Energy & Fuels 2004, 18, 883-888
883
Evolution of Carbon Microstructures during the Pyrolysis of Turkish Elbistan Lignite in the Temperature Range 700-1000 °C Billur Sakintuna and Yuda Yu¨ru¨m* Faculty of Engineering and Natural Sciences, Sabanci University, Tuzla, Istanbul 34956, Turkey
Sevil C¸ etinkaya Department of Chemistry, Hacettepe University, Beytepe, Ankara 06532, Turkey Received November 19, 2003. Revised Manuscript Received March 2, 2004
Chars and activated carbons were produced from raw, HCl-washed, and HCl/HF-washed Elbistan lignites at 700 °C, 800 °C, 900 °C, and 1000 °C. The pyrolysis and activation reactions increased the BET areas (m2/g carbon) of the acid-washed samples almost 10-fold. The increase of the BET areas (m2/g carbon) by increasing the temperature of pyrolysis or activation from 700 °C to 1000 °C was explained with the burn-out of carbon which led to the development of porosity. The values of the stacking heights, Lc of HCl/HF-washed samples seemed to increase from 1.0 nm to 1.5 nm, the average number of graphene sheets increased from 2.8 to 4.4, and the lateral size of the crystallites, La, increased very faintly from 5.0 nm to 5.5 nm when the pyrolysis temperature was increased from 700 °C to 1000 °C. Activation reactions performed at the same temperature range did not change the stacking heights. The values of Lc for activated HCl/HFwashed samples stayed almost constant in the same range as for the carbonized samples within 1.0-1.5 nm. This indicated that oxidative reactions during activation did not alter the stacking heights of the crystallites significantly in the temperature range of 700-1000 °C. The results presented in the present work can be considered as indications for the development of turbostratic (fully disordered) structures in the temperature range of 700-1000 °C.
Introduction Carbonaceous materials such as graphite, soot, chars, coke, and coals have characteristic structural properties which differ from mostly amorphous to completely ordered graphitic crystalline structure. The degree of order in these structures clearly depends on the thermal treatment of the material as well as the type of precursor of the carbonaceous material. The structures can be characterized by various parameters such as interlayer spacing d, the stacking height Lc, and the lateral size of the crystallites La, Figure 1. Currently used X-ray diffraction techniques for the measurement of graphene sheet size and turbostratic crystallite thickness were developed by Warren1,2 in 1941 and by others.3-7 In the most widely accepted model of the structure of “turbostratic” carbon, the atoms are arranged in layers but stacked randomly instead of the order ABABAsthe sequence of graphite and interlayer spacings also occur * Author to whom correspondence should be addressed. Phone: 90216-483 9512. Fax: 90-216-483 9550. E-mail:
[email protected]. (1) Warren, B. E. Phys. Rev. 1934, 2, 551. (2) Warren, B. E. Phys. Rev. 1941, 59, 693. (3) Brindley, G. W.; Mering, J. Acta Crystallogr. 1951, 4, 441. (4) Hirsch, P. B. Proc. R. Soc. Ser. A 1954, 226, 143. (5) Diamond, R. Acta Crystallogr. 1957, 10, 359. (6) Diamond, R. Acta Crystallogr. 1958, 11, 129. (7) Short, M. A.; Walker, P. L., Jr. Carbon 1963, 1, 3.
Figure 1. Model for a stack of layers of a carbon crystallite.
randomly. It is generally accepted8 that the development of turbostratic structures of carbon takes place above 1200 °C. Information about these parameters is significant to comprehend the processes such as pyrolysis, gasification, graphitization, etc. Oberlin9 demonstrated that the 700 °C-1500 °C range is the most significant for the production of activated carbons of large surface areas. In this stage, basic structural units are claimed to associate and the number of layers reaches to 8-10. The aim of this study was to investigate the probability of the development of turbostratic carbon structures at a relatively lower temperature range of 7001000 °C during the preparation of activated carbons from a low-rank lignite. (8) Bacon, R. In Chemistry and Physics of Carbon; Walker, P. L., Jr., Thrawer, P. A., Eds.; Marcel Dekker: New York, 1973; Vol. 9, pp 1-95. (9) Oberlin, A. Carbon 1984, 22, 521.
