ADSORPTIVE REMOVAL OF SULFUR COMPOUNDS IN KEROSENE BY USING RICE HUSK ACTIVATED CARBON

Yoshie Shimizu, Seiji Kumagai, Koichi Takeda, Akita Prefectural University, Yurihonjo 015-0055, JAPAN Yasuhiro Toida, Japan Energy Corporation, Toda 335-8502, JAPAN Yukio Enda, Industrial Technology Center of Akita Prefecture, Akita 010-1623, JAPAN

Abstract
Japanese rice husks were carbonized in N2 at 400C for 1 h and then were activated in CO2 at 850C for 1 h. The capacities of rice husk activated carbons (RHACs) to adsorb refractory sulfur compounds of dibenzothiophenes (DBTs) were evaluated correlating with their textural and chemical characteristics. The RHACs of 0.5 g were soaked in commercial kerosene of 15.0 g at 10C for 100 h. A micro-porous activated carbon fiber (ACF) having large specific surface area (2336 m2/g) and total pore volume (1.052 cm3/g) was also tested for a comparison purpose. The largest DBTs adsorption capacity (0.069 mg-S/g-Ads and 77 % performance to the ACF) was observed, despite much smaller specific surface area (473 m2/g) and total pore volume (0.267 cm3/g) compared to those of the ACF. Predominant adsorption sites for DBTs in the RHACs were deemed to be slit-shaped micropores of which pore width was less than 0.7 nm.

Introduction
Dibenzothiophenes and its derivatives (hereafter called DBTs) included in fuel oils such as kerosene and diesel oil are refractory poly-aromatic sulfur compounds. In particular, alkyl-substituted DBTs, i.e., 4-methyldibenzothiophene (4-MDBT) and 4, 6- dimethyldibenzothiophene (4, 6-DMDBT), are much refractory (Song and Ma, 2003 and Macaud et al., 2000). Modified hydrodesulfurization under catalysis (Pawelec et al., 2003, Shu et al., 2005 and Zeng et al., 2006) and adsorptive desulfurization using porous materials (Jayne et al., 2005, Ma et al., 2002 and Ng et al., 2005) have been proposed to remove efficiently the refractory DBTs in fuel oils. Selective desulfurization using activated carbons (Ania and Bandosz, 2005 and Zhou et al., 2006), which can be performed at ambient temperature and pressure, has been intensively studied. That is because it seems to be more realistic than the other means on economical basis and operational safety. Toida demonstrated that activated carbon having a specific pore structure attained an efficient and selective removal of DBTs (Toida, 2003). The specific activated carbon is required to have a composite micro-meso porous structure with a large pore volume. A huge quantity of rice husk (ca. 3 million tons) is produced every year as agricultural waste in Japan. In the present study, rice husk was converted into activated carbon intended for a removal of DBTs from kerosene by employing CO2 gas activation method. Kerosene is widely used in Japan as a home heating fuel for portable and installed kerosene heaters. Kerosene can be readily purchased at any filling stations or be delivered to homes. Kerosene is a promising fuel oil as hydrogen source of fuel cells for stationary home-use in Japan. The DBTs adsorption capacity of the rice husk activated carbon (RHAC) in commercial kerosene was evaluated, with a correlation with textural and chemical characteristics. A micro-porous activated carbon fiber (ACF) was also tested for a comparison with the RHAC.

Materials and Methods

Material Preparation Japanese rice husk samples were obtained by rice threshing performed in autumn of 2002, 2003 and 2004. The sample description of the rice husks used is shown in Table 1. The raw rice husk of 10 g was carbonized and activated in a stainless steel cylinder (SUS 304) whose respective external and bore diameters were 50 and 44 mm. The rice husk was at first carbonated at 400C for 1 h in N2 flow at 500 mL/min. The rice husk was further heated in N2 flow and finally activated at 850C for 1 h in CO2 flow at 500 mL/min. The carbonization and activation process is shown in Figure 1. The mass change resulting from the carbonization and activation was measured as mass yield. The commercially available ACF (FR-25, Kuraray Chemical Co., Ltd., Japan) used for the comparison purpose was also given to the similar tests and analysis. Table 1. Sample description of the Japanese rice husks used in the present study. Sample ID AK-OH AK-NI AK-OG HI-HO KO-TO Rice brand Akita Komachi Akita Komachi Akita Komachi Hitomebore Koshihikari Production place Ohgata, Akita Prefecture Nishiki, Akita Prefecture Ogachi, Akita Prefecture Honjo, Akita Prefecture Toyooka, Shizuoka Prefecture Production year 2002 2003 2003 2003 2004

