Chemical Engineering Science 64 (2009) 5240 -- 5246

Contents lists available at ScienceDirect

Chemical Engineering Science

journal homepage: www.e lsev ier .com/ locate /ces

Desulfurization of high-sulfur jet fuel bymesoporous�-complexation adsorbents

Hao Chen, Yuhe Wang, Frances H. Yang, Ralph T. Yang∗Department of Chemical Engineering, University of Michigan, Ann Arbor, MI 48109, USA

A R T I C L E I N F O A B S T R A C T

Article history:Received 26 June 2009Received in revised form 25 August 2009Accepted 26 August 2009Available online 2 September 2009

Keywords:Desulfurization of jet fuelSBA-15MCM-41AgNO3

Fuel desulfurization

Desulfurization of JP-5 jet fuel (1172ppmw S) was investigated by �-complexation adsorption withAgNO3 supported on mesoporous silica SBA-15 and MCM-41. The average pore sizes of AgNO3/SBA-15and AgNO3/MCM-41 were 48.8 and 19.1Å, respectively. The results of JP-5 desulfurization showed thatsignificant sulfur breakthrough occurred at ∼10.0 and ∼15.0mL/g by AgNO3/SBA-15 and AgNO3/MCM-41,respectively, at a space velocity of 1.25h−1. The spent AgNO3/MCM-41 was regenerated by a simpleprocess (heating in air at 200 ◦C) and ∼50% of the sulfur capacity was recovered after the first cycle.Molecular orbital calculations show that Cu+ (as that in CuY zeolite) formed stronger �-complexationbonding with the thiophenic compounds than Ag+ (in AgNO3), as evidenced by experimental heats ofadsorption. However, pore diffusion limitation of the large sulfur molecules (alkylated benzothiophenes)became an important factor for desulfurization of high sulfur jet fuels such that the AgNO3-supportedmesoporous sorbents yielded substantially better results than Cu(I)Y, although Cu(I)Y was better for amodel fuel that contained only small sulfur molecules. Among all sorbents that have been investigated,the AgNO3/MCM-41 sorbent showed the best desulfurization performance for high sulfur jet fuels.

© 2009 Elsevier Ltd. All rights reserved.

1. Introduction

Because of the high energy density, ready availability, safety andconvenient storage, gasoline, diesel and jet fuel are the best sourcesfor hydrogen for fuel cell applications (Hernández-Maldonado andYang, 2004a). In particular, military jet fuels are considered by manyas an excellent fuel source for hydrogen for on-board fuel cell gen-erator systems, and are desired for use in various military applica-tions. For certain applications, such as distributed shipboard powergeneration, high flashpoint liquid fuels are absolutely mandatoryfor realization of such systems, as a result of safety considerations(Ma et al., 2002). However, jet fuels generally contain very highconcentrations of sulfur. For example, JP-5 high flashpoint jet fueltypically contains 1000–1500ppm of sulfur (e.g., the JP-5 jet fuelcontains around 1270ppmw sulfur, with a maximum specificationof 3000ppm, Wang et al., 2006). Similarly, JP-8 jet fuel also has amaximum specification of 3000ppm of sulfur. F-76 marine distil-late fuel currently can contain up to 10000ppm of sulfur, althougha specification change will reduce this to 5000ppm (Heinzel et al.,2005). When the hydrogen feeding the fuel cell is produced via cat-alytic autothermal reforming of jet fuels, sulfur compounds that canpoison the follow-on processing operations and ultimately the elec-trode catalyst of the fuel cell will also be generated simultaneously.

∗ Corresponding author. Tel.: +17349360771; fax: +17347647453.E-mail address: yang@umich.edu (R.T. Yang).

0009-2509/$ - see front matter © 2009 Elsevier Ltd. All rights reserved.doi:10.1016/j.ces.2009.08.031

Therefore, reducing the sulfur concentration of jet fuels is a key issuefor the utilization of jet fuels as a hydrogen source for on-board fuelcells. Removal of sulfur containing compounds from transportationfuels is an important operation in petroleum refining, and is achievedby catalytic processes operated at high pressures and elevated tem-peratures. Today, refineries rely on hydrodesulfurization (HDS) pro-cesses to reduce sulfur levels, but achieving deep-desulfurizationlevels would require increasing existing reactor sizes and increasinghydrogen consumption (Hernández-Maldonado and Yang, 2004b).There has been much recent interest in using adsorbents for selec-tive desulfurization of fuels (Haji and Erkey, 2003; McKinley andAngelici, 2003; Chica et al., 2004; Jeevanandam et al., 2005; Kim et al.,2006; King and Faz, 2006; Jiang and Ng, 2006; Laborde-Boutet et al.,2006; Etemadi and Yen, 2007; Yang et al., 2007a,b, 2008; Seredychand Bandosz, 2007; Ko et al., 2007; Nuntang et al., 2008; Landauet al., 2008, 2009; Jeon et al., 2009). Additionally, for military fuelcell generators, HDS is not a viable option due to the requirementof high purity, high pressure hydrogen, large reactors, valving, ac-tuators and associated equipment. In systems that are sensitive topower density, weight and volume, these systems lose applicability(Heinzel et al., 2006).