10.1021/ef0301809 CCC: $27.50 © 2004 American Chemical Society Published on Web 05/01/2004
884
Energy & Fuels, Vol. 18, No. 3, 2004
Sakintuna et al.
Table 1. Proximate and Elemental Analyses of Elbistan Lignite proximate analysis
%, dry
volatiles fixed carbon Ash
44.8 20.9 34.3
elemental analysis
%, dmmf
C H N S O (by difference)
53.0 5.8 1.8 3.6 35.8
Table 2. Elemental Carbon Analyses of Activated Carbon Samples Obtained at Different Temperatures (%, dmmf) activation temperature, °C % C, dmmf
700 58.82
800 44.99
900 47.23
1000 45.15
Experimental Section Materials. Elbistan lignite was used in the study. Analysis of the Elbistan lignite is presented in Table 1. The lignite sample was ground under a nitrogen atmosphere to 100 µm size and stored under nitrogen. The lignite sample was demineralized with HCl and HF by standard methods.10 Two liters of 6 N HCl was added to 200 g of coal. The slurry was stirred for 24 h under a nitrogen atmosphere, then it was filtered and washed with distilled water until the filtrate became neutral. Consecutively, 1.6 L of aqueous (40%) HF was added to HCl-washed coal and this mixture was stirred for ∼24 h under a nitrogen atmosphere. After filtering, the demineralized coal was washed with 1 L of distilled water and dried at 100 °C for 24 h under a nitrogen atmosphere. Pyrolysis Experiments. Raw, HCl- and HCl-HF treated coal samples were dried at 100 °C under an inert atmosphere. An 8.00 g sample was placed into a porcelain crucible and then placed into a furnace purged with ultrahigh-purity nitrogen. Coal samples were heated to four different temperaturess700, 800, 900, and 1000 °Cswith a heating rate of 10 °C/min, under a nitrogen flow of 100 mL/min for 120 min. After the pyrolysis experiments, chars were cooled to room temperature under the nitrogen flow. Activation Experiments. Chars were activated under a carbon dioxide flow of 100 mL/min at the final pyrolysis temperature for an additional 2 h. The system was cooled to room temperature under the nitrogen flow. After that, carbons were taken out from the system. Chars obtained from pyrolysis experiments and carbons from activation experiments were stored for physical characterization tests. Elemental carbon analyses of the activated carbon samples obtained at different temperatures are presented in Table 2. Surface Analysis. Surface areas of activated carbons were measured by an ASAP2000 Accelerated Surface Area and Porosimetry system manufactured by Micromeritics Co., USA, using nitrogen gas. The surface area of the samples was determined by using a BET equation in the relative pressure range of between 0.05 and 0.25, over five adsorption points. All the BET areas measured in the present study as m2/g sample were converted to BET areas as m2/g carbon values using the carbon percentages given in Tables 1 and 2. XRD Measurements. XRD measurements of the carbonized and activated product samples were done with a Bruker axs advance powder diffractometer fitted with a Siemens X-ray gun and equipped with Bruker axs Diffrac PLUS software. The sample was rotated (15 rpm) and swept from 2θ ) 10° through to 90° using default parameters of the program. The X-ray generator was set to 40 kV at 40 mA. All the XRD measure(10) Yu¨ru¨m, Y.; Kramer, R.; Levy, M. Thermochim. Acta 1985, 94, 285.
Figure 2. X-ray diffraction pattern of raw Elbistan lignite. Q: quartz, C: calcite, G: gypsum, P: pyrite, CL: clay minerals, K: kaolinite, Ar: aragonite, F: feldspar, Cr: cristobalite. ments were repeated at least three times, and the results reported were the average of these measurements. The XRD patterns were analyzed for the structural parameters using the classical Debye-Scherer equations:
Lc ) 0.90 λ/β002 cos θ002 La ) 1.94 λ/β100/101 cos θ100/101 n ) Lc/d002 where β represents full-width at half maximum, fwhm (in radians of θ), and n is the number of graphene sheets. The peak positions of the (002) peak were measured and Bragg’s Law was used to calculate the interlayer spacing d002. Lc, La, and d002 are indicated in Figure 1. The full widths were calculated at the half-maxima (fwhm) with the Bruker axs Diffrac PLUS software provided with the Bruker axs advance powder diffractometer of the peak positions of (002) and (100)/ (101) peaks.
