Ind. Eng. Chem. Res. 2001, 40, 2679-2684
2679
Membrane Crystallizers Efrem Curcio,† Alessandra Criscuoli,† and Enrico Drioli*,†,‡ Research Institute on Membranes and Modeling of Chemical Reactors and Department of Chemical Engineering and Materials, University of Calabria, via P. Bucci, Cubo 17/C, Arcavacata di Rende (CS) 87030, Italy
Well-controlled crystallization is the best method for preparing materials that are extremely pure. Furthermore, it can also be highly advantageous by operating the crystallization so that the crystals pick up impurities from water. Crystal growth generally starts at solute concentrations at which the nucleation occurs; the growth rate, depending on the supersaturation degree, is determined by a combination of the nature of the growing crystal surface and the diffusional rate. The purpose of this work is to introduce an innovative methodology, the membrane crystallization, to produce crystals from solutions. For the present application, direct contact membrane distillationsa separation process based on hydrophobic microporous membraness has been employed to reach the supersaturation in the crystallization of NaCl from aqueous solutions: volatile components diffuse through the membrane pores as vapor by applying a temperature difference across the membrane which creates a gradient of the equilibrium partial pressures; that is the driving force for the operation. The experiments have been carried out in a laboratory plant for studying the distribution of crystal dimensions, nucleation, and growth rates as a function of the retention time, slurry density, temperature, and supersaturation level of the solution. The kinetic data and product size distribution obtained have been compared with the ones by traditional crystallizers. Introduction Crystallization is an excellent technique for purification of chemical species by solidification from a liquid mixture: many materials are marketed in crystalline form, and large amounts of product may be obtained from impure solutions in one processing step. The attraction of this method is heightened when one considers that crystallization may operate at lower temperatures and requires lower energy than other separation processes. Generally, a distinction between crystallization from the solution and crystallization from the melt occurs. The term “melt crystallization” indicates the separation of mixture components without addition of diluent solvent: the solid phase is formed by cooling of the melt. On the other hand, most of the industrial applications involve crystallization from the solution; in this case, the solution can be directly or indirectly cooled and/or the solvent is evaporated to effect crystallization. As is well-known, crystals obtained by evaporation are formed because of the less solvent available for the ions dissolved; that determines a metastable or supersaturation state in which the nucleation and the crystal growth can occur. A membrane crystallizer has been conceived as an alternative technology for producing crystals from supersatured solutions; the use of the membrane distillation (MD) technique23 in the concentration of a solution by solvent removal in the vapor phase is proposed for this new application. * To whom all correspondence should be addressed. Fax: +39-0984-402103. E-mail:
[email protected]. † Research Institute on Membranes and Modeling of Chemical Reactors. E-mail:
[email protected], a.criscuoli@ irmerc.cs.cnr.it. ‡ Department of Chemical Engineering and Materials.
