Environ. Sci. Technol. 2002, 36, 3010-3019
Electrosorption of Inorganic Salts from Aqueous Solution Using Carbon Aerogels C H R I S T O P H E R J . G A B E L I C H , * ,† TRI D. TRAN,‡ AND I. H. “MEL” SUFFET§ Metropolitan Water District of Southern California, La Verne, California 91750, Lawrence Livermore National Laboratory, Livermore, California 94550, and University of California, Los Angeles, California 90624
Capacitive deionization (CDI) with carbon aerogels has been shown to remove various inorganic species from aqueous solutions, though no studies have shown the electrosorption behavior of multisolute systems in which ions compete for limited surface area. Several experiments were conducted to determine the ion removal capacity and selectivity of carbon aerogel electrodes, using both laboratory and natural waters. Although carbon aerogel electrodes have been treated as electrical double-layer capacitors, this study showed that ion sorption followed a Langmuir isotherm, indicating monolayer adsorption. The sorption capacity of carbon aerogel electrodes was approximately 1.0-2.0 × 10-4 equiv/g aerogel, with ion selectivity being based on ionic hydrated radius. Monovalent ions (e.g., sodium) with smaller hydrated radii were preferentially removed from solution over multivalent ions (e.g., calcium) on a percent or molar basis. Because of the relatively small average pore size (4-9 nm) of the carbon aerogel material, only 14-42 m2/g aerogel surface area was available for ion sorption. Natural organic matter may foul the aerogel surface and limit CDI effectiveness in treating natural waters.
Introduction Capacitive deionization (CDI) with carbon aerogel electrodes is a novel technology for removing ionic species from aqueous solutions. This electrochemical process is conducted at ambient conditions and low voltages (e.g., 1 V) and requires no high-pressure pumps, membranes, distillation columns, or thermal heaters. CDI could be developed into a method for desalination, as CDI removes ions by charge separation and therefore may avoid the scaling problems commonly associated with membrane and distillation processes. Also, this approach could offer an attractive, energy-efficient alternative to thermal and membrane desalination processes. The most likely use of carbon aerogel CDI would be as an alternative to reverse osmosis for the initial salinity-reduction step with mildly brackish waters. CDI with high-surface-area carbon electrodes has been considered for desalination since the 1960s (1-5). These earlier attempts used activated carbon powder or fibers as the ion sorption media. Performance issues associated with * Corresponding author phone: (909)392-5113; e-mail: cgabelich@ mwdh2o.com. † Metropolitan Water District of Southern California. ‡ Lawrence Livermore National Laboratory. §University of California, Los Angeles. 3010
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 13, 2002
FIGURE 1. Carbon aerogel microstructure: transmission electron micrograph at 500 000 x magnification. electrodes consisting of activated carbon powders or fibers included large pressure drops, low electrical conductivity, binder degradation, and low overall capacity. The CDI approach in this study, because of the specific properties of the high-surface-area carbon aerogels used, can potentially overcome these limitations. The manufacturing process for carbon aerogels has been described elsewhere (6-8). Carbon aerogels are derived from the pyrolysis of resorcinol-formaldehyde polymeric gels. Depending on their application, carbon aerogels can be prepared in different forms, such as monolithic blocks, beads, or thin sheets. Thin sheets of carbon paper impregnated with carbon aerogel material are useful for electrochemical applications such as CDI. Carbon aerogels are unique, porous materials consisting of interconnected, uniform carbonaceous particles (3-30 nanometers [nm]) with small ( Mg2+ > K+ > Na+ For anions of interest, the sorption of solutes should obey the following order:
SO42- > NO3- > Br- > ClThese relationships typically hold for waters with less than 1000 mg/L total dissolved solids (TDS) (18). However, initial laboratory testing using carbon aerogel electrodes showed that these relationships did not hold for carbon aerogel sorption. In the early 1970s, electrosorption studies using porous carbon electrodes demonstrated preferential sorption of divalent ions from complex waters (2, 3). Investigators attributed the preferential sorption of divalent ions to the fact that a chemical equilibrium was achieved in the electric double layer at the electrode surface. Trainham and Newman (19) studied metal-ion removal using a porous carbon electrode. However, this work concentrated on metal species (e.g., Cu and Hg) that tend to plate (electrochemically bond), rather than sorb, onto the electrode surface. Previous research with CDI using carbon aerogels demonstrated differing breakthrough curves for 100 µS/cm NaCl
and NaNO3 solutions (9). These data suggest that ion selectivity based on ionic size and degree of complexation may have been occurring. Though chloride and nitrate anions have the same charge, their atomic weight difference (35 and 62 amu, respectively) may play a role in terms of effective charge affinity for the electrode surface. Additional studies with CDI have demonstrated the removal of various inorganic cations and anions common to potable water supplies as well as radionuclides and heavy metals (1, 10, 11, 20). No recent studies have demonstrated the effects of ion charge and size on sorption to carbon-based electrochemical desalination devices. To gain a better understanding of carbon aerogel-based CDI’s potential as a practical desalination technology, the adsorption phenomena associated with these processes need to be addressed. Representative ions of varying ionic radius, charge, and mass were used to understand the electrosorption dynamics of ideal and complex aqueous solutions. Additionally, experiments were expanded to include more practical solutions: (1) an unfiltered blend of 75% Colorado River water with 25% water from the California State Water Project and (2) blended water augmented with 14 mg/L artificial natural organic matter (NOM) (7 mg/L total organic carbon [TOC]). These two streams closely resemble typical potable water that is currently available for municipal use.
