Comparison of chars obtained under oxy-fuel and conventional pulverized coal combustion atmospheres

Comparison of Chars Obtained under Oxy-Fuel and ConventionalPulverized Coal Combustion Atmospheres

Angeles G. Borrego* and Diego Alvarez

Instituto Nacional del Carbón, CSIC, P.O. Box 73, 33080 OViedo, Spain

ReceiVed June 22, 2007. ReVised Manuscript ReceiVed August 15, 2007

The combustion of coal in conventional power plants produces large amounts of CO2 which contributes tothe greenhouse effect. One of the ways to approach the CO2 emissions abatement is to burn the coal in anO2/CO2 atmosphere which eliminates the need of a separation step. In this study, two coals of different rank(a high volatile and a low volatile bituminous) have been burned in a drop tube reactor using O2/N2 andO2/CO2 mixtures with increasing oxygen content from 0 to 21%. Various oxygen concentrations have beenselected for each set of experiments in order to follow both the progress of combustion and the influence ofoxygen content in the devolatilization behavior of coal. Results show that a higher amount of O2 in CO2 thanin N2 is needed to achieve similar burnout levels. Significant differences were found in the influence of oxygencontent on the devolatilization behavior of the lower and higher rank coal. The limited amount of oxygen inthe reacting atmosphere resulted in volatile release inhibition for the high volatile bituminous coal, whereasthe more plastic low volatile coal was hardly affected. The presence of variable amounts of oxygen in CO2

had a small influence on the char particle appearance. The chars from both the combustion series (O2/N2) andthe oxy-fuel series (O2/CO2) were similar for each parent coal in terms of reactivity and micropore surfacearea measured by CO2 adsorption. The main difference between both series of chars relied on the surface areadetermined by N2 adsorption (SBET) and on the size distribution of pores which was shifted to a larger size forthe oxy-fuel series. The difference between both series of chars was larger for the high volatile bituminouscoal chars than for the low volatile bituminous coal chars. This might have important implications for combustionunder the diffusion-controlled regime.

Introduction

The combustion of fossil fuels for energy production resultsin the generation of greenhouse gases, with CO2 as the majorcontributor, which are emitted to the atmosphere. There is ageneral agreement on the need to reduce the emissions of CO2,although the degree of compromise of the different governmentsand the ways to approach the problems are rather different.1

The coal fired power plants are among the best candidates toinstall systems for CO2 capture because they are stationarysources emitting large amounts of CO2. There are variouspossible routes to concentrate the CO2 for further storage whicheither act after the combustion process (i.e., amine scrubbing,calcination–recarbonation cycles) or prior to combustion throughthe utilization of a decarbonized fuel. The combustion of coalin a nitrogen-free atmosphere, most known as oxy-fuel technol-ogy, is one of the ways to approach the problem in which theneed of a CO2 separation step can be eliminated.2 In thistechnology, coal combustion occurs in an oxygen atmospherediluted with recycled CO2 to reduce the boiler temperature andto ensure the volume of gas. As a consequence, the flue gaseswill consist mainly of CO2 plus H2O which can be easilyseparated by condensation. The state of the art of this technologyincluding research findings and testing scale results has been

recently reviewed.3 The replacement of N2 by CO2 in thereacting atmosphere may have a number of consequences whichaffect different aspects of coal combustion; on the one hand,the flame propagation speed,4 flame stability, and flame tem-peratures are lower in the O2/CO2 environment than in O2/N2.5,6

The larger specific heat of CO2 compared to N2 seems to beresponsible for the high amounts of oxygen diluted in carbondioxidesbetween 30 and 35% depending on gas impurities5–7srequired to match the air combustion temperatures. In addition,although the CO2-char reaction rate would be much slower thanthe O2-char reaction rate, it could also contribute to the overallefficiency of the process, particularly at the high temperaturesand CO2 concentrations prevailing on the boiler.8 Despite theexistence of works comparing aspects such as heat transfer,flame temperature, and reaction rate under conventional and oxy-fuel technology conditions,3 there are few works focusing ondifferences in char structure, specific surface area, chemicalreactivity, and swelling behavior which are needed formodeling.8–10 Two papers addressing these topics were presented

* Corresponding author. E.mail: [email protected]. Fax: +34 985297662. Phone: +34 985119090.

(1) Intergovernmental Panel on Climate Change. Carbon Dioxide Captureand Storage. http://www.ipcc.ch/.

(2) Anheden, M.; Yan, J.; Smedt, G. de ReV. IFP 2005, 60, 485–495.

