Heavy Metal Adsorption to a Chelating Resin in a Binary Solid Fluidized Bed

Transcription

1 Heavy Metal Adsorption to a Chelating Resin in a Binary Solid Fluidized Bed By Jianzhong Yang and Albert Renken* The addition of inert particles of lighter density and saller diaeter increases considerably the ass transfer coefficient in coparison to that of ono-coponent active particles at the sae liquid velocity. This effect was applied to eliination of copper ions by adsorption on a chelating resin. An intensification of the fil ass transfer coefficient in binary syste leads to a 15 % increase of the usable adsorbent efficiency. 1 Introduction Wastewater generated by electroplating contains a substantial aount of etal ions and the recovery of these etal ions is becoing of increasing iportance for the protection of the environent. The developent of efficient separation processes is leading to an increased interest in the engineering aspects. Many techniques can be used for pre-concentration of trace aounts of etal ions, and aong these, adsorption of etal ions with chelating resins is an effective and powerful separation technology. Adsorption processes are traditionally carried out in fixed beds [1±5] due to the high concentration of solids and the obtainable unifor residence tie. However since the wastewater to be treated often contains solid ipurities leading to a plugging of the fixed bed, the liquid ust be clear to avoid colun blocking. Another proble concerns shrinking and swelling of the resin during the adsorption/desorption processes resulting in poor liquid distribution caused by channeling and foration of dead zones [6±7]. Therefore any experiental studies have been conducted in fluidized beds, which allow treatent of turbid liquids while avoiding the channeling probles [8±14]. For a proper design of the fluidized bed adsorber, an adequate atheatical odel is required which takes into account the hydrodynaics and adsorption kinetics. Menoud et al [15] studied the adsorption kinetics of a chelating resin (Chelaine JPS-Chiie, Neuchâtel, Switzerland [16]) and found that the overall adsorption rate is liited by external ass transfer. Therefore intensification of the ass transfer should increase the perforance of the fluidized bed adsorber for such an adsorption syste. According to our previous work [17], the addition of inert particles of heavier density and saller diaeter could be used to increase the fil ass transfer coefficient. In this work, this technique is applied to the aforeentioned syste. A coparison is ade of copper ion reoval in onocoponent and binary solid systes, thereby providing the inforation to understand the changing tendency of adsorption process at different ass transfer coefficients and to evaluate the influence extent of sall/heavier inert particles on adsorption perforance. ± [*] Dr. J. Yang, Prof. Dr. A. Renken, Laboratory of Cheical Reaction Engineering, Swiss Federal Institute of Technology, CH-1015 Lausanne, Switzerland. 2 Adsorption Model in a Fluidized Bed The governing equations for the adsorption of copper ion in a fluidized bed can be derived on the base of the research results of the equilibriu isother and hydrodynaics of liquid and solid particles [15]. In the developent of the atheatical odel, several siplifying assuptions are ade as follows : l Adsorption rate is liited by a liquid fil resistance and surface reaction. l Surface reaction between adsorbate and adsorption site is described by a reversible second order reaction. Adsorption equilibriu behavior is represented by the Languir isother. l Adsorbent particles are spherical with a ean diaeter d p. These particles are relatively iobile and distributed uniforly in the colun. l Axial dispersion of liquid is negligible. A aterial balance of the cation in the obile phase for such a syste gives rise to c " t ˆ z R i (1) The adsorption rate fro the bulk solution to the surface of particles is : R i ˆ k L a c c i (2) Where a is the surface area of the particles per reactor unit volue; c is the bulk concentration; k L, the fil ass transfer coefficient; the teporal point concentration of solute c i can be calculated by a ass balance at the liquid-solid c " i t ˆ k L a c c i 1 " q p t The reaction rate in the solid phase is related to the interface concentration c i and described by the Languir t ˆ k f c i q q K d q (4) K d ˆ kr k f (5) ± 1) List of sybols at the end of the paper. Che. Eng. Technol. 23 (2000) 11, Ó WILEY-VCH Verlag GbH, D Weinhei, /00/ $ $ /0 1007