Figure 1. Carbonization and activation process to prepare RHAC. Compositional Analysis The ash content of the samples (0.1 g) was precisely determined from the residual-ash ratio after incineration at 800C for 1 h in air flow (100 mL/min) using a thermogravimetric analyzer (TGA-51; Shimadzu Corp., Japan). A CHN/S analyzer (2400 I; PerkinElmer Inc., USA) was used to determine the hydrogen, carbon, nitrogen and sulfur contents of the samples. All the samples were dried at 105C for 3 h prior to analyses. The content of oxygen excluding that in ash was determined to be a difference of hydrogen, carbon, nitrogen, sulfur, and ash. Textural Characterization Nitrogen adsorption isotherms at -196C were obtained using a gas adsorption analyzer (Autosorb-3B; Quantachrome Instruments Inc., USA). The sample of 0.05 g was degassed at 200C for > 3 h prior to the isotherm measurement. The BET specific surface area (SBET) was based on the BET theory (Bansal and Goyal, 2005) and was calculated using the N2 adsorbed volume at relative pressures of 0.05-0.1. Liquid N2 volume, referring to the N2 adsorbed volume at a relative pressure of 0.995, was determined to be the total pore volume (Vt). The mean pore width (Wmean) was calculated to be 2Vt/SBET, in which pores were assumed to be slits. Activated carbons consist of graphite-like sheets between which small slit-shaped pores exist (EI-Sayed and Bandosz, 2002). In the present study, according to Sing et al. (1985), micropores were distinguished into ultlamicropores (width less than 0.7 nm) and supermicropores (width from 0.7 to 2.0 nm). The pore size distribution was evaluated using density functional theory (Olivier, 1998 and Ryu et al., 1999). The DFT software developed by Quantachrome Instruments Inc. (version 1.62) was used to obtain the pore size distribution, in which a slit pore equilibrium model was employed. The volumes of ultramicropores (Vu), supermicropores (Vs), and mesopores (Vm) were obtained using the pore size distribution data obtained by the above DFT method. Adsorptive DBTs Removal Test Commercial kerosene produced in Japan Energy Corporation, which was Grade-1 quality regulated in Japan Industrial Standard (JIS-K2203), was employed for the adsorptive desulfurization test. The sample was dried at 130 C in atmospheric air for 3 h prior to the kerosene immersion. The sample of 0.5 g was immersed in the above kerosene of 15.0 g in a 30 mL glass vessel at 10C for 100 h. Kerosene was not stirred during the immersion, but shaken a few time everyday by hand. The adsorption time of 100 h was confirmed to be sufficient to reach the adsorption equilibrium. The total sulfur content in kerosene was evaluated using a sulfur analyzer based on ultraviolet fluorescence method (9000LLS; Antek Instruments, Inc., USA). The sulfurs embedded in DBTs were identified and quantified using a gas chromatograph-inductively coupled plasma mass spectrometer (GC-ICP-MS, GC 6890N; ICP-MS 7500CS; Agilent Technologies Inc., USA). DBTs are defined as sulfur compounds which contain double aromatic rings and of which molecular mass is larger than that of DBT (molecular mass: 184) in the GC-ICP-MS system. The total sulfur content in the untreated kerosene was evaluated to be 14.1 mass ppm, while the content of sulfur in DBTs therein was 3.5 mass ppm. DBTs adsorption capacity (CDBTs) was represented by the mass of sulfur in adsorbed DBTs divided by the mass of the adsorbent (RHAC sample).

