Electrodialysis for Recovering Salts from a Urine Solution Containing

Feb 21, 2006 - A laboratory electrodialysis stack from Elektrolyse Projekt BV ..... drinking water showed that pesticides were not permeated in this p...
0 downloads 0 Views 248KB Size
Environ. Sci. Technol. 2006, 40, 2414-2420

Electrodialysis for Recovering Salts from a Urine Solution Containing Micropollutants WOUTER PRONK,* MARTIN BIEBOW, AND MARKUS BOLLER Eawag, Swiss Federal Institute of Aquatic Science and Technology, CH-8600 Du ¨ bendorf, Switzerland

Electrodialysis was investigated for the separation of micropollutants from nutrients in anthropogenic urine. In a continuously operated process, the nutrients were concentrated up to a factor of 3.2. The concentration factor was limited by water transport across the membrane. Water transport was caused by osmosis and electroosmosis, and a model was developed to describe these phenomena. The removal of several spiked micropollutants was investigated in continuous electrodialysis experiments. Ethinylestradiol was removed completely during the whole operating period. Diclofenac and carbamazepine were initially retained, but limited permeation (510%) occurred after longer operating times (90 days). Retentions of propranolol and ibuprofen were also high initially, but substantial breakthroughs occurred during extended operation. Considerable adsorption on the membranes was observed for all compounds. The permeation mechanism of several compounds appears to depend on the adsorbed amount on the membrane, which indicates that partitioning and diffusion mechanisms play an important role in the permeation transport. Partial desorption occurred in leaching experiments with polarity reversal, and almost quantitative desorption was observed after incubation of the membranes with Filter Count Gel Solution. Because environmental concentrations are much lower than the concentrations spiked here, it can be anticipated that operation without significant permeation is possible in practice during extended periods of time.

Introduction It is current practice in municipal wastewater treatment to collect several waste streams in sewers and transport them to centralized wastewater treatment plants. Advantages of this concept include the facilitation of transport, (e.g., several diluted streams enable the transport of more concentrated streams) and the benefits of scale in the treatment. Disadvantages of this concept include the dilution of micropollutants and nutrients. This dilution hinders the effective removal of micropollutants and the effective recovery of nutrients. Micropollutants in wastewater can include a wide range of compounds from personal care products to pesticides and pharmaceutical compounds (1). The environmental and ecotoxicological effects of these compounds remains largely unknown (2, 3), and it is therefore important to prevent their * Corresponding author e-mail: [email protected]; phone: 41-1-823-5381; fax: 41-1-823-5389. 2414

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 7, 2006

diffusion. Most of these compounds are only partly degraded in integrated biological wastewater treatment plants (4), while other compounds appear to be almost completely persistent (5). Source control and treatment of separate streams such as urine can provide more sustainable scenarios. It is remarkable that the largest part not only of the pharmaceuticals but also of the nutrients in wastewater stem from urine: about 80% of the nitrogen and 50% of the phosphorus in wastewater originate from urine (6). Thus, separate collection and treatment of urine could provide an attractive alternative. Since phosphorus is a depletable raw material, its recovery would result in a more sustainable scenario from the viewpoint of material conservation. Nitrogen is not a depletable raw material, but it is present in urine in the form of ammonia or urea, which is a valuable material for use as a fertilizer. This would replace the chemical synthesis of these compounds, which requires energy and auxiliary materials. Therefore, an ecological advantage of nitrogen recovery from urine can be attained if this scenario is associated with a lower ecological burden than the conventional one (7). The nutrients in urine reflect the components necessary for plant growth, which makes it suitable as a fertilizer in agricultural applications (8, 9). However, the concentrations of nutrients in urine are low compared to commercial fertilizers. The transport and storage volumes of untreated urine are consequently relatively high, which results in higher costs. Electrodialysis (ED) processes can be applied for removal of salts from wastewaters (10-12), for the recovery of products such as organic acids (13, 14), for the treatment of drinking water (15), and numerous other applications. In principle, organic compounds such as micropollutants can be separated from the product if they are retained by the membrane. The apparent molecular weight cutoff of electrodialysis membranes in the separation of natural organic matter was reported to be around 200 D (16). Since the molecular weight of pharmaceutical and hormonal compounds lies around or above this value, perspectives exist for the removal of these compounds. If full rejection of micropollutants were to take place, the urine treatment would result in a concentrate stream containing the nutrients (salts) and a diluate stream containing the micropollutants. An additional advantage for the use of ED in urine treatment is that the membranes provide a barrier for the removal of hygienic risk factors such as bacteria, viruses, and pathogenic proteins which can occur in urine. Electrodialysis has been studied for the removal of ions from wastewaters (17), and combinations with other unit operations were also examined (12). However, the separation between inorganic compounds (nutrients) and micropollutants by electrodialysis has not been previously reported. The present study investigated the concentration of nutrients as well as the retention of micropollutants from urine using batch and continuous electrodialysis processes. In principle, this process could also be used for the separation of micropollutants and nutrients from other waste streams including wastewater treatment plant effluent, and this paper also could serve as an input for further investigations within the field.

