Fuel 113 (2013) 216–220

Contents lists available at SciVerse ScienceDirect

Fuel

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

Efficient solvent regeneration of Basolite C300 used in the liquid-phaseadsorption of dibenzothiophene

0016-2361/$ - see front matter � 2013 Elsevier Ltd. All rights reserved.http://dx.doi.org/10.1016/j.fuel.2013.05.065

⇑ Corresponding authors. Fax: +34 915 854 760.E-mail addresses: j.m.campos@icp.csic.es (J.M. Campos-Martin), jlgfierro@icp.

csic.es (J.L.G. Fierro).1 http://www.icp.csic.es/eqsgroup/.

G. Blanco-Brieva a, J.M. Campos-Martin a,⇑, S.M. Al-Zahrani b, J.L.G. Fierro a,⇑a Sustainable Energy and Chemistry Group (EQS), Instituto de Catálisis y Petroleoquímica, CSIC, c/Marie Curie, 2 Cantoblanco, 28049 Madrid, Spain1

b Chemical Engineering Department, College of Engineering, King Saud University, Riyadh, Saudi Arabia

h i g h l i g h t s

� The spent C300 adsorbent can be regenerated by washing with alcohols.� Initial adsorption capacity of the fresh C300 can be recovered after regeneration with methanol.� After three desulfurization–regeneration cycles the adsorption capacity remain constant.

a r t i c l e i n f o

Article history:Received 4 February 2013Received in revised form 10 May 2013Accepted 20 May 2013Available online 10 June 2013

Keywords:Metal–organic frameworksSulfur compoundsAdsorptionRegeneration

a b s t r a c t

The metal–organic framework (MOF) Basolite C300 displays excellent performance for the removal ofdibenzothiophene (DBT) in a model system consisting in DBT in a mixture of 2,2,4-trimethylpentane(TMP). With the aim to exploit this property in a continuous operation, it is imperative to develop a sim-ple and effective regeneration procedure. Our previous work (Fuel 105 (2013) 459–465) demonstratedthat there is only partial thermal regeneration of this system at 473 K, which makes this approach of lim-ited value. In the present work, we developed a significantly more effective regeneration procedure inwhich the DBT-saturated MOF C300 system was washed with polar solvents, such as methanol, ethanoland isopropanol. It was found that regeneration using methanol as a solvent at temperatures near ambi-ent (304 K) is highly effective and allows for the recovery of the initial adsorption capacity. After severalregeneration cycles, both the crystallinity and the initial sorption capacity of the C300 system are essen-tially unchanged.

� 2013 Elsevier Ltd. All rights reserved.

1. Introduction

In recent years, sulfur removal from hydrocarbon fuels, such asdiesel and gasoline, has resulted in a global effort to prevent airpollution that includes the deactivation of exhaust catalysts [1,2].Mandatory environmental fuel specifications introduced by theEuropean Standards Organization (CEN) and the federal govern-ment of the United States (EPA) require that refineries reduce thelevels of sulfur in gasoil for heating purposes and in marine dieselto 1000 ppmw by 2008 and 2010, respectively. Additionally, thereis a demand for ultra-low-sulfur fuel (preferably down to0.1 ppmws) [3].

In the petroleum industry, low-sulfur fuels are often obtainedthrough hydrocracking or hydrotreating processes (HDS) [4]. Theorganic S-compounds are converted to H2S and the correspondinghydrocarbon. These reactions are conducted under high hydrogen

pressure (typically 40 bar) and at temperatures around 670 K inrefinery units using Ni(Co)–Mo(W) sulfide phases that are depos-ited on alumina substrates. Although current HDS processes arehighly efficient and technologically important, they do not appearto be suitable for the production of ultra-clean fuel. While they effi-ciently eliminate aliphatic and alicyclic sulfur compounds andthiophenes, they are much less effective at removing the more ste-rically hindered benzothiophene (BT), dibenzothiophene (DBT),and 4,6-dimethyldibenzothiophene (DMDBT) fuel contaminantsdue to the sterically hindered adsorption of these compounds onthe catalyst surface [5]. Among the novel technologies proposedfor deep fuel desulfurization, adsorption appears to be a soft tech-nology for sulfur removal because hydrogen is not required, andthe process operates under mild conditions [6–8].

