Environ. Sci. Technol. 2003, 37, 1992-1998
Comparing Nafion and Ceramic Separators Used in Electrochemical Purification of Spent Chromium Plating Solutions: Cationic Impurity Removal and Transport KUO-LIN HUANG† AND THOMAS M. HOLSEN* Department of Civil and Environmental Engineering, Clarkson University, Potsdam, New York 13699 TSE-CHUAN CHOU Department of Chemical Engineering, National Cheng Kung University, Tainan 70101, Taiwan J. ROBERT SELMAN Department of Chemical and Environmental Engineering, Illinois Institute of Technology, Chicago, Illinois 60616
This study focuses on the electrolytic regeneration of spent chromium plating solutions. These solutions contain a significant amount of chromium and a lesser amount of other heavy metals, which makes them a significant environmental concern and an obvious target for recycling and reuse. The type of separator used is extremely critical to the performance of the process because they are the major resistance in the transport-related impurity (Cu(II), Ni(II), and Fe(III)) removals from contaminated chromic acid solutions. A Nafion 117 membrane and a ceramic diaphragm separator traditionally used in the industry were tested for comparison. It was found that the mobilities of Cu(II) and Ni(II) were similar and higher than that of Fe(III) using both separators. The mobility of each cation was smaller in the Nafion membrane than in the ceramic diaphragm. The measured conductivity of the ceramic diaphragm was slightly higher than that of Nafion membrane. However, the Nafion membrane was much thinner than the ceramic diaphragm resulting in the system using the Nafion membrane having higher impurity removal rates than the system using the ceramic diaphragm. The removal rates were approximately equal for Cu(II) and Ni(II) and lowest for Fe(III). Both current and initial concentration affected the removal rates of the impurities. Modeling results indicated that a system using a Nafion separator and a small catholyte/anolyte volume ratio was better than a system using a ceramic separator for removing impurities from concentrated plating solutions if the impurities transported into the catholyte are deposited or precipitated.
Introduction Hexavalent (hard) chromium plating is widely used in the chromium electroplating industry, and its use will most likely * Corresponding author phone: (315)268-3851; fax: (315)268-7636; e-mail:
[email protected]. † Present address: Center of General Education, Chang Jung Christian University, Tainan 711, Taiwan. 1992
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 9, 2003
continue due to the difficulty in finding alternative processes for specific applications (1). As contaminants (i.e. Cu(II), Ni(II), and Fe(III)) build up in chromic acid baths during use, a portion of the baths must be discarded even though it still contains large amounts of chromic acid (2). Since disposal of chromic acid is difficult, it is increasingly attractive to regenerate the waste chromium liquors (1). One of the regeneration techniques that can be employed is electrolytic separation. In this process, a cell with two compartments divided by a separator is used to separate purified and contaminated solutions. The materials generally used for the separator are a cation exchange membrane or porous ceramic diaphragm (3). Electrolytic regeneration can simultaneously remove contaminants such as Cu(II), Ni(II), and Fe(III) and oxidize Cr(III) to Cr(VI). Removal relies on the physical transport of the contaminants driven by an electrical field (migration) across the separator, while oxidation is primarily associated with an electrochemical reaction at the anode. Accordingly, removal is strongly related to the applied current/potential and the type of separator used, whereas oxidation relies on the nature of anode surface and the applied current/potential. As a result, the removal efficiency due to migration is, in general, much lower than that of reoxidation. For example, the system used by the Bureau of Mines showed poor copper removal but efficient Cr(III) reoxidation (4). An overall current efficiency of 5% for copper removal after 24 h with an initial copper concentration of 20 g/L in waste chromic acid was seen. Mandich et al. (3) and Guddati et al. (5) also reported similar removals using a porous pot technique (PPT). However, Cr (III) reoxidation at the Pb anode was not as efficient in their experiments, and the ceramic separators used were sometimes clogged, probably by metal hydroxides, resulting in high cell voltage requirements. Removal of up to 60% of the nickel (in orthophosphoric acid) and 52% of the iron (in sodium monophosphate) was observed by Pattanayak et al. (6) using a similar porous pot technique, and they reported that chromic acid appeared to be the most suitable catholyte for purification when more than one metallic impurity was present in the spent solution. More recently, Mondal et al. (7) indicated that a Bi-doped PbO2 anode had faster oxidation kinetics for trivalent chromium than a PbO2 coated lead anode using a similar PPT process. In general, the removal of Cr(III) from plating baths is a minor concern compared to the removals of other impurities because it is fairly easily oxidized to Cr(VI) resulting in a much faster removal rate than for the other contaminants. Therefore, the removals of the non-chromium impurities (Cu(II), Ni(II), and Fe(III)) through the separators were focused on in this study. A Nafion membrane was used as the separator to see if removal rates could be increased and clogging could be avoided. A ceramic separator commonly used in the regeneration process was used for comparison. Modeling incorporating transport parameters was used to compare the two systems.