Our group has recently developed a class of adsorbents thatrely on �-complexation bonding to selectively remove organosulfurmolecules from commercial fuels (Hernández-Maldonado and Yang,2004a; Wang et al., 2006, 2008, 2009; Wang and Yang, 2007). It wasfound that Cu(I)-Y zeolite exhibited highly selective removal for sul-fur compounds from transportation fuels. It was concluded that Cu+

H. Chen et al. / Chemical Engineering Science 64 (2009) 5240 -- 5246 5241

in Cu(I)-Y zeolite can remove sulfur compounds by �-complexationadsorption with thiophenic compounds (Hernández-Maldonado andYang, 2004a). Our results also indicted that metal halides supportedon active carbon, such as CuCl/AC or PdCl2/AC, are effective in thedesulfurization of JP-5 (Wang et al., 2006; Wang and Yang, 2007).More recently, we have found that PdCl2/SBA-15 and PdCl2/MCM-41were excellent adsorbents for desulfurizing very-high-sulfur com-mercial jet fuels such as an JP-5 that contained 847ppmw S (Wanget al., 2008). For this JP-5, significant breakthrough (< 10ppmwS) occurred at about 15.0 and 50.0mL/g with PdCl2/MC-41 orPdCl2/SBA-15, respectively. Although the regeneration result of thespent PdCl2/SBA-15 was unsatisfactory as only around 48% capacityof PdCl2/SBA-15 could be recovered, these findings demonstratedthat the support has an important effect on the performance ofadsorbents in the desulfurization of jet fuels (Wang et al., 2008).Regeneration is an important aspect that should be considered,especially for on-board applications. Safety, high efficiency andavailability of reactants need to be considered for the regenerationof saturated adsorbents. Heating the spent adsorbents using airis an effective choice for the regeneration of saturated adsorbent(Hernández-Maldonado and Yang, 2004a). Therefore, a convenientregeneration sorbent, Cu2O/MCM-41 was developed for the deepdesulfurization of a JP-5 light fraction (841ppmw S). Although thespent Cu2O/MCM-41 can be completely regenerated by heating inair, the product fuel of 10mL/g sorbent still contained ∼50ppmw S(Wang et al., 2009).

In our previous study for the bulk separation of C4 olefins fromparaffins, a �-complexation capable sorbents, AgNO3/SiO2, wasfound to have excellent capacities and selectivities for this ap-plication (Padin et al., 1999). AgNO3/SBA-15 and AgNO3/MCM-41adsorbents were prepared and studied in this work. The resultsshow that the supported AgNO3 is an excellent adsorbent for selec-tive desulfurization from the jet fuel; ∼15.0mL of the JP-5 can bedesulfurized per gram adsorbent from 1172ppmw S to< 1ppmwS with AgNO3/MCM-41 adsorbent, and ∼50% capacity of the spentAgNO3/MCM-41 can be recovered upon air regeneration after the1st regeneration cycle.

2. Materials and methods

2.1. Adsorbents preparation

Monolayer AgNO3/SBA-15 or AgNO3/MCM-41 was prepared byimpregnating SBA-15 or MCM-41 with an appropriate amount ofaqueous solution of silver nitrate (AgNO3). The weight ratios of sil-ver nitrate to the support for AgNO3/SBA-15 and AgNO3/MCM-41were, respectively, 0.43 and 0.60, based on estimate for monolayerspreading of AgNO3. First, a 1.2M solution of AgNO3 was prepared.A volume of solution equal to the total pore volume of sample wasintroduced and mixed with the support so that a weight ratio of0.43 or 0.60 was achieved. The sample was then heated for 12h at105 ◦C in air to remove the water (Padin et al., 1999; Padin and Yang,2000; Xie and Tang, 1990). Before its use, the sample was heated insitu at 105 ◦C in flowing nitrogen. In the earlier work of Padin andYang (2000), the dispersion of AgNO3 on various supports (includingmesoporous silica MCM-41), by various methods (including sponta-neous thermal spreading and wet impregnation, as used here), anddetail characterization have been discussed. The appropriate tem-peratures of sample treatment (as used in this work) have also beendiscussed.

