Bulk Chlorine Uptake by Polyamide Active Layers of Thin-Film

Feb 7, 2014 - We studied the volume-averaged chlorine (Cl) uptake into the bulk region of the aromatic polyamide active layer of a reverse osmosis mem...
0 downloads 9 Views 985KB Size
Article pubs.acs.org/est

Bulk Chlorine Uptake by Polyamide Active Layers of Thin-Film Composite Membranes upon Exposure to Free ChlorineKinetics, Mechanisms, and Modeling Joshua Powell, Jeanne Luh, and Orlando Coronell* Department of Environmental Sciences and Engineering, Gillings School of Global Public Health, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina 27599, United States S Supporting Information *

ABSTRACT: We studied the volume-averaged chlorine (Cl) uptake into the bulk region of the aromatic polyamide active layer of a reverse osmosis membrane upon exposure to free chlorine. Volume-averaged measurements were obtained using Rutherford backscattering spectrometry with samples prepared at a range of free chlorine concentrations, exposure times, and mixing, rinsing, and pH conditions. Our volume-averaged measurements complement previous studies that have quantified Cl uptake at the active layer surface (top ≈ 7 nm) and advance the mechanistic understanding of Cl uptake by aromatic polyamide active layers. Our results show that surface Cl uptake is representative of and underestimates volume-averaged Cl uptake under acidic conditions and alkaline conditions, respectively. Our results also support that (i) under acidic conditions, N-chlorination followed by Orton rearrangement is the dominant Cl uptake mechanism with N-chlorination as the rate-limiting step; (ii) under alkaline conditions, N-chlorination and dechlorination of N-chlorinated amide links by hydroxyl ion are the two dominant processes; and (iii) under neutral pH conditions, the rates of N-chlorination and Orton rearrangement are comparable. We propose a kinetic model that satisfactorily describes Cl uptake under acidic and alkaline conditions, with the largest discrepancies between model and experiment occurring under alkaline conditions at relatively high chlorine exposures.



INTRODUCTION Reverse osmosis (RO) and nanofiltration (NF) membranes for water desalination and reuse have a thin-film composite structure with a top ultrathin (∼20−200 nm) active layer supported by a polysulfone porous support (∼20−50 μm) backed by a mat of polyester fibers (∼300 μm).1 The top active layer constitutes the main barrier to water and solute permeation and is typically made of aromatic polyamide.1−3 One significant drawback to the use of polyamide RO/NF membranes is the high sensitivity of polyamide to degradation by free chlorine, which is the main disinfectant used in water purification applications. RO/NF membranes tend to suffer from biological fouling, and polyamide sensitivity to free chlorine forces treatment plants to use less effective disinfectants (e.g., combined chlorine). Membrane performance degradation from exposure to free chlorine is thought to occur due to chlorine (Cl) uptake and amide link scission in the polyamide active layer.4−6 Thus, understanding the kinetics and mechanisms of Cl uptake and amide link scission is of great © 2014 American Chemical Society

importance for the development of chlorine-resistant RO/NF membranes. In this paper, we focus on the study of Cl uptake, which has been proven to cause membrane performance degradation before amide link scission is detected.4,6−10 The current understanding of the mechanisms of Cl uptake by aromatic polyamide upon chlorination has been gained mostly through studies with model amide compounds.5,7,11−16 Briefly, chlorine uptake can occur at the amidic nitrogen (Nchlorination) or at the ring adjacent to the amidic nitrogen (ring chlorination).4,5 N-Chlorination5,11 results from the reaction k1

−CONH− + free chlorine → −CONCl− + OH− Received: Revised: Accepted: Published: 2741

(1)

October 25, 2013 January 7, 2014 February 7, 2014 February 7, 2014 dx.doi.org/10.1021/es4047632 | Environ. Sci. Technol. 2014, 48, 2741−2749

Environmental Science & Technology

Article

and pH conditions. Volume-averaged measurements were made using Rutherford backscattering spectrometry (RBS).31,32 Among other tasks, we compared volume-averaged Cl uptake to surface Cl uptake, evaluated whether volume-averaged results are consistent with the Cl uptake mechanisms proposed in the literature on the basis of studies with model compounds, and modeled Cl uptake under acidic and alkaline conditions.

