Catalytic gasification of coal using eutectic salts: reaction kinetics with binary and ternary eutectic catalysts

Fuel 82 (2003) 305–317 www.fuelfirst.com

Catalytic gasification of coal using eutectic salts: reaction kinetics with binary and ternary eutectic catalystsq Atul Shetha, Yaw D. Yeboahb,*, Anuradha Godavartya, Yong Xub, Pradeep K. Agrawalc a

The University of Tennessee Space Institute, Tullahoma, TN 37388-8897, USA Department of Engineering, Clark Atlanta University, Atlanta, GA 30314, USA c School of Chemical Engineering, Georgia Institute of Technology, Atlanta, GA 30332, USA b

Received 20 September 2001; revised 29 May 2002; accepted 12 July 2002; available online 5 September 2002

Abstract Kinetic studies of the catalytic steam gasification of Illinois No. 6 coal were carried out using binary and ternary eutectic salt mixtures in a fixedbed reactor. The effects of major process variables such as temperature, pressure, catalyst loading and steam flow rate were evaluated for the binary 29% Na2CO3 –71% K2CO3 and ternary 43.5% Li2CO3 –31.5% Na2CO3 –25% K2CO3 eutectic catalyst systems. A Langmuir–Hinshelwood rate expression was developed to explain the reaction mechanism for steam gasification using the binary and ternary catalysts. The activation energy of the ternary catalyst (98 kJ/mol) was less than that of the binary catalyst (201 kJ/mol) or single salt such as K2CO3 (170 kJ/mol). The molar heats of adsorption for the ternary and binary catalysts were exothermic and about 180 and 92 kJ/mol, respectively. The molten nature of the ternary eutectic at the gasification temperatures and its lower activation energy favored higher gasification rates compared to the single and binary alkali metal salts. q 2002 Elsevier Science Ltd. All rights reserved. Keywords: Coal gasification; Steam; Eutectic catalysts; Coal char; Kinetic modeling

1. Introduction The reactivity of carbonaceous materials such as graphite and coal char towards CO2 and steam is strongly enhanced by the presence of alkali metal salts [1,2]. However, the exact role that the salts play in these processes is not completely understood and details of the catalytic mechanisms remain controversial. For a catalyst to function satisfactorily in carbon or char gasification, a three-phase interface must be maintained between the carbonaceous substrate, the catalyst phase and the gaseous oxidant. Whatever the detailed mechanism of the catalytic process is, the overall rate of gasification should be enhanced by improving the contacting of catalyst with carbon [3], as long as the gaseous oxidant still has ready access to this interface. Molten catalyst salts are better able to penetrate the coal structure and, hence, improve accessibility of the unavailable carbon sites in the interior of the coal/char. Previous studies of the alkali-catalyzed oxidation of graphite [4] have * Corresponding author. Tel.: þ 1-404-880-6619; fax: þ1-404-880-6615. E-mail address: [email protected] (Y.D. Yeboah). q Published first on the web via Fuelfirst.com—http://www.fuelfirst.com

shown that oxidation rates increase rapidly at temperatures in the vicinity of the melting point of the active catalyst phase. Therefore, it might be possible that eutectics, which melt at significantly lower temperatures than the pure salt constituents, would exhibit enhanced catalytic activity at lower temperatures. On the other hand, if the carbon surface becomes coated with a film of molten salt, kinetics would be limited by diffusion of the gaseous reactant through the film of salt and the overall reaction rate and apparent activation energy of the gasification process may be reduced. Scientists working at the General Electric (GE) Corporate Research and Development Center [5] evaluated the behavior of binary and ternary eutectic salt catalysts in gasification reactions of graphite and coal. They used a thermogravimetric analyzer (TGA) to carry out gasification runs at atmospheric pressure using CO2 and steam and restricted carbon conversion to less than 10– 20%. The eutectic catalysts were prepared by fusion of finely ground salt mixtures having compositions corresponding to the eutectic melting temperatures. McKee et al. [5] and Tetsuya et al. [6] found that the gasification rates of coal char and graphite in CO2 and steam in the temperature range 700– 900 8C can be considerably increased by the addition of

0016-2361/03/$ - see front matter q 2002 Elsevier Science Ltd. All rights reserved. PII: S 0 0 1 6 - 2 3 6 1 ( 0 2 ) 0 0 2 5 5 - 7

306

A. Sheth et al. / Fuel 82 (2003) 305–317

Nomenclature ð2dC=dtÞoverall overall gasification rate in a typical gasification process (g/s) ð2dC=dtÞthermal thermal gasification rate (g/s) ð2dC=dtÞcatalytic catalytic gasification rate (g/s) r gasification rate of carbon in the bed (g/s) C remaining carbon in the bed at any instant time, t (g) r=C specific gasification rate (s21) pH2O partial pressure of steam in the bed (kPa) k1 reaction rate constant k2 adsorption constant k apparent reaction rate constant Xc the carbon conversion at time t, defined as Xc ¼ ðC0 2 CÞ=C0 C0 the initial carbon content in the bed t gasification reaction time (s) NK 29% Na2CO3 – 71% K2CO3 binary catalyst LNK 43.5% Li2CO3 –31.5% Na2CO3 – 25% K2CO3 ternary catalyst binary and ternary eutectic alkali salt catalysts. The reduced melting points of the eutectics probably resulted in enhanced catalytic activity at the lower gasification temperatures by achieving a better dispersion of the salt phases on the carbonaceous substrates. However, there were some major issues that were not addressed in the earlier work. These included the potential enhancement in gasification performance at high carbon conversion levels (i.e. . 90%); the effect of representative gas atmosphere (e.g. CO and H2); the effect of gasification pressure; and the effect of catalyst impregnation techniques. In addition to uniform catalyst distribution, the effectiveness of the candidate eutectic salt mixtures in maintaining good contact with the retrieving carbon matrix can only be evaluated and be meaningful at high carbon conversion levels (e.g. . 90%). Unfortunately, this was not addressed in the earlier studies. In the fluidized-bed gasifier, during gasification, the gas atmosphere contains varying concentrations of species such as CO, H2, CO2, H2O and H2S. In the literature, it has often been shown that gases like CO and CO2 inhibit the catalysis of the C – H2O reaction by calcium, potassium, and sodium; and H2 inhibits the catalysis by calcium [7]. Thus, the evaluation of eutectic salt mixtures needs to be carried out using the appropriate gas atmosphere under representative conditions. The chief difficulty in the gasification reactor involves the degree of contact between the coal and the catalyst. When the catalyst is simply mixed and introduced into the system, the degree of contact is generally poor until the catalyst is melted in the reactor. In the moving-bed type gasifier, the degree of contact achieved is relatively high,

