Fuel 89 (2010) 3218–3225

Contents lists available at ScienceDirect

Fuel

journal homepage: www.elsevier .com/locate / fuel

Supported silver adsorbents for selective removal of sulfur speciesfrom hydrocarbon fuels

Sachin Nair, Bruce J. Tatarchuk *

Department of Chemical Engineering, Auburn University, Auburn, AL 36849, USA

a r t i c l e i n f o a b s t r a c t

Article history:Received 9 February 2010Received in revised form 23 April 2010Accepted 9 May 2010Available online 20 May 2010

Keywords:DesulfurizationSulfur sorbentOxygen chemisorptionRefined fuels

0016-2361/$ - see front matter � 2010 Elsevier Ltd. Adoi:10.1016/j.fuel.2010.05.006

* Corresponding author. Tel.: +1 334 844 2023; faxE-mail address: tatarbj@auburn.edu (B.J. Tatarchuk

Dispersed silver oxides on supports such as TiO2, c-Al2O3 and SiO2 were observed to be effective desul-furizing agents for refined fuels at ambient conditions. TiO2 was determined to be the most stable supportfor silver oxide. Ag (4 wt%)/TiO2 demonstrated a saturation sulfur capacity of 6.3 mgS/g for JP5 fuel con-taining 1172 ppmw sulfur. This high affinity for sulfur translated to one sulfur heterocycle associatedwith every two surface Ag atoms in the sorbent even in the presence of a 160-fold excess of other aro-matics found in the fuel. A unique attribute of these sorbents was that they were thermally regenerableat 450 �C using air as a stripping medium over multiple cycles. Desulfurization characteristics also variedwith fuel composition. Variation in desulfurization performance between JP5, JP8 and a light fraction JP5were established and associated with the differences in sulfur composition of these fuels. The effects ofsurface area, porosity and crystal structure of the sorbent on sulfur capacity are also presented.

� 2010 Elsevier Ltd. All rights reserved.

1. Introduction

Sulfur content in transportation fuels have been continuallyregulated around the world due to its environmental effects as wellas the effects on engines and catalytic systems. Sulfur removal hasalso gained prominence in recent years due to the need for sulfur-free reformation fuels for various applications including fuel cells.Under the Tier 2 Vehicle and Gasoline sulfur program, the EPArequires total sulfur content of gasoline to be limited to 30ppmwand 15ppmw for diesel. Standards developed by the Europeanstandards organization (CEN) mandates a maximum sulfur contentof 10 ppm by 2009. Military logistic fuels such as JP5 and JP8 arenot bound by these regulations. It is thus not feasible to use refor-mate gases derived from these fuels in fuel cell systems withoutsome kind of sulfur abatement technology.

Hydrodesulfurization (HDS) has been the most widely used andeffective sulfur abatement technology in refining. Even though themajority of low-sulfur hydrocarbon fuels are derived through HDS,there are limitations. Production of ultra low-sulfur fuels requirecatalyst volumes significantly larger than presently employedusing the known reaction pathways for hydrotreating [1]. Adosorp-tive desulfurizing units can provide low sulfur fuel for sulfur intol-erant systems such as fuel cells and catalyst beds. Operability of adesulfurizer at ambient conditions without the requirement forhydrogen provides many advantages over conventional systems.The use of sorptive systems in combination with traditional HDS

ll rights reserved.

: +1 334 844 2085.).

units would reduce the costs of retrofitting them when lower sul-fur standards are enacted. Some of the resulting additional operat-ing costs may also be reduced. Several emergent technologies havediverged from HDS to provide low sulfur products. Adsorption orsorption, catalytic oxidation and pervaporation are among themost promising.

Catalytic oxidation and extraction are being pursued as anotheralternative to sorptive sulfur removal [2–5]. Biological sulfurremoval has also been demonstrated using several varieties ofmicroorganisms [6–10]. Treatises that detail various deep desul-furization technologies is dealt with elsewhere [11,12]. Oxidationof sulfur species in liquid hydrocarbon feeds have been carriedout through the use of peroxide solutions [3,13,14], alkanols andmolecular oxygen [15] followed by purification steps. The effec-tiveness of these processes have been enhanced by the use of ultra-sound [5]. Pervaporation techniques have achieved significantprominence in the recent years with the advancement in mem-brane technology [16–19]. Since the development of the S-Braneprocess by Grace Davison in 2002 [20], several manufacturers havetaken this path to desulfurization.

Several adsorbents have been reported to have an excellentcapacity for sulfur removal. A majority of these adsorbents havetransition metal components. Transition metal oxides have beenreported to be effective desulfurization agents [21–28]. Nickelhas been shown to be effective in its reduced metallic form[29–31]. Supported chloride salts of copper and palladium havealso been effectively demonstrated [32,33]. Selective desulfuriza-tion agents have been developed through ion exchange of Cu, Ag,Ce, Ni and other metal ions into zeolite structures [34–40]. The

Table 1Properties of supports used for the preparation of silver sorbents.

Support Vendor Grade BET surface area(m2/g)

Pore volume(ml/g)

TiO2 St. GobainNor Pro

Type 1 153 0.46

SiO2 GraceDavison

Grade 21 319 0.80

c-Al2O3 Alfa Aesar Catalystsupport

256 1.2

S. Nair, B.J. Tatarchuk / Fuel 89 (2010) 3218–3225 3219

active metal components have been supported on silica, alumina[41–43] and activated carbon [41,43]. ZnO based sorbents wereinitially commercialized in the ZSorb process developed by Philipspetroleum corporation [44] followed by improvements from sev-eral researchers [27,45].

