Technical viability and exergy analysis of membrane crystallization: Closing the loop of CO2 sequestration

International Journal of Greenhouse Gas Control 12 (2013) 450–459

Contents lists available at SciVerse ScienceDirect

International Journal of Greenhouse Gas Control journal homepage: www.elsevier.com/locate/ijggc

Technical viability and exergy analysis of membrane crystallization: Closing the loop of CO2 sequestration P. Luis ∗ , D. Van Aubel, B. Van der Bruggen Department of Chemical Engineering, Laboratory for Applied Physical Chemistry and Environmental Technology, K.U. Leuven, W. de Croylaan 46, B-3001 Leuven, Belgium

a r t i c l e

i n f o

Article history: Received 12 June 2012 Received in revised form 24 November 2012 Accepted 28 November 2012 Available online 31 December 2012 Keywords: Membrane crystallization Sodium carbonate Sodium chloride Magnesium chloride Mass transfer Exergy analysis

a b s t r a c t The potential of membrane crystallization has been studied for the recovery of Na2 CO3 from aqueous streams, as a tool that can be used in CO2 sequestration. The influence of various crystallization conditions (i.e., concentration and flowrate of the Na2 CO3 solution, concentration and flowrate of the osmotic solution and kind of osmotic solution: NaCl or MgCl2 ) on the process performance has been determined. The concentration of the osmotic solution was found to be the key parameter that may condition the applicability of the system. The flowrate of the Na2 CO3 solution has also an important influence on the mass transfer through the membrane. Thus, concentration polarization may occur since the transmembrane flux increases with the flowrate. In addition, the characterization of crystals demonstrated that Na2 CO3 ·10H2 O crystals were obtained. An exergy analysis was made for the membrane contactor. The higher the concentration of NaCl, the higher the exergy of the outlet streams. It yields a positive variation of exergy, reaching a maximum value at a specific Na2 CO3 concentration. In addition, a range of concentrations can be established in order to operate under conditions of positive exergy variation. Increasing the flowrates of the feed or osmotic solutions has also an effect on the exergy variation, reaching an asymptotic value. © 2012 Elsevier Ltd. All rights reserved.

1. Introduction Carbon dioxide is the most critical greenhouse gas and its capture and recovery for further reuse or storage is being intensively researched to develop most efficient and modern methods (Luis et al., 2012). Several technologies and strategies are considered, depending on the application (Aaron and Tsouris, 2005; Kuramochi et al., 2012; Luis et al., 2011; Merkel et al., 2010). Both pre-combustion and post-combustion oriented technologies involve the separation of CO2 from a gas mixture, composed of CO2 and H2 in the first approach and CO2 diluted in air and other combustion gases, such as sulphur dioxide or nitrogen oxides, in post-combustion. Absorption using amines (e.g., using monoethylamine, MEA) is the reference technology to separate CO2 from flue gases in a post-combustion scenario, where an absorber removes CO2 and a regenerator (or stripper) releases the CO2 in a concentrated form and the original solvent is recovered (Aaron and Tsouris, 2005; Ebner and Ritter, 2009; Olajire, 2010). For the precombustion capture approach, due to the high concentration of CO2 in the high pressure gas, physical solvents (mixture of dimethyl ethers of polyethylene glycol in Selexol process or refrigerated

∗ Corresponding author. Tel.: +32 16 322348; fax: +32 16 322991. E-mail address: [email protected] (P. Luis). 1750-5836/$ – see front matter © 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.ijggc.2012.11.027

methanol in Rectisol process) are used to take advantage of the high pressure of the gas stream (Figueroa et al., 2008). Both processes are very intensive from the point of view of energy and material consumption (Luis et al., 2012). Changing the organic solvent for other more environmentally friendly solutions, such as water-based solutions, are proposed in the recent literature (ElNaas et al., 2010; Mansourizadeh et al., 2010; Zhang et al., 2010). For example, sodium hydroxide has shown high CO2 removal efficiency but its further reuse is still subject of research. When an aqueous alkaline solution is used as the liquid absorbent, CO2 reacts with the hydroxyl ions within the liquid film as follows (Mansourizadeh et al., 2010): CO2 + OH− → HCO3 − −

−

HCO3 + OH → CO3

(R1) 2−

(R2)

where reaction (R1) is the second order in the forward direction and the carbonization reaction (R2) is much faster than reaction (R1). Thus, the carbonate is formed in solution and its reuse is conditioned by the further purification steps. Alkaline absorption has also achieved much attention in the direct capture of CO2 from ambient air, “air capture”, in order to manage dispersed emissions, such as transportation (Mahmoudkhani and Keith, 2009). But the main challenge that has to be solved to close the loop by reusing the carbonate (e.g., NaCO3 ) or recovering the reagent (e.g., NaOH) and the pure CO2 is the development of a process that allows

