Solvent Extraction Research and Developent, Japan, Vol. 22, No 1, 95 11 (215) Notes Separation of Nitrogen Heterocyclic Copounds fro Model Oils by an Eulsion Liquid Mebrane Hiroaki HABAKI, Yoshiyuki SHIMADA and Ryuichi EGASHIRA * Departent of International Developent Engineering, Tokyo Institute of Technology, 12-1, O-okayaa-2-choe, Meguro-ku, Tokyo 152-855, Japan (Received March 28, 214; Accepted May, 3, 214) The separation of nitrogen heterocyclic copounds by an (O/W)/O eulsion liquid ebrane was studied. uinoline and 2-ethylnaphthalene were selected as the representative coponents of nitrogen heterocyclic and hoocyclic copounds, and a toluene or heptane solution of these two copounds was used as the odel feed oil. Deionized water or an aqueous solution of 1,3-butandiol was used as the ebrane solution. uinoline preferably pereated through the eulsion liquid ebrane and could be separated fro 2-ethylnaphthalene in the odel feed oils. The addition of 1,3-butandiol in the ebrane solution could enhance the pereation rates. The separation selectivity of quinoline relative to 2-ethylnaphthalene was larger with the toluene feed oil copared with the heptane solution. 1. Introduction Soe nitrogen heterocyclic copounds are useful as interediates for the production of agricultural cheicals, edicines, perfues, and any other useful cheicals. Coal tar and soe coal-derived oils are regarded as the possible sources for these valuable nitrogen heterocyclic copounds. On the other hand, soe fractions produced fro petroleu refinery processes also contain these copounds which occasionally cause soe trouble by reducing the catalytic efficiency or shorten its lifetie, and degradation in the quality of the fuel oils due to autoxidation or cheical reaction during storage [1]. Hydrotreatent is the ost popular way to reove these coponents, which effectively decoposes the coponents but the process is expensive and consues uch hydrogen [1,2]. For both type of oils, effective techniques to separate the heterocyclic copounds have been expected to develop. To recover these copounds, acidic and basic extraction was reported [3], and soe drawbacks were pointed out, such as corrosion of the equipent, difficulties in solvent recovery, and so on. As alternative ethods, ordinary liquid-liquid extraction without reaction [4-7] and (O/W)/O liquid ebranes [9-11] have been studied. In these reports, catalytically cracked diesel oil, light cyclic oil, and coal tar absorption oil were used as source oils for the nitrogen heterocyclic copounds. In a study of the separation of coal tar absorption oil, the distribution coefficients of nitrogen heterocyclic and aroatic copounds were easured, in which water and coal tar absorption oil diluted by toluene or heptane were used as the aqueous and organic phases to select the favorable extract phase for the eulsion liquid ebrane separation. The distribution coefficients of quinolone, indole and ethylnaphthalene were affected by the diluents and were larger with heptane as the - 95 -
extract phase than those with toluene [11]. However, the effects of the diluents or of the ajor coponent in the feed phase on the (O/W)/O eulsion liquid ebrane pereation have not been reported yet. In this work, toluene or heptane solutions of quinoline and 2-ethylnaphthalene were used as odel feed oils to study the effects of the type of feed solution on the separation of nitrogen heterocyclic copounds by the (O/W)/O eulsion liquid ebrane technique. Coal and petroleu derived oils are generally rich in aroatic and paraffin coponents, and the feed solutions of toluene and heptane were prepared to siulate these oils. First the liquid liquid equilibriu was easured with the odel feed oils and the eulsion liquid ebrane separations were carried out under various conditions. 2. Experiental 2.1 Liquid-liquid equilibriu The experiental conditions for the batch liquid liquid equilibriu studies are shown in Table 1. Two types of odel ixtures were used as the feed organic phases. uinoline and 2-ethylnaphthalene were selected as representative dicyclic copounds of nitrogen heterocyclic and hoocyclic copounds, which were reported to be contained in coal tar adsorption oil [6] and catalytically cracked oil [4]. The ixture of quinoline and 2-ethylnaphthalene was diluted with toluene, Feed 1, or heptane, Feed 2, to represent the feed oil of an aroatic or paraffin rich fraction. Deionized water or an aqueous solution of 1,3-butandiol was used as the aqueous phase. 