(3) Buhre, B. J. P.; Elliot, L. K.; Sheng, C. D.; Gupta, R. P.; Wall, T. F.Prog. Energy Combust. Sci. 2005, 31, 283–307.

(4) Kiga, T.; Takano, S.; Kimura, N.; Omata, K.; Okawa, M.; Mori, T.Energy ConVers. Manage. 1997, 38, 129–34.

(5) Liu, H.; Zailani, R.; Gibbs, B. M. Fuel 2005, 84, 833–840.(6) Croiset, E.; Thambimuthu, K. V.; Palmer, A. Can. J. Chem. Eng.

2000, 78, 402–407.(7) Chui, E. H.; Douglas, M. A.; Tan, Y. Fuel 2003, 82, 1201–1210.(8) Shaddix, C. R.; Murphy, J. J. 20th Pittsburg Coal conference,

Pittsburgh, PA, 2003.(9) Elliot, L. K.; Liu, Y.; Buhre, B. J. P.; Martin, J.; Gupta, R. P.; Wall,

T. F. Proceedings of the ICCS&T 2005, Okinawa, Japan, CD-12pp.(10) Alvarez, D.; Fernández-Domínguez, I.; Borrego, A. G. Proceedings

of the ICCS&T 2005, Okinawa, Japan, CD-6pp.

Energy & Fuels 2007, 21, 3171–3179 3171

10.1021/ef700353n CCC: $37.00 2007 American Chemical SocietyPublished on Web 10/06/2007

at the last International Conference on Coal Science andTechnology held in Okinawa. The one by Alvarez et al.10 wasthe basis of this extended version, which also contains additionalinformation on petrography11 and textural characterization ofchars. These results will be compared with those of Elliot etal.9 who followed an approach similar in many aspects to theone reported in this work.

Experimental Section

1. Coal Samples. Two coals of different rank, ground and sievedto 36–75 µm, have been selected for this study. Ultimate analysesof coals were performed using a LECO CHN600 instrument forcarbon, nitrogen, and hydrogen, a LECO SC132 instrument forsulfur, and a LECO VTF900 instrument for oxygen. Proximateanalyses were carried out following the standard procedures desc-ribed in UNE 32-019-84 for volatile matter and ISO-1171/1981for ash contents. Standard petrographic (maceral ISO 7404-3; 1994and random reflectance ISO 7404-5; 1994) analyses were carriedout on the coal fractions.

2. Char Preparation. Coal chars were prepared in a drop tubereactor at 1300 °C under nine different O2/N2 ratios (ranging from0 to 21% oxygen), and also under four different O2/CO2 atmo-spheres (0–21% oxygen). The reactor, whose scheme is shown inFigure 1, is a furnace which surrounds two concentric alumina tubes(70 and 50 mm inner diameter, 1.30 and 1 m long, respectively).The reacting gas (600 L h-1) was injected at the bottom of theouter cylinder and was preheated while flowing upwards. When atthe top of the outer cylinder, the gas was forced onto the innertube through a flow straightener, and the gases flowed downwardsand left the reactor through a water-cooled collection probe. Thefuel particles were entrained (1 g min-1) by a jet of nonpreheatedgas (300 L h-1) to a water-cooled injection probe placed on top ofthe inner tube. The estimated residence time of the particles in thereactor was 0.3 s. The chars left the reactor through the collectionprobe, and an extra nitrogen flow was added to the exhaust gasesin order to quench the reaction and improve the collection efficiencyin the cyclone.

Coal burnout was calculated by the ash tracer, which is a massbalance between the ashes entering and leaving the reactor. Theconversion is then calculated as

conversion (%)) [1- ( ashcoal

100- ashcoal)(100- ashchar-comb

ashchar-comb)] × 100

The use of the ash tracer implies a number of assumptions includingthat mineral matter transformations occurring in the drop tubereactor (DTR) and during the proximate ash determination aresimilar and that the cyclone is able to recover 100% residualmaterial both organic and inorganic.12 Both are unrealistic assump-tions because typically ashes from proximate tests have a powderlikeappearance indicating limited melting, whereas extensive meltingoccurs in the reactors and the boilers, as shown by the abundanceof aluminosilicate spheres. Just a difference of 200 °C in theoperating conditions of the DTR (1000 vs 1300 °C gas tempera-tures) has shown to have different effects on the melting behaviorof the mineral matter.13 The opposite effect might be observed forminerals such as carbonates, which are completely decomposedduring the proximate test but are not able to fully decompose inthe short residence times of the reactor. On the other hand, liberationof fine ashes is a complex process which does not only depend onthe characteristics of the parent coal and its mineral matter but alsodepends on the operating conditions and oxygen concentration.14