P. Luis et al. / International Journal of Greenhouse Gas Control 12 (2013) 450–459

obtaining the dried carbonate (e.g., Na2 CO3 ). It may be reused directly as reagent in the industry (e.g., in the ceramic industry) or it can be converted back to sodium hydroxide and pure CO2 by means of, for example, a causticization process with Ca(OH)2 (Mahmoudkhani and Keith, 2009). This work is focused on developing a process in which the carbonate, Na2 CO3 , can be obtained in optimal conditions for further reuse. Membrane crystallization is proposed as the technology to crystallize Na2 CO3 since it has demonstrated a high potential for the recovery of salts from concentrates (Curcio et al., 2010; Drioli et al., 2004, 2005; Ji et al., 2010; Kim, 2011). Crystallization of ionic salts (Kieffer et al., 2009a,b; Watanabe and Akashi, 2009), metal ions (Tang et al., 2010), low molecular organic acids (Curcio et al., 2003; Kuhn et al., 2009), proteins and pharmaceutical compounds (Di Profio et al., 2005, 2007, 2009; Zarkadas and Sirkar, 2006; Zhang et al., 2008) are examples of the applicability of this technology. In addition, the main advantages of membrane crystallization have been already demonstrated: (1) it is possible to control the maximum level of supersaturation due to a defined mass transfer through the membrane (Charcosset, 2009); (2) the membrane induces heterogeneous nucleation; (3) size, shape and purity of crystals can be controlled; (4) there is a significant reduction of energy consumption compared to conventional crystallization by means of cooling or evaporation (Drioli et al., 2006); and (5) comparable or slightly higher nucleation rates with respect to batch crystallizers or tubular precipitators have been obtained (Zarkadas and Sirkar, 2006). Furthermore, the use of membranes has been already considered to satisfy the requirements established by the “process intensification” strategy (Criscuoli and Drioli, 2007; Drioli and Curcio, 2007; Van Gerven and Stankiewicz, 2009). The basic concept of membrane crystallization consists of a technique in which crystal nucleation and growth is carried out across a well-controlled pathway, starting from an undersaturated solution, by adjusting solution composition (supersaturation) by means of a membrane (Di Profio et al., 2010). The membrane acts as a barrier that separates the feed stream (with the compound to crystallize) and the stripping stream, which consists of a hypertonic solution of inert salts (NaCl, CaCl2 , etc.) that enhances mass transfer of water in the vapor phase from the feed to the stripping side, resulting in the supersaturation of the feed stream under isothermal conditions. Thus, the membrane is not selective and the separation is based on the phase equilibrium (Curcio and Drioli, 2005). Examples of hypertonic solutions can be the brines obtained from integrated membrane desalination systems, reaching concentrations higher than 250 g/L of salt (Criscuoli and Drioli, 1999), of which the recovery and/or reuse is subject of further study (Charcosset et al., 2010; Rodríguez-DeLaNuez et al., 2012; Van der Bruggen et al., 2003; Van der Bruggen and Braeken, 2006). In addition, the porous surface of the membrane induces heterogeneous nucleation (formation of nuclei locally close to the surface) while the crystal growth can be produced in a separate place. The driving force for the isothermal transport of volatile compounds through the membrane is given by a partial pressure gradient due to an activity difference between the solutions contacting the membrane (Curcio and Drioli, 2005). This gradient induces evaporation of the solvent at the liquid/membrane interface on the feed side, the migration of the solvent in vapor phase through the porous of the membrane and its recondensation at the membrane/liquid interface on the stripping side. The continuous removal or solvent from the feed solution increases the solute concentration and generates supersaturation (Di Profio et al., 2010). The main requirement is the use of hydrophobic membranes in order to ensure the non-wettability of the membrane. If membrane wetting occurs, the resistance to mass transfer that the membrane itself produces will increase significantly, reducing the flux of the volatile compound through the membrane and decreasing, thus, the process efficiency increasing the operation or investment costs

451

Table 1 Characteristics of the membrane contactor and hollow fibers. Contactor type

Liqui-Cel® 1 × 5.5 MinimoduleTM

Module configuration Housing/potting

Hollow fibers Polycarbonate/polyurethane

Membrane type

Celgard® X50-215 Microporous Fiber

Membrane material Porosity Effective pore size Inner diameter/outer diameter Active surface area Number of fibers Burst strength Contact angle ()

PP (hydrophobic) 40% 0.04 ␮m 300 ␮m/220 ␮m 0.18 m2 2300 27 bar 112◦ (Azoury et al., 1986)

(more time or larger membranes should be required to obtain the same production of crystals). The nucleation rate, i.e., the rate at which the nuclei are formed, in a membrane crystallizer depends on the concentration of crystals in the magma, hydrodynamics (e.g., stirrer speed, pump impeller, rotation rate) and the difference of concentration between the mother liquor (feed solution) and the value at equilibrium (solubility) (Curcio and Drioli, 2005). Different conditions lead to different supersaturation dynamics, affecting the quantity and quality of the crystals produced (Zhang et al., 2008). Thus, their evaluation and control of supersaturation are crucial to adjust the working conditions to a specific membrane. In this work, the crystallization of NaCO3 is studied systematically in order to evaluate the effect of the following variables on the process performance and crystal characteristics: (i) concentration of NaCO3 in the feed solution; (ii) concentration of salt in the stripping solution; (iii) flow rate of the feed and osmotic solution; (iv) kind of stripping agent (NaCl, MgCl2 ); (v) stirring velocity of the feed solution. The crystal structure is evaluated and an exergy analysis of the membrane contactor is made, aiming at establishing the reference conditions and the technical viability of the crystallization process using a membrane contactor. 2. Experimental 2.1. Materials and experimental procedure Deionized water (18.2 M cm−1 ) was used to prepare the feed and stripping solutions. NaCO3 (anhydrous, Riedel-de Haën, purity >99.5%) was used to prepare the feed solution, and NaCl (AnalaR NORMAPUR, purity >99.7%) and MgCl2 ·6H2 O (Applichem, purity >99%) were used for the osmotic solutions. The equivalent MgCl2 concentration in g/L was considered in the calculations. A hollow fiber membrane contactor (1 × 5,5 MiniModuleTM Liqui-Cel® , Membrana GmbH, Germany) was used as membrane crystallizer. Details are given in Table 1. Two peristaltic pumps (Watson Marlow 503S-Belgium and Gilson Minipuls IIIThe Netherlands) recirculated the feed stream and the stripping stream from the cylindrical glasses to the membrane contactor in a counter-current mode with flow rates ranging from 500 to 2500 ␮m/s. The feed consisted of Na2 CO3 solutions with concentrations 150 and 200 g/L. Aqueous solutions of NaCl or MgCl2 with concentrations ranging between 35 and 300 g/L were used as stripping agent. The feed and stripping solutions were introduced in the lumen and shell sides of the contactor, respectively. The experiments were performed at room temperature (20 ±1 ◦ C). A balance was placed below the feed stream glass cylinder in order to measure the weight of the solution over time. Fig. 1 shows the setup of the process. The stripping solution was stirred with a magnetic mixer (Fisher Scientific, Belgium) at 300 rpm. A similar

