small.Instead, some reduction in particle combustion ratesduring the later stage of char burnout occurred due tothe increased resistance to oxygen transport and heatlosses caused by the relatively large quantity of mineralmatter admixed with organic materials. It is thereforesuggested that both inertinite content and mineralmatter should be included to explain the poor burnoutof chars in some cases.

In practical situations of pf coal combustion andgasification, coals are not subjected to demineralization,and the typical pf size is 50 to 150 µm. Therefore, it isworth investigating the particle swelling behavior andchar structure under the conditions relevant to thenormal pf coal conversion processes. In the presentstudy, different maceral concentrations are obtainedthrough density separation using sink-float techniqueswithout demineralization. Char samples are preparedin a drop tube furnace, and are analyzed using differenttechniques.

2. Experimental Procedure

2.1. Coal Sample Preparation. Parent Coal. Two Aus-tralian coals were selected, and their properties are shown inTable 1. Coal A is a high volatile sub-bituminous coal withhigh liptinite content (24.9%). The volatile matter content ofcoal A is 40.2%, and the vitrinite reflectance is 0.44%. Coal Bis a bituminous coal with a high inertinite content (34.7%).Its volatile matter content is 29.7%, with a vitrinite reflectance0.73%. The coal samples were crushed to the size range +63-90 µm.

Density Separation. Density-fraction samples of the abovetwo coals were prepared using the sink-float method. Theparticle size of the coal in the present study was selected inthe pf size range (+63-90 µm), and demineralization was notperformed. CsOOCH (water solution) was used as the sink-

(8) Gilfillan, A.; Lester, E.; Cloke, M.; Snape, C. Fuel 1999, 78 (14),1639-1644.

(9) Cai, H. Y.; Megaritis, A.; Messenbock, R.; Vasanthakumar, L.;Dugwell, D. R.; Kandiyoti, R. Fuel 1998, 77 (12), 1283-1289.

(10) Cai, H. Y.; Megaritis, A.; Messenbock, R.; Dix, M.; Dugwell, D.R.; Kandiyoti, R. Fuel 1998, 77 (12), 1273-1282.

(11) Dyrkacz, G. R.; Horwitz, E. P. Fuel 1982, 61 (1), 3-12.(12) Dyrkacz, G. R.; Bloomquist, C. A. A.; Ruscic, L. Fuel 1984, 63

(10), 1367-1373.(13) Choi, C. Y.; Dyrkacz, G. R.; Stock, L. M. Energy Fuels 1987, 1

(3), 280-286.(14) Cloke, M.; Gilfillan, A.; Lester, E. Fuel 1997, 76 (13), 1289-

1296.(15) Rubiera, F.; Parra, J. B.; Arenillas, A.; Hall, S. T.; Shah, C. L.;

Pis, J. J. Fuel 1999, 78 (14), 1631-1637.(16) Sugawara, K.; Tozuka, Y.; Kamoshita, T.; Sugawara, T.; Sholes,

M. A. Fuel 1994, 73 (7), 1224-1228.(17) Kawashima, H.; Yamashita, Y.; Saito, I. J. Anal. Appl. Pyrolysis

2000, 53, 35-50.(18) Gilfillan, A.; Lester, E.; Cloke, M. DGMK Tagungsber.; Vol. 9702

(Proceedings ICCS ’97, Vol. 1), 1997; pp 239-242.(19) Gibbins, J. R.; Beeley, T. J.; Crelling, J. C.; Scott, A. C.;

Skorupska, N. M.; Williamson, J. Energy Fuels 1999, 13 (3), 592-601.

Table 1. Properties of Raw Coal Samples

proximate analysis % (a.d. basis)ultimate analysis % (daf) maceral analysis volume %(MMF)

coalair driedmoisture ash VM FC C H O N S V L I Rvmax(%)

A 6.70 12.10 40.20 41.00 78.60 6.10 13.70 1.07 0.48 62.00 24.90 13.10 0.44B 2.20 14.80 29.70 53.30 83.70 5.45 8.60 1.81 0.47 62.10 3.10 34.70 0.73

Char from Density Fractions of Pulverized Coal Energy & Fuels, Vol. 17, No. 5, 2003 1161

float separation medium. Both coals were separated into fivedensity fractions, i.e., F1.25, S1.25-F1.30, S1.30-F1.35, S1.35-F1.50, and S1.50.

Ultimate analysis and proximate analysis have been carriedout on the density-fraction samples, and the results arepresented in Table 2. With increasing the coal density, VMcontent (db basis) decreases sharply, while ash content in-creases. Both the carbon content and hydrogen content gener-ally decrease with increasing coal density, while oxygencontent increases. For coal A, the carbon content decreases5.5% (daf), while H content decreases 0.3% (daf) from the F1.30density fraction to the S1.50 density fraction. For coal B, thecarbon content decreases from 84.4% (daf) to 81.4% (daf), whileH content decreases from 5.50% (daf) to 5.26% (daf) from theF1.30 density fraction to the S1.50 density fraction. Petro-graphic analysis was carried out to determine the maceralcomposition of each coal sample based on the measurementof a reflectogram, and the results are presented in Table 2.As raw coal samples have not been subjected to demineral-ization, a very high level of ash is contained in the heavydensity fractions, i.e., F1.50 and S1.50. However this does notchange the general trend, the inertinite content increaseswhile the liptinite + vitrinite content decreases as the rawcoal density increases. The trend is more apparent for coal B,which has an inertinite content of around 10% (daf) for lightdensity fractions and around 60% (daf) for heavy densityfractions. The particle size distributions of raw coal samplesare measured using the Malvern Laser Sizer. The resultssuggest that the particle size distribution of raw coal samplesis independent of the coal density.

