Air Reactivity of Petroleum Cokes - ACS Publications - American

Nov 24, 2006 - Rio Tinto Aluminium, Thomastown, VIC 3074, Australia ... The apparent activation energy of the coke-air reaction derived from the extra...
0 downloads 0 Views 504KB Size
Ind. Eng. Chem. Res. 2007, 46, 3265-3274

3265

Air Reactivity of Petroleum Cokes: Role of Inaccessible Porosity Kien N. Tran and Suresh K. Bhatia* DiVision of Chemical Engineering, The UniVersity of Queensland, St. Lucia, QLD 4072, Australia

Alan Tomsett Rio Tinto Aluminium, Thomastown, VIC 3074, Australia

This paper presents a detailed study of the air reactivity of petroleum cokes measured at temperatures between 400 and 600 °C using a combination of characterization techniques and reactivity measurements. The microstructure of the coke was found to comprise an essentially inaccessible pore system at low temperatures of 77-273 K used in characterization, and it is more accessible to oxygen at higher temperatures of about 773 K used in oxidation. The correlation of reactivity data using the random pore model suggests that the true micropore area is significantly larger than that measured using physical gas adsorption methods. The difference in surface area can be attributed to the low kinetic energy of gas molecules at the lower temperatures of characterization; as a result, they are unable to overcome the pore mouth energy barrier. By examining the variation of coke structure with burnoff level, we find that most of the internal reaction occurs in pores in the narrow pore width range 1-2 nm. For pores greater than 2 nm, however, the surface areas remains essentially constant with burnoff level. The apparent activation energy of the coke-air reaction derived from the extracted rate constants falls in the range 145-160 kJ/mol. 1. Introduction The air reactivity of petroleum coke is known to depend on the microstructure and impurity content.1-7 Hume has shown that metal impurities such as sodium, vanadium, iron, and nickel strongly catalyze the coke-air reaction, while sulfur decreases the reactivity by binding with the free sodium to form inactive complexes.4 On the other hand, Rey Boero6 and Tyler7 showed the air reactivity is directly proportional to the available surface area of the coke. However, these studies did not distinguish between oxygen attacks on different parts of the internal surface of the coke structure during the gasification process. In a study of coal char gasification with air, Feng and Bhatia found that the micropore surfaces, particularly in pores smaller than 1 nm, are underutilized during the gasification process.8 This behavior was partly attributed to blockage of reactive edge sites by functional groups or molecules composed of disorganized matter that cross-links the crystallites. These smaller micropores consist of stable basal plane sites in crystallites, which are essentially unreactive.9 This is not the case for micropores greater than 1 nm, which are more likely to be intercrystallite spaces composed largely of reactive edge sites. There is some literature that details the modeling of the impact of coke structure on reactivity.6 The standard air reactivity measurement employed by the aluminum smelting industry measures the ignition temperature of the coke, then converts this temperature to an air reactivity value expressed in percentage per minute.4,5 This method provides no information about the relationship between the structure and reactivity. The aim of this study was to investigate air reactivity of high sulfur petroleum cokes within a temperature range 400-600 °C. A number of different experimental techniques were employed to characterize the structure of reacted and unreacted cokes. The results show that the pore network is initially largely inaccessible, a problem which is overcome after about 5% * To whom correspondence should be addressed. Tel: +61-7-33654263. Fax: +61-7-3365-4199. E-mail: [email protected].