while the possibility of intimate contact in a jet flow bed type gasifier is significantly lower. In a fluidized-bed gasifier, results will vary directly with the degree of catalyst contact. For eutectic salt mixtures, the catalysts are assumed to be easily spread out over the surface of the coal because of their low melting points. However, initial distribution of such salt mixtures in the coal prepared at room temperature need to be homogeneous so that at gasifier conditions, the catalysts will penetrate into the coal matrix and be present at the reacting carbon sites. This can be achieved only if a proper method of initial catalyst application is employed to ensure good catalyst distribution as well as good penetration into the coal matrix. Also, the earlier GE studies included eutectic salt mixtures that contained halogen compounds. Under the Clean Air Act amendment of 1990, the emissions of HCl/Cl2 type ‘Air Toxics’ and halogen compounds are restricted to less than 10 tons/year to avoid costly control measures. In addition, chloride related stress corrosion under high steam pressure conditions might warrant the use of costly materials for the construction of gasifier and accessories. Hence, the fate of Cl2, Br2 and F2 introduced via catalysts as well as their possible adverse effects need to be understood well before such salt mixtures can be considered for catalyzing coal gasification process. In the present study, bench scale experiments were carried out on Illinois No. 6 coal in a high-pressure, hightemperature fixed-bed gasifier using binary and ternary eutectic salt catalysts. The effects of major process variables such as temperature, pressure, catalyst loading and addition methods, and steam flow rate on the gasification kinetics were evaluated using steam as the gasifying agent. A reaction kinetics model, valid up to high carbon conversion (. 90%), was also developed to relate the effects of temperature and partial pressure of steam on the gasification reaction. This paper presents the results of the kinetic studies with the 29% Na2CO3 – 71% K2CO3 binary and 43.5% Li2CO3 – 31.5% Na2CO3 – 25% K2CO3 ternary eutectic catalysts [8,9]. These catalyst systems are referred to in this paper as the NK and LNK catalyst systems, respectively. The identification and selection of suitable eutectic catalysts and the evaluation of the methods of addition to the coal were carried out in CO2 and steam gasification experiments and are reported in a separate and complementary article [8].

2. Experimental Illinois No. 6 coal sample obtained from the Penn State Coal Sample Bank was used for the bench scale experiments on catalytic coal gasification using the identified binary and ternary eutectic salt mixtures. The compositional data for the parent coal used in this study is given in Table 1. The catalyzed coal and pyrolyzed char samples were not analyzed for similar elemental compositions. The fixed

A. Sheth et al. / Fuel 82 (2003) 305–317 Table 1 Compositional data for the Illinois No. 6 coal used (hv Cb rank) [8,9] Proximate analysis (wt%)

Ultimate analysis (wt%)

H2O Ash Volatiles Fixed C

Ash C H N S O

13.20 11.62 35.44 39.74

11.62 57.33 3.98 0.99 4.80 8.07

carbon and ash contents of each sample were, however, determined using TGA to calculate overall carbon conversion for the given sample. The composition and eutectic temperatures of the catalysts used are given in Table 2. 2.1. Sample preparation The preparation of the coal-catalyst sample involved several steps. These are as follows. 2.1.1. Preparation of the eutectic catalysts The individual salts used for preparing the eutectic salt mixtures included the Aldrich Chemical Company’s Li2CO3 (grade 99 þ %), Na2CO3 (grade 99.5 þ %) and K2CO3 (grade 99 þ %). The binary and ternary catalysts were prepared by fusion of finely ground salt mixtures having compositions corresponding to the eutectic melting temperatures, as obtained from published phase diagrams [10]. The eutectic mixtures were heated in air at temperatures at least 100 8C above their reported respective eutectic points. The ternary eutectic salt, Li2CO3 – Na2CO3 – K2CO3 (LNK), formed a single phase liquid at its eutectic point. After cooling, the resulting glassy solidified LNK mixture was crushed and finely ground in an agate mortar and stored in a glove box under inert conditions. In the case of the binary eutectic salt, Na2CO3 – K2CO3 (NK), it formed a solid solution at its eutectic point and was thoroughly ground and stored under inert conditions. 2.1.2. Catalyst addition method The incipient wetness method of catalyst addition gave enhanced gasification rates over the physical mixing method in the TGA studies [8]. However, in the fixed-bed reactor studies the differences were less pronounced. Physical mixing of the catalyst to the raw coal was, therefore, chosen for the fixed-bed gasification studies reported in this paper due to its simplicity as a method of catalyst addition [8,9,