Size-Fraction Samples. In addition to the density-fractionsamples, size-fraction samples were prepared for both coals.Four size ranges are selected, i.e., +45-63, +63-75, +75-90, and +90 µm. Raw coal samples were analyzed beforepyrolysis experiments were conducted. Ultimate and proxi-mate analysis data show there are no major changes inproximate and elemental composition among the different sizefractions for the same coal, which are similar to that in Table1. PSD of size-fraction raw coal samples were measured usingboth Malvern Laser Sizer and SEM image analysis. Morphol-ogies of raw coal particles were also examined under SEM.

Char Sample Preparation in DTF. The char samples wereprepared in a drop tube furnace at atmospheric pressure, innitrogen gas, using density-separated samples of both coals.Two peak temperatures, i.e., 1373 K and 1573 K, were used.A fluidized-bed feeder has been used in the present experi-ments to feed in coal samples. The primary flow gas carriescoal particles into the furnace from the top through a water-cooled feeding probe. The coal feed rate is ∼2 g/h. Secondarygas flow enters the furnace from the top. The residence timeis calculated from and controlled by the gas flow rates, and is

typically 300-400 ms in the present study. The schematicdiagram of the reactor has been described elsewhere.20 Atypical heating rate of pf particles in the drop tube furnace is1-5 × 104 K/s, by calculation.21 The water-cooled samplingprobe was extended into the furnace from the bottom, andchars were collected in the cyclone. Coal samples were driedat 110 °C in the oven before pyrolysis experiments. The flowrate of the primary gas, secondary gas, and quenching gas andsuction rate were carefully adjusted to ensure that the furnaceatmosphere was reducing. The feeding probe, sampling probe,furnace, and the cyclone were completely cleaned before eachexperiment run so that char samples were not contaminated.

2.2. Char Characterization. Determination of WeightLoss. Proximate analysis was carried out to determine thevolatile content, fixed carbon, and ash content of char samples.The weight losses (∆W) and Q factors during the pyrolysis inthe DTF were determined using the total ash tracer. Q factoris defined as Q ) ∆W/(VM0 - R), where R is the volatile mattercontent of the char, and VM0 is the proximate volatile contentof the parent coal. The ash tracer technique has been employedoften and is described in the literature.21-23

Particle Size Distribution of Chars. The particle size distri-bution of the char and coal was measured using a MalvernLaser Sizer. The char morphology was observed under ascanning electronic microscope (SEM). Here, char particleswere spread on stubs, and coated with gold or carbon. Two orthree stubs were used for each char sample. Char particle sizedistribution was also measured through SEM image analysisusing the ImageTools V2.00 software provided by UTHSCSADental Diagnostic Science.24 The final swelling ratio of the charwas obtained by comparing the average char particle size tothe average raw coal particle size.

Porosity of Individual Char Particles. To analyze charstructures, char samples-in-section were prepared using resinas the binding material. Each sample was then coated withcarbon, and analyzed under SEM. The porosity of individualchar particles was analyzed through SEM image analysisusing the same ImageTools software.24 The char structure type(i.e., group I, group II, or group III) of each char particle was

(20) Yan, L. CCSEM analysis of minerals in pulverised coal and ashformation modelling. Ph.D. Thesis, Dept. of Chemical Engineering,University of Newcastle (N.S.W.), Newcastle, Australia, xxi, 250 leaves,2000.

(21) Jamaluddin, A. S.; Truelove, J. S.; Wall, T. F. Combust. Flame1986, 63 (3), 329-337.

(22) Badzioch, S.; Hawksley, P. G. W. Ind. Eng. Chem., Process Des.Dev. 1970, 9 (4), 521-530.

(23) Yeasmin, H.; Mathews, J. F.; Ouyang, S. Fuel 1998, 78 (1), 11-24.

(24) Dove, S. B.; U. D. D. Science; 2002; ImageTool ApplicationVersion 3.0 Final; UTHSCSA Dental Diagnostic Science; http://www.ddsdx.uthscsa.edu/dig/download.html; Internet website.