Results and Discussion The minerals identified by XRD in the raw Elbistan lignite sample were calcite and quartz in abundant quantities, gypsum, pyrite, kaolinite, and other clay minerals, aragonite, feldspar, and cristobalite in minor quantities, Figure 2. Quartz, gypsum, calcite, clay minerals, and greenalite were observed in lignite used for activated carbon production for environmental purposes by Skodras et al.11 The XRD of HCl-washed and HCl/HF-washed lignite samples are presented in Figure 3. The XRD traces of the acid-washed lignite samples contained fewer peaks than that of the raw lignite sample. The missing peaks in Figure 3b belonged to the carbonate minerals (calcite, aragonite) and sulfate minerals (gypsum) originally present in the raw lignite that were leached by HCl-washing. HF was used to dissolve the silicate minerals and peaks attributed to silicate minerals (quartz, kaolinite and other clay minerals, feldspar, and cristoballite) are absent in the XRD of HCl/HF-washed lignite sample, Figure 3c. HCl and HF washings did not dissolve the pyrite minerals, therefore the only type of minerals present in the HCl/ (11) Skodras, G.; Orfanoudaki, Th.; Kakaras, E.; Sakellaropoulos, G. P. Fuel Process. Technol. 2002, 77-78, 75.
Evolution of C Microstructures in Pyrolysis of Lignite
Figure 3. X-ray diffraction patterns of (a) raw, (b) HCl-, and (c) HCl/HF-washed lignite samples.
Figure 4. Change of BET areas (m2/g carbon) of chars obtained from the pyrolysis of raw, HCl-, and HCl/HF-washed lignite samples with pyrolysis temperature.
Figure 5. Change of BET areas (m2/g carbon) of carbons obtained after the activation of chars of raw, HCl-, and HCl/ HF-washed lignite samples with activation temperature.
HF-washed sample was pyrite minerals, and peaks due to this mineral are easily identified in Figure 3c near 2θ ) 33°, 37°, and 56°. The BET area of raw, HCl- and HCl/HF-washed Elbistan lignite samples were calculated as 7.0 m2/g carbon, 7.0 m2/g carbon, and 4.5 m2/g carbon, respectively. HCl-washing dissolved minerals such as carbonates, sulfates, etc., and further treatment with HF eliminated all silicate minerals. The surface area of the samples seemed to stay unchanged with acid washings. The effect of temperature of pyrolysis and activation experiments on the BET areas of the HCl-washed and HCl/HF-washed samples are presented in Figures 4 and 5. The pyrolysis and activation reactions increased the BET areas of the acid-washed samples almost 10-fold,
Energy & Fuels, Vol. 18, No. 3, 2004 885
(250-2608 m2/g carbon). The rapid increase in the BET areas might be attributed to the widening of existing pores or the creation of new pores. Although no effects of acid washings were observed in the BET areas of nonthermally treated samples, BET areas of the acidwashed samples increased significantly after pyrolysis and activation experiments. Removal of the alkali and alkaline earth metal cations and silicates present in the raw lignite might have loosened the structure so that the volatiles that formed during the pyrolysis and the burn-out of carbon during activation reactions might have opened micropores in the structure of the chars and carbons while diffusing out from the inner zones. The increase of the BET areas parallel to the raising of the temperature of pyrolysis or activation experiments from 700 °C to 1000 °C could be explained with the loss of volatile matter which led to the development of porosity (Figures 4 and 5). This effect was amplified in the case of HCl/HF-washed lignite samples, the BET of these samples increased to 749 m2/g carbon at 700 °C, and further increases were observed at higher temperatures: 1098 m2/g carbon (800 °C), 1429 m2/g carbon (900 °C), and 2608 m2/g carbon (1000 °C), Figure 5. This showed that the chars obtained from HCl/HF-washed lignite might have possessed higher adsorption capacity than raw and HCl-washed chars. This might be an indication to the development of more pores in this type of samples. On the other hand the BET areas of the non acidwashed samples decreased when the temperature of pyrolysis and activation experiments were raised from 700 °C (363 m2/g carbon) to 1000 °C (46 m2/g carbon), Figure 5. Mixed oxides such as