Experimental Methods Three bench-scale CDI units, requiring only water and power hookups (Figure 3), were manually operated. A typical CDI stack consisted of two cells attached in series (in terms of solution flow). Each cell consists of two parallel carbon aerogel electrode plates separated by a 1.52 mm gap for solution flow. The carbon aerogel papers, 10 cm wide × 20 cm long × 0.015 cm thick (Ocellus, Inc., Alameda, CA), were mounted to titanium supports with four 0.5-cm conductive epoxy strips (Magnabond 3386; Magnolia Plastics, Inc., Chamblee, GA). Three carbon aerogel composites were tested, using various carbon aerogel pore size distributions and paper thicknesses (Table 1). Details of their preparation and electrochemical characterization have been presented elsewhere (21). The Brunauer-Emmet-Teller (BET) surface area and pore size distribution were obtained using a five-point N2 gas adsorption technique (ASAP 2000; Micromeritics, Norcross, GA). The average pore size and pore size distribution were determined from the desorption branch according to a theory developed by Barrett, Joyner, and Halenda (22). Each carbon aerogel sheet weighed 2 g, with a total active surface area of approximately 8.16 × 106 cm2. The composite morphology and microstructure was examined by transmission electron VOL. 36, NO. 13, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3011
TABLE 1. Capacitive Deionization Test Cell Data CDI unit
aerogel compositea
BET (m2/g)
mean pore size (nm)
aerogel sheet dimensions (cm)
total mass of aerogel (g)
#1 #2 #3
R-F aerogel pyrolyzed with N2 at 1050 °C R-F aerogel pyrolyzed with N2 at 800 °C R-F aerogel pyrolyzed with N2 at 800 °C
400 590 590
4 9 9
10 × 20 × 0.015 10 × 20 × 0.015 10 × 20 × 0.030
8 8 16
a
R-F aerogelsresorcinol-formaldehyde aerogel (11).
TABLE 2. Statistical Design Matrices To Evaluate the Effect of Ion Properties on Carbon Aerogel Sorption a. effect of charge on sorption
b. effect of mass on sorption
c. effect of ionic radius on sorption
test
cation
anion
test
cation
anion
test
cation
1 2 3 4 5
Na+ Na+ Mg2+ Mg2+ Na+, Mg2+
BrSO42BrSO42Br-, SO42-
6 7 8 9 10
K+ K+ Rb+ Rb+ K+, Rb+
ClBrClBrCl-, Br-
11 12 13 14 15
Na+ Na+ K+ K+ Na+, K+
anion BrNO3BrNO3Br-, NO3-
TABLE 3. Chemical Data of Test Ions ion
massa (amu)
charge
ionic radiusb (pm)
hydrated radiusc (pm)
sodium potassium rubidium magnesium chloride bromide nitrate sulfate
22.990 39.098 85.468 24.305 35.453 79.904 62.005 96.064
+1 +1 +1 +2 -1 -1 -1 -2
116 152 166 86 167 182 165 244
358 331 329 428 331 330 335 379
a
FIGURE 3. Capacitive deionization flow schematic for single-pass mode. microscopy (JEM-200CX; JEOL USA, Inc., Peabody, MA). Aqueous solution was pumped in by a peristaltic pump (model 185; Micropump, Vancouver, WA) from the bottom and exited from the top of the stack. The maximum voltage used in these experiments was 1.6 V with an applied flow rate of 100 mL/min. Supporting equipment included a flowmeter (model GF-2560; Gilmont Instruments, Barrington, IL), online conductivity meters (SC-170; Suntek, Taiwan) at both inlet and outlet, a multimeter (GDT-295A; GB Instruments, Milwaukee, WI), and a power supply (Instek GPR 18-10HD; Taipei Hsien, Taiwan). Applied voltage, amperage, and effluent conductivity were recorded via a waveform browser (WINDAQ 2000; Dataq Instruments, Akron, OH) onto a personal computer (LTE ELITE 4/75CX; Compaq, Houston, TX). Regeneration of the CDI unit was conducted in closedloop operation and began immediately after the sorption phase. The original test water was repumped back through the shorted-out CDI unit. Pumping was initiated at the same flow rate as the sorption stage. When the conductivity reached an asymptotic level, or after approximately 20 min of regeneration time, the pump was shut off. The final conductivity, temperature, and pH values were recorded. After each experiment, the CDI unit was flushed with deionized water. After the unit was cleaned, the polarity to the CDI unit was reversed in order to extend the life of the carbon aerogels. 3012
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 13, 2002
Reference 23.
b
Reference 24. c Reference 25.