Table 2. Properties of Density-Separated Coal Samples

Coal & Coal Sample

A B

S1.25-F1.30 S1.30-F1.35 S1.35-F1.50 S1.50 F1.25 S1.25-F1.30 S1.30-F1.35 S1.35-F1.50 S1.50

Proximate Analysis(db)(%)Ash 4.1 10.8 25.5 45.7 0.8 2.7 7.2 26.0 45.2VM 38.7 40.4 38.8 27.3 41.2 37.9 33.8 28.1 21.7FC 57.2 48.8 35.7 27.0 58.0 59.4 59.0 45.9 33.1

Ultimate Analysis(daf) (%)C 84.3 80.8 80.2 78.8 84.38 83.23 83.85 84.9 81.41H 5.88 5.7 5.68 5.5 5.49 5.72 5.4 5.0 5.26N 2.07 1.64 1.54 1.98 2.07 1.99 1.90 1.91 1.69O 7.15 11.42 12.13 13.1 8.67 8.63 8.44 7.8 10.78S 0.57 0.48 0.45 0.58 0.10 0.43 0.41 0.39 0.86

Maceral Content (MMMF)(%)V 67.8 35.4 42.6 59 78.8 86.6 55.4 33.4 33.8L 25.8 59.2 46.6 24.6 8.6 3.2 7.4 3.2 6.0I 6.4 5.4 10.8 16.4 12.6 10.2 37.2 63.4 60.2

1162 Energy & Fuels, Vol. 17, No. 5, 2003 Yu et al.

determined according to the 3-fold char structure classificationsystem developed by Bailey and Benfell,4,5,25 mainly on thebasis of its porosity in the present study and internal config-uration. As such, the char structure distribution of each charsample was measured. It may be noted that the averageporosity measured using image analysis is always lower thanthe true porosity as a result of the following: (1) in cross-sectionsamples, char particles are not cut across the maximumgeometric diameter; and (2) microporosity cannot usually bemeasured because of the resolution required for the measure-ment.

The image analysis for porosity measurement involves thefollowing three steps, as illustrated in Figure 1:

(i) Image Separation. Each individual char particle (in grayscale) is separated from the whole picture, so that each charimage subjected to the ImageTools measurement contains onlyone char particle. Figure 1A shows single char cross-sectionimages separated from the whole SEM image.

(ii) Measurement of the Total Char Particle Area. In thisstep, the gray scale image is converted into binary images byadjusting the threshold of the image, shown in Figure 1B.Then, the area of the whole char particle, Ach, is measured bythe “object analysis function” of the ImageTools.

(iii) Measurement of Inner Particle Pore Area, Ap. These arethe black objects inside the white area, as shown in Figure1C. The porosity is then given by Ap/Ach (%). The structuretype, i.e., group I, group II, or group III, of the measured charparticles is then determined on the basis of the porosity value.

3. Results and Discussion

Results of DTF chars prepared using density-sepa-rated samples demonstrate the heterogeneity of charchemistry, including weight loss, swelling, char mor-phology, char internal structure, porosity, etc.

3.1. Results of Char Samples of Density Frac-tion. Weight Loss. Figure 2 shows weight losses (dafbase) of density-fraction samples during pyrolysis at apeak temperature 1573 K for coal A (Figure 2a) and coalB (Figure 2b). As expected, there is a clear trend thatthe weight loss during pyrolysis decreases with increas-

ing density for both coals examined, although thedecrease for coal A is significantly smaller than coal B.The decrease in the weight loss for the heavy densityfraction is the result of the lower liptinite and vitrinitecontent (and consequently higher inertinite content).Liptinites and vitrinites generate higher volatiles duringpyrolysis than inertinite macerals1. Coal A containsmuch less inertinites, 13.1%, compared to coal B, whichcontains 34.1% inertinite. This may provide someexplanation as to the smaller decrease in the weight lossfor the higher density samples in coal A.

Decreases in the Q factors for heavy density-fractionsamples are also observed for both coals, as shown inFigure 2. The value of the Q factor is mainly associatedwith tar generation and release.26 The excess volatileyields (Q factor > 1.0) are mostly the consequence ofreduced tar cracking and less carbon deposition, which

(25) Bailey, J. G.; Tate, A.; Diessel, C. F. K.; Wall, T. F. Fuel 1990,69 (2), 225-239.

Figure 1. Image analysis procedure of porosity measurement of individual char particles using cross-section SEM images; (A)gray scale image separated from whole cross-section SEM image, (B) binary image converted from picture A, (C) color-reversedimage converted from image B.

Figure 2. Weight losses of density fractions of coals A and Bduring pyrolysis at 1573 K; (a) coal A, (b) coal B.


is achieved by the reduction of contact between reactivetars and hot particle surfaces at high heating rates.26

Hodek et al.27 found the tar yields decrease in thesequence exinite (liptinite) f vitrinite f inertinite. Thisconclusion supports the observations of the presentstudy. With increasing coal density, liptinite and vit-rinite content decreases while the inertinite contentincreases. Therefore, the Q factor is smaller for theheavy density fractions as a result of lower tar yields.

Char Particle Size Distribution (PSD) and Swelling.Particle size distributions (PSD) of char samples ofdensity fractions of coal A and B measured usingMalvern Laser Sizer are presented in Figure 3. Asexpected, a shift of char particle sizes to a larger sizerange for light density fractions (F1.25 and F1.30) forboth coals, while there is little shift for heavy densityfractions (F1.50 and S1.50).

The trend is more clearly revealed in Figure 4 bycomparing the PSD of chars to that of raw coal samples,in (a) and (c) for F1.30, in (b) and (d) for F1.50; for bothcoals, as such, higher swelling ratios are expected forthe light density-fraction samples. PSD shifts in lightdensity fractions for coal A are larger than that of coalB, by comparing Figures 4a and 4c, implying largerswelling ratios for coal A. This is probably due to ahigher liptinite and volatile content in coal A. From the

open literature,26,28-30 it is generally recognized that coalparticle swelling during devolatilization is contributedby fluidity and volatile yields. The surface tension hasa major role in retaining volatile gases in the liquid coalparticle, which causes high swelling of coal particles.