where amide reacts with free chlorine, producing N-chlorinated amide and hydroxyl ion. Reaction 1 is reversible as given by k2

−CONCl− + OH− → −CONH− + free chlorine

(2)

Ring chlorination can occur from the direct reaction between the ring adjacent to the amidic nitrogen [−CONH−(C6H4)−] and molecular chlorine (Cl2) or by a mechanism referred to as Orton rearrangement (or indirect ring chlorination).5,12,15,16 The Orton rearrangement is a two-step mechanism as given by



MATERIALS AND METHODS Membranes. Experiments were performed with samples (2.5 × 5.0 cm2 coupons) of the SWC4+ RO membrane and of its polysulfone support (Hydranautics, Oceanside, CA). The coupons were thoroughly rinsed with and stored in ultrapure water (UPW, ≥18 MΩ·cm) before use. The SWC4+ membrane has an uncoated aromatic polyamide active layer, as confirmed by ATR-FTIR and RBS analyses reported elsewhere2,32 (see the Supporting Information). RBS analyses indicated that the polyamide active layer of clean, nonchlorinated SWC4+ samples had the approximate composition C0.510H0.345O0.073N0.071Cl0.0004, which is in close agreement with the theoretical composition of fully cross-linked polyamide (C0.5H0.33O0.083N0.083).1,4,21,31 Given the proprietary nature of membrane manufacturing, the reasons that account for the differences between the theoretical and experimental compositions are unknown. The trace Cl content we obtained in nonchlorinated SWC4+ membranes may be the result of controlled exposure to free chlorine by the manufacturer to enhance water permeability.33 Reagents, Chemicals, and Free Chlorine Solutions. Reagent grade or better ∼5% w/w sodium hypochlorite (NaOCl), 12.1 N hydrochloric acid (HCl), sodium hydroxide (NaOH), 15.8 N nitric acid (HNO3), and silver nitrate (AgNO3) were purchased from Fisher Scientific (Hampton, NH). All solutions were prepared using UPW. All free chlorine solutions were prepared by dilution of the ∼5% w/w NaOCl stock solution with UPW in glassware protected from light to minimize photodegradation of free chlorine. The concentration of total free chlorine (CFC) in the stock solution was measured using the standard iodometric method I (Method 4500-Cl B)34 and was determined to be CFC = 51 400 ppm (where ppm units throughout the paper refer to mg/L as Cl2). The total free chlorine concentrations in test solutions were approximated as the summation of the concentrations of HOCl and OCl− (see the speciation diagram in the Supporting Information)35 and were measured using spectrophotometric analyses as described by Lei et al.36 All procedures were performed at room temperature (20 ± 1 °C). Chlorination of Membrane Samples. Most membrane chlorination tests were performed by exposing membrane coupons to free chlorine solutions without mixing, using the following exposure conditions: (i) pH 4 and 10 each at CFC = 75, 750, and 7500 ppm and (ii) pH 5 and 7.5 each at CFC = 750 ppm. We also performed tests under mixing conditions (700 rpm with a stir bar and a magnetic stirrer) at pH 4 and CFC = 750 ppm. The chloride (Cl−) concentrations in free chlorine solutions with CFC = 750 ppm at pH 4, 5, 7.5, and 10 were determined to be 0.025, 0.024, 0.020, and 0.015 M, respectively. In general, four replicates were obtained for each chlorination condition tested. The pH of free chlorine solutions was adjusted by addition of HCl or NaOH. Membrane exposure to free chlorine was expressed in three ways: exposure time [t (h)], exposure to total free chlorine [CTFC (ppm·h) = ∫ CFC ·dt], and exposure to hypochlorous acid [CTHOCl (ppm·

−CONCl(C6H4)− + H+ + Cl− ↔ −CONH(C6H4)− + Cl 2

(3a)

and −CONH(C6H4)− + Cl 2 → −CONH(C6H3Cl)− + HCl (3b)

