Environ. Sci. Technol. 2006, 40, 2746-2752
Micellar Enhanced Ultrafiltration for Arsenic(V) Removal: Effect of Main Operating Conditions and Dynamic Modelling F . B E O L C H I N I , * ,† F . P A G N A N E L L I , ‡ I. DE MICHELIS,§ AND F. VEGLIO Å § Dipartimento di Scienze del Mare, Universita` Politecnica delle Marche, Via Brecce Bianche, Ancona, Italy, Dipartimento di Chimica, Facolta` di S.M.F.N., Universita` degli Studi “La Sapienza”, P.le A. Moro, 5, 00185 Roma, Italy, and Dipartimento di Chimica, Ingegneria Chimica e Materiali, Universita` degli Studi di L’Aquila, 67040 Monteluco di Roio, L’Aquila, Italy
In this work arsenic removal by micellar enhanced ultrafiltration (MEUF) was investigated using cetylpyridinium chloride (CPC) in ceramic membrane apparatus. Permeability tests and discontinuous diafiltration tests were performed in different operating conditions to evaluate the effect of membrane pore size (20 and 50 nm), transmembrane pressure, pH, surfactant concentration (1-3 mM), and arsenic concentration (10-40 mg/L) on permeate flux decline, arsenic, and CPC rejections. These preliminary experimental results showed that a ceramic membrane with large pore size allows treament of high fluxes of concentrated arsenic-bearing solutions even by using low surfactant concentrations. Arsenic concentration in the permeate was at the 1 ppm level, with feed As concentrations (10 ppm) that are larger than those generally used in MEUF studies and with CPC amounts that are lower than the usual ones. In addition, operating conditions adopted in these tests obtained CPC concentrations in the permeate always lower than its critical micellar concentration (0.9 mM). Dynamic simulations of discontinuous two-step diafiltration tests allowed a simple and adequate representation of the performance of the process especially for 1 mM CPC, while discrepancies for 2.5 mM CPC level denoted complex interactions between CPC and As.
Introduction Membrane processes can be used as cost-effective and environmentally safe separation techniques as an alternative to sorption onto active carbon and synthetic resins (1). The attractive low-energy characteristics of ultrafiltration can be exploited also for the removal of low dimension ionic species (such as heavy metals) by using surfactant-based separation processes such as micellar-enhanced ultrafiltration (MEUF) (2-10). The effectiveness of MEUF processes is strictly related to the achievement of high permeate fluxes and high rejections of both heavy metals and surfactant. Low permeate * Corresponding author phone: +39 071 2204225; fax: +39 071 2204650; e-mail:
[email protected]. † Universita ` Politecnica delle Marche. ‡ Universita ` degli Studi “La Sapienza”. § Universita ` degli Studi di L’Aquila,. 2746
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 8, 2006
fluxes are observed working at high pressure, low retentate flow velocity, and high viscosity of feed solution (11). Surfactant rejection is increased by high retentate flow velocity due to the attenuation of concentration polarization produced by fast retentate flux (reduced surfactant gradient across the membrane) (11). On the other hand, opposite effects on surfactant rejections were observed by increasing transmembrane pressure: the increase of surfactant in the permeate as pressure increases was explained by taking into account micelle deformation, micelle decomposition, and high surfactant concentration near the membrane (increased surfactant gradient across the membrane), while the opposite effect (diminution of surfactant in permeate as pressure increases) was explained by a presieving effect sometimes observed for concentration polarization (11, 12). The improvement of surfactant rejection can be then approached by the modification of membrane module or material in order to reduce the concentration polarization (13). Nevertheless, concentration polarization itself can be also exploited to design MEUF working below CMC and reducing surfactant release and consumption. In fact, the concentration polarization can result in the formation of micelles near the membrane surface even below the surfactant CMC (14). In addition, the presieving of gel layer associated with concentration polarization can have a crucial effect on metal rejection when the surfactant