307

11]. After addition and mixing of the catalyst, the coalcatalyst sample was devolatilized. 2.1.3. Devolatilization/pyrolysis When coal is introduced into a gasification atmosphere, it is rapidly pyrolyzed to produce volatiles and a char. Bench gasification studies are, therefore, typically conducted on chars produced by prior devolatilization of the coal. This is because most bench units are incapable of handling the tars evolved during the pyrolysis process. In this study, the devolatilization process was carried in a Barnstead Thermolyne—Model F48015 muffle furnace at atmospheric pressure. The muffle furnace housed a cylindrical stainless steel container to maintain complete inert atmosphere during pyrolysis. The container had a heavy lid that had an inlet for the inert gas (N2) and an outlet for the volatiles and tars. A 20 g coal sample was taken for every batch of pyrolysis and purged with N2 (grade 5.0) initially. The muffle furnace was then set to a furnace temperature of 1023 K (750 8C) for the pyrolysis to begin. There was continuous purging of the inert gas to help remove the volatiles. The outlet gases were bubbled into a conical flask partially filled with water to allow the tars and other heavy volatile compounds to condense before the gases were released into the atmosphere through the hood. It took about 15 – 20 min for the furnace to reach 1023 K from room temperature. The pyrolysis was then carried out for an additional 3 h from the time the desired temperature of 1023 K was reached in the muffle furnace. At the end of the pyrolysis, the furnace was cooled to room temperature by continuous purging with the inert gas. This cooling took about 1 h. 2.1.4. Sieving In the nitrogen-purged dry box, the devolatilized/ pyrolyzed char was crushed in an agate mortar and sieved to obtain char particles ranging in size form 2 30 mesh (0.595 mm) to þ 100 mesh (0.149 mm). This fraction was collected in small glass bottles, suitably labeled and then stored in a nitrogen-purged dry box. 2.1.5. Catalytic steam gasification The catalytic steam gasification experiments were carried out in a high-pressure, high-temperature fixed-bed gasifier system. The gasifier (constructed of 3/4 in. type 304 stainless steel tubing with wall thickness of 0.065 in.) was typically operated with a downdraft gas flow regime and in a

Table 2 Composition and eutectic temperatures of the eutectic salt mixtures [9,10] Eutectic salt composition (mol%)

Eutectic temperature (K)

State at eutectic point

Nomenclature

43.5% Li2CO3 – 31.5% Na2CO3 –25% K2CO3 29% Na2CO3 –71% K2CO3

673 416

Liquid Solid solution

LNK NK

308

A. Sheth et al. / Fuel 82 (2003) 305–317

Fig. 1. Schematic of the high-pressure high-temperature fixed-bed gasifier setup.

Table 3 Experimental conditions for steam gasification runs using the LNK and NK eutectic catalysts Run

Temperature (K)

Pressure (MPa)

Catalyst loading (wt%)

Steam flow rate (ml/h)

Steam-to-initial carbon ratio (mol/h/mol)

Catalyst

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

923 1005 1044 1005 1005 1005 1005 1005 1005 964 923 1005 1044 1005 1005 1005 1005

0.44 0.44 0.44 1.12 2.14 0.44 0.44 0.44 0.44 0.44 0.44 0.44 0.44 0.44 0.44 0.44 0.44

10 10 10 10 10 5 15 10 10 10 10 10 10 5 15 10 10

13.8 13.8 13.8 13.8 13.8 13.8 13.8 5.0 8.7 13.8 13.8 13.8 13.8 13.8 13.8 5.0 8.6

5.38 5.38 5.38 5.38 5.38 5.16 5.88 1.95 3.39 5.66 6.10 6.10 6.10 5.37 6.68 2.21 3.80

LNK LNK LNK LNK LNK LNK LNK LNK LNK LNK NK NK NK NK NK NK NK

A. Sheth et al. / Fuel 82 (2003) 305–317

309

8610C gas chromatograph (GC). Experiments were also performed using a TGA to determine the fixed carbon content of the char sample before and after the gasification step. The GC and TGA data were analyzed and reconciled to determine the reactivity of the char samples and to obtain a rate expression to explain the kinetics of the gasification reaction. Using the two eutectic catalysts (LNK and NK), a set of 17 experiments were performed on steam gasification, each lasting for about 4 h, to study the effects of temperature, pressure, catalyst loading, and steam flow rate. The operating conditions for these runs are tabulated in Table 3. Runs 1– 10 were carried out using the ternary eutectic (LNK) system, whereas Runs 11– 17 were performed using the binary eutectic (NK) system as the catalyzing agent. Fig. 2. Carbon conversion with time at different temperatures with NK catalyst (pressure ¼ 0.44 MPa; catalyst loading ¼ 10 wt%; steam flow rate ¼ 13.8 ml/h; steam to carbon ratio ¼ 6.1 mol/h/mol). B 1055 K; X 1044 K; O 923 K.

differential fixed-bed mode. The reactor was packed with ceramic beads to support the char sample towards its center using two 200-mesh stainless steel screen baskets. Temperature of the bed was measured using a type K thermocouple. A schematic of the high-pressure hightemperature fixed-bed gasifier is shown in Fig. 1 and consisted of gas/steam feeding and preheating units, the reactor and 3-zone Lindberg furnace, back pressure regulator, condenser and dryer units, and gas analysis (gas chromatograph) system. The differential char bed accommodated about 2.5 g of char during each experimental run. The exit gases from the steam gasification reaction were analyzed for carbonaceous species using an off-line SRI

2.1.6. Determination of fixed carbon (before and after gasification) The samples were pyrolyzed in microgram quantities under inert conditions using nitrogen (grade 5.0) at a heating rate of 12 8C/min in a TGA. The system was brought to 1023 K (750 8C) and held at this temperature for about 2 h. A typical weight loss profile with respect to time generated by the TGA, then helped to determine the amount of moisture, volatile matter and the combined ash and fixed carbon content in the sample. The fixed carbon and ash contents were determined by the oxidation of the pyrolyzed residue that was left using air (grade 5.0) as the oxidizing agent. The fixed carbon content, before and after gasification, and the GC analysis were used to determine the overall carbon conversion.

3. Results and discussion Reproducibility runs that were carried out with the LNK catalyst with different sizes of initial charge (1.7 –2.5 g) and at different reaction conditions gave error bars in the carbon conversion data within ^ 5 up to 90% carbon conversion. Also in the data analysis, rate data were used only up to about 90% carbon conversion. Based on these and regression analysis and round off errors, it is estimated that the error bars in the various data plots in the figures provided in this paper are about ^ 5 –10%. 3.1. Effect of temperature

Fig. 3. Carbon conversion with time at different temperatures with LNK catalyst (pressure ¼ 0.44 MPa; catalyst loading ¼ 10 wt%; steam flow rate ¼ 13.8 ml/h; steam to carbon ratio <5.5 mol/h/mol). X 1044 K; B 1005 K; þ 964 K; O 923 K.