1,3-Butandiol was selected to enhance the solubility of the organic copounds, especially the nitrogen heterocyclic copounds [8, 1]. All the cheicals used were of analytical grade. The specified aounts of feed organic and aqueous phases, R and E, were brought into contact in an Erleneyer flask in a constant teperature shaker at 298 K. After equilibriu, the organic and aqueous phases, R 1 and E 1, were split using a separation funnel. The obtained phases were analyzed using a coercial gas chroatograph with a flae ionization detector, (GC-21, Shiadzu Corp.) to deterine the copositions in both phases. 2.2 Eulsion liquid ebrane pereation The principal experiental conditions of the (O/W)/O eulsion liquid ebrane pereation are suarized in Table 2. The toluene or heptane solutions of quinoline and 2-ethylnaphthalene was used as the feed oils for Feed 1 or Feed 2, the sae as was used in the liquid-liquid equilibriu easureents. Aqueous ebrane solutions were prepared fro deionized water, plus saponin as an eulsifying agent, and 1,3-butandiol as an additive to enhance the pereation rate. All the cheicals used were also of analytical grade. The initial ass fraction of the additive in the ebrane solution was fixed as.25. The feed oils and the aqueous ebrane solutions were eulsified by the coercially available hoogenizer (MULTIDISPERSER PB-95, SMT Co. Ltd.). The prepared eulsions were contacted with the solvents in a stirred vessel, ade of Pyrex glass equipped with a six-flat-blade turbine type ipeller and four baffles, the sae apparatus as used in the previous study [11]. Toluene or heptane was used as the solvents for the feed oils of Feed 1 or Feed 2, respectively. The tie to start the stirring in the vessel was defined as t = h. After a specified contacting tie, the extract was analyzed by the sae gas chroatograph as used in the liquid-liquid equilibriu easureent, in order to deterine the coposition in the extract. The deulsification of the obtained eulsion was not carried out and the coposition in the raffinate was calculated by the ass balance equation based on the analysis of the extract. A tracer to detect ebrane instability was not used in these - 96 -
easureents. According to our previous studies [11, 12], the effects of ebrane instabilities due to ebrane breakage and echanical entrainent rates were very sall relative to the pereation rate. Thus the effects of the ebrane instabilities were considered to be negligible in this study as well. Table 1. Experiental conditions of liquid-liquid equilibriu Organic phase Feed 1 (, 2MN and T ixture) (z,=.8, z2mn,=.25, zt,=.67) Feed 2 (, 2MN and Hp ixture) (z,=.8, z2mn,=.25, zhp,=.67) Aqueous phase Aqueous solution of BuD (CBuD,= or.25) Mass of organic phase, R [kg] 5 1-3 Mass ratio of organic and aqueous phases, R/E [ ] 1 Teperature [K] 298 Table 2. Experiental conditions of eulsion liquid ebrane pereation Feed Feed 1 (, 2MN and T ixture) (x,=.8, x2mn,=.25, xt,=.67) Feed 2 (, 2MN and Hp ixture) (x,=.8, x2mn,=.25, xhp,=.67) Solvent Hp (Feed 1) or T (Feed 2) Mebrane liquid Aqueous solution of saponin and BuD (CS=.3, CBuD,= or.25) Stirring velocity in pereation [h -1 ] 18 Total volue of stirring vessel [ 3 ] 4. 1-4 Volue fraction of feed oil in O/W eulsion [ ].5 Volue fraction of eulsion in total solution [ ].25 Teperature [K] 298 Operation tie, t [hr].44 3.1 Liquid-liquid equilibriu 3. Results and Discussion The distribution coefficient, i, and separation selectivity, β,2mn, was defined as, C i i, zi C C2MN β (1), (2),2MN z z2mn 2MN where C i and z i represent the ass fractions of coponent i in the aqueous and organic phases, respectively. The obtained i and β,2mn values are listed in Table 3. The i s for quinoline,, were larger than those for 2-ethylnaphthalene, 2MN, in all cases. The and 2MN values for Feed 1 were saller than those for Feed 2, respectively. The higher polarity feed oil, i.e. the Feed 1 toluene solution, should show lower distribution coefficients than that of the lower polarity solution, i.e. the Feed 2 heptane solution. In the case of Feed 1, toluene being a ore polar diluent should dissolve in the aqueous phase and the other polar coponents, quinolone and ethylnaphthalene, will