Cu(I)Y (VPIE, i.e., vapor-phase ion exchange) adsorbent was syn-thesized following the previous recipe developed in our laboratory(Hernández-Maldonado and Yang, 2004b).

The particle or crystal sizes of the CuY zeolite were around 0.5�m.SBA-15 consisted of rope-like domains with sizes of around 1�m,

Fig. 1. GC-FPD chromatograms of JP-5 light fraction (A) and model jet fuel (B).

which were aggregated, whereas MCM-41 consisted of irregularspheres with sizes of 1–3�m.

2.2. Adsorbent characterization

The BET surface areas of the samples were measured by physicaladsorption of N2 at −196 ◦C using Micromeritics ASAP 2010. Poresize distribution was also analyzed by N2 adsorption at −196 ◦C. TheBJH method was used for calculating the mean pore size and porevolume (Barrett et al., 1951).

2.3. Reagents

A JP-5 jet fuel sample obtained from the Naval Business Center,NAVSEA, Philadelphia, was used in the desulfurization experiments.The sulfur concentration of the JP-5 was analyzed with a flame pho-tometric detector-gas chromatograph (FPD-GC). For the JP-5 sam-ples, the as-received JP-5 with 1172ppmw total sulfur concentrationand a light fraction of the JP-5 with 841ppmw sulfur concentrationwere used. The analysis result of 1172ppmw-S concentration canbe found in our previous work (Wang et al., 2006), and the analysisresult for the JP-5 light fraction is shown in Fig. 1.

5242 H. Chen et al. / Chemical Engineering Science 64 (2009) 5240 -- 5246

A model fuel with 150ppmw total sulfur concentration wasalso prepared in this work as follows: 10ml benzene and 90mln-octane were well mixed first, and then 0.0226g benzothiopheneand 0.0251g methyl-benzothiophene were dissolved in the abovesolution and well stirred. After well mixed, the model jet fuel waskept in the flask. The corresponding analysis result is also shown inFig. 1. The sulfur concentrations were 75ppmw-S for both benzoth-iophene and methyl-benzothiophene, or total 150ppmw-S.

2.4. Fixed-bed adsorption/breakthrough experiments

All dynamic adsorption or breakthrough experiments were per-formed in a vertical custom-made quartz adsorber equipped with asupporting glass frit as described elsewhere (Hernández-Maldonadoet al., 2005). The setup consisted of a low-flow liquid pump, feedtanks, and a heating element. Initially, the adsorbent was loaded in-side the adsorber, and pretreated in situ using dry gases to avoid ex-posure to atmospheric moisture. The gases used for activation werepretreated (dried) inline before contacting the adsorbent using a col-umn of 3A-type zeolite. After activation treatment, the adsorbentbed was washed with a sulfur free hydrocarbon (e.g., n-octane) to re-move any entrapped gas. After allowing the liquid hydrocarbon headto disappear, the fuel was allowed to contact the bed in a down-flow direction. The effluent was sampled periodically until satura-tion was achieved, which depended on the adsorption dynamics andthe amount of adsorbent.

Breakthrough adsorption curves were generated by plotting thetransient total sulfur concentration normalized by the feed total sul-fur concentration vs. cumulative fuel volume normalized by totaladsorbent weight. The adsorption amounts (normalized per adsor-bent weight) were obtained by integration (Hernández-Maldonadoet al., 2005).

2.5. Gas chromatographic analysis

All the fuels from the desulfurization of JP-5 collected during thebreakthrough experiments were analyzed using a Shimadzu GC-17Av3 unit equipped with an EC-5 capillary column and a flame photo-metric detector (FPD). The sulfur detection limit was approximately50ppb. More details on the GC analysis could be found elsewhere(Hernández-Maldonado et al., 2005).

2.6. Adsorbent regeneration

In this work, the feasibility for regeneration of the AgNO3/MCM-41 was investigated by heating the spent AgNO3/MCM-41 adsorbentin flowing air. The saturated sample (1.0 g) by JP-5 was heated at200 ◦C for 24h in flowing air of 50mL/min. Afterward, the regener-ated sorbent was further tested for desulfurization of JP-5.

3. Ab initio molecular orbital computation

The nature and bond energy for �-complexation bonding betweenthe sorbate molecule and sorbent metal site can be calculated andunderstood bymolecular orbital (MO) theory. Molecular orbital stud-ies on the �-complexation bonding for thiophene, benzene, benzoth-iophene and 4,6-dimethyldibenzothiophene on sorbent surfaces andzeolites have been reported in our earlier work (Yang et al., 2005). Inthis work, similar MO studies were extended to a new sorbent, silvernitrate, and with an additional adsorbate, 2-methybenzothiophene.The Gaussian 03 package (Frisch et al., 2004) and Cerius2 molecu-lar modeling software (Cerius2, version 4.6, Accelrys, San Diego, CA)were used for all MO calculations. Geometry optimizations were per-formed at density functional theory (DFT) level using effective corepotentials (ECPs) (Russo et al., 1995).