The char particle size distributions (by volume) of thedensity fractions of coal A obtained through SEM imageanalysis using size bin I are shown in Figure 5. Sizebin I is the same as the size bin used for the MalvernSizer measurements. The volume distribution wasconverted from the number distribution. Trends similarto those from Malvern Sizer measurements are appar-ent. Light density-fraction samples (F1.25 and F1.30)have large particle size shifts, Figure 5a, and theirparticle sizes are distributed in a larger size range, asshown in Figure 5b.

Figure 6 shows particle size distributions (by volume)of coals A and B, using size bin II, obtained throughSEM image analysis. Size bin II is a more refined sizerange. The larger particle size shifts observed for lightdensity-fraction samples are similar in the cumulativecurves of Figures 6a and 6c for both coals and to thoseshown in Figure 5a. However, size distributions inFigures 6b and 6d show that the F1.25 and F1.30 aremost likely to have bi-model or tri-model peaks. The sizerange of peak I is the same as the heavy density-fractionsamples and the raw coal particle size range. Peaks IIand III are larger particle sizes contributed by thoseparticles that experienced more swelling during pyroly-sis. It is logical to consider that particles of peaks II andIII evolve from liptinite- and vitrinite-containing coalparticles, while chars forming peak I come from coalparticles containing mainly inertinite maceral or withhigh ash content. However, peaks II and III are notobvious in heavy density-fraction samples, i.e., F1.50and S1.50, as shown in Figures 6b and 6d. For the highdensity samples, the particles are distributed within sizepeak I, which has the same size range as that of theraw coal samples. This is because the particles of high-density samples either have high inertinite content orhigh ash content. The inertinite maceral will exhibit lessswelling due to its lower volatile yield and absence offluidity and will therefore retain the size distributionof the raw or un-transformed coal particles. Highmineral matter content would have a significant effecton coal pyrolysis behavior in that: (a) ash grainsconsume energy during heating, therefore reducing theparticle heating rate; (b) ash grains displace organicmaterial in the given particle, and therefore serve toreduce volatile yields; (c) ash grains are usually solidat pyrolysis temperatures, therefore the particle viscos-ity is high; (d) when ash content is very high, ash grainsmay isolate organic materials, therefore the pyrolysisreactions occur in isolated sections of the particle, andthe whole particle does not exhibit fluidity and soften-ing.

The swelling of char particles plotted against parentcoal fraction density is shown in Figure 7 and Figure 8.Char swelling, measured using both the Malvern Sizer

(26) Howard, J. B. Fundamentals of coal pyrolysis and hydropy-rolysis. In Chemistry of coal utilization; Elliott, M. A., Ed.; John Wiley& Sons: New York, 1981; p 665.

(27) Hodek, W.; Kirschstein, J.; van Heek, K.-H. Fuel 1991, 70, 424-428.

(28) Gao, H.; Murata, S.; Nomura, M.; Ishigaki, M.; Qu, M.; Tokuda,M. Energy Fuels 1997, 11 (3), 730-738.

(29) Smith, K. L. The structure and reaction processes of coal;Plenum Press: New York, 1994.

(30) Oh, M. S.; Peters, W. A.; Howard, J. B. AIChE J. 1989, 35 (5),775-792.

Figure 3. Particle size distributions (by volume) of DTF charsof density fractions of coals A and B prepared at 1573 K,measured using the Malvern Laser Sizer; (a) Coal A, (b) CoalB.


and SEM image analysis, clearly shows a decrease withincrease in parent particle density. The swelling ratiosmeasured using the two methods are generally consis-tent with each other. The mean swelling ratio of lightdensity-fraction chars is 25∼60% higher than that ofchar samples from the heavy density-fraction samples.Mean swelling ratios of coal A are slightly higher thanthat of the same density fraction of coal B. Interestingly,the experimental measurements show that the swellingratio for the F1.30 density fraction is the highest, thisbeing more apparent for coal A, as shown in Figure 7.The decrease in the swelling ratio of F1.25 fraction isattributed to possible char fragmentation during theplastic stage. This observation is consistent with previ-ous works by Gilfillan et al.8,18 using density-separatedsamples, for which maximum swelling occurred at F1.30density. Figure 8 also compares experimental resultswith model predictions using a char structural evolutionmodel developed by the authors and reported else-where.31 The results demonstrate that the model pre-dicts the experimental trend of swelling. As highswelling of particles results in high porosity, it wouldbe expected that the porosity of char samples from thelight density samples will be higher than that of heavydensity chars.

Char Structure and Porosity. Internal structures ofDTF chars prepared from density-fraction samples ofcoals A and B are shown in cross-sectional SEM images

(31) Yu, J. L.; Strezov, V.; Lucas, J.; Liu, G. S.; Wall, T. Proceedingsof the Combustion Institute, 2002, pp 467-473.

Figure 4. A comparison of PSD (by volume) of DTF chars of light and heavy density fractions of coals A and B; (a) F1.30 of coalA, (b) F1.50 of coal A, (c) F1.30 of coal B, (d) F1.50 of coal B.