conversion of the solid (also termed burnoff). In addition, the random pore model10 is applied to correlate the reactivity with the microstructure of the coke, and it yields a consistent interpretation once the pore network becomes accessible (conversion >5%). 2. Experimental Section 2.1. Materials. A high sulfur petroleum coke (HSC) (∼3 wt % S) was used. The coke was calcined at five different temperatures, 1000, 1100, 1150, 1225, and 1300 °C, using a two-step laboratory calcination process. The green coke was first calcined using a rotary kiln to 900 °C and then flash calcined using a high-temperature oven at the final heat treatment temperature for 30 min. The physical and chemical properties of these five coke samples are listed in Table 1. Each coke sample was named as HSC-temperature. For example, HSC-1000 stands for high sulfur coke calcined at 1000 °C. In addition, a high sulfur industrial coke calcined in a gas-fired rotary kiln to 1225 °C was also included in the study. This industrial coke sample was named HSC-1225p. Sample HSC1225p was crushed and sieved into eight different particle size fractions, 355-500, 250-355, 212-250, 180-212, 106-180, 90-106, 75-90, and 45-75 µm, to study the variation of reactivity with particle size as well as the variation of structure with burnoff level. The particle size range 180-212 µm was used for the air reactivity measurements of the laboratory calcined cokes. 2.2. Air Reactivity Measurements. Reactivity measurements were conducted using a SETARAM (SETSYS 16/18) thermogravimetric analyzer (TGA). In each experiment, approximately 1-5 mg of sample was used, spread evenly as a single layer in a crucible. The reaction experiments were conducted under isothermal conditions in flowing air at 100 mL/min. During the heating period (20 K/min), the sample was protected from oxidation using high purity nitrogen. When the desired temperature was reached, the sample was held at temperature for 10 min and

10.1021/ie061017e CCC: $37.00 © 2007 American Chemical Society Published on Web 11/24/2006

3266

Ind. Eng. Chem. Res., Vol. 46, No. 10, 2007

Table 1. Physical and Chemical Properties of All Petroleum Coke Samples sample property

HSC-1000

HSC-1100

HSC-1150

HSC-1225

HSC-1300

HSC-1225p

3.18 0.95 0.13 95.7 50 nm

0.20 0.33 0.010

Argon Area (m2/g) 0.23 0.22 0.007

0.24 0.31 0.008

0.15 0.27 0.015

0.16 0.27 0.018

H < 2 nm 2 nm < H < 50 nm H > 50 nm argon BET area (m2/g)

15 111 40 1.66

Argon Volume (×105 cm3/g) 15 87 27 1.44

14 116 34 1.19

9 129 60 1.18

9 146 66 1.24

g d002 (nm) Lc (nm) La (nm) Me (nm)

0.61 0.342 4.39 1.53 13.0

0.67 0.344 4.07 2.40 11.9

0.66 0.344 4.28 2.53 12.5

0.65 0.344 5.17 2.65 15.0

CO2 Volume (×105 cm3/g) 8 3 0 0 8 3

XRD Analysis 0.68 0.344 3.38 2.10 9.8

of sources. On the basis of these data, the values of ks differ by a factor of 0.4-3.5 relative to the value given by Tyler. This means that the surface area computed above would be expected to fall in the range 0.2-2.5 m2/g. This range of surface area is still well within the range of surface area reported in the litereature1-7 as well as greater than the values obtained from adsorption measurement. This further supports our argument that the accessibility of gas molecules is temperature dependent. Bae and Bhatia14 carried out argon adsorption isotherm measurements for coals at different temperatures. They found that argon adsorption was substantially higher at a high temperature (313 K). At the same time, molecular simulations performed by Nguyen and Bhatia15 show that it is possible for adsorbate molecules to enter an inaccessible or apparently blocked pore, provided that the adsorbate molecule has enough kinetic energy to overcome the energy barrier at the entrance of a blocked pore. Since the gasification reaction of coke was carried out at relatively high temperatures, the oxygen molecules appear to have enough kinetic energy to overcome the potential barrier of the apparent blocked micropores. Thus, argon and CO2 encounter problems of inaccessibility at the low temperatures of 87 and 273 K, respectively, and yield low apparent surface areas and pore volumes. On the other hand, at the high temperature reaction of 773 K, oxygen does not encounter this difficulty and accesses the whole reactive surface of the carbon. This explains why the apparent true micropore surface area based on oxygen reactivity is much greater than the measured surface area. The argon BET area closely matches the apparent true micropore area estimated from the reactivity data. This could be coincidental since most of the total surface area calculated using the DFT method is associated with the macropore region as shown in Tables 2 and 3. Therefore, the BET area mainly comprises the surface area in the macropore region as well as the external surface area. 3.6. Effects of Calcination Levels on Air Reactivity. The reactivity of the laboratory calcined coke samples HSC-1000, HSC-1100, HSC-1150, HSC-1225, and HSC-1300 were measured in air at three different temperatures, 600, 540, 470 °C, using the particle size fraction 180-212 µm. The conversion-