The final conductivity, temperature, and pH were recorded before and after the sorption and regeneration phases. A series of 15 experiments was conducted to evaluate the effects of charge, mass, and ionic radius on the electrosorption efficiency of carbon aerogel electrodes. Ions selected for evaluation were sodium (Na+), magnesium (Mg2+), potassium (K+), rubidium (Rb+), chloride (Cl-), bromide (Br-), nitrate (NO3-), and sulfate (SO42-). Table 2 shows the design test matrix. Physical-chemical data for each of the above ions are shown in Table 3. All experiments were conducted using 1 L of 0.005 M salt solutions made up in deionized water. Reagents for these experiments included the following: certified ACS-grade potassium chloride (J. T. Baker Chemical Co., Phillipsburg, NJ); potassium bromide (Spectrum Quality Products, Inc., New Brunswick, NJ); rubidium chloride (Spectrum); rubidium bromide (Spectrum); magnesium sulfate (Spectrum); magnesium bromide (Aldrich Chemicals, Milwaukee, WI); sodium bromide (J. T. Baker); sodium nitrate (J. T. Baker); and potassium nitrate (J. T. Baker). Stock solutions were prepared daily. Additional experiments were conducted with more complex streams, using 1-L samples of raw water that consisted of 75% Colorado River water and 25% California state project water with and without a small amount of organic contaminant (i.e., NOM). These experiments were conducted to determine the effects of NOM on CDI performance as well as to help validate data collected during the electrochemical sorption experiments. Experiments were run in triplicate, with and without 14 mg/L supplemental NOM. The NOM was isolated in Norway (Hellerudmyra, Oct-96, and Hellerudmyra, May-96) by reverse osmosis and evaporative techniques according to Christy, Bruchet, and Rybacki (26). To develop an adsorption isotherm for carbon aerogels, the CDI unit was reconfigured from the standard two-cell
FIGURE 4. Multiple sorption/regeneration cycles for closed-loop operation. Data collected in a closed-loop configuration using 1 L, 0.004 M NaCl solutions at 1.0 V, 100 mL/min flow rate, and ambient pH (∼pH 7). stack into six-cell and 14-cell stacks. This increased the total amount of aerogel material by 3- and 7-fold, respectively. These experiments were conducted using three replicates of 1 L, 0.004 mol/L NaCl solutions at 1.4 V, 100 mL/min flow rate, and ambient pH (∼pH 7).
Results and Discussion Experiments involving standard deionization conditions were initially conducted to assess the system’s operational stability and reproducibility. For ease of operation and data collection, a series of 14 experiments was conducted using a two-cell stack in a closed-loop configuration. Successive deionization and regeneration were completed on 1 L of deionized water with 0.004 mol/L NaCl (250 mg/L). Other operating parameters included pH 7, 1.0 V, and 100 mL/min. Both sorption and regeneration phases were limited to 10 min in duration, with the electrode applied voltage being reversed between successive cycles. This alternating polarity approach was demonstrated by Tran and Lenz (patent pending (27)) to enhance overall sorption capacity and minimize electrode performance degradation. Figure 4 shows the ratio of outlet conductivity to inlet conductivity over 12 consecutive sorption/regeneration cycles. The unsymmetrical nature of the electrosorption (deionization) and desorption (regeneration) curves is characteristic of the different kinetics associated with ion uptake onto, and removal from, the carbon surface. Although this work did not focus on studying these rates, it did identify favorable experimental parameters for maximizing the system’s efficiency. Between each sorption/regeneration cycle, the conductivity reading returned within 2.6% of the previous value after only 10 min. Error associated with measuring solution conductivity is expected to be in the same range. No appreciable decline in carbon aerogel performance was observed after 14 experiments. Adsorption Isotherm. Langmuir, Freundlich, and BET isotherms were graphed using data from the two-, five-, and 14-cell CDI experiments. Neither the Freundlich nor the BET isotherms accurately modeled the data. The charge felt by the ions in solution is distributed over a three-dimensional, two-layer region in a widely accepted double-layer theory (11, 13, 28). The electric double-layer theory posits that for weakly charged surfaces (like CDI), some ions may adsorb (stick) to the surface in a closely compacted layer (Stern layer),
FIGURE 5. Langmuir adsorption isotherm for carbon aerogel sorption: Ce ) concentration of final solution at equilibrium (g/L), x/m ) mass of solute sorbed (g) per mass of aerogel (g). Data collected using 1 L, 0.004 M NaCl solutions at 1.4 V, 100 mL/min flow rate, and ambient pH (∼ pH 7). but many ionic species remained throughout the diffuse (approximately 100-200 Å) layer (Gouy-Chapman layer). However, the Langmuir isotherm best described the data (R2 ) 0.90), suggesting that for modeling purposes, monolayer coverage of the carbon aerogel surface area may be assumed (Figure 5). Researchers using activated carbon cloth electrodes found that sorption of mercury(II) acetate was Langmuirian as well (29). Effects of Ion Properties on Electrosorption Capacity. The effects of ion charge, size, and mass on electrosorption capacity were investigated in several series of single-solute experiments containing one or more representative ions of varying properties. Cations (Ca2+, Na+, Mg2+, and Rb+) and anions (Cl-, Br-, NO3-, and SO42-) were selected based on their presence in natural water, their availability, or their ease of use as reagents. Each of the test matrices in Table 2 was designed to isolate a test variable (e.g., ion charge) to the greatest extent possible. For example, to evaluate the effects of charge on sorption potential, test ions were selected that had similar atomic mass and ionic radius. It should be noted, however, that some of ions used, and their respective concentrations in these matrices, are not representative of those found in natural surface waters (e.g., rubidium and bromide). Deionization experiments were conducted with 0.005 M solution in a single-pass configuration using a twoVOL. 36, NO. 13, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3013
FIGURE 6. Effect of ion charge on carbon aerogel performance. All experiments were conducted using 1 L of 0.005 M salt solutions made up in deionized water, 1.4 V, 100 mL/min, and ambient pH (∼pH 7).