Result and Discussion

Compositional, Textural, DBTs Adsorption Characteristics of Different Types of RHACs Five types of rice husks activated at the temperature of 850C for 1 h were provided to compositional analysis. The mass yield of the carbonization and activation process and the elemental composition of the above RHACs are shown in Table 2. The level of mass yield is found to be related to the type of rice husk. The RHAC with the lower mass yield is composed of the higher-level carbon and the lower-level ash. High-content carbon and low-content ash are particularly observed on the activated AK-NI and KO-TO. Table 2. Mass yield and elemental composition of different types of rice husks activated at 850C for 1 h. Mass yield (mass%) AK-OH 33.2 AK-NI 26.1 AK-OG 30.6 HI-HO 34.2 KO-TO 26.8 *: by difference and not in ash Sample ID H (mass%) 0.4 0.5 0.5 0.6 0.5 C (mass%) 35.0 47.3 40.5 35.5 43.6 N (mass%) 0.5 0.7 0.8 0.5 0.6 S (mass%) 0 0 0 0 0 Ash (mass%) 61.5 47.7 55.1 60.7 52.2 O* (mass%) 2.6 3.8 3.1 2.7 3.1

Figure 2 shows nitrogen adsorption isotherms on the different types of the rice husks activated at 850C for 1 h. All the isotherms display a gradual increase in the slope at the relative pressure higher than 0.1 and a hysteresis loop at a high relative pressure, belonging to type IV. A hysteresis loop at a high relative pressure shows an existence of mesopores. The type of rice husk has an influence on the isotherm at only low relative pressure (< 0.1) at the activation conditions of 850C and 1 h. Table 3 shows SBET, Vt, and Wmean of the different types of rice husks activated at 850C for 1 h. Large SBET and Vt were observed on the activated KO-TO and AK-NI, with high-content carbon and low-content ash, while the activated AKOH and HI-HO with low-content carbon and high-content ash provide small SBET and Vt. Large SBET and Vt are found to be responsible for a reduction of the pore width.

Figure 2. Nitrogen adsorption isotherms at -196C on different types of rice husks activated at 850C for 1 h.

Table 3. BET specific surface area (SBET), total pore volume (Vt), and mean pore width (Wmean) of different types of rice husks activated at 850C for 1 h. Sample ID AK-OH AK-NI AK-OG HI-HO KO-TO SBET (m2/g) 334 460 388 325 473 Vt (cm3/g) 0.207 0.261 0.231 0.200 0.267 Wmean (nm) 1.24 1.13 1.19 1.23 1.13

The DBTs adsorption capacities of different types of RHACs were evaluated with a relation to the volume of pores at the similar activation conditions (see Figure 3). High values of CDBTs were observed on the activated KO-TO and AK-NI (0.069 and 0.066 mg-S/g-Ads, respectively), which are accompanied by the large Vu and Vs. All the RHACs show a similar level of Vm (0.065-0.070 cm3/g), indicating that the volume of mesopores was not determined by the contents of carbon and ash, but by the activation conditions. Comparable volumes of mesopores were observed on all the RHACs, demonstrating that increases in Vu and Vs enlarge the DBTs adsorption capacity. The textural and DBTs adsorption characteristics of micro-porous ACF were also evaluated. The textual characteristics and the DBTs adsorption capacities of the ACF and the KO-TO activated at 850C for 1 h, with the largest adsorption capacity in the RHAC samples, are compared in Table 4. Remarkably larger Vs of the ACF (0.614 cm3/g) should contribute to larger SBET and Vt which are respectively 5 and 4-fold greater than those of the activated KO-TO. Vu and Vm of the ACF are about twice larger than those of the activated KO-TO. However, CDBTs of the ACF was 0.090 mg-S/g-Ads, while that of the activated KO-TO was 0.069 mg-S/g-Ads (77 % performance to the ACF). If all the sulfur in DBTs was removed from the used kerosene, the DBTs adsorption capacity should be 0.105 mg-S/g-Ads. The above DBTs adsorption results imply that larger Vs which provides larger SBET and Vt is not so useful for the DBTs adsorption. The roles of Vu and Vm on the DBTs adsorption seem to be important.

Figure 3. The relationship between DBTs adsorption capacity in commercial kerosene and the pore volumes of ultramicropore, supermicropore, and mesopore on different types of the activated rice husks.