Without disregarding the limitations of the ash tracer for burnoutestimation, its extended use to express conversion in combustionexperiments and the good repeatability of the reactor (differencesin char ash between runs around 0.6% for coals with 15% ashcontent have been recorded) justifies its utilization also in this work.Moreover, it is mainly used for comparative purposes betweenexperiments performed at a single temperature varying only thereaction atmosphere, and therefore, the transformations of themineral matter in the reactor for each of the parent coals areexpected to be similar.

3. Char Characterization. Two widely used methods todetermine the pore surface area of carbon from gas adsorptionisotherms were applied in this study to selected char samples, usingCO2 at 0 °C and N2 at -196 °C as adsorptives. The equipmentused was a Micromeritics ASAP 2020 instrument. Chars wereoutgassed under vacuum prior to gas adsorption experiments inorder to eliminate moisture or condensed volatiles, which couldprevent the adsorbate accessibility. The heating rate used was 5°C min-1 with holding temperatures of 90 °C (1 h) and 350 °C (4h). This temperature is well below the char preparation temperatureand is not expected to modify the structure of the char. CO2

adsorption isotherms were performed at 0 °C up to a pressure of0.035 Torr, and the Dubinin–Radushkevich (D–R) equation15 wasapplied to the adsorption data. The Brunauer–Emmett–Teller (BET)theory was applied to the N2 adsorption data to obtain the surfacearea.16 These two methods can be regarded as complementary, giventhe difficulties of CO2 to fill large micropores and the slow diffusionof N2 in the small micropores.17 As some of the samples containedlarge amounts of mineral matter, the surface area data are expressedon an ash-free basis considering a surface area for the ashes of 0.8m2 g-1. The isotherms were analyzed using the Micromeriticsdensity functional theory (DFT) software package DFT plus. Thepore size distribution was obtained in the size range 0.4–1.0 nmfor CO2 adsorption and in the range 0.4–250 nm for N2 adsorption.Considering the DFT distributions, the pore volumes in themicropore and mesopore range were further split into micropores(<1.4 nm under both CO2 and N2) and supermicropores (1.4–2.0nm). Also, two groups were calculated for the mesopores (2–5 and5–50 nm).

(11) Alvarez, D.; Fernández-Domínguez, I.; Borrego, A. G. 57th AnnualMeeting of the ICCP, Patras, Greece, 2005; p 36.

(12) Carpenter, A.; Skorupska, N. M. Coal combustion—analysis andtesting; IEA CR/64: London, 1993; p 97.

(13) Alonso, M. J. G.; Borrego, A. G.; Alvarez, D.; Menéndez, R. Fuel1999, 78, 1501–1513.

(14) Buhre, B. J. P.; Hinkley, J. T.; Gupta, R. P.; Nelson, P. F.; Wall,T. F. Fuel 2006, 85, 185–193.

(15) Dubinin, M. M.; Radushkevich, L. V. Proc. Acad. Sci. USSR 1947,55, 331–335.

(16) Brunauer, S.; Emmett, P.; Teller, E. J. Am. Chem Soc. 1938, 60,309–315.

(17) Jagiello, J.; Thommes, M. Carbon 2004, 42, 1227–1232.

Figure 1. Scheme of the drop tube reactor used for the preparation ofthe chars.

3172 Energy & Fuels, Vol. 21, No. 6, 2007 Borrego et al.

Char combustion was isothermally recorded using a Perkin ElmerTGA7 thermal analysis system. A small quantity of char (13 mg)was homogeneously spread at the bottom of the platinum crucibleand then heated up to 550 °C under a nitrogen flow (35 cm3 min-1)at a heating rate of 25 °C min-1. After weight stabilization, nitrogenwas replaced by air at the same flow rate and the temperature wasmaintained until combustion was completed. Under these operationconditions, diffusional constraints have shown to start at temper-atures ranging from 570 to 680 °C for high and low bituminouscoals, respectively,18 and therefore, the kinetic control of the reactionis ensured. The reactivity was calculated as

R) 1m0

· dmdt

where m0 is the initial ash-free mass of coal.The chars were mixed with an epoxy resin to prepare a pellet

for petrographic analysis. Samples were carefully polished to avoiddamage to the fragile walls of the chars and then analyzed using areflected light microscope, oil immersion objectives, and 1λ retarderplate. For char reflectance analysis, 100 random reflectance readingswere taken on the char walls. Additionally, scanning electronmicroscopy (SEM) studies were carried out on the chars to visualizethe topographical aspect of the particles and to give insight intothe combustion mode of the particles.

