Green Chemistry Dynamic Article Links

Cite this: Green Chem., 2011, 13, 1907

www.rsc.org/greenchem PAPER

Desulfurization and denitrogenation of gasoline and diesel fuels by meansof ionic liquids†

Antje R. Hansmeier,‡ G. Wytze Meindersma* and Andre B. de Haan

Received 22nd February 2011, Accepted 26th April 2011DOI: 10.1039/c1gc15196g

Ionic liquids can extract mono- and poly-aromatic sulfur and nitrogen compounds from gasolineand diesel and they perform better than conventional solvents. The extraction capacity of severalionic liquids for these heterocyclic compounds is determined and compared to the extractioncapacity for aromatic hydrocarbons. Furthermore, the experimental results obtained are evaluatedin view of the results reported in literature. It is shown that the ionic liquids investigated in thiswork are able to extract sulfur as well as nitrogen-containing aromatics in preference to aromatichydrocarbons. Moreover, the ionic liquids [3-mebupy]N(CN)2, [4-mebupy]N(CN)2 and[bmim]C(CN)3 are superior to sulfolane, which has been used as a benchmark and also outperformthe ionic liquids reported in literature so far. Finally, it has been shown that nitrogen-containinghetero-aromatics are significantly better extracted than sulfur-containing hetero-aromatics.

Introduction

Petrochemical streams contain, in addition to mono- and/orpoly-aromatic components, heterocyclic components compris-ing sulfur and nitrogen compounds. Since these compoundsare responsible for the formation of smog, sour gases, acidrain and NOx emissions,1 the heterocyclic components haveto be removed. From all refinery streams contributing to thegasoline blending pool, FCC Gasoline is, with up to 2.5 wt.%sulfur content, the main sulfur source for carburant fuels.2 Inorder to meet the current compulsory limits for the sulfurcontent in carburant fuels, i.e. gasoline (petrol) and dieselfuels, the sulfur content has to be reduced to 10 ppm.2–4

Furthermore, numerous refinery processes are affected by thepresence of sulfur- and nitrogen-containing components; inparticular nitrogen compounds act as catalyst poison in thedesulfurization process.5 In Fig. 1, typical sulfur and nitrogenaromatics present in gasoline and diesel fuel are shown. Whilegasoline contains mono-aromatic components, the sulfur andnitrogen components present are, besides mercaptans, thiopheneand pyrrole; whereas diesel comprises poly-aromatics and,hence, contains higher sulfur and nitrogen aromatics such asbenzothiophene, dibenzothiophene, indole and carbazole andthe like.5

Conventional desulfurization processes apply hydrotreatingwhere the organic sulfur components are converted to H2S and

Eindhoven University of Technology, SPS, P.O. Box 513, 5600 MBEindhoven, The Netherlands. E-mail: g.w.meindersma@tue.nl† Electronic supplementary information (ESI) available: Additionalexperimental data. See DOI: 10.1039/c1gc15196g‡ Current affiliation for Antje Hansmeier is: Evonik Degussa GmbH.

Fig. 1 Aromatic sulfur (left) and nitrogen (right) components presentin gasoline and diesel fuel.

the corresponding hydrocarbons by means of catalysts based onCoMo or NiMo.5,6 Subsequently, the H2S formed is convertedto elemental sulfur by means of the Claus process.7

The reactivity of the sulfur components depends stronglyon the molecular structure. This means that paraffinic com-ponents, such as thiols, thiolethers and disulfides are readilyconverted, whereas aromatic components, such as thiophene,benzothiophene and dibenzothiophene are less reactive for

This journal is © The Royal Society of Chemistry 2011 Green Chem., 2011, 13, 1907–1913 | 1907

Dow

nloa

ded

by V

ande

rbilt

Uni

vers

ity o

n 06

/05/

2013

09:

59:0

1.

Publ

ishe

d on

10

June

201

1 on

http

://pu

bs.r

sc.o

rg |

doi:1

0.10

39/C

1GC

1519

6GView Article Online / Journal Homepage / Table of Contents for this issue

hydrodesulfurization (HDS). Particularly, dibenzothiopheneand derivatives, such as methyldibenzothiophenes and the like,are difficult to convert since the adsorption on the catalystsurface is sterically hindered. The hydrodesulfurization reactiontakes place in a reactor that is charged by the feed andexcess hydrogen gas at 300–345 ◦C and 30–100 bar.5,6 Nitrogencomponents are simultaneously converted to ammonia and thecorresponding hydrocarbons.6

However, in order to meet the future required compulsorysulfur limit for diesel fuel, HDS is limited due to the lowconversion rates of the higher aromatics. The low sulfur limitscan only be met by extreme operation conditions in terms ofpressure and residence time. Therewith, the process becomesuneconomic. Furthermore, in order to fulfil the gasoline sulfurlimits, deep desulfurization of all streams contributing to thegasoline-pool has to be carried out. In particular, the sulfurcontent of FCC gasoline, which is nowadays the main S-source inthe gasoline-pool, has to be decreased. Although, in comparisonto sulfur components in diesel fuel, HDS of thiophene is reachedmore easily; the bottleneck of this process is the decrease inoctane number due to simultaneous hydrogenation of the olefinspresent in the stream.3,5 In the case of diesel fuels, due to thesevere HDS conditions that are necessary to produce ultra lowsulfur diesel, the cetane number is affected as well.