Table 4. A comparison of textual characteristics and DBTs adsorption capacity for activated carbon fiber (ACF) and KO-TO activated at 850C for 1 h, with the largest adsorption capacity in RHAC samples. SBET (m2/g) Vt (cm3/g) Wmean (nm) Vu (cm3/g) Vs (cm3/g) Vm (cm3/g) CDBTs (mg-S/g-Ads) ACF 2336 1.052 0.901 0.201 0.614 0.128 0.090 Activated KO-TO 473 0.267 1.128 0.105 0.065 0.070 0.069

Ania and Bandosz (2006) evaluated the performance of various activated carbons from different carbon precursors as a DBT adsorbent in hexane liquid at 178-S mass ppm. It was shown that the volume of micropores of which width was < 0.7 nm (defined as ultramicropores in this study) governed the amount of the adsorbed DBT, and the adsorption capacity had a linear relation with the volume of those pores. It was also mentioned that mesopores controlled the kinetics of the DBT adsorption process. The size of the DBT molecule is reported to be 0.65 nm (Ania and Bandosz, 2006). DBTs having a plane structure can enter slit-shaped ultramicropores in parallel to the wall surface, which produces a high potential for the DBTs adsorption. Meille et al. (1999) studied the mechanism of DBTs adsorption on a catalyst, suggesting a significance of electrons in the aromatic rings participating in the adsorption mechanism of DBTs. The interaction of electrons on graphitelike sheets constituting slit-shaped pores in activated carbons between electrons in DBTs is likely to further promote an entrance of DBTs in parallel to the wall surface of slit-shaped ultramicropores. Therefore, predominant adsorption sites of DBTs in the RHACs are deemed to be ultramicropores. An activated carbon of 0.5 g was immersed at 25C in a model fuel of 10.5 g, in which benzothiophene, DBT, 4-methyl DBT, and 4, 6-dimethyl DBT (each about 100 S-mass ppm) were added (Zhou et al., 2006). The largest adsorption capacity was 7.0 mg-S/g-Ads, which is about 100-fold higher than that of the RHAC greatest in this study. The differences between the present study and the above preceding researches are the sulfur content in the solute. The sulfur content in the preceding works is much higher than that of the kerosene used. At the high sulfur content, it is likely that ultramicropores volume dominates the capacity of DBTs adsorption and mesopores volume has influence on the kinetics of DBTs adsorption. However, at the low sulfur content as in the present study, the DBTs adsorption sites in ultramicropores might not be fully filled. Hence, not only ultramicropores but also mesopores, of which important role is leading DBTs to the adsorption sites, determined the DBTs adsorption capacity. Very large Vs (0.614 cm3/g) of the ACF, resulting in very large SBET and Vt, was not useful for increasing CDBTs. Even though SBET and Vt of the RHACs are not so large, the RHACs have more suitable pore structure which can efficiently adsorb DBTs in the kerosene.

Conclusions
Japanese rice husks were at first carbonated in N2 and then activated in CO2 at 850C for 1 h. The textural and chemical characteristics of the obtained RHACs as well as the ACF having much larger SBET and Vt were evaluated in order to correlate their DBTs adsorption capacity in the kerosene. A feature of the RHACs was very high-content ash making up about half of their mass. The largest CDBTs (77 % performance to the ACF) appeared on the RHAC, with the high-content carbon and the low-content ash in all the RHAC samples. Not only ultramicropores working as the DBTs adsorption sites but also mesopores leading DBTs into the ultramicropores were suggested to determine the DBTs adsorption capacity. A contribution of supermicropores on the DBTs adsorption was minor. Supermicropores, which were much produced in the ACF and provided its large SBET and Vt, were not be beneficial as long as the DBTs removal in the kerosene. Therefore, the RHACs displayed a decent performance when compared to the ACF despite much smaller SBET and Vt, indicating that RHACs produced a textural property suitable to an efficient removal of DBTs in the kerosene.

Acknowledgement
This research was in part supported by Industrial Technology Research Grant Program in 2006 from New Energy and Industrial Technology Development Organization (NEDO) of Japan.