Results and Discussion

The coals were selected for this study on the basis of theirrank difference. Coal COS is a Colombian vitrinite-rich highvolatile bituminous coal with moderate inertinite and lowliptinite contents, and SMK is a Western Canadian low volatilebituminous coal with relatively high inertinite content (Table1). Both coals are low in sulphur, and the ash content of COSis very low. Overall, the chemical parameters follow theexpected trends for the differences in rank between the two coals(Table 1), with low rank coal (COS) having lower vitrinitereflectance and carbon content and higher volatile matter,hydrogen and oxygen contents than the high rank coal (SMK).The reflectance of inertinite also increases with rank and had alarge interval of variation in COS and a symmetric distributionin SMK centered around 2%.

1. Coal Conversion. The coal burnouts calculated by theash tracer are plotted in Figure 2 against the amount of oxygenin the reacting gas. Significant differences were observed in thebehavior of the two coals for both the combustion (O2/N2) andthe oxy-fuel series (O2/CO2). As expected, the conversions ofboth series were higher for the lower rank coal than for thehigher rank coal at any of the conditions tested, but the observedtrends deserve some further attention. The conversion ofthe higher rank coal continuously increased from 29 to 88%with increasing oxygen content, approaching well to a polyno-mial three-order expression. The conversion of SMK under aninert atmosphere (N2) was about 10 points higher than the

volatile matter content of the coal determined by proximateanalyses, indicating an enhanced devolatilization at the hightemperatures and heating rates occurring in the reactor. Theenhanced volatile release with increasing heating rate andtemperature is a known effect,19,20 and it would not requirefurther discussion if a similar trend would have been obtainedfor the low rank coal. In the case of COS, the volatile yield inthe reactor doubled the proximate volatiles and the differencesin conversion at variable oxygen concentrations were rathersmall (79–98%) considering the large range of oxygen contents(0–21%). As the errors associated with the ash tracer (brieflyoutlined in the Experimental Section) can introduce largedifferences in conversion, particularly for such low ash coal asCOS, a significant number of runs was repeated to confirm theresults. The burnouts plotted in Figure 2 were calculatedaveraging the char ashes for the runs available. The largeststandard deviation recorded for the conversions averaged was1.7% and corresponded to the runs performed under 7% O2 inN2. The plot of Figure 2 shows that the lowest conversion wasnot observed for the char obtained under an inert atmosphere(0% O2) but for the run performed under 5% O2 in N2. Thisindicates that the oxygen is not exclusively involved in thecombustion of volatiles but it had also some effect on thedevolatilizing particles.21 A similar effect has been observedfor other bituminous coals22 and would indicate that limitedamounts of oxygen inhibited to a certain extent the volatilerelease. Oxygen could be involved in cross-linking at the surfaceof the resolidifying char, reducing the swelling of the particles.However, with oxygen levels high enough, both homogeneousand heterogeneous reactions will take place extensively, andthe weight losses on combustion will be higher than theinhibiting effect of oxygen on pyrolysis.

The conversions of the oxy-fuel chars (O2/CO2) were alwayslower than those of the combustion chars (O2/N2) for equivalentoxygen concentrations (Figure 2), although the differencesdecreased at high oxygen concentrations. The conversion ofthe COS oxygen-free oxy-char was intermediate between theproximate volatile yield (43%) and the conversion of the

(18) Alonso, M. J. G.; Borrego, A. G.; Alvarez, D.; Menéndez, R. FuelProcess. Technol. 2001, 69, 257–272.

(19) Jamaluddin, A. S.; Truelove, J. S.; Wall, T. F. Combust. Flame1986, 63, 329–337.

(20) Solomon, P. R.; Fletcher, T. H.; Pugmire, R. J. Progress in coalpyrolysis. Fuel 1993, 72, 587–597.

(21) Senneca, O.; Salatino, P.; Masi, S. Proc. Combust. Inst. 2005, 30,2223–2230.

(22) Milenkova, K. S.; Borrego, A. G.; Alvarez, D.; Xiberta, J.;Menéndez, R. Proceedings of the 12th ICCS 2003, Cairns, Australia, CD-10pp.