The affinity of silver for sulfur is evident from the fact that silvertarnishes due to the formation of multi-layer silver sulfide at roomtemperature. Several other researchers and inventors have utilizedthis fact for applications including sulfur removal from liquid fuels[46–50]. Silver based catalysts have been widely employed for oxi-dation reactions such as ethylene oxidation, propene epoxidationand oxidative dehydrogenation [51–56]. These catalysts have beencharacterized in great detail. The surface morphology of silver cat-alysts varies extensively between the methods used for prepara-tion, weight loading and the supports used. X-ray diffraction hasbeen used with limited success on supported silver catalysts. Dif-fraction patterns for silver metal or silver oxides were not discern-able at low loadings [55,57,58]. This has been attributed to theoverlapping of the Ag peaks to that of the support materials[55,58]. Dispersion of transition metal ions on supports are rou-tinely accomplished by chemisorption of gases, such as H2 andCO. This technique is difficult with Ag because H2 and CO do notadsorb strongly and reversibly to provide well-defined monolayercoverages [59]. Oxygen has been successfully chemisorbed on Agresulting in reproducible dispersion measurements. The formationof a monolayer of oxygen required for accurate measurements onreduced Ag surfaces have been reported between 170 and 220 �Cwith a stoichiometry of one oxygen atom to one silver atom[59–64].

Adsorptive desulfurization processes address some of the inad-equacies associated with the HDS process. Sorptive sulfur unitsoperating at ambient conditions in the absence of hydrogen pro-vide a cost effective and scalable alternative. Sorbents maybedeveloped to selectively capture sulfur heterocycles with alkyl sidechains that have been shown to be resistant to hydrodesulfuriza-tion [65–67]. Thus the process is most suited for fuel cell applica-tions where reformation is required at the point of use. Supportedsilver oxides were tested as desulfurization compositions for high-sulfur fuels and model fuel compositions. The effect of the varia-tion in dispersion of Ag oxides on desulfurization performancewas observed using TiO2, c-Al2O3 and SiO2 as supports. Sorbentswith varying support pore structure and Ag loading were tested.Facile regeneration was carried out in air for 10 cycles. Variationof sorbent performance with fuel chemistry was also examined.

The adaptation of Ag/TiO2 adsorbents to commercial applica-tions would warrant consideration of risk associated with toxicol-ogy of these materials. The effect of Ag maybe considerednegligible due to low loadings and a minimum probability of themetal to dislodge from TiO2. Therefore the most risk associatedwith exposure may result from inhalation of TiO2 particles duringhandling. TiO2 has been shown to be quite benign by various tox-icology studies [68–70]. These materials may therefore be handledwith normal precautions taken with catalytic/adsorbents.

Table 2Surface pre-treatment steps for Ag/TiO2 sorbents for oxygen chemisorption.

Pre-treatmentstep

Temperature(�C) Time(min) Conditions

Surfacecleaning/moistureremoval

150 30 Vacuum(3.99 � 10�11 kPa)

Reduction 300 60 Hydrogen(101.32 kPa)Removal of

physicallyadsorbed H2

300 60 Vacuum(3.99 � 10�11 kPa)

2. Experimental

2.1. Sorbent preparation

TiO2 support was obtained from St. Gobain Norpro GradeST61120 (Type 1) as 3.2 mm pellets. Grade 21 SiO2 obtained fromGrace Davison Co. and catalyst support Grade c-Al2O3 obtainedfrom Alfa Aesar were also used as supports (Table 1). The pelletswere crushed and sieved to size and dried in a convection ovenfor at least 6 h at 110 �C prior to use. Incipient wetness impregna-tion was used to disperse the metals on the supports using aque-

ous solutions of Cu (NO3)2, Mn(NO3)2, Ni(NO3)2 and Co(NO3)2. Allthe nitrates used were of 99.9% purity obtained from Alfa AesarCo. The concentration of the impregnating solution was adjustedto obtain the required metal loading on the support. Typicallythe volume of impregnating solution was maintained at 90% ofthe pore volume of the support. The resulting particles were thendried at 110 �C for 6 h followed by calcination in air at 400 �C for2 h. A maximum of 15% variation in sulfur capacity was observedbetween batches of sorbent of the same composition. Therefore,for every case presented in this work, the same batch of sorbentwas used to provide a consistent basis for comparison.

2.2. Sorbent characterization

BET surface areas, pore volumes and pore size distributionswere obtained using a Quantachrome AS1 surface area and poresize analyzer using nitrogen adsorption at 77 K. The QuantachromeAS1 automated surface analysis module was used to carry outselective oxygen chemisorption on the sorbent. The instrumentutilized the static chemisorption technique. Selective oxygenchemisorption was used to determine the morphology of the sup-ported Ag sorbent. A clean Ag surface was obtained by following aseries of in situ pre-treatment steps (Table 2). The primary step wasan evacuation step followed by hydrogen reduction to providereproducible O2 uptakes. All surface Ag species are assumed tobe in the reduced metallic state prior to uptake measurements. Iso-therms were generated from O2 uptake at 170 �C. Oxygen uptake atzero relative pressure was obtained by extrapolating the linear partof the isotherm to the abscissa. Physical or weak adsorption wasobserved to be negligible at 170 �C. Therefore the combined oxy-gen uptake (physisorbed and chemisorbed) was considered for dis-persion estimations. Dispersion (D) was estimated from theexpression:

D ¼ NmSM100L

where Nm represented the oxygen uptake (lmol/g), S the adsorptionstoichiometry, M the molecular weight of metal and L the percent-age metal loading. The active metal surface area (s) was estimatedfrom the expression:

Fig. 1. Schematic of laboratory scale desulfurization unit used for adsorption–regeneration studies.

0

1

2

3

4

5

Mn Co Ni Cu Ag

Satu

ratio

n Su

lfur C

apac

ity [m

g/g]

Metal

SiO

Fig. 2. Saturation sulfur capacities of 4.0 wt% transition metal ions supported onSiO2 determined using JP5 fuel with 1172 ppmw sulfur.