Figs. 2 and 3, respectively, show the degree of conversion vs time for catalytic steam gasification of char at several temperatures for the LNK and NK catalyst systems. For similar temperatures, the LNK catalyst system gave better conversion levels compared to the NK catalyst system. At 923 K, the reaction was not complete in both catalyst systems, although the conversions were much higher with LNK catalysts than with NK catalysts. At 1005 K, the NK catalyst yielded only 85% conversion,

310

A. Sheth et al. / Fuel 82 (2003) 305–317

Fig. 4. Effect of temperature on [CO2]/[CO] molar ratio for the LNK system (pressure ¼ 0.44 MPa; catalyst loading ¼ 10 wt%; steam flow rate ¼ 13.8 ml/h; steam to carbon ratio <5.5 mol/h/mol). X 1044 K; B 1005 K; þ 964 K; O 923 K.

Fig. 5. Carbon conversion with time at different operating pressures with the LNK catalyst (temperature ¼ 1005 K; catalyst loading ¼ 10 wt%; steam flow rate ¼ 13.8 ml/h; steam to carbon ratio <5.5 mol/h/mol). X 2.14 MPa; B 1.12 MPa; O 0.44 MPa.

whereas the reaction under similar conditions was almost 99% complete with the LNK catalyst. The variation in conversion levels between the two catalyst systems is attributed to the molten nature of the ternary eutectic salt (melting point of LNK is less than the gasification temperature whereas the binary eutectic salt remained a solid solution). We believe this liquid nature of the LNK at the gasification temperature provides a better dispersion of the LNK catalyst during gasification, thereby enhancing the reaction rate. Based on the scanning electron micrographs (SEM) of the catalysts [12], the LNK catalyst provides more porosity and crystallinity in comparison to the NK catalyst. At 923 K, though the LNK catalyst (with a melting point of about 673 K) is in the liquid phase, the reaction does not go to completion due to lower gasification rates. This is explained from the thermodynamics of the gasification

reaction, where the DGreaction at this temperature is þ 3.7 kJ/mol [13]. Thus, the system does not have enough energy at 923 K to overcome the necessary energy barrier and proceed in the forward direction, at an acceptable gasification rate. The water –gas reaction during steam gasification is given by Eq. (1).

Table 4 Specific gasification rates for Runs 1–17 Run

Catalyst

1/C(2dC/dt ) (min21)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

LNK LNK LNK LNK LNK LNK LNK LNK LNK LNK NK NK NK NK NK NK NK

0.004 0.0208 0.0301 0.0177 0.0168 0.0037 0.0281 0.0053 0.0153 0.0135 0.001 0.0158 0.024 0.0018 0.0394 0.0128 0.0087

C þ H2 O ! CO þ H2

ð1Þ

As gasification temperature was increased, the CO production was relatively higher for the LNK catalyst system than the NK catalyst system. Fig. 4 shows the effect of temperature on the product distribution for the LNK system. The [CO2]/[CO] ratio increased with a decrease in temperature, a trend consistent with the thermodynamics of the water – gas shift reaction [13]. The ratio was practically the same for runs above 1005 K until conversions were above about 0.8. At 923 K, the conversion was limited to 0.6 and no CO2/CO ratio is available beyond this conversion limit. A similar trend was observed for the binary catalyst system. The fact that the ratio [CO2]/[CO] remained practically constant over a rather broad range of fractional conversions (# 0.8) suggests that the product distribution is controlled by thermodynamics and not kinetics. 3.2. Effect of pressure Pressure has a significant effect on the methanation reaction in the presence of CO and H2 during the gasification process [14]. In our study, the main focus was to increase the overall gasification rate of the char sample. Runs 2, 4 and 5 (Tables 3 and 4) using the LNK catalyst demonstrated the effect of system pressure on the overall gasification rate. No significant effect of pressure on the carbon conversion and the overall gasification rate was

A. Sheth et al. / Fuel 82 (2003) 305–317

Fig. 6. Carbon conversion with time at different catalyst loadings for the LNK catalyst (temperature ¼ 1005 K; pressure ¼ 0.44 MPa; steam flow rate ¼ 13.8 ml/h; steam to carbon ratio <5.5 mol/h/mol). X 15 wt% LNK; B 10 wt% LNK; O 5 wt% LNK.

observed in these runs as shown in Fig. 5. This indicated that the reaction rate was independent of pressure in the pressure range studied. These results are supported by the literature, where a similar effect of the system pressure was seen when K2CO3 was used as the catalyst for steam gasification of coal [15]. Similar behavior was observed for the NK catalyst as well. 3.3. Effect of catalyst loading Increase in catalyst loading increases the metal/carbon or M/C ratio. Runs 2, 6 and 7 using the LNK catalyst (Fig. 6) and Runs 12, 14 and 15 using the NK catalyst (Fig. 7) were performed by varying the amount of catalyst loading to the amount of raw coal taken initially. At low catalyst loading

Fig. 7. Carbon conversion with time for different catalyst loadings for the NK catalyst (temperature ¼ 1005 K; pressure ¼ 0.44 MPa; steam flow rate ¼ 13.8 ml/h; steam to carbon ratio <5.5 mol/h/mol). X 15 wt% NK; B 10 wt% NK; O 5 wt% NK.