Therefore, numerous research activities are carried out inorder to develop new technologies for sulfur removal frompetrochemical streams.2,3 The use of ionic liquids as extractionsolvents has been evaluated as well. Van Veen et al. concludedthat their application for desulfurization is limited due to the co-extraction of aromatic hydrocarbons.3 However, as discussed inthe literature,8 next to the sulfur content, the content of aromatichydrocarbons in carburant fuels also needs to be decreased, dueto stricter environmental legislation. Therefore, in this case, theco-extraction of aromatics is regarded as an advantage sincethis allows for the removal of both types of compounds in oneseparation step.

Hence, the use of ionic liquids in order to remove thesulfur aromatics by means of extraction can be a promisingnew technology. This is also indicated by several authors inthe literature.9–15 Besides, Xie et al.16 have reported that poly-aromatics are better extracted than mono-aromatics, while theeconomic evaluation made by Meindersma et al.17 shows thatan ionic liquid-based extraction process is beneficial in termsof investment and operational costs. Therefore, in particular forhigher sulfur aromatics, extraction can be an (economically)attractive possibility. Moreover, several authors have shownthat the simultaneous extractive removal of sulfur and nitrogencomponents with ionic liquids is possible.4,15,16

Ionic liquids are molten salts that consist mostly of largeorganic cations in combination with a vast variety of mainlyinorganic anions. The advantages of ionic liquids as extractionsolvents for the removal of heterocyclic aromatic hydrocarbonsare their negligible vapour pressure, their broad liquid tempera-ture range and their tailorability. The latter enables the creationof one specific extraction solvent while the two aforementionedproperties are valuable characteristics in terms of process design.For example, the low vapour pressure enables the recovery ofthe dissolved components by simple evaporation instead of, e.g.distillation or back-extraction.18

Table 1 Sulfur and nitrogen model feed

Sulfur model feed[wt%]

Nitrogen model feed[wt%]

Thiophene 0.5 —Benzothiophene 0.5 —Dibenzothiophene 2.0 —Pyrrole — 0.5Indole — 0.5Carbazole — 250 ppmToluene 5 5Tetralin 28 28n-Heptane 64 65.975

Although for nitrogen components present in carburant fuelsno limitations exist yet, it is likely that within the near futuremaximum concentrations of nitrogen permitted in fuels will bereleased. Moreover, the low sulfur content of diesel fuel enablesthe use of effective NOx removal in the car itself. Therefore,in this work, the removal of sulfur and nitrogen compoundsby means of ionic liquids has been studied. Numerous authorshave shown that the extraction of only sulfur components9–15 orthe simultaneous removal of sulfur and nitrogen componentswith ionic liquids is possible.4,15,19 However, the ionic liquidsused for these purposes are often not suitable for an industrialprocess since they are chemically unstable,4,9,13,17 exhibit ahigh molecular weight which requires large solvent amounts,5

or have too low selectivity and/or capacity for aromatichydrocarbons.5,10,12 The total sulfur and nitrogen removal willbe investigated in comparison to mono- and poly-aromatics,representatives for gasoline and diesel carburant fuels. Therefore,two model feeds have been created, each consisting of toluene,tetralin, heptane and the three sulfur or nitrogen componentsshown in Fig. 1. The model feed compositions are given inTable 1.

The ionic liquids investigated in this work are chosen basedon their COSMO-RS s-profile.20 As described above, the ionicliquids appeared to be suitable candidates for the removal ofmono- and poly-aromatic hydrocarbons20–23 and are, therefore,evaluated in this work for the removal of sulfur and nitrogencomponents.

For the application of ionic liquids in refinery processing, theinfluence of complex feed mixtures on the extraction has to beinvestigated. Due to the vast amount of different componentsin real feed mixtures, effects, such as emulsification, complexformation etc., might occur. Therefore, since experiments withmodel feeds only give limited information about the extractionfrom a feed with multiple components, the sulfur and nitrogenremoval from real refinery streams from the BP-refinery inRotterdam, The Netherlands, was also investigated. The refin-ery streams studied were: LCCS (Light Catalytically CrackedSpirit), comparable to FCC gasoline, LGO (Light Gas Oil),comparable to diesel and LLCO (Low Sulfur Light Cycle Oil).