FIGURE 7. Carbon aerogel breakthrough curves for charge experiments. Data indicate that saturation of the carbon aerogel surface occurred. Water quality samples taken at the end of each experiment. cell CDI stack. The experiments were run for 10 min at 1.4 V until saturation, as indicated by the constant effluent conductivity measurements. It is of interest that the stack capacity remained constant after these 36 experiments. There was no poisoning, scaling, or electrode degradation under these operating conditions. Molar balances for each solution, before and after CDI experiments, were greater than 95%. The total moles of ions removed from solution were determined from the difference between the ion concentrations before and after the deionization step. The maximum (equilibrium) ion electrosorption capacity was achieved in these experiments based on constant effluent conductivity at the end of each experimental run. The effects of ion charge (valency) in terms of mol/g aerogel electrode are shown in Figure 6. Here, variance on both cations (Na+ and Mg2+) and anions (Br- and SO42-) was elucidated in four series of experiments (tests 1-4, Table 2). Figure 7 shows the breakthrough curves for four experiments, each containing solutes from each of the ions above. Complete saturation of the carbon aerogels was achieved in each experiment. The capacities for monovalent ions (Na+ and Br-) were determined to be approximately 1.6 × 10-4 mol/g aerogel. The spread in the Br- data, however, was large (between 1.3 and 1.8 × 10-4 mol/g aerogel). The 3014
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 13, 2002
capacities for divalent species with a monovalent counterion (series 2 and 3, Table 2 and Figure 6) were 5.1 × 10-5 mol/g. Interestingly, with both divalent ions, capacities appeared to increase 2-fold (series 4). The molar stoichiometry in three out of four series of experiments reflects electrical neutrality in the electrolytes. Experiments involving Na2SO4, on the other hand, showed an excess of Na+ being removed (or a larger-than-expected depletion in Na+ in solution phase, based on SO42- analysis). At this point, the authors do not understand this discrepancy. The normalized equivalent capacity data (mol × valence/g aerogel) show that divalent ions (SO42- and Mg2+) have a considerably lower capacity than that of single-valent species (Na+, Br-) when one or more ionic species is monovalent (series 1-3). However, when both ions are divalent (as in series 4), the capacities appear to be enhanced. The results suggest that other properties of the dissolved ions, such as ionic radius and charge, may be affecting the electrosorption capacity. The effects of ion mass were investigated in several series of experiments (tests 6-9, Table 2, and Figure 8). Monovalent cations (K+ and Rb+) and anions (Br- and Cl-) with large differences in atomic mass were selected. The saturated capacities were determined to be about 1.5 to 1.8 × 10-4 mol/g aerogel for all ions in solution. The large scatter in Rb+ data was attributed to analytical error. This value is consistent with that obtained for the other single-valent ion pair (NaBr) in the charge effect study above. The effects of ionic radius are illustrated in Figure 9 (tests 11-14, Table 2). Monovalent cations (Na+ and K+) and anions (Br- and NO3-) with large ion radius variance were investigated in single-solute deionization experiments. The average capacities for all ions were determined to be about 1.5 to 2.0 × 10-4 mol/g aerogel. Again, this value is the same with other single-valent ion pairs studied elsewhere in this work. The data for the three series of experiments clearly show that ion charge had the single greatest effect on electrosorption behavior. The ion mass and radius had no observable effect under these experimental conditions. There was evidence that the counterion charge also played an important role in an individual ion’s sorption capacity. The hydrated radius of an ion is a function of charge and ionic radius (i.e., charge-to-radius ratio). Therefore, as with ion the exchange and membrane separation processes, ion hydrated radius
FIGURE 8. Effect of ion mass on carbon aerogel performance. All experiments were conducted using 1 L of 0.005 M salt solutions made up in deionized water, 1.4 V, 100 mL/min, and ambient pH (∼pH 7).