452

P. Luis et al. / International Journal of Greenhouse Gas Control 12 (2013) 450–459

Fig. 1. Experimental setup for membrane crystallization.

magnetic mixer was used for the feed solution to study the influence of the stirring velocity (60, 120 and 180 rpm) on the crystal characteristics. The variation of weight over time allows the calculation of the transmembrane flux J, which indicates the rate of solvent extraction. The flux is calculated from a central difference equation:



wf (ti+1 ) − wf (ti−1 ) 1 dVf,tot  1 J(ti ) = − ≈− A dt  Awater ti+1 − ti−1

(1)

ti

in which Wf (g) is the weight of the reservoir containing the Na2 CO3 solution at time ti (min), A (m2 ) is the membrane area and water (g/mL) is the density of water. Because water permeates from the fiber to the permeate side, the Na2 CO3 concentration on the fiber side increases until it reaches saturation, while the NaCl/MgCl2 concentration decreases. This is referred to as a positive flux. At the end of the experiments, the crystals were collected, dried under vacuum and weighed, before performing the analytical measurements described in Section 2.2. The concentration in the shell and tube sides changes throughout the contactor and therefore, the flux also changes. In order to consider an overall driving force in the contactor, a logarithmic activity dependency is considered and the overall mass transfer coefficient, K, can be obtained experimentally according the following equation: J = K · P vap (T )

((af,in − ap,out ) − (af,out − ap,in )) ln[(af,in − ap,out )]/(af,out − ap.in )]

2.2.2. Dynamic laser scattering (DLS) After filtration of the crystals, dynamic laser scattering (DLS) was used to detect small crystals in the liquid phase. A sample of 3 mL was placed in a ALV-CGS3 goniometer at different angles and a wavelength of 632.8 nm. The sample holder was sprayed with toluene, which has the same refractive index as the glass of the sample holder, in order to avoid interferences in the measurement.

(2)

in which the subscripts f and p refer to the feed (tube side) and permeate (shell side), respectively, at the inlet (in) and the outlet (out) of the contactor in a countercurrent mode. These activities are calculated as the product between the molar fraction of water and the water activity coefficients (a = x·), which are calculated from the Debye–Hückel theory as described in Appendix A. The molar fractions at the outlet of the contactor (xf,out and xp,out ) are calculated from mass balances assuming that no salts can permeate. The vapor pressure was calculated with Eq. (A.4) in Appendix A.

2.2.3. Total water fraction (TWF) To analyze the fraction of water in the crystal samples, the following procedure was applied. First a portion of crystals was weighed and put on a platter. These platters were then put into a chamber at 25 ◦ C for 24 h. After this first period, the platters were first weighed again and then placed into an oven at 105 ◦ C. Then after 2 h, 4 h and finally 72 h these platters were weighed again. Between 4 h and 72 h, no difference in weight was observed. It was assumed that only water can evaporate, thus, the total water fraction can be calculated: TWF =

Winitial − W105 ◦ C, 72 h Winitial − Wplatter

(3)

in which Winitial is the initial gross weight, W105◦ C,72h is the gross weight after a 72 h period at a temperature of 105 ◦ C and Wplatter is the weight of the platter. 2.2.4. Ion chromatography To investigate possible impurities of Cl− in the crystals, ion exchange chromatography (IC) is used. 1 g of sample is dissolved into 100 mL of MilliQ water. This solution is then placed in a shaker at 160 rpm for 24 h. The sample is processed by the ICS-2000, Ion

Sample (after at least 24h)

Vacuum iltration Liquid fracon

2.2. Analytical methods

Crystals Total water fraction (TWF)

pH measurement

(Water content in crystals)

(Supersaturation)

Fig. 2 shows the different methods used to characterize of crystals and the supernatant solution.

Dynamic Laser Scattering (DLS) (Investigate remaining crystals in the liquid fraction)

2.2.1. pH measurements To assess the remaining degree of supersaturation in the liquid phase, the pH is measured with a pH meter of Orion, model 420A. It was calibrated using a pH 4 and a pH 12 buffer.

Ion Chromatography (IC) & X-ray luorescence (XRF) (Purity of crystals)

Microscopy (Shape and aggregation of crystals)

Fig. 2. Schematic overview of the different analytical methods used.

P. Luis et al. / International Journal of Greenhouse Gas Control 12 (2013) 450–459

453

Fig. 3. Transmembrane flux as a function of time: (a) NaCl solution as osmotic solution and (b) MgCl2 solution as osmotic solution. Feed solution: pure water.

Chromatography System to yield qualitative insights of the presence of Cl− .

et al., 1995):



Ex = G · Cp (T − T0 ) − Cp T0 ln 2.2.5. X-ray fluorescence (XRF) To gather information on the chemical elements present and the stoichiometry in the crystal sample, X-ray fluorescence (XRF) is used. The obtained samples were analyzed with a X-ray diffraction meter (PW 2400, Philips) working at 40 kV and 20 mA at room temperature to obtain the X-ray diffraction pattern and determine the composition of the crystals.