2 Where the paraeters k f and k r are the rate constants of forward and reverse reaction, respectively; K d, the dissociation constant; q, axiu capacity of resin. The initial and boundary conditions would be c=0 q=0 att=0 (6) c=c 0 at z = 0, t > 0 c ˆ 0 at z = h, t > 0 z The ass balance for the adsorbate is a partial differential equation. This equation can be reduced to an ordinary differential equation using the finite difference ethod. It is assued that a fluidized bed of total volue V can be divided into N equal size cells in series and that the liquid flowing upwards passes through the bed. Furtherore, in order to copare the odel with different experiental results, it is convenient to use the following diensionless variables: f ˆ c c 0 (9) ˆ q q 0 (10) ˆ t ˆ Qc 0 t q 0 (11) ˆ q 0 Qc 0 (12) The variables f and n are diensionless concentrations in liquid and solid phases respectively; h, the diensionless tie; s, the tie required for saturating all the resin in the reactor; c 0, the concentration of cation at the inlet of the colun; q 0, the dynaic capacity of the resin which is in equilibriu with c 0 ;, the ass of dry resin; Q, the flow rate of liquid. After transforation with these diensionless variables, the ass balance for the j th cell becoes: df j d ˆ N f j 1 f j DaI kl f j f i;j (13) df i;j d ˆ DaI kl f j f i;j d j d (14) d j d ˆ DaI kf f i;j j 1 j (15) While developing the differential equations, several diensionless groups are introduced. Soe explications for these paraeters are given as follows: l DaI kl is a Daköhler nuber, which is defined by: kla DaI kl ˆ (16) " r DaI kl represents the nuber of fil ass transfer units which is the ratio between the hydraulic residence tie and a characteristic tie of adsorption [18]. For a fixed residence tie, if the value DaI kl is larger, the adsorption will be carried out at a higher rate and the concentration of solute at the exit of the colun is lower. l DaI kf is another Daköhler I nuber, which is given by : DaI kf ˆ kf q (17) Q This is an expression for the nuber of transfer units due to the intrinsic adsorption kinetics. Two transfer units DaI kl and DaI kf can be cobined as the resistances in series to yield the overall nuber of transfer units [19]: 1 DaI 1 1 (18) DaI kl DaI kf l w is a residence tie nuber required to saturate all the resin in the reactor: ˆ ˆ q 0 " r c 0 V l n is an equilibriu factor which is defined by: (19) ˆ q ˆ Kd 1 (20) q 0 c 0 The equilibriu factor depends on the concentration of solution c 0 at the inlet of the colun for a continuous adsorption process. If n!, the concentration of solution c 0 is in the linear region of the Languir isother curve; if n! 1 the dynaic capacity q 0 approxiates to the axiu capacity of the resin, the c 0 and q 0 are in the plateau region of the Languir isother curve. The response at the exit of the colun is given by the concentration f N and Eqs. (13), (14) and (15) ust be solved siultaneously subject to the boundary conditions (f j±1 =1 when j = 1 and h > 0) and the initial conditions (f j, f i,j, and q j are zero when h =0(1 j N)). This two-resistance odel can be used as a sole resistanceliiting odel by giving a high value for another resistance coefficient. When the reaction rate is instantaneous, only the fil ass transfer resistance controls the adsorption rate, the teporal point concentration c i in Eq. (2) can be directly calculated with the Languir equilibriu: c i ˆ qk d (21) q q Using this theoretical fraework and a SiuSolv progra (Dow Cheical Copany) [20], the contributions of ass transfer processes and kinetics ay be estiated and copared with the experiental results. 3 Experiental 3.1 Materials The adsorbent used in this study is a resin, Chelaine Standard Ò (JPS-Chiie, Neuchâtel, Switzerland) [16]. The 1008 Ó WILEY-VCH Verlag GbH, D Weinhei, /00/ $ /0 Che. Eng. Technol. 23 (2000) 11