The ability to use the adsorption process at ambient tempera-ture and pressure would be a major advance in the petroleumrefinery industry. One of the challenges of the adsorptive desulfur-ization process is that the sorbent material should remove only thesulfur compounds without adsorbing the other aromatics andolefins in the fuel. For on-board or onsite fuel cell applications,

G. Blanco-Brieva et al. / Fuel 113 (2013) 216–220 217

the adsorptive desulfurization should be performed at near ambi-ent temperatures. It is also mandatory that the adsorbent materialbe regenerable to reduce costs.

Several sorbents have been previously studied: zeolites, silica,carbon, alumina [9,10] and metal–organic frameworks (MOF) sys-tems [1,11–14], with the latter being the most promising candi-date. Metal–organic frameworks are crystallized hybrid poroussolids built up from inorganic subunits (clusters, chains, or layers)that are connected through organic linkers, such as carboxylates orphosphonates [15]. They are composed of metal ions and organiclinkers, the latter of which act as bridging ligands between the me-tal ions to form highly ordered frameworks. They exhibit a hugevariety of structures that sometimes have very large pores andhigh surfaces areas, which are qualities that lead to excellentadsorption.

We recently reported the use of MOF systems as sorbents fororganosulfur compounds present in transportation fuels andshowed that this system is an efficient complement to the conven-tional hydrodesulfurization process [12]. However, the thermalregeneration of this sorbent results in a partial recuperation ofthe original adsorption capacity [16]. Part of this incomplete regen-eration can be attributed to the limited regeneration temperature,as the adsorbent is not stable at high temperatures. The aim of thiswork is to study the regeneration of a MOF (BASF C300) used in theadsorptive desulfurization of fuels via solvent washing.

2. Experimental section

2.1. Materials

Solvents, dibenzothiophene and the Basolite C300 [Cu3(C9H3

O6)2] metal–organic framework were purchased from Sigma–Al-drich and used without further purification.

2.2. Adsorbent characterization

The thermogravimetric analyses of the used MOFs were per-formed with a Perkin-Elmer TGS2 instrument using a heating rateof 10 K min�1 under a nitrogen flow (60 mL min�1).

The X-ray diffraction (XRD) patterns of the fresh and used C300samples were recorded using a Seifert 3000P vertical diffractometerand nickel-filtered Cu Ka radiation (k = 0.1538 nm) using the stan-dard powder diffraction procedures. A standard glass slide was usedfor the background corrections. The crystallinity of the C300 MOFsystem in both the fresh and the used samples was studied to detectany possible deterioration of the crystalline structure when used inthe adsorption experiments. The crystallinity percentage wasdetermined using the ratio of the sum of the relative intensity ofthe five most intense peaks, according to the following equation:

%crystallinity ¼P5

i¼1IrelsampleP5

i¼1Irelsample

� 100 ð1Þ

In this calculation, the fresh C300 sample was taken as the standard(100% crystallinity).

The textural properties were determined using the adsorption–desorption nitrogen isotherms recorded at 77 K with a Micromer-itics TriStar 3000. The specific area was calculated by applyingthe BET equation to the values of the nitrogen adsorption iso-therms within the relative pressure (P/P0) range from 0.03 to 0.3and using a value of 0.162 nm2 for the cross-section of an adsorbednitrogen molecule at 77 K. The pore size distributions were com-puted by applying the Horvath-Kawazoe model to the desorptionbranch of the nitrogen adsorption–desorption isotherms.

X-ray photoelectron spectra (XPS) were acquired with a VGEscalab 200R spectrometer equipped with a hemispherical

electron analyzer utilizing a Mg Ka (hm = 1253.6 eV) non-mono-chromatic X-ray source. The samples were degassed in the pre-treatment chamber at room temperature for 1 h prior to beingtransferred into the instrument’s ultra-high vacuum analysischamber. The high resolution Cu2p, O1s, S2p and C1s spectra werescanned several times at pass energy of 20 eV, to obtain good sig-nal-to-noise ratios. The binding energies (BE) were referenced withrespect to the BE of the C1s core-level spectrum at 284.9 eV. Theinvariance of the peak shapes and widths at the beginning andend of the analyses indicated a constant charge across all of themeasurements. The peaks were fitted using a non-linear leastsquare fitting routine after background subtraction and a properlyweighted sum of Lorentzian and Gaussian component curves. Thesurface atomic ratios were estimated from peak areas, normalizedto silicon, and corrected using the corresponding sensitivity factors[17].