=

3. Exergy analysis The total energy of a system can be divided in two parts: exergy and anergy. The exergy is the part of the energy that can be converted through reversible transformation from one form to another, and the anergy is the part of the energy that is wasted to the environment as heat in conditions of complete degradation (Drioli et al., 2006; Molinari et al., 1995). The exergy can be thus defined as the maximum amount of work obtained by the evolution of a system with reversible transformations from the initial state to the equilibrium state with the environment, this is, the work available in the system because of its non-equilibrium with respect to the reference conditions (Molinari et al., 1995). If the intensive parameters governing the system are temperature, pressure and composition, the exergy is described by the following equation (Criscuoli and Drioli, 1999; Drioli et al., 2006; Macedonio et al., 2007; Molinari

Table 2 Distance per pixel that correspond to a specific magnification. Magnification

Distance per pixel (␮m/px)

5× 10× 20× 50×

1.3 0.65 0.31 0.13

+ ExP

T0

+

(p − p0 ) − NS RT0 ln x1 

+ ExC

 (4)

in which G is the mass flow, CP is the specific heat of the solution,  is the solution density, and T0 (=293.15 K) and P0 (0.1 MPa) are the temperature and pressure in the reference state, respectively. NS is the moles of solvent for unit weight of the solution, defined as: NS =

2.2.6. Microscopy The crystal samples are examined in an Olympus microscope at 5×, 10× and 20× magnifications. The images of these samples are processed with the HiPic 8.3.0 computer program. A calibration was also performed on the pixel/distance relationship. This relation is shown in Table 2.

ExT

T

1000 −

ci

/

(5)

MWs

and x1 is the molar fraction of the solvent: xi =

NS +



NS

(ˇi ci /MWi )

(6)

ci is the weight concentration of the i-component per liter of solution, MWS and MWi are the molecular weight of the solvent and of the i-component, respectively, and ˇi is the number of particles generated by the dissociation of the component i in the solution. In addition, the exergy variation between the outlet and inlet streams, Ex, can be calculated from: Ex =



i

Exout −



j

Exin

(7)

This variation of exergy can be positive or negative, meaning exergy transferred to components and exergy destroyed by components (Drioli et al., 2006; Cerci, 2002). 4. Results and discussion 4.1. Influence of concentration of osmotic solution In order to obtain a reference case (maximum flux) for the crystallization experiments, two series of experiments were carried out using distilled water as feed solution and NaCl or MgCl2 solution in the permeate side. Different concentrations of NaCl or MgCl2 ranging from 90 to 324 g/L and 152 to 378 g/L, respectively, were used. The flow rate through the fibers was fixed at 85 mL/min and through the shell at 45 mL/min. The driving force causes the permeation of water from the feed side to the permeate side, rendering a positive flux. Fig. 3 shows the transmembrane flux as a function of time for the different concentrations of electrolyte. It can be observed that the flux when MgCl2 solution is used is slightly higher than that of NaCl. This is expected because of the higher ionic strength of the

454

P. Luis et al. / International Journal of Greenhouse Gas Control 12 (2013) 450–459

Different concentrations of NaCl solution (50 to 300 g/L) are contacted with the Na2 CO3 solution, with a concentration of 150 and 200 g/L. During these experiments, the velocity through the fibers is kept at 1000 ␮m/s (5.24 mL/min). The flow rate through the shell is kept at 150 mL/min. Fig. 5 shows the results of the transmembrane flux, indicating as expected a higher flux for lower concentrations of Na2 CO3 . After about 2 h of experiment, quite steady state conditions are achieved. For high concentrations of Na2 CO3 , there is an initial time where the flux declines more drastically due to the decrease of the driving force. Fig. 6 shows the average flux achieved for both systems. There is a slight difference in the mass transfer when 150 g/L and 200 g/L Na2 CO3 solutions are used. Because the water activity depends much more on the NaCl concentration than on the Na2 CO3 concentration, there is a smaller influence of the Na2 CO3 concentration on the flux. Furthermore, the NaCl concentration at which the flux becomes negative (water permeates from the osmotic solution to the feed solution) can be determined: 75 g/L NaCl for 150 g/L Na2 CO3 solution, and 100 g/L NaCl for 200 g/L Na2 CO3 solution. Working at lower NaCl concentrations makes the process technically unviable. Taking these results into account, Fig. 7 can be plotted as a reference to consider when selecting the concentration of the osmotic solution. The experimental overall mass transfer coefficient K has been calculated from Eq. (2). Fig. 8 shows K as a function of the

Fig. 4. Average transmembrane flux as function of the osmotic solution concentration. Feed solution: pure water.

MgCl2 solution. For a better comparison, Fig. 4 shows the average flux when both osmotic solutions are used. 4.2. Influence of concentration of feed solution When a Na2 CO3 solution is used as the feed, the smaller driving force compared to pure water produces a decrease of the transmembrane flux. 1,8

35 g/l NaCl 75 g/l NaCl 100 g/l NaCl 150 g/l NaCl 200 g/l NaCl 250 g/l NaCl 300 g/l NaCl

Transmembrane flux (ml.m-2min-1)

1,6 1,4 1,2 1,0 0,8 0,6 0,4 0,2 0,0 -0,2 -0,4 0

50

100

150

200

250

Time (min)

Transmembrane flux (ml.m-2min-1)

1,5

50 g/l NaCl 100 g/l NaCl 150 g/l NaCl 200 g/l NaCl 250 g/l NaCl 300 g/l NaCl

1,0

0,5

0,0

-0,5 0

50

100

150

200

250

Time (min) Fig. 5. Transmembrane flux as a function of time for an initial Na2 CO3 concentration of (a) 150 g/L and (b) 200 g/L, in the feed solution (fiber side) and different concentrations of NaCl in the osmotic solution (shell side). The velocity of the solution through the fibers is 1000 ␮m/s and the flow rate through the shell is 150 mL/min. The dotted line marks zero flux.