Table 1. Proximate, Ultimate, and Petrographic Analysis of theSize Fractions of the Coals

daf % vol mmf %

coal

ashdb% VM C H N O S

Rvr

(%)Rir

(%) V L I

COS 1.1 40.7 80.3 5.8 1.5 11.8 0.5 0.60 1.26 84.4 2.4 13.2SMK 16.2 17.5 91.3 6.0 0.9 2.0 0.3 1.53 2.08 61.6 0.0 38.4

a MV ) volatile matter; Rr ) random reflectance; V ) vitrinite; L )liptinite; I ) inertinite; db ) dry basis; daf ) dry-ash-free basis; mmf )mineral-matter-free basis; vol ) volume.

Figure 2. Conversion of the high volatile (COS) and low volatilebituminous coal char (SMK) under various reacting atmospheres.

Comparison of Chars Energy & Fuels, Vol. 21, No. 6, 2007 3173

pyrolysis char (88%). Conversely, the weight loss of SMK oxy-char (11.6%) was lower than the proximate volatile mattercontent of the coal (17.5%). This suggests that CO2 could beinvolved in cross-linking at the char surface, having a similareffect on the resolidifying char to that previously described fora limited amount of oxygen. A different result was reported byElliot et al.9 who attributed to C–CO2 reaction the higher masslosses found under CO2 compared to N2 at high temperatures.

2. Appearance of the Chars. Particles obtained from thelow rank coal under nitrogen were mainly isotropic, ratherrounded in shape, and showed a limited amount of secondarydevolatilization voids within the walls (Figure 3). A largeramount of secondary vesicles was typically found in the particlesobtained under a limited amount of O2 (2.5 and 5%) in thecombustion series (O2/N2), whereas chars prepared with 15%and higher oxygen contents had very thin char walls and porositycrossed the walls from side to side. This is also observed in thescanning electron micrographs of Figure 4 in which the lowconversion char showed rather intact walls with a granulatedappearance and with pores that did not fully penetrate the wallsand the extensively burned char showed large coalesced voids.The combustion pattern appears to indicate a diffusion-controlledregime of combustion. In the oxy-fuel series (O2/CO2), somelow reflecting chars were found in the CO2 char, which wereno longer present in the samples prepared with a certain amountof O2 (Figure 3). In addition, more vesiculated particles werefound in the oxy-fuel series, and network structures with voidsdistributed throughout the particle surface were more commonthan in the combustion series, indicating a lower capacity of

the bubbles to coalesce.23 In particular, the lower reflectanceparticles found in the CO2 char typically had a network structure.The secondary porosity observed in the chars obtained under alimited amount of oxygen together with the thickening of thewalls will be probably a consequence of further devolatilizationof a material not yet consolidated. This could be due to self-heating of the particles promoted by oxygen. Previous studieshave shown that char reflectance is sensitive to the preparationtemperature;13 therefore, this possibility was checked throughthe measurement of reflectance of char walls, as it is knownthat increased O2 content produces higher peak temperaturein the particles.24 Figure 4 shows the reflectance profiles forvarious COS chars for the combustion and oxy-combustionseries. A significant increase in reflectance is observed after thepassage through the reactor from 0.6% in the parent coal tohigher than 6% in the chars due to the carbon enrichment andthe volatile release. In addition, systematically higher reflec-tances were recorded for the oxy-fuel chars obtained undersimilar oxygen concentrations. The reflectance values of thecombustion series varied within a narrow interval for thedifferent conditions (6.19–6.45%), whereas a larger variationwas observed between the reflectance of the CO2 char (7.45%)and those with higher oxygen content in the oxy-fuel series(∼6.8%).

Coal SMK had a significant amount of vitrinite and inertiniteand therefore many char particles contained material havingexperienced different behavior. Vitrinite in this coal passes uponheating through a plastic stage in which aromatic clusters havethe chance to grow and coalesce to form well ordered anisotropicdomains (Figure 6). The shape of pure vitrinite-derived charsis cenospheric and tended to multichambered particles as theoxygen content in the reacting atmosphere increased. The CO2

char generated also multichambered particles approachingnetworks. Inertinite exhibited a variety of optical textures: asignificant part remained isotropic with no or very little alteration(small degassing bubbles); other inertinites, typically the leastreflecting ones, fused and developed anisotropy. Even foranisotropic inertinites, the lower plasticity of this materialcompared to vitrinite is reflected in the network vs cenosphericshape and in the smaller size of the anisotropic domains whichindicates less mobility of the aromatic clusters.25 As seen inthe micrographs of Figure 6, the size of the anisotropic domainsin the walls of SMK char is smaller than that of the measuringfield of the microscope even using high magnification. Thisimplies that a measure in this region will be an average ofdomains with different orientation with large scatter and lower

(23) Yu, J.; Lucas, J. A.; Wall, T. F. Prog. Energy Combust. Sci. 2007,33, 135–170.