3220 S. Nair, B.J. Tatarchuk / Fuel 89 (2010) 3218–3225

s ¼ NmSAm

166

Am being the cross-sectional area occupied by each active surfaceatom (8.6960 Å2/Ag atom)

The average crystallite size was estimated from the expression:

d ¼ 100LfsZ

Here Z represented density of Ag (10.5 g/ml), s the active metalsurface area and f the shape factor (6 for the assumed sphericalshape). Comprehensive treatises on chemisorption of gases on dis-persed metals are also available [71,72].

2.3. Analysis of sulfur

A Varian CP3800GC equipped with a Pulsed Flame PhotometricDetector (PFPD) containing a sulfur specific optical filter was usedto determine the sulfur content of the sorbent bed outlet. The GCcolumn employed was a Restek crossboard column of length30 m, inner diameter of 0.25 mm and 0.25 lm df. The PFPD detec-tor was calibrated using standards of subsequent dilutions of bothreal fuels as well as sulfur heterocycles in n-octane. The injectoroperated at split ratios between 0 and 80 using a 1 ll injection vol-ume. The lower detection limit of the PFPD was observed to be20 ppbw total sulfur with the injector in splitless mode. Sulfur con-tent was also determined using an Antek 9000VS Total Sulfur Ana-lyzer. The instrument was calibrated using sulfur standards asmentioned above. The lower detection limit of the instrumentwas observed to be 200 ppbw.

2.4. Saturation sulfur capacity

Preliminary analysis of sulfur capacity was carried out throughsaturation tests wherein a known mass of the sorbent compositionwas agitated gently in a known weight of JP5 fuel for 48 h. An esti-mation of the sulfur capacity was thus obtained from the sulfurcontent of the resulting fuel.

2.5. Sorbent breakthrough performance

The breakthrough characteristics of the sorbents were deter-mined in a packed column configuration where the challenge fuelflowed from the bottom of the bed to the top. It was observed thatthis configuration minimized channeling and wall-slip effects giv-ing consistent breakthrough data. Ten grams of the sorbent wasused in all the breakthrough studies. The bed was contained inquartz tubing supported on both ends by quartz wool. The tubeID was 16 mm and the length varied depending on the sorbentcomposition as well as bed loading. The varied bed lengths em-ployed in this work are mentioned at the appropriate instances.Dead spots in the bed were avoided by tapping the bed on the sideprior to testing to ensure consistent packing. None of the sorbentswere activated in situ nor was any step followed to wash the bedwith a sulfur-free solvent to remove trapped air. The fuel flow ratewas maintained by peristaltic pump typically at 0.5 ml/min(LHSV � 2 h�1). The bed output was sampled at regular intervalsfor analysis of sulfur content. The adsorption/regeneration systemis shown schematically in Fig. 1. The concept of t1/2 was used toestimate the saturation capacity for cases where sorption wasnot carried out to bed exhaustion.

2.6. Challenge fuels

JP5 with a total sulfur content of 1172 ppmw and light fractionJP5 with a sulfur content of 582 ppmw were obtained from NAV-

SEA Philadelphia. JP8 fuel was obtained from TARDEK and had a to-tal sulfur content of 630 ppmw sulfur. A model fuel consisting ofapproximately 3500 ppmw Benzothiophene (98% Alfa Aesar Co.)in n-octane was used as a model fuel for some of the breakthroughtests. Sharper breakthroughs obtained using model fuels were use-ful in identifying the differences in breakthrough performance ofsorbents compared to real fuels. Model fuels also reduced the ef-fects of competitive adsorption posed by sulfur-free aromaticspresent in natural fuels. The major sulfur species in the fuels wereidentified using analytical standards obtained from Chiron AS.

3. Results and discussion

3.1. Sorbent formulation

Preliminary data on sulfur removal performance with JP5 fuelwere obtained from saturation tests carried out by contacting sor-bent compositions with fuel for 48 h to give sufficient time todetermine equilibrium sulfur capacity. Cu, Ni, Mn, Co and Ag werethe transition metals supported on Grade 21 SiO2. Since thermalregeneration in air is ideal for multi-cycle operation, these compo-sitions were tested for sulfur capacity in their oxidized form. Onegram of sorbent was contacted with 5.0 g of JP5 (1172 ppmw sul-fur) and agitated mechanically for 48 h. Saturation sulfur capacitywas estimated from sulfur content of the remaining fuel. The sulfur

S. Nair, B.J. Tatarchuk / Fuel 89 (2010) 3218–3225 3221

capacities of the compositions are shown in Fig. 2. It was observedthat Ag demonstrated the highest sulfur capacity of 4.95 mgS/g. Inorder to explain the high affinity of Ag for sulfur heterocycles, sev-eral aspects of the system need analysis such as the composition ofAg phase at the interface with respect to oxidation state and sur-face morphology, the presence and contribution of surface func-tional groups among others. Chemical characterization andmechanistic details of sulfur removal is under investigation. Froman application perspective, the observed higher sulfur capacity ofsupported silver warranted the following work.

3.1.1. Effect of supportAg (4 wt%)/SiO2 showed the highest affinity for sulfur among

Mn, Co, Ni and Cu. Ag was further supported on TiO2 and c-Al2O3

0

200

400

600

800

1000

1200

0 50 100 150 200

Out

let t

otal

sul

fur c

once

ntra

tion

[ppm

w]

Time [min]

Ag(4%)/Al O

Ag(4%)/SiO

Ag(4%)/TiO

Fig. 3. Breakthrough characteristics of 4 wt% Ag supported on TiO2, c-Al2O3 andSiO2 for JP5 with a total sulfur content of 1172 ppmw sulfur.

Fig. 4. Oxygen uptake of 4.0 wt% Ag/TiO2, Ag/SiO2, and Ag/c-Al2O3.

Table 3Pore structure and Ag dispersion of 4 wt% Ag/TiO2, Ag/SiO2 and Ag/c-Al2O3.