3.1. Geometry optimization and bond energy calculations

Frequency analysis was used to verify that all geometry optimizedstructures were true minima on the potential energy surface. Theoptimized structures were then used for bond energy calculationsaccording to the following expression:

Eads = Eadsorbate + Eadsorbent − Eadsorbent−adsorbate (1)

where Eadsorbate is energy of free adsorbate, Eadsorbent is energy offree adsorbent and Eadsorbent−adsorbate is energy of the adsorbate/adsorbent system. A higher value of Eads corresponds to a strongeradsorption.

4. Results and discussion

4.1. Adsorbent characterization

Fig. 2 shows the adsorption–desorption isotherms of N2 at−196 ◦C for the SBA-15 and MCM-41 based adsorbents. Fig. 3 showsthe pore size distributions of SBA-15 and MCM-41 based adsorbents.The corresponding BET surface areas, pore sizes and pore volumesare listed in Table 1.

4.2. Molecular orbital calculation results

For the CuY zeolite, a cluster model was chosen for the zeoliteframework structure which represents well the chemistry of thezeolite (Yang et al., 2005). The zeolite anion is denoted as Z−. Theadsorption bond energies for all relevant adsorbate molecules andthree sorbents (CuZ, CuCl and AgNO3) were calculated and are shownin Table 2.

First, it is seen that the �-complexation bonding is strongerfor thiophene than benzene for both Cu+ and Ag+. This result as-sures selectivity for thiophene over benzene by sorbents containingthese cations, i.e., selective desulfurization. Also, from Table 2, it isseen that the �-complexation bonding for all aromatic adsorbatemolecules is considerable stronger for Cu+ than Ag+, i.e., Cu+ >Ag+. These results are consistent with that reported earlier, and arealso consistent with experimental heats of adsorption data (Yanget al., 2003). Comparing different anions, it is seen that the zeolite

Fig. 2. Nitrogen adsorption–desorption isotherm plots for SBA-15 (�) MCM-41 (•),AgNO3/SBA-15 (�) and AgNO3/MCM-41 (©); P0 is 760mmHg.

H. Chen et al. / Chemical Engineering Science 64 (2009) 5240 -- 5246 5243

Fig. 3. BJH pore volume plots of SBA-15 (�), MCM-41 (�), AgNO3/SBA-15 (�) andAgNO3/MCM-41 (©).

Table 1N2 adsorption characterization of different sorbents.

Adsorbent Metalloading(mmol/g)

BET surfacearea (m2/g)

BJH poresize (Å)a

BJH porevolume(cm3/g)b

SBA-15 – 901 58.7 1.14AgNO3/SBA-15 1.77 408 48.8 0.55MCM-41 – 1225 28.7 1.17AgNO3/MCM-41 2.21 490 19.1 0.27

aBJH desorption average pore diameter (4V/A).bBJH desorption cumulative pore volume.

Table 2Energy of adsorption (�E, in kcal/mol) for different adsorbate/adsorbent (Z: zeoliteanion using cluster model).

Adsorbate �E on CuZa �E on CuCl �E on AgNO3

Thiophene 21.4 13.5 10.6Benzene 20.5 12.4 10.1Benzothiophene 22.9 16.2 12.04,6-Dimethyldibenzothiophene 23.5 18.4 13.62-Methylbenzothiophene 23.3 17.2 13.2

aCuZ: Z denotes zeolite anion using cluster model for zeolite framework.

anion (Z−) is more electronegative than Cl−, resulting in stronger�-complexation bonding with CuZ than CuCl. These results indicatethat CuY zeolite would be the best sorbent for desulfurization (with-out considering the pore diffusion/limitation effects).

4.3. Fixed-bed adsorption experiment

From Fig. 4 and Table 3, it can be seen that the AgNO3/SBA-15adsorbent was capable of removing 0.32 and 0.91mmol of sulfur pergram at breakthrough and saturation, respectively. A significant ordetectable breakthrough (∼10ppmw S) occurred at about 10.0mL/gfor 1.0 g of AgNO3/SBA-15. In comparison, the pure SBA-15 adsorbentcould only remove 0.01 and 0.17mmol of sulfur per gram at break-through and saturation, respectively. It was concluded that AgNO3was an effective selective metal salt for removing sulfur compoundsfrom jet fuel.