(24) Murphy, J. J.; Shaddix, C. R. Combust. Flame 2006, 144, 710–729.

(25) Yu, J.; Lucas, J.; Wall, T.; Liu, G.; Sheng, C. Combust. Flame2004, 136, 519–532.

Figure 3. Optical micrographs showing the appearance of high volatilebituminous COS chars generated under different gas atmospheres (left,O2/N2 series; right, O2/CO2 series). Reflected light, oil immersion, 1λretarder plate. The most remarkable difference is the more vesiculatedCO2 char compared to the pyrolysis counterpart.

Figure 4. Scannning electron microscopy (SEM) images of slightlyburned (2.5% O2 in N2) and extensively burned (air) COS chars showingthe fine grained texture of the walls and the progress of combustionthrough pore enlargement and coalescence.


reliability. Therefore, in this case, measurements were takenon isotropic inertinite. The histograms showing the distribution

of inertinite reflectance in SMK chars are shown in Figure 7.As in the case of COS, the reflectances were higher for the oxy-fuel inertinites than for those in the combustion series. In thiscase, a clear drop in reflectance is observed with increasingoxygen content in the reacting gas. The average reflectance ofinertinite in the parent coal was 2.08% (Table 1), and it raisedto 6.14 and 6.87% for the nitrogen and CO2 chars, respectively.The reflectance of inertinite was lower than that reached by lowrank vitrinite, confirming its lower ability to be transformed evenat the drastic temperatures occurring in the reactor.

The combustion in SMK vitrinite proceeded through the limitsof the domains, as shown by the deep engrave observed in themore extensively burned char (Figure 8). The combustion markscoinciding with the limits of the domains are well observed inthe image from the combustion char prepared with 10% O2

content and burned at 61% conversion.3. The Size of Char Porosity. In a previous study, we

showed that the presence of a variable amount of oxygen inthe combustion atmosphere hardly affected the microporesurface area (SCO2), which is mainly due to the structural porosityleft by the reorganization of the carbonaceous material duringresolidification,26 but strongly affected the mesopore surface area(SBET).27 This was so for coals generating disordered isotropicmaterial typically resulting in large micropore surface areas andfor coals passing through a plastic stage and generating charswith low microporosity. The plot in Figure 9 indicates a similarbehavior for the coals of this study, in which the decrease inSCO2 which occurs at relatively high burnouts for COS andcontinuously for SMK can be assigned to micropore wideningdue to combustion. Figure 9 also shows rather similar SCO2 forthe oxy-fuel and the combustion series, indicating that thedevelopment of microporosity determined by CO2 adsorptionis more related to the nature of the parent coal and otheroperating conditions such as temperature or heating rate thanto the “inert” gas accompanying the oxygen. A very differentsituation is observed in the case of surface areas determinedfrom N2 adsorption isotherms. In this case, the oxy-fuel seriesshowed systematically higher SBET values than the combustion

(26) Feng, B.; Bhatia, S. K. Carbon 2003, 41, 507–523.(27) Alvarez, D.; Borrego, A. G. Energy Fuels 2007, 21, 1085–1091.

Figure 5. Reflectance histograms of vitrinite-derived material from COS for a set of chars prepared under similar oxygen concentrations. Solidsquares, oxy-fuel series; void squares, combustion series.

Figure 6. Optical micrographs showing the appearance of low volatilebituminous SMK chars generated under different gas atmospheres (left,O2/N2 series; right, O2/CO2 series). Reflected light, oil immersion, 1λretarder plate.


series, as also reported by Elliot et al.,9 and the differences weregreater for the high volatile bituminous coal (Figure 9). It iswell known that N2 has difficulties to diffuse in the microporesof chars, often resulting in less volume of micropores determinedby N2 adsorption than by CO2 adsorption.26 This has also beenobserved in the chars of this study in which pore volumesdetermined by CO2 adsorption were always higher than themicropore volumes (<2 nm) determined by N2 adsorption(Figure 10).