Experimental Section The device used for experiments consisted of a separator with two Teflon seals sandwiched between two acrylic compartments (Figure 1). The anode and cathode Pb foil (Aldrich) areas were 15 and 10 cm2, respectively. Both electrodes were connected by a DC power supply. One of the separators used was a Nafion 117 membrane (Dupont) which is much thinner and less porous than the ceramic diaphragm used (0.9 cm thick, 40% porosity with an average pore size 10.1021/es026037w CCC: $25.00
2003 American Chemical Society Published on Web 03/20/2003
TABLE 1. Initial Concentrations (C0, mM) of the Impurities for Each Experimenta Nafion separator experiment ion Cu(II) Ni(II) Fe(III)
ceramic separator
1
2
3
4
5
6
7
0.5A, MC 78.4 79.8 72.6
0.8A 86.9 106.5 92.0
0.2A 86.9 106.5 92.0
HC 189.6 187.1 171.6
LC 36.3 35.0 34.8
Cer0.5A 89.6 89.9 77.9
Cer0.8A 96.2 94.6 80.7
a 0.5A, MC: operated at 0.5 A with initially middle concentrations of impurities. HC: operated at 0.5 A with initially higher concentrations of impurities. LC: operated at 0.5 A with initially lower concentrations of impurities. Cer0.5A: operated at 0.5 A using ceramic separator.
FIGURE 1. Device for removing impurities from spent chromiumelectroplating baths. Magnetic stir bars were used to mix the anolyte and catholyte. Cr(III) can be oxidized to Cr(VI) at the anode, and Cr(VI) can be reduced to Cr(III) at the cathode. The direct reduction of Cr(VI) to Cr is very difficult without the addition of sulfate. Oxygen and hydrogen evolution also occurred at the anode and cathode, respectively, during operation. of 0.5 µm (1 µm maximum), Hard Chromium Plating Consultants). Nafion 117 (equivalent wt of 1100) is a cationexchange membrane with hydrophobic fluorocarbon backbone and hydrophilic sulfonate fixed site (8). The dry thickness of the Nafion membrane is 0.01778 cm, and its ion exchange capacity is 0.91 mequiv/g (9). The porosity of Nafion membrane equilibrated with the contaminated 2.12 M chromic acid solution was about 25% (10). The ceramic diaphragm was not ion selective and consisted primarily of alumina. Both separators had the same projected area of 10 cm2. All experiments were performed at room temperature (25-26 °C). The anolyte and catholyte used were 2.12 M chromic acid prepared from CrO3 (Aldrich). The anolyte (contaminated chromium plating solution) contained Cu(II), Ni(II), Fe(III), and Cr(III) impurities (3, 11) added as metal sulfates (CuSO4, NiSO4‚6H2O, FeSO4‚7H2O, and Cr2(SO4)3‚10H2O, Aldrich). Most of the added Fe(II) was quickly oxidized to Fe(III) by the Cr(VI) in the chromic acid and by anodic oxidation during electrolytic operation. The chromate/sulfate ratio was maintained at 100/1 by adding BaCO3 (Aldrich) to precipitate the excess sulfate anion. Both anolyte (1000 mL) and catholyte (1000 mL) were stirred with magnetic stir bars. During the experiments, the pHs of the anolyte and catholyte were stable and similar, approximately 0.2-0.3. Cu, Ni, and Fe were analyzed by atomic absorption (AA) spectrophotometer (Hitachi Z-6100). The conductivity of the ceramic diaphragm was determined using an AC impedance approach (12, 13). In these tests the ceramic diaphragm was soaked in the contaminated 2.12 M CrO3 for more than 24 h before the measurements. Two Pt planar electrodes were tightly sandwiched with a ceramic diaphragm flooded with the contaminated chromic acid solution. These two Pt electrodes were connected to an EG&G Model 5210 lock in amplifier and an EG&G 273A potentiostat during operation. The applied frequency varied from 100 K to 10 Hz at 0.1 V. The projected areas of the two Pt foils and ceramic diaphragm were 1 cm2. Data were recorded by Echem (M270) software (from EG&G) and plotted as imaginary resistance (Zim) vs real resistance (Zr) from high to low frequency. In an impedance diagram, the intercept (R) on the Zr (real) axis at high frequency (far away from the occurrence of electrochemical reaction) represents the ohmic resistance. The conductivity is κ ) (1/R)(L/A), where L and A are the thickness and projected area of the membrane, respectively.