Fig. 4. Breakthrough of total sulfur in a fixed-bed adsorber with SBA-15 (�), MCM-41(©), monolayer AgNO3/SBA-15 (�) and monolayer AgNO3/MCM-41 (�) for JP-5 Jetfuel (1172ppmw-S) at room temperature. Ci is the total sulfur concentration of thefeed at flow rate 1.0mL/40min, based on 1g of sorbents (S.V. = 1.25h−1).

Fig. 4 also shows the desulfurization performance for JP-5 onMCM-41 and AgNO3/MCM-41 adsorbents. Table 3 summarizes theresults obtained from the corresponding breakthrough and satura-tion curves. The AgNO3/MCM-41 adsorbent exhibited higher adsorp-tion capacity that was capable of removing 0.49 and 1.0mmol ofsulfur per gram at breakthrough and saturation, respectively. As seenin Fig. 4, nearly 15.0mL of product fuel in which sulfur concentrationwas < 10.0ppmw could be obtained by the desulfurization of JP-5(1172ppmw S) with the fresh AgNO3/MCM-41 adsorbent. It can beseen that the sulfur capacity of AgNO3/MCM-41 for JP-5 was higherthan that of AgNO3/SBA-15. From Table 1, the AgNO3/MCM-41 con-tained more silver than that of AgNO3/SBA-15, 2.21 vs. 1.17mmol/g.In addition, the BET surface area of AgNO3/MCM-41 was larger thanthat of AgNO3/SBA-15, 490m2/g vs. 408m2/g. However, the averagepore size of AgNO3/MCM-41 was smaller than that of AgNO3/SBA-15,19.1Å vs. 48.8Å. The pore volume of AgNO3/MCM-41 was smallerthan that of AgNO3/SBA-15, 0.27 cm3/g vs. 0.55 cm3/g. Since the poresizes for both sorbents were considerably larger compared with thesizes of the sorbate (sulfur)molecules, the pore size difference (19.1Åvs. 48.8Å) is not a factor in determining the desulfurization perfor-mance, as we have observed previously with similar mesoporoussorbent materials (Wang et al., 2009). The higher sulfur capacityof the AgNO3/MCM-41 was likely caused by the larger BET surfacearea and higher silver loading of the adsorbent. From these results,MCM-41 was a better support than SBA-15 for the AgNO3 supportedadsorbents in the desulfurization of JP-5. This order was in agree-ment with our previous conclusion for the desulfurization of JP-5 bySBA-15 and MCM-41 supported cuprous oxide (Padin et al., 1999).Table 3 also represents the ratios of the adsorbed sulfur moleculesper silver cation at breakthrough and saturation. It can be seen thatthe ratios of sulfur to sliver metal were 0.51 and 0.45mmol/mmol forAgNO3/SBA-15 and AgNO3/MCM-41 at saturation, respectively. Thisresult indicates that during the desulfurization of JP-5, almost equalutilization of AgNO3 occurred on AgNO3/SBA-15 and AgNO3/MCM-41. This is in agreement with the above conclusion that the diffusionof sulfur compounds is not a key factor for these adsorbents.

The pore (or aperture) size of the CuY zeolite is approximately 7Å.It is similar to that of the sulfur molecules in the jet fuel (i.e., alkyl-benzothiophenes). The strong diffusion limitation of desulfurization

5244 H. Chen et al. / Chemical Engineering Science 64 (2009) 5240 -- 5246

Table 3Sulfur capacity at breakthrough and saturation over fresh adsorbents for JP-5 (1172ppmw S).

Adsorbent Adsorption capacity (mmol S/g) Ratio of sulfur on metal (mmol/mmol)b

Breakthrougha Saturation Breakthrougha Saturation

SBA-15 0.01 0.17 – –AgNO3/SBA-15 0.32 0.91 0.18 0.51MCM-41 0.02 0.22 – –AgNO3/MCM-41 0.49 1.0 0.22 0.45

aThe concentration of sulfur was ∼10ppmw S at breakthrough.b1.77mmol/g of silver on SBA-15 adsorbent or 2.21mmol/g of silver on MCM-41 adsorbent.

Fig. 5. Breakthrough of total sulfur in a fixed-bed adsorber with Cu(I)Y andAgNO3/MCM-41 for Model Jet fuel (150ppmw-S) at room temperature. Ci is thetotal sulfur concentration of the feed at flow rate 1.0mL/40min, based on 1g ofsorbents (S.V. = 1.25h−1).

was manifested by the long residence time that were needed fordesulfurization. The strong diffusion limitation has also been shownby the strong dependence of desulfurization on the flow rates (Liet al., 2006).