3 adsorbate is a copper ion solution prepared with pentahydrate cupric sulfate (Fluka-AG, Buchs, Switzerland). The inert particles are glass beads (Microbeads, Brugg, Switzerland). The ean particle sizes and other properties for glass beads and resins are suarized in Tab Experiental Protocol and Methods The fluidized bed reactor previously used [15] is odified and used for onitoring the effect of inert particles on the adsorption process. A colun is coprised of five Plexiglas eleents (52 inner diaeter and 102 height) which are designed with a therostatic double jacket. During the adsorption experient, the copper ion solution (~30 pp) in the tank is circulated through a heat exchanger using a peristaltic pup, then fed into the colun. The flow rate is easured with a calibrated rotaeter. The ph is easured with an on-line ph eter and the teperature is easured with a therocouple, the results are recorded autoatically with a data acquisition syste (LabVIEW for windows, version 3.0, National Instruents Corporation). To deterinate the concentration of the copper ion, the effluent at the outlet of the colun is continuously withdrawn by a peristaltic pup. In parallel, an EDTA solution (0.1 M) is supplied at the sae flow rate, thus coplex solution of copper ion-edta produces a easurable color. The optical density of the coplex solution of copper ion-edta is easured with an on-line UV/VIS spectrophotoeter (Diode array spectrophotoeter 8452A, Hewlett Packard) at 732 n [20]. The results are recorded by a coputer and converted to concentration according to the Labert-Beer law [21]. The color of the resin changes fro yellow to green and then becoes Table 2. Operation conditions. dark blue during the adsorption process. The resin at the botto of the colun is loaded initially and this can be observed clearly fro the color change. The resin particles shrink due to the loading of copper ion, therefore the height of the suspension layer decreases during the adsorption process. The ean height is taken into account for calculating the reactor volue. The diaeter of loaded resin (340 l) is used to deterinate the ass transfer surface area. For coparison, the operating conditions are kept identical for the two kinds of experients. The only difference is that the glass beads (30 l) are added to the colun for the experients with the binary syste. The detailed operating conditions for experients MIX3 and NOR5 are given in Tab. 2. Table1. Properties of solid particles in binary syste. 4 Results and Discussion Full Paper Paraeter Sybol Value Units Resin (Chelaine) Mean voluetric diaeter d p 414 l Particle porosity e p 0.81 ± Mass of dry resin per unit volue of swollen resin r ps kg ±3 Apparent density r ap 1064 kg ±3 Archiedes nuber Ar 46 ± Inert particles (glass beads) Mean voluetric diaeter d p 103 l Apparent density r p 2450 kg -3 Archiedes nuber Ar 15 ± Typical adsorption experiental results (points) are shown as a function of diensionless tie h in Fig.1 and Fig.2. The ph profiles (dotted curves) are also given in the sae figures. A coparison is ade firstly between the fil ass transfer and intrinsic adsorption kinetics with the two-resistance odel. By using the fil ass transfer coefficient and the intrinsic forward rate constant which are obtained with siulation, the nuber of transfer units for fil ass transfer DaI kl and the nuber of transfer units for the intrinsic adsorption kinetics DaI kf are calculated by Eq. (16) and Eq. Operating paraeters Sybol MIX3 NOR5 Units Operation teperature T C ph for preconditioning of resin ph ± Height of free settled bed h Height of resin in fluidized bed h Voidage of reactor e r ± Liquid flow rate Q s Volue of reactor V Residence tie s s Feed concentration of Cu 2+ c ol -3 Mass of dry resin kg Volue of free settled resin V Voidage of free settled resin e ± Volue of free settled glass beads V 0g ± 3 Surface area per unit volue a Che. Eng. Technol. 23 (2000) 11, Ó WILEY-VCH Verlag GbH, D Weinhei, /00/ $ /0 1009

