An Improved Extractor for Dual-Flow Extraction. Hideyuki NISHIZAWA, Yukako WATANABE, Sakiko OKIMURA and

Transcription

1 ANALYTICAL SCIENCES JUNE 1989, VOL An Improved Extractor for Dual-Flow Extraction Hideyuki NISHIZAWA, Yukako WATANABE, Sakiko OKIMURA and Kyoritsu College of Pharmacy, Shibakoen, Minato, Tokyo 105, Japan Yoshihiro ABE An improved apparatus for dual-flow countercurrent extraction was constructed. Further, a simple method to evaluate the extractor was developed, by which the effective plate number could be calculated by a couple of measurements of fractional recoveries. To find the optimum operation condition, an apparatus with a shorter column (10 cm in length) was prepared, the efficiency of which was evaluated as being about one plate per cm of the column length by the method. It was also shown that the distribution constant of a compound could be calculated from the fractional recoveries measured by dual-flow countercurrent extraction. The distribution constants, thus obtained, agreed with values determined independently by liquid-liquid chromatography, which settled theoretical correlations between chromatography and extraction. Keywords Dual-flow countercurrent extraction, extraction extractor, liquid particle chromatography, liquid particle We previously reported a new-type liquid-liquid chromatography and named it liquid particle chromatography (LPC)', the principle of which involved the application of that used in countercurrent extraction, reported by Schutze et al.2 Their apparatus was constructed by inserting a rod into a pipe; its rotation (driven by a rotary motor) generated liquid particles of the dispersed phase in the continuous phase. Though the extractor used no more than 10 effective extraction plates, acids in the petroleum oil could be successfully separated. The efficiency of the column of LPC (60 cm in length) was not sufficient for chromatographic separation, but the results prompted us to improve the apparatus for dual-flow countercurrent extraction with a greater number of plates. Bartels3 and Rometsch4 had shown that dual-flow countercurrent extraction is a very efficient technique for the separation of mixtures. However, this method has not been widely used, because there has not existed a simple and efficient apparatus for laboratory use. We presently wish to report on an improvement in the dual-flow countercurrent extractor, working principle of which is similar to that of Schutze's. The extractors could be operated continuously with a large number of plates (approximately 100) by selecting the sizes of the inner and the outer pipes, using magnets for rotation and adopting pumps for accurate flow rates that are obtainable today. To determine the optimum operating condition, it was necessary to measure the efficiency of the extractor. Although there are some experimental methods available to evaluate a countercurrent extractor6, by which the plate number can be calculated graphically, they are not practical for measurements of the efficiency of an apparatus with high plate numbers (>10). Based on the theoretical investigatious of Bartels3 we developed a simple method to estimate the efficiency of the extractor. That is, a single calculation by an equation using a couple of values of fractional recoveries provides the effective plate number of the extractor without using graphical calculations. There may also be experimental difficulties in measuring the low concentration of the recovered solution from an extractor with a large number of plates. We decided to calculate the plate number per unit length of the column of the extractor using a shortcolumn apparatus with only a few plates. Here we describe: (1) a simple method to evaluate a countercurrent extractor with theoretical investigations and (2) a method to measure the effective number of plates per unit length of the improved apparatus with a short column apparatus. The improved apparatus with a large number of plates and its applications are detailed in the following report. We would like to name this extraction technique liquid particle extraction (LPX). Theory Method for the calculation of the number of theoretical