Sorbent/support BET surface area(m2/g) Pore volume(ml/g) A

Ag (4%)/TiO2 114.2 0.27 6.Ag (4%)/c-Al2O3 252.5 0.56 4.Ag (4%)/SiO2 267.0 0.50 3.

in order to identify most effective support for desulfurization of li-quid fuels. The nature of the support influences the dispersion ofsilver besides being involved in strong interactions with Ag. Theyhave also been reported to effect the activity of Ag based catalystsfor ethylene as well as propylene oxidation [73]. Such metal sup-port interactions are likely to influence the sulfur capacity of thesesorbents and their stability during thermal cycling. High tempera-ture thermal treatment of TiO2 and c-Al2O3 in air improved sulfurcapacity significantly compared to support particles dried at110 �C. This was demonstrated in breakthrough experiments car-ried out using a model fuel consisting of 3500 ppmw benzothio-phene in octane. Blank TiO2, c-Al2O3 and SiO2 were prepared byfollowing identical steps as with the supported Ag sorbents exceptthat HNO3 of similar concentration was used for impregnation. TheTiO2 blank showed the highest capacity among the three supportsat 19.82 mg/g compared to a capacity of 17.34 mg/g demonstratedby c-Al2O3. However SiO2 support did not demonstrate thisimprovement in sulfur capacity following the thermal treatment.Thus calcination only generated active centers on the TiO2 andc-Al2O3. The calcined TiO2 support demonstrated a capacity of19.82 mg/g compared to 5.58 mg/g demonstrated by the dried sup-port. The reason for this improvement in sulfur capacity is underinvestigation.

Ag loading of 4 wt% were obtained on the three supports byvarying the concentration of the AgNO3 impregnating solutionkeeping the volume of impregnation equal to the pore volume ofthe respective supports. Breakthrough for the supported Ag sor-bents were observed using JP5 fuel with total sulfur content of1172 ppmw shown in Fig. 3. At 10 ppmw breakthrough capacity,Ag/SiO2 showed capacity of 0.49 mg/g in comparison to 1.87 byc-Al2O3 and 0.82 by TiO2. However SiO2 showed the highest satu-ration capacity of 9.74 mg/g compared to 6.97 by c-Al2O3 and 5.6by TiO2.

O2 chemisorption and N2 adsorption were used to determine Agdispersion and surface characteristics. Oxygen uptake on 4 wt% Agon TiO2, SiO2 and c-Al2O3 are shown in Fig. 4. The highest disper-sion of Ag was observed on TiO2 despite both SiO2 and c-Al2O3 hav-ing twice the BET surface area (Table 3). Even though smallercrystallites have been noted to be prone to sintering in reducingatmospheres at elevated temperatures, TiO2 was observed to be asubstantially more stable support during thermal cycling in oxidiz-ing conditions. Silver dislodged from the surface of SiO2 andc-Al2O3 was observed on the tubing as well as reactor walls aftera few cycles of operation unlike TiO2. TiO2 has been reported tobe a more stable support for Ag by other researchers as well[74]. Thus TiO2 was chosen as the primary support material forfurther desulfurization studies.

3.1.2. The effect of pore structure of TiO2

The pore structure of a sorbent affects the transport or diffusionof large aromatic sulfur species of a fuel between active sites on theinterior of the sorbent particle and bulk of the fuel. Thus transportimpacts the desulfurization performance as well as the regenera-bility of the sorbent. Similar to support chemistry, support porestructure also influences the dispersion of metals on the surface.Generally higher surface area supports tend to result in a higherdispersion of the active metals. Larger pore sizes facilitate higher

ctive metal surface area(m2/g) Dispersion% Avg. crystal size(nm)

69 34.2 3.4262 23.0 5.1195 20.3 5.79

Table 4Pore structure and dispersion of 4 wt% Ag supported on three grades of TiO2.

Type Vendor BET surfacearea(m2/g)

Pore volume(ml/g)

Avg. porediameter(nm)

Active metal surfacearea(m2/g)

Dispersion% Avg. crystallitesize(nm)

1 St. Gobain Nor Pro 114.2 0.27 9.61 6.69 34.4 3.412 Alfa Aesar 98.35 0.36 14.59 4.73 24.4 4.833 Alfa Aesar 38.48 0.23 24.02 4.61 23.7 4.96

3222 S. Nair, B.J. Tatarchuk / Fuel 89 (2010) 3218–3225

mass transfer rates for the heavy molecular weight sulfur aromat-ics. Ag (4 wt%)/TiO2 was prepared on three types of TiO2 supportswith surface characteristics as listed in Table 4.

The breakthrough performance of three sorbents using a modelfuel with approximately 3500 ppmw benzothiophene with n-oc-tane is shown in Fig. 5. Ag supported on Type 1 TiO2 demonstratedthe highest sulfur capacity of 28.74 mg/g saturation capacity fol-lowed by type 2 TiO2 at 27.87 mg/g. Type 3 TiO2 having the lowestsurface area among the three had the lowest sulfur capacity of9.29 mg/g. Thus the sulfur capacity was observed to be moredependent on of the specific surface area than the pore volumeof the sorbent. Higher surface area materials tend to have more de-fects on the surface leading to better dispersion of metals which re-sults in more active sites. The pore sizes of the three types of TiO2

used were much larger than the dimensions of the sulfur aromaticsin fuels. Thus the pore size of the support did not influence desul-furization performance of the sorbent.

3.1.3. The effect of Ag loadingThe silver loading in the sorbents tested so far were maintained

at 4.0 wt%. Since metal loading affects its dispersion and influencesthe interactions with the support, sorbents with loading varyingbetween 2 and 20 wt% were prepared and tested for desulfuriza-tion performance. JP5 fuel with a total sulfur content of

0

500

1000

1500

2000

2500

3000

3500

4000

0 100 200 300

Out

let s

ulfu

r con

cent

ratio

n [p

pmw

]

Time [min]

Grade 1Grade 2Grade 3

Fig. 5. Breakthrough performance of 4 wt% Ag supported on Grade 1 TiO2, Grade 2TiO2 and Grade 3 TiO2 using supports with different pore structure for a model fuelcontaining 3500 ppmw benzothiophene in n-octane.