time and reaction rate evolution data for all the laboratory calcined coke samples are shown in Figure 9. At all reaction temperatures, the reactivity is highest for the lowest calcination temperature coke and decreases with increasing calcination temperature. Oberlin has carried out extensive study of the carbonization and graphitization of many carbonaceous materials, including petroleum coke.17 She found the development of the crystallite structure of carbon is consistent across a wide range of carbonaceous materials. For example, during the temperature range 700-1300 °C, the interlayer defects in the basic structural units of carbon reduce and the crystallites start to grow larger, with the largest crystallites at the highest calcination temperature. The XRD analysis, particularly the crystallite sizes, Lc and La (Table 3), confirm the crystallite growth with increasing calcination, except for the material calcined at 1000 °C. This latter deviation was consistently obtained when the XRD test on sample HSC-1000 was repeated several times, and is most likely related to the nature of the Shi et al.20,21 algorithm, which provides the size of pseudocrystallites larger than those considered in the Sherrer equation. As a result of the crystallite growth and layer realignment, the specific accessible reactive crystallite edge area and the overall number of active sites decreases with increasing calcination temperature. This process is known as thermal deactivation. The reaction rates for all laboratory calcined coke samples are also shown in Figure 9, depicting a rate maximum at a conversion of 0.4, which is consistent with the industrial coke sample HSC-1225p. Similar to the industrial coke case, the conversion-time data were correlated using the random pore model. The correlation involved simultaneous fits of the conversion-time data at all three reaction temperatures, for a given calcination temperature. This permitted determination of a single ψ value for that calcination temperature, while also providing estimates of the values of ke at the three reaction temperatures. Since experimental data of the variation of pore structure with burnoff level for the laboratory calcined cokes was not obtained, a simple correlation study using the random pore model was carried out to determine the conversion level at which the

3272

Ind. Eng. Chem. Res., Vol. 46, No. 10, 2007

Figure 9. Air reactivity data for all laboratory calcined coke samples measured at (a, b) 600 °C, (c, d) 540 °C, and (e, f) 470 °C. The figures on the left-hand side show the time-conversion data while those on the right-hand side show the corresponding reaction rate data.

majority of the micropore inaccessibility is cleared. For the calculation procedure, the original conversion-time data were corrected to consider the solid at an arbitrary conversion as the initial solid. Conversions in the range from 1% to 20% were chosen for the initial solid, and the variation of the fitted value of ke with the shift in conversion, Xshift, was determined. Xshift is defined as the conversion at the point at which the solid is free of inaccessible porosity. Then, the corrected conversion is given by eq 5. A somewhat similar procedure has previously been used by Aarna and Suuberg27

Xnew )

Xold - Xshift 1 - Xshift

(5)

where Xold lies between Xshift and unity. The results for sample HSC-1150 are plotted in Figure 10. For this, the data for all the three temperatures (460, 500, 540 °C) were fitted simultaneously, yielding a common value of ψ and the value of ke at each temperature. It is seen that ke shows a large increase when the initial solid lies in the 0-5% conversion range, and then gradually levels out at the higher shifts in conversion. Similar results were obtained for the remaining laboratory calcined coke samples. The relatively constant value of ke for a shift beyond about 5% conversion in the initial solid indicates that most of the micropore inaccessibility is cleared by 5% conversion at all calcination temperatures. This is consistent with the results for the industrial calcined coke sample HSC-1225p, based on the helium densities.