In the first step, an N-chlorinated amide link is dechlorinated by hydrochloric acid, resulting in the production of unchlorinated amide and Cl2. In the second step, the ring adjacent to the amidic nitrogen is chlorinated by Cl2. The relative importance and other details of how the mechanisms described above apply to Cl uptake by polyamide active layers upon exposure to free chlorine are not fully agreed upon yet. For example, while some authors6,7,13,17,18 have proposed that hypochlorous acid (HOCl) is the species responsible for Nchlorination, others5,11,19,20 have proposed that it is hypochlorite ion (OCl−). Likewise, there is no agreement on whether direct ring chlorination is an important Cl uptake mechanism,4,5,7,13,19 and the reversibility of N-chlorination by OH− is often ignored in the discussion of the effects of pH on Cl uptake.6,17,19,21 Various studies6,8,17,21−24 have quantified Cl uptake at active layer surfaces (top ≈ 7 nm) by X-ray photoelectron spectroscopy (XPS) and thus identified the factors that promote increased Cl uptake (i.e., decreasing pH and increasing exposure time and free chlorine concentration in solution). However, given that the transport of water and contaminants through RO/NF membranes is the result of partitioning and diffusion phenomena controlled by the physicochemical properties of the bulk region of the active layer,1,25,26 and not only of its surface, there is a need to characterize Cl uptake in the polyamide bulk region (i.e., the volume-averaged, not surface, Cl uptake). Current literature shows that volume-averaged Cl uptake has only been qualitatively characterized by attenuated total reflectanceFourier transformed infrared (ATR-FTIR) spectroscopy.6,9,17,21−24,27−30 Thus, the following three gaps exist in the literature regarding the study of Cl uptake by polyamide active layers upon exposure to free chlorine: (1) there is no quantitative data confirming that the Cl uptake observed at the active layer surface by XPS is representative of the volumeaveraged Cl uptake in the bulk region of the active layer, (2) there is no quantitative data confirming that the Cl uptake mechanisms that have been proposed on the basis of studies with model compounds are consistent with observations for the bulk region of the active layer, and (3) we could not find a study that modeled the kinetics of Cl uptake into polyamide active layers. Accordingly, we measured the volume-averaged Cl uptake by the fully aromatic polyamide active layer of a seawater RO membrane upon exposure to free chlorine at a range of free chlorine concentrations, exposure times, and mixing, rinsing, 2742

dx.doi.org/10.1021/es4047632 | Environ. Sci. Technol. 2014, 48, 2741−2749

Environmental Science & Technology

Article

h) = ∫ CHOCl dt]. To calculate CTFC and CTHOCl, we monitored CFC and the pH of free chlorine solutions with time and used an HOCl/OCl− acidity constant of pKa = 7.5. Rinsing of Membrane Samples after Chlorination. To stop the reaction between free chlorine and membrane samples, most chlorinated samples were thoroughly rinsed using the following sequence of solutions: UPW (30 min), 0.1 M NaNO3 at pH 10 (10−12 h), and UPW (30 min). The purpose of the 0.1 M NaNO3 solution was to rid the samples of any ionically bound Cl−, and alkaline conditions (pH 10) were used to release the fraction of chlorine reversibly bound to the membrane.5,11,13,27 For various purposes described in detail in the Results and Discussion, some samples were rinsed with UPW only; for some others, 0.1 M NaNO3 at pH 4.0−5.5 replaced 0.1 M NaNO3 at pH 10 in the sequence described above. After rinsing, membrane coupons were dried between two Whatman qualitative grade filter papers by applying fingertip pressure and stored overnight in open Petri dishes for air drying. After rinsing and before drying, some samples were also immersed in dilute alkaline AgNO3 solutions to measure the concentration of carboxylate groups in the active layer as described in detail elsewhere.32 Quantification of Volume-Averaged and Surface Chlorine (Cl) Content in Polyamide Active Layers. The volume-averaged elemental composition of polyamide active layers, including the elemental fraction of chlorine (Cl) as atom/atom (e.g., 0.061 atom/atom indicates that 61 out of every 1000 atoms are Cl atoms), was quantified using RBS as described elsewhere.31,32,37 We used a target system developed to facilitate the RBS analyses of polymeric membrane samples,37 helium ion fluences below the membrane damage threshold of 1 × 1014 He/cm2,38 an analysis area of ≈400 mm2 for any given sample, and SIMNRA v6.0639 to analyze RBS spectra. The surface Cl content in membrane samples was obtained by XPS using a beam area of analysis of 300 × 700 μm2. From our experimental settings it can be calculated that ≈95% of the XPS signal was collected from within ≈7 nm of the active layer surface or less (see Supporting Information). More specific details about our RBS and XPS analyses settings can be found in the Supporting Information. Extensive details about RBS analyses can be found elsewhere.31,32,37,39,40