Table 5Surface properties of Ag/TiO2 sorbents with different silver loadings.

Ag loading (wt%) BET surface area(m2/g) Pore volume (ml/g) A

4.00 114.2 0.27408.00 89.33 0.2303 1

12.00 79.22 0.2115 120.00 57.92 0.1331 1

1172 ppmw sulfur was used as the challenge. While Ag loadingwas varied, O2 chemisorption was used to examine the dispersionof Ag. N2 adsorption was also used to determine BET surface areaand pore volume of the sorbent. Specific surface areas and porevolumes of the sorbents decreased with Ag loading. This indicatesa progressive clogging of pores with increasing Ag loading. The ef-fect of silver loading on its dispersion is summarized in Table 5.Generally increasing metal loading has the effect of reducing itsdispersion. This is due to the agglomeration of metal atoms intocrystallites. This trend was observed in the case of Ag/TiO2 sor-bents as well. Oxygen chemisorption indicated that the averagecrystal size increased from 3.4 nm to 6.9 nm when the loadingwas increased from 4 to 20 wt% Ag while the dispersion decreasedfrom 34% to 17%.

From the breakthrough data shown in Fig. 6, the sorbents withlower Ag loadings had the highest sulfur capacity. The differencein sulfur capacity was minimal between 2% and 4% Ag loadingwith the 4% performing marginally better. The loss in capacitywas significantly larger at the higher loadings of Ag. The activeAg surface area increased from 6.69 m2/g to 14.31 m2/g between2% and 20% loadings. Sulfur capacity was lost despite this in-crease in Ag surface area. This indicates that the sulfur capacityis associated with a highly dispersed phase of Ag. Loss in sulfurcapacity at the higher loadings may also be associated with loss

ctive metal surface area(m2/g) Avg. crystal size(nm) Dispersion%

6.69 3.4 34.440.7 4.1 28.662.05 5.3 22.354.31 6.9 17.04

0

200

400

600

800

1000

1200

0 50 100 150

Out

let s

ulfu

r con

cent

ratio

n [p

pmw

]

Time [min]

Ag(2wt%)/TiOAg(4wt%)/TiOAg(8wt%)/TiOAg(12wt%)/TiOAg(20wt%)/TiO

Fig. 6. Breakthrough performance of Ag/TiO2 sorbent with the Ag loading variedbetween 2 and 20 wt% for JP5 fuel with 1172 ppmw sulfur.

5 10 15 20 25Elution Time [min]

BT2MBT

5 MBT

3,5 DMBT

2,3,4 TMBT

2.3,6TMBT

DBT

2,3 DMBT

2,5 DMBT

2,4 DMBT

2,5,7 TMBT

JP5

JP8

LJP5

Fig. 8. Chromatograms of JP5, JP8 and light fraction JP5 showing sulfur heterocyclespresent.

0

200

400

600

800

1000

1200

0 50 100 150 200

Out

let t

otal

sul

fur c

once

ntra

tion

[ppm

w]

Time [min]

JP5LJP5JP8

JP8 (693ppmw)

LJP5 (582ppmw)

JP5(1172 ppmw)

Fig. 9. Performance comparison of Ag (4 wt%)/TiO2 sorbent with JP5 (1172 ppmw),JP8 (693 ppmw) and light fraction JP5 (582 ppmw) fuels.

6

7

]

BreakthroughSaturation

S. Nair, B.J. Tatarchuk / Fuel 89 (2010) 3218–3225 3223

in active desulfurization centers on the support that are shieldedby the larger Ag crystallites.

3.2. Multi-cycle performance

Sulfur sorbents that operate in the liquid phase have relativelylow capacity. It is thus impractical to use a single use sorbent forsulfur removal from hydrocarbon fuels. Regenerability is mostcost-effectively carried out using air rather than solvents andreducing gases such as hydrogen. The Ag (4 wt%)/TiO2 sorbentwas taken through 10 cycles of adsorption in JP5 fuel, regenerationusing air. Following an adsorption step, the fluid held-up in the bedwas drained downward to the sump. Air at room temperature wasused as a blow-down medium for approximately 10 min to removefuel from the particle interstices. The bed was then heatedexternally using a furnace to a temperature of 220 �C at a rate of10 K/min and held for 1 h. This step ensured the vaporization offuel held in the pores. Bed temperature was then raised to 450 �Cat a rate of 10 K/min and held for another 2 h. The bed was readyfor the next adsorption cycle after cooling down to room tempera-ture in flowing air. The resulting breakthrough curves obtainedfollowing the 10 regeneration cycles are shown in Fig. 7. Desulfur-ization performance was uniform for the 10 cycles tested.

3.3. Effect of varying fuel composition

The sulfur content of a fuel varies depending on the origin of thecrude, refining and blending operations. Refining processes such ashydrocracking and HDS significantly influence the sulfur content.Therefore, JP5, JP8 and a light fraction JP5 with respective sulfurcontents of 1172, 630 and 582 ppmw were studied for desulfuriza-tion using the Ag (4.0 wt%)/TiO2 sorbent. PFPD chromatograms ofthese fuels are shown in Fig. 8. Sulfur content varies with thesource of the crude and between different batches of the same fuel[75]. The sulfur molecule contributing to approximately 20% of thetotal sulfur content in JP5 was 2,3-dimethyl benzothiophene. Tri-methyl benzothiophenes were found to be in higher concentrationin JP8 compared to JP5. From the chromatograms it was also ob-served that fractionation of JP5 resulted in the separation of amajority of the heavier sulfur aromatics such as 2,3,4-tri-methylbenzothiophenes (TMBT) to the heavier fraction.

