Organic matter composition more important than ... - ACS Publications

Article Cite This: Ind. Eng. Chem. Res. 2018, 57, 3742−3752

pubs.acs.org/IECR

Organic Matter Composition More Important than Concentration in Ion Exchange Demineralization of Different Water Qualities for the Production of Steam Evelyn De Meyer,*,† Bart Peeters,‡ Marjolein Vanoppen,† Kim Verbeken,§ and Arne R. D. Verliefde† †

Particle and Interfacial Technology Group, Department of Applied Analytical and Physical Chemistry, Faculty of Bioscience Engineering, Ghent University, Coupure Links 653, 9000 Ghent, Belgium ‡ Environmental Department, Monsanto Europe N.V., Scheldelaan 460, 2040 Antwerp, Belgium § Department of Materials, Textiles and Chemical Engineering, Faculty of Engineering and Architecture, Ghent University, Technologiepark 903, 9052 Zwijnaarde, Belgium S Supporting Information *

ABSTRACT: Fresh water becomes a limited resource in the industry. In order to help chemical industries use other water sources and close their water cycle for the production of steam, a well-founded insight on the challenges and possibilities of switching from one specific water quality to another is needed. A case study for Monsanto Europe N.V. was carried out, but the main findings hold for many more applications, as many industries struggle with water scarcity and feel the need to reuse wastewater. Besides demineralization by ion exchange (IEX), the total organic carbon (TOC) concentration, composition, and the formation of organic acids under boiler conditions were investigated for two different water qualities (Antwerp tap water and wastewater after reverse osmosis (RO) treatment). The comparison included the effect of TOC composition on its removal by IEX and the potential corrosiveness of TOC compounds. Despite tap water showing a more efficient and higher TOC removal compared to RO permeate (93% and 57%, respectively), tap water led to more organic acid formation under boiler conditions. Including the composition of the organic matter in TOC corrosiveness assessment may offer multiple advantages, not in the least economic benefits, when less treatment is required in order to meet the TOC limit value.

1. INTRODUCTION Increasing fresh water shortages demonstrate the importance of the use of alternative water resources for the production of demineralized water. In Flanders and the southern parts of The Netherlands, including the Ghent, Antwerp, and Zeeland seaports with highly developed industry, there is an increasing groundwater salinization, due to both natural and anthropological influences.1 In order to ensure industrial water production in an economic and ecological efficient manner in the future, industry needs to diversify its water sources, shifting from the traditional groundwater to alternative sources such as (brackish) surface water and reuse of municipal or industrial wastewater.2,3 The improvement of water resource management is also included in the 2030 Agenda of Sustainable Development Goals of the United Nations, where wastewater is referred to as “the untapped resource”.4 These alternative water streams, however, are often characterized by a lower quality and higher quality variation than groundwater sources, for example, with respect to their total organic carbon (TOC) concentration and composition, next to variations in inorganic composition.5 The presence of potentially harmful organic matter in the reused water after a specific demineralization treatment is of most concern for companies with steam-water cycles. The residual amount of organic matter is of interest as TOC is a © 2018 American Chemical Society

precursor for the formation of organic acids under boiler conditions (high temperature and pressure) due to hydrothermolysis.6 As these organic acids are known to decrease pH and to induce or enhance corrosion mechanisms in steam, strict TOC guidelines are put in place. During the first condensate corrosion (which takes place in a steam-water cycle where the steam is converted to the first initial water droplets, mostly due to a drop in pressure), these acids provide a local pH drop, which is detrimental for the materials used in the steam-water cycle.7−9 The remaining TOC concentration after demineralization is therefore often limited to values below 100 ppb, in order to meet the guideline of the VGB PowerTech e.V., the international technical association for generation and storage of power and heat.10 However, the VGB TOC guideline does not take into account the effect of TOC composition on potential corrosiveness, as it only limits the overall TOC concentration. However, it has been observed that different organic compounds may lead to different amounts and types of organic acids.6 Characterization of the organic matter is thus Received: Revised: Accepted: Published: 3742

December 7, 2017 February 26, 2018 February 26, 2018 February 26, 2018 DOI: 10.1021/acs.iecr.7b05059 Ind. Eng. Chem. Res. 2018, 57, 3742−3752