2 340 ANALYTICAL SCIENCES JUNE 1989, VOL. 5 Assuming that a solute is fed from the top of the extractor as an queous solution, which is the case when m=1 in Fig. 1, Eq. (4) reduces to E-1 Rhf - E n+1 _ 1 (5) where Rhf indicates the fractional recovery of the aqueous phase, that is, the ratio of the quantity (per unit time) in the aqueous extract and that in the initial feeding solution. Similarly, if a solute is fed from the bottom of the extractor as an organic solution, the fractional recovery in the organic phase (Rsf) is given by Eq. (3) by substituting 1 for n: Fig. 1 Ideal diagram of a dual-flow countercurrent extractor. U This diagram is slightly modified from the reported one.3 H, aqueous phase; S, organic phase; f, sample solution; w, organic extract; b, aqueous extract. plates from the fractional recoveries According to Bartels et a1.3 when a solute is introduced continuously at the middle point of a dualflow countercurrent extractor (f in Fig. 1), the ratio of the amount of solute extracted to the organic phase (w) to the water phase (b) in its stationary state can be described as w (En- 1)Em b = Em_ 1 (1) (w/b=n/m,when E= 1) where n and m are the number of theoretical plates of the top (m) and the bottom (n) of the extractor, respectively (Fig. 1). E is the extraction factor, and is defined as E = SD / H, (2) where D is the distribution constant of the solute; S and H are the flow rates' of the organic phase (S) and the aqueous phase (H), respectively. Using Eq. (1), the fraction in the organic phase (R,) is given by Em+n - Em Rs=w/(w+b)= E m+n_1. (3) The fraction in the aqueous phase (Rh) is Em -1 Rh = 1- RS = E m+n _ 1 (R, = n/(m+n), Rh = m/(m+n), when E= 1). (4) Rsf = Em+' - Em Em(E-1) Em+1 _ 1 = Em+i -1 (6) In this case, the number of theoretical plates is considered to be m. In an apparatus with the number of theoretical plates equal to n, Eq. (6) reduces to Rsf - En(E-1) E n+t -1 (6') From Eqs. (5) and (6'), the fractional recovery in the aqueous phase (1- Rhf) and that in the organic phase (1- Rsf) are and 1 - Rhf = E(En -1) E~1-1 En Rsf = E n+l _ 1. (8) The following equation is obtained by arranging Eqs. (7) and (8): 1-Rhf E_ 1 -. R sf. (9) While the values of Rhf and Rsf can be obtained experimentally, the experimental value of E (the extraction factor) is obtained using Eq. (9). Equations (5) and (6') can be combined to give Rsf / Rhf = En. (10) The number of the theoretical plates, n, is obtained by Eq. (11), which is derived from Eqs. (9) and (10): n - l log[(rsf) /(Rhf)] 11 og [(1-Rhf)/ (1- Rsf)] ( ) Thus, it is possible to obtain the number of effective plates (n) simply by measuring the fractional recoveries (7)

3 ANALYTICAL SCIENCES JUNE 1989, VOL (Rsf and Rhf) of a sample.8 Since E is known from Eq. (9), the distribution constant of the sample (D) is also obtained experimentally using D = (1- Rhf)H/ (1- Rsf)S. (12) Based on these considerations, a column of a "half structure" was used for measurements of the fractional recoveries, Rsf and Rhf, to obtain the effective plate number, n. However, in the case with an extractor with a high effective plate number, Rsf and Rhf become too small to calculate plate numbers accurately. An extractor with a short column was, hence, constructed (Fig. 2) and the plate number per unit length was determined. Experimental Apparatus Improved apparatus for dual-flow countercurrent extracton, liquid particle extraction (LPX) apparatus. The improved extractor with high plate numbers is detailed in the following reports Apparatus used for evaluation of the extractor. An evaluation of the extractor was carried out using the apparatus illustratd in Fig. 2. The structure of the apparatus is similar to that described in the following paper but has a shorter inner pipe (10 cm in length), and has no sample inlet at the middle point. It is just the "half structure" of LPX apparatus. A detailed description is given in the legend of the figure. The absorptiometry of the extracts was performed with a U-3200 spectrophotometer (Hitachi, Tokyo) Measurement of distribution constant. Using the previously reported liquid-liquid chromatography' (LPC), the distribution constant (D) was calculated using D = (Vr- Vo) / Vs, (13) where Vr is the retention volume; Vs and Vo are the volumes of the stationary and the mobile phases in the column, respectively. After LPC separation, both phases were taken out from the column and their volumes measured. Table 1 shows the distribution constants between water and 1-butanol of salicin, phenyl-a- and phenyl-/3- D-glucopyranoside, p-nitrophenyl-/3-d-glucopyranoside and esculin determined by LPC with a column of 18 mm i.d.x 16 mm o.d.x60 cm in length. Analytical conditions are given in the table. Chemicals Deionized water (>2 MSS) was used in this work. 1- Butanol was distilled with a small amount