References
Ania, C. O. and T. J. Bandosz. 2005. Importance of Structural and Chemical Heterogenecity of Activated Carbon Surfaces for Adsorption of Dibenzothiophene, Langmuir, 21:7752-7759. Ania, C. O. and T. J. Bandosz. 2006. Metal-loaded Polystyrene-based Activated Carbon as Dibenzothiophene Removal Media via Reactive Adsorption, Carbon, 44:2404-2412. Bansal, R. C. and M. Goyal. 2005. Activated Carbon Adsorption, CRC Press Inc., Boca Raton, Florida. EI-Sayed, Y. and T. J. Bandosz. 2002. Acetaldehyde Adsorption on Nitrogen-containing Activated Carbons, Langmuir, 18: 3213-3218. Jayne, D., Y. Zhang, S. Haji and C. Erkey. 2005. Dynamics of Removal of Organosulfur Compounds from Diesel by Adsorption of Carbon Aerogels for Fuel Cell Applications, International Journal of Hydrogen Energy, 30:1287-1293. Ma, X., L. Sun and C. Song. 2002. A New Approach to Deep Desulfurization of Gasoline, Diesel and Jet Fuel by Selective Adsorption for Ultra-clean Fuels and for Fuel Cell Applications, Catalysis Today, 77:107-116. Macaud, M., A. Milenkovic, E. Schulz, M. Lemaire and M. Vrinat. 2000. Hydrodesulfurization of Alkyldibenzothiophenes: Evidence of Highly Unreactive Aromatic Sulfur Compounds, Journal of Catalysis, 193:255-263. Meille, V., E. Schulz, M. Lemaire and M. Vrinat. 1999. Hydrodesulfurization of 4-methyl-dibenzothiophene: A Detailed Mechanistic Study, Applied Catalysis A: General, 187:179-186. Ng, F. T. T., A. Rahman, T. Ohasi and M. Jiang. 2005. A Study of the Adsorption of Thiophenic Sulfur Compound using Flow Calorimetry, Applied Catalysis, B: Environment, 56:127-136. Olivier, J. P. 1998. Improving the Models Used for Calculating the Size Distribution of Micropore Volume of Activated Carbons from Adsorption Data, Carbon, 36:1469-1472. Pawelec, B., S. Damyanova, R. Mariscal, J. L. G. Fierro, I. Sobrados, J, Sanz and L. Petrov. 2003. HDS of Dibenzothiophene over Polyphosphates Supported on Mesoporous Silica, Journal of Catalysis, 223:86-97. Ryu, Z., J. Zheng, M. Wang and B. Zhang. 1999. Characterization of Pore Size Distributions on Carbonaceous Adsorbent by DFT, Carbon, 37:1257-1264. Shu, Y., Y. K. Lee and S. T. Oyama. 2005. Structure-sensitivity of Hydrodesulfurization of 4, 6-dimethyldibenzothiophene over Silica Supported Nickel Phosphide Catalysts, Journal of Catalysis, 236:112-121. Sing, K. S. W., D. H. Everett, R. A. W. Haul, L. Moscou, R. A. Pierotti, J. Rouquerol, T. Siemieniewska. 1985. Reporting Physisorption Data for Gas/Solid Systems, Pure and Applied Chemistry, 57:603-619. Song, C. and X. Ma. 2003. New Design Approaches to Ultra-clean Diesel Fuels by Deep Desulfurization and Deep Dearomatization, Applied Catalyst B: Environmental, Vol. 41, pp. 207-238, 2003. Toida, Y. 2003. Adsorption Desulfurization Agent for Desulfurizing Petroleum Fraction and Desulfurization Method Using the Same, International Patent WO03-097771, Japan Energy Corporation. Zeng, S., J. Blanchard, M. Breysse, Y. Shi, X. Su, H. Nie and D. Li. 2006. Mesoporous Materials from Zeolite Seeds as Supports for Nickel-tungsten Sulfide Active Phases. Part 2. Catalytic Properties for Deep Hydrodesulfurization Reactions, Applied Catalysis A: General, 298:88-93. Zhou, A., M. Xiaoling and C. Song. 2006. Liquid-Phase Adsorption of Multi-Ring Thiophenic Sulfur Compounds on Carbon Materials with Different Surface Properties, Journal of Physical Chemistry. B., 110:4699-4707.