The pore distribution of the DFT model has been used toquantify the pore volume in various size intervals (Figure 10).The comparison between the pore volume measured by CO2

adsorption and the various pore intervals measured by N2

adsorption also provides some insight into the structure of the

chars. The pyrolysis char of COS had a microporosity networkonly accessible to CO2 in reasonably long equilibration times.Combustion in O2/N2 atmosphere (5% O2) resulted in amicroporosity more easily accessible to N2, and further increaseof oxygen (10% O2) enlarged the size and amount of pores andapproached the volume of micropores measured by both CO2

and N2 adsorption. The N2 (77 K) isotherm of the oxygen-freeCOS oxy-char indicated a higher volume of micropores thanthat of the pyrolysis counterpart and also a significant mesoporevolume. The increase of oxygen in CO2 (from 0 to 5% O2)enlarged the volume of both micro- and mesopores, but furtherincrease in O2 (10%) mainly enlarged the mesopore volume.

Overall, the pore volumes of SMK chars were lower thanthose of COS chars regardless of the preparation atmosphere.

Figure 7. Reflectance histograms of inertinite-derived material from SMK for the set of chars prepared under similar oxygen concentrations. Solidromb, oxy-fuel series; void romb, combustion series.

Figure 8. Scannning electron microscopy (SEM) images of the pyrolysis char (N2) and chars obtained under 10% O2 combustion (left) and oxy-combustion atmospheres showing the carved limits of the anisotropic domains through which combustion preferentially proceeded.


As in the case of COS, similar micropore volume was obtainedby CO2 adsorption for both O2/N2 and O2/CO2 series in SMK.Despite the pore volumes derived from N2 adsorption of SMK,those of the oxy-fuel series were overall larger than those ofthe combustion series; the differences were smaller than thosein the COS chars. The most remarkable difference between thepore size intervals of SMK chars determined by N2 isothermswas the absence of micropores in the oxygen-free pyrolysis charand the presence of supermicropores and small size mesoporesin the oxygen-free oxy-char. Both the size and amount of poreswere similar in the 5% O2 members of SMK chars, in whichthe major pore volume corresponded to micropores with sizesunder 1.4 nm. This size range was no longer present in the moreextensively burned 10% O2 chars. This indicates that a limitedamount of oxygen cleaned or opened the mouth of the smallmicropores but further oxygen concentration only enlarged thepores without making other micropores accessible. A largedifference was observed in the volume of supermicroporesbetween the two 10% O2 chars.

The evolution of pore size with burnout indicates that forthe high volatile bituminous coal both opening of microporemouths making accessible otherwise closed porosity and mi-cropore widening operated, whereas for the low volatile coal acertain amount of oxygen is needed to make more easilyaccessible the micropores but further combustion essentiallyenlarged the existing porosity. The larger amount of mesoporesin the oxy-fuel chars might be relevant to facilitate combustionin a diffusion-controlled regime.

4. Char Reactivity. Char reactivities were measured underkinetic control in a thermobalance. The evolution of reactivitywith conversion typically reached a maximum (Rmax) at about10% conversion for COS chars and 25% conversion for SMKchars, which is the value used to compare the reactivity of the

various chars (Figure 11). The reactivities of the high volatilecoal chars were always higher than those of the low volatilecoal chars and reached a maximum for the 15% O2 COS char.The reactivity of SMK chars slightly decreased with increasingO2 concentration in the reacting gas. For both coals, thereactivities of the oxy-fuel and conventional combustion charsobtained under equivalent oxygen concentrations were similar,indicating that the resolidified char generated during thecombustion process had a rather similar structure. This is alsoconfirmed by the similar appearance of the chars observed byoptical microscopy and the similar specific surface area (SCO2).

As reactivity is directly related to the surface area in whichoxygen attack can occur, the effect of the surface area in thereactivity can be discounted to calculate the intrinsic reactivity.The reactivities were corrected using the values of SCO2, as it isexpected that at such a low temperature reaction is slow enoughto allow oxygen to reach the micropores. In Figure 12, theintrinsic reactivity is plotted vs burnout. No significant variationis observed in intrinsic reactivity through the whole burnoutinterval in the case of SMK chars, and the value which deviatedmore from the trend was that of the 21% O2 oxy-char. The dropin SCO2 in this char due to pore coalescence relatively increasedthe intrinsic reactivity. A different situation was observed forCOS chars. In this case, a decrease in reactivity with increasingburnout is observed for the oxy-fuel series up to around80% burnout and then reactivity increased again for extensivelyburned chars. The whole combustion series fits in the highburnout part of this trend and only an increase in reactivity withincreasing burnout is observed. This is a different result to thatgenerally reported (i.e., Davis et al.28) in which extensivelyburned chars are affected by thermal annealing which decreasesthe reactivity. It must be kept in mind that the experimentalconditions used here for the extensively burned char onlyindicated overstoichiometric combustion conditions and nolonger residence times that would have favored the ordering ofthe carbonaceous structure and also the decrease in reactivityof the material.