concentration is lower than its CMC (2, 6, 15). Membrane pore size also plays an important role in this context (12). An increase of the molecular weight cutoff (MWCO) of the membrane can cause an earlier development of concentration-polarization regime and consequently reduce the surfactant release in the permeate due to presieving effect. As a consequence, even for a very large pore size (50 KDa MWCO), the vast majority of micelles can be rejected (12). According to this finding, high MWCO membranes present extremely good rejection characteristics with minimum area requirement and capital cost. In this work arsenic removal by MEUF was investigated using cetylpyridinium chloride (CPC) and a cross-flow ceramic membrane apparatus. The effects of some operating factors on permeate flux, arsenic, and CPC rejections were investigated, in particular transmembrane pressure, pH, CPC concentration, and As concentration. The novelty was in evaluating the possible advantages of using large-pore membranes (20 and 50 nm) and reduced surfactant concentrations (1-3 mM) for treating high fluxes of concentrated arsenic-bearing solutions (10-40 mg/L). An example of such concentrated solutions is given by acid mine drainage (AMD), which is a common problem in different areas with a past history of mining activities. Oxidation of pyritic ores is the major cause of acidic drainage which is characterized by low pH (2-4) and toxic metal concentrations often well above maximum contaminant levels for drinking water (16). In particular arsenic is one element that is common in AMD being generally associated with metal sulfide deposits that are responsible for acid drainage (17-19). AMD waters can be characterized by a wide range of As concentration depending on the specific site geochemistry and the type of past mining and metallurgical activities: literature data denoted As concentrations in AMD ranging from 2-4 ppm (20-24) up to 8 (19), 14 (20), 64, and 95 ppm (24, 25). Furthermore AMD is generally characterized by oxidizing redox conditions in which As(V) species prevail over As(III) ones (4, 24). Another example of concentrated arsenic-bearing waste can be found in biohydrometallurgical processes. In fact, biooxidation of arsenic-bearing refractory gold ores is performed as a pretreatment in order to enhance gold 10.1021/es052114m CCC: $33.50
2006 American Chemical Society Published on Web 03/14/2006
TABLE 1. Operating Conditions and Experimental Results in Permeability Tests: Membrane Pore Dimensions, Average Values of [CPC], [As], pH, and Eh in the Retentate; Lower Boundary Limits for As(V) Species (LB Eh); Membrane Resistances (RTOT) Regressed by Permeate Flux vs TMP Plots; Average Values of Arsenic Retention Coefficient (σAs) and Surfactant Retention Coefficient (σCPC) test set
test identifier
pore dimensions (nm)
[CPC]R (mM)
[As]R (mg/L)
pHR
EhR (mV)
LB Eh (mV)
RTOT (MPa h m-1)
R2 (-)
σAs (-)
σCPC (-)
1
P1 P2 P3
50 50 50
2.23 2.48 2.51
9.0 9.1 8.9
7.52 7.68 6.51
225 216 279
-83 -104 34
1.81 1.96 1.95
0.993 0.997 0.997
0.84 0.84 0.84
0.85 0.86 0.86
2
P4 P5 P6 P7
50 50 50 50
1.20 1.50 0.95 1.02
8.9 9.3 9.1 8.6
7.41 6.76 6.65 6.61
232 264 268 283
-69 11 21 25
1.90 2.11 3.35 0.90
0.998 0.998 0.997 0.999
0.88 0.89 0.89 0.82
0.78 0.85 0.83 0.83
3
P8 P9 P10
20 20 20
1.48 1.12 2.72
8.1 9.1 8.5
7.13 6.67 7.00
183 239 273
-32 19 -16
13.46 10.13 15.10
0.999 0.999 0.999
0.85 0.78 0.75
0.96 0.92 0.95
4
P11 P12
50 50
7.1 8.2
7.84 10.29
194 36
-125 -360
1.04 0.86
0.999 0.997
0.59 0.38
TABLE 2. Batch Diafiltration Tests Using 50 nm Membrane: Operating Conditions Adopted in the Different Test Sets ([CPC]0, [As]0, and pH); Average Values of Permeate Flux (JP), Potential (Eh), Arsenic Retention Coefficient (σAs), and Surfactant Retention Coefficient (σCPC) test set
test identifier
[CPC]0 (mM)
[As]0 (mg/L)
pH
Eh (mV)
LB Eh (mV)
JP (ml/min)
σAs (-)
σCPC (-)
5
D1 D2 D3 D4 D5 D6 D7 D8
0.75 0.89 0.69 0.65 1.90 2.07 1.58 1.74
8.1 39.2 7.9 42.6 8.3 39.9 7.9 37.1
7.76 8.03 10.11 9.96 7.65 8.03 10.04 10.03
229.8 189.6 84.4 56.8 222.4 182.4 71.3 51.8
-114.1 -149.2 -347.7 -337.6 -99.9 -149.2 -343.0 -342.3
71.3 67.3 69.6 64.4 69.6 80.2 73.4 65.3
0.89 0.78 0.87 0.40 0.77 0.54 0.74 0.54
0.88 0.78 0.82 0.53 0.75 0.52 0.70 0.68
6
D9 D10 D11 D12
6.7 41.7 8.0 33.1
7.86 8.11 9.74 10.02
191.4 138.8 85.7 59.9