Figure 10. Variation of effective rate constant ke with shift in initial solid to conversion Xshift, at various temperatures, for sample HSC-1150.

On the basis of the above results, the reactivity data of all laboratory cokes were corrected for 5% shift conversion and fitted using the random pore model. The extracted effective rate constant, structural parameter, and apparent activation energy are summarized in Table 4. The activation energy was evaluated from the rate constants following an Arrhenius-type relationship:

(

ke ) A0 exp -

E RT

)

(6)

Here, E is the apparent activation energy (J/mol), R is the gas constant (8.314 J/(mol K)), and T is the reaction temperature (K). Figure 11 depicts an Arrhenius plot of the rate constants, confirming a linear relationship for each sample. For comparison

Ind. Eng. Chem. Res., Vol. 46, No. 10, 2007 3273 Table 4. Rate Constants, Pore Structure Parameters, and Activation Energies of Laboratory Calcined Cokes Based on Correlation with Xshift ) 5% sample model parameters ke (min-1) 600 °C ke (min-1) 540 °C ke (min-1) 470 °C y activation energy, E (kJ/mol)

HSC-1000

HSC-1100

HSC-1150

HSC-1225

HSC-1300

0.0663 0.0134 0.0019 5 147

0.0509 0.0123 0.0015 2 147

0.0403 0.0093 0.0011 4 151

0.0203 0.0045 0.0006 19 149

0.0121 0.0026 0.0003 56 154

purpose, the rate constant of sample HSC-1225p is also shown in Figure 11. The industrial coke lies on the same line of the laboratory calcined coke of similar heat treatment temperature indicating that the laboratory calcination process adequately mimics the industrial process. The observed correlation suggests the coke reactivity is a strong function of calcination temperature. From Table 4, the activation energy of the laboratory calcined coke falls in the range 145-160 kJ/mol. This is in agreement with the reported values of Rey Boero6 and Tyler.7 A slight increase in activation energy with calcination temperature is noted, which may reflect the effect of thermal deactivation and increased graphitization. 3.7. Effect of Calcination Temperature on Structure. Using the same technique for estimating the true micropore surface area for the industrial coke sample HSC-1225p, the intrinsic rate constant reported by Tyler7 was used to estimate the true micropore surface area of all laboratory calcined coke samples from the effective rate constants computed in the previous section. Figure 12 depicts the variation of the apparent micropore surface area and structural parameter ψ, with calcination temperature for all laboratory calcined cokes. The micropore area plotted in Figure 12a is the arithmetic average of all three surface areas obtained from the three different reaction temperatures, considering the small difference in activation energy between that obtained here and the value reported by Tyler.7 The micropore area shows a decreasing trend with increasing calcination temperature. The decrease in surface area is most likely due to the decrease in the microporosity as a result of the increase in graphitization level of the coke microtexture at higher calcination temperatures. The decrease in porosity is also reflected by the apparent increase of ψ with calcination temperature as shown in Figure 12b. A large ψ can be readily seen to be associated with low porosity, following the definition of this parameter. From Table 3, the laboratory calcined coke has a highly graphitized carbon structure with approximately 60-70% of organized carbon structure. The reactive micropore area is believed to be the void volume between carbon crystallites and imperfections in crystallites. With an increase in calcination

Figure 11. Arrhenius plots of the effective rate constant for the coke-air reaction.