K2O-Na2O-Al2O3SiO2-H2O, CaO-Al2O3-SiO2-H2O, and K2O-Na2OFeO-MgO-Al2O3-SiO2-H2O have melting liquids in their phase diagrams at atmospheric pressures in temperature ranges of 450-700 °C, 400-750 °C, and 640-750 °C, respectively.12 The reason for this decrease in the BET areas (m2/g carbon) might be the melting of the minerals, containing similar oxides, originally present in the raw lignite sample, at temperatures between 700 °C and 1000 °C during pyrolysis and activation experiments and probably forming glassy structures which probably clogged the porosity in the carbons formed. The XRD trace of HCl/HF-washed lignite sample (Figure 3) also showed the presence of a broad peak at 2θ ) 22°, corresponding to (002) reflection of carbon due to the stacking structure of aromatic layers.13 Broadening of the (002) peak is interpreted in terms of the presence of crystallites perpendicular to aromatic layers. The interlayer spacings, d002, of chars and activated carbons obtained in the present work are presented in Table 2. The d002 results reported were the average of at least three measurements. d002 values seemed to change in the range of 0.35-0.39 nm. Similar values of d002 were also reported recently.14 d002 values measured by Sharma et al.14 for carbons produced at 1000 °C and 1400 °C were 0.38 and 0.37 nm, respectively. However, a correlation between the d002 values and pyrolysis and (12) Phase Diagrams for Ceramists, Vol. VIII; Mysen, O. B., Ed.; The American Ceramic Society Inc., Washington, DC, 1990; pp 7982, 7991, 8025. (13) Yoshizawa, N.; Maruyama, K.; Yamada, Y.; Zielinska-Blajet, M. Fuel 2000, 79, 1461. (14) Sharma, A.; Kyotani, T.; Tomita, A. Carbon 2000, 38, 1977.
886
Energy & Fuels, Vol. 18, No. 3, 2004
Figure 6. Change of X-ray diffraction patterns of the chars of carbonized HCl/HF-washed lignite samples with temperature.
activation temperatures could not be done; this was due to relatively low temperatures utilized in the present work. In particular, a carbon that is thermally treated at temperatures below 1800 °C is generally very disordered; it contains turbostratic (fully disordered) structures. The reasons for this disorder are the presence of local stacking faults, random shifts between adjacent layers, variable interspacing values, unorganized carbons that are not a part of layer structure, and strain in the layers.15 The results presented for d002 in the present work can be considered as indications to the presence of turbostratic (fully disordered) structures. The d002 values are typical of the turbostratic structure of carbon; it is also interesting to note that in a recent report,16 while the development of turbostratic structures in untreated cellulose took place above 1200 °C, in the present study using a lignite as a carbon precursor, turbostratic structures started to appear at a much lower temperature range of 700-1000 °C. XRD patterns of the chars obtained from the pyrolysis of the HCl/HF-washed samples at 700 °C-1000 °C are presented in Figure 6. These XRD pattern were analyzed to obtain the structural parameters of the chars obtained after pyrolysis reactions and carbons obtained after activation reactions. The XRD pattern of the char obtained from HCl/HF-washed sample at 700 °C is presented to show the calculation of fwhm value utilizing the Bruker axs Diffrac PLUS software provided with the Bruker axs advance powder diffractometer in Figure 7. Lc and La calculated using Debye-Scherer equations are not exactly equal to the stacking height and lateral size of the crystallites because these equations in fact can actually be derived for highly graphitized carbons and are not suitable for turbostratic (fully disordered) carbons. Therefore, these can be used as convenient relative estimates of actual stacking height and lateral size of the crystallites.17 The actual crystallite sizes, therefore, are likely to be slightly greater than the calculated values. (15) Babu, V. S.; Seehra, M. S. Carbon 1996, 34, 1259. (16) Oberlin, A.; Bonnamy, S.; Lafdi, K. Structure and texture of carbon fibers. In Carbon Fibers, 3rd ed.; Donnet, J.-B., Wang, T. K., Reboyuillat, S., Peng, J. C. M., Eds.; Marcel Dekker: New York, 1998; Chapter 2. (17) Gurudatt, K.; Tripathi, V. S. Carbon 1998, 36, 1371.