FIGURE 9. Effect of ionic radius on carbon aerogel performance. All experiments were conducted using 1 L of 0.005 M salt solutions made up in deionized water, 1.4 V, 100 mL/min, and ambient pH (∼pH 7). dictated removal phenomena. However, unlike ion exchange or reverse osmosis, CDI with carbon aerogels showed better removal characteristic of smaller, monovalent ions as opposed to larger, divalent ions. Saturation of Carbon Aerogel Surface. To better understand the mechanism governing carbon aerogel sorption, the saturated equilibrium adsorption capacities were calculated for each compound based on the amount of aerogel surface area covered by the sorbed species, using both the ionic area and hydraulic area of each ion (Table 4). Complete saturation of the carbon aerogels was assumed for each experimental condition. Table 4 presents the carbon aerogel surface area coverage using both ionic area and hydrated area. For example, the molar sorption capacity for sodium bromide (1.73 × 10-4 mol Na/g aerogel) was multiplied by Avogadro’s number (6.022 × 1023) and the two-dimensional area based on the ionic radius (4.2 × 10-20 m2 for Na) to generate the surface
area covered by sodium ions (4.4 m2/g aerogel). These data show that approximately 0.9-7.6 m2/g aerogel surface was covered by cations and 5.9-11.9 m2/g aerogel surface was covered by anions. Using ionic area, the anion sorption capacity was significantly greater than for cations (R ) 0.05). When the hydrated radius was used to calculate the amount of surface area occupied by each ion, the surface coverage greatly increased (Table 4). As the “hydration sphere radius” increased, the planar surface area increased by the square of the radius. For example, the saturation adsorption capacity for sodium from the sodium bromide experiments increased from 4.4 to 42 m2/g aerogel. Based on these data, the surface coverage of the carbon aerogel surface ranged between 14 and 42 m2/g aerogel. When hydrated-ion surface coverage was used, the cation and anion populations were statistically equivalent (R ) 0.05). Tests using various ion pairs and multiple solutes (tests 5, 10, and 15; Table 2) gave similar results; hydrated-ion VOL. 36, NO. 13, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3015
TABLE 4. Effect of Ion Hydration on Sorption Capacity of Carbon Aerogelsa ionic area hydrated area (m2/g aerogel) (m2/g aerogel) variable a. charge
b. ionic radius
c. mass
test no.
solute
1 2 3 4 6 7 8 9 11 12 13 14
NaBr Na2SO4 MgBr2 MgSO4 NaBr NaNO3 KBr KNO3 KCl KBr RbCl RbBr
cation anion cation anion 4.4 4.1 0.9 1.4 4.4 3.7 7.0 7.4 7.0 7.0 7.6 7.4
11.7 5.9 8.5 10.7 11.7 8.4 11.2 8.8 8.5 11.2 9.0 11.9
42 39 22 34 42 35 33 35 33 33 30 29
39 14 28 26 39 35 37 36 33 37 35 39
a All data are average values. All experiments conducted at 1.4 V, 100 mL/min, and ambient pH (∼pH 7).
TABLE 5. Effect of Multiple Solutes on Ion Sorption Capacity to Carbon Aerogelsa ionic area (m2/g aerogel) variable
solutes
cation
a. charge NaBr, MgSO4 Na 1.7 Mg 0.7 total 2.4 b. ionic NaBr, KNO3 Na 1.7 radius K 4.2 total 5.9 c. ionic KCl, RbBr K 3.4 mass Rb 4.2 total 7.6
hydrated area (m2/g aerogel)
anion
cation
anion
Br 4.7 SO4 4.4 total 9.1 Br 5.1 NO3 4.0 total 9.1 Cl 4.3 Br 4.6 total 8.9
Na 16 Mg 17 total 34 Na 16 K 20 total 36 K 16 Rb 16 total 33
Br 16 SO4 12 total 28 Br 17 NO3 16 total 33 Cl 17 Br 15 total 32
a All data are average values. All experiments conducted at 1.4 V, 100 mL/min, and ambient pH (∼pH 7).
surface coverage ranged from 28 to 36 m2/g aerogel (Table 4). Therefore, although there was equal surface area coverage for sodium and magnesium, this resulted in significantly greater sodium removal on a molar basis because of sodium’s smaller hydrated size. It should be noted that the hydrated radius data for sulfate is consistently less than that for other ions (e.g., 12-26 m2/g aerogel for sulfate compared to 35-42 m2/g aerogel for sodium). This discrepancy is believed to be attributable to the inaccuracy of the hydrated radius reported for sulfate. Cation Versus Anion Sorption. Previous data indicated that the electrosorption capacity of carbon-based electrodes is limited by their poor ability to accommodate anions (1). This finding was attributed to the cation affinity of carbonyl groups on the electrode surface. However, when the hydrated radius is used to calculate the saturation adsorption capacity (Table 5), no apparent difference between sorption of cations and anions was observed (approximately 33 m2/g aerogel for both cations and anions) (tests 5, 10, and 15; Table 2). In addition, on the basis of equivalent/g aerogel, the data showed slightly lower sorption capacities than those demonstrated for activated carbon electrodes (1.6 × 10-4 versus 2.5 × 10-4 equivalents/g aerogel) (9). However, difficulties associated with other carbon-based electrodes, such as poor conductivity, binder degradation, and decreasing sorption capacity, were not observed when carbon aerogel electrodes were used. Utilization of Carbon Aerogel Porosity. Using either ionic radius or hydrated radius data, the surface coverage of the carbon aerogel surface is very low considering the estimated 3016
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 13, 2002