Results and Discussion Electrolytic Removal of Cu(II), Ni(II), and Fe(III) Impurities. The removal of Cu(II), Ni(II), and Fe(III) impurities from the
anolyte were mainly caused by cation migration and diffusion driven by the electrical field and concentration gradient, respectively. (Osmosis was not important in this system since ionic strength on both sides of the membrane was similar.) The species flux resulting from ohmic and concentration gradients can be described with the following form of the Nernst-Planck equation (8, 14, 15)
Nι ) -zιuιFCι∇φ - Dι∇Cι
(1)
where Nι is the flux, zι is the charge on the ion, uι is the mobility, cι is the species concentration, F is Faraday’s constant, ∇φ is the potential gradient, and Dι is the diffusivity. The two terms on the right side of eq 1 represent fluxes caused by migration and diffusion, respectively. In a strict sense eq 1 is only valid in a dilute solution (14-17); however, it is commonly extended to more concentrated solutions (14) and membranes (17-20). Impurity Removal in Systems Using the Nafion and Ceramic Separators. Five impurity removal experiments using the Nafion separator and two experiments using the ceramic separator were performed (Table 1). In the first experiment using 0.5 A (current density: 0.05 A/cm2 effective separator area), the removal of Cu(II) was similar to that of Ni(II) but faster than the removal of Fe(III) (Figure 2). The buildup of impurities in the catholyte were consistent with their removals in the anolyte. Impurity removals in the other four experiments are quite similar to that found in experiment 1. The operation time, average removal rates, total efficiencies, and energy consumption of the three impurities for the above experiments using the Nafion separator are summarized in Table 2. As can be seen, the average removal rates and fluxes of the impurities increase if the current or initial concentration increases. (The time period used also affects the calculated average removal rate. Since the removal rates declined with time, the longer the time used in calculation, the lower the removal rate obtained, if the other conditions were constant.) Impurity removal rates using the ceramic separator system were much smaller than those measured using the Nafion membrane. For example, at 0.5 A, the average removal rate for each impurity was about one-half of that in the Nafion separator system even though the concentration of each impurity in the ceramic separator system was slightly higher than that in the Nafion system. Even though the diffusion coefficients of the three cations in the ceramic diaphragm are all higher than in the Nafion 117 (10), the ceramic separator is much thicker than the Nafion resulting in a smaller overall mass transfer coefficient. Since the mobility of each metallic cation in the ceramic is also greater than that in the Nafion separator, it can be inferred that the ceramic has a smaller potential gradient or greater separator conductivity than the Nafion (eq 1). The current (It) associated with impurity transport (including migration and diffusion) through the separator VOL. 37, NO. 9, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1993
FIGURE 2. Variation of dimensionless impurity concentration against time in (a) anolyte and (b) catholyte operated at 0.5 A (using Nafion separator).