MD is a relatively recent process framed in the most general class of membrane contactors, an emerging technology in which the membrane promotes the masstransfer operations acting not as a selective barrier but accomplishing the separation on the principle of phase equilibria. The termally driven membrane contactors strategy is one of the innovative approaches for improving various important processes and for designing new scientific routes. The crucial requisite for the process is represented by the ability of the membrane to sustain the vapor-liquid equilibrium interfaces. Aqueous solutions of inorganic substances show high values of the surface tension; because of this, the hydrophobic condition is guaranteed for several microporous membranes such as polypropylene (PP), Teflon (PTFE), and poly(vinylidene fluoride) (PVDF) having a nominal pore size of around 0.2 µm. The efficiency of MD operations is limited by concentration and temperature polarization phenomena because of the mass- and heat-transfer resistances at the boundary layers adjacent to the membrane.18,24 Several typologies of conducting the operation14 exist, depending on the method used to recover the vapor once it has migrated along the membrane. Various previous publications9,10,17 showed that direct contact MD (DCMD) is the most simple, economical, and efficient configuration for the concentration of aqueous inorganic streams by membrane distillation. DCMD has been investigated as an alternative technique for desalination,6,12 the concentration of fruit juices,5,7 the split of azeotropic mixtures,15 wastewater treatment and removal of organic components from aqueous streams,4,8 and separation of liquid mixtures.16 In DCMD, a microporous hydrophobic membrane is in contact with the hot feed at the retentate side and with a condensing solution at the permeate side (also
10.1021/ie000906d CCC: $20.00 © 2001 American Chemical Society Published on Web 05/10/2001
2680
Ind. Eng. Chem. Res., Vol. 40, No. 12, 2001
called the “distillate side”) at different temperatures and compositions.13 At both layers of the membrane, vapor-liquid equilibrium is established, giving rise to a vapor pressure gradient: vapor molecules migrate through the membrane pores from the warmer to the cooler side. The main advantages of MD over conventional thermally driven processes are the total rejection of ions, macromolecules, colloids, cells, and other nonvolatiles, the low operating temperatures and energy requirements, the less demanding membrane mechanical properties, and the reduced global size.20 Furthermore, remote control, easy automatic, soft, and gentle operations are other important characteristics. Membrane Crystallizer Strategy The advantages of membrane crystallization can be elucidated by comparing it with the most important type of conventional crystallization equipment in use today: the circulating-magma crystallizer. Although a number of different varieties and features are included within this category, common and general principles of work can be identified. In this unit, the feed is introduced in the circulating pipe together with the slurry leaving the crystallization body; the total stream is pumped through a tube-and-shell heat exchanger where the temperature increases by about 5-10 °F. The heated magma, returned to the body through a recirculation line, mixes with the body slurrysthat raises its temperature locally near entrysand causes evaporation at the liquid surface. The partial evaporation of the solvent allows one to reach the supersaturation level, generating the deposit of suspended crystals. Because the solvent evaporation and the solute crystallization occur in the same location, the temperature gradients between the surface and the bulk of the body often compromise the suspension uniformity of the solid products. On the contrary, the membrane crystallizer apparatus does not deal with such a problem because this unit is characterized by the dissociation of the two fundamental moments marking crystallization from the solution process: the solvent evaporation occurring inside the membrane module (where the flowing solution is below the supersaturation condition) and the crystallization stage (performed in a separate tank on the retentate line operating in the metastable regime of supersaturation). Thus, the crystals produced are expected to show improved size distribution and global quality. The control and the prediction of the particle size distribution of the solids leaving the system are important objectives in the design and use of crystallizers. When the crystals must undergo successive treatments such as washing, filtering, further crystallization, or reaction with other chemicals, adequate dimensions and size uniformity are required. If the crystals represent the final product, the quality control criteria require single crystals in nonaggregated form, uniform in size, and high purity and with no phenomena of cake formation in their package during storing and transporting. A schematic representation of the continuous membrane crystallizer plant and relative connections has been depicted in Figure 1.
Figure 1. Schematic representation of a membrane crystallizer plant.