FIGURE 10. Pore distribution of carbon aerogels pyrolyzed with N2 at 800 °C and 1050 °C. surface area as determined by surface area analysis (BET surface area of 400 m2/g aerogel, or Langmuir surface area of 540 m2/g aerogel). Given the microporous nature of the carbon aerogel structure (median pore size 4 nm), the Langmuir surface area may be more appropriate for determining the actual surface area. However, both of these surface areas may be misleading in terms of actual surface area available for ion sorption. Both BET and Langmuir surface areas were calculated using pressurized nitrogen. Given that the bond length for nitrogen is approximately 0.1 nm, the surface area was calculated using a high level of porosity. This level of porosity probably does not allow for the effective mass transfer of solutes to the inner aerogel structure. Assuming an electric double-layer formationswhich previous researchers with carbon aerogel have assumeds the electric double-layer thicknesses at 25 °C for 10-1 and 10-4 M electrolytes have been shown to be 1 and 20 nm, respectively (9). Therefore, the actual surface area available for sorption may be much lower than that indicated by BET analysis. Recent data have indicated that the double layer at the carbon-electrolyte interface is primarily formed in the mesopore region (greater than 3 nm pore size) (11). If singlelayer (i.e., Langmuir) sorption is assumed (as shown in this study), not only does the measured aerogel surface area increase (540 m2/g aerogel), but so does the level of porosity available to sorption. However, the same mass transfer limitations would still be in effect if Langmuirian sorption behavior were assumed. Effects of Carbon Aerogel Preparation on Deionization Capacity. New types of carbon aerogels have been reported that have higher surface areas and larger average pore sizes (21, 30). Pyrolysis temperature during the carbonization of the aerogel was reported to produce carbon aerogels with higher surface area and larger pore size. Thicker electrode sheets have also been recently prepared. These kinds of electrode materials are expected to have higher capacities for electrosorption. These two types of materials were reproduced and tested in several deionization experiments to study their preparative effects on sorption capacities. Figure 10 compares the pore size distribution of two types of resorcinol-formaldehyde carbon aerogels prepared at two different pyrolysis temperatures. Pyrolyzing the carbon aerogel material with N2 at lower temperatures (800 °C) increased the overall BET surface area (590 m2/g aerogel) and slightly increased the average pore size (9 nm). Table 6 shows the results from the carbon aerogel composite experiments. As reported earlier, each of the CDI units tested exhibited increased salt removal with increased applied voltage. At 1.4 V, the carbon aerogel pyrolyzed with N2 at 800 °C (CDI unit #2) exhibited 26% higher sorption potential when compared to the carbon aerogels pyrolyzed with N2 at 1050 °C (CDI unit #1). In addition, doubling the thickness of the carbon aerogel (CDI unit #3) did not increase the sorption capacity of the carbon aerogel material. In fact, the thicker carbon aerogels pyrolized at 800 °C exhibited
FIGURE 11. Capacitive deionization TOC removal from Colorado River water. All experiments were conducted using 1.4 V, 100 mL/min, and pH 8.3.
TABLE 6. Experimental Results from Various Carbon Aerogelsa removal conductivity (µs/cm) pH CDI voltage (10-4 mol unit (V) initial final regenerate initial final NaCl/g C)b #1 #2 #3
1.0 1.4 2.0 1.0 1.4 2.0 1.0 1.4 2.0
536 536 539 530 530 538 534 569 540
455 434 403 477 457 455 443 444 396
501 506 525 507 506 518 495 523 498
6.2 6.0 6.0 6.6 6.5 6.6 6.6 6.7 6.5
7.5 8.3 4.9 5.0 5.6 4.6 7.0 7.7 4.0
1.7 10.1 13.3 5.3 7.5 8.0 4.6 6.0 7.2
a All experiments conducted at 1.4 V, 100 mL/min, and ambient pH (∼pH 7). Note: Flow rate was 100 mL/min. b As measured by conductivity.
20% lower sorption capacity than the thinner carbon aerogels pyrolyzed at 800 °C. This finding may be attributed to the high density and tortuosity of the carbon aerogel structure, which prevented the ions from penetrating past the exposed surface area of the carbon aerogel. Therefore, thicker carbon aerogelssexcept for the additional structural support they providesmay not be advantageous, as sorption capacity is
not increased. In light of the relatively high cost of carbon aerogel, the use of a thicker material may not be cost-effective. Additional experiments are needed to reproduce these results and to further study these various effects. The thickness of an electroactive layer, for example, is critical in this current flow-by design, in which ion transfer to the interior electrode surface may be rate-limiting under practical deionization conditions. Using the hydrated ion surface coverage data for sodium, approximately 40 m2/g carbon aerogel, pyrolyzed at 1050 °C, is actively involved with mass transfer. Interpolating between the pore size distribution data (Figure 10), a 40 m2/g surface area coverage corresponds to sorption in the pore size range above 20 nm. With a mean pore size of 4 nm, more than 90% of the carbon aerogel surface area was unavailable for ion sorption. Other researchers working with carbon aerogels showed that pyrolyzing the carbon aerogel material with nitrogen at 800 °C produced the highest electrical capacitances: experiments conducted using 1.0 V and 5 M KOH solution (30). The higher electrical capacitance coupled with the larger pore volume distribution may have led to higher overall ion sorption capacities for the carbon aerogels pyrolized at 800 °C. Effects of NOM on Inorganic Salt Removal. Table 7 shows the results from the tests conducted with natural waters,
TABLE 7. Sorption Capacity of Carbon Aerogel Electrodes Using Colorado River Watera Colorado River water with 14 mg/L NOM (9.8 mg/L total TOC)
Colorado River water (2.6 mg/L total TOC)
parameter calcium magnesium potassium sodium sulfate chloride nitrate fluoride silica total cations total anions a
sorption capacity
hydrated area
sorption capacity
hydrated area
(× 10-5 mol/g aerogel) 2.91 2.23 0.38 9.43 3.50 9.09 0.01 0.06 0.01 14.9 13.4
(m2/g aerogel) 9.3 7.7 0.8 23 9.5 19 0.02 0.14 0.02 41 29
(× 10-5 mol/g aerogel) 1.88 2.06 0.34 7.98 2.59 7.38 0.01 0.01 0.02 12.3 10.0
(m2/g aerogel) 6.0 7.1 0.7 19 7.0 15 0.02 0.02 0.03 33 22
All experiments conducted at 1.4 V, 100 mL/min, and pH 8.3. All data are average values.