2.3. Adsorption measurements

A model diesel fuel (MDF) was prepared for the adsorptionexperiments. This fuel contained a molar concentration of dibenzo-thiophene (DBT) in a mixture of 2,2,4-trimethylpentane (TMP).Prior to use, the MOF samples were degassed under vacuum to re-move any adsorbed water or solvents. Both fresh and regeneratedMOF C300 sorbents (see below) were tested in a liquid-phase glassbatch reactor operating at atmospheric pressure with constant stir-ring. The procedure involved suspending 1 g of C300 in 100 g of asolution of the sulfur compound (DBT) in TMP. This mixture waskept at 304 K with vigorous stirring for 72 h to reach the thermo-dynamic equilibrium of adsorption. The sorbent saturated withthe organosulfur compound was then separated from the liquidphase via filtration for subsequent characterization, while the li-quid phase was analyzed by GC-FID to evaluate the sulfur com-pound concentration. All measurements were performed at leastthree times for reproducibility test.

2.4. Regeneration treatment

The regeneration of the used MOF was accomplished using thefollowing procedure: the saturated adsorbent was put in contactwith the solvent with vigorous stirring at 304 K for 72 h. The solidwas then recovered by filtration and dried at 373 K to remove anyadsorbed solvent or water.

3. Results and discussion

We began the study with fresh C300 samples because thesesamples were used in the adsorption of DBT in a model systemconsisting in DBT in a mixture of 2,2,4-trimethylpentane (TMP),under the conditions selected in our previous works [12,16]: atemperature of 304 K, an initial S-concentration of 1 wt%, and atime period of 72 h. Under these conditions, the adsorption equi-librium is reached with an adsorption capacity of 58.1 g S/kgsorbent.

Attempts were made to regenerate the spent C300 sample bywashing the adsorbent with three different polar solvents: metha-nol, ethanol and isopropanol (50:1 solvent:sorbent). The solventswere selected based on their Hildebrand solubility factors (Table1) [18,19], as the higher polarity helps to remove the sulfur com-pounds. Consequently, solvents with a d value higher than 22 areexpected to be good candidates for the extraction of these com-pounds (Table 1). After regeneration, the samples were reusedfor the adsorption of DBT. The results shown in Table 2 indicatethat the extraction efficiency depends on the polarity of the sol-vent. The adsorption capacity for fresh samples and samples

Table 1Hildebrand solubility parameters.

Solvent d d (SI)

Methanol 14.28 29.7Ethanol 12.92 26.2Isopropanol 11.50 24.9

Table 2Summary of the extent of DBT adsorption at 304 K after the regeneration with polarsolvents.

Sorption capacity(gS/kg adsorbent)

Standard deviation(gS/kg adsorbent)

Fresh C300 58.10 1.02Methanol 50:1 57.60 1.32Ethanol 50:1 51.97 1.47Isopropanol 50:1 48.58 0.90

Table 3Textural properties of fresh and regenerated MOF Basolite C300.

Adsorbent BET surfacearea (m2/g)

Porediameter(nm)

Microporevolume(mL/g)

C300 1277 0.61 0.78C300 regenerated with methanol 964 0.57 0.64C300 regenerated with ethanol 783 0.50 0.48C300 regenerated with

isopropanol569 0.51 0.33

Table 4Evolution of C300 crystallinity after the use and regen-eration with different solvents at 304 K.

Crystallinity (%)

Fresh C300 100Used DBT 82.53Methanol 97.03Ethanol 94.40Isopropanol 87.10

300 400 500 600 70020

40

60

80

100

Wei

ght%

Temperature (K)

C300 Fresh C300 Methanol C300 Ethanol C300 Isopropanol C300 Used with DBT

Fig. 1. Thermogravimetric profiles of the fresh and used MOF C300 samples.

218 G. Blanco-Brieva et al. / Fuel 113 (2013) 216–220

regenerated with methanol is close to coincidence, thus indicatingthat methanol washing can almost completely regenerate theadsorbent. The use of ethanol and isopropanol as solvents resultedin the incomplete recovery of the initial adsorption capacity. Theregeneration capacity of the solvent can be correlated to the valueof the Hildebrand solubility parameter (d). It appears that an in-crease in the value of the d parameter results in a higher adsorptioncapacity recovery.