TABLE 2. Operation Time (T), Average Removal Rates, Total Efficiencies (E, Sum of the Efficiency of the Three Species), and Energy Consumption (EC, in KJ per mol of Impurity Removal) of Impurities for the Experiments Using the Nafion and Ceramic Separators average removal rate mM/h Cu(II) Ni(II) Fe(III) E,b % EC, kJ/mol
experiment label
T, h
0.5A, MC 0.8A 0.2A HC LC
89.5 79.3 79.3 94.5 90.1
0.16 0.22 0.12 0.39 0.057
Nafion 0.19 0.32 0.17 0.45 0.060
0.090 0.13 0.068 0.23 0.020
5.3 5.0 10.6 12.7 1.6
12500 15400 5600 5200 40900
106.6 90.8
0.080 0.134
Ceramic 0.096 0.160
0.044 0.069
2.6 2.7
33500 37300
0.5A 0.8A
can be calculated by It ) AFΣzjNj, where Nj (mol/s‚cm2) and zj are the flux and charge of the species j, respectively, and A and F are the projected separator area and the Faraday’s constant, respectively (14, 19). The efficiency equals It/I, where I is the total current applied. The total efficiencies in these experiments were low, and the total current efficiency (migration only) were even lower indicating that most of the 1994
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 9, 2003
cations transported through the Nafion separator were protons. The system using the Nafion separator had a greater total efficiency than that using the ceramic separator. The total efficiency and energy consumption were current, concentration, and operational time dependent. In general, the total efficiency decreased and energy consumption increased as the applied current increased. Both migration and diffusion contribute to the contaminant flux. To compare their relative importance the diffusion flux was calculated using data from the diffusion alone (no current) experiments (10), and the migration flux was calculated by subtracting this flux from the total measured flux. This results show that when the current density increased, the total flux increased due to an increase in the migration flux (Figure 3(a)). At equivalent current (0.5 A), the increase of initial concentration also resulted in the increase in the total flux due to an increase in the concentration gradient across the separator (Figure 3(b)). These results were in agreement with the Nernst-Planck equation. In general, the anode-to-cathode cell voltages were all fairly constant (variations < 5%) during experiments. The larger the applied current, the higher the cell voltage required. Initial concentration did not affect the required cell voltage. However, compared to the system with the Nafion separator, the system with ceramic separator required about 1 V higher cell voltages. The relatively constant cell voltages over times
FIGURE 4. The impedance curve for the ceramic diaphragm in the contaminated chromic acid solution. Topsall data and bottoms data measured at high frequency (Zr ) ∼2.1 Ω at Zim ) 0 Ω). L is the thickness of the separator, i is the current density, and κ is the conductivity of separator. FIGURE 3. (a) Average Cu(II) diffusion, migration, and total fluxes for the system using a Nafion separator. The migration flux is equal to the total flux (after applying current) minus the diffusion flux (diffusion only). Diffusion experiment: initial Cu(II) ) 95 mM, operation time ) 95 h; experiments in which current was varied: initial Cu(II) ) 90 mM, operation time ) 90 h. (b) Variation of average Cu(II) flux with initial anolyte Cu(II) concentration for the system using Nafion separator operating at 0.5 A. The operation times were 90, 90, and 95 h for the three experiments with different initial Cu(II) concentrations (from low to high), respectively. of >100 h indicate there was no clogging in the ceramic. Mandich et al. (3) and Guddati et al. (5) reported that their ceramic membranes were clogged at about 6 and 40 h after applying current operating at 1 and 0.37 A, respectively, probably by hydroxides of the metal impurities due to a pH increase in the catholyte. The catholyte volumes (1 L) in these experiments were significantly larger than in their experiments (100 mL) so the catholyte pH remained below 1 and no metal hydroxides were formed. Both the use of chromic acid as the catholyte and Nafion as a separator (resulting in faster proton transfer) were helpful for preventing a pH increase and sludge formation in the catholyte. The yellow to orange cathode deposits (possibly due to dichromate or Cr2O3 compounds) observed in this work were also different than the dark-green deposits (possibly due to coprecipitated iron-nickel compounds) observed in their experiments.
