Journal of Colloid and Interface Science 372 (2012) 121–129

Contents lists available at SciVerse ScienceDirect

Journal of Colloid and Interface Science

www.elsevier .com/locate / jc is

A comparative kinetic study on ultra-deep hydrodesulfurization of pre-treatedgas oil over nanosized MoS2, CoMo-sulfide, and commercial CoMo/Al2O3 catalysts

Hamdy Farag a,b,⇑, Isao Mochida c

a Department of Material Process Engineering, Graduate School of Engineering, Kyushu University, Motooka 744, Fukuoka 819-0395, Japanb Chemistry Department, Faculty of Science, Mansoura University, 35516, Egyptc Institute for Materials Chemistry and Engineering, Kyushu University, Fukuoka 816-8580, Japan

a r t i c l e i n f o

Article history:Received 16 November 2011Accepted 9 January 2012Available online 18 January 2012

Keywords:HydrodesulfurizationGas oilMoS2

Catalyst

0021-9797/$ - see front matter � 2012 Elsevier Inc. Adoi:10.1016/j.jcis.2012.01.019

⇑ Corresponding author at: Department of Materiauate School of Engineering, Kyushu University, MotoJapan. Fax: +81 92 802 2792.

E-mail addresses: hamdy@chem-eng.kyushu-u.ac(H. Farag).

a b s t r a c t

Unsupported nanosized MoS2 and CoMo-sulfide catalysts were synthesized, and their catalytic perfor-mances for the deep hydrodesulfurization (HDS) of treated gas oil were investigated as compared withthat of a CoMo/Al2O3 catalyst. The HDS reactions were carried out in a batch autoclave reactor at340 �C and 3 MPa H2. The CoMo-sulfide catalyst shows the highest activity and can reduce the sulfur con-tent to less than 10 ppm. The decrease in total sulfur content as a function of reaction time was found tofollow pseudo-second order kinetics (empirical form). The change in the concentration of some individualrepresentative sulfur-containing species in gas oil as a function of time was found to follow pseudo-first-order kinetics. However, the change in combined concentration of these species in the gas oil during HDSwith the reaction time was found to corroborate pseudo-second-order kinetics. A kinetic model approachwas proposed from which an estimation of the intrinsic kinetic data can be achieved. The model fitted theobtained data reasonably well, suggesting its potential for better assessment of the catalytic activity inthe HDS of real feedstock. The study reveals that ranking of catalyst activities using model refractory sul-fur-containing compounds does not necessarily imply a typical rank in case of investigating the realfeedstocks.

� 2012 Elsevier Inc. All rights reserved.

1. Introduction

Catalytic hydrodesulfurization (HDS) is recognized as being themost effective and reliable process for reducing the sulfur contentin gas oil to meet stringent regulations recently adopted all overthe world; a sulfur limit of 15 ppm or less has been alreadyadopted by the developed countries [1–5]. For a typical gas oil, thissulfur content corresponds to a significant high conversion level ofall the sulfur-containing compounds. ‘‘Sulfur-free’’ diesel has beenproposed as a future specification for use in a number of applica-tions [6]. For instance, use in fuel cell applications is increasinglypromising as the existence of sulfur-containing compounds, evenin low amounts, can have a severe poisoning effect on noblecatalysts. The poisoning effect on catalysts is such that a catalysteffective for fuel cell reactions may easily become ineffective orshort-lived in the presence of even traces of such species [6]. Thekey factor in the HDS process is the identity of the catalyst wherea high activity is mandatory. Refineries most often use CoMo- and

ll rights reserved.

l Process Engineering, Grad-oka 744, Fukuoka 819-0395,

.jp, hamdy.farag@gmail.com

NiMo-based catalysts [1,2]. However, recently, a series of alterna-tive catalyst candidates relying on unsupported metal sulfidephases have been proposed for use in the hydrotreatment of refin-ery streams [7–11]. One such group of catalysts, termed ‘bulk cat-alysts’, is formed from the transition metal sulfide alone or ismixed with other metal compounds, usually by co-precipitationtechniques, where there is no catalyst carrier or support [8]. For in-stance, nanosized MoS2 catalysts have been found to show promiseactivity in the HDS of model refractory sulfur-containing compo-nents [10,12,13]. It would be of interest to investigate further thebehavior of such catalysts with real feedstocks such as gas oil.Gas oil consists of a mixture of hundreds of paraffins, naphthenes,aromatics, polyaromatic sulfur-containing compounds, and polyar-omatic nitrogen-containing compounds. The catalyst activity inHDS is seriously affected by a variety of contaminants containedin feedstocks [14–20]. The existence of H2S and nitrogen-contain-ing compounds, for instance, showed dramatically detrimental ef-fects on catalyst activity. However, the extent of response of onecatalyst to such species depends on the catalyst identity [12,13].The HDS reaction rate is markedly affected by the existence of eventrace amounts from such species. Polyaromatic sulfur-containingcompounds are well known to be the most refractory sulfur-con-taining compounds in middle distillates [1,2]. For effective HDSactivity, and especially for achieving ultra-low sulfur fuel to meet