Figure 1. Representative RBS spectra of nonchlorinated and chlorinated (a) SWC4+ membrane samples and (b) polysulfone support samples. All free chlorine solutions used were prepared at pH 4 with a total free chlorine concentration of CFC = 750 ppm. Symbols indicate experimental data and solid lines indicate SIMNRA simulations of the RBS spectra. (a) SWC4+ membrane samples with total free chlorine exposures of CTFC = 0 ppm·h (red squares), CTFC = 15 ppm·h (yellow circles), and CTFC = 7500 ppm·h (blue diamonds); the corresponding Cl contents measured by RBS were 0.04, 3.5, and 6.1% atom/atom, respectively. (b) Polysulfone support samples with total free chlorine exposures of CTFC = 0 ppm·h (red squares) and CTFC = 30 000 ppm·h (blue diamonds); no Cl content was detected by RBS analyses in the polysulfone support samples.

Effect of Mixing on the Volume-Averaged Uptake of Chlorine by Polyamide Active Layers. Inconsistent experimental design has been noted in the literature with respect to mixing conditions (i.e., lack or absence of mixing) during sample chlorination or during rinsing after chlorination.6,8,17,20,23,24,28−30,33,42 Thus, we evaluated whether different mixing conditions alone contribute toward differences in measured Cl uptake. The results presented in Figure 2 show that mixing conditions during sample chlorination or rinsing did not play a role in Cl uptake kinetics. Thus, we performed all chlorination and rinsing procedures in the absence of mixing. The independence of Cl uptake kinetics on mixing conditions indicates that Cl uptake is reaction-limited, not diffusionlimited. Chlorine Uptake Reversibility upon Alkaline Rinsing. Previous studies27 reported that the pH history of chlorinated samples affects the final surface Cl content in the active layer. This is likely the result of the dechlorination of N-chlorinated amide links by OH− (see reaction 2).5,11 Thus, we compared for samples chlorinated under acidic (pH 4) and alkaline (pH 10) conditions the volume-averaged Cl contents measured after acidic (pH 4.0−5.5) and alkaline (pH 10) rinsing. This comparison served to assess the extent to which the measured Cl content is affected by the pH of the rinsing solution and the relative importance of N-chlorination as a final chlorination product in the pH range studied. Figure 3 presents the Cl content measured as a function of exposure time in the polyamide active layer of membrane samples chlorinated at pH 4 and 10. We compared samples rinsed with UPW only (pH ≈5.5), UPW followed by 0.1 M



RESULTS AND DISCUSSION Measuring Volume-Averaged Chlorine Uptake by the Polyamide Active Layer. Figure 1a shows RBS spectra of samples chlorinated at pH 4 using no (CTFC = 0 ppm·h), relatively low (CTFC = 15 ppm·h), and relatively high (CTFC = 7500 ppm·h) free chlorine exposures. Analyses of the RBS spectra indicate that the Cl content (0.061 atom/atom) detected at the high exposure condition was ≈122 times higher than in the nonchlorinated sample (0.0005 atom/atom). Given that the detection limit of the Cl content measurements under our RBS analyses settings was 0.0004 atom/atom, the results in Figure 1a indicate that we can use RBS to readily monitor Cl uptake in the range of free chlorine exposures tested. Figure 1b presents RBS spectra of polysulfone support samples before and after chlorination. No evidence of Cl uptake by the polysulfone support was observed, and therefore, the polysulfone support does not contribute to the Cl signal in RBS spectra of chlorinated SWC4+ samples, consistent with the reported chemical stability of polysulfone-based membranes against free chlorine attack.6,41 2743