Fresh and regenerated samples were analyzed by nitrogenadsorption isotherms. All samples exhibited a type I isotherm ofthe IUPAC classification (not show), isotherms reach a plateau atrelative pressures as low as 0.1, and then remain essentially flatthroughout the entire relative pressure interval. This kind of iso-therms is typical of solids with a microporous system with almostno mesoporosity as was expected for a MOF [20]. Fresh C300 sam-ple shows a very high BET surface area (1277 m2/g) (Table 3) with apore diameter of 0.61 nm. Regenerated samples show a cleardecrease in the BET surface area, pore diameter and microporesvolume (Table 3), textural properties decreases depending on thesolvent used in the regeneration treatment, methanol > etha-nol > isopropanol. Textural data of regenerated samples are relatedwith the adsorption properties of regenerated samples, becausethe sample who has lost less surface area, pore volume or diameter(treated with methanol) recover almost completely the adsorptioncapacity, while sample who has lost more surface area, pore vol-ume or diameter (treated with isopropanol) recover less adsorp-tion capacity.

The MOF structure is quite flexible and can be modified by theadsorption of guest molecules [15]. Changes in the crystallinity ofC300 are therefore expected to occur during the adsorption ofS-containing molecules. Accordingly, a measurement of the crys-tallinity of the C300 substrate can be used to evaluate the possiblechanges brought on by the regeneration treatment. After the

adsorption of sulfur containing organic compounds, the crystallin-ity decreases with respect to the original MOF (Table 4). A close tototal recovery of crystallinity is observed after the sample is regen-erated using the methanol solvent, and the recovery is conse-quently lower for samples regenerated when ethanol orisopropanol is used as the solvent (Table 4). Again, these resultssuggest that the S-containing molecules can be released from thesorbent material using solvent washing.

Fig. 1 shows the thermogravimetric profiles of the fresh andused MOF C300 samples used in the adsorption of the DBT com-pound. The fresh C300 sample shows two major weight losses:one below 420 K and another within the range of 550–570 K. Theweight loss of the sample below 420 K is due to the desorptionof adsorbed water, which can be easily recognized using thechange in color of the sample because the hydrated material isblue-turquoise and the dehydrated material is dark violet. Thesamples used in the adsorption of organosulfur compounds notonly show the weight losses above but also show a smaller lossnear 450 K, which is attributed to the desorption of the S-contain-ing compound and because the adsorption capacity cannot becompletely regenerated with a thermal treatment [16] at low tem-peratures (<473 K); some of the sulfur compounds must be des-orbed at higher temperatures. The large weight loss that takesplace at 550–570 K (Fig. 1) is unambiguously attributed to thedecomposition of the C300, as confirmed by the final weight loss

Table 5Binding energies (eV) of the core levels of the C300 samples used in the adsorption ofDBT.

Sample Cu2p3/2 S2p S/Cu at

C300 used 935.3 163.9 (43)169.2 (57) 0.149C300 used and regenerated

with Methanol935.3 163.8 (57)169.2 (43) 0.019

C300 used and regeneratedwith Ethanol

935.3 163.9 (41)169.2 (59) 0.028

C300 used and regeneratedwith2-Propanol

935.2 164.0 (0)169.2 (100) 0.047

In parentheses are peak percentages.

Table 6Extent of DBT adsorption at 304 K after the regeneration with polar solvents usingdifferent solvent/sorbent ratios.

Solvent/adsorption ratio

Sorption capacity (gS/kgadsorbent)

Standard deviation (gS/kgadsorbent)

Methanol 25:1 47.00 0.45Methanol 50:1 57.60 1.32Methanol 75:1 57.20 0.82Ethanol 35:1 31.00 1.10Ethanol 50:1 51.97 1.47Ethanol 75:1 53.10 1.52Isopropanol 25:1 46.65 1.87Isopropanol 50:1 48.58 0.90Isopropanol 75:1 47.75 1.37

Table 7Adsorption of DBT at 304 K with MOF C300 regenerated with methanol (50:1).

Uses Sorption capacity(gS/kg adsorbent)

Standard deviation(gS/kg adsorbent)

Fresh 58.10 1.021st regeneration cycle 57.63 0.312nd regeneration cycle 56.29 0.633rd regeneration cycle 56.28 0.65

G. Blanco-Brieva et al. / Fuel 113 (2013) 216–220 219

of approximately 24.3%, which is expected for the completedecomposition of the Basolite [Cu3(C9H3O6)2] structure into CuO.The samples regenerated using solvent washing do not present aweight loss near 450 K, and their behavior is similar to that ofthe fresh sample.