The evolution of char reflectance with burnout (Figure 13)did not support increased order in char structure duringcombustion which would have resulted in a rise of reflectance.On the contrary, a decrease in reflectance was observed for bothchars with burnout. The trend of the oxy-fuel and combustionseries matched well for both chars, and only the most extensivelyburned oxy-char showed a higher reflectance than that expectedfrom the trend. Despite the fact that reflectance was measuredon different materials (vitrinite in COS chars and inertinite inSMK chars), the trends paralleled each other, indicating a rathersimilar evolution. The difference in reflectances was smallerfor COS chars than for SMK chars. This is in agreement withthe greater ability of COS vitrinite to transform in the reactoras derived from its theoretic volatile matter content (43%)compared to the theoretic volatile matter content of SMKinertinite (17%).29

Conclusions

The comparative study of chars obtained under typicalcombustion (O2/N2) and oxy-combustion (O2/CO2) conditionswith similar oxygen concentrations from a high volatile and alow volatile bituminous coal revealed remarkable similaritiesin reactivity, micropore surface area (SCO2), and morphology.

(28) Davis, K. A.; Hurt, R. H.; Yang, N. Y. C.; Headley, T. J. Combust.Flame 1995, 100, 31–40.

(29) Borrego, A. G.; Marbán, G.; Alonso, M. J. G.; Alvarez, D.;Menendez, R. Energy Fuels 2000, 14, 117–126.

Figure 9. Surface area determined by CO2 adsorption (top) and N2

adsorption (bottom) of the chars prepared under different atmospheres.


The main difference between both series of chars relied onsurface area determined by N2 adsorption (SBET) and on sizedistribution of porosity which was shifted to larger size for theoxy-fuel series. The difference between both series of chars was

larger for the high volatile bituminous coal chars than for thelow volatile bituminous coal chars. This might have importantimplications for combustion under the diffusion-controlledregime.

The differences found in intrinsic reactivity and char reflec-tance between both series of chars can be attributed to thedifferent burnout levels of the chars, which were lower for the

Figure 10. Distribution of pore size determined applying the DFT model to CO2 (line) and N2 (bars) isotherms of the chars.

Figure 11. Maximum reactivity (Rmax) measured at 550 °C in air in athermobalance against oxygen content in the gas preparation atmosphereof the chars.

Figure 12. Intrinsic reactivity of char with relation to burnout.


oxy-fuel series than the combustion series for equivalent oxygenconcentrations.

The reflectance of isotropic material in chars has been usedto follow the course of combustion. The drop in reflectance withthe combustion progress does not support an increase in

structural order that could be expected from higher peaktemperatures in the particles but better an increase in themesoporosity that can decrease the optical density of thematerial. Both vitrinite- and inertinite-derived material reflec-tance decreased as combustion progressed despite the fact thatinertinite reached a lower reflectance than vitrinite due to itslower volatile matter content and also lower ability to reorganizeits structure during the passage through the reactor.

The presence of a limited amount of oxygen in the reactingatmosphere has a significant effect on volatile inhibition for thehigh volatile bituminous coal compared to the pyrolysis (N2)char. This effect is less marked for the low volatile bituminouscoal. The char prepared in CO2 for both coals had lower volatileyields than the pyrolysis counterparts. This could indicate thatCO2 is participating in the cross-linking reactions at the charsurface, reducing the plasticity of the material and preventingcoalescence of aromatic clusters in the carbonaceous structure.

Acknowledgment. Financial support of Principality of Asturiasand ENDESA through the Project PC04-03 and Ministry forEducation through PSE-02 is gratefully acknowledged. I. Fernández-Dominguez is thanked for the CO2 adsorption isotherms of the oxy-fuel series.

EF700353N

Figure 13. Evolution of char reflectance with burnout. Vitrinite-derivedmaterial was measured for COS and isotropic inertinite-derived materialfor SMK.


Documents

Comparison of chars obtained under oxy-fuel and conventional pulverized coal combustion atmospheres