dx.doi.org/10.1021/es4047632 | Environ. Sci. Technol. 2014, 48, 2741−2749

Environmental Science & Technology

Article

by 0.1 M NaNO3 at pH 10 allowed for the evaluation of whether chlorine atoms covalently bound to the polyamide structure were released upon alkaline rinsing. The results in Figure 3 show that, for samples chlorinated under acidic conditions, neither the presence of a relatively high concentration of anions (i.e., nitrate) in the rinsing solution nor the pH of the rinsing solution significantly affected the volumeaveraged Cl content measured in the active layer; this conclusion was drawn from unpaired t test analyses, which showed that only two out of seven exposure times resulted in a statistically significant difference (95% confidence) in Cl content for samples rinsed under different pH conditions. For samples chlorinated under alkaline conditions, all but the lowest exposure time resulted in a statistically significantly lower (≈20%) Cl content for samples rinsed under alkaline conditions than under acidic conditions. Thus, the results indicate that (i) rinsing chlorinated samples with UPW or concentrated NaNO3 solutions ensures that ionically bound chloride is present at most at negligible concentrations in the membrane after rinsing and (ii) Cl uptake reversibility upon alkaline rinsing is negligible for samples chlorinated under acidic conditions and minor (≈20%) for samples chlorinated under alkaline conditions. Given that Cl uptake reversibility was at most minor, unless otherwise specified, we prepared all other samples using alkaline rinsing, as we needed to rid the samples of any trace of reversibly bound chlorine for additional sample analyses out of the scope of this paper. Relative Importance of N-Chlorination and Ring Chlorination as Final Chlorination Products under Acidic and Alkaline Conditions. From the evidence in the literature that only N-chlorination is reversible but ring chlorination is not,5,11,15 the conclusion from the previous section that Cl uptake reversibility is detectable for samples chlorinated at alkaline pH, but not for samples chlorinated at acidic pH, suggests that N-chlorination as a final chlorination product is significant at alkaline pH. This conclusion is consistent with that of Antony et al.30 for the ATR-FTIR analysis of chlorinated membrane samples. The ≈20% reversibility of Cl uptake observed at pH 10 in Figure 3 does not mean that only ≈20% of the Cl uptake is due to Nchlorination but rather that N-chlorination is significant enough that a measurable fraction of it is reversed by alkaline rinsing. Consistent with this assessment, Hardy et al.11 readily quantified N-chlorination at pH 9.5, and Soice et al.7 did not observe ring chlorination at pH ≥10, therefore indicating that N-chlorination is likely dominant at pH 10. Our results for Cl uptake reversibility also support the results of previous studies with model amide compounds7 and polyamide active layers27,30 that indicate that ring chlorination is dominant at acidic pH, at which Cl uptake reversibility is negligible, but not at alkaline pH; this observation is consistent with the first-order dependence on H+ of the rate-limiting step of the Orton rearrangement mechanism (reaction 3a).15,16 Do et al.6 speculated that the dominant ring chlorination mechanism in free chlorine solutions at acidic pH was direct ring chlorination; thus, we tested whether direct ring chlorination was dominant in our experiments at pH 4 by comparing the Cl uptake results at pH 4 to those at pH 5 (see Figure 4a). Despite the fact that the concentration of Cl2 at pH 4−5 is low, the Cl2 concentration at pH 4 is 10 times higher than at pH 5 (i.e., Cl2 accounts for 0.62% and 0.06% of chlorine species at pH 4 and pH 5, respectively;35 see the Supporting Information). As a result, if direct ring chlorination by Cl2 were

Figure 2. Effect of mixing on the volume-averaged chlorine content (atom/atom) measured in the aromatic polyamide active layer of SWC4+ membrane samples as a function of exposure time (t). All samples were chlorinated with free chlorine solutions at pH 4 with a total free chlorine concentration of CFC = 750 ppm. For samples labeled as “No mixing” (blue diamonds), no mixing was provided during chlorination or rinsing. For samples labeled as “Mixing-FC” (green triangles), mixing was provided during exposure to free chlorine. For samples labeled as “Mixing-rinse” (red squares), mixing was provided during rinsing. Mixing was provided at 700 rpm by a magnetic stir plate and a stir bar.