Experimental section

Materials and methods

The ionic liquids used in this work were purchased from Iolitec,Merck and Sigma Aldrich. Toluene (p.a., 99.9%), n-heptane

1908 | Green Chem., 2011, 13, 1907–1913 This journal is © The Royal Society of Chemistry 2011

Dow

nloa

ded

by V

ande

rbilt

Uni

vers

ity o

n 06

/05/

2013

09:

59:0

1.

Publ

ishe

d on

10

June

201

1 on

http

://pu

bs.r

sc.o

rg |

doi:1

0.10

39/C

1GC

1519

6G

View Article Online

(p.a., >99%) and ethylbenzene (>99%) were purchased fromMerck. Tetralin (99%), thiophene (98%), benzothiophene (95%),pyrrole (97%), indole (98.5%), carbazole (96%) and acetone (p.a.> 99%) were purchased from Fluka. Dibenzothiophene (98%)has been purchased from Aldrich. Prior to the experiments, theionic liquids have been dried in a rotary evaporator (BuchiRotavapor R-200) at 100 ◦C and under reduced pressure.Subsequently, the water content of the ionic liquids has beendetermined by means of Karl Fischer titration and was in allcases found to be less than 0.1 wt%.

Equipment and experimental procedure

Liquid–liquid extraction experiments were carried out in jack-eted glass vessels with a volume of about 70 mL. The vessels wereclosed with a PVC cover through which a stirrer shaft was led.For each experiment, 10 mL of the feed and 10 mL ionic liquidhave been added and the mixture was stirred (1200 rpm) for15 min to reach equilibrium. In previous work, Meindersmaet al. reported that a mixing time of 5 min is sufficient toreach equilibrium.24 Nevertheless, in order to make sure thatthe phase equilibrium is reached in every case, the extractionexperiment has been continued for 15 min. After stirring, thetwo phases were allowed to settle for about 1 h. This has beendone according to the procedure described by Meindersma etal.24 For phase mixing, two stainless steel propellers, one in thebottom phase and one at the phase interface, with an electronicstirrer (Ika Eurostar) were used. Constant temperature (± 0.1 ◦C)was maintained by means of a water bath (Julabo F32-MW).

Analysis

After equilibrium was reached, a sample of 0.5 mL of each phasewas taken and analysed by gas chromatography (Varian CP-3800). Acetone was added to the samples to avoid phase splittingand to maintain a homogeneous mixture. Ethylbenzene, and incase of the real feed analysis, butanol (0.2 mL for the raffinatesamples and 0.1 mL for the extract phase samples) was used asan internal standard for the GC-analysis. The compositions ofthe components in the samples were analysed by a Varian CP-3800 gas chromatograph with an WCOT fused silica CP-SIL5CB column (50 m ¥ 0.32 mm; DF = 1.2 mm) and a Varian 8200AutoSampler. Since ionic liquids have no vapour pressure, theycannot be analyzed by GC, therefore, only the hydrocarbons ofthe extract and raffinate phase were analyzed and the amountof ionic liquid was calculated by means of a mass balance. Inorder to avoid inaccuracy of the analysis caused by fouling ofthe GC by the ionic liquid, a liner and a pre-column have beenused. Furthermore, measurements were carried out in duplicateto increase accuracy.

The deviation in the calibration curves of 1% and a possiblecontamination of the gas chromatograph can cause a variancein the mole fractions (estimated on 1%). The averages of thetwo measurements were used in our results. The relative averagedeviation in the compositions is about 2.5%.

Results and discussion

To characterize the suitability of a solvent in liquid–liquidextraction the solute distribution ratio (Di) and the selectivity

(S) are widely used parameters. Hence, these parameters areused to present the experimental screening results for thetwo model feeds. The solute distribution coefficient is definedas:

Dsulphur-aromatic = wILsulphur-aromatic/worg

sulphur-aromatic (1)

Dnitrogen-aromatic = wILnitrogen-aromatic/worg

nitrogen-aromatic (2)

Daliphatic = wILaliph/worg

aliph (3)

The selectivity S is derived from the ratio of the distributioncoefficients, according to:

S = Darom/Daliph (4)

The results of the ionic liquids are compared to sulfolane,since sulfolane is the commercial extraction solvent with thehighest aromatic capacity, mostly used for aromatics extractionfrom different petroleum fractions.25,26 Because part of thesulfur in fuels is present in the form of aromatic compounds,sulfolane is a suitable solvent for the extraction desulfurizationof FCC gasoline.3 It is to be expected that some amount ofaromatics will also be extracted. This effect is positive withrespect to EU restrictions regarding aromatics content in dieseland gasoline fuels. Therefore, sulfolane serves as a benchmarkfor the desulfurization, although the reported desulfurizationefficiency of one step is in the range of 50 to 65%. For aromatichydrocarbons, Meindersma et al. showed that for an extractionprocess based on ionic liquids, the influence of capacity is higherthan the influence of selectivity.17 It is assumed that this also canbe applied for sulfur- and nitrogen-containing aromatics. Thus,a prerequisite for an ionic liquid used as an extraction solventfor hetero-aromatic hydrocarbons is a capacity higher than, orat least equal to, that of sulfolane. Additionally, the selectivityof a suitable ionic liquid is preferably equal to that of sulfolaneor higher.