of sodium borohydride. Salicin, esculin, phenyl-$-d-glucopyranoside, and p-nitrophenyl-/3-d-glucopyranoside were purchased from Tokyo Kasei (Tokyo). Phenyl-a-D-glucopyranoside was obtained from Sigma Chemical Co. (St. Louis). After shaking a mixture of water (21) and 1-butanol (21) in a 41 separatory funnel, two phases were separated at 25 C and the upper and the lower phases were used as a 1-butanol phase and an aqueous phase, Table 1 Distribution constant of glucosides measured by LPCa Fig. 2 A short column dual-flow extractor. Column: a glass pipe (16 mm o.d.x 10 cm, in which has a magnet (N), and air tight plugs on both ends) is inserted into another glass pipe (19 mm i.d.x33 cm), and sustained by two plastic plugs (15 cm in length, about 11 cm of which is inserted into the column). There is a gap (3 mm) between the plug (16 mm o.d.) and the glass pipe, where the liquid particles settle and a separation of the two phases occurs. B, boundary of two phases; M, motor (RW-7 with 2GK5K gear box, Oriental Motor, Tokyo); N,S,N, magnets which are set on the edge of a hexagonal plate, whose rotation by motor (M) causes the revolution of the inner pipe; p1, handmade polypropylene plugs; DH, a difference in height between the outlet of the organic extract (w) and the aqueous one (b), which is set to keep the boundary (B) at the position lower than the rotating pipe; P, and P2, pumps (880-PU, JASCO, Tokyo). All tubings were of polytetrafluoroethylene. a. Conditions of LPC: column, 18 mm i.d.x 16 mm o.d.x 60 cm; stationary phase, 1-butanol (water saturated, Vs=35 ml); mobile phase, water (1-butanol saturated, V =8 ml); eluent, l-butanol saturated water (0.45 ml/min); at 550 min', 25 C. b. The standard deviations were calculated by eight times of experiments.

4 342 ANALYTICAL SCIENCES JUNE 1989, VOL. 5 respectively. Measurement of the efficiency with a short column extractor Using the apparatus as drawn in Fig. 2, the optimum conditions were surveyed using three glucosides as samples. The ratio of the flow rates, S/ H, was adjusted so that SD/ H=.1 in accordance with the value of D, which was obtained by LPC (Table 1). The value of the fractional recovery (Rhf) changes greatest with a change of n under the condition E(=SD/ H) =1. Similarly, Rsf varies the greatest around E 1. Initially, a sample was dissolved in the aqueous phase (initial concentration: Co mg/ ml) and fed from the top of the column; a fresh organic phase was fed from the bottom. The concentration of the aqueous extract (Cf) was measured after the system became stationary (40 min, monitored by flow cell). The fractional recovery of Rhf was Cf/ Co. Then, without changing the operating conditions, the sample was fed as a 1-butanol solution from the bottom and the fresh aqueous phase was fed from the top. The fractional recovery of Rsf was also determined at the stationary state. The number of theoretical plates (n) was calculated using Eq. (11) with the Rhf and Rsf, thus obtained (cf. Fig. 3). The initial concentration and the concentration after extraction were measured by an absorptiometric method at the wavelength given in Table 1. Results and Discussion Factors affecting the efficiency of the column As shown in Fig. 3, an increase in the total flow rates of solvents, which causes an increase in the number of liquid particles per unit length in the column and may result in an increase in the interface, as well as an increase in the rotation of the inner pipe, improved the efficiency. However, at a total flow rate of >3.5 ml/ Fig. 3 Evaluation of the column; the number of theoretical plates calculated under different conditions using three glucosides as samples. (a) salicin (D=0.60); (b) phenyl-q-d-glucopyranoside (D=1.17); (c) p-nitrophenyl- $-D-glucopyranoside (D=2.03). Flow rates of 1-butanol (S ml/ min) and the water phase (H ml/min) are indicated as S/H. Table 2 Effective plate number (n) per unit length measured by the short column extractor a. Extraction factor (E) and distribution constant (D) were calculated from the fractional recoveries, Rhf and Rsf using the following equations: E (1-Rhf)/(1-Rsf), D=(1-Rhf)XH/(1-Rsf)XS. b. These values were determined by LPC (cf. Table 1). c. 25 C, 500 min'. Rhf and Rsf: fractional recovery in the aqueous and the organic phase, respectively. S: flow rate of the organic phase; H: flow rate of the aqueous phase. Flow rates were measured by weighing the liquids per unit time comparing with the weight of the liquids pipetted by a transfer pipet.