122 H. Farag, I. Mochida / Journal of Colloid and Interface Science 372 (2012) 121–129

the recent environmental regulations, a catalyst must be effectivein removing all refractory sulfur-containing compounds [21]. Theprocess is, in fact, still in need of further investigation to fullyunderstand the chemistry associated with removing these refrac-tory species. The HDS of such species proceeds through two paral-lel reaction pathways: (1) the direct hydrogenolysis of the C–Sbonds and (2) the hydrogenation of one of the benzene rings priorto rapid hydrogenolysis. A catalyst of heightened selectivity to-ward one of these catalytic routes for the HDS of polyaromatic sul-fur-containing compounds is important for ultra-deep HDS. Littleinformation is available in literature regarding the HDS of realfeedstocks (e.g., gas oil) over unsupported nanosized bulk metalsulfide catalysts in comparison with the conventional CoMo- andNiMo-based catalysts [22–29]. The results may be of interest asthe data obtained from gas oil and model sulfur compounds arenot always straightforward. The HDS of model refractory sulfur-containing compounds (dibenzothiophene and 4,6-dim-ethyldibenzothiophene) over nanosized MoS2 proceeds mainlyvia a significant contribution from the hydrogenation route [10–13]. On the other hand, the conventional catalysts, that is, CoMo-and NiMo-based catalysts, perform the reaction via a considerablecontribution from both these routes. As such species and theirderivatives are the most refractory sulfur-containing compoundsin HDS, it would be worthwhile investigating their reaction perfor-mance in real feedstocks. Studies have shown that the activity forthe hydrogenation pathway is much more important than the di-rect desulfurization pathway (DDS) in the HDS of these compo-nents [7,8,13]. In this study, a feedstock of gas oil wasinvestigated for HDS over synthesized unsupported nanosizedMoS2, CoMo-sulfide, and commercial CoMo/Al2O3 catalysts. Toalleviate or avoid inhibition by nitrogen species, pre-hydrotreatedgas oil was used. The catalytic properties of the synthesized nano-sized MoS2, CoMo-sulfide, and commercial CoMo/Al2O3 catalysts indeep HDS of this gas oil under refinery operating conditions werecompared. The kinetics of the HDS was also studied.

2. Experimental section

2.1. Materials

Pre-hydrotreated gas oil was provided from the Japan refinery.The initial sulfur content of this gas oil was 360 ppm S with addi-tional specifications listed in Table 1. All chemicals in this studywere purchased from the Wako Chemical Co. and were used with-out any further purification.

Table 1Specification of gas oil.

Density (g/cm3, @15 �C) 0.835S content (w/w-ppm) 360Nitrogen (w/w-ppm) 3

Chemical composition (vol%)Paraffin 78.3Olefin 0Mono-aromatics 19.3Di-aromatics 2.4Tri-aromatics 0

Boiling points (�C)10 vol% 249.550 vol% 29790 vol% 357.5Cetane index 61.8

Measurements were conducted by refinery pro-vided this oil.

2.2. Catalysts

Three catalysts, synthesized nanosized MoS2 (MS), synthesizedCoMo-sulfide (CMS), and commercial CoMo/Al2O3 catalysts, wereimplemented in this study. The nanosized MoS2 catalyst was syn-thesized according to the method described in detail elsewhere[10]. Briefly, ammonium heptamolybdate tetrahydrate was thestarting molybdenum precursor. The molybdenum precursor washeated under the continuous flow of an Ar atmosphere to 300 �C.Then, the inlet gas was converted to H2S/H2 (10% (v/v)), with a flowrate of 60 SCCM. The temperature was raised to 830 �C at 2 �C/min.The sample was maintained at 830 �C and a flow of H2S/H2 of 10%(v/v) for 3 h. Thereafter, the reactor was flushed using Ar forapproximately half an hour. The obtained sample was kept in adesiccator for further investigation. The sample was milled usingan agitating-media mill with inner volume of 80 ml for 24 h usingzirconia beads of a diameter of 3 mm. The CMS catalyst was syn-thesized where MoS2 was used as a preliminary support for the co-balt phase. The unsupported CoMo sulfide catalysts with Co/(Co + Mo) atomic ratio of ca. 0.05 were synthesized as follows.Firstly, thermal treatment of ammonium tetrathiomolybdate(NH4)2MoS4 was carried out under a continuous flow of 10% v/vH2S/H2 gas mixture at 420 �C. This step yields the MoS2 phase.Then, 0.5 g of cobalt nitrate hexahydrate (Co(NO3)2�6H2O) was dis-solved in distilled water, to which 20 mL of 1 M ammonia solutionwas added. The ammonical cobalt solution was poured into a slur-ry solution containing 3.2 g of the previously synthesized MoS2.The slurry was then subjected to ultrasonic vibration for 1 h, stir-red in air over a water bath at 45 �C for approximately 6 h to gentlyevaporate the solvent, and then dried in a vacuum oven overnightat 120 �C. The precursor was sulfided again under a 10% v/v H2S/H2

gas mixture at 400 �C; thereafter, the sulfiding gas was purged for30 min using Ar. Specifications of the catalysts are summarized inTable 2. Immediately before carrying out the catalytic tests in theHDS, each catalyst was subjected once more to sulfidation usingthe H2S/H2 gas mixture. The BET-surface areas were measuredusing an automatic sorptomatic apparatus (Quantachrome) at li-quid nitrogen temperature. Activation treatments were carriedout at 300 �C. X-ray diffraction studies were carried out in a RigakuDiffractometer using Cu Ka radiation (k = 1.54056 Å) and operatingat 43 kV and 30 mA. Transmission electron microscopy (TEM) wasused for lattice imaging and electron diffraction of nanoparticles.Imaging were obtained in a Hitachi 9000UHR working at an accel-erating voltage of 200 kV.

2.3. Catalytic activity

Catalyst performances were investigated for the HDS of gas oil.The reactions were carried out in a high pressure batch systemusing a magnetically stirred microautoclave reactor (100 ml

Table 2Specifications of catalysts.