Figure 5. PSD (by volume) of the DTF chars of coal A throughSEM image analysis using size bin I as for Malvern Sizermeasurements; (a) Cumulative, (b) Distribution.


presented in Figure 9 and Figure 10. The structuralcharacteristics of the chars are consistent with descrip-tions in the literature.6,32,33 However, a significantobservation from the present experiments is the pro-nounced changes that occur with increasing coal densityfor both coals. Light density-fraction samples, i.e., F1.25and F1.30, generate very porous particles, mostly groupI chars with a cenospheric structure, as shown inFigures 9a and 9b and in Figure 10a. The cenosphericstructure has the configuration of a single central void

surrounded by a thin outer shell (∼5 µm in thickness).The porosity of this shell is usually very low. Qualita-tively, it can be seen from the images (Figures 9a and9b) that the population of group I chars in the F1.30fraction is larger than that of the F1.25 fraction. Thisis consistent with the swelling ratio results. Twosomewhat conflicting reasons possibly explain the dif-ferences: (1) low density fractions develop high fluidity,therefore easily break up during the liquid stage owingto intensive generation of volatiles; (2) some of theparticles in coal A may behave like low rank coalparticles, for which cross-linking reactions occur at lowtemperature, which prevent the development of fluidityand swelling. Proposition (1) will lead to the formationof small cenospheric chars, marked as “SC” in Figures

(32) Wu, H. Ash formation during pulverised coal combustion andgasification at pressure. Ph.D. Thesis, Dept. of Chemical Engineering,University of Newcastle, Newcastle, Australia, xix, 156 leaves, 2000.

(33) Bailey, J. G. The origin of unburnt combustibles in coal. Ph.D.Thesis, Dept. of Geology, University of Newcastle, Newcastle, Austra-lia, 383 leaves, 1993.

Figure 6. PSD (by volume) of the DTF char samples of coals A and B through SEM image analysis using size bin II; (a) Cumulativeof coal A, (b) Distribution of coal A, (c) Cumulative of coal B, (d) Distribution of coal B.

Figure 7. Swelling ratios of DTF chars from density fractionsof coal A measured using different approaches as indicated.

Figure 8. Comparisons of model-predicted swelling ratios ofDTF chars from density fractons of coal B with experimentaldata measured through different methods.


9a and 9b and Figure 10a. This is confirmed from theobservations of char morphology discussed in the sub-sequent section. While proposition (2) leads to relativelysolid chars with tenui-network char structure (group IIor III) in light density fractions, marked as “TN” inFigure 9a.

Chars from high density-fraction coal samples, i.e.,F1.50 and S1.50, have a very solid structure with lowporosity, as shown in Figure 9c and Figures 10c and10d. Solid particles can be relatively pure carbonparticles, marked as “SG”, or carbon particles with someincluded ash grains, marked as “SM”. In the S1.50fraction, some excluded pure mineral grains can alsobe seen, marked as “M” in Figure 10d. The char samplefrom the medium density fraction, i.e., F1.35, containsa mixture of particles with different structures, asshown in Figure 10b, with porosity and wall thicknessin between.

Comparing the structure of coal A to that of coal B, itcan be seen that coal A has a noticeable population ofchars with a tenui-network structure, which is a con-figuration of many voids distributed inside the particle,in particular, for the medium density and the highdensity samples. The tenui-network structure has beendescribed in the literature.25 With the same densityfractions in coal B, however, there exists a significantamount of thick-walled cenospheric chars, which consist

of a large central void surrounded by a thick and porousouter shell, marked as “TC” in Figures 10b and 10c.These observations are attributed to the maceral com-positions in their parent coals. Coal A contains highliptinites, and coal B contains high inertinites. Inertinitecan contribute to the formation of thick-walled cenos-pheric chars.

The results of the porosity for both coals measuredthrough SEM image analysis are shown in Figure 11.A marked decrease in the porosity with increasingparent coal density is exhibited. For coal A, the averageporosity in the S1.25-F1.30 fraction is 86.8%, and 42.5%for the S1.35-F1.50 fraction. In coal B, the averageporosity decreases from above 80% in the light densityfractions, F1.25 and F1.30, to around 20% in the heavyfraction, S1.50. Consistent with the observations onswelling and char structure, the porosity shows amaximum value for F1.30 density fraction for both coals;this is more noticeable in coal A. The decrease in thechar porosity has also been predicted using the charstructural evolution model reported in the previouspaper.31

Figure 12 shows the char type distribution of densityfractions of coal A (Figure 12a), and coal B (Figure 12b).The figure clearly shows that group I chars are mainlygenerated in light density-fraction samples, i.e., F1.25and F1.30, while group III chars are yielded from heavy

Figure 9. SEM images of cross-sections of chars from density-fraction samples of coal A prepared in DTF at 1573 K; (a) F1.25,(b) S1.25-F1.30, (c) S1.35-F1.50; Isgroup I chars, IIsgroup II chars, IIIsgroup III chars; SGssolid carbon particle, SMssolidparticle with included mineral grains, SCssmall cenospheric char particle, TNstenui-network structure, TCsthick-walledcenospheric char.