Sakintuna et al.
Figure 7. Method to calculate fwhm data (from the X-ray diffraction pattern of the char obtained from the pyrolysis of HCl/HF-washed lignite sample at 700 °C).
Figure 8. Change of fwhm and Lc values of chars obtained from the pyrolysis of HCl/HF-washed lignite samples with temperature.
Figure 9. Change of average number of graphene sheets of chars obtained from the pyrolysis of HCl/HF-washed lignite samples with temperature.
The change of the calculated values of the stacking height, Lc, of HCl/HF-washed samples with temperature in pyrolysis experiments is presented in Figure 8. The values of Lc of HCl/HF-washed samples seemed to increase from 1.0 to 1.5 nm when the pyrolysis temperature was increased from 700 °C to 1000 °C. In general terms, the results indicated that Lc increased monotonically with an increase in the temperature of pyrolysis. The change of the average number of graphene sheets in relation to the values of Lc is presented in Figure 9. The average number of graphene sheets increased from 2.8 to 4.4 with the increase of pyrolysis temperature
Evolution of C Microstructures in Pyrolysis of Lignite
Figure 10. Change of fwhm and La values of chars obtained from the pyrolysis of HCl/HF-washed lignite samples with temperature.
from 700 °C to 1000 °C. The lateral size of the crystallites, La, increased very faintly from 5.0 to 5.5 nm when the pyrolysis temperature was increased from 700 °C to 1000 °C, Figure 10. These results indicated that lateral size of the crystallites of the chars did not change significantly within the pyrolysis temperature range studied, but the stacking height and in relation with this the number of graphene sheets per stack increased about 60% within the same temperature range. Kercher and Nagle18 reported that, with increased pyrolysis temperature, the crystallite boundary area decreased from turbostratic crystallite growth and the number of graphene sheets increased. The reason for the increase in the stacking height could be related to heat treatment temperature only. Fujimoto and Shiraishi19 observed that above 1000 °C continuing growth of the carbon layer plane might be due to the increase of van der Waals forces between the layers, which resulted in an increases of the stacking number, hence the coalescence of the units was probably accelerated also at temperatures between 700 °C and 1000 °C, as it was also observed in the present work. Feng et al.20 stated that La and Lc could be seen to increase slightly through the heat treatment at 1150 °C for heating times up to 700 min, while the interlayer spacing d002 was essentially unchanged during heat treatment. It is also suggested that the impurities in the raw coal char enhanced the structural ordering during heat treatment.20 The samples investigated in the present work were all mineral-free, and therefore the effect of minerals on the size of crystallites was not studied in the present work. The fwhm of (002) peaks in the XRD patterns of the chars obtained in pyrolysis experiments stayed relatively constant (Figure 10) from 700 °C to 1000 °C, but the intensity of the (002) peak decreased over the same temperature range (Table 2). This observation was not parallel to those of Kercher and Nagle18 who found an increase in the intensity of (002) peak as the temperature of pyrolysis was increased from 500 °C to 1400 °C. They attributed this increase to an increased amount of carbon from large turbostratic crystallites; therefore the decrease of the intensity of the (002) peaks in the (18) Kercher, A. K.; Nagle, D. C. Carbon 2003, 41, 15. (19) Fujimoto, H.; Shiraishi, M. Carbon 2001, 39, 1753. (20) Feng, B.; Bhatia, S. K.; Barry, J. C. Carbon 2002, 40, 481.
Energy & Fuels, Vol. 18, No. 3, 2004 887
Figure 11. Change of fwhm and Lc values of carbons obtained from the activation of chars of HCl/HF-washed lignite samples with temperature.