311

of 5 wt%, the conversion was only 60% complete for the LNK system and 35% complete for the NK system. Upon increasing the loading to 10 wt%, the reaction was almost 99% complete for Run 2 (using LNK) in comparison to about 85% completion for Run 12 (using NK). Further increasing the loading to 15 wt% gave complete conversion in the binary catalyst system. At low catalyst loading of 5 wt%, the fraction of the catalyst lost by the reaction of the catalyst with the mineral matter content of the coal during gasification, might be sufficient to yield low gasification rates and lower overall carbon conversions. Based on our catalyst recovery, regeneration and recycle work by water washing of the gasifier char [9], we can say that recovery of Li, K, and Na was not complete. Less than 100% recovery of these alkali metals suggests that some of them were present in the gasified chars as insoluble alumino-silicates. The presence of such insoluble alumino-silicates have been confirmed by Liu and Zhu [1] and Formella et al. [16]. Single salts such as K2CO3 showed a similar behavior at low loadings in steam gasification [16]. There was no significant rise in the overall gasification rate upon increasing the catalyst loading from 10 to 15 wt% in the LNK system. This may be due to the saturation of the available active sites in the carbon matrix by the increased number of cations available in the system. Mims and Pabst [17] showed that catalytic activity is due to the active complex formed between the alkali metal and the acidic groups on the coal surface. Once the acidic groups have formed complexes with alkali metals, any additional catalyst may play little or no role in the gasification process. SEM analysis of the pyrolyzed and gasified chars provided images of porosity and catalyst distributions in the samples [12]. Based on these images it was found that at 5 wt% loading the catalyzed coal sample was highly porous and this porosity decreased with increasing catalyst loading. The elemental mapping of sulfur and potassium in the 15 wt% catalyzed samples showed very dense distribution of sulfur and potassium [12]. Based on these results, it was hypothesized that probably at higher loadings, the active sites became saturated. Schumacher and co-workers [15] made similar observations with the K2CO3-catalyzed coal. Microstructural and X-ray diffraction characterizations were used to study the effect of catalyst distribution on the coal matrix after gasification [9,12]. It was found that the gasified chars of the LNK system were highly porous and crystalline in nature. This crystalline nature was attributed to the formation of K2S. However, in the case of the NK system, the chars were porous but showed the presence of potassium polysulfides (K2S4, K2S5 and K2S6). Similarly, from our catalyst recovery, reactivation and recycle work, we found that all the alkali metals were not extracted by water washing [9]. In the case of the LNK system, about 50 – 75% of Na and K, and 5 –8% of Li were recovered. For the NK system, the water wash recovery of sodium was about 35 –58% and for potassium was about 42– 53%. These results show that the steam and char have significant

312

A. Sheth et al. / Fuel 82 (2003) 305–317

Fig. 8. Effect of metal (catalyst) to carbon (M/C) ratio on specific gasification rate for the NK and LNK catalysts. V NK; B LNK.

effect on the physical state of the alkali metal salts in the eutectic mixtures (by changing their chemical composition and form through chemical interactions with the mineral matter content). For example, reactions may convert carbonate catalysts to an alumino-silicate or sulphate that is less soluble. The effect of the metal-to-carbon (M/C) ratio on the specific gasification rate for the two catalyst systems is shown as a combined plot in Fig. 8. The figure indicates that M/C ratio may be independent of the type of catalyst in the range studied, especially for M/C # 3%. For the same catalyst loading, however, the ratio M/C was generally higher for the LNK system than for the NK system. This may explains the reason for higher conversion levels and

Fig. 9. Carbon conversion with time at different steam flow rates for the LNK catalyst (temperature ¼ 1005 K; pressure ¼ 0.44 MPa; catalyst loading ¼ 10 wt%). X 13.8 ml/h; B 8.70 ml/h; O 5.04 ml/h.

Fig. 10. Carbon conversion with time at different steam flow rates for the NK catalyst (temperature ¼ 1005 K; pressure ¼ 0.44 MPa; catalyst loading ¼ 10 wt%). X 13.80 ml/h; B 8.70 ml/h; O 5.04 ml/h.

overall gasification rates for the ternary catalyst system when compared to the binary catalyst system. The above phenomenon infers that the reaction kinetics depend on the number of alkali metal cations available on the active sites of coal for enhancing the gasification process. At high catalyst loadings (. 15 wt%) or M/C . 4%, the NK catalyst system appeared to show higher gasification rate than the LNK catalyst system. Further studies are needed to understand the behavior of both catalysts, especially at higher loadings. 3.4. Effect of steam flow rate Experiments were performed at steam flow rates varying from 5 to 14 ml/h, corresponding to steam-to-initial carbon molar ratios in the range of 2:1 to 6:1 (mol/h of steam/mol of initial carbon in the bed). Runs 2, 8 and 9 using the LNK catalyst (Fig. 9) and Runs 12, 16 and 17 (Fig. 10) using the NK catalyst depict the effect of steam flow rate. As the initial steam flow rate was increased, the extent of carbon conversion increased for the ternary system. Further increasing the flow rate above 9 ml/h, corresponding to a steam to carbon ratio of about 3.4, gave no significant increase in the overall gasification rate in the LNK system. This could be due to the saturation of the active site. Addition of excess steam, beyond that needed for overall reaction, provided no effect on the overall gasification rate. In the case of the binary system, it gave 100% conversions at lower water flow rates compared to about 80 – 85% conversion at higher steam flow rates. The NK behavior was quite different from that of the ternary catalyst. Detailed microscopic analysis of the catalyst distribution is needed to understand the behavior of the NK catalyst during gasification. The fractional amount of CO produced (on a molar basis) increased from less than 10% to about 30 – 35% upon

A. Sheth et al. / Fuel 82 (2003) 305–317

Fig. 11. Effect of steam flow rate on the water–gas shift reaction (as measured by the CO2/CO molar ratio with time) for the LNK system (temperature ¼ 1005 K; pressure ¼ 0.44 MPa; catalyst loading ¼ 10 wt%). X 13.8 ml/h; B 8.7 ml/h; O 5.04 ml/h.

increasing the steam flow rates for the LNK system (Fig. 11). In the case of the NK system, the fraction of CO dropped from 40% to about 30% with the increase in steam flow rates. However, both catalysts behaved in accordance with the thermodynamics of the shift reaction, by showing a clear rise in the [CO2]/[CO] molar ratio with the increase in the steam flow rate and thereby enhancing the forward reaction. The rise in this ratio with the increase in the steam flow rate plotted as a function of carbon conversion in the bed for NK catalyst system is shown in Fig. 12. 3.5. Comparison of LNK and NK catalyst systems In this paper, the catalyst systems were physically mixed with the coal on weight basis. Based on the chemical