4 Figure 1. Experiental data and odeling of noral adsorption of copper ion onto Chelaine resin. Figure 2. Experiental and odeling results on the adsorption of Cu 2+ onto Chelaine resin in the suspension of glass beads. (17), respectively. The nuber of transfer units for the adsorption of copper ion on chelating resin (experient NOR5) is found to be respectively 145 for DaI kf, and 16 for DaI kl. The overall nuber of transfer units DaI is calculated by Eq. (18), and has a value of 14. The nuber of transfer units for the intrinsic adsorption kinetics is about 9 fold higher than that for the fil ass transfer. This indicates that the surface interaction is uch faster than the fil ass transfer and the overall adsorption rate is liited ainly by the fil ass transfer for this syste. Since the fil ass transfer resistance appears to be the controlling step for the overall adsorption rate, a one paraeter odel, in which only the fil ass transfer is considered, is sufficient. The siulated curve describes the experiental data reasonably well, however, at the second breakthrough point, the two-resistance odel fits better than the one resistance odel (results not shown here) [20]. This suggests that the surface reaction becoes ore iportant when the resin approaches saturation. In other words, there is a lateral interaction between adsorbed olecules, the assuption in the Languir odel does not held true for this case. Since the volue of resin particles decreases with loading of copper ion, the adsorbed cations exert an effect on the interaction between new solute and active sites. A decrease of k f is possibly due to steric change of the resin or pore blockage by adsorbed cations. Koloini and Zuer [22] thought that after approxiately 80 % of the resin capacity is exhausted the internal resistance to the adsorption process becoes iportant and the overall rate of the ion exchange decreases continuously. This behavior has been observed by other authors [18, 23] for protein adsorption in fixed beds. Therefore the two-resistance odel is recoended for the following process siulation. For the sake of coparison of the influence of inert particles on the fil ass transfer coefficient, the forward rate constant and isother equilibriu constant are assued to be fixed (k f =2 0 ±2 3 ol ±1 s ±1, K d = ol ±3 ) [14]. The liquid fil ass transfer coefficient is a free paraeter, with a value deterined by fitting the experiental data. As typical exaples siulation results for a ono-coponent (NOR5) and a binary-ixture systes (MIX3) are shown in Fig.1 and Fig.2 respectively. The prediction of the odel (curve) for fluidized bed perforance results in a good agreeent. The values for the ass transfer coefficient obtained with the siulations are ±5 ( s ±1 ) for the ono-coponent syste and ±5 ( s ±1 ) for the binary ixture syste. The fil ass transfer coefficient is substantially higher in the binary-ixture syste and this confired once again that the sall/heavier inert particles can be applied to intensify the external ass transfer. Fro the engineering point of view, the ain requireent for rational design is an estiate of the dynaic capacity of the fluidized bed. Assuing that the adsorption colun is used to reove a trace ipurity fro a strea, a axiu perissible concentration at the outlet of the colun is required, due to the legislation for the provision of water containation. For exaple, the perissible concentration of Cu 2+ in the effluent water should be less than 0.5 pp in Switzerland and Gerany [24±25]. During the adsorption process, the effluent concentration at the outlet of the colun increases gradually. The operation should be stopped when the effluent concentration reaches the axiu perissible concentration. The corresponding tie ay be called break tie h b. The dynaic capacity of the bed depends on the break tie and can be calculated fro the following ass balance: R f Rb q ˆ q 0 1 f d " r c 0 Vdf (22) 0 0 Assuing that the quantity of etal ions in the liquid phase of the reactor is negligible copared to the adsorbed one, the usable fraction of the total capacity of adsorbents ay be expressed using the following approxiate relationship: ˆ q R b 1 f d (23) q 0 0 Defined in this way, the usable fraction of the total capacity of adsorbents in the colun can be represented by the shaded area in Fig Ó WILEY-VCH Verlag GbH, D Weinhei, /00/ $ /0 Che. Eng. Technol. 23 (2000) 11