P. Luis et al. / International Journal of Greenhouse Gas Control 12 (2013) 450–459

455

Fig. 6. Average transmembrane flux as function of the osmotic solution concentration. Feed solution: Na2 CO3 solution of 150 g/L and 200 g/L. The dotted line marks the zero flux.

concentration of NaCl for two Na2 CO3 solutions (150 and 200 g/L). It can be observed that K does not show a dependency of the feed concentration, except in the region between 50 and 100 g/L due to lack of confidence for those values caused by the

180 160

NaCl (g/L)

140

Posive flux

100

80 60

Negave flux

40

4.4. Characterization of crystals

20 0

0

25

50

75

100 125 Na2 CO3 (g/L)

4.3. Influence of flowrates Fig. 9 shows the overall mass transfer coefficient as a function of the velocity through the fibers (feed solution). It can be observed that increasing the velocity of the feed solution has a positive effect on the mass transfer through the membrane. The concentration polarization phenomenon may explain these results. High velocity improves the mixing at the membrane interface and avoids the back flux of water due to an opposite driving force. Variation of the flowrate on the shell side (osmotic solution) was also studied within the range from 50 to 350 mL/min with a Na2 CO3 concentration of 200 g/L and NaCl concentration of 300 g/L, keeping the velocity through the fibers at 1000 ␮m/s. However, a clear influence was not obtained and conclusions cannot be inferred.

200

120

zero flux region. The average value, excluding the 3 outliers, is K = (4.88 ± 0.39) × 10−11 mPa−1 s−1 .

150

175

200

Fig. 7. Areas of negative and positive flux as a function of the concentration in the feed (Na2 CO3 ) and osmotic (NaCl) solutions.

Fig. 8. Overall mass transfer coefficient as a function of the NaCl concentration for two Na2 CO3 solutions. The dotted line marks the average total mass transfer coefficient (calculated without the outliers).

4.4.1. pH of the liquid fraction The average pH of the liquid fraction is 11.4 ± 0.3, which is related to the high remaining concentration of Na2 CO3 . This involves limitations of the discharge of the streams to the environment and an approach based on recirculation and maximum recovery has to be followed.

Fig. 9. Mass transfer coefficient as a function of the velocity through the fibers. Na2 CO3 concentration = 200 g/L; NaCl concentration = 300 g/L; flowrate through the shell = 150 mL/min.

456

P. Luis et al. / International Journal of Greenhouse Gas Control 12 (2013) 450–459

Table 3 The fraction of water in the different crystal samples determined by TWF analysis in an oven at 105 ◦ C over a period of 72 h. Sample

TWF

Na2 CO3 NaHCO3 Na2 CO3 ·10H2 O Average of 15 samples 60 rpm stirring 120 rpm stirring 180 rpm stirring

3.72% 36.58% 52.01% 64.08 ± 1.15% 64.36% 64.15% 64.39%

4.4.2. Total water fraction Table 3 shows the total water fraction (TWF) calculated with Eq. (3) for stock material of Na2 CO3 , NaHCO3 and Na2 CO3 ·10H2 O as a reference. It also shows the average TWF of 15 crystal samples and its standard deviation. This table also shows the different results for 3 velocities of stirring of the feed solution. It is clear that there is no statistical difference between these 3 samples and the average for the 15 samples. It can be observed that there is about 12% more water in the samples than in pure Na2 CO3 ·10H2 O, which indicates that the crystal samples are Na2 CO3 ·10H2 O. Thus, the crystallization process can be improved significantly in order to obtain more pure Na2 CO3 crystals (e.g., Na2 CO3 ·7H2 O or eventually Na2 CO3 ·1H2 O). 4.4.3. Determination of Cl− impurities with IC In order to assess the presence of Cl− ions as impurity, ion chromatography was performed in several crystal samples. Table 4 shows the results for crystals obtained at different conditions. The reference sample was prepared using stock Na2 CO3 . From the results, it can be concluded that some Cl− from the osmotic solution is contaminating the feed stream. The highest concentration of Cl− was obtained when operating with low flowrate in the shell side (osmotic solution) and low concentration of Na2 CO3 (lower driving force). 4.4.4. XRF analysis Table 5 gives an overview of the impurities found with the XRF analysis. The reference material was stock Na2 CO3 . It can be observed that there are already some impurities present in the stock Na2 CO3 ; especially the presence of Mg influences the ionic strength of the solution. In addition, there is a significant amount of

Al (apart from Cl− which was discussed from the Cl− measurements using IC) in the stock solution, which increases after performing the experiment due to the concentration of the solution. Thus, starting from a solution with impurities may be a critical problem to obtain crystals with high purity. Since the feed stream is expected to be produced from the reaction between CO2 and NaOH, it is important to use high purity NaOH in order to avoid such contamination. 4.4.5. Microscopy analysis Fig. 10 shows the micrographs for the crystal sample at 5× and 20× magnifications. In Fig. 10a and b, the formation of rectangular and rounded shape crystals and some aggregates of smaller particles can be observed. Fig. 10c and d allows a better observation of the crystal surface. The black spots (non-transparent zones) are associated to a phase transition due to the water vaporization. Fig. 10d represents the same crystal and surface as Fig. 10c but some minutes later. The black spots grow over time, which proves that changes in the crystal structure take place after the experiments. 4.4.6. DLS measurements of the liquid fraction After vacuum filtration, the liquid fraction was measured by means of DLS in order to examine the presence of small crystals or nuclei. Fig. 11 shows the normalized particle distribution for a filtrated sample obtained from an experiment with Na2 CO3 solution of 200 g/L, NaCl solution of 300 g/L, the velocity through the fibers was 1000 ␮m/s and the flowrate through the shell was 150 mL/min. The average radius is calculated as 1861 nm but the particles present a large polydispersity. 4.5. Exergy analysis An exergy analysis has been made under several experimental conditions of concentrations and flowrate in order to evaluate the system in terms of useful energy that is available. From Eq. (4), the term related to the change of exergy due to variation in temperature is not applicable since the system works at the reference temperature (293.15 K) and the pressure term is negligible since a very small pressure drop is produced in the hollow fiber membrane contactor. Thus, only the variation of exergy due to changes in concentration is significant and the linear fitting obtained for the flux of water through the membrane (Fig. 6) has been considered in the calculations.