Article

Industrial & Engineering Chemistry Research

and there is no worse effect on the organic acid formation during hydrothermolysis. Therefore, this study performed a critical assessment on the removal of TOC from the RO effluent by a complete IEX demineralization unit and a clear comparison with the TOC removal from tap water. In order to investigate the removal of natural organic matter, from two completely different water qualities, throughout a complete IEX desalination treatment, a world-unique full lab-scale demineralization unit was designed and built, consisting of the five different parts (WAC−SAC−degasser−WBA−SBA). Both demineralization and TOC removal were studied as opposed to many studies solely focusing on TOC removal. Due to the design of the lab-scale IEX demineralization unit, the effect of each building block on TOC removal could be investigated. Mass balances were included to determine the regeneration efficiency of organic matter on the WBA and SBA between successive cycles of demineralization. This made an investigation possible on the effect of different water qualities on the resin capacity. In addition, LC-OCD characterization of the different feed waters allowed one to link TOC removal to TOC composition. To investigate the possible degradation of the organic matter and to give an indication of their contribution to corrosion, both feedwater streams (i.e., tap water and RO permeate produced from biologically treated wastewater) and their respective treated streams by IEX were subjected to hydrothermolysis in a lab-scale flow-through boiler, simulating reallife boiler conditions. The LC-OCD method was used to characterize the organic matter of both feedwater streams in an attempt to link the initial TOC concentration and characterization to corrosiveness, measured as organic acid formation.

fundamental, since each organic compound is vulnerable to hydrothermolysis to a different extent. In addition to the polarity rapid assessment method (PRAM) and charge chromatography, liquid chromatography in combination with organic carbon detection (LC-OCD) is a well-known TOC characterization method.11−13 Nevertheless, to date, no literature is available linking LC-OCD chromatograms with the formation of organic acids. In general, organic matter can be removed by various techniques, typically based on thermal or physical separation and chemical degradation. Some potential water treatment techniques, which govern partial or complete TOC removal, are membrane distillation (MD), reverse osmosis (RO), ion exchange (IEX), activated carbon, and ozonation (although the latter may result in the formation of undesired byproducts).14−19 The choice for a particular method for demineralization in industry depends on a number of factors, of which cost is often the most important. When TOC removal is required, the choice of treatment technology depends on the initial TOC concentration and composition and the desired end water quality. When the aim is the production of high end water quality, in other words the production of ultrapure water, IEX is one of the most appropriate methods as a final treatment step, concerning both demineralization and TOC removal.20 Nevertheless, despite the large number of studies on TOC removal by IEX described in the literature, there are still many gaps in knowledge that require the necessary attention. For example, most IEX research on TOC removal is performed on the removal of organic compounds by weak or strong anion exchange resins alone, without considering the full IEX demineralization unit consisting of a weak and strong cation exchange column (WAC and SAC), followed by a CO2 degasser and a weak and strong anion exchange column (WBA and SBA).21−24 In addition, most studies investigate the removal of model components as a simulation of organic matter.25 This kind of research is fundamental for the basic knowledge on IEX mechanisms but is a simplification of the complex wastewater composition in reality. In addition, the condition in which the ion exchange resins are used is important to achieve full demineralization. The WAC and SAC need to be in the H+-form and the WBA and SBA need to be in the OH−-form for the production of water after ion exchange, instead of salts which are formed when the resins are in the Na+- and Cl−-form, respectively. Studies on TOC removal, including those for the production of drinking water, using a WBA or SBA in the Cl−-form are therefore less representative.26,27 In a complete IEX demineralization unit, the proceeding treatment with the WAC and SAC has a major impact on the combined TOC removal mechanisms in the WBA and SBA, as the pH decreases significantly after the CO2 degasser and a difference in acidity has a large effect on TOC affinity for the anion resin. Among other companies, Monsanto Europe N.V. (located in the harbor of Antwerp, Belgium) wants to decrease the intake of tap water for the production of demineralized water, which is fed to the steam boilers for the production of steam and electricity. Here, the goal is to reuse a large fraction of the company’s biologically treated wastewater, after ultrafiltration (UF) and RO. However, the effect of potentially harmful organic matter in the reused water after RO treatment (and consequent demineralization by IEX) in the steam-water cycle is of most concern. The switch to another water type can be made only when the TOC guideline after IEX treatment is met

2. MATERIALS AND METHODS 2.1. Characterization and Analysis of Feedwater Streams. The two feedwater streams under investigation were Antwerp tap water and RO permeate. The tap water originates from surface water in the Meuse (the Albert channel). The surface water is treated with flocculation, flotation, sand filtration, granular activated carbon, UV light, and chlorine by the Belgian drinking water company WaterLink. The produced tap water is distributed to households and industrial companies across Antwerp. The RO permeate was produced in the wastewater treatment plant (WWTP) at Monsanto Europe N.V. Here, the industrial wastewater of 8 chemical plants (for the production of plasticizers, rubber chemicals, polymers, adhesives, and herbicides) is treated in a conventional activated sludge WWTP. Through successive anoxic (denitrifying) and aerobic conditions, the contaminants are biologically removed from the water. The sludge age is about 30 days, and the hydraulic retention time amounts to 1 day. In a pilot infrastructure, tests were executed on the biologically treated wastewater to evaluate the technical feasibility of consecutive UF and RO in order to reuse part of the currently discharged water. The intended purpose is the production of demineralized water and steam in the central utilities (boiler unit) on the Monsanto site. In the UF installation, a standard Zeeweed 1500 UF membrane from Suez was used, and in the RO installation, AG4040LF low fouling membranes from GE Water were used. A more in-depth water quality analysis of the two feedwater streams (RO effluent versus tap water) was conducted, to obtain clear insights in the composition of the organic matter in the different feed streams, as well as a clear insight in the ionic 3743