density fractions, i.e., F1.50 and S1.50. The mediumdensity sample contains a mixture of different types ofchars. Benfell6 and Liu et al.7,34 have related the group

I char population to the vitrinite content of the parentcoal using empirical correlations. Table 3 compares thecalculated population of the Group I chars using theempirical correlations as indicated with experimentalresults from density-fraction samples. Large discrep-ancies exist for medium to heavy density coal fractionswith high ash levels. The present experiments indicatethat the correlation between the group I char populationand vitrinite (including liptinite) content from currentpetrographic analysis methods may not be straightfor-ward. This may be attributed to the following. (1) Thevitrinite content by current petrographic analysis is notgiven on coal particle number basis. It is thereforedifficult to quantify the proportion of pure vitrinite,vitrinite-dominated, or inertinite-dominated particles,etc. Also, any synergetic effects of maceral grains arenot clear. For instance, with an individual coal particle,how much vitrinite content will be required to result inthe formation of group I char? (2) It is obvious that themineral matter in coal is doing some work, in additionto the dominant role of coal macerals. It has beenreported in the open literature that liptinite and vit-rinite macerals develop high fluidity, while inertinitegenerally exhibits no fluidity during heating.1,35,36 How-ever, the included ash grains also significantly influencethe thermoplastic properties of the coal particles during

(34) Liu, G.-S.; Wu, H.; Benfell, K. E.; Lucas, J. A.; Wall, T. F. Proc.sAnnu. Int. Pittsburgh Coal Conf. 1999, 16, 817-826.

Figure 10. SEM images of cross-sections of char samples of coal B prepared in DTF at 1573 K; (a) F1.25, (b) S1.30-F1.35, (c)S1.35-F1.50, (d) S1.50; Isgroup I chars, IIsgroup II chars, IIIsgroup III chars; SGssolid carbon particle, Fsfoam structure,SMssolid particle with included mineral grains, SCssmall cenospheric char particle, SGssolid carbon particle, Msexcludedmineral grains, TCsthick-walled cenospheric char.

Figure 11. The porosity of DTF chars of the density-fractionsamples of both coals, measured through image analysis (charsample prepared at 1573 K).


pyrolysis, be it in a negative sense. At pyrolysis tem-perature, most of the ash grains remain solid, hencedecrease the apparent fluidity of the whole coal particle.Therefore, it is reasonable to conclude that ash has animpact on the physical structure of the residual char.This observation supports the conclusion of Gibbins etal.19 that both coal maceral and ash should be consid-ered to provide reasonable explanation to pf coal com-bustion behavior.

Char Morphology. The morphology of the chars wasexamined under the SEM for all the density samplesprepared in the DTF, and the results are presented inFigure 13 (coal A) and Figure 14 (coal B). There appearno major differences in the char morphology between

the two coals. However, significant changes in the charmorphology occur with increasing parent coal densityfor both coals. In general, from Figures 14a and 14b,chars from the two light density fractions have largerparticle size, and tend to have a round shape. They looksomewhat like “petrified balloons”, and their surfacesare usually smooth with some flow patterns. Someparticles have one large “blow-hole”, providing evidenceof the intensive release of the volatiles during pyrolysis.This observation agrees well with others in the litera-ture.8 Two possible reasons are proposed for the forma-tion of the surface-flow patterns: (1) they may be thewall between bubbles that have burst at the surface ofthe char particle; or (2) they may have resulted fromvolatile gases travelling along the particle surface, dueto the pressure difference within the surface layer. Fromthe cross-sectional SEM images discussed previously,the first proposition appears to be more essential. Thisinformation has also been confirmed by the SEM imagesof chars prepared at high pressures by Wu et al.32,37

Char morphologies from the high density coal samplesare remarkably different, as shown in Figures 13 and14d and 14e. Most of the particles have an irregularshape. No flow patterns were observed on the surfaceof these particles. These char particles usually havemany small open pores, some with cracks on theirsurfaces.

Table 4 lists typical char morphologies observed inthe present study, groups A to H, their characteristics,and occurrence. This list is somewhat similar to themorphology classification system by Littlejohn et al.38

However, Table 4 provides more detailed informationon the basis of the observation in the present experi-ments. The significant differences in the morphology ofchar particles are attributed to their thermoplasticproperties during pyrolysis. Particles from heavy densitysamples do not develop fluidity during heating, theygenerate less volatiles, and the volatiles are releasedthrough open pores or cracks at the particle surface.These particles form group C, D, or H chars. It can beexpected that little morphology changes will occur forthese particles during pyrolysis. Some particles maydevelop local fluidity that, although the whole particledoes not soften or swell, some parts may be liquefiedduring evolution. These particles are most likely togenerate group C or F chars. Particles that develop highfluidity during pyrolysis may have three paths: (i) Theparticle may break up to form several liquid dropsduring the plastic stage as a result of very high fluidityand intensive generation of volatiles. The smallerparticles then continue to pyrolyze and swell. Theseparticles will most likely generate group E or B chars;(35) van Krevelen, D. W. Coalstypology, physics, chemistry, consti-

tution, 3rd ed., completely revised; Elsevier: Amsterdam, New York,1993.