The desulfurization results on Cu(I)Y (VPIE) and AgNO3/MCM-41were also compared as shown in Figs. 5–7. Fig. 5 gives a direct com-parison between these two sorbents for themodel jet fuel. Themodeljet fuel contained only benzothiophene and methyl-benzothiophene(at 75ppmw-S each). As seen from Fig. 1, the main sulfur com-pounds in the JP-5 were alkylated benzothiophenes withmuch largermolecules. As shown above, Cu+ forms stronger bonds with the sul-fur molecules than Ag+. For the model fuel, the sulfur moleculeswere sufficiently small compare to the aperture size of the Cu(I)Yzeolite, hence diffusion was not a factor for the desulfurization per-formance. As a result, the sulfur capacity of Cu(I)Y was higher thanthat of AgNO3/MCM-41 for the model fuel, as shown in Fig. 5.

Figs. 6 and 7 show the sulfur breakthrough curves for the lightfraction of JP-5 (841ppmw-S) on these two sorbents. It is seen thatthe desulfurization capacity of the AgNO3/MCM-41 was much higherthan that of Cu(I)Y zeolite. This result is a manifestation of the porediffusion limitation for the sulfur molecules in the Cu(I)Y zeolite(∼7Å pores), as compared with the case of AgNO3/MCM-41 with∼19Å pores. The sulfur molecules in the JP-5 were mostly dimethyl-benzothiophene, trimethylbenzothiophene and a small amountof tetramethylbenzothiophene, while that in model jet fuel werebenzothiophene and methylbenzothiophene. So the larger sulfur

Fig. 6. Breakthrough of total sulfur in a fixed-bed adsorber with AgNO3/MCM-41for JP-5 light fraction (841ppmw-S) at room temperature. Ci is the total sulfurconcentration of the feed at flow rate 1.0mL/40min, based on 1g of sorbents(S.V. = 1.25h−1).

molecules in JP-5 light fraction could diffuse more freely in thelarge pores of AgNO3/MCM41 and not so in the much smaller poresof Cu(I)Y. As a result, AgNO3/MCM41 showed better desulfurizationresult for JP-5.

4.4. Adsorbent regeneration

After saturation by JP-5, regeneration of the spent AgNO3/MCM-41 was conducted in situ by heating in flowing air at 200 ◦C for24h. Fig. 8 shows the total sulfur breakthrough curves for JP-5 (1172ppmw S) with fresh AgNO3/MCM-41 and regeneratedAgNO3/MCM-41 sorbents. Three desulfurization–regeneration cy-cles were performed. There was some loss of the sulfur capacityafter the first cycle. However, there was no further loss after the sec-ond cycle, as shown in Fig. 8. The sulfur capacities at breakthroughand saturation over the fresh sorbent and regenerated sorbents arelisted in Table 4. The results show that approximately 0.16mmol ofsulfur per gram at breakthrough can be removed, for the second andthird regeneration cycles. From Fig. 8, more than 5.0ml of “sulfurfree” (well below 1ppm) fuel can be obtained over 1.0 g of regener-ated AgNO3/MCM-41 sorbents after the second regeneration cycle.Nearly 50% total sulfur capacity of fresh sorbent could be recoveredat saturation.

As mentioned, metal halides supported on MCM-41 and SBA-15(Wang et al., 2008) and Cu2O supported on the same meso-porous silics supports (Wang et al., 2009) have been studied for the

H. Chen et al. / Chemical Engineering Science 64 (2009) 5240 -- 5246 5245

Fig. 7. Breakthrough of total sulfur in a fixed-bed adsorber with Cu(I)-Y (VPIE), forJP-5 light fraction (841ppmw-S) at room temperature. Ci is the total sulfur concen-tration of the feed at flow rate 1mL/20min, based on 1g of sorbents (S.V. = 2.5h−1).

Fig. 8. Breakthrough of total sulfur in a fixed-bed adsorber with fresh and regeneratedAgNO3/MCM-41 for JP-5 Jet fuel (1172ppmw-S) at room temperature. Ci is the totalsulfur concentration of the feed at flow rate 1.0mL/40min, based on 1g of sorbents(S.V. = 1.25h−1).

Table 4Sulfur capacity at breakthrough and saturation over regenerated AgNO3/MCM-41for JP-5 (1172ppmw-S).

Adsorbent Breakthroughloading(mmol/g)a

Saturationloadingv(mmol/g)

Fresh 0.49 1.01st regenerated AgNO3/MCM-41 0.16 0.502nd regenerated AgNO3/MCM-41 0.16 0.52

aThe concentration of sulfur was ∼10ppmw S at breakthrough.

desulfurization of JP-5. A comparison of the results showed thatAgNO3/MCM-41 was superior because it yielded the largest regen-erable “sulfur free” fuel product. It is concluded that the AgNO3/

MCM-41 sorbent is the best sorbent for desulfurization of highsulfur jet fuels.