Breakthrough characteristics of the fuels using Ag (4 wt%)/TiO2

sorbent are shown in Fig. 9. A summary of sulfur capacities islisted in Fig. 10. At 10 ppmw breakthrough concentration thehighest capacity was demonstrated for light fraction JP5 at4.22 mg/g. Highest saturation capacity was demonstrated for JP5

Fig. 7. Multi-cycle performance of Ag/TiO2 sorbent tested with JP5 fuel; regener-ated in air for 10 cycles.

0

1

2

3

4

5

JP8 JP5 Light JP5

Sulfu

r cap

acity

[mg/

g

Fig. 10. Sulfur capacities for Ag (4 wt%)/TiO2 sorbent for JP5, JP8 and light fractionJP5; considering breakthrough threshold at 10 ppmw.

at 6.32 mg/g. The difference in breakthrough performance be-tween JP8 and JP5 was significant. Analysis of the bed outlet after30 min. (Fig. 11) shows that the tri-methyl benzothiophenes

Fig. 11. Chromatograms of bed output after 30 min showing the higher concen-tration of tri-methyl benzohtiophenes.

3224 S. Nair, B.J. Tatarchuk / Fuel 89 (2010) 3218–3225

prevalent in JP8 were not removed by the sorbent resulting inpremature sulfur breakthrough for JP8. Thus fractionation of fuelsprovides an advantageous separation of sulfur species for adsorp-tive desulfurization.

4. Conclusions

The high affinity of Ag for sulfur observed in the tarnishing ofsilver in air was also observed to be true for sulfur in hydrocarbonfuels. The Ag (4 wt%)/TiO2 sorbent demonstrated a saturation sul-fur capacity of 6.3 mg/g for JP5 (1172 ppmw sulfur) and 2.9 mg/gfor JP8 (630 ppmw sulfur) at ambient conditions. Thermal treat-ment of acidic supports such as TiO2 and c-Al2O3 was observedto generate desulfurization centers in addition to the sulfur capac-ity demonstrated by the Ag phase. Higher sulfur capacity was dem-onstrated by sorbents with higher dispersion of Ag. Therefore theactive desulfurization centers are associated with a highly dis-persed oxide phase of Ag. Thermal regeneration conditions forthe sorbent using air were also established and demonstrated for10 cycles with minimal loss in capacity. Differences in desulfuriza-tion efficiency were observed with varying fuel chemistry and therespective sulfur aromatics involved. The sorbent showed the bestperformance with light fraction JP5 due to the lower concentrationof tri-methyl benzothiophenes. The mechanism of sulfur removalfor these sorbents is currently under investigation.

Acknowledgements

Support from the Office of Naval Research and US Army isgratefully acknowledged; also the contributions of Donald Cahela,Alexander Samokhvalov and Wenhua Zhu of the Center for Micro-fibrous Materials at Auburn University and Hongyun Yang atIntramicron Inc. for their inputs on various aspects of this work.

References

[1] Inoue S, Takatsuka T, Wada Y, Hirohama S, Ushida T. Distribution functionmodel for deep desulfurization of diesel fuel. Fuel 2000;79:9.

[2] Ali FM, Al-Malki A, El-Ali B, Martinie G, Siddiqui NM. Deep desulphurization ofgasoline and diesel fuels using non-hydrogen consuming techniques. Fuel2006;85:1354–63.

[3] Garcia-Gutierrez JL, Fuentes AG, Hernandez-Teran EM, Murrieta F, Navarrete J,Jimenez-Cruz F. Ultra-deep oxidative desulfurization of diesel fuel with H2O2

catalyzed under mild conditions by polymolybdates supported on Al2O3. ApplCatal A: Gen 2006;305:15–20.

[4] Liu S, Wang B, Cui B, Sun L. Deep desulfurization of diesel oil oxidized by Fe (VI)systems. Fuel 2008;87:422–8.

[5] Mei H, Mei WB, Yen FT. A new method for obtaining ultra-low sulfur diesel fuelvia ultrasound assisted oxidative desulfurization. Fuel 2003;82:405–14.

[6] Grossman JM, Lee KM, Prince CR, Garrett KK, George NG, Pickering JI. Microbialdesulfurization of a crude oil middle-distillate fraction: analysis of the extentof sulfur removal and the effect of removal on remaining sulfur. Appl EnvironMicrobiol 1999;65:181–8.

[7] Klein J. Biological processing of fossil fuels. Appl Microbiol Biotechnol1999;52:2–15.

[8] Ma T, Tong M, Zhang Q, Liang F, Liu R. Screening, identification of the strainFDS-1 for microbial desulfurization specially and its use in diesel oildesulfurization. Weishengwu Xuebao 2006;46:104–10.

[9] Mcfarland LB, Boron JD, Deever W, Meyer AJ, Johnson RA, Atlas MR. Biocatalyticsulfur removal from fuels: applicability for producing low sulfur gasoline. CritRev Microbiol 1998;24:99–147.

[10] Zakharyants AA, Murygina PV, Kalyuzhnyi VS. Biodesulfurization ofdibenzothiophene and its derivatives. Uspekhi Sovremennoi Biologii2005;125:104–14.

[11] Rang H, Kann J, Oja V. Advances in desulfurization research of liquid fuel. OilShale 2006;23:164–76.

[12] Song C. An overview of new approaches to deep desulfurization for ultra-cleangasoline, diesel fuel and jet fuel. Catal Today 2003;86:211–63.

[13] Campos-Martin MJ, Capel-Sanchez CM, Fierro GJ. Highly efficient deepdesulfurization of fuels by chemical oxidation. Green Chem 2004;6:557–62.

[14] Defilippis P, Scarsella M. Functionalized hexagonal mesoporous silica as anoxidizing agent for the oxidative desulfurization of organosulfur compounds.Ind Eng Chem Res 2008;47:973–5.

[15] Ma X, Zhou A, Song C. A novel method for oxidative desulfurization of liquidhydrocarbon fuels based on catalytic oxidation using molecular oxygencoupled with selective adsorption. Catal Today 2007;123:276–84.