Modeling As discussed previously, both migration and diffusion cause Cu(II), Ni(II), and Fe(III) to move through the membrane into the catholyte. A general mathematical model developed for the system based on mass balance and transport mechanisms can be used to obtain ionic mobilities by fitting data and to understand the importance of each of the processes involved in impurity removal. Conductivity Measurements. Since ∇φ is assumed to be uniform across the separator, it can be approximately expressed by ∇φ ) ∆V/L ) i/κ assuming that the potential drop caused by the concentration gradient is negligible (14, 16, 17), where ∆V is the potential drop across the separator,
The measured ohmic resistance of the ceramic was approximately 2.1 Ω (Figure 4) corresponding to a conductivity of 0.43 (Ω-cm)-1. This value is slightly smaller than the 0.52 (Ω-cm)-1 reported for a solution of 2.5 M CrO3 + 0.025 M H2SO4 (21) which is reasonable since the membrane and metallic impurities decrease the conductivity of the solutions (22). The conductivity of the Nafion was not measured, and the value of 0.38 S/cm reported in a literature (20) was used. Modeling Approach. According to Newman (14), the material balance for an ion species i in the bulk electrolyte can be expressed as
∂Ci/∂t ) -∇Ni + Ri
(2)
where Ri is the chemical reaction in the electrolyte. Convection is not included directly because convection in both Nafion and ceramic is complex due to the presence of the wall in the mesoporous or microporous medium. The effective mobility and effective diffusivity do include a possible electrosmotic effect. However this effect appears to be small since there is a linear relationship between the calculated migration flux and current density (at a constant diffusion flux) for Cu(II) transport in the Nafion (Figure 3(a)). Combining eqs 1 and 2, the mass balance equations for the three control volumes (separator, anolyte, and catholyte) can be written as follows (18, 20, 23). I. Within the Separator
( )( ) ( ) ( )
∂Cmi ∂Cmi ∂φmi ∂2Cmi ) ziuiF + Dei + ∂t ∂x ∂x ∂x2 ∂2φmi ziFuiCmi 0 < x < L (3) ∂x2 II. In the Anolyte
[
( )
( )] (
∂Cai Am ∂φmi ∂Cmi ) z Fu C + Dei ∂t Va i i mi ∂x ∂x
(
Aanode sii Va mF
VOL. 37, NO. 9, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
)
9
(4)
1995
III. In the Catholyte
[
( )
( )] (
∂Cci ∂φmi ∂Cmi Am z Fu C + Dei )∂t Vc i i mi ∂x ∂x
(
)
Acathode sii Vc mF
(5) The initial and boundary conditions are as follows:
IC 1: Cmi ) Cmio for 0 < x < L (at t ) 0)
(6)
IC 2: Cai ) Caio
(7)
IC 3: Cci ) Ccio
(8)
BC 1: Cmi|x ) 0 ) K′(Cai)n
(9)
BC 2: Cmi|x ) L ) K′(Cci)n
(10)
where Cmi is the concentration of ion i in membrane, Cmio is the initial concentration of ion i in membrane, Cai is the concentration of ion i in anolyte, Cci is the concentration of ion i in catholyte, Caio is the initial concentration of ion i in anolyte, Ccio is the initial concentration of ion i in catholyte, Am is the effective membrane area for transport, Aanode is the effective area of anode, Acathode is the effective area of cathode, Va is the volume of anolyte, Vc is the volume of catholyte, Dei is the effective or apparent diffusivity of ion I, F is Faraday’s constant, 96487 C/equivalent, R is the universal gas constant, 8.3143 J/K mol, T is temperature, K, i is the current efficiency for ion I, i is the current density (i ) I/Aanode or I/Acathode), I is the total current, m is the number of electrons transferred in redox reaction (Ox + me- f Red), ui is the mobility, and K′ and n are constants of Freundlich isotherm for ionic partition in separator. The above mass balance equations are based on the following important assumptions: 1. The Nernst-Planck equation with experimentally determined diffusivities (10) and conductivities can be applied in the separator domain (18, 20, 23). 2. The ion concentrations are uniform in bulk solution (complete mixing). 3. There are no kinetic limitations for the electrode reactions. 4. No electrochemical reactions involving Cu2+, Ni2+, and Fe3+ occur at the anode. 