Sulfur

In Fig. 2, the capacity and selectivity of the investigated ionicliquids for thiophene (a) and dibenzothiophene (b) based onweight fractions are depicted.

The black lines shown in both diagrams indicate the extractioncapacity and selectivity of sulfolane for thiophene (a) anddibenzothiophene (b), respectively. It becomes apparent that forboth sulfur aromatics, the pyridinium-based ionic liquids [3-mebupy]N(CN)2 and [4-mebupy]N(CN)2, [4-mebupy]SCN aresuitable candidates since the capacity, as well as the selectivity,are higher than those of sulfolane. In case of the investigatedimidazolium-based ionic liquids, only [bmim]C(CN)3 fulfilsthe criteria and is superior to sulfolane. [Bmim]N(CN)2 onlyhas a higher selectivity, while the selectivity of [bmim]SCN iscomparable to that of sulfolane.

The comparison of the distribution coefficients in Fig. 3shows that the extraction in the investigated ionic liquids is inthe order dibenzothiophene > thiophene > toluene > tetralin.In case of sulfolane, tetralin is better extracted than toluene,thus resulting in the extraction order dibenzothiophene >

thiophene > tetralin > toluene.

This journal is © The Royal Society of Chemistry 2011 Green Chem., 2011, 13, 1907–1913 | 1909

Dow

nloa

ded

by V

ande

rbilt

Uni

vers

ity o

n 06

/05/

2013

09:

59:0

1.

Publ

ishe

d on

10

June

201

1 on

http

://pu

bs.r

sc.o

rg |

doi:1

0.10

39/C

1GC

1519

6G

View Article Online

Fig. 2 Capacity and selectivity of several ionic liquids for thiophene(left) and dibenzothiophene (right) at T = 313.15 K.

Fig. 3 Comparison of distribution coefficients of sulfur model feed atT = 313.15 K.

The interactions between ionic liquids and aromatic hy-drocarbons are based on p–p-interactions where the aromaticsolute is enclosed between the ionic liquid cations in a so-called sandwich-like structure.27 In case of sulfur-containingaromatic components, the interaction with the ionic liquid isalso based on p–p-interaction and thus clathrate formation ofthe ionic liquid cations and the enclosed solvent.28 However,the interactions between sulfur components and ionic liquidsare, due to a higher p-electron density,12,14 stronger than foraromatic hydrocarbons resulting in a higher capacity for the

sulfur components (Fig. 3). Furthermore, the quadrupole mo-ment of poly-aromatics is stronger than for mono-aromaticsresulting in a stronger attraction between cations and solutemolecules.14,29 Thus, the p–p stacking between poly-aromaticsis more pronounced, causing closer packing between them butalso between cations and poly-aromatics.30 This explains thehigher capacity for dibenzothiophene compared to thiophene.The same observation has also been reported elsewhere20 whereit was shown that the higher the poly-aromatic character of acomponent of a diesel model feed is, the better the interactionwith the ionic liquid and thus the higher the extraction.

Additionally, hetero-atoms have the possibility to form hydro-gen bonds with the ionic liquid cation,14 which might also be afactor participating in the interaction between sulfur aromaticsand ionic liquid cations.

This is also visible in Fig. 3, where the extraction capacity ofthe investigated ionic liquids is in the order [3-mebupy]N(CN)2 ~[bmim]C(CN)3 > [4-mebupy]N(CN)2 > [4-mebupy]SCN >

[bmim]N(CN)2 > [bmim]SCN. In earlier work,31,32 it has alreadybeen shown that pyridinium-based ionic liquids are superiorto imidazolium-based ionic liquids with the same anion. Thishas also been reported by Liu et al. and Holbrey et al.who have made the same observation for feeds containingsulfur components.14,33 Pyridinium cations have an aromaticring with a pronounced p-system enabling interaction with thearomatic components through p–p and cation–p interactions.34

In the cases of [3-mebupy]N(CN)2, [4-mebupy]N(CN)2 and [4-mebupy]SCN, compared to the other ionic liquids investigated,the lattice is, due to the anion, obviously less closely packed,therewith enabling a higher solubility of the aromatic compo-nents in pyridinium-based ionic liquids, explaining the highercapacity. The same holds for [bmim]C(CN)3. In comparison to[bmim]SCN and [bmim]N(CN)2, the anion C(CN)3