Table 4 Determination of Cl− impurities in the obtained crystals. Initial Na2 CO3 concentration (g/L)

Initial NaCl concentration (g/L)

Velocity through the fibers (␮m/s)

Shell flowrate (mL/min)

Cl− concentration (g/kg)

150 200 200 200 200 Reference sample

300 300 300 300 300

1000 600 1000 1300 1000

150 150 350 150 150

4.2903 0.6316 0.4462 1.2242 0.8653 0.3096

Table 5 Impurities in crystals, measured with XRF. Na (wt%) Experiment aa Experiment bb Experiment cc Reference

86.80 83.99 86.32 95.22

± ± ± ±

Mg (wt%) 1.45 1.88 0.16 0.46

9.13 ± 1.29 11.43 ± 1.81 Nd 3.69 ± 0.43

Al (wt%) 1.94 1.70 0.22 0.48

± ± ± ±

0.46 0.55 0.17 0.14

Cl (wt%) 1.00 1.19 1.28 0.336

± ± ± ±

0.24 0.29 0.08 0.075

Nd: not determined. a Experiment a: Na2 CO3 concentration = 200 g/L; NaCl concentration = 300 g/L; velocity through the fibers = 1100 ␮m/s; shell flowrate = 150 mL/min. b Experiment b: Na2 CO3 concentration = 200 g/L; NaCl concentration = 300 g/L; velocity through the fibers = 600 ␮m/s; shell flowrate = 150 mL/min. c Experiment a: Na2 CO3 concentration = 200 g/L; NaCl concentration = 300 g/L; velocity through the fibers = 1000 ␮m/s; shell flowrate = 150 mL/min; with stirring in feed solution.

P. Luis et al. / International Journal of Greenhouse Gas Control 12 (2013) 450–459

457

Fig. 10. Micrographs of crystals: (a) and (b) are obtained at 5× magnification, and (c) and (d) at 20× magnification. The black dots are the result of a phase transition.

Fig. 12 shows the variation of exergy for different concentrations of Na2 CO3 in the feed solution when different concentrations of NaCl are used. It is interesting to highlight the presence of a maximum of exergy, which would indicate the best conditions in which the system is operating from an exergetic point of view. Negative

values of the exergy variation mean that the exergy is destroyed by the components, thus, the working conditions should be those that fit in the positive region. In addition, it can be observed that lower concentrations than 150 g/L of NaCl in the osmotic solution always involve a loss of exergy.

0.2 0.1

0 ∆Ex (KJ.h-1 )

0

100

150

200

250

300

350

-0.2 -0.3 -0.4 -0.5 -0.6

Fig. 11. Normalized particle distribution from DLS measurements in the filtrated sample.

50

-0.1

NaCl 300 g/L NaCl 250 g/L NaCl 200 g/L NaCl 150 g/L NaCl 100 g/L NaCl 50 g/L Na2 CO3 (g.L -1 )

Fig. 12. Variation of exergy as a function of the concentration in the feed (Na2 CO3 ) and the osmotic (NaCl) solution. Feed flowrate = 5.24 mL/min; osmotic solution flowrate = 150 mL/min.

458

P. Luis et al. / International Journal of Greenhouse Gas Control 12 (2013) 450–459

0.12 0.1 0.08

∆Ex (KJ.h-1 )

0.06 0.04 0.02 0 0

5

10

15

20

-0.02 -0.04

NaCl 300 g/L - Na2CO3 175g/L

-0.06

NaCl 250 g/L - Na2CO3 150g/L

-0.08 Flow rate Na2 CO3 (mL .min-1 ) 0.09 0.08 0.07

the feed solution has to be high enough to minimize its resistance to mass transfer. The characterization of the crystals demonstrated that Na2 CO3 ·10H2 O crystals are obtained. Also, some impurities such as Mg, Cl and Al were observed but they are assumed to be introduced by the stock material used to prepare the solutions. Nevertheless, this confirms the importance of the feed composition since it may condition the purity of the crystals. In addition, an exergy analysis of the membrane contactor has been made aiming at studying the effect that the concentration of Na2 CO3 in the feed, the concentration of NaCl in the osmotic solution and the flowrates of both streams causes on the availability of useful energy. An optimum of maximum exergy as a function of the concentrations was observed, which also indicates that the concentration of the osmotic solution must be always higher than 150 g/L in order to avoid the destruction of exergy by the components in the system. In addition, increasing the flowrates reaches an asymptotic value of exergy, which may change at high flowrates due to the effect of the pressure drop in the contactor. Acknowledgements

∆Ex (KJ.h-1 )

0.06

The Research Council of KU Leuven and FWO-Vlaanderen (Research Foundation Flanders) are gratefully acknowledged for providing a postdoctoral grant to P. Luis. Also, P. Luis acknowledges the support by a Marie Curie – CIG Career Integration Grant (PCIG9-GA-2011-294218).

0.05 0.04 0.03 0.02

NaCl 300 g/L- Na2CO3 175g/L

0.01

NaCl 250 g/L - Na2CO3 150g/L

Appendix A.

0 0

10

20 30 Flow rate NaCl (mL .min-1 )

40

50

Fig. 13. Variation of exergy as a function of the flow rates of (a) the feed solution and (b) the osmotic solution.