5. ∇φ is uniform across the membrane (linear potential profile). Based on assumption 5, the third term on the right-hand side of eq 3 can be eliminated. The third term on the righthand side of eq 4 also can be eliminated due to assumption 4. It is also assumed that the potential drop caused by the concentration gradient is negligible compared to that due to ohmic resistance so ∇φ ) ∂φmi/∂x ) i/κ. A Fortran program developed by Ahmed et al. (20) was used to model the ion transport within the separator and the concentration variation (with time) in the electrolytes. The system equations (eqs 3-5) with initial and boundary conditions (eqs 6-10) were numerically solved using a combination of the orthogonal collocation method (24, 25) and numerical integration using Gear’s method (26). A series of orthogonal polynomials (Shifted Legendre Polynomials) was used to simulate the solution function, and the sum of errors was forced to approach zero by a weighting function. The coefficients of the polynomials were chosen to minimize the weighted residuals (20, 24). The roots of the polynomials specified the collocation points in the controlled domain (e.g., the separator) where the differential equations were forced to be satisfied. In the Fortran program, the set of nonlinear partial differential equations (PDEs) were converted into a set of linear ordinary differential equations (ODEs) and a subroutine based on Gear’s method with time step control was used to solve the set of ODEs. 1996
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 9, 2003
FIGURE 5. (a) Model fitting to the anolyte Cu(II), Ni(II), and Fe(III) concentration variations with time operating at 0.5 A for the system using a Nafion separator. D: data; M: model fitting line. (b) Model fitting to the anolyte Cu(II), Ni(II), and Fe(III) concentration variations with time operating at 0.5 A for the system using a ceramic separator. D: data; M: model fitting line.
TABLE 3. Main Input Parameters (Conductivity and Diffusivity) and Mobilities Obtained by Data Fitting for Each Impurity input parameter
separator Nafion Ceramic
a
ion species
separator conductivity, S/cm
measured diffusivity,a 10-7 cm2/s
fitted parameter mobility, 10-5 cm2/V‚s
Cu(II) Ni(II) Fe(III) Cu(II) Ni(II) Fe(III)
0.38b 0.38b 0.38b 0.43 0.43 0.43
11.8 10.9 5.6 36.4 30.4 16.8
3.5 3.3 2.5 4.6 3.9 3.2
Data from ref 10.
b
The conductivity reported by Ahmed et al. (20).
Mobilities Obtained from Model Fitting. The major input parameters required to obtain mobilities are membrane conductivity and diffusivity. The former and the latter were measured in this study and a previous study (10), respectively. However, the Nafion membrane conductivity used here was from the literature (20). The model was able to fit the data very well (Figure 5). The fitted electrolytic mobilities (17) are summarized in Table 3. The estimated mobilities of the three species are in the order of Cu(II) > Ni(II) > Fe(III), the same as their diffusivities (10). When they were used as input to the model, good agreement (errors < 5%) between measured data and fitted curves was found. The estimated membrane-phase Ni(II) mobility (3.3 × 10-5 cm2/V‚s) is slightly lower than that reported by Lehmani et al. (27) (5.0 × 10-5 cm2/V‚s) using 1-2 M nitric acid. The fitted mobility (Uf) and the mobility calculated using the Nernst-Einstein equation (Uc ) DzF/RT) and diffusivity (D)