5. Conclusion

Desulfurization of a JP-5 (1172ppmw S) for on-board fuel cellapplications was investigated by �-complexation adsorption withSBA-15 or MCM-41 supported AgNO3. It was found that significantbreakthrough occurred near 15.0mL/g for desulfurization of the JP-5by AgNO3/MCM-41. The spent AgNO3/MCM-41 was regenerated byheating in air at 200 ◦C, and the regenerated adsorbent was tested forre-use. The results showed that ∼50% of the sulfur capacity could berecovered for re-use. Compared with all sorbents developed for thehigh sulfur jet fuels, AgNO3/MCM-41 showed the best desulfurizationperformance.

Acknowledgments

This work was supported by NSF Grant CBET 0852129 and Officeof Naval Research via NAVSEA. Discussions with Mr. John M. Heinzeland Dr. Donald J. Hoffman were helpful.

References

Barrett, E., Joyner, L., Halenda, P., 1951. The determination of pore volume and areadistributions in porous substances. I. Computations from nitrogen isotherms.Journal of the American Chemical Society 73, 373–380.

Chica, A., Strohmaier, K., Iglesia, E., 2004. Adsorption, desorption, and conversion ofthiophene on H-ZSM5. Langmuir 20, 10982–10991.

Etemadi, O., Yen, T.F., 2007. Surface characterization of adsorbents in ultrasound-assisted oxidative desulfurization process of fossil fuels. Journal of Colloid andInterface Science 313, 18–25.

Frisch, M.J., Trucks, G.W., Schlegel, H.B., 2004. Gaussian, Inc., Wallingford CT.Haji, S., Erkey, C., 2003. Removal of dibenzothiophene from model diesel by

adsorption on carbon aerogels for fuel cell applications. Industrial andEngineering Chemistry Research 42, 6933–6937.

Heinzel, J., Cervi, M., Hoffman, D., Nickens, A., Kuseian, J., 2005. Fuel cell systemmodels for US navy shipboard application. In: USFCC Fuel Cell Seminar, PalmSprings, CA.

Heinzel, J., Cervi, M., Hoffman, D., Nickens, A., 2006. Advances in Logistic FuelsDesulfurization. American Chemical Society, San Francisco, CA.

Hernández-Maldonado, A.J., Yang, R.T., 2004a. Desulfurization of transportation fuelsby adsorption. Catalysis Reviews—Science and Engineering 46, 111–150.

Hernández-Maldonado, A.J., Yang, R.T., 2004b. Desulfurization of diesel fuels byadsorption via �-complexation with vapor-phase exchanged Cu(I)-Y zeolites.Journal of the American Chemical Society 126, 992–993.

Hernández-Maldonado, A.J., Qi, G.S., Yang, R.T., 2005. Desulfurization of commercialfuels by �-complexation: monolayer CuCl/�-Al2O3. Applied Catalysis B 61,212–218.

Jeevanandam, P., Klabunde, K.J., Tetzler, S.H., 2005. Adsorption of thiophenes outof hydrocarbons using metal impregnated nanocrystalline aluminum oxide.Microporous and Mesoporous Materials 79, 101–110.

Jeon, H.J., Ko, C.H., Kim, S.H., Kim, J.N., 2009. Removal of refractory sulfur compoundsin diesel using activated carbon with controlled porosity. Energy Fuels 23,2537–2543.

Jiang, M., Ng, F.T.T., 2006. Adsorption of benzothiophene on Y zeolites investigatedby infrared spectroscopy and flow calorimetry. Catalysis Today 116, 530–536.

Kim, J.H., Ma, X.L., Zhou, A.N., Song, C.S., 2006. Ultra-deep desulfurization anddenitrogenation of diesel fuel by selective adsorption over three differentadsorbents: a study on adsorptive selectivity and mechanism. Catalysis Today111, 74–83.

King, D.L., Faz, C., 2006. Desulfurization of Tier 2 gasoline by divalent copper-exchanged zeolite Y. Applied Catalysis A 311, 58–65.

Ko, C.H., Park, J.G., Park, J.C., Song, H., Han, S.S., Kim, J.N., 2007. Surface status andsize influences of nickel nanoparticles on sulfur compound adsorption. AppliedSurface Science 253, 5864–5867.

Laborde-Boutet, C., Joly, G., Nicolaos, A., Thomas, M., Magnoux, P., 2006. Selectivity ofthiophene/toluene competitive adsorptions onto NaY and NaX zeolites. Industrialand Engineering Chemistry Research 45, 6758–6764.

Landau, M.V., Herskowitz, M., Agnihotri, R., Kegerreis, J.E., 2008. Ultradeepadsorption–desulfurization of Gasoline with Ni/Al–SiO2 material catalyticallyfacilitated by ethanol. Industrial and Engineering Chemistry Research 47,6904–6916.