Retentate and distillate streams are the foremost lines: they converge, in a countercurrent way, toward the membrane module containing microporous hydrophobic membranes, where the solution is concentrated by solvent evaporation. Crystallization occurs inside a body where a centrifugal pump takes and sends the mother liquor to the membrane module on the retentate side. A crucial requirement for a membrane crystallizer is to prevent crystal deposition inside the membrane; because the solubility of solids in solution increases with temperature, a suitable heating must guarantee that the temperature of the solution flowing along the membrane is fairly high to be always under saturation condition. In this regard, it is necessary to consider that, along the capillary module, thermal exchange phenomena between cold and hot streams cause a progressive reduction of temperature, depending on the fluiddynamic regime. Besides, a filter in the suction lift pipe stops crystal drag, avoiding the blockage of the membrane pores that would prejudice the system performance. The solution coming out from the module is returned back to the magma reservoir. Also on the distillate line a centrifugal pump ensures the countercurrent recycle of the cold stream that removes from the saline solution water vapor passing across the membrane pores: a reservoir represents the drawing and picking tank for the distillate. The conceptual elaboration of the membrane crystallizer might be included in the innovative strategy of the “process intensification” approach. Process intensification consists of the development of novel apparatuses and techniques that, compared to those commonly used today, are expected to bring dramatic improvements in manufacturing and processing, essentially decreasing the equipment size/production capacity ratio, energy consumption, or waste production.19 The available area between the cold and warm streams of the hollow fiber membrane contactor is substantially larger than that in the heat exchanger equipment in the conventional crystallizers. The specific area with a fiber of 10-3 mm inner diameter can be about 104 m2/m3, at least 1 order of
Ind. Eng. Chem. Res., Vol. 40, No. 12, 2001 2681
magnitude higher than those of traditional shell-andtube units. Therefore, the large mass-transfer area offers the possibility of creating a compact contactor with a high geometrical surface enclosed in a small volume. The crystallization of sodium chloride from an aqueous solution is a typical system of most inorganic crystallizing systems; the experimental work presented in this paper is aimed at evaluating the performance of the proposed membrane apparatus applied to this system. In all types of crystallizers, the particle size distribution is determined by the interaction between nucleation and growth rates. In this work, the quantification of kinetic parameters has been made by an idealized crystallizer model, the mixed-suspension-mixed-productremoval model (MSMPR) characterized by equipment operating at steady state, containing a mixed-suspension of magma with a uniform supersaturation degree.1 Kinetic data allowed one to obtain a semiempirical power law relation describing the nucleation rate of sodium chloride. Nucleation kinetics related to the hydrodynamics, physical solution properties, geometry, and design characteristics have frequently been the object of theoretical and experimental studies for a conventional crystallization apparatus;11,22 then, the resulting kinetic expression has been compared with those available in the literature. Mathematical Relationships The MSMPR crystallizer technique represents the most popular means for evaluating the growth rate G and the nucleation rate B0, according to the exponential decrease at steady state of the crystal size distribution (CSD)
n ) n0 exp(-L/Gt)
(1)
proposed by Randolph and Larson3 for systems which obey some assumptions such as perfect mixing, no classification at withdrawal, uniform shape factor, sizeindependent growth rate, etc. The relation which connects the nucleation rate B0 with the zero-size particle density n0 is
B0 ) Gn0
(2)
The moment equations allow one to derive the product size distribution;21 the cumulative mass distribution, defined as the weight fraction up to the size L, can be obtained from the equation 3
W(L) ) 1 -
[ ( ) ( )] 1 L
∑ i)0 i! Gt
i
exp -
L
Gt
(3)
Finally, the kinetic data obtained from the slope and intercept of the semilog population density plot are correlated by an empirical expression:
B0 ) kmiGj
(4)
Laboratory Apparatus The design of the laboratory apparatus used for this work has been made according to the scheme shown in Figure 1 and previously described.