CoMo/Al2O3

Metal (%, in the oxide form as received)Co 3Mo 12.4Surface area (m2/g, BET-N2) 210Pore volume (ml/g) 0.4

MSSurface area (m2/g, BET-N2) 100Pore volume (ml/g) 0.5

CMSSurface area (m2/g, BET-N2) 47Pore volume (ml/g) 0.3

0 20 40 60 80 100

0

100

200

300

MS

2θIn

tens

ity, C

PS

0

100

200

300

CMS

Fig. 1. Powder XRD pattern of synthesized MS and CMS catalysts after sulfidation

H. Farag, I. Mochida / Journal of Colloid and Interface Science 372 (2012) 121–129 123

capacity). This reactor was designed to allow for the withdrawal ofsample at time intervals during the reaction. In all runs, the cata-lysts were investigated at typical reaction conditions. Prior to thereaction, the catalyst was sulfided at 400 �C for 3 h, by a streamof 10 v/v% H2S/H2 with a flow rate of 60 SCCM. In a typical exper-iment, 15 ml of gas oil was loaded into the reactor with 0.02–0.1 gfreshly sulfided catalyst (nanosized MS, CMS, and/or CoMo/Al2O3

catalysts). The stirrer speed was adjusted to 1000 rpm, and thereactor purged several times with H2 before the reaction pressurewas set at 3 MPa H2. All the HDS reactions were finally carried outat 340 �C. During the reaction, small aliquots of less than 0.1 mLwere withdrawn at various time intervals for analysis. Total sulfurcontent in the hydrotreated gas oil during the reaction was deter-mined using a gas chromatograph (GC) equipped with sulfurchemiluminescence detector (CLD) and methylsiloxane capillarycolumn (0.32 mm � 50 m, HP6890) with the aid of a standard gasoil sample. Identification of some sulfur species in the hydrotreat-ed gas oil was achieved using standard sulfur-containing com-pounds. The quantity of some individual sulfur-containingspecies was estimated from standard calibrations of analogousknown samples in terms of the GC-CLD chromatogram area. Theconversion of total sulfur was calculated as follows:

HDS ð%Þ ¼ SI � SF

SI � 100

where SI and SF are the concentration of total sulfur in ppm (weight)in the feedstock, before and after the HDS reaction, respectively.

with 10 v/v% H2S/H2 gas at 400 �C.

Fig. 2. Transmission electron microscopy (TEM) micrographs of the MS (a) and CMS(b) catalysts after sulfidation with 10 v/v% H2S/H2 gas at 400 �C.

3. Results

3.1. Catalyst characterizations

Fig. 1 reports the XRD patterns obtained for the synthesized MSand CMS catalysts. Broad peaks at 2h = 14, 34, and 59 characterizedfor MoS2 hexagonal structure were observed. The patterns illus-trate typical scans that are characteristic of poorly crystallineMoS2 structure. They are similar to those reported in JCPDS datafor crystalline MoS2-2H (PDF: 6-97I). The XRD peaks indicate a verypoorly crystalline structure of MoS2 characterized by a low stack-ing of layers along the c direction. Inclusion of Co-phase intoMoS2 did not make any important change in the XRD pattern ofMoS2. The lack of obtaining any peak assigned for Co-phase canbe attributed to the low loading amount of Co-phase. The averagecrystallite sizes of MS and CMS catalysts estimated from the XRDpatterns in the (002) plane were 4 and 3.8 nm, respectively. TheBET surface areas for MS and CMS catalysts were estimated to be100 and 47 m2/g, respectively. Surface characteristics of catalystsare reported in Table 2. TEM micrographs are reported in Fig. 2aand b, for MS and CMS catalysts, respectively. These images showthe well-known characteristic fringes of the layered MoS2-phases.The number of stacked layers generally varies between 3 and 7.

3.2. Hydrodesulfurization of gas oil

The pre-hydrotreatment of the gas oil affects the nitrogen levelsignificantly. The HDS of the sulfur-containing compounds in thisgas oil proceeds with minimal inhibition owing to the nitrogen-containing species [16,18]. This gas oil contains a complex mixtureof sulfur-containing components of more than 60 different species.Fig. 3 shows the HDS reaction data represented by the change insulfur content of pre-hydrotreated gas oil as a function of reactiontime for nanosized MS, CMS, and CoMo/Al2O3 catalysts. It is inter-esting to note that both the nanosized MS and CMS catalysts exhi-bit higher activity than the commercial CoMo/Al2O3 catalyst. Thedata obtained over all catalysts are well matched with second or-

der kinetics in respect to the total sulfur content. The pseudo-sec-ond-order plots of the conversion of sulfur compounds in the pre-treated gas oil over the present catalysts are shown in Fig. 4. It

0

70

140

210

280

350

0 200 400 600 800Time, min

Tota

l S in

ppm

CMS

MS

CoMo/Al2O3

Fig. 3. HDS kinetics of gas oil (decrease in total sulfur content with time) over MS,CMS, and CoMo/Al2O3 catalysts.