313

composition, the LNK system contained about 0.020 g-atoms of alkali metals per gram of catalyst mix. In comparison, the NK system contained about 0.015 g-atoms of alkali metals per gram of catalyst mix. Thus, for a similar weight percentage loading on the coal, the LNK system provides more atoms of alkali metals than the NK system. Also, based on our TGA study carried out with single salts [8,9], lithium salts were found to be more reactive than the other alkali metal salts. It is possible then that the higher gasification activity at a similar catalyst loading exhibited by the LNK catalyst system may be due to higher concentrations of alkali metals, especially lithium. The sodium contents of the LNK and NK catalyst systems were 0.006 and 0.004 g-atoms/g of catalyst, respectively. Similarly, the potassium contents of the LNK and NK catalyst systems were 0.005 and 0.011 g-atoms/g of catalyst, respectively. Since based on the single salt study in TGA, we had found potassium carbonate to be more active than sodium carbonate, this would suggest that the NK should have more activity. But this is opposite of what was found in the experiments. We believe that in addition to the differences in the physical state of the eutectics at the gasification temperature, the presence of lithium and the overall higher alkali metal concentration in the LNK catalyst system are responsible for the higher activity observed in the LNK system. McKee [2] previously studied the pathways by which the ternary salts react. From his early work at General Electric Company on graphite, he proposed reaction mechanisms for the sodium, potassium and lithium-based catalysts for the C – H2O and C –CO2 gasification processes. In general, he showed that oxysalts of the alkali metals were effective catalysts and were involved in a cyclic series of elementary reactions. The mechanism for the C –CO2 gasification was also supported by Wood and Sancier [18], Moulijn et al. [19], Mims and Pabst [17], Fox and White [20], and later by McKee and Chatterji [4]. Lizzio et al. [21] also supported this mechanism based on their work utilizing transient kinetics method. In the ternary and binary catalyst systems evaluated in the present study, probably similar reaction pathways as those proposed by McKee [2] are responsible for the gasification. 3.6. Kinetic modeling Kinetics of catalytic steam gasification of coal/graphite has been studied extensively by several researchers [15,22, 23]. A Langmuir – Hinshelwood mechanism is generally proposed for the steam gasification of coal.

Fig. 12. Effect of steam flow rates on the water–gas shift reaction (as measured by the CO 2 /CO ratio with time for the NK catalyst (temperature ¼ 1005 K; pressure ¼ 0.44 MPa; catalyst loading ¼ 10 wt%). X 13.8 ml/h; B 8.7 ml/h; O 5.04 ml/h.

CðsÞ þ H2 OðgÞ ! COðgÞ þ H2ðgÞ

ð2Þ

H2 OðgÞ ! H2 OðadÞ

ð3Þ

CðsÞ þ H2 OðadÞ ! H2ðadÞ þ COðadÞ

ð4Þ

H2ðadÞ ! H2ðgÞ

ð5Þ

314

A. Sheth et al. / Fuel 82 (2003) 305–317

Hinshelwood type of rate expression is obtained 2rc ¼

k1 pH2 O C 1 þ k2 pH2 O

ð7Þ

where p H2O is the partial pressure of steam in the bed (kPa), k1 is the reaction rate constant, k2 is the adsorption constant, rc is the gasification rate of carbon in the bed (g/ min), C is the remaining carbon in the bed at any instant time (g). Since excess steam was used in all the runs reported in this study (Table 3), it is fair to assume that in any given run (at the specified value of temperature, partial pressure of steam ( p H2O), and catalyst loading), the term k1 pH2 O 1 þ k2 pH2 O Fig. 13. Plots of fitted first-order rate expression with respect to carbon vs. time for 10 wt% LNK catalyst loading at different experimental run conditions. O Run 1, 923 K; 0.44 MPa; 13.8 ml/h of steam flow: A Run 2, 1005 K; 0.44 MPa; 13.8 ml/h of steam flow: K Run 5, 1005 K; 2.14 MPa; 13.8 ml/h steam flow: B Run 8, 1005 K; 0.44 MPa; 5.04 ml/h steam flow.

remains unchanged during the entire duration of the run (, 4 h). It would then appear from Eq. (7) that the transient behavior of a run could be expressed by a first-order equation 2rc ¼ kC

ð8Þ

where COðadÞ ! COðgÞ

ð6Þ

Eq. (2) represents the overall gasification step. Eqs. (3) –(6) represent the individual reaction steps. Water molecule is initially adsorbed on the active sites of the coal matrix and then reacts with carbon to produce CO and H2 species, which subsequently desorb. Product inhibition due to H2 or CO was neglected. If one assumes that the surface reaction between an adsorbed water molecule and carbon to be the rate-determining step, the following Langmuir –

k¼

k1 pH2 O 1 þ k2 pH2 O

Eq. (7) or (8) can be rewritten as 2

dC ¼ kC dt

ð9Þ

or 2lnð1 2 Xc Þ ¼ kt

ð10Þ

where Xc, the carbon conversion at time t, is defined as Xc ¼ ðC0 2 CÞ=C0 and C0 is the initial carbon content in the bed. Thus, a linearity in the plot of 2lnð1 2 Xc Þ vs time would provide the first test of the validity of the Langmuir– Hinshelwood kinetics as represented by Eq. (7). Fig. 13 shows 2lnð1 2 Xc Þ vs time plots for the LNK catalyst; similar plots for the NK catalyst system are shown in Fig. 14. In both cases, reasonably linear fits are obtained. In view of the initial transients associated with the run start-up, the linear fits are quite acceptable. Similar first-order behavior with respect to carbon has been reported earlier [13,22,23]. In general, the gasification reaction would occur thermally (homogeneously) as well as catalytically. The overall char gasification can therefore be expressed as:       2dC 2dC 2dC ¼ þ ð11Þ dt overall dt catalytic dt thermal Fig. 14. Plots of fitted first-order rate expression with respect to carbon vs. time for the NK catalyst system at 0.44 MPa pressure and other experimental run conditions. O Run 11, 923 K; 10 wt% loading; 13.8 ml/h of steam: A Run 12, 1005 K; 10 wt% loading and 13.8 ml/h of steam: K Run 14, 1005 K; 5 wt% loading; 13.8 ml/h of steam: B Run 17, 1005 K; 10 wt% loading; 8.6 ml/h of steam flow.