Landau, M.V., Herskowitz, M., Hoffman, T., Fuks, D., Liverts, E., Vingurt, D., Froumin,N., 2009. Ultradeep hydrodesulfurization and adsorptive desulfurization of dieselfuel on metal-rich nickel phosphides. Industrial and Engineering ChemistryResearch 48, 5239–5249.

5246 H. Chen et al. / Chemical Engineering Science 64 (2009) 5240 -- 5246

Li, Y.W., Yang, F.H., Qi, G.S., Yang, R.T., 2006. Effects of oxygenates and moistureon adsorptive desulfurization of liquid fuels with Cu(I)Y zeolite. Catalysis Today116, 512–518.

Ma, X.L., Sun, L., Song, C.S., 2002. A new approach to deep desulfurization of gasoline,diesel fuel and Jet fuel by selective adsorption for ultra-clean fuels and for fuelcell applications. Catalysis Today 77, 107–116.

McKinley, S.G., Angelici, R.J., 2003. Deep desulfurization by selective adsorptionof dibenzothiophenes on Ag+/SBA-15 and Ag+/SiO2. Chemical Communications,2620–2621.

Nuntang, S., Prasassarakich, P., Ngamcharussrivichai, C., 2008. Comparative study onadsorptive removal of thiophenic sulfurs over Y and USY zeolites. Industrial andEngineering Chemistry Research 47, 7405–7413.

Padin, J., Yang, R.T., Munson, C.L., 1999. New sorbents for olefin/paraffin separationsand olefin purification for C4 hydrocarbons. Industrial and Engineering ChemistryResearch 38, 3614–3621.

Padin, J., Yang, R.T., 2000. New sorbents for olefin–paraffin separations by adsorptionvia �-complexation: synthesis and effects of substrates. Chemical EngineeringScience 55, 2607–2616.

Russo, T.V., Martin, R.L., Hay, P.J., 1995. Effective core potentials for DFT calculations.Journal of Physical Chemistry 99, 17085–17087.

Seredych, M., Bandosz, T.J., 2007. Template-derived mesoporous carbons withhighly dispersed transition metals as media for the reactive adsorption ofdibenzothiophene. Langmuir 23, 6033–6041.

Wang, Y.H., Yang, F.H., Yang, R.T., Heinzel, J.M., Nickens, A.D., 2006.Desulfurization of high-sulfur jet fuel by Pi-complexation with copper andpalladium halide sorbents. Industrial and Engineering Chemistry Research 45,7649–7655.

Wang, Y.H., Yang, R.T., 2007. Desulfurization of liquid fuels by adsorption on carbon-based sorbents and ultrasound-assisted sorbent regeneration. Langmuir 23,3825–3831.

Wang, Y.H., Yang, R.T., Heinzel, J.M., 2008. Desulfurization of jet fuel by �-complexation with metal halides supported on MCM-41 and SBA-15 mesoporousmaterials. Chemical Engineering Science 63, 356–365.

Wang, Y.H., Yang, R.T., Heinzel, J.M., 2009. Desulfurization of jet fuel JP-5 by MCM-41 and SBA-15 supported cuprous oxide for fuel cell applications. Industrial andEngineering Chemistry Research 48, 142–147.

Xie, Y.C., Tang, Y.Q., 1990. Spontaneous monolayer dispersion of oxides and saltsonto surfaces of supports: applications to heterogeneous catalysis. Advances inCatalysis 37, 1–43.

Yang, F.H., Hernández-Maldonado, A.J., Yang, R.T., 2005. Selective adsorptionof organosulfur compounds from transportation fuels by �-complexation.Separation Science Technology 39, 1717–1732.

Yang, H.Y., Sothen, R., Cahela, D.R., Tatarchuk, B.J., 2008. Breakthrough characteristicsof reformate desulfurization using ZnO sorbents for logistic fuel cell powersystems. Industrial and Engineering Chemistry Research 47, 10064–10070.

Yang, L.M., Wang, Y.J., Huang, D., Luo, G.S., Dai, Y.Y., 2007a. Preparation of highperformance adsorbents by functionalizing mesostructured silica spheres forselective adsorption of organosulfur compounds. Industrial and EngineeringChemistry Research 46, 579–583.

Yang, R.T., Hernandez-Maldonado, A.J., Yang, F.H., 2003. Desulfurization oftransportation fuels with zeolites under ambient conditions. Science 301, 79–81.

Yang, X.X., Cao, C.D., Klabunde, K.J., Hohn, K.L., Erickson, L.E., 2007b. Adsorptivedesulfurization with xerogel-derived zinc-based nanocrystalline aluminum oxide.Industrial and Engineering Chemistry Research 46, 4819–4823.