124 H. Farag, I. Mochida / Journal of Colloid and Interface Science 372 (2012) 121–129

must be noted that this second order fit is simply an empirical pre-sentation of data and provides no description of the reaction mech-anism. For comparison, the activity was normalized per unitsurface area, per unit metal content, as well as per unit volumeof catalyst. The normalization based on unit volume was calculatedusing an experimental packing density of 1.35, 1.40, and 0.70 g/cm3 for MS, CMS, and the commercial CoMo/Al2O3 catalysts,respectively. The estimated data are presented in Table 3. The rateconstant, estimated with respect to surface area, for the nanosizedMS catalyst in the HDS of pre-treated gas oil is ca. 10 times largerthan that for CoMo/Al2O3. Furthermore, the CMS shows an order ofmagnitude higher activity than that of the CoMo/Al2O3 catalyst.Furthermore, the activity of CMS catalyst normalized to total metalconcentration is ca. 4 times larger than that of CoMo/Al2O3. Thistrend in activity may also be observed for the reaction of some se-lected singly investigated sulfur-containing species. This meansthe nanosized MS and/or CMS may have more potentially activesites per unit surface area than the CoMo/Al2O3 catalyst.

The conversion of selected sulfur-containing compounds in gasoil, that is, dibenzothiophene (DBT), 4-methyl-dibenzothiophene(4-MDBT), and 4,6-dimethyldibenzothiophene (4,6-DMDBT), canbe well approximated by applying pseudo-first-order kinetics aspresented in Figs. 5a, 6a, and 7a for the HDS reaction over MS,CMS, and CoMo/Al2O3 catalysts, respectively. These results are in

0

0.02

0.04

0.06

0.08

0.1

0.12

0.14

1 10 100 1000Time, min (log scale)

1/S

cont

ent,

ppm

CMS

MS

CoMo/Al2O3

Fig. 4. Pseudo-second-order plots of the HDS of gas oil over MS, CMS, and CoMo/Al2O3 catalysts at 340 �C under 3 MPa H2.

agreement with those reported for such reactants when used sin-gly. Alkyl-substituted dibenzothiophenes were selected for investi-gation because of their low reactivity in the HDS reaction. Thereactivity of these species toward HDS over the present catalystscan be ranked as follows: DBT > 4-MDBT > 4,6-DMDBT. Results inTables 1 and 2 demonstrate that the CMS catalyst is the most activecatalyst for the HDS of gas oil in comparison with the othercatalysts.

Table 4 shows the total sulfur content as well as the sulfur contentin a form of some identified sulfur species in the gas oil after the HDSreaction over the present catalysts. The data show that the catalystactivities are ranked in decreasing order as follows: CMS > MS > Co-Mo/Al2O3 catalysts. One may notice that the sulfur content in theform of alkyl-substituted dibenzothiophene makes up a significantratio of the total sulfur contained in this gas oil. This highlights theimportance of finding a catalyst of potential activity to transformsuch species for achieving an ultra-low sulfur fuel.

4. Discussion

4.1. Kinetics of hydrodesulfurization of some model sulfur-containingcompounds in gas oil

Although the kinetics of the HDS of some model sulfur com-pounds have been reported widely in literature [1,15,28–36], stud-ies on the HDS kinetics of real feedstocks are relatively limited.This is probably owing to the presence of a large variety of sulfurcompounds with different reactivities [27,36]. The refractory sul-fur-containing compounds accumulate in the high boiling feed-stocks making up a significant ratio of the total sulfur content.Thus, to achieve ultra-deep-HDS, it is necessary to remove suchspecies effectively. It is important to evaluate the kinetic behaviorof such species inside the real feedstock. A better insight into thepotential difference between the present catalysts in the HDS ofthe pre-treated gas oil can be gained by investigating the kineticbehavior of some representative individual sulfur compounds inthis feed. Figs. 5a, 6a, and 7a show pseudo-first-order plots of theHDS of DBT, 4-MDBT and 4,6-DMDBT in gas oil over MS, CMS,and CoMo/Al2O3 catalysts, respectively. Pseudo-first order kineticsrepresents satisfactorily the experimental data with regression fac-tors in the range 96–99%. These results are generally in agreementwith those reported in literature for similar model species [4,22].The estimated kinetic data are described in Table 5. One may seethat 4,6-DMDBT is the most refractory sulfur-containing com-pound toward HDS. The rate of disappearance of DBTs over allexamined catalysts decreased in the following order DBT� 4-MDBT > 4,6-DMDBT, indicating that alkyl groups next to the sulfuratom hinder the HDS reactions, and the degree of inhibition in-creases as the number of alkyl groups around this sulfur increasesin agreement with other studies in the literature [5,15]. Accordingto the estimated pseudo-first order rate constant, 4,6-DMDBT is ca.2.4, 1.7, and 2.3 times less reactive than DBT in the HDS over CMS,MS, and CoMo/Al2O3 catalysts, respectively. This is well known tobe attributed to the steric hindrance exerted by the substitutedmethyl groups in the 4- and 6-positions, which limits the approachof the compounds to the catalytic active sites. The reactivity of theDBT components over the CoMo/Al2O3 and metal-sulfide catalystssuggests that the HDS of these sulfur species proceeds by means ofa typical mechanism over these catalysts. A similar behavior wasobserved for the HDS of model DBT components over analogouscatalysts [30–33,37] The reactivity trend for the HDS of the repre-sentative refractory sulfur-containing components over all thepresent catalysts was the same. It is clear that the nanosized me-tal-sulfide catalysts are significantly more active for the HDS ofthe refractory sulfur-containing compounds than the CoMo/Al2O3

Table 3Pseudo-second order rate constants and half-life of the sulfur content in gas oil hydrodesulfurization over nanosized metal sulfides and CoMo/Al2O3 catalysts.