-127.1 -159.6 -322.7 -341.7
121.0 124.3 122.8 118.5
0.58 0.26 0.51 0.37
extraction yields and gives a liquor leach with a significant arsenic content (26). The combination of investigated operating conditions was therefore aimed at improving the cost-effectiveness of the process taking advantage of the concentration-polarization and presieving effects associated with high fluxes, but also minimizing membrane fouling by using large pore sizes and low surfactant concentration. This would allow the reduction of surfactant leakage on one hand, and the economical treatment of large streams of highly polluted waters on the other.
Materials and Methods Membrane Apparatus. The ultrafiltration apparatus was a tangential flow laboratory pilot plant Membralox XLAB3 (EXEKIA, Bazet, France), with a single tube ceramic membrane Membralox T1-70 (pore diameter either 20 or 50 nm, membrane surface area 50 cm2) (Figure S1 Supporting Information). Chemicals. Arsenic pentoxide (As2O5) was analytical reagent grade with a purity of 99.9%. The cationic surfactant was cetylpyridinium chloride (CPC) [C16H33(N(C5H5))+Cl-] with a purity of 100%. Permeability Tests. Metal-surfactant bearing solutions (CPC, As, and pH as specified in Table 1) were introduced in a temperature-controlled glass reactor (3 L volume) and kept at constant temperature (30 °C) by a tank jacket connected to a thermostatic bath (CRIOTERM 10-80). Each emulsion was fed to the membrane module by a recirculation pump (tangential velocity 7 m/s, nominal flowrate 1 m3/h).
During each permeability test transmembrane pressure (TMP) was changed from 85 to 370 KPa. For each TMP, samples of permeate and retentate were collected to measure pH, Eh, As, and CPC concentrations. The permeate flux was also monitored during each permeability test by measuring the volume of permeate collected during a fixed time. A further test with a synthetic AMD was performed with a 50 nm membrane cut off, CPC 0.95 mM, pH 7, and two levels of transmembrane pressure: 130 and 230 kPa. Batch Diafiltration Tests. These ultrafiltration tests were performed according to the two-step procedure described as follows. Step 1. Metal-surfactant bearing solution (volume of 3 L; CPC, As, and pH as specified in Table 2) was introduced in the feed reactor (30 °C) and fed to the membrane module at constant TMP (185 KPa). Only retentate was recirculated to the feed reactor while permeate came out of the system and its flux was measured. Both permeate and retentate were sampled in order to measure pH, Eh, As, and CPC. The first step ended when final volume in the feed reactor was 1 L. Step 2. Metal-bearing solution (volume of 2 L; same composition as 1st step according to Table 2 but without CPC) was added to the feed reactor, to obtain a 3 L final volume. As in the 1st step, the obtained solution was fed to the membrane module at constant TMP (185 KPa). Also in this 2nd step only retentate was recirculated and permeate flux monitored. Both permeate and retentate were sampled and analyzed (pH, Eh, As, and CPC). The 2nd step ended when 1 L volume remained in the feed reactor. VOL. 40, NO. 8, 2006 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2747
Analytical Procedures. pH and Eh were measured by specific electrodes. Cetylpyridinium was determined by means of kit LANGE LCK 331: the cationic surfactant reacts with bromophenol blue and forms complexes that are extracted into chloroform and read at 410 nm by a CADAS 50 Dr. LANGE spectrophotometer. Replicated determinations revealed an experimental error of about 5%. Arsenic was determined by a flame atomic absorption spectrophotometer (Varian SpectrAA200), with an arsenic quantitation limit of 1 ppm. The same procedure was used both in the presence and in the absence of surfactant, as verified by means of determinations on control samples. In the case of the experimental test with a synthetic acid mine drainage, arsenic concentration was determined through inductively coupled plasma with optical emission spectroscopy (Varian VistaMPX). Measures were performed in triplicate. Membrane Maintenance. Membrane cleaning was worked out as follows: (1) pure water at 1.3 bar and room temperature for .5 h; (2)P3-Ultrasil 75 1% at 1.3 bar and room temperature for 1.5 h, with a backwashing device; (3) repeat step 1 twice.