have liited solubility. In the case of Feed 2, the polar coponents of quinolone and ethylnaphthalene should preferentially dissolve in the aqueous phase to ake and 2MN larger. This trend was confired in the previous studies [6, 9, 11]. The addition of 1,3-butandiol ade i higher, siilar to the previous study [1]. The separation selectivity, β,2mn, was higher with deionized water than that for the aqueous solution of 1,3-butandiol. The addition of 1,3-butandiol had a large influence on the distribution of 2-ethylnaphthalene, thus reducing the selectivity. The β,2mn s values for Feed 1 were higher than those for Feed 2 regardless of the additive. - 97 -
Table 3. Distribution coefficients and separation selectivity C BuD,= C BuD,=.25 Feed 1 Feed 2 Feed 1 Feed 2 8.9 1-3 9.5 1-3 4.2 1-2 5.2 1-2 2MN 7.5 1-5 9.6 1-5 1.3 1-3 2. 1-3 β,2mn 118 99 34 26 3.2 Eulsion liquid ebrane pereation For all cases, the (O/W)/O eulsion liquid ebrane pereations could be stably operated. No phase inversion was observed and the phase separation of the eulsion and extract was readily achieved. The yield of coponent i, Y i, overall pereation coefficient, P x,i, and separation selectivity of quinoline to 2-ethylnaphthalene, β,2mn, were defined in the following equations, yi E Yi, x, R i i,e Ni Px, i a xi y i V (3), (4) T i,r β,2mn d d E y E y 2MN 2MN,E Px,2MN dt dt x x 2MN,E,R y 2MN,R y 2MN P x, (5) For the derivation of Eq. (4), the effects of echanical entrainent and ebrane breakage rates caused by ebrane instabilities were considered so sall as to be negligible relative to those of the pereation rates [11]. 1-1 1 1-2 1-1 yi [-] 1-3 CBuD 2MN Feed1.25 Feed2.25 1-4.1.2.3.4.5 8 xi [-] 1-2 1-3.1.2.3.4.5 Figure 1. Tie courses of y i and x i. Figure 1 shows the tie courses of the ass fractions in the raffinate and extract phases, x i and y i, respectively. In all cases, the ass fractions of quinoline in the extract phase, y s, were higher than those of 2-ethylnaphthalene, y 2MN s, though x 2MN, was uch higher than x,. The value was larger than 2MN and quinoline could preferably pereate through the eulsion liquid ebrane relative to - 98 -
2-ethylnaphthalene. After around t =.2 h, the concentration difference of quinoline in the raffinate and extract phase, x - y, was so sall and x and y approxiately attained plateau values. On the other hand, the concentration difference of 2-ethylnaphthalene between the raffinate and extract phases, x 2MN - y 2MN, was still large at t =.2 h, and 2-ethylnaphthalene continued to pereate through the eulsion liquid ebrane. The addition of 1,3-butandiol to the ebrane solution enhanced the pereations of both quinoline and 2-ethylnaphthalene through the eulsion liquid ebrane. This was because the distribution coefficients of the pereates increased by addition of 1,3-butandiol in the ebrane solution, as entioned above. The pereation of quinoline and 2-ethylnaphthalene were affected by the type of the feed solution, and the pereation rate was a little larger with Feed 1, than that with Feed 2. Figure 2 shows the tie courses of yield of coponent i, Y i. Y was uch higher than Y 2MN and the eulsion liquid ebrane could effectively separate these two coponents. The yields were higher with 1,3-butandiol than those without 1,3-butandiol. The pereation rate of quinoline was large in the initial stage relative to the rate of 2-ethylnaphthalene. In the range of t >.2 h, the concentration difference of quinoline becae sall thus reducing the pereation rate and attained plateau values for each condition. In the case of 2-ethylnaphthalene, large concentration differences were aintained during the pereation runs, and the yields linearly increased in all cases. Y attained a values greater than.9 for Feed 1, though i was saller with the toluene feed oil solution than that with the heptane solution. 1.8 1 5 1 2 Yi [-].6.4.2 Feed1 Feed2 CBuD 2MN.25.25.1.2.3.4.5 Pi a [kg h 1 3 ] 1 4 1 3 1 2.1.2.3.4.5 8 β,2mn [-] 1 1 CBuD key Feed1.25 Feed2.25 1.1.2.3.4.5 Figure 2. Tie courses of Y i. Figure 3. Tie courses of P x,i a. Figure 4. Tie courses of β,2mn. (keys are sae as in Figure 2) Figure 3 shows the tie courses of the overall voluetric pereation coefficient of coponent i, P x,i a, estiated by Eq. (4). The P x,i a value for quinoline was larger than that for 2-ethylnaphthalene. The addition of 1,3-butandiol to the ebrane solution enhanced both quinoline and 2-ethylnaphthalene P x,i a values. The overall pereation coefficient, P x,i, for a one diensional onolayer liquid ebrane [11] can be expressed as; P x, i ρm Di i,r (6) δ P x,i should be proportional to the distribution coefficient, however P x,i a was larger for Feed 1 than that for Feed 2 though i was saller for Feed 1. This ight occur because of the conditions of (O/W)/O - 99 -