5 Figure 3. Usable efficiency of total capacity of adsorbents in the colun. For given operating conditions, the effluent concentration at the outlet of colun changes during the adsorption process and fors an asyptotic breakthrough. The shape of the breakthrough curve is depending on the ass transfer rate as shown in Fig. 4. With increasing ass transfer coefficient, the effluent concentration at the outlet of the colun decreases and the break tie h b is postponed. Therefore the usable fraction of absorbents increases with increasing the ass transfer coefficient. Due to the higher ass transfer rate in a binary fluidized syste, the break tie increases fro h b1 to h b2, resulting in a 15% higher usable fraction of the adsorbent. Figure 5. Influence of ass transfer coefficient on usable efficiency of adsorbents in the colun. 5 Conclusions Adsorption processes of copper ions onto a chelating resin in fluidized beds are studied experientally and siulated with theoretical odels. Coparison of the nuber of transfer units for fil ass transfer and the nuber of transfer units for intrinsic adsorption kinetics indicate that the overall adsorption rate is liited ainly by the fil ass transfer resistance for this syste. Sall/heavier inert particles (glass beads) are used to ix with the resin particles during the adsorption of etal ion in the liquid fluidized bed. The experiental and theoretical results prove that the addition of inert glass beads exerts a significant positive effect on the fil ass transfer coefficient. For an adsorption process in which surface fil resistance is the liiting step, an increase of the fil ass transfer coefficient will result in a decrease of the effluent concentration at the outlet of the reactor and a postponeent of the break tie h b or prolongation of the operating tie resulting in a significant increase of the usable fraction of adsorbent g as defined in Eq.(23). Figure 4. Influence of ass transfer coefficient on break tie. The influence of the ass transfer coefficient on the break tie and therefore the usable fraction of the adsorbent is suarized in Fig. 5, where the relationship between the usable fraction g is plotted as function of the ass transfer coefficient k L in the fluidized bed adsorber. As shown in Fig. 5 for the given operating conditions of experient Mix3, the usable fraction is nearly zero, if the ass transfer coefficient is less then 1.3 * 10 ±5 (/s). That eans that the concentration at the outlet of the colun is higher than the perissible concentration at the beginning of the operation process. With increasing ass transfer coefficient, the usable fraction g increases sharply in the range of k L =1.3 ~ 3 * 10 ±5 (/s), and is then followed by a gradual increase at higher values. Acknowledgeent Financial support by the Coission for Technology and Innovation, Switzerland is gratefully acknowledged. Sybols used Received: January 10, 2000 [CET 1189] a [ 2 ±3 ] surface area per unit volue of reactor Ar [±] Archiedes nuber, = d 3 g=2 p ap c [ol ±3 ] bulk concentration of liquid in reactor c i [ol ±3 ] cation concentration at the liquid ±solid interface Che. Eng. Technol. 23 (2000) 11, Ó WILEY-VCH Verlag GbH, D Weinhei, /00/ $ /0 1011

6 c 0 [ol ±3 ] cation concentration at inlet of the reactor d p [] ean particle diaeter DaI kf [±] Daköhler I nuber = k f q Q DaI kl [±] Daköhler I nuber = k L a " r f [±] diensionless liquid phase concentration h [] height of particle layer in fluidized bed h 0 [] height of particle layer in free settled bed K d [ol ±3 ] Languir isother equilibriu constant k f [ 3 ol ±1 s ±1 ] rate constant of forward reaction k L [s ±1 ] liquid-solid ass transfer coefficient k r [s ±1 ] rate constant of reverse reaction [kg] ass of dry resin q [ol kg ±1 ] concentration of adsorbate in adsorbent q [ol kg ±1 ] axiu adsorption capacity of resin q 0 [ol kg ±1 ] dynaic capacity of resin which is in equilibriu with c 0 Q [ 3 s ±1 ] liquid flow rate t [s] tie T [C] teperature u [s ±1 ] superficial liquid velocity V 0 [ 3 ] volue of free settled resin particles in the reactor V 0g [ 3 ] volue of free settled glass beads in the reactor V [ 3 ] volue of the reactor z [] axial coordinate along the colun Greek sybols e 0 [±] settled bed porosity e p [±] particle porosity e r [±] reactor porosity g [±] usable efficiency of total capacity of adsorbents h [±] diensionless tie h b [±] break diensionless tie r [kg ±3 ] density of liquid solution r ap [kg ±3 ] apparent particle density r ps [kg ±3 ] ass of dry resin per unit volue of swollen resin l [Pa s] liquid viscosity s [s] hydraulic residence tie s [s] tie required for saturating all the resin in reactor n [±] diensionless solid phase concentration w [±] resistance tie nuber to saturate all the resin in reactor References [1] McDougall, G. J.; Fleing, C. A., Extraction of precious etals on activated carbon, in Ion Exchange and Sorption Processes in Hydroetallurgy, Ed. by Streat, M. and Naden, D., Critical Reports on Applied Cheistry 19 (1987) pp. 56±126. [2] Clifford, D.; Zhang, Z., Modifying ion exchange for cobined reoval of uraniu and radiu, J. A. Water Works Assoc. 