(36) Kidena, K.; Katsuyama, M.; Murata, S.; Nomura, M.; Chikada,T. Energy Fuels 2002, 16, 1231-1238.

(37) Wu, H. W.; Bryant, G.; Benfell, K.; Wall, T. Energy Fuels 2000,14, 282-290.

(38) Littlejohn, R. F. J. Inst. Fuel 1967, 40, 128.

Table 3. Comparisons of Estimated Group I Char Populations with Experimental Measurements from Density Fractionsof Coals A and B

coal A coal B

F1.25 F1.30 F1.35 F1.50 S1.50 F1.25 F1.30 F1.35 F1.50 S1.50

calculateda 86.0 87.2 87.7 84.9 81.9 83.9 85.2 70.9 57.0 58.7calculatedb 87.6 89.0 89.6 86.3 82.8 85.1 86.6 69.9 53.6 55.6experiments 70.8 80.8 - 2.6 - 75.0 74.2 8.3 0.0 2.7a Correlation by Liu et al.:7 nGrpI (%) ) 0.6 × P + 0.53 × vitr + 37. b Correlation by Benfell:6 nGrpI (%) ) 0.994 × P + 0.621 × vitr

+ 29.87.

Figure 12. The char type distribution of the density fractionsof coals A and B (prepared at 1573 K); (a) coal A, (b) coal B.


(ii) Some particles may swell to a large extent togenerate cenospheric chars, and the bubble may ruptureduring the plastic stage. However, the holes may notbe closed again because of the very intensive release ofvolatiles through the holes or as a result of high surfacetension. These particles generate group A char particleswith a large blow hole at the surface. (iii) If the particleswells but the bubble does not rupture or the blow holeis closed again after the bubble ruptures, then group Bchars would be generated. Coal particles that have amixture of two different maceral components generategroup F chars. Group G char particles evolve from coalparticles with liptinite macerals. These particles obvi-ously experienced high swelling and high conversionduring pyrolysis. Overall, light density samples generateGroup A, B, E, or G chars, and heavy density samplesyield group C, D, or H chars.

To allow comparisons of char morphology, two pa-rameters, i.e., elongation and roundness, are measuredthrough SEM image analysis, and the results are shownin Figure 15 (coal A) and Figure 16 (coal B). Theelongation is defined as the ratio of the major axislength to the minor axis length of a measured object,while the roundness describes how close the measuredobject is to a perfect rounded shape, of which theroundness is 1. The data demonstrate that more mor-phology changes occur in the light density-fraction charsamples for both coals; the gap between the raw coaland char curves is greatest at lower densities. As such,these particles have experienced more extensive physi-cal and chemical transformation as a result of moreintense volatile evolution. The high density-fractionparticles have experienced much less physical changesduring heating. The trends agree well with the above-

Figure 13. SEM morphology of char particles of coal A prepared in DTF at 1573 K; (a) F1.25, (b) S1.25-F1.30, (c) S1.30-F1.35,(d) S1.35-F1.50, (e) S1.50.


mentioned char structure and swelling results withrespect to parent coal density.

3.2. Swelling of Size-Fraction Samples. Particlesize distribution and swelling ratios of char samples ofsize fractions of both coals were measured using Mal-vern Laser Sizer, and the results are shown in Figure17 and Figure 18. The observed trend is that smallerparticle size fractions showed more swelling than largersize fractions, as shown in Figure 18. This observationis consistent with the results reported in the litera-ture.28,39 Smaller particles will be heated at a higherheating rate under the same external condition, com-pared to larger ones; also, small particles have smallertemperature gradients, therefore, smaller particles swell

more readily. As the particle size decreases, volatilesescape easily from the coal particle, which reduces thechances of secondary reactions, hence, the existence ofan optimum particle size range for maximum swellingis likely. This optimum swelling particle size may becoal type-dependent, and strongly affected by heatingconditions, such as heating rate and system pressure.It is logical that high heating rates will shift this particlesize range to smaller values.

The results of limited samples measured throughSEM image analysis on swelling of coal A confirms theresults from the Malvern Sizer measurements thatchars of smaller size fractions have larger swelling ratio,as shown in Figure 19a. On the other hand, themorphology of the chars appears not to be sensitive tothe raw coal particle sizes, as shown in Figure 19b.

(39) Hackert, G.; Wirtz, S.; Kremer, H. International Conference onCoal Combustion, Brisbane, 1999; pp 717-722.

Figure 14. SEM morphology of char particles of coal B prepared in DTF at 1573 K; (a) F1.25, (b) S1.25-F1.30, (c) S1.30-F1.35,(d) S1.35-F1.50, (e) S1.50.


Figure 15. Comparisons of morphological parameters of DTFchars of coal A with raw coal, analyzed through SEM imageanalysis; ) raw coal samples of density fractions, × DTF charsamples of density fractions after pyrolysis at 1573 K; (a)elongation, (b) roundness.

Figure 16. Comparisons of morphology of DTF chars of coalB with raw coal samples analyzed through SEM imageanalysis; ) raw coal samples of density fractions, × DTF charsamples of density fractions after pyrolysis at 1573 K; (a)elongation, (b) roundness.