5 ANALYTICAL SCIENCES JUNE 1989, VOL min and! or at a rotation of >600 min', it became difficult to stabilize the phase ratio in the column. Thus, a total flow rate of <_3.0 ml/ min and a rotation of min' were chosen as the standard condition for further experiments. Efficiency of the column The number of effective plates (n) of the short column (10 cm) was about 10, obtained with three different samples (Table 2). The distribution constant of each sample (Da) was calculated by using Eq. (12). The distribution constant obtained by chromatography (Db) is also given in Table 2. Small differences between Da and Db are considered as being experimental errors. Since the volume of the stationary phase in LPC became a little larger because of the dead volume of the column, smaller values for the distribution constants were obtained. In conclusion, we constructed an improved apparatus for dual-flow countercurrent extraction, and developed a simple method to evaluate the efficiency experimentally. The experimental procedures of the method are: (1) to measure the fractional recovery of the aqueous solution after extraction with the organic phase, (2) without changing any operation conditions, to measure the fractional recovery of the organic sample solution after extraction with the aqueous phase, (3) to calculate the plate number from the fractional recoveries thus obtained. There is no change in the state of the column during the above-mentioned operations. The effective plate number of the extractor increases with an increase in the revolution of the inner pipe and the total flow rates of both phases. However, it becomes difficult to maintain the stability of the extractor under a high number of revolutions (>600 min') and/or high flow rates (>3.5 ml/ min). The effective plate height is considered as being about 1 plate/ cm under the usual operating conditions ( min' and <_3.0 ml/ min). The coincidence of the two values of the distribution constants, which were obtained independently from LPC and LPX, showed the correctness of the abovementioned theoretical investigations. The extractive separation of a two-component mixture by another type of apparatus which has a column of 100 cm in length shall be demonstrated in a following report. References and Notes 1. H. Nishizawa, Y. Watanabe and Y. Abe, Anal. Sci, 3, 97 (1987). 2. H. G. Schutze, W. A. Quebedeaux and H. L. Lochte, Ind. Eng. Chem., Anal. Ed, 10, 675 (1938). 3. C. R. Bartels and G. Kleiman, Chem. Eng. Progr., 45, 589 (1949). 4. R. Rometsch, Hely. Chim. Acta, 33,184 (1950). 5. H. Nishizawa, Y. Watanabe, S. Okimura and Y. Abe, Anal. Sci., 5, 345 (1989). 6. R. E. Kirk and D. F. Othmer, ed., "Encyclopedia of Chemical Technology", Supplement, pp , Interscience Encyclopedia Inc., New York (1957). 7. The unit of weight per unit time for the flow rate was adopted in the reference 1. However, we used volume per unit time, because it is more practical and there is no need to change equations. 8. The plate number (n) can be calculated also from Eqs. (2) and (10), if D is known. But, our method can provide both the values of E (from Eq. (9)) and n (from Eq. (11) through a measurement of R,f and Rhf without knowing D. (Received September 8, 1988) (Accepted April 4, 1989)