However, the kinetics of such mixed gasification cannot be separated easily. At the temperatures employed in this study, thermal gasification rates are usually negligible compared to the catalytic gasification rates. We, therefore, ascribed the observed rates solely to the catalytic process as

A. Sheth et al. / Fuel 82 (2003) 305–317

315

Fig. 15. Determination of k1 and k2 from a plot of Eq. (12) for the LNK catalyst at different temperatures. X 923 K; B 964 K; K 1005 K; B 1044 K.

expressed by Eq. (7). Eq. (7) can be inverted to give a linearized form as: C 1 k þ 2 2 ¼ rc k1 pH2 O k1

ð12Þ

The calculated specific rate data (2 rc/C ) from the experimental results for different temperatures and partial pressures of steam were used to fit this linearized rate expression. The resulting linearity following Eq. (12) is shown in Fig. 15 for the LNK system. The k1 and k2 values determined from the slopes (1/k1) and intercepts (k2/k1) of the linear plots in Fig. 15 were used to obtain the slopes and intercepts of Arrhenius type plots (ln ki vs 1/T ) as shown in Fig. 16. A similar procedure was carried out to obtain a rate expression for the NK system in Fig. 17. The obtained values of the energy or heat of adsorption and the

Fig. 17. Determination of k1 and k2 from a plot of Eq. (12) for the NK catalyst at different temperatures. O 923 K; B 1005 K; V 1044 K.

preexponential or frequency factors for both catalyst systems are given in Table 5. From the slopes of the Arrhenius plots for the adsorption rate constant, k2, one can see that the values in both catalyst systems are negative (2 180 kJ/mol for LNK and 2 92 kJ/mol for the NK), indicating that the molar heat of adsorption is exothermic in contrast to the water – gas reaction. Generally chemisorption is an exothermic process [24]. It may, however, when accompanied by dissociation, be endothermic [24]. The obtained values of the heat of adsorption support the Langmuir– Hinshelwood adsorption model without dissociation proposed in this study. The activation energy for the LNK system was found to be a little less than half the activation energy of the NK system. This shows that the gasification reaction is less temperature sensitive and better enhanced using the ternary catalyst system than the binary catalyst system. As shown in Table 2, at the eutectic temperature the LNK catalyst exists as a homogeneous liquid phase. On the other hand, the NK catalyst exists as a homogeneous solid phase. Even though the gasification temperatures (964 – 1044 K) were higher than the eutectic temperatures of the LNK (673 K) and NK (416 K) catalysts, it is not known whether at 964 –1044 K, the NK catalyst system still exists as a solid solution or liquefies prior to that. If it is present as a solid phase than gasification rate can be lower due to solid –solid interaction. Also, our single salt studies carried out in the TGA system Table 5 Rate parameters for steam gasification

Fig. 16. Arrhenius plots for the reaction (k1) and adsorption (k2) rate constants for the LNK catalyst. A ln k1 vs. 1/T; B ln k2 vs. 1/T.

Catalyst

Parameter

ki

Ei or DHads (kJ/mol)

LNK LNK NK NK

k1 k2 k1 k2

9.2 £ 103 kPa21 s21 3.7 £ 10212 kPa21 2.0 £ 109 kPa21 s21 3.4 £ 1027 kPa21

98.6 2180.3 201.5 291.9

316

A. Sheth et al. / Fuel 82 (2003) 305–317

[15]. This allows higher gasification rates for the ternary catalyst when compared to single and binary alkali metal salts. The value of the activation energy obtained for the LNK system was consistent with that obtained at General Electric during their steam gasification studies in a TGA [5]. Based on the results of this study, the derived Langmuir– Hinshelwood type rate equation for the two catalyst systems are given by Eqs. (13) and (14).   211 863 154:1 exp pH2 O 21 dC T    ¼ 21 695 C dt ð13Þ pH2 O 1 þ 3:7 £ 10212 exp T ðLNK systemÞ

Fig. 18. Comparison of the model predictions with the experimental results for the specific gasification rates of the LNK catalyst at 10 wt% catalyst loading.

have shown that the catalytic activity of alkali metal carbonates in CO2-gasification increased in the order of Li . Cs . K . Ca , Na . raw coal [8,9]. This was also demonstrated by the earlier GE study involving steam gasification of graphite using single salts of Na, K and Li [2]. Since the LNK catalyst contains lithium whereas the NK catalyst does not, it is possible that the better performance exhibited by the LNK catalyst may be due to the lithium. Additional studies will be needed to determine the role of the individual alkali metals in these eutectic salt mixtures. The activation energy was lower for the ternary catalyst (98.6 kJ/mol) compared to the binary catalyst (201 kJ/mol) and the literature single salt (K2CO3) value of 170 kJ/mol

  224 241 3:3 £ 107 exp pH2 O 21 dC   T  ¼ 11 050 C dt pH2 O 1 þ 3:4 £ 1027 exp T

ð14Þ

ðNK systemÞ The derived rate expression for each catalyst system was used to estimate the specific gasification rates ð2rc =CÞ at different steam flow rates and the results obtained were found to be in close comparison to the experimental results. The consistency of the experimental results to the model is shown in Figs. 18 and 19 for the two catalyst systems, respectively. The model seems to over predict the specific gasification rates at higher temperatures in both cases. This could be attributed to the reaction rate being probably more influenced or controlled by pore- and inter-particle diffusion instead of surface chemical reaction at higher temperatures [1]. The model needs to be modified after additional experimentation to incorporate such effects.