Catalyst Apparent rate constants, half-life and relative activity at different consideration

ka t1/2a RAd kb t1/2

b RA kc t1/2c RA ke RA

CoMo/Al2O3 1 � 10�4 28 1 1 � 10�4 20 1 5 � 10�7 5555 1 9 � 10�4 1MS 3 � 10�4 9 3 2 � 10�4 13 2 50 � 10�7 555 10 5 � 10�4 0.5CMS 20 � 10�4 1 20 14 � 10�4 2 14 425 � 10�7 65 85 33 � 10�4 4

Factor of error: ±2–7%.a Apparent rate constant, ppm(wt)�1 min�1 gcat�1, normalized per unit weight of catalyst and the corresponding half lifetime, minute.b Apparent rate constant, ppm(wt)�1 min�1 mL cat�1, normalized per unit volume of catalyst and the corresponding half-life, minute.c Apparent rate constant, ppm(wt)�1 min�1, normalized per unit surface area of catalyst and the corresponding half-life, minute.d Relative activity calculated as the ratio of k of the nanosized MoS2 catalysts to that of the CoMo/Al2O3 catalyst.e Apparent rate constant, ppm(wt)�1 min�1, normalized per unit of total metal (Co + Mo) concentration and the corresponding half-life, minute.

(a)

0 100 200 3000

0.02

0.04

0.06

Experimental

Fitted

Time, min1/

S co

nten

t

0 100 200 3000

1

2

3

4

DBTFitted4-MDBT

Fitted4,6-DMDBTFitted

Time, min

Ln

(S in

ppm

)

(b)

Fig. 5. Pseudo-first-order kinetics of DBT, 4-MDBT, and 4,6-DMDBT (a) and second order plot of combined content of DBT, 4-MDBT, and 4,6-DMDBT (b) in gas oil HDS over MScatalyst at 340 �C and 3 MPa H2.

(a) (b)

0 100 200 300 4000

0.1

0.2

Experimental

Fitted

Time, min

1/S

cont

ent

0 100 200 300 4001

0

1

2

3

DBT

Fitted

4-MDBT

Fitted

4,6-DMDBT

Fitted

Time, min

Ln

(S in

ppm

)

Fig. 6. Pseudo-first-order kinetics of DBT, 4-MDBT, and 4,6-DMDBT (a) and second order plot of combined content of DBT, 4-MDBT, and 4,6-DMDBT (b) in gas oil HDS overCMS catalyst at 340 �C and 3 MPa H2.

H. Farag, I. Mochida / Journal of Colloid and Interface Science 372 (2012) 121–129 125

catalyst. The superior activity of metal-sulfide catalysts over theCoMo/Al2O3 catalyst can be attributed to their higher hydrogena-tion activity, which is considered essential for the HDS of such al-kyl DBTs [10,12,15] and their high tolerance for the HDS reactioninhibitors [38,39].

4.2. Kinetics of combined hydrodesulfurization reaction of DBT,4-MDBT and 4,6-DMDBT

Various studies have been devoted to exploring the aggregatedkinetic behaviors of sulfur-containing components in real

(a) (b)

0 100 200 3000.01

0.02

0.03

0.04

ExperimentalFitted

Time, min

1/S

cont

ent

0 100 200 3002

0

2

4

DBTFitted4-MDBT

Fitted

4,6-DMDBTFitted

Time, min

Ln

(S in

ppm

)

Fig. 7. Pseudo-first-order kinetics of DBT, 4-MDBT, and 4,6-DMDBT (a) and second order plot of combined content of DBT, 4-MDBT, and 4,6-DMDBT (b) in gas oil HDS overCoMo/Al2O3 catalyst at 340 �C and 3 MPa H2.

Table 4Distribution of some refractory sulfur-containing compounds in gas oil before and after the HDS reaction over nanosized metal sulfides and commercial CoMo/Al2O3 catalysts.

Sulfur species S (ppm) S (%) Catalyst

CoM/Al2O3 (242 min) MS (205 min) CMS (260)

S (ppm) %Conv. S (ppm) %Conv. S (ppm) %Conv.

Total S in gas oil 360a 100.0 207 42.5 115 68.1 23 94DBT 5.1 1.4 0 100.0 1.1 78.4 0 1004-MDBT 19.1 5.3 5.8 69.6 5.4 71.7 1.2 954,6-DMDBT 67.8 18.8 21 69.0 14.1 79.3 5 93

Reaction conditions: 340 �C and 3 MPa H2; 0.1 g catalyst was used in all tests.a Total sulfur content in the gas oil before the HDS reaction.

Table 5Pseudo-first-order rate constants of some refractory sulfur-containing compounds in the HDS of gas oil over nanosized MS, CMS and commercial CoMo/Al2O3 catalysts.

Sulfur component Catalyst

CMS MS CoMo/Al2O3

Rate constants k (s�1 g cat�1 10�4)

DBT 18.8 11.7 11.04-MDBT 13.5 10.5 7.84,6-DMDBT 7.8 7.1 4.7

First Order half-life

DBT 6.1 9.9 10.54-MDBT 8.6 11.0 15.24,6-DMDBT 14.8 16.3 24.6

Combined rate constants (s�1 g cat�1) (1st); (ppm�1 s�1 g cat�1) (2nd)

k2nda 8.3 � 10�5 2.6 � 10�5 1.3 � 10�5

k1stb 40.1 � 10�4 29.3 � 10�4 23.5 � 10�4

(t1/2)1stc 8.6 11.7 15.0

(t1/2)2ndc 6.7 10.8 16.0

Factor of error: ±2–7%.NB: Initial experimental points were not considered on estimating the kinetic parameters.

a Pseudo-second-order rate constant for combined DBT, 4-MDBT, and 4,6-DMDBT.b Pseudo-first-order rate constant for combined DBT, 4-MDBT, and 4,6-DMDBT.c Half-life of second order and average half-life of first order in minute for conversion of the sulfur species.