Table 1. Structural Parameters of an Enka Microdyn MD020CP-2N Module material type no. of fibers packing density external diameter of the fibers membrane thickness length of the fibers available area nominal pore size shell diameter
PP hollow fiber 40 70% 1.8 mm 120 µm 45 cm 0.1 m2 0.20 µm 2.1 cm
The membrane module MD020CP-2N, supplied by Microdyn, contains hydrophobic PP hollow fibers of 0.1 m2 interfacial area; the main characteristics are specified in Table 1. Four platinum thermocouples (Pt100) placed at the inlet and outlet of the module on the retentate and permeate lines allow a quantification of the thermal drop, generally comprised within 3-4 °C. On an equal number of Oterman barometers, disposed in a similar way, it is possible to read operative pressures that have to be lower than the pore penetration threshold for liquid (breakthrough pressure). Both retentate and distillate lines are also furnished with Brooks Instruments mass flowmeters with a capacity of up to 5.7 L/min. A cooling bath (Neslab model RTE300D) provides the maintenance of the established temperature value by removing the heat flux that the retentate transfers to the permeate stream; the warming of the retentate line is made by a Haake D1 heater. The estimation of the transmembrane flux occurs by evaluating weight variations in the distillate tank with a Precisa 300S balance. Experimental Procedures and Results and Discussion Step One: Concentration of the Solution from an Undersaturation to a Metastable Phase. Initial tests led to a rise of the concentration level of the saline solution from an initial value to a final one slightly higher than the saturation point. This operation has been carried out in a batch mode. The crystallization plant has been loaded with aqueous solutions of sodium chloride of known concentration (28.30 g of NaCl/100 g of H2O); the permeate stream, consisting of demineralized water, has been recirculated in a countercurrent mode with respect to the retentate one. Both streams have been supplied to the module with a flow rate of 100 L/h. After a transitory state, the temperatures observed at the module entrance for the retentate and permeate sides were of 29 and 9 °C, respectively; the operative pressures ranged from 1 to 1.2 absolute atmospheres. The crystallization tank worked at 25 °C and atmospheric pressure. From an initial value of approximately 1.4 × 10-4 kg/ (m2 s), the water flux progressively decreased in concomitance with the increase of the solution concentration and the consequent decrease of the solution activity coefficient; at the end of the test, the reduction was of about 20%. The increase of the concentration has been estimated by refractometric analysis using an established appropriate correlation between the instrumental
2682
Ind. Eng. Chem. Res., Vol. 40, No. 12, 2001
Figure 2. Trend of the water transmembrane flux and an increase of the concentration of the solution versus time in the concentration step.
measure (expressed in Brix degree) and the effective concentration of the tested solution. Figure 2 shows the water transmembrane flux obtained during a test mode at the operating conditions reported above; moreover, the salinity increase, also reported on this figure, exhibits a linear trend with a rate of 0.2 kg of NaCl/100 kg of H2O/h. The batch run was stopped after about 40 h at the degree of supersaturation established (saturation at the operating conditions used equals 36.15 g of NaCl/100 g of H2O). Step Two: Crystallization from Solution. After achievement of a solution saturation level that guarantees the starting of the crystallization, the operation was carried in a continuous mode, introducing a feed stream of fresh undersaturated solution into the retentate line; the flow rate was adjusted in order to achieve, on the basis of the crystallizer equipment volume, the desired retention time. The supersaturation degree requested was fulfilled by means of simple tuning between characteristics of a supplied solution and water flux removed by DCMD. The retention time in most commercial equipments is on the order of hours, though nucleation in a supersaturated solution may occur in less than a second. Because the number of nuclei formed influences the size and shape of the crystals obtained, the effects of an upset in the system can be serious. In this study, three to four retention periods have passed before the effects of an upset can be considered negligible. The membrane distillation performances, in the supersaturation stage, have been investigated by monitoring the water flux trend as illustrated in Figure 3. The slight decline of the transmembrane flux that appears after 1 h of operation is due to the recirculation in the system of some particles formed on the filter surface. Crystals obtained show the characteristic cubic form in accordance with the expected geometry of the NaCl crystallized.
Figure 3. Transmembrane flux in supersaturation conditions.
Figure 4. Cumulative size distribution of crystals per unit of volume versus particle size at operative conditions indicated in the legend.
The CSD of the total crystals contained in a known volume of magma was measured by a screen analysis performed via a video microscope. Experimental data relative to the cumulative mass fractional distribution of averaged slurry samples as a function of the particle size are shown in Figure 4 at three different operative conditions (details are reported in the correspondent legend); the curve represents a nonlinear regression fit of the data by eq 3. When the particle dimension is correlated to the cumulative percent function (on the log probability scale), the crystal distribution is generally characterized by the coefficient of variation, defined as
CV ) 100
PD84% - PD16% 2PD50%
(5)
where CV ) coefficient of variation, as a percentage, and PD ) particle dimension at the percent indicated.