- provides,due to its size, more room for the poly-aromatic componentsin the lattice. Therefore, the capacity of this ionic liquid ishigher than for the other 1-butyl-3-methylimidazolium-basedionic liquids. This is confirmed by Zhang et al., who reportedthat the extraction capacity of an ionic liquid strongly dependson the space between cation and anion. This means, the tighterthe packing, the less space for enclosed solutes in the ionic liquid,whereas a lower compactness allows for a facile restructuring ofthe ionic liquid and therewith a higher capacity for the dissolvedcomponent.31

Since the suitability for sulfur extraction of an ionic liquid isevaluated compared to sulfolane, this means that for the sulfurmodel feed investigated the ionic liquids [4-mebupy]N(CN)2, [4-mebupy]SCN, [bmim]C(CN)3 and [3-mebupy]N(CN)2 are theresulting suitable candidates because these ionic liquids have, inall four cases, a higher capacity and a higher selectivity thansulfolane.

Comparing these results to the extraction of aromatichydrocarbons,20–23 it is obvious that the same ionic liquids thatwere found to be suitable for the removal of aromatic hydrocar-bons are also promising solvents for desulfurization. Moreover,it is apparent that the ionic liquid [4-mebupy]SCN, which has alower capacity but higher selectivity than sulfolane for aromatichydrocarbons, exhibits a higher capacity than sulfolane forsulfur aromatics. Nevertheless, this result corresponds with thedata reported by Holbrey et al.33

1910 | Green Chem., 2011, 13, 1907–1913 This journal is © The Royal Society of Chemistry 2011

Dow

nloa

ded

by V

ande

rbilt

Uni

vers

ity o

n 06

/05/

2013

09:

59:0

1.

Publ

ishe

d on

10

June

201

1 on

http

://pu

bs.r

sc.o

rg |

doi:1

0.10

39/C

1GC

1519

6G

View Article Online

Table 2 Comparison of different ionic liquids for the extraction of sulfur components

Thiophene Dibenzothiophene

Ionic liquidDthioph

[g g-1]Feed[ppm]

Removed[ppm]

Removed[%]

DDBT

[g g-1]Feed[ppm]

Removed[ppm]

Removed[%] Ref.

[3-mebupy]N(CN)2 0.92 2700 2108 78 1.78 12400 10642 86 twb

[bmim]C(CN)3 0.92 2700 2072 77 1.79 12400 10673 86 twb

[4-mebupy]N(CN)2 0.89 2700 2047 76 1.69 12400 10544 85 twb

[4-mebupy]SCN 0.80 2700 1902 70 1.60 12400 10453 84 twb

Sulfolane 0.77 2700 1747 65 1.17 12400 9717 78 twb

[bmim]N(CN)2 0.74 2700 1664 62 1.10 12400 9565 77 twb

[omim]BF4 0.66 16000 6360 40 2.33 18500 13020 70 10[bmim]SCN 0.47 2700 1138 42 0.73 12400 8660 70 twb

[hmmpy][Tf2N] 0.49 16000 5247 33 2.03 18500 12462 67 32[opy]BF4 0.79 160 70 44 1.79 160 105 66 14[beim]Bu2PO4 0.91a 1300 630 48 1.72 1400 900 63 35

— — — — 0.73 500 211 42 36[bmim]Bu2PO4 0.57a 1900 697 37 1.60 1580 981 62 37

— — — — 1.30 500 283 57 36[eeim]Et2PO4 0.78a 1600 690 44 1.61 2100 1300 62 35

— — — — 0.92 500 240 48 36[hpy]BF4 0.70 160 65 41 1.42 160 95 59 14[omim]Me2PO4 — — — — 1.34 500 286 57 36[emim]N(CN)2 0.72 579 241 42 1.30 500 283 57 38[emim]Et2PO4 0.51a 1150 392 34 1.27 2150 1221 57 37

0.56 500 179 36 36[obim]Bu2PO4 — — — — 1.23 500 275 55 36[beim]Et2PO4 — — — — 1.19 500 272 54 36[oeim]Et2PO4 — — — — 1.19 500 272 54 36[emim]Me2PO4 0.50a 1400 460 33 1.17 1450 790 54 35

— — — — 0.56 500 179 36 36[hmim]Me2PO4 — — — — 1.03 500 254 51 36[hbim]Bu2PO4 — — — — 1.01 500 251 50 36[bbim]Bu2PO4 — — — — 0.90 500 237 47 36[heim]Et2PO4 — — — — 0.89 500 235 47 36[bmim]Me2PO4 — — — — 0.69 500 204 41 36[bmim]Cl/AlCl3

c — — — 4.0 500 225 45 5, 13[bpy]BF4 0.53 160 55 34 0.77 160 70 44 14[bmim]FeCl4 — — — — — 5000 2100 42 39[mmim]Me2PO4 0.059a 800 50 6.25 0.46 1050 350 33 35, 37