126 H. Farag, I. Mochida / Journal of Colloid and Interface Science 372 (2012) 121–129

feedstock [34,35,40]. The HDS of gas oil is often treated by a com-bined kinetic model because of the existence of a rather high num-ber of various sulfur-containing components. Therefore, a widerange of reactivities is expected for these components. While theHDS reactions of individual sulfur-containing compounds exhibit

mostly first order kinetics, the overall reaction of the mixture inreal feedstock often shows higher order kinetics [34,35]. In thepresent study, the change in concentration of some refractory sul-fur-containing components (i.e., DBT, 4-MDBT, and 4,6-DMDBT)contained in gas oil as a function of time was found to fit pseu-

0 50 100 150 200 2504

6

8

10

Experimental

Fitted

Time, min

Ln

(S c

onte

nt a

s D

BT

x4M

DB

Tx4

6DM

DB

T

ig. 8. Pseudo-first-order kinetics of hydrodesulfurization of multiplied content ofBT, 4-MDBT, and 4,6-DMDBT in gas oil over MS catalyst at 340 �C and 3 MPa H2.he theoretical fitting was fulfilled according to Eq. (4).

H. Farag, I. Mochida / Journal of Colloid and Interface Science 372 (2012) 121–129 127

do-first order kinetics reasonably well. Thus, there is a need tosolve these peculiarities in light of detailed kinetic analysis.

As most model sulfur-containing compounds have proven tofollow pseudo-first order kinetics in the HDS reaction, it may bemore interesting to evaluate the kinetic behavior of these specieswhen they exist together in real feedstock under the same reactionconditions. It has already been mentioned that the overall reactionof total sulfur removal follows pseudo-second order kinetics(Fig. 2). The half-life of a compound is highly dependent on the or-der it follows. For a first order reaction, the half-life is independentof the initial concentration. Physically, this means that whateverthe initial concentration is, a half of its content will be transformedvia the reaction in a typical time; that is, the concentration dropsby a factor of two for every t1/2 increment in time. Unlike the firstorder reaction, the second order half-life is inversely proportionalto the initial concentration of the component. The kinetics of thecombined concentrations of DBT, 4-MDBT and 4,6-DMDBT followa second order equation as described in Figs. 5a, 6a, and 7a. How-ever, this second order has no implication, in terms of kinetics, onthe mechanism of the HDS of the components. Some interestingpoints can be estimated and extracted from investigating thebehavior of the combination of these sulfur-containing compo-nents based on first-order kinetics. Table 5 shows the dataestimated from a second-order representation of combined DBT,4-MDBT, and 4-MDBT in comparison with their individual first-order manipulation. A model approach to treat these datakinetically is described below.

4.3. Mathematical approach

For first-order reactions, the following equations were applied:

dCAi

dt¼ ki � CAi ð1Þ

CAi ¼ CAi� � exp�ki �t ð2Þ

LnCAi

CAi�

� �¼ �ki � t ð3Þ

where CAi is the concentration of component i at reaction time t andCAio is the initial concentration of component i. The overall rate isthe algebraic sum of the rates for the reaction of the individual spe-cies. Combining the pseudo-first-order equations of the three se-lected sulfur-containing components, for instance, yields thefollowing:

LnCA1 � CA2 � CA3

CA1� � CA2� � CA3�

� �¼ �ðk1 þ k2 þ k3Þ � t ¼ �k1st � t ð4Þ

The half-life of the reaction for the combined components isestimated to be:

n � Lnð2Þ ¼ ðk1 � tð1=2Þ1 þ k2 � tð1=2Þ2 þ k3 � tð1=2Þ3Þ ð5Þ

Assuming

tð1=2Þ1 þ tð1=2Þ2 þ tð1=2Þ3 þ . . . tð1=2Þn

n

� �¼ tð1=2Þ1st ð6Þ

Then

tð1=2Þ1st ¼n � Lnð2Þ

k1stð7Þ

where n is the number of all sulfur-containing components in thefeedstock and k1st is the sum of pseudo-first-order rate constantsof each sulfur-containing compound in the real feedstock. Thechange in combined concentrations of DBT, 4-MDBT, and 4,6-

DMDBT was also proven to fit the second-order kinetic equationand can be described as follows:

1PCAi

� �¼ 1P

CAi�þ k2nd � t ð8Þ

This may lead to the following approximation:

LnCA1� � CA2� � � � � � CAi�

CA1 � CA2 � � � � � �CAi

� �� k1st

k2nd� 1P

CAi� 1P

CAi�

� �ð9Þ

Eq. (9) is a linear equation that can be easily solved to estimate theratio of k1st/k2nd. All these functions were solved to fit the obtainedexperimental data using the least square regression method withthe aid of Mathcad software.