concentration in the catholyte. When the catholyte/anolyte volume ratio was 1 L/1 L, the catholyte did not reach Cu(II) concentrations greater than the initial anolyte concentrations in either system, even after 50 days of operation at 0.5 A (Figure 6(a)). In agreement with the experimental results, the modeling results shows that initially the system using the ceramic separator has a smaller decrease in the anolyte Cu(II) concentrations and smaller increase in the catholyte Cu(II) concentrations than the system using the Nafion separator. However after 45 days, the ceramic separator system performance will be better than the Nafion system. This crossover time decreases as the catholyte/anolyte volume ratio decreases. In addition, the ceramic separator system is able to concentrate the impurity (Cu(II)), while the Nafion separator system only reaches catholyte Cu(II) concentrations equal to that of the original anolyte after about 30 days of operation. This difference is mainly due to increased back diffusion of Cu(II) from the catholyte to the anolyte in the Nafion separator system. If the catholyte/anolyte volume ratio is decreased to 0.1 L/1 L, the ceramic separator system theoretically will be much better than the Nafion separator system (Figure 6(b)). The ceramic separator system starts to concentrate the Cu(II) in the catholyte after about 5 days and can reach 5 times the initial concentration at around 50 days of operation. As mentioned earlier, however, this catholyte volume (0.1 L) may be too small to keep a low pH in the catholyte (H+ is consumed by hydrogen evolution in the catholyte) resulting in the precipitation of metal hydroxides as reported by Mandich et al. (3) and Guddati et al. (5). They used a catholyte/ anolyte volume ratio of 0.1 L/7 L, and their ceramic was clogged by the metal hydroxides causing the cell voltage to increase sharply to over 30 V after 6 and 40 h of operation at 1 A and at 0.37 A, respectively. In contrast, it is likely that the Nafion separator would not be clogged by these metal hydroxides because the high proton flux through the membrane will keep the catholyte pH low. Under these conditions, the Nafion separator system may be economically viable even though the Nafion membrane ($0.12/cm2) is more expensive than the ceramic diaphragm ($0.072/cm2). This possibility needs further investigation. FIGURE 6. (a) Model of the anolyte and catholyte Cu(II) concentration variations with time for systems using a Nafion and ceramic separators operating at 0.5 A with a catholyte/anolyte volume ratio ) 1 L/1 L. Assuming no impurity deposition in the catholyte; projected separator area ) 10 cm2. (b) Model of the anolyte and catholyte Cu(II) concentration variations with time for systems using a Nafion and ceramic separators operating at 0.5 A with a catholyte/anolyte volume ratio ) 0.1 L/1 L. Assuming no impurity deposition in the catholyte; projected separator area ) 10 cm2. measured in previous work (10) are significantly different, particularly for the ceramic membrane (Uc ) ∼2.6 and 6 Uf, for the Nafion and ceramic separators, respectively). A somewhat smaller difference was seen by Pourcelly et al. for Ca2+ in a Nafion 117 membrane separator (28). These differences are probably due to the fact that the NernstEinstein equation was derived using a statistical mechanics approach (29) and is strictly applicable only at infinitely dilute solutions (14, 15, 17) so it cannot be directly applied in the separator domain. A detailed discussion of reasons for deviation from the Nernst-Einstein relationship are presented by Auclair et al. (30). Engineering Significance. Since waste reduction is a primary concern of this study, it is important to determine if it is possible to concentrate the impurities in the catholyte. One way to achieve this goal may be to use a small volume of catholyte. Therefore a modeling study was used to evaluate the effect of the catholyte/anolyte volume ratio on the
Acknowledgments The authors would like to acknowledge the contributions of Professor Khalili and Dr. M. Muntasir Ahmed at IIT, Professor Der Tau Chin at Clarkson University, and Dr. Ming-Chang Yang at National Cheng Kung University. Although the research described in this article has been funded wholly or in part by the United States Environmental Protection Agency through Grant R827125 it has not been subjected to the Agency’s required peer and policy review and therefore does not necessarily reflect the views of the Agency and no official endorsement should be inferred.