— — — — 0.11 500 50 10 36[emim]Bu2PO4 — — — — 0.47 500 160 32 36[bmim]OcSO4 0.70 — — — 1.9 500 150 30 5, 13[bmim]Cl — — — — — 500 90 18 13[bmim]MeSO4 — — — — — 500 90 18 13[bmim]MeSO3 — — — — — 500 90 18 13[bmim]BF4 — — — — 0.7 500 80 16 5, 13[bmim]CF3SO3 — — — — — 500 70 14 13[bmim]BF4 — — — — — 1000 120 12 15[bmim]PF6 — — — — 0.9 500 60 12 5, 13[bmim]PF6 — — — — — 1000 100 10 15[emim]BF4 — — — — — 1000 30 3 15[bmim]Cu2Cl3 — 685 160 23.4 — — — 40

a 3-methyl-thiophene b this work c oil : IL = 5 : 1

Evaluation of sulfur model feed results

Several research groups have investigated different ionic liquidsfor the removal of sulfur components from gasoline anddiesel fuels. Hence, for the sulfur components thiophene anddibenzothiophene, the results after one extraction step in termsof capacity and with view of the amount of sulfur removed aresummarized and compared in Table 2.

Additionally, the data for sulfolane have been incorporatedas a benchmark for the evaluated ionic liquids. As mentionedabove, van Veen et al. reported sulfolane as a promising extrac-tion solvent for sulfur removal from petrochemical streams.3

Due to the lower price of sulfolane, ionic liquids have to besignificantly superior in order to be able to compete with theconventional solvent. This means that the capacity and selec-tivity of the investigated ionic liquids have to be considerablyhigher than those of sulfolane.

The sulfur model feed in this work, see Table 1, contained threesulfur compounds in a solvent mixture, which were extractedsimultaneously. Most other authors report extraction of onlyone sulfur compound from a single solvent, such as dodecane.The initial volumes of the feed and IL phases were both 10 mL,but after extraction, the volume of the extracted phase wasaround 15 mL, because part of the solvent (toluene and tetralin)

This journal is © The Royal Society of Chemistry 2011 Green Chem., 2011, 13, 1907–1913 | 1911

Dow

nloa

ded

by V

ande

rbilt

Uni

vers

ity o

n 06

/05/

2013

09:

59:0

1.

Publ

ishe

d on

10

June

201

1 on

http

://pu

bs.r

sc.o

rg |

doi:1

0.10

39/C

1GC

1519

6G

View Article Online

is co-extracted with the three sulfur compounds. The amountof sulfur compounds removed was calculated from the resultsof the analysis of the sulfur compounds in the raffinate phase,because analysis of the raffinate phase is more reliable than thatof the extract phase containing the IL.

From Table 2, it is apparent that most of the ionic liquidsevaluated show satisfying results for sulfur removal after oneextraction step. This means that in one step 30–80% of thethiophene and 30–90% of the dibenzothiophene (DBT) can beremoved. With respect to a multi-stage extraction, these arepromising results since it indicates that the required maximumsulfur content of 10 ppm for carburant fuels can be met. Thishas also been reported by several authors.5,10,13–15 Nevertheless,the comparison of ionic liquids with conventional extractionsolvents, e.g. sulfolane, has to be made.

As it is obvious from Table 2, only the four ionic liquids[3-mebupy]N(CN)2, [4-mebupy]N(CN)2, [4-mebupy]SCN and[bmim]C(CN)3 are superior to sulfolane. These ionic liquidsoutperform sulfolane in terms of capacity by up to 20% (thio-phene) and even 53% in case of dibenzothiophene. Furthermore,as visible in Fig. 2, the selectivity of these ionic liquids issignificantly higher than the sulfolane selectivity.

Nitrogen

The same ionic liquids investigated for the sulfur model feedhave been evaluated for the removal of nitrogen as well. Afterthe LLE-experiments were carried out for all ionic liquidsand sulfolane, the amount of the nitrogen components inthe raffinate was below the detection limit. This leads to theconclusion that the extraction of nitrogen-containing hetero-aromatic compounds with the investigated ionic liquids is >99%and, therewith, clearly higher than for sulfur components. Thisis confirmed by several authors who report that nitrogen-containing aromatic components are significantly better ex-tracted than sulfur-containing aromatics.4,5,15,32

Removal of sulphur and nitrogen compounds fromrefinery feeds

Based on the results of the model feeds, the ionic liquid [3-mebupy]N(CN)2 has been chosen as the extraction solvent forfurther investigation. The evaluation of real feeds in comparisonto model feeds is essential in order to study the extractionbehaviour of ionic liquids with highly complex feed mixturesand, therewith, to evaluate the suitability of ionic liquids forindustrial aromatic extraction processes. In particular, for theextraction of sulfur and nitrogen components with ionic liquids,the study of real feeds is inevitable since industrial extractionprocesses for these components do not exist yet.