Experimental Section Synthetic and Natural Urine. Experiments were carried out with both synthetic and natural urine. The natural urine was obtained from the facilities of EAWAG, Du¨bendorf (Switzerland), which are equipped with no-mix toilets and 10.1021/es051921i CCC: $33.50

 2006 American Chemical Society Published on Web 02/21/2006

FIGURE 1. (Left) schematic presentation of ion migration in two cell pairs of an electrodialysis stack, and (right) an ED setup in continuous experiments (CEM ) cation exchange membrane, AEM ) anion exchange membrane)

waterless urinals. Details of this system have been described before (18). No special measures were taken to prevent urea hydrolysis, which implies that the urea had been fully hydrolyzed by spontaneous microbiological activity in the storage tank. The composition of the synthetic urine was based on the average composition of natural urine after hydrolysis (19). The composition and main properties of both solutions are shown in Table 1, Supporting Information. Comparison of the concentrations of dissolved compounds in both solutions shows that the stored natural urine is somewhat diluted in comparison to the synthetic variant, which reflects its composition at source. This dilution may be caused by occasional flushing and cleaning operations in the urine collecting system at Eawag. Before electrodialysis, the natural urine solutions were permeated through a polypropylene microfiltration membrane (pore size 0.2 µm) in order to protect the membrane unit from suspended solids. Micropollutants. Micropollutants were selected as described before (20). The compounds selected (ethinylestradiol, diclofenac,carbamazepine propranolol, and ibuprofen) all are excreted in urine to more than 65%. Experiments were carried out with radioactively labeled and nonlabeled compounds. Some properties of the (nonlabeled) compounds are shown in Table 2, Supporting Information. The characteristics of the labeled compounds are summarized in Table 3, Supporting Information. In the experiments with nonlabeled compounds, these substances were first dissolved in a concentrated stock solution (2 mM) and then added to the urine solution up to a concentration of around 10 µM. Analytical Methods. Chloride, sulfate, and phosphate were analyzed with ion chromatography (Compact IC type 761, Metrohm, Herisau, Switzerland, with column IonPac AS12A, Dionex, Sunnyvale, CA). Potassium, sodium, calcium, and magnesium were determined with the ICP-OES (inductively coupled plasma-optical emission spectrometer, Spectro Analytical Instruments, Kleve, Germany). COD was measured with HACH test tubes (HR, test tube 435, HACH Company, Loveland, CO). Total ammonia was determined photometrically (reaction with bromocresol purple) by flow injection analysis (FIA, Ismatec AG, Glattbrugg, Switzerland) and urea was measured after hydrolysis to ammonia using urease (Merck 16493). Alkalinity was determined by titration analysis. Nonlabeled pharmaceuticals and the hormones were measured with HPLC using a Chromcart HPLC column with 125/4 Nucleosil 100-5 C18 packing (125 mm × 4 mm), and a CC 8/4 Nucleosil 100-5 C18 precolumn with UV and fluorescence detection. Labeled compounds were analyzed using a HewlettPackard TRI-CARB 2200CA liquid scintillation analyzer.

Samples (1 mL) were mixed with 9 mL of Lumagel Gold scintillation counting gel (LUMAC-LSC B. V.), and measured in triplicate. Desorption of radio-labeled micropollutants from membranes was measured by incubating a defined area of membrane (1 cm2) in a scintillation-counting vial along with 10 mL of Packard Filter Count Gel Solution scintillation liquid (complete LSC cocktail for counting membrane filters, Packard, Meriden, CT) and the liquid was counted on a Hewlett-Packard TRI-CARB 2200CA scintillation counter. Electrodialysis Equipment. A laboratory electrodialysis stack from Elektrolyse Projekt BV (Bussum, Holland) with an effective area of 49 cm2 per membrane sheet was used for all experiments. Electrodialysis membranes by Mega a.s., Prague, Czech Republic, were used, which are claimed to have a low tendency of fouling by organic compounds. The ion-exchange groups are tertiary amine groups for the anionselective membrane (Ralex AMH-PES) and sulfonate groups for the cation-selective membrane (Ralex CM-PES). The thickness of the membranes (dry) was 0.45 and 0.55 mm, their resistance was less than 9 and 8 Ω.cm2, and their ionexchange capacity was 2.2 and 1.9 mval/g (0.20 resp. 0.16 mval/cm2) for the cation- and anion-exchange membranes, respectively. The binder of the membrane was polyethylene, and the textile fibers of the membrane reinforcement were made of polyethersulfone (all data derived from manufacturer fact sheet). The experimental setup consisted of the ED stack with recirculating streams of concentrate, diluate, and electrode rinse (see Figure 1). Recirculation vessels and hoses were thermostated to maintain a constant temperature of 25 °C in all experiments. Recirculation flow rates in all compartments were set at 65 L/h. Both batch and continuous experiments were carried out. At the start of each electrodialysis experiment, the circuits of concentrate and electrode rinse were filled with NaCl solutions of 0.01 and 0.2 M, respectively. In the batch experiments, the diluate compartment was filled with stored natural urine at the start of the experiment. In the continuous experiments, stored natural urine was added continuously to the diluate circuit; diluate was removed by an overflow, thus maintaining a constant volume in this circuit. No feed was applied to the concentrate compartment: the concentrate overflow was generated by water transport through the membranes (see Figure 1). The volume of each circuit was adjusted to 1000 mL, except for the batch experiment with labeled ethinylestradiol, which was carried out with 500 mL in all circuits. All batch experiments were carried out with two cell pairs and a total voltage of 3.6 V. The continuous experiments were carried out with five cell pairs and a total voltage of 9 V. Highly accurate measurements of electric current (error