Results and Discussion Permeability Tests. Four different sets of permeability tests (Table 1) were performed in order to investigate the effects of different operating factors such as membrane pore dimensions (20 and 50 nm), surfactant concentration (around 1 and 2.5 mM), and pH (6.5-7.8). The system was monitored for Eh, permeate flux, arsenic, and CPC concentrations for transmembrane pressure (TMP) ranging from 85 to 370 KPa. Both pH and Eh in retentate and permeate streams remained quite constant as transmembrane pressure was increased (experimental data not shown here). Mean values of Eh in the retentate (Eh) were reported in Table 1. In the investigated range of conditions the lower boundary of potential for As(V) species (LB Eh in Table 1) was calculated by empirical correlations (4). Comparing measured and calculated Eh it is possible to see that arsenic is mainly present as arsenate ions in the form of H2AsO4- (for pH between 2.24 and 6.88) and HAsO42- (for pH between 6.88 and 11.44) (4). This is in agreement with the fact that most surface waters are characterized by the predominance of As(V) species, while in groundwaters As(III) species are also present (4, 27). In all the investigated conditions, permeate flux (Jp) dependence on transmembrane pressure (TMP) can be represented by a linear trend (28). The apparent total membrane resistance (RTOT) evaluated by linear regressions of Jp versus TMP data were reported in Table 1 for the different permeability tests. The coefficients of determinations (R2) show that the linear model is an adequate approximation of the system for all experiments. Results in Table 1 show that membrane resistance estimates for Set 1 and Set 2 using a 50 nm membrane are very similar to the values obtained without surfactant (Set 4). This confirms that permeate flux in the presence of CPC at both levels (around 1 and 2.5 mM) is not significantly reduced because of membrane fouling and presieving for concentration polarization. On the other hand, an evident increase of membrane resistance can be observed when a 20 nm membrane is used (Set 3). Nevertheless, even in this case linear models for RTOT estimates are still adequate. Arsenic and CPC retention coefficients (σAs and σCPC) were evaluated from retentate concentration (CR) and permeate concentration (CP) for each component:
σ)1-
CP CR
(1)
Average values of retention coefficients for both As and CPC 2748
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 8, 2006
FIGURE 1. CPC released in the permeate during permeability tests (Table 1). are reported in Table 1. Arsenic retention coefficients measured during permeability tests do not seem to be significantly affected by the change of both transmembrane pressure and operating factors considered here (CPC concentration, pH, and membrane pore dimensions). Nevertheless, the effect of adding the surfactant in the systems is evident (Set 1-3 versus Set 4). In all the investigated conditions of Sets 1-3 the average values of σAs ranged from 0.75 to 0.89 while blank tests without CPC (Set 4) present lower retention coefficients (Table 1). Partial retention of arsenic even without surfactant micelles is mainly due to the negative charge of membrane matrix (isoelectric point of ceramic membranes is slightly below 6) (11), which can provide arsenate repulsion (Donnan-exclusion effect) (15). Arsenic concentration in the permeate is at the 1 ppm level in all the investigated conditions. Even though this value is larger than the general legal limits for this pollutant in effluents, it should be noted that the system was operated with As concentrations (10-40 ppm) that are larger than those generally used in MEUF studies (