4.4. Catalyst performances

Figs. 5a, 6a and 7a show the pseudo-first order plots of changeof DBT, 4-MDBT, and 4,6-DMDBT concentration in gas oil as afunction of time. The change in combined concentration of DBT,4-MDBT, and 4,6-DMDBT in the form of 1/

PCAi during the HDS

reaction as a function of time was found to corroborate second-order behavior (Figs. 5b, 6b and 7b). The slope of this linear lineis the empirical second order rate constant where the interceptequals [1/sum of initial concentration of all sulfur-containingcompounds in this mixture]. Fig. 8 shows an example of the linearrepresentation between the logarithm of the multiplied concentra-tion of DBT, 4-MDBT, and 4,6-DMDBT as a function of time for theHDS reaction over the MS catalyst. The slope of this linear linerepresents the sum of the pseudo-first order rate constants for allsulfur-containing compounds as described in Eq. (4). One maysafely assume that under the present reaction conditions, acombined first-order reaction for some representative sulfur-containing components fits the second-order equation reasonably(Figs. 5b and 8). Although this assumption is limited to exist undercertain circumstances only, one can use it to deduce some usefulconclusions on this matter. During the manipulation of fittingthese models to match the experimental results using the Mathcadsoftware program, it was noticed that the half-life estimated fromthe second-order empirical equation was not very different fromthe average half-life estimated from the individual first-order plotof these components. The result is described in Table 5. The follow-ing approximation can thus be suggested:

tð1=2Þ2nd � tð1=2Þ1st ð10Þ

FDT

128 H. Farag, I. Mochida / Journal of Colloid and Interface Science 372 (2012) 121–129

If the total number of sulfur-containing components in the realfeedstock is known, then the sum of the apparent pseudo-first-or-der rate constants of all sulfur-containing components in the realfeedstock can be estimated according to Eq. (7) (the empirical sec-ond order manipulation). Therefore, an intrinsic kinetic interpreta-tion of the observed data can be evaluated.

The HDS rate constant of CMS for total sulfur removal was anorder of magnitude higher than that of MS or the conventional cat-alyst, irrespective of the basis on which these activities were esti-mated. This indicates an existence of potential synergy betweenCo- and Mo-sulfide phases in case of CMS catalyst [41]. However,when considering the activity according to the unit surface area,the CMS catalyst was approximately 85-fold more active than thecommercial CoMo/Al2O3 catalyst. On the other hand, there is adifference in the relative catalytic performance in the HDS ofpre-treated gas oil and the combination of DBT, 4-MDBT, and4,6-DMDBT in the pre-treated gas oil (Fig. 9). The CMS catalystexhibits a higher relative catalytic activity for the HDS of pre-treated gas oil than that for the representative aggregate of DBT,4-MDBT, and 4,6-DMDBT. Thus, one should be careful in terms ofranking catalysts based on the criteria of their activities for theHDS reaction when utilizing model sulfur-containing componentsalone. The present result reveals that there is still room for furtherdevelopment of the unsupported metal-sulfide catalysts to displaymore superior activity. While the CMS catalyst shows the highestactivity among other catalysts for the conversion of individualsulfur-containing components, its performance for the HDS ofpre-treated gas oil is much more pronounced. This suggests thatin the evaluation of the relative performance of catalysts for theHDS reaction, it is recommended to extend the investigation tothe real feedstock and not to rely solely on investigating model sul-fur-containing components. The unsupported catalysts performthe HDS reaction mainly via the hydrogenation route [9,13]. Thus,a catalyst of potential hydrogenation activity with tolerance fornitrogen inhibition may be promising for fulfilling ultra-deepHDS of real feedstock.

The results in Table 4 show that a considerable amount of sulfurin gas oil exists as DBT, 4-MDBT, and 4,6-DMDBT. The sulfur con-tent in these components represents approximately 25% of the allsulfur content in the pre-treated gas oil before reaction. Interest-ingly, even after the HDS reaction, the sulfur content representedby these components still constitutes from 13% to 24% of the totalsulfur. The rest of this percentage is supposed to be comprisedfrom the derivatives of such sulfur species. This result indicatesthat at high HDS conversion, the least refractory sulfur-containingcompounds are dominating. These results stress the importance of

0

3

6

9

12

15

18

CMS MS CoMo/Al2O3

Rat

io o

f k2n

d/k 2n

d(C

oMo/

Al2

O3)

Real feedstockDBT+4-MDBT+4,6-DMDBT

Fig. 9. Relative catalyst performances in the HDS of real feedstock (pre-treated gasoil) and aggregate of (DBT + 4-MDBT + 4,6-DMDBT) in the pre-treated gas oil withrespect to second order manipulation.

fully understanding the reaction mechanism of these components.The marked differences in reactivities of these sulfur-containingcompounds in gas oil have important implications on catalystselection and optimum reaction conditions for performing ultra-deep HDS. As we move toward ultra-low sulfur fuel, the chemistryshould be focused and directed toward investigating in more detailthe reaction behavior of these compounds.

5. Summary and conclusions

The HDS of gas oil was studied over two synthesized unsup-ported catalysts, that is, MS and CMS and a commercial CoMo/Al2O3 catalyst under typical operating conditions. The data showthat the nanosized CoMo-sulfide catalyst was the most activeamong the catalysts studied. The decrease in total sulfur contentin gas oil during the HDS reaction with reaction time was foundto fit pseudo-second-order kinetics reasonably well. The HDS ofcertain individual sulfur-containing components in gas oil exhibitspseudo-first-order kinetics. However, the reaction of the overallmixture shows pseudo-second-order kinetics. The data obtainedwere analyzed based on a suggested kinetic model approach. Theresults also imply that a catalyst with potential hydrogenationactivity is preferred to the traditional CoMo-based catalysts toachieve gas oil of ultra-low sulfur content. The obtained data sug-gest that a more developed unsupported nanosized metal-sulfidecatalyst can provide promise for superior activity in the HDSreaction.