All the samples were studied using X-ray photoelectron spec-troscopy (XPS) to reveal the chemical environment around the ex-posed atoms of the C300 substrate. The line shape of the Cu2p levelshows the binding energy, the line splitting, and the satellite struc-ture characteristics of the Cu(II) compounds. The binding energy ofthe most intense Cu2p3/2 peak at 935.0–935.1 eV (Table 2), and theintensity ratio of the satellite to the principal peak (Isat/ICu2p3/2) of0.54 ± 0.02 are similar to the reported values for the cupric organicsalts and coincide with the reported values for the fresh C300 sam-ples [12,16]. The examination of the energy region for the S 2pcore-level spectra also provides relevant chemical information.Upon applying a peak-fitting procedure to the experimental S 2pspectra, two sulfur species were detected (Table 5). One appearedat a binding energy of 163.9 eV, which is typical of S–C bonds in or-ganic compounds [21,22], and another appeared at approximately169.2 eV, which originates from S(VI) in sulfone-like species [21].No S 2p component of a sulfoxide species has been observed near166.0 eV [15]. The observation of the highly oxidized S(VI) speciescan be taken as conclusive evidence of the strong interaction be-tween the S-atom of the adsorbate and the MOF surface via a truechemisorption process. As explained in an earlier study [12,16], theappearance of these oxidized S(VI) species was unexpected giventhe high stability of the S-atom in the aromatic rings of DBT, thelack of oxidant present in the liquid phase and the lack of changein the chemical state of the Cu2+ ions in the C300 sample duringthe adsorption process. A tentative explanation for this behavioris based on the presence of chemisorbed oxygen, like happens oncarbons [21].

Despite the limitations of XPS analysis on these types of sam-ples due to the electron mean free path in solid surfaces (typically2–3 nm), the chemical information corresponds to the topmostatomic layers, and some interesting information can be obtained.The proportion of each sulfur species depends on the sample ana-lyzed. The used sample exhibited a slightly higher proportion ofthe S(VI) species, and this proportion is similar to the sampletreated with ethanol. On the contrary, the sample treated withisopropanol only exhibits the presence of a S(VI) species, and thesample regenerated with methanol shows a higher proportion ofa S(II) species. All three of the treated samples showed a lower sur-face concentration of sulfur than the used samples, and the surfaceatomic ratio of S/Cu decreases with the efficiency of the regenera-tion (methanol < ethanol < isopropanol). This experimental dataindicates that the most effective solvent studied was methanol,which is in agreement with the adsorption data; however, it isnot the most efficient solvent for removing the S(II) species.

The effect of the ratio of solvent/used MOF that is necessary forthe regeneration of the spent sample was also studied. The

adsorption capacity recovered using a small amount of solventwas not efficient but increased as the ratio of solvent/used MOF in-creased. The adsorption capacity recovered reached a maximumand remained stable near a solvent/solid ratio of 50–75:1 (Table6). The minimum amount of solvent necessary to regenerate thespent adsorbent depended on the solvent used, and the ratio was50:1 for methanol and near 75:1 for ethanol and isopropanol.Methanol was the solvent that yielded the highest regenerationof adsorption capacity independent of the solvent/solid ratio em-ployed. We consider methanol to be the best regeneration solventbecause it can recover the adsorption capacity with a lower sol-vent/solid ratio. For this reason, we tried to test the regenerationproperties of methanol using several cycles of DBT adsorption/methanol regeneration of the C300. After three desulfurization-regeneration cycles, no significant loss in the sulfur adsorptioncapacity was observed (Table 7). These data indicate that theadsorption procedure can be utilized and regenerated severaltimes.

4. Conclusions

The spent C300 adsorbent can be easily regenerated via wash-ing with low molecular weight alcohols. The almost completerecovery of the adsorption capacity of the fresh C300 can be ob-tained following the regeneration with methanol. This method isclearly more effective than thermal treatments [22]. The crystallin-ity of the spent C300 decreases due the high adsorption level of theorganosulfur compound (58.1 g S/kg sorbent), but for the sorbentregenerated with methanol at 304 K, the crystallinity is almostcompletely recovered. Three desulfurization–regeneration cycleswere performed without any evident loss of adsorption capacity.The solvent used in the regeneration of the spent sorbent can thenbe treated in a small HDS reactor to remove the sulfur, and theremaining desulfurized organic moiety can then be blended withthe fuel.