DOI: 10.1021/acs.iecr.7b05059 Ind. Eng. Chem. Res. 2018, 57, 3742−3752

Article


Figure 1. Schematic view of the lab-scale setup of the demineralization unit, in which a weak acid cation (WAC) resin is followed by a strong acid cation (SAC) resin, a degassing unit, a weak basic anion (WBA) resin, and strong basic anion (SBA) resin. The TOC concentration and the conductivity of the demineralized water were measured continuously.

2.2. Lab-Scale Ion Exchange Demineralization Setup. The demineralization unit of Monsanto Europe N.V. consists of a “classical” ion exchange demineralization, whereby a WAC resin is followed by a SAC resin, a degassing unit, a WBA resin. and a SBA resin. This sequence was maintained in the lab-scale setup, and a schematic overview is shown in Figure 1. In order to remove the inorganic carbon as CO2 in the degassing unit, N2 gas was purged through the column. At the outlet of the SBA column, the TOC concentration and the conductivity of the demineralized water were measured continuously with a Sievers 900 online TOC analyzer (Colorado, US) and a WTW conductivity probe LR 925/01-P IDS (Weilheim, DE), respectively. The Sievers 900 online TOC analyzer had a sampling interval of 4 min. Dimensioning of the lab-scale setup was also based on the demineralization unit of Monsanto Europe N.V. The resin types used in the lab-scale setup and their volumes are provided in Table S4. All resin types were supplied by Lanxess Lewatit (Leverkusen, DE). Each column was made of glass and had an integrated frit with porosity P0 at the bottom. The internal diameter of the columns was 2.6 cm. A peristaltic WatsonMarlow 530S pump (Watson-Marlow, US) was used to transport each solution to the setup. 2.3. Assessment of the Background TOC Concentration of the Lab-Scale Demineralization Setup. TOC contamination is one of the most problematic issues during experiments which intend low TOC concentrations. The TOC contamination by the environment and the setup (such as

composition of both water streams. The ionic analysis was performed with the 930 Compact IC Flex (Metrohm, CH). A Metrosep C6-150/4.0 column and a Metrosep C4 guard/4.0 column were used for cation analysis, with 1.7 mM HNO3 and 1.7 mM DPCA as eluent solution. The anion analysis was done with a Metrosep A Supp 16-250/4.0 column and a Metrosep A Supp 16 guard/4.0 column, with an eluent solution of 7.5 mM Na2CO3 and 0.75 mM NaOH. Cation and anion analyses after the SAC and SBA, respectively, were done with the same method. In addition, size-exclusion chromatography (LCOCD) for both water samples was performed to assess potential differences in dissolved organic carbon (DOC) composition of the two water streams. The feedwater streams were sampled and stored in 1000 L containers, in order to minimize variation in both concentration and composition over consecutive experiments. The composition of the feedwater streams is provided in Table S1. In general, tap water is subjected to seasonal variation in concentration and composition; the TOC values for the year 2016 are given in the Supporting Information, as well as the average ion concentration of tap water in the years 2012 until 2016, with the minima and maxima, in Figure S1 and Table S2, respectively. The RO treatment at Monsanto N.V. was in a test phase only for several months; therefore, no information is available for the variation in concentration and composition during a whole year. Nevertheless, the variation in RO permeate water quality was determined during each week before sample was taken. The results are shown in Table S3. 3744


Article

Industrial & Engineering Chemistry Research tubing and glassware) was limited, as a stable TOC background value as low as 50 ppb TOC was obtained when flushing the whole setup with DI water. 2.4. Operation and Regeneration Parameters. In operation mode, the empty bed contact time of the water streams in the SAC and SBA column was 3 min, resulting in a flow rate of 16.30 m/h. After each normal operation cycle, the resins were exhausted and hence needed a regeneration. Regeneration was carried out in counter flow compared to the normal operation mode. For the regeneration of the resins, the solutions had the same concentrations and bed volumes as applied at the demineralization unit of Monsanto N.V. The cation columns (WAC and SAC) were regenerated with a 4.5% HCl solution for 40 min at an upward velocity of 2.60 m/h, resulting in a contact time of 26 min and 1.55 bed volumes (BV) of regenerant. The anion columns (WBA and SBA) were regenerated with a 2% NaOH solution for 20 min at an upward velocity of 5.76 m/h, resulting in a contact time of 14