[16] Qi Rongbin, Wang Yujun, Li Jiding, Zhao Changwei, Zhu Shenlin. Pervaporationseparation of alkane/thiophene mixtures with pdms membrane. J Membr Sci2006;280:545–52.

[17] Qi R, Zhao C, Li J, Wang Y, Zhu S. Removal of thiophenes from n-octane/thiophene mixtures by pervaporation. J Membr Sci 2006;269:94–100.

[18] Qi R, Wang Y, Chen J, Li J, Zhu S. Pervaporative desulfurization of modelgasoline with Ag2O-filled PDMS membranes. Sep Purif Technol2007;57:170–5.

[19] Qi R, Wang Y, Li J, Zhu S. Sulfur removal from gasoline by pervaporation: theeffect of hydrocarbon species. Sep Purif Technol 2006;51:258–64.

[20] Balko J, Glaser R, Wormsbecher R, Wynn W. Reduce your tier 2 gasoline sulfurcompliance costs with Grace Davison S-Brane technology. In: 2002 NPRAannual meeting AM-02-21.

[21] Sughrue EL, Johnson MM, Dodwell GW, Reed LE, Bares JE, Gislason JJ, et al.Desulfurization and sorbents for same. US patent 6656877.

[22] Gislason JJ, Schmidt R, Welch MB, Simon DE, Morton RW. Desulfurization ofcracked gasolines and diesel fuels using cadmium oxide and a promoter. USpatent application 2004007130.

[23] Price AG, Gislason JJ, Dodwell GW, Morton RW, Parks GD. Desulfurization ofhydrocarbon stream using novel compositions containing manganese oxide.US patent application 2003166464.

[24] Simon DE, Morton RW, Schmidt R, Gislason JJ, Welch MB. Reduced-valencemetal-promoted niobium oxide and tantalum oxide as petroleumdesulfurization sorbents. US patent application 2004040887.

[25] Morton RW, Gislason JJ, Welch MB, Simon DE, Schmidt R. Promoted gallium orindium oxides as bulk and supported desulfurization catalysts for petroleumfeedstocks. US patent application 2004063578.

[26] Morton RW, Gislason JJ, Schmidt R, Welch MB. Reduced-valence metal-promoted molybdenum oxide and tungsten oxide as petroleumdesulfurization sorbents. US patent application 2004040890.

[27] Khare GG. Novel sorbents for desulfurization of gasoline or diesel fuel. USpatent 6683024.

[28] Morton RW, Gislason JJ, Schmidt R, Welch MB. Desulfurization and novelcompositions for same. US patent 7220704.

[29] Kim HJ, Ma X, Zhou A, Song C. Ultra-deep desulfurization and denitrogenationof diesel fuel by selective adsorption over three different adsorbents: a studyon adsorptive selectivity and mechanism. Catal Today 2006;111:74–83.

[30] Velu S, Ma X, Song C, Namazian M, Sethuraman S, Venkataraman G.Desulfurization of JP-8 jet fuel by selective adsorption over a Ni-basedadsorbent for micro solid oxide fuel cells. Energy Fuels 2005;19:1116–25.

[31] Ma X, Sprague M, Song C. Deep desulfurization of gasoline by selectiveadsorption over nickel-based adsorbent for fuel cell applications. Ind EngChem Res 2005;44:5768–75.

[32] Hernández-Maldonado AJ, Gongshin Q, Yang RT. Desulfurization ofcommercial fuels by p-complexation: monolayer CuCl/c-Al2O3. Appl Catal B2005;61:7.

[33] Wang Y, Yang HF, Yang RT, Heinzel MJ, Nickens DA. Desulfurization of high-sulfur jet fuel by p-complexation with copper and palladium halide sorbents.Ind Eng Chem Res 2006;45:7649–55.

[34] Bhandari MV, Ko H, Chang PG, Jung H, Sang-Sup C, Soon-Haeng K, et al.Desulfurization of diesel using ion-exchanged zeolites. Chem Eng Sci2006;61:2599–608.

[35] Hernández-Maldonado AJ, Yang RT. New sorbents for desulfurization of dieselfuels via p-complexation. AIChE J 2003;50:791–801.

[36] Velu S, Ma X, Song C. Selective adsorption for removing sulfur from jet fuelover zeolite-based adsorbents. Ind Eng Chem Res 2003;42:5293–305.

[37] King LD, Li L. Removal of sulfur components from low sulfur gasoline usingcopper exchanged zeolite Y at ambient temperature. Catal Today2006;116:526–9.

S. Nair, B.J. Tatarchuk / Fuel 89 (2010) 3218–3225 3225

[38] Xue M, Chitrakar R, Sakane K, Hirotsu T, Ooi K, Yoshimura Y, et al. Preparationof cerium-loaded y-zeolites for removal of organic sulfur compounds fromhydrodesulfurizated gasoline and diesel oil. J Colloid Interface Sci2006;298:535–42.

[39] Hernández-Maldonado AJ, Yang FH, Qi G, Yang RT. Desulfurization oftransportation fuels by p-complexation sorbents: Cu(I)-, Ni(II)-, and Zn(II)-zeolites. Appl Catal B 2005;56:111–28.

[40] Hernandez-Maldonado AJ, Yang RT. Desulfurization of diesel fuels via p-complexation with nickel(ii)-exchanged x- and y-zeolites. Ind Eng Chem Res2004;43:1081–90.

[41] Kim JH, Ma X, Zhou A, Song C. Ultra-deep desulfurization anddenitrogenation of diesel fuel by selective adsorption over three differentadsorbents: a study on adsorptive selectivity and mechanism. Catal Today2006;111:74–83.

[42] Jeevanandam P, Klabunde JK, Tetzler HS. Adsorption of thiophenes out ofhydrocarbons using metal impregnated nanocrystalline aluminum oxide.Microporous Mesoporous Mater 2005;79:101–10.

[43] Hernández-Maldonado AJ, Yang RT. New sorbents for desulfurization of dieselfuels via p-complexation. AIChE J 2004;50:791–801.