Literature Cited (1) Pletcher, D.; Walsh, F. C. Industrial Electrochemistry; Blackie Academic & Professional: New York, 1993; pp 60-171 and 331384. (2) Beszedits, S. Chromium removal from industrial wastewaters. In Chromium in the Natural and Human Environment; Nriagu, J. O., Nieboer, E., Eds.; John Wiley & Sons: New York, 1988; pp 231-265. (3) Mandich, N. V.; Li, C.-C.; Selman, J. R. Plating and Surface Finishing 1997, Dec., 82-90. (4) Ng, P. K.; Snyder, D. D. J. Membrane Sci. 1983, 13, 327-336. (5) Guddati, S. L.; Holsen, T. M.; Li, C.-C.; Selman, J. R.; Mandich, N. V. J. Appl. ElectroChem. 1999, 29, 1129-1133. (6) Pattanayak, J.; Mandich, N. V.; Mondal, K.; Wiltowski, T.; Lalvani, S. B. Environ. Technol. 1999, 20, 317-323. VOL. 37, NO. 9, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1997
(7) Mondal, K.; Mandich, N. V.; Lalvani, S. B. J. Appl. Electrochem. 2001, 31, 165-173. (8) Hamann, C. H.; Hamnett, A.; Vielstich W. Electrochemistry; WileyVCH: New York, 1998; pp 178-179. (9) Bontha, J. R.; Pintauro, P. N. Chem. Eng. Sci. 1994, 49, 38353851. (10) Huang, K.-L. Ph.D. Dissertation, Clarkson University, NY, 2001. (11) Krishnan, R. M.; Sriveeraraghavan, S.; Jayakrishnan, S.; Kennady, C. J.; Sekar, R.; Baby, N. Bull. Electrochem. 1999, 15, 195-198. (12) Samec, Z.; Trojanek, A.; Langmaier, J. J. Electrochem. Soc. 1998, 145, 2740-2746. (13) Millet, P. J. Membrane Sci. 1990, 50, 325-328. (14) Newman, J. S. Electrochemical Systems; Prentice-Hall: NJ, 1991; pp 241-257. (15) Prentice, G. Electrochemical Engineering Principles; PrenticeHall: NJ, 1991; pp 21-23 and 152-156. (16) Bard, A. J.; Faulkner, L. R. Electrochemical Methods: Fundamentals and Applications; John Wiley & Sons: New York, 2001; pp 137-155. (17) Koryta, J.; Dvorak, J.; Kavan, L. Principles of Electrochemistry; John Wiley and Sons: New York, 1993; pp 85-87 and 110-116. (18) Verbrugge, M. W.; Hill, R. F. J. Electrochem. Soc. 1990, 137, 886893. (19) Ahmed, M. I.; Holsen, T. M.; Selman, J. R. J. Appl. Electrochem. 2001, 31, 1381-1387. (20) Ahmed, M. I.; Chang, H. T.; Selman, J. R.; Holsen, T. M. J. Membrane Sci. 2002, 197, 63-74.
1998
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 9, 2003
(21) Vashchenko, S. V.; Solodkova, L. N.; Solov’eva, Z. A. Russ. J. Electrochem. 2000, 36, 947-951. (22) Benaben P.; Popakul J. Plating Surface Finishing 2000, Dec., 88-91. (23) Verbrugge, M. W.; Hill, R. F. J. Electrochem. Soc. 1990, 137, 37703777. (24) Ramirez, W. F. Computational Methods for Process Simulation; Butterworth: Boston, 1989; pp 395-405. (25) Villadsen, J. V.; Stewart, W. E. Chem. Eng. Sci. 1967, 22, 14831501. (26) Finlayson, B. A. The Method of Weighted Residuals and Variational Principles; Academic Press: New York, 1972. (27) Lehmani, A.; Turq, P.; Perie, M.; Perie, J.; Simonin, J.-P. J. Electroanal. Chem. 1997, 428, 81-89. (28) Pourcelly, G.; Sistat, P.; Chapotot, A.; Gavach, C.; Nikonenko, V. J. Membrane Sci. 1996, 110, 69-78. (29) Bird, R. B.; Stewart, W. E.; Lightfoot, E. N. Transport Phenomena, 2nd ed.; John Wiley & Sons: New York, 2002; pp 528-530. (30) Auclair, B.; Nikonenko, V.; Larchet, C.; Metayer, M.; Dammak, L. J. Membrane Sci. 2002, 195, 89-102.
Received for review August 8, 2002. Revised manuscript received January 3, 2003. Accepted February 12, 2003. ES026037W