VOL. 36, NO. 13, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3017
FIGURE 12. Carbon aerogel performance on natural water with 2.6 mg/L TOC. Mean water quality data (n ) 3) taken using 1.4 V, 100 mL/min, and pH 8.3. with and without supplemental NOM. The data are presented in terms of molar sorption capacity and hydrated-ion surface coverage (see previous discussion). Based on a paired t-test (R ) 0.05), the sorption capacity, in terms of these variables, was significantly lower when a high concentration of NOM was present. In addition to individual ions measured, the total cation and anion sorption capacities were also significantly lower for Colorado River water with 9.8 mg/L TOC present. Figure 11 shows the TOC removal and recovery from the CDI unit. When lower levels of TOC (2.3 mg/L TOC) were present, better NOM recovery was observed (92% versus 30% for 9.8 mg/L TOC samples). These data suggest that CDI with carbon aerogels may reach a saturation point with respect to sorbing NOM. Sorption mechanisms for NOM may include electrostatic attraction as well as physical enmeshment with the carbon aerogel structure, with the result that the NOM is not being removed during regeneration. In conclusion, the presence of NOM in the source water appears to reduce the inorganic sorption capacity of the carbon aerogel material. Pretreatment for NOM removal would aid the operational efficiency of the CDI process using carbon aerogels. Figure 12 shows influent water quality data and the removal percentage of various ions from a blend of Colorado River water and California state project water. No correlation between influent water quality data and percent removal was observed. These data show that in a competitive environment (i.e., when multiple ions of varying valences are present), the sorption of the divalent species is limited. Based on carbon aerogel-based CDI’s inability to preferentially sorb divalent ions, and its limited sorption capacity, further research and development are required before this technology can be useful for commercial applications.
Acknowledgments Funding for this project was graciously provided by the American Water Works Association Research Foundation, the Electric Power Research Institute, and the California Energy Commission. Special thanks are extended to Stephanie Duin, formerly with the University of California, Los Angeles, for conducting the bench-scale CDI testing. The authors also thank Metropolitan’s Water Quality Laboratory staff for conducting the inorganic analyses.
Literature Cited (1) Caudle, D. D.; Tucker, J. H.; Cooper, B. B.; Arnold, B. B.; Papastamataki, A. Research and Development Progress Report 188; U.S. Department of the Interior (USDI): Washington, DC, 1966. (2) Johnson, A. M.; Venolia, A. W.; Newman, J.; Wilbourne, R. G.; Wong, C. M.; Gillam, W. S.; Johnson, S.; Horowitz, R. H. The Electrosorb Process for Desalting Water; R&D Progress Report 516 (USDI Publication 200 056); Office of Saline Water: Washington, DC, 1970. 3018
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 13, 2002
(3) Johnson, A. M.; Newman, J. Desalting by Means of Porous Carbon Electrodes. J. Electrochem. Soc. 1971, 118, 510-517. (4) Oren, Y.; Soffer, A. Water Desalting by Means of Electrochemical Parametric Pumping: I. The Equilibrium Properties of a Batch Unit Cell. J. Appl. Electrochem. 1983, 13, 473-487. (5) Matlosz, M.; Newman, J. Experimental Investigation of a Porous Carbon Electrode for the Removal of Mercury from Contaminated Brine. J. Electrochem. Soc. 1986, 133, 1850-1859. (6) Wang, J.; Angnes, J.; Tobias, H.; Roesner, R.; Hong, K.; Glass, R.; Kong, F. M.; Pekala, R. Carbon Aerogel Composite Electrodes. Anal. Chem. 1993, 65, 2300. (7) Kaschmitter, J. L.; Mayer, S. T.; Pekala, R. W. U.S. Patent No. 5,260,855; 1993. (8) Pekala, R. W.; Alviso, C. T. Carbon Aerogels and Xerogels. In Novel Forms of Carbon; MRS Symposium Proceedings Vol. 270; Renschler, C. L., Pouch, J. J., Cox, D. M., Eds.; Materials Research Society: Pittsburgh, 1992. (9) Farmer, J. C.; Fix, D. V.; Mack, G. V.; Pekala, R. W.; Poco, J. F. Capacitive Deionization of NaCl and NaNO3 Solutions with Carbon Aerogel Electrodes. J. Electrochem. Soc. 1996, 143, 