Figure 12. Change of the average number of graphene sheets of carbons obtained from the activation of chars of HCl/HFwashed lignite samples with temperature.
present report could be explained by the loss of amounts of carbon from large turbostratic crystallites which is quite evident that as the temperature is increased carbon losses are expected to increase in pyrolysis reactions. Crystallite growth involves mass transport processes and may be due to either motion and mergence of layers coming in contact or atomic diffusion.21 The latter mechanism, involving the detachment of carbon atoms from layers, is more likely at temperatures higher than 1000 °C and the slight increase in the values of La in the present work could be explained by the mergence of layers coming in contact. Activation reactions performed at the same temperature range did not change the stacking heights. The values of Lc for activated HCl/HF-washed samples stayed almost constant in the same range as for the carbonized samples within 1.0-1.5 nm, Figure 11. The reason the stacking heights of the HCl/HF-washed coal does not change after activation at different temperature is that the stacking height only reduces at high conversion levels (above 75%) during gasification. All the carbons tested in the present work might have conversion levels lower than that. The cause of the decrease of the stacking height at high conversion only is that the gasification occurs from the edge of the crystallites and moves inward. The values of the average number of graphene sheets of activated samples are also similar to those of carbonized samples, Figure 12. These results indicated that activation reactions in the same temperature range did not have any effect on the values of Lc and the number of graphene sheets. This indicated that oxidative reactions during activation did not alter (21) Belenkov, E. A. Inorg. Mater. 2001, 37, 928.
888
Energy & Fuels, Vol. 18, No. 3, 2004
the stacking heights of the crystallites significantly in the temperature range of 700-1000 °C. Since 2θ ) 43 °C peak was missing in all of the XRD patterns of activated samples, the values of the lateral size of the crystallites, La, could not be calculated. Conclusions (a) BET areas (m2/g carbon) of the acid-washed samples increased significantly after pyrolysis and activation experiments when compared with those of the raw lignite samples. Removal of the alkali and alkaline earth metal cations and silicates which constituted 34% of the lignite might have loosened the structure so that the volatiles that formed during the pyrolysis and activation reactions might have opened micropores in the structure of the chars and carbons while diffusing out from the inner zones. (b) The minerals identified by XRD in the raw Elbistan lignite sample were abundant calcite and quartz quantities, minor quantities of gypsum, pyrite, kaolinite, and other clay minerals, aragonite, feldspar, and cristobalite. (c) The BET areas (m2/g carbon) of the non acidwashed samples decreased when the temperature of pyrolysis and activation experiments were raised from 700 °C to 1000 °C. The reason for this decrease in the BET areas was the decomposition and melting and of the minerals originally present in the raw lignite sample at temperatures between 700 °C and 1000 °C during pyrolysis and activation experiments and covering of the porosities by the glassy material formed. (d) The interlayer spacing, d002, in chars and activated carbons seemed to stay constant in the range of 0.350.39 nm, and broadening of the (002) peak is interpreted in terms of small dimensions of crystallites perpendicular to aromatic layers. The results presented in this work might be considered as indications for the presence
Sakintuna et al. Table 3. Change of Interlayer Spacing with Pyrolysis and Activation Temperature
sample
d002, nm
relative intensity of (002) peak
HCl/HF-washed raw HCl/HF-washed carbonized at 700 °C HCl/HF-washed carbonized at 800 °C HCl/HF-washed carbonized at 900 °C HCl/HF-washed carbonized at 1000 °C
0.35 0.37 0.35 0.39 0.35
1.0 1.0 0.9 0.7 0.7
of turbostratic (fully disordered) structures in the temperature range of 700-1000 °C. (e) Lc values of the chars obtained in pyrolysis experiments increased monotonically with the increase in the temperature of pyrolysis. The average number of graphene sheets also increased in the temperature range 700 °C-1000 °C. The lateral size of the crystallites, La, increased very faintly from 5.0 to 5.5 nm when the pyrolysis temperature was increased from 700 °C to 1000 °C. These results indicated that lateral size of the crystallites of the chars did not change significantly within the pyrolysis temperature range studied but the stacking height and in relation with this the number of graphene sheets per stack increased about 60% within the same temperature range. (f) Activation reactions performed at the same temperature range did not change the stacking heights. The values of Lc for activated HCl/HF-washed samples stayed almost constant in the same range as for the carbonized samples within 1.0-1.5 nm. The values of the average number of graphene sheets of activated samples are also similar to those of carbonized samples. These results indicated that oxidative reactions during activation reactions in the same temperature range 700-1000 °C did not have any effect on the values of Lc and the number of graphene sheets. EF0301809