temperature, the imperfections in crystallites decrease. Also, growth in crystallite size occurs, as is evident in the values of Lc and La, which effectively reduces the void volume between the carbon crystallites. Hence, the observed decrease in micropore surface area and increasing value of structural parameter, ψ, with increasing calcination temperature is expected. 4. Conclusions The air reactivity of calcined petroleum cokes has been studied in the temperature range between 400 and 600 °C. At all reaction temperatures, the coke reactivity decreases with increasing calcination temperature. This is due to the thermal deactivation of active sites. A number of important observations were derived from the correlation study of the reactivity data using the random pore model. The microstructure of the coke appears to be initially largely inaccessible to gas molecules, and the micropore blockages are cleared by about 5% conversion. In addition, the micropore surface area estimated from the reactivity data using the random pore model suggests that the accessibility of gas molecules into the micropore structure is strongly dependent on the temperature of the reacting/adsorbing molecule. The accessibility of gas molecules is highest for oxygen at the high reaction temperature 400-600 °C, and decreases for CO2 at 273 K and argon at 87 K; in that order. Characterization of partially oxidized coke samples provided additional insight into the evolution of the internal structure

Figure 12. Variation of (a) average apparent micropore surface area and (b) structural parameter ψ, with calcination temperature for laboratory calcined cokes.

3274

Ind. Eng. Chem. Res., Vol. 46, No. 10, 2007

during the coke-air reaction. The reaction of coke with air initially occurs at the external surface of the grains and quickly proceeds into the internal pore structure. The internal reaction progresses gradually with an increase in conversion, with most of the internal reaction concentrated in the narrow pore size range 1-2 nm. At a high conversion level, most of the internal structure has been gasified leading to significant loss of connectivity in the coke structure. In addition, our results help to clarify the observation of the reactivity of petroleum coke as observed by industrial researchers.1-4 Hume4 noted that certain types of petroleum coke showed a marked increase in reactivity over an initial activation period, spanning a conversion range up to 15%. Our results indicate that this activation period represents the time required to open the inaccessible porosity of the coke. The new understanding has significance in industrial applications involving coke reactions, such as carbon anodes in aluminum smelting and blast furnaces. Acknowledgment The financial support of this research provided by the Australian Research Council and by Rio Tinto Aluminium is gratefully acknowledged. Notation A0 ) pre-exponential factor of Arrhenius equation CA0 ) initial concentration of gaseous reactant d002 ) average interlayer spacing of crystallite determined from XRD analysis g ) fraction of organized carbon of the carbon phase H ) slit pore width ke ) effective rate constant, ksS0/Fc(1 - 0) ks ) intrinsic rate constant for surface reaction La ) average crystallite width determined from XRD analysis Lc ) average crystallite height determined from XRD analysis L0 ) length of overlapped system per unit volume at t ) 0 Me ) average number of layer per crystallite determined from XRD analysis n ) reaction order with respect to the reactant gas R0 ) initial particle radius S0 ) reaction surface area per unit volume at t ) 0 t ) time T ) temperature x ) overall conversion Greek Letters 0 ) initial porosity ψ ) structural parameter, 4πL0(1 - 0)/S02 Fc ) true density of carbon skeleton τ ) dimensionless time, ksCA0nS0t/(1 - 0)