References

[1] K.H. Choi, Y. Sano, Y. Korai, I. Mochida, Appl. Catal., B 53 (2004) 275.[2] Y. Yoshimura, M. Toba, H. Farag, K. Sakanishi, Catal. Surv. Asia 8 (1) (2004) 47.[3] C. Song, X. Ma, Appl. Catal., B: Environ. 41 (2003) 207.[4] P. Steiner, E.A. Blekkan, Fuel Process. Technol. 79 (1) (2002) 1.[5] K. Segawa, K. Takahashi, S. Satoh, Catal. Today 63 (2000) 123.[6] K. Tawara, T. Nishimura, H. Iwanami, T. Nishimoto, T. Hasuike, Ind. Eng. Chem.

Res. 40 (10) (2001) 2367.[7] S. Eijsbouts, S.W. Mayo, K. Fujita, Appl. Catal., A: General 322 (2007) 58.[8] L. Kane, S. Romanow, Hydrocarbon Proces. 80 (8) (2001) 28.[9] B. Yoosuk, J.H. Kim, C. Song, C. Ngamcharussrivichai, P. Prasassarakich, Catal.

Today 130 (1) (2008) 14.[10] H. Farag, K. Sakanishi, M. Kouzu, A. Matsumura, Y. Sugimoto, I. Saito, J. Mol.

Catal. A: Chem. 206 (2003) 399.[11] Y. Iwata, K. Sato, T. Yoneda, Y. Miki, Y. Sugimoto, A. Nishijima, H. Shimada,

Catal. Today 45 (1998) 353.[12] H. Farag, Appl. Catal., A: General 331 (1) (2007) 51.[13] H. Farag, K. Sakanishi, J. Catal. 225 (2) (2004) 531.[14] T. Kabe, Y. Aoyama, D. Wang, A. Ishihara, W. Qian, M. Hosoya, Appl. Catal., A

209 (2001) 237.[15] S.T. Sie, Fuel Process Technol. 61 (1) (1999) 149.[16] K. Choi, Y. Korai, I. Mochida, J.W. Ryu, W. Min, Appl. Catal., B 50 (2004) 9.[17] G.C. Laredo, J.A. De los Reyes, J.L. Cano, J.J. Castillo, Appl. Catal., A 207 (2001)

103.[18] Y. Sano, K.H. Choi, Y. Korai, I. Mochida, Appl. Catal., B 53 (2004) 169.[19] Y. Sano, K. Sugahara, K.H. Choi, Y. Korai, I. Mochida, Fuel 84 (2005) 903.[20] P. Zeuthen, K.G. Knudsen, D.D. Whitehurst, Catal. Today 65 (2001) 307.[21] M. Macaud, A. Milenkovic, E. Schulz, M. Lemaire, M. Vrinat, J. Catal. 193 (2000)

255.[22] Z. Varga, J. Hancsók, G. Nagy, G. Pölczmann, D. Kalló, Top. Catal. 45 (1–4)

(2007) 203.[23] S.W. Lee, J.W. Ryu, W. Min, Cat. Surv. Asia 7 (2003) 271.[24] T. Fujikawa, M. Kato, H. Kimura, K. Kiriyama, M. Hashimoto, N. Nakajima, J. Jpn.

Petrol. Inst. 48 (2005) 106.[25] N. Kunisada, K.H. Choi, Y. Korai, I. Mochida, K. Nakano, Appl. Catal., A 269 (1–2)

(2004) 43.[26] T. Fujikawa, O. Chiyoda, M. Tsukagoshi, K. Idei, S. Takehara, Catal. Today 45

(1998) 307.[27] A. Al-Barood, H. Qabazard, A. Stanislaus, Pet. Sci. Technol. 23 (7–8) (2005) 749.[28] S.K. Bej, A.K. Dalai, J. Adjaye, Pet. Sci Technol. 20 (2002) 867.[29] R.M. Navarro, P. Castaño, M.C. Álvarez-Galván, B. Pawelec, Catal. Today 143 (1–

2) (2009) 108.[30] M. Houalla, D.H. Broderick, A.V. Sapre, N.K. Nag, V.H.J. de Beer, B.C. Gates, J.

Catal. 61 (1980) 523.[31] M. Egorova, R. Prins, J. Catal. 225 (2004) 417.[32] M.J. Girgis, B.C. Gates, Ind. Eng. Chem. Res. 30 (1991) 2021.[33] M. Houalla, N. Nag, A.V. Sapre, D.H. Broderick, B.C. Gates, AIChE J. 24 (1978)

1015.

H. Farag, I. Mochida / Journal of Colloid and Interface Science 372 (2012) 121–129 129

[34] T.C. Ho, Stud. Surf. Sci. Catal. 127 (1999) 179.[35] J. Ancheyta, M.J. Angeles, M.J. Macías, G. Marroquín, R. Morales, Energy Fuels

16 (1) (2002) 189.[36] X. Ma, K. Sakanishi, T. Isoda, I. Mochida, Ind. Eng. Chem. Res. 33 (1994) 218.[37] H. Farag, I. Mochida, K. Sakanishi, Appl. Catal., A 194–195 (2000) 147.

[38] H. Farag, A. El-Hendawy, K. Sakanishi, M. Kishida, I. Mochida, Appl. Catal., B 91(1-2) (2009) 189.

[39] M. Schmidt, H. Topsoe, Hydrocarbon Eng. 16 (3) (2011) 38.[40] M. Te, C. Fairbridge, Z. Ring, Pet. Sci. Technol. 21 (1–2) (2003) 157.[41] L. Kaluza, D. Gulková, Z. Vít, M. Zdrazil, Appl. Catal., A 324 (1–2) (2007) 30.