Since real refinery feeds contain a vast amount of components,the experimental results have been analyzed quantitatively foronly the key components present in the sulfur and nitrogenmodel feeds. This has been done by comparing the amounts ofthese components in the feed prior to the measurement and inthe raffinate phase after the experiment.

Thiophene was extracted at 25% and pyrrole at 45% fromLCCS (Light Catalytically Cracked Spirit), both values lowerthan for the model feeds. Indole was extracted at 69% from

LGO (Light Gas Oil) and at 64% from LLCO (Low Sulfur LightCycle Oil) and carbazole was extracted at 38% from LGO. Allthese values are lower than for the model feeds used. This can beexplained by a lower extraction efficiency due to the vast amountof components present in the real feed that have competingeffects. On the other hand, it is obvious that some values ofthe real feed analysis deviate considerably from the model feedresults and, therewith, from the expected range of extractedreal feed component quantities. Furthermore, dibenzothiophenecould not be detected in the real feeds LGO and LLCO, whileit should be present in these feeds. Nevertheless, the real feedresults confirm that both the sulfur and nitrogen aromaticcomponents can be extracted, although the extraction efficiencyis less than for the investigated model feeds.

Conclusions

The pyridinium-based ionic liquids [3-mebupy]N(CN)2, [4-mebupy]N(CN)2, [4-mebupy]SCN, as well as the imidazolium-based ionic liquids [bmim]C(CN)3, [bmim]N(CN)2 and[bmim]SCN have been investigated for the removal of sulfur-and nitrogen-containing hetero-aromatics from petrochemicalstreams. In the case of sulfur, it can be concluded thatthe ionic liquids [3-mebupy]N(CN)2, [4-mebupy]N(CN)2, [4-mebupy]SCN and [bmim]C(CN)3 are superior to sulfolane,which has been used as a benchmark. These ionic liquids exhibitan up to 20% higher capacity in case of thiophene and, inthe case of dibenzothiophene, even up to 53%. Finally, it wasobserved that nitrogen-containing hetero-aromatic componentsare significantly better extracted than sulfur components.

Experiments with real refinery streams confirmed that sulfurand nitrogen compounds can be removed from these streams,albeit with a lower efficiency than with model feeds.

Acknowledgements

We acknowledge BP Chemicals for their financial support forthis work.

References1 C. D. Cooper, in Kirk-Othmer Encyclopedia of Chemical Technology

(5th Edition), John Wiley & Sons, Inc., 2007, vol. 26, pp. 667–729.

2 I. V. Babich and J. A. Moulijn, Fuel, 2003, 82, 607–631.3 E. Ito and J. A. R. van Veen, Catal. Today, 2006, 116, 446–460.4 L. L. Xie, A. Favre-Reguillon, S. Pellet-Rostaing, X. X. Wang, X.

Z. Fu, J. Estager, M. Vrinat and M. Lemaire, Ind. Eng. Chem. Res.,2008, 47, 8801–8807.

5 J. Eßer, P. Wasserscheid and A. Jess, Green Chem., 2004, 6, 316–322.6 J. G. Speight, in Kirk-Othmer Encyclopedia of Chemical Technology

(5th Edition), John Wiley & Sons, Inc., 2006, vol. 18, pp. 639–682.

7 H. Fischer, Chem. Ing. Tech., 1967, 39, 515–520.8 W. Dabelstein, A. Reglitzky, A. Schutze and K. Reders, in Ullmann’s

Encyclopedia of Industrial Chemistry, Wiley-VCH Verlag GmbH &Co. KGaA, 2000, DOI: 10.1002/14356007.a16_719.pub2, electronicversion, 34 pp.

9 U. Domanska and M. Laskowska, J. Chem. Thermodyn., 2009, 41,645–650.

10 L. Alonso, A. Arce, M. Francisco, O. Rodriguez and A. Soto, AIChEJ., 2007, 53, 3108–3115.

11 J.-l. Wang, D.-s. Zhao, E.-p. Zhou and Z. Dong, J. Fuel Chem.Technol., 2007, 35, 293–296.

1912 | Green Chem., 2011, 13, 1907–1913 This journal is © The Royal Society of Chemistry 2011

Dow

nloa

ded

by V

ande

rbilt

Uni

vers

ity o

n 06

/05/

2013

09:

59:0

1.