86 (1994) pp. 214±227. [3] Becker, N. S. C.; Eldridge, R. J., Selective recovery of ercury(ii) fro industrial wastewaters. II. Anion exchange of Hg(II) chlorocoplexes, React. Poly. 22 (1994) pp. 131±138. [4] Petruzzelli, D.; Tiravanti, G.; Santori, M.; Passino, R., Chroiu reoval and recovery fro tannery wastes: Laboratory investigation and field experience on a 10 3 /d deonstration plant, Wat. Sci. Tech. 30 (1994) pp. 225±233. [5] Petruzzelli, D.; Passino, R.; Tiravanti, G., Ion exchange process for chroiu reoval and recovery fro tannery wastes, Ind. Che. Eng. 34 (1995) pp. 2612±2617. [6] Yang, J.; Menoud, P., Adsorption de Øtaux lourds à l'aide de la røsine Chelaine, Copte-rendu 1, EPF, Lausanne [7] Schidt, V., DØveloppeent d'un syst e fluidisø pour la røcupøration des Øtaux lourds des eaux industrielles par des røsines chølatantes, Th se No. 1247, IGC-DC, Ecole Polytechnique FØdØrale, Lausanne [8] Biscans, B., Chroatographie d'øchange d'ions en couche fluidisøe. Extraction des protøines du lactosøru, Th se ENSIGC, Toulouse [9] Biscans, B.; Riba, J. P.; Couderc, J. P., Continuous equipent for ion exchange in a fluidized bed. Prospects and probles, Int. Che. Eng. 28 (1988) pp. 248±256. [10] Han, K. N; Liu, J., Reoval of cadiu and zinc ions by liquid fluid-beds, in: Eerging Proc.Technol. Cleaner Environ. Proc. Syp., Soc.Min. Metall. Expl., Littleton. Colo 1992, pp. 175±183. [11] Zarraa, M. A., Mass transfer during the reoval of dissolved heavy etals fro wastewater flows in fluidized beds of ion exchange resins, Che. Eng. Technol. 15 (1992) pp. 21±25. [12] Nakaura, S.; Ito, Y.; Iwafuchi, K.; Shirakashi, M., Iproveents in uraniu adsorption ability of a circulating fluidized bed adsorber, J. Japan Che. Eng. 27 (1994) pp. 32±38. [13] Huang, L.; Hao, O. J.; Chen, J. M.; Buglass. R. L., Metal toxicity on sulfate reducing bacteria, in: Proceedings of 25th Mid-Atlantic Ind. Waste Conf. (1995) pp. 541±550. [14] Menoud, P., RØcupØration de Øtaux lourds par une røsine chølatante, Th se No. 1563, IGC-DC, Ecole Polytechnique FØdØrale, Lausanne [15] Menoud, P.; Cavin, L.; Renken, A., Modeling of heavy etals adsorption to a chelating resin in a fluidized bed reactor, Che. Eng. Proc. 37 (1998) pp. 89±101. [16] Jeanneret-Gris, G., Brevet europøen, [17] Yang, J.; Renken, A., Intensification of ass transfer in liquid fluidized beds with inert particles, Che. Eng. Proc. 37 (1998) pp. 537±544. [18] Boyer, P. M.; Hsu, J. T., Effect of ligand concentration on protein adsorption in dye-ligand adsorbents, Che. Eng. Sci. 47 (1992) pp. 241± 251. [19] Arnold, F. H.; Blanch, H. W., Analysis of affinity Chroatography. II. Rate theory and the easureent of biological binding kinetics, J. Chroat. 355 (1986) pp. 13±27. [20] Yang, J., Hydrodynaics and liquid-solid ass transfer in non-conventional fluidized beds, Thesis No. 1919, IGC-DC, Ecole Polytechnique FØdØrale, Lausanne [21] Cavin, L., Etude cinøtique de la døsorption de Øtaux lourds sur la røsine ChØlaine par coplexation avec un poly re soluble, Travail de diplôe, EPF, Lausanne [22] Koloini, T.; Zuer, M., Ion exchange with irreversible reaction in deep fluidized beds, Canadian Che. Eng. 57 (1979) pp. 183±190. [23] Horstann, B. J.; Chase, H. A., Modeling the affinity adsorption of iunoglobulin G to protein iobilized to agarose atrices, Che. Eng. Res. Des. 67 (1989) pp. 243±254. [24] ConfØdØration HelvØtique, Ordonnance sur le døverseent des eaux usøes, Ordonnance , (1975) pp. 1±16. [25] Hartinger, L., Handbuch der Abwasser und Recycling Technik, 2nd ED. Carl Hanser Verlag, München, Wien Ó WILEY-VCH Verlag GbH, D Weinhei, /00/ $ /0 Che. Eng. Technol. 23 (2000) 11