Ind. Eng. Chem. Res., Vol. 40, No. 12, 2001 2683
Figure 5. Cumulative size particle distribution data for three slurry samples drawn from different locations of the membrane crystallizer at the operative conditions indicated in the legend.
Figure 6. Typical plot of the logarithm of crystal population density versus crystal size.
Table 2. Values of Coefficients of Variation (CV) at Different Crystallization Conditions
Table 3. Semiempirical Power Law Relations for Different Conventional Crystallizer Configurations (Impeller Speed N Generally Ranges from 10 to 30 rps)
residence time t (s)
magma density M (kg/m3)
CV (%)
1800 4300 8600
2.2 3.2 3.8
57.2 51.7 42.3
Some experimental evaluations of this parameter are reported in Table 2; the value of about 50%, obtained when the product is removed from a conventional mixedsuspension crystallizer, provides a first comparison between traditional equipment and the membrane crystallizer.2 Figure 5 shows the cumulative mass fraction distribution for crystals produced at a retention time t of 4300 s and a magma density M of 3.2 kg/m3. Data refer to three slurry samples, extracted from different positions in the membrane crystallizer, to verify that the system was well-mixed: the overall agreement of plotted points confirms this assumption. Figure 6 reports the estimated population density data versus crystal size for a representative set of averaged samples at t ) 4300 s and M ) 3.2 kg/m3; a linear regression of eq 1 is used to determine both the growth rate G and the effective nuclei density n0 for each run. An empirical correlation which describes, for the investigated system, the nucleation rate B0 including homogeneous, primary, and secondary nucleation, as a function of the growth rates G and the slurry density M, can be expressed as
B0 ) 7.30 × 1018M1.00G2.08
(6)
with a correlation coefficient of 0.92. The relationship (6) can be checked against the power law relations reported in Table 3 relative to conventional crystallizer configurations. Although a substantial agreement exists in regards to the exponents on G and M, a sensible discordance on the value of the k coefficient is quite apparent because
crystallizer
volume (m3)
correlation
MSMPR MSMPR pilot unit pilot unit Weston Point Swenson Evap.
0.055 0.091 1.33 1.0-1.8 280 121
8 × 1016N2MG2 (3.5/11) × 1016N2MG2 (0.4/1.5) × 1016N2MG2 (0.5/1.5) × 1016N2MG2 2.5 × 1016N2MG2 (10/20) × 1016N2MG2
of the differences in the hydraulic characteristics of MSMPR crystallizers, as suggested by Grootscholten et al.11 Conclusion In this work an innovative technology, the membrane crystallization, has been employed to produce crystals from supersatured solutions. The experimental tests, aimed at obtaining sodium chloride crystals from an aqueous solution of this electrolyte, have been carried out in a laboratory plant for studying the distribution of crystal dimensions, nucleation, and growth rates as a function of the retention time, slurry density, and supersaturation level of the solution. The kinetic parameters, investigated by using a MSMPR crystallizer model, have been joined in a power law relation describing the nucleation rate of sodium chloride as a function of the growth rate and magma density. The achieved results, in substantial agreement with those reported in the literature, confirm the interesting potentialities of the membrane crystallization strategy in redesigning traditional crystallizers. Symbols B0 ) nucleation rate [no./(m3 s)] CV ) coefficient of variation [%] G ) growth rate [m/s] K ) constant in eq 4 L ) characteristic crystal dimension [µm] M ) slurry density [kg/m3]