Figure 3. Effect of rinsing solution on the volume-averaged chlorine content (atom/atom) measured in the aromatic polyamide active layer of SWC4+ membrane samples as a function of exposure time (t). All samples were chlorinated with free chlorine solutions having a total free chlorine concentration of CFC = 750 ppm. Solid symbols correspond to samples chlorinated at pH 4. Empty symbols correspond to samples chlorinated at pH 10. Rinsing was performed with ultrapure water only (pH 5.5) (diamonds), ultrapure water followed by 0.1 M NaNO3 at pH 4.0−5.5 and ultrapure water (circles), and ultrapure water followed by 0.1 M NaNO3 at pH 10 and ultrapure water (triangles). All data points have error bars representing the standard deviation between 3 and 15 replicate samples. For various exposure conditions, including most at pH 10, error bars are hidden by the symbols.

NaNO3 at pH 4.0−5.5 and UPW, and UPW followed by 0.1 M NaNO3 at pH 10 and UPW. The comparison between rinsing with UPW only and UPW followed by 0.1 M NaNO3 at pH 4.0−5.5 allowed for the evaluation of whether rinsing with UPW only was sufficient to rid the membrane of ionically bound Cl−. The comparison between rinsing with UPW followed by 0.1 M NaNO3 at pH 4.0−5.5 and UPW followed 2744

dx.doi.org/10.1021/es4047632 | Environ. Sci. Technol. 2014, 48, 2741−2749

Environmental Science & Technology

Article

Figure 4. Effect of pH on the volume-averaged chlorine uptake by the polyamide active layer of SWC4+ membrane samples chlorinated at pH 4 (diamonds), pH 5 (circles), pH 7.5 (squares), and pH 10 (triangles). All chlorine solutions had a total free chlorine concentration of CFC = 750 ppm. Chlorine content in the active layer (atom/atom) is plotted as a function of (a) exposure time (t) and (b) exposure to hypochlorous acid (CTHOCl). All continuous lines correspond to simulation lines calculated using eq 8 and E = 0.952, 0.761, 0.364, and 0.224 at pH 4, 5, 7.5 and 10, respectively. We used the concentrations of nitrogen (0.071 atom/atom) and chlorine (0.0004 atom/atom) in nonchlorinated samples as the initial concentrations of −CONH− and [Cl]. The dashed line was calculated using d[Cl]/dt = r1 − r2 − r3 with k3 = 5 × 1011 M−2·s−1 and a chloride concentration of 0.020 M.

Figure 5. Effect of total free chlorine concentration (CFC) on the volume-averaged chlorine uptake by the aromatic polyamide active layer of SWC4+ membrane samples as a function of exposure time (t). Chlorination experiments were performed at (a) pH 4 and (b) pH 10. The inset shows that when chlorine uptake is plotted as a function of total free chlorine exposure (CTFC), all data sets for one given pH collapse into one single trend. All lines correspond to simulation lines calculated using eq 8. At pH 4, E = 0.625, 0.952, and 0.995 at CFC = 75, 750, and 7500 ppm, respectively. At pH 10, E = 0.023, 0.224, and 0.81 at CFC = 75, 750 and 7500 ppm, respectively. We used the concentrations of nitrogen (0.071 atom/atom) and chlorine (0.0004 atom/atom) in nonchlorinated samples as the initial concentrations of −CONH− and [Cl].

the dominant Cl uptake mechanism at pH 4, then Cl uptake would be noticeably lower at pH 5 because direct ring chlorination is dependent on the concentration of Cl2.4,6 Since Figure 4a shows that there is no significant difference in Cl uptake in the pH 4−5 range, we conclude that the dominant Cl uptake mechanism at pH ≥4 is not direct ring chlorination by Cl2 but Orton rearrangement. This conclusion is consistent with results for model amide compounds and powdered polyamides, which showed that ring chlorination of amide compounds was dependent on the presence of the N−H bond.7,13 Given that Cl2 is known to cause direct ring chlorination under acidic conditions,5,12,15 direct ring chlorination may become more dominant at pH