Acknowledgement

We thank the financial support of our research sponsor: TheKing Saud University, Riyadh (Saudi Arabia).

220 G. Blanco-Brieva et al. / Fuel 113 (2013) 216–220

References

[1] Achmann S, Hagen G, Hämmerle M, Malkowsky IM, Kiener C, Moos R. Sulfurremoval from low-sulfur gasoline and diesel fuel by metal–organicframeworks. Chem Eng Technol 2010;33:275–80.

[2] Bartholomew CH, Agrawal PK, Katzer JR. Sulfur poisoning of metals. In: EleyHPDD, Paul BW, editors. Advances in catalysis. Academic Press; 1982. p.135–242.

[3] Zhang X, Chan SH, Li G, Ho HK, Li J, Feng Z. A review of integration strategies forsolid oxide fuel cells. J Power Sources 2010;195:685–702.

[4] Babich IV, Moulijn JA. Science and technology of novel processes for deepdesulfurization of oil refinery streams: a review. Fuel 2003;82:607–31.

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

[6] Salem ABSH, Hamid HS. Removal of sulfur compounds from naphtha solutionsusing solid adsorbents. Chem Eng Technol 1997;20:342–7.

[7] Hernández-Maldonado AJ, Yang RT. Desulfurization of transportation fuels byadsorption. Cat Rev – Sci Eng 2004;46:111–50.

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

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

[10] Sano Y, Choi K-H, Korai Y, Mochida I. Adsorptive removal of sulfur and nitrogenspecies from a straight run gas oil over activated carbons for its deephydrodesulfurization. B: Environ 2004;49:219–25.

[11] Cychosz KA, Wong-Foy AG, Matzger AJ. Liquid phase adsorption bymicroporous coordination polymers: removal of organosulfur compounds. JAm Chem Soc 2008;130:6938–9.

[12] Blanco-Brieva G, Campos-Martin JM, Al-Zahrani SM, Fierro JLG. Effectiveness ofmetal–organic frameworks for removal of refractory organo-sulfur compoundpresent in liquid fuels. Fuel 2011;90:190–7.

[13] Shi F, Hammoud M, Thompson LT. Selective adsorption of dibenzothiopheneby functionalized metal organic framework sorbents. Appl Catal B: Environ2011;103:261–5.

[14] Peralta D, Chaplais G, Simon-Masseron A, Barthelet K, Pirngruber GD. Metal–organic framework materials for desulfurization by adsorption. Energy Fuels2012;26:4953–60.

[15] Ferey G. Hybrid porous solids: past, present, future. Chem Soc Rev 2008;37:191–214.

[16] Blanco-Brieva G, Campos-Martin JM, Al-Zahrani SM, Fierro JLG. Thermalregeneration of the metal organic frameworks used in the adsorption ofrefractory organosulfur compounds from liquid fuels. Fuel 2013;105:459–65.

[17] Wagner CD, Davis LE, Zeller MV, Taylor JA, Raymond RH, Gale LH. Empiricalatomic sensitivity factors for quantitative analysis by electron spectroscopy forchemical analysis. Surf Interface Anal 1981;3:211–25.

[18] Hansen CM. The three dimensional solubility parameter-key to paintcomponent affinities. J Paint Technol 1967;139:104.

[19] Wiehe IA. Polygon mapping with two-dimensional solubility parameters. IndEng Chem Res 1995;34:661–73.

[20] Li J-R, Kuppler RJ, Zhou H-C. Selective gas adsorption and separation in metal–organic frameworks. Chem Soc Rev 2009;38:1477–504.

[21] Seredych M, Hulicova-Jurcakova D, Bandosz TJ. Effect of the incorporation ofnitrogen to a carbon matrix on the selectivity and capacity for adsorption ofdibenzothiophenes from model diesel fuel. Langmuir 2009;26:227–33.

[22] Jiménez Mateos JM, Fierro JLG. X-ray photoelectron spectroscopic study ofpetroleum fuel cokes. Surf Interface Anal 1996;24:223–36.