2684
Ind. Eng. Chem. Res., Vol. 40, No. 12, 2001
N ) crystal population density [no./m4] n0 ) nuclei population density [no./m4] PD ) particle dimension [µm] t ) retention (or drawdown) time [s] W ) cumulative mass distribution function [1/µm]
Literature Cited (1) McCabe, W. L.; Smith, J. C. Unit Operations of Chemical Enginering; McGraw-Hill Book Co.: New York, 1976. (2) Perry, J. H. Perry’s Chemical Eengineers’ Handbook; McGrawHill Book Co.: New York, 1987. (3) Randolph, A. D.; Larson, M. A. Theory of particulate processes; Academic Press: New York, 1971. (4) Calabro`, V.; Drioli, E.; Matera, F. Membrane distillation in the textile wastewater treatment. Desalination 1991, 83, 209. (5) Calabro`, V.; Jiao, B. L.; Drioli, E. Theoretical and Experimental Study on membrane distillation in the concentration of orange juice. Ind. Eng. Chem. Res. 1994, 33, 1803. (6) Criscuoli, A.; Drioli, E. Energetic and exergetic analysis of an integrated membrane desalination system. Desalination 1999, 124, 243. (7) Curcio, E.; Barbieri, G.; Drioli, E. Direct contact membrane distillation in fruit juice concentration. Ind. Bevande 2000, 166, 113. (8) Drioli, E.; Wu, Y.; Calabro`, V. Membrane distillation in the treatment of aqueous solution. J. Membr. Sci. 1987, 33, 277. (9) Fane, A. G.; Schofield, R. W.; Fell, C. J. D. The efficient use of energy in membrane distillation. Desalination 1987, 64, 231. (10) Godino, M. P.; Pena, L.; Rincon, C.; Mengual, J. I. Water production from brines by membrane distillation. Desalination 1996, 108, 91. (11) Grootscholten, P. A. M.; Brekel, L. D. M.; Jong, E. Effect of scale-up on secondary nucleation kinetics for the sodium chloride-water system. Chem. Eng. Res. Des. 1984, 26, 179. (12) Lagana`, F.; Drioli, E.; Barbieri, G.; Criscuoli, A. Integrated membrane operations in desalination processes. Desalination 1999, 122, 141.
(13) Lagana`, F.; Barbieri, G.; Drioli, E. Direct contact membrane distillation: modelling and concentration experiments. J. Membr. Sci. 2000, 166/1, 1. (14) Lawson, K. W.; Lloyd, D. R. Membrane distillation, review. J. Membr. Sci. 1997, 124, 1. (15) McDowell, J. K.; Davis, J. F. A characterization of diffusion distillation for azeotropic separation. Ind. Eng. Chem. Res. 1988, 27, 2139. (16) Sarti, G. C.; Gostoli, C.; Bandini, S. Separation of liquid mixtures by membrane distillation. J. Membr. Sci. 1993, 80, 21. (17) Sarti, G. C.; Gostoli, C.; Matulli, S. Low Energy Cost Desalination Processes Using Hydrophobic Membranes. Desalination 1995, 56, 277. (18) Schofield, R. W.; Fane, A. G.; Fell, C. J. D. Heat and mass transfer in membrane distillation. J. Membr. Sci. 1987, 33, 299. (19) Stankiewicz, A. I.; Moulijn, A. J. Process Intensification: Transforming Chemical Engineering. Chem. Eng. Prog. 2000, 96/ 1, 22. (20) Tomaszewska, M. Membrane Distillation. Environ. Prot. Eng. 1999, 25/1, 37. (21) Toyokura, K. New aspects of industrial crystallization. J. Chem. Eng. Jpn. 1995, 28/4, 361. (22) Zumstein, R. C.; Rousseau, R. W. Utilization of industrial data in the development of a model for crystallizer simulation. AIChE Symposium Series; AIChE: New York, 1987; Vol. 83, p 130. (23) Bodell, B. R. Silicone rubber vapor diffusion in saline water distillation. U.S. Patent 3,361,645, 1968. (24) Drioli, E.; Molero, L. P.; Criscuoli, A. Membrane Distillation. EOLSS Encycl. 2000, in press.
Received for review October 24, 2000 Revised manuscript received March 19, 2001 Accepted March 21, 2001 IE000906D