Publ

ishe

d on

10

June

201

1 on

http

://pu

bs.r

sc.o

rg |

doi:1

0.10

39/C

1GC

1519

6G

View Article Online

12 S. G. Zhang and Z. C. Zhang, Green Chem., 2002, 4, 376–379.13 A. Bosmann, L. Datsevich, A. Jess, A. Lauter, C. Schmitz and P.

Wasserscheid, Chem. Commun., 2001, 2494–2495.14 H. Gao, M. Luo, J. Xing, Y. Wu, Y. Li, W. Li, Q. Liu and H. Liu, Ind.

Eng. Chem. Res., 2008, 47, 8384–8388.15 S. Zhang, Q. Zhang and Z. C. Zhang, Ind. Eng. Chem. Res., 2004,

43, 614–622.16 L.-L. Xie, A. Favre-Reguillon and M. Lemaire, Selective removal

of nitrogen-containing compounds from straight-run diesel fuel usingILs, International Solvent Extraction Conference (ISEC) Tucson,Az, USA, 2008, 1325–1330.

17 G. W. Meindersma and A. B. de Haan, Chem. Eng. Res. Des., 2008,86, 745–752.

18 A. Heintz, J. Chem. Thermodyn., 2005, 37, 525–535.19 L. L. Xie, A. Favre-Reguillon, X.-X. Wang, X. Fu, M. Vrinat and M.

Lemaire, Ind. Eng. Chem. Res., 2009, 48, 3973–3977.20 A. R. Hansmeier, PhD thesis, Eindhoven University of Technology,

2010.21 A. R. Hansmeier, M. Minoves Ruiz, G. W. Meindersma and A. B. de

Haan, J. Chem. Eng. Data, 2010, 55, 708–713.22 A. R. Hansmeier, M. Jongmans, G. W. Meindersma and A. B. d.

Haan, J. Chem. Thermodyn., 2010, 42, 484–490.23 G. W. Meindersma, A. R. Hansmeier and A. B. de Haan, Ind. Eng.

Chem. Res., 2010, 49, 7530–7540.24 G. W. Meindersma, A. Podt and A. B. de Haan, Fuel Process.

Technol., 2005, 87, 59–70.25 R. P. Tripathi, A. R. Ram and P. B. Rao, J. Chem. Eng. Data, 1975,

20, 261–264.

26 Y. Yorulmaz and F. Karpuzcu, Chem. Eng. Res. Des., 1985, 63, 184–190.

27 J. D. Holbrey, W. M. Reichert, M. Nieuwenhuyzen, O. Sheppard, C.Hardacre and R. D. Rogers, Chem. Commun., 2003, 476–477.

28 B.-M. Su, S. Zhang and Z. C. Zhang, J. Phys. Chem. B, 2004, 108,19510–19517.

29 G. L. Heard and R. J. Boyd, Chem. Phys. Lett., 1997, 277, 252–256.30 B. K. Mishra, V. K. Bajpai, V. Ramanathan, S. R. Gadre and N.

Sathyamurthy, Mol. Phys., 2008, 106, 1557–1566.31 J. Zhou, J. Mao and S. Zhang, Fuel Process. Technol., 2008, 89, 1456–

1460.32 M. Francisco, A. Arce and A. Soto, Fluid Phase Equilib., 2010, 294,

39–48.33 J. D. Holbrey, I. Lopez-Martin, G. Rothenberg, K. R. Seddon, G.

Silvero and X. Zheng, Green Chem., 2008, 10, 87–92.34 S. Tsuzuki, M. Mikami and S. Yamada, J. Am. Chem. Soc., 2007,

129, 8656–8662.35 X. Jiang, Y. Nie, C. Li and Z. Wang, Fuel, 2008, 87, 79–84.36 E. Kuhlmann, M. Haumann, A. Jess, A. Seeberger and P. Wasser-

scheid, ChemSusChem, 2009, 2, 969–977.37 Y. Nie, C. Li, A. Sun, H. Meng and Z. Wang, Energy Fuels, 2006, 20,

2083–2087.38 G. Yu, X. Li, X. Liu, C. Asumana and X. Chen, Ind. Eng. Chem.

Res., 2010, 50, 2236–2244.39 N. H. Ko, J. S. Lee, E. S. Huh, H. Lee, K. D. Jung, H. S. Kim and M.

Cheong, Energy Fuels, 2008, 22, 1687–1690.40 C. Huang, B. Chen, J. Zhang, Z. Liu and Y. Li, Energy Fuels, 2004,

18, 1862–1864.

This journal is © The Royal Society of Chemistry 2011 Green Chem., 2011, 13, 1907–1913 | 1913

Dow

nloa

ded

by V

ande

rbilt

Uni

vers

ity o

n 06

/05/

2013

09:

59:0

1.

Publ

ishe

d on

10

June

201

1 on

http

://pu

bs.r

sc.o

rg |

doi:1

0.10

39/C

1GC

1519

6G

View Article Online