In order to calculate the water activity required in Eq. (2), a relation between the water activity and the concentration of salts in that solution is needed. This is given by the following equations (Sandler, 2006).



ln(xs s ) − ms vMAB

√ √ ˛ ıI 1 − |z+ z− | I(B I) + 3 2



(A.1)

Fig. 13a and b shows the influence of the flowrate of the feed solution and the osmotic solution, respectively. In both cases, an asymptotic tendency is observed, which determines the minimum flowrate that should be used. However, the flowrate also affects the mass transfer through the membrane due to, for example, the increase of the concentration polarization at low flowrates. Thus, higher flowrates than the minimum given by the exergy analysis may be necessary to avoid a decrease of the transmembrane flux. In addition, the asymptotic tendency is not expected for a very large flow beyond the experimentally studied region since the term of exergy that considers the effect of pressure may have influence and decrease the available energy.

where ms is the amount of kg in one mole of solvent (e.g., 0.018 kg/mol for water), vi is the number of ions coming from the dissociation of the electrolyte and v is the sum of these different ions, MAB is the concentration of the electrolyte, ˛, B and ı are the constants of the model (Tables A1 and A2), z is the valence of the ion after the electrolyte is dissociated and I, the ionic strength is defined as:

5. Conclusions

I=

An experimental study of the operating conditions that affect the performance of membrane crystallization (i.e., concentration in the feed and osmotic solutions, flowrates) has been carried out in order to determine the technical viability of the crystallization of Na2 CO3 using a membrane contactor. Results showed that the concentration of the osmotic solution is a key parameter to achieve a realistic approach based on a high driving force and the flowrate in

and (y) =

3 y3



1 + y − 2 ln(1 + y) −

1 1+y

 (A.2)

 1 1 zi2 Mi = MAB vi zi2 = MAB Q 2 2 ions ions

(A.3)

Table A1 Debye–Hückel coefficients.

NaCl MgCl2 Na2 CO3

B

ı

Ref.

1.0000 1.7309 0.8953

0.1370 0.1196 0.0150

Sandler (2006) Meier (1982) Green and Perry (2008)

Table A2 The first Debye–Hückel coefficient as a function of temperature. Temp. (◦ C)

˛

0

10

20

30

40

50

60

70

80

90

100

1.129

1.146

1.164

1.184

1.206

1.230

1.255

1.283

1.313

1.345

1.379

P. Luis et al. / International Journal of Greenhouse Gas Control 12 (2013) 450–459

The relation between the vapor pressure of water (Pvap ) and the temperature of the solution is empirically given by the following Antoine equation. The result is in mmHg, while the temperature is in ◦ C. P vap (T ) = 108.07131

1780.68 233.426 + T

(A.4)

References Aaron, D., Tsouris, C., 2005. Separation of CO2 from flue gas: a review. Separation Science and Technology 40, 321–348. Azoury, R., Garside, J., Robertson, W.G., 1986. Habit modifiers of oxalate crystals precipitated in a reverse osmosis system. Journal of Crystal Growth 76, 259–262. Cerci, Y., 2002. Exergy analysis of a reverse osmosis desalination plant in California. Desalination 142, 257–266. Charcosset, C., Kieffer, R., Mangin, D., Puel, F., 2010. Coupling between membrane processes and crystallization operations. Industrial and Engineering Chemistry Research 49, 5489–5495. Charcosset, C., 2009. A review of membrane processes and renewable energies for desalination. Desalination 245, 214–231. Criscuoli, A., Drioli, E., 1999. Energetic and exergetic analysis of an integrated membrane desalination system. Desalination 124, 243–249. Criscuoli, A., Drioli, E., 2007. New metrics for evaluating the performance of membrane operations in the logic of process intensification. Industrial and Engineering Chemistry Research 46, 2268–2271. Curcio, E., Di Profio, G., Drioli, E., 2003. Recovery of fumaric acid by membrane crystallization in the production of l-malic acid. Separation and Purification Technology 33, 63–73. Curcio, E., Drioli, E., 2005. Membrane distillation and related operations—a review. Separation and Purification Review 34, 35–86. Curcio, E., Ji, X., Quazi, A.M., Barghi, S., Di Profio, G., Fontananova, E., Macleod, T., Drioli, E., 2010. Hybrid nanofiltration–membrane crystallization system for the treatment of sulfate wastes. Journal of Membrane Science 360, 493–498. Di Profio, G., Curcio, E., Drioli, E., 2010. Supersaturation control and heterogeneous nucleation in membrane crystallizers: facts and perspectives. Industrial and Engineering Chemistry Research 49, 11878–11889. Di Profio, G., Perrone, G., Curcio, E., Cassetta, A., Lamba, D., Drioli, E., 2005. Preparation of enzyme crystals with tunable morphology in membrane crystallizers. Industrial and Engineering Chemistry Research 44, 10005–10012. Di Profio, G., Stabile, C., Caridi, A., Curcio, E., Drioli, E., 2009. Antisolvent membrane crystallization of pharmaceutical compounds. Journal of Pharmaceutical Sciences 98, 4902–4913. Di Profio, G., Tucci, S., Curcio, E., Drioli, E., 2007. Controlling polymorphism with membrane-based crystallizers: application to form I and II of paracetamol. Chemistry of Materials 19, 2386–2388. Drioli, E., Curcio, E., 2007. Perspective membrane engineering for process intensification: a perspective. Journal of Chemical Technology and Biotechnology 82, 223–227. Drioli, E., Curcio, E., Criscuoli, A., Di Profio, G., 2004. Integrated system for recovery of CaCO3 , NaCl and MgSO4 ·7H2 O from nanofiltration retentate. Journal of Membrane Science 239, 27–38. Drioli, E., Curcio, E., Di Profio, G., 2005. State of the art and recent progresses in membrane contactors. Chemical Engineering Research and Design 83 (A3), 223–233. Drioli, E., Curcio, E., Di Profio, G., Macedonio, F., Criscuoli, A., 2006. Integrating membrane contactors technology and pressure-driven membrane operations for seawater desalination. Energy, exergy and costs analysis. Chemical Engineering Research and Design 84, 209–220. Ebner, A.D., Ritter, J.A., 2009. State-of-the-art adsorption and membrane separation, processes for carbon dioxide production from carbon dioxide emitting industries. Separation Science and Technology 44, 1273–1421. El-Naas, M.H., Al-Marzouqi, M., Marzouk, S.A., Abdullatif, N., 2010. Evaluation of the removal of CO2 using membrane contactors: membrane wettability. Journal of Membrane Science 350, 410–416. Figueroa, J.D., Fout, T., Plasynski, S., McIlvried, H., Srivastava, R.D., 2008. Advances in CO2 capture technology—the U.S. Department of Energy’s Carbon Sequestration Program. International Journal of Greenhouse and Gas Control 2, 9–20. Green, D.W., Perry, R.H., 2008. Perry’s Chemical Engineers’ Handbook, 8th ed. McGraw Hill.