(2) Fischer, W. K.; Perruchoud, R. C. Influence of coke calcination parameters on petroleum coke quality. In Anodes for the Aluminium Industry; Fischer, W. K., Mannweiler, W., Keller, F., Perruchoud, R. C., Buhler, U., Eds.; R & D Carbon Limited: Sierre, Switzerland, 1995; pp 21-37. (3) Hume, S. M. Ph.D. Thesis, University of Auckland, Auckland, New Zealand, 1993. (4) Hume, S. M.; Fischer, W. K.; Perruchoud, R. C. A model for petroleum coke reactivity. Light Met. 1993, 525-531. (5) Milani, F.; Melendi, J. Air reactivity studies on calcined petroleum coke. Light Met. 1992, 641-647. (6) Rey Boero, J. F. The reaction of petroleum cokes with air. Carbon 1987, 25, 477-483. (7) Tyler, R. J. Intrinsic reactivity of petroleum coke to oxygen. Fuel 1986, 65, 235-240. (8) Feng, B.; Bhatia, S. K. Variation of pore structure of coal chars during gasification. Carbon 2003, 41, 507-523. (9) Hurt, R. D.; Dudek, D. R.; Longwell, J. P.; Sarofim, A. F. The phenomenon of gasification-induced carbon densification and its influence on pore structure evolution. Carbon 1988, 26, 433-449. (10) Bhatia, S. K.; Perlmutter, D. D. A random pore model for fluidsolid reactions: I. Isothermal, kinetic control. AIChE J. 1980, 26, 379386. (11) Gavalas, G. R. A random capillary model with application to char gasification at chemically controlled rates. AIChE J. 1980, 26, 577-585. (12) Feng, B.; Bhatia, S. K. Percolative fragmentation of char particles during gasification. Energy Fuels 2000, 14, 297-307. (13) Su, J.-L.; Perlmutter, D. D. Effect of pore structure on char oxidation kinetics. AIChE J. 1985, 31, 973-981. (14) Bae, J.-S.; Bhatia, S. K. High pressure adsorption of methane and carbon dioxide on coal. Energy Fuels, 2006, 20, 2599-2607. (15) Nguyen, T. X.; Bhatia, S. K. Determination of pore accessibility in disordered nanoporous materials. J. Phys. Chem. C, in press. (16) Bhatia, S. K. Reactivity of chars and carbons: New insights through molecular modeling. AIChE J. 1998, 44, 2478-2493. (17) Oberlin, A. Carbonization and graphitization. Carbon 1984, 22, 521-541. (18) A practical guide for the preparation of specimens for X-ray fluorescence and X-ray diffraction analysis; Burke, V. E., Jenkins, R., Smith, D. K., Eds.; Wiley-VCH: New York, 1998. (19) X-ray diffraction procedures for polycrystalline and amorphous materials; Klug, H. P., Alexander, L. E., Eds.; Wiley: New York, 1974. (20) Shi, H.; Reimers, J. N.; Dahn, J. R. Structure-Refinement program for disordered carbons. J. Appl. Crystallogr. 1993, 26, 827-836. (21) Shi, H. Ph.D. Thesis, Simon Fraser University, Burnaby, BC, Canada, 1993. (22) Franklin, R. E. The interpretation of diffuse X-ray diagrams of carbon. Acta Crystallogr. 1950, 3, 107-121. (23) Powder surface area and porosity; Lowell, S., Shields, J. E., Eds.; Chapman and Hall: London, 1984. (24) Szekely, J.; Evans, J. W.; Sohn, H. Y. Gas-Solid Reactions; Academic Press: New York, 1976. (25) Sahimi, M.; Gavalas, G. R.; Tsotsis, T. T. Statistical and continuum models of fluid-solid reactions in porous media. Chem. Eng. Sci. 1990, 45, 1443-1502. (26) Bhatia, S. K.; Gupta, J. S. Mathematical modeling of gas-solid reactions: Effect of pore structure. ReV. Chem. Eng. 1994, 8, 177-258. (27) Aarna, I.; Suuberg, E. M. Changes in reactive surface area and porosity during char oxidation. Proceedings of the 27th Symposium (International) on Combustion: The Combustion Institute, 1998; pp 29332939. (28) Rybak, W. Intrinsic reactivity of petroleum coke under ignition conditions. Fuel 1988, 67, 1696-1702.

Literature Cited (1) Mannweiler, W. Petroleum coke production: Delayed coking and calcining. In Anodes for the Aluminium Industry; Fischer, W. K., Mannweiler, W., Keller, F., Perruchoud, R. C., Buhler, U., Eds.; R & D Carbon Limited: Sierre, Switzerland, 1995; pp 15-19.

ReceiVed for reView August 3, 2006 ReVised manuscript receiVed September 29, 2006 Accepted October 2, 2006 IE061017E