Table 4. Typical Morphology of DTF Char Particles in SEM Images in the Present Study


There are no major differences in the morphology, asmeasured by roundness, of both chars and raw coalparticles among the three measured size fractions of coalA. However, there are noticeable differences in theroundness from the raw coal to char samples for all thesize fraction samples, as shown in Figure 19b. SEMimages also support these observations. Figure 20ashows SEM images of the morphology, and Figure 20bshows the cross-sections of DTF chars from the +75-90 size fraction of coal A. The observed char morphologyis described in Table 4, including particles with onelarge blow hole and flow patterns (BH), particles withclosed surface and flow patterns (CS), small roundparticles which are possibly small cenospheric chars(SC), and particles with irregular shapes thought to besolid (SD). Three types of chars are also apparentlydistributed in the cross-sectional image, i.e., group I,group II, and group III chars.

3.3. Effects of Pyrolysis Temperature on Swell-ing of DTF Chars. A limited number of char sampleshave also been prepared at 1373 K to investigate effectsof pyrolysis temperature on swelling. Table 5 comparesthe swelling results of char prepared at two differenttemperatures as indicated, measured using the MalvernSizer. For all the samples examined, the higher pyroly-sis temperature under the present experimental condi-tions resulted in higher swelling. This may be explainedby the following: when pyrolyzed at a higher temper-ature, (1) the coal particles have a slightly higherheating rate; and (2) devolatilization takes place athigher temperature, hence at a higher reaction rate,yielding more volatiles in a very short time. This causeshigher swelling of coal particles during pyrolysis.

Conclusions

Observations of DTF-prepared chars using density-separated coal samples demonstrate heterogeneousbehavior, including weight loss, swelling, and charchemistry, of coal particles in pf size from the same coalunder the same heating conditions. Influences of ashon char formation, in addition to the dominant role ofmacerals, are also identified.

For the both coals, weight losses and swelling ratiosof DTF chars decrease with an increase of the coaldensity. Correspondingly, the char porosity of thedensity fractions decreases significantly from above 80%to ∼20% as the coal density increases. Internal char

Figure 17. PSD of DTF chars of size fractions of coals A andB prepared at 1573 K, measured using the Malvern LaserSizer; (a) coal A, (b) coal B.

Figure 18. Swelling ratios of DTF chars of size fractions ofcoals A and B.

Figure 19. Swelling ratios and roundness of DTF chars ofsize fractions of coal A measured through SEM image analysis;(a) swelling ratio, (b) roundness.


structures change drastically with an increase of thecoal density. Light density-fraction samples, i.e., F1.25and F1.30, generate group I chars with high porosityand thin walls. On the contrary, heavy density-fractionsamples, i.e., F1.50 and S1.50, yield group III chars,with solid char structure and very low porosity. The charsample from the medium density fraction contains amixture of different types of char particles with amoderate porosity and wall thickness. Morphology ofchars among different density fractions is also signifi-cantly different. Char particles from light density frac-tions appear as “petrified balloons” and have a relativelyround shape and a smooth surface with flow patterns,some with “blow-holes”, while char particles from heavydensity fractions have an irregular shape with manysurface pores and often highly cracked.

It is obvious that light density char particles haveexperienced intensive softening and swelling due to thedevelopment of fluidity and intensive volatile genera-tion. With these particles, their volatile release duringpyrolysis is largely governed by the behavior of bubbles

within the particle. However, the particles of heavydensity-fraction samples are most likely to have main-tained their physical structures during the course ofdevolatilization due to the absence of fluidity, owing tothe high inertinite content and high ash level. Therefore,these particles do not exhibit softening and swelling,and their volatiles are released through the pore open-ings or cracks via the hydrodynamic process.

Observations of various size fraction samples revealedthat particles with smaller size tend to have higherswelling under the same heating condition. In addition,it is believed that particles in smaller size fractions havea higher likelihood of a mono-maceral behavior duringheating while larger particles may consist of more thanone type of macerals. Char samples prepared at lowerpyrolysis temperature in the DTF have a smallerswelling ratio in the temperature range under thepresent experimental conditions.

Acknowledgment. The authors acknowledge thesupport provided by the Cooperative Research Centrefor Coal in Sustainable Development, which is fundedin part by the CRC program of Australia. The authorsalso thank Dr. Gary Bryant and Dr. Liza Elliott forpreparing the density-fraction coal samples, Dr. GregoryBrowning helped with the DTF experiments and Mr.Steve Richardson helped with safety issues during thecourse of the experiments.

EF030022H

Figure 20. DTF char SEM morphology and cross-section images of +75-90 size fraction of coal A; (a) char morphology, (b)cross-section; Isgroup I chars, IIsgroup II chars, IIIsgroup III chars; SDssolid particle, Cscenospheric char, BHschar with ablow hole, TNsTenui-network structure, SCssmall cenospheric char particle, Fsfoam structure.

Table 5. A Comparison of Swellings of DTF CharsPrepared at Different Pyrolysis Temperatures as

Indicated, Measured Using Malvern Sizer

coal A coal B

swelling ratio 1573 K 1373 K 1573 K 1373 K

F1.30 1.55 1.31 1.32 1.10F1.35 1.28 1.10F1.50 1.35 1.10


Documents

Swelling and Char Structures from Density Fractions of Pulverized Coal