459

Ji, X., Curcio, E., Al Obaidani, S., Di Profio, G., Fontananova, E., Drioli, E., 2010. Membrane distillation–crystallization of seawater reverse osmosis brines. Separation and Purification Technology 71, 76–82. Kieffer, R., Mangin, D., Puel, F., Charcosset, C., 2009a. Precipitation of barium sulphate in a hollow fibers membrane contactor. Part I. Investigation of fouling. Chemical Engineering Science 64, 1759–1767. Kieffer, R., Mangin, D., Puel, F., Charcosset, C., 2009b. Precipitation of barium sulphate in a hollow fibers membrane contactor. Part II. Influence of process parameters. Chemical Engineering Science 64, 1885–1891. Kim, D.H., 2011. A review of desalting process techniques and economic analysis of the recovery of salts from retentates. Desalination 270, 1–8. Kuhn, J., Lakerveld, R., Kramer, H.J.M., Grievink, J., Jansens, P.J., 2009. Characterization and dynamic optimization of membrane-assisted crystallization of adipic acid. Industrial and Engineering Chemistry Research 48, 5360–5369. Kuramochi, T., Ramírez, A., Turkenburg, W., Faaij, A., 2012. Comparative assessment of CO2 capture technologies for carbon-intensive industrial processes. Progress in Energy and Combustion Science 38, 87–112. Luis, P., Van der Bruggen, B., Van Gerven, T., 2011. Non-dispersive absorption for CO2 capture: from the lab to the industry. Journal of Chemical Technology and Biotechnology 86, 769–775. Luis, P., Van Gerven, T., Van der Bruggen, B., 2012. Recent developments in membrane-based technologies for CO2 capture. Progress in Energy and Combustion Science 38, 419–448. Macedonio, F., Curcio, E., Drioli, E., 2007. Integrated membrane systems for water desalination: energetic and exergetic analysis, economic evaluation, experimental study. Desalination 203, 260–276. Mahmoudkhani, M., Keith, D.W., 2009. Low-energy sodium hydroxide recovery for CO2 capture from atmospheric air—thermodynamic analysis. International Journal of Greenhouse and Gas Control 3, 376–384. Mansourizadeh, A., Ismail, A.F., Matsuura, T., 2010. Effect of operating conditions on the physical and chemical CO2 absorption through the PVDF hollow fiber membrane contactor. Journal of Membrane Science 353, 192–200. Meier, P.C., 1982. Two-parameter Debye–Hückel approximation for the evaluation of mean activity coefficients of 109 electrolytes. Analytica Chimica Acta 136, 363–368. Merkel, T.C., Lin, H., Wei, X., Baker, R., 2010. Power plant post-combustion carbon dioxide capture: an opportunity for membranes. Journal of Membrane Science 359, 126–139. Molinari, R., Gagliardi, R., Drioli, E., 1995. Methodology for estimating saving of primary energy with membrane operations in industrial processes. Desalination 100, 125–137. Olajire, A.A., 2010. CO2 capture and separation technologies for end-of-pipe applications: a review. Energy 35, 2610–2628. Rodríguez-DeLaNuez, F., Franquiz-Suárez, N., Santiago, D.E., Veza, J.M., Sadhwani, J.J., 2012. Reuse and minimization of desalination brines: a review of alternatives. Desalination and Water Treatment 39, 137–148. Sandler, S.I., 2006. Chemical, Biochemical, and Engineering Thermodynamics, 4th ed. John Wiley & Sons, USA. Tang, B., Yu, G., Fang, J., Shi, T., 2010. Recovery of high-purity silver directly from dilute effluents by an emulsion liquid membrane–crystallization process. Journal of Hazardous Materials 177, 377–383. Van der Bruggen, B., Braeken, L., 2006. The challenge of zero discharge: from water balance to regeneration. Desalination 188, 177–183. Van der Bruggen, B., Lejon, L., Vandecasteele, C., 2003. Reuse, treatment, and discharge of the concentrate of pressure-driven membrane processes. Environment Science and Technology 37, 3733–3738. Van Gerven, T., Stankiewicz, A., 2009. Structure, energy, synergy, times the fundamentals of process intensification. Industrial and Engineering Chemistry Research 48, 2465–2474. Watanabe, J., Akashi, M., 2009. Formation of various polymorphs of calcium carbonate on porous membrane by electrochemical approach. Journal of Crystal Growth 311, 3697–3701. Zarkadas, D.M., Sirkar, K.K., 2006. Antisolvent crystallization in porous hollow fiber devices. Chemical Engineering Science 61, 5030–5048. Zhang, W., Li, J., Chen, G., You, W., Jiang, Y., Sun, W., 2010. Experimental study of mass transfer in membrane absorption process using, membranes with different porosities. Industrial and Engineering Chemistry Research 49, 6641–6648. Zhang, X., Zhang, P., Wei, K., Wang, Y., Ma, R., 2008. The study of continuous membrane crystallization on lysozyme. Desalination 219, 101–117.