dispersion. The properties of the O/W eulsion should be affected by the properties of the inner oil, and accordingly the (O/W)/O dispersion should change the specific surface area in the vessel, a. The easureent of the (O/W)/O dispersion is necessary to clarify the pereation echanis. Figure 4 shows the tie courses of the separation selectivities of quinoline relative to 2-ethylnaphthalene, β,2mn, estiated by Eq. (5). The β,2mn value was reduced by the addition of 1,3-butandiol. The addition of 1,3-butandiol enhanced not only quinoline pereation but also 2-ethylnaphthalene pereation. The selectivity was larger with Feed 1 than with Feed 2. This trend confored with the results of the distribution coefficients easured in the liquid-liquid equilibriu, and the effects of the aterial syste also influenced the selectivity. 4. Conclusion The separation of nitrogen heterocyclic copound fro a odel oil by the eulsion liquid ebrane technique was studied. The distribution coefficient of quinoline was larger than that of 2-ethylnaphthalene. The coefficients were larger with the feed solution rich in heptane than those with the solution rich in toluene. 1,3-Butandiol, used as an additive, enhanced the distribution coefficients of both coponents but reduced the separation selectivity. In the case of the eulsion liquid ebrane separation, quinoline preferably pereated through the eulsion liquid ebrane, and could be separated relative to 2-ethylnaphthalene in the odel feed oils. The addition of 1,3-butandiol enhanced the pereation rates of both coponents. The rates were larger with the toluene feed oil than those with the heptane solution probably because the specific surface area in the (O/W)/O dispersion ight have been larger with the toluene solutions. The separation selectivity of quinoline relative to 2-ethylnaphthalene decreased with the addition of 1,3-butandiol, and it was larger with the heptane feed solution copared to that for the toluene feed solution, which followed the trend of the distribution coefficients. For clarification of the pereation echanis it is necessary to study not only the pereation rates of the coponents, but also the (O/W)/O dispersion. Noenclature a = specific surface area [ 1 ] C i = ass fraction of coponent i in aqueous or ebrane phase [ ] D i = diffusivity of coponent i in ebrane liquid [ 2 h 1 ] E = ass of extract phase [kg] i = distribution coefficient of coponent i [ ] N = pereation rate [kgh 1 ] P x,i = overall pereation coefficient of coponent i [kgh 1 2 ] R = ass of raffinate phase [kg] V T = total volue of vessel [ 3 ] x i = ass fraction of coponent i in raffinate phase [ ] Y i = yield of coponent i [ ] y i = ass fraction of coponent i in extract phase [ ] z i = ass fraction of coponent i in organic phase [ ] β,2mn = separation selectivity of relative to 2MN [ ] δ = effective thickness of liquid ebrane [] ρ ML = density of the liquid ebrane [kg 3 ] <Subscript> = at initial state 2MN = 2-ethylnaphthalene Hp = heptane i = coponent i = quinoline T = toluene References 1) B.D. Batts, A.Z. Fathoni, Energy Fuels, 5, 2-11 (1991). 2) D. Bohlann, E. Dohler, H. Lier, J. Che. Technol., 33, 358-364 (1981). 3) M.A. Wecher, D.R. Hredy, Fuel Sci. Technol. Int., 7, 423- (1989). - 1 -
4) J. i, Y. Yan, Y. Su, F. u, Y. Dai, Energy Fuels, 12, 788-791 (1998). 5) M. Matsuoto, Y. Inooto, K. Kondo, J. Mebr. Sci., 246, 77-81 (25). 6) C. Sali, J. Saito, R. Egashira, J. Jpn. Pet. Inst., 48, 6-66 (25). 7) C. Sali, R. Egashira, J. Jpn. Pet. Inst., 49, 326-334 (26). 8) R. Egashira, K. Watanabe, Solvent Extr. Res. Dev., Jpn., 14, 63-69 (27). 9) R. Egashira, N. Hara, M. Nagai, 7th World Congress of Cheical Engineering C23-4 (25). 1) H. Habaki, B. Dejin, R. Egashira, ICSST8 FP-7 (28). 11) Y. Shiada, D. Bi, H. Habaki, R. Egashira, J. Che. Eng. Jpn, 46, 376-382 (213). 12) H. Habaki, M. Haruna, R. Egashira, J. Jpn. Pet. Inst., 56, 34-311 (213). - 11 -