[44] Phillips starts up new gasoline desulfurization unit. Oil Gas J 2001;99:1.[45] Mesters CM. Catalyst particles and its use in desulphurisation. US patent

7297655.[46] Kimberlin CN Jr., Mason RB. Desulfurization of hydrocarbons by adsorption of

sulfur compounds by metallic silver. US patent 2791540.[47] Miller JW, Ward JW. Catalytic absorbent. US patent 4582819.[48] Price AG, Gislason JJ, Dodwell GW, Morton RW. Desulfurization and novel

compositions for same. US patent 7105140.[49] Klabunde K, Sanford BR, Jeevanandam P. Inorganic and organic sulfur

compound removal from petroleum refining streams by adsorption-desulfurization over impregnated porous metal oxides. US patent application2004260139.

[50] Yang RT, Hernandez-Maldonado AJ, Yang FH. Desulfurization of transportationfuels with zeolites under ambient conditions. Science 2003;301:79–81.

[51] Natal M, Bhasin MM, Soo H, Liu AC. Supported silver oxidation catalysts forpreparation of alkylene oxides by directly oxidizing alkylenes. US patentapplication 2007123932.

[52] Verykios EX, Stein PF, Coughlin WR. Oxidation of ethylene over silver:adsorption, kinetics, catalyst. Catal Rev 1980;22:197–234.

[53] Toreis N, Verykios EX. Oxidation of ethylene over silver-based alloy catalystsand silver-gold alloys. J Catal 1987;108:161–74.

[54] Minahan DM, Hoflund GB, Epling WS, Schoenfeld DW. Study of Cs-promoted,[a]-alumina-supported silver, ethylene epoxidation catalysts: III.Characterization of Cs-promoted and nonpromoted catalysts. J Catal1997;168:393–9.

[55] Oliveira D, Lange A, Wolf A, Schüth F. Highly selective propene epoxidationwith hydrogen/oxygen mixtures over titania-supported silver catalysts. CatalLett 2001;73:157–60.

[56] Liu Q, Cao Y, Dai W, Deng J. The oxidative dehydrogenation of methanol over anovel low loading Ag/SiO2–TiO2 catalyst. Catal Lett 1998;55:87–91.

[57] Luo M, Yuan X, Zheng X. Catalyst characterization and activity of Ag–Mn, Ag–Co and Ag–Ce composite oxides for oxidation of volatile organic compounds.Appl Catal A 1998;175:121–9.

[58] Bogdanchikova N, Meunier CF, Avalos-Borja M, Breen PJ, Pestryakov A. On thenature of the silver phases of Ag/Al2O3 catalysts for reactions involving nitricoxide. Appl Catal B 2002;36:287–97.

[59] Strohmayer ED, Geoffroy LG, Vannice AM. Measurement of silver surface areaby the H2 titration of chemisorbed oxygen. Appl Catal 1983;7:189–98.

[60] Czanderna WA. The adsorption of oxygen on silver. J Phys Chem1964;68:2765–72.

[61] Yeung LK, Gavriilidis A, Varma A, Bhasin MM. Effects of 1,2 dichloroethaneaddition on the optimal silver catalyst distribution in pellets for epoxidation ofethylene. J Catal 1998;174:1–12.

[62] Verykios EX, Stein PF, Coughlin WR. Influence of metal crystallite size andmorphology on selectivity and activity of ethylene oxidation catalyzed bysupported silver. J Catal 1980;66:368–82.

[63] Smeltzer WW, Tollefson LE, Cambron A. Adsorption of oxygen by a silvercatalyst. Can J Chem 1956;34:1046–60.

[64] Kholyavenko MK, Rubanik YM, Chernukhina AN. Chemisorption method forthe determination of the surface area of Ag on a carrier. Kinet Katal1964;5:505–12.

[65] Gates CB, Topsoe H. Reactivities in deep catalytic hydrodesulfurization:challenges, opportunities, and the importance of 4-methyldibenzothiopheneand 4, 6-dimethyldibenzothiophene. Polyhedron 1997;16:3213–7.

[66] Ma X, Sakanishi K, Isoda T, Mochida I. Quantum chemical calculation on thedesulfurization reactivities of heterocyclic sulfur compounds. Energy Fuels1995;9:33–7.

[67] Ma X, Sakanishi K, Mochida I. Hydrodesulfurization reactivities of varioussulfur compounds in diesel fuel. Ind Eng Chem Res 1994;33:218–22.

[68] Wang J, Zhou G, Chen C, Yu H, Wang T, Ma Y, et al. Acute toxicity andbiodistribution of different sized titanium dioxide particles in mice after oraladministration. Toxicol Lett 2007;168:176–85.

[69] Gurr J, Wang S, Chen AS, Chien-Hung JK. Ultrafine titanium dioxide particles inthe absence of photoactivation can induce oxidative damage to humanbronchial epithelial cells. Toxicology 2005;213:66–73.

[70] Lee PK, Trochimowicz JH, Reinhardt FC. Pulmonary response of rats exposed totitanium dioxide (TiO2) by inhalation for two years. Toxicol Appl Pharmacol1985;79:179–92.

[71] Wachs EI. Characterization of catalytic materials. Massachusetts: Butterworth-Heinemann, Boston; 1992.

[72] Trapnell W. Chemisorption. New York: Academic Press; 1955.[73] Sullivan D, Hooks P, Mier M, Hal WJ, Zhang X. Effect of support and preparation

on silver-based direct propylene epoxidation catalyst. Top Catal 2006;38:303–8.[74] Seyedmonir RS, Strohmayer DE, Guskey GJ, Geoffroy GL, Vannice MA.

Characterization of supported silver catalysts. III: effects of support,pretreatment, and gaseous environment on the dispersion of Ag. J Catal1985;93:288–302.

[75] Edwards T. Liquid fuels and propellants for aerospace propulsion: 1903–2003.J Propul Power 2003;19:1089–107.