4. Conclusions The following conclusions may be drawn from the bench scale studies of the effects of various process variables (temperature, pressure, steam flow rate and catalyst loading) on the catalytic steam gasification of Illinois No. 6 coal using the ternary (LNK) and binary (NK) eutectic catalysts.

Fig. 19. Comparison of the specific gasification rates calculated from the model with the experimental results for the 10 wt% NK catalyst loading.

1. At similar temperatures, the ternary LNK catalyst gave better carbon conversions and gasification rates than the binary NK catalyst. This was attributed to better catalyst dispersion resulting from the molten nature of the ternary eutectic in comparison to the solid solution of the binary eutectic at the gasification temperatures. It may also be attributed to the presence of the lithium (based on the observed order of catalytic activity of the individual alkali metal components of the eutectic salts) and the slightly higher g-atoms of metal per gram of catalyst matrix for the LNK catalyst.

A. Sheth et al. / Fuel 82 (2003) 305–317

2. There was a significant effect of catalyst loading on the gasification reaction in both catalyst systems. Beyond 10 wt%, however, the effect was marginal, especially for the LNK catalyst. The gasification rates and conversion levels were found to increase with the increase in the metal-to-carbon (M/C) ratio. Below 10 wt% catalyst loading, the specific gasification rate increased linearly with the M/C ratio and indicated the gasification rate to be nearly independent of the catalyst type and more dependent on the concentration of the alkali metals. 3. Within experimental error, there was no effect of system pressure on the gasification rate in the LNK and NK systems. 4. The effect of steam flow rate showed a different behavior in the two catalyst systems. With increase in steam flow rate, the carbon conversion levels in the LNK system increased. However, the NK system appeared to give higher overall conversions at lower flow rates. While the CO trends appeared to differ, the CO2/CO molar ratios of both catalysts were in accordance with the thermodynamics of the shift reaction and suggest the product distribution is controlled by thermodynamics and not kinetics. Additional work is needed to understand the observed differences between the two catalysts. 5. A simple Langmuir – Hinshelwood type rate model, excluding the effect of hydrogen inhibition, provided a reasonably good fit to the experimental results at different temperatures, pressures and steam flow rates. 6. The activation energy of the NK system (201 kJ/mol) was higher than that of the LNK system (98 kJ/mol) indicating a better catalytic activity by the liquid ternary catalyst. The molar heats of adsorption for the LNK and NK systems were exothermic and about 180 and 92 kJ/mol, respectively. The exothermic results support the Langmuir – Hinshelwood adsorption without dissociation model proposed.

Acknowledgements We would like to thank the Department of Energy for its support of this project under grant number DE-FG2697FT97263. We also wish to acknowledge the support of

317

Clark Atlanta University, the University of Tennessee Space Institute and the Georgia Institute of Technology.

References [1] [2] [3] [4] [5] [6] [7] [8] [9]

[10] [11] [12] [13] [14]

[15] [16] [17] [18] [19] [20] [21]

[22] [23] [24]

Liu ZL, Zhu HL. Fuel 1986;65:1334 –8. McKee WD. Carbon 1982;20:59–66. Spiro CL, McKee DW, Kosky PG, Lamby E. Fuel 1984;63:686–91. McKee DW, Chatterji D. Carbon 1975;13:381–90. McKee WD, Spiro CL, Kosky PG, Lamby EJ. Fuel 1985;64:805– 9. Tetsuya H, Kozo N, Masashi A, Yoshiyuki N. Appl Catal 1991;67: 189–202. Walker Jr. PL, Matsumoto S, Hanzawa T, Muira T, Ismail INK. Fuel 1983;62:140 –9. Yeboah YD, Xu Y, Sheth A, Godavarty A, Agrawal P. Accepted for publication in Carbon. Yeboah YD, Xu Y, Sheth A, Agrawal P. Catalytic gasification of coal using eutectic salts mixtures. Final Technical Report to the Department of Energy under Contract No. DE-FG26-97FT97263; December 2001. Levin EM, Robbins CR, McMurdie HF. Phase diagrams for ceramists. American Ceramic Society; 1964. Godavarty A. Catalytic coal gasification using eutectic salt mixtures. Master’s Thesis. Knoxville: University of Tennessee; August 1999. Godavarty A, Agarwal A. Energy Fuels 2000;14:558–65. Chase Jr. MW, Davies CA, Downey Jr. JR, Frurip DJ, McDonald RA, Syverud AN. J Phys Chem Ref Data 1985;14:535–600. Gallagher JE, Euker Jr CA. Catalytic coal gasification for SNG manufacture. 6th Annual International Conference on Coal Gasification, Liquefaction and Conversion to Electricity, Pittsburgh; July 31 – August 2, 1979. Schumacher W, Mu¨hlen JH, Heek KHV, Ju¨ntgen H. Fuel 1986;65: 1360– 3. Formella K, Leonhard P, Sulimma A, Heek KHV, Ju¨ntgen H. Fuel 1986;65:1470 –2. Mims CA, Pabst JK. International Conference Coal Science, Du¨sseldorf, FRG; 1981. p. 730–6. Wood BJ, Sancier KM. Catal Rev—Sci Engng 1984;26:233–79. Moulijn JA, Cerfontain MB, Kapteijn F. Fuel 1984;63:1043 –7. Fox DA, White AH. Ind Engng Chem 1931;23:259– 66. Lizzio AA, Jiang H, Radovic LR. On the kinetics of carbon (char) gasification: reconciling models with experiments. Carbon 1990; V28(1):7–19. Vadovic JC, Eakman JM. Kinetics of potassium catalyzed gasification. Annual ACS meeting, Florida; September 1978. Tuyl ED, William JT. Fuel 1986;65:58–62. Bond GC. Heterogeneous catalysis—principles and applications, 2nd ed. Oxford: Clarendon Press; 1990.