159169. (10) Farmer, J. C.; Tran, T. D.; Richardson, J. H.; Fix, D. V. The Application of Carbon Aerogel Electrodes to Desalination and Waste Treatment. In Proceedings of the Separation Science and Technologies Conference; American Institute of Chemical Engineers: Los Angeles, 1997. (11) Pekala, R. W.; Mayer, S. T.; Poco, J. F.; Kaschmitter, J. L. Structure and Performance of Carbon Aerogel Electrodes. In Novel Forms of Carbon II; MRS Symposium Proceedings Vol. 349; Renschler, C. L., Cox, D. M., Pouch, J. J., Achiba, Y., Eds.; Materials Research Society: Pittsburgh, 1994. (12) Mayer, S. T.; Pekala, R. W.; Kaschmitter, J. L. The Aerocapacitor: An Electrochemical Double-Layer Energy-Storage Device. J. Electrochem. Soc. 1993, 140, 446-451. (13) Farmer, J. C.; Fix, D. V.; Mack, G. V.; Pekala, R. W.; Poco, J. F. Capacitive Deionization of NH4ClO4 Solutions with Carbon Aerogel Electrodes. J. Appl. Electrochem. 1996, 26, 1007-1018. (14) Wiesner, M. R.; Aptel, P. Mass Transport and Permeate Flux and Fouling in Pressure-Driven Processes. In Water Treatment Membrane Processes; Mallevialle, J., Odendaal, P. E., Wiesner, M. R., Eds.; McGraw-Hill: New York, 1996. (15) Faust, S. D.; Aly, O. A. Chemistry of Water Treatment, 2nd ed; Ann Arbor Press: Ann Arbor, MI, 1998. (16) Chang, R. Chemistry; 4th ed; McGraw-Hill: New York, 1991. (17) Weber, W. J., Jr. Physicochemical Processes for Water Quality Control; Wiley-Interscience: New York, 1972. (18) Abrams, I. M.; Benezia, L. Ion-Exchange Polymers. In Encyclopedia of Polymer Science and Technology; John Wiley & Sons: New York, 1967. (19) Trainhaim, J. A.; Newman, J. Flow-Through Porous Electrode Model: Application to Metal-Ion Removal from Dilute Streams. J. Electrochem. Soc. 1977, 124, 1528-1540. (20) Tran, T. D.; Farmer, J. C.; De Pruneda, J. H.; Richardson, J. H. Electrosorption on Carbon Aerogel Electrodes as a Means of Treating Low-Level Radioactive Wastes and Remediating Contaminated Ground Water. In Proceedings of the Sixth International ASME Conference on Radioactive Waste Management and Environmental Remediation; American Society of Mechanical Engineers: New York, 1997. (21) Tran, T. D.; Lenz, D.; Kinoshita, K.; Droege, M. Effects of Processing Conditions on the Physical and Electrochemical Properties of Carbon Aerogel Composites. In Materials for Electrochemical Energy Storage and Conversion IIsBatteries, Capacitors and Fuel Cells; MRS Symposium Proceedings Vol. 496; Materials Research Society: Warrendale, PA, 1998. (22) Barrett, E. P.; Joyner, L. G.; Halenda, P. P. The Determination of Pore Volume and Area Distributions in Porous Substances. I. Computations from Nitrogen Isotherms. J. Am. Chem. Soc. 1951, 73, 373. (23) CRC Handbook of Chemistry and Physics, 73rd ed.; Lide, D. R., Ed.; CRC Press: Boca Raton, FL, 1992. (24) Huheey, J. E. Inorganic Chemistry, 3rd ed.; Harper & Row: New York, 1983. (25) Nightingale, E. R., Jr. Phenomenological Theory of Ion Solvation: Effective Radii of Hydrated Ions. J. Phys. Chem. 1959, 63, 1381-1387. (26) Christy, A. A.; Bruchet, A.; Rybacki, D. Characterization of Natural Organic Matter by Pyrolysis-GC/MS. Envir. Intl. 1999, 25, 1-9.
(27) Tran, T. D.; Lenz, D. V. Alternating-Polarity Operation for Complete Regeneration of Electrochemical Deionization System. Lawrence Livermore National Laboratories Case No. IL-10314, patent filed January 21, 1999. (28) Yang, K.-L.; Ying, T.-Y.; Yiacoumi, S.; Tsouris, C.; Vittoratos, E. S. Electrosorption of Ions from Aqueous Solutions by Carbon Aerogel: An Electrical Double-Layer Model. Langmuir 2001, 17, 1961-1969. (29) Jayson, G. G.; Sangster, J. A.; Thompson, G.; Wilkinson, M. C. Adsorption and Electrosorption of Mercury(II) Acetate onto Activated Charcoal Cloth from Aqueous Solution. Carbon 1987, 25, 523-531.
(30) Tran, T. D.; Kinoshita, K. Carbon for Supercapacitors and LiIon Batteries in EVS. In Exploratory Research and Development of Batteries and Supercapacitors for Electric and Hybrid Vehicles; Adams, W. A., Ed.; Electrochemical Society: Pennington, NJ, 1998; Vol. II.
Received for review September 10, 2001. Revised manuscript received February 22, 2002. Accepted April 19, 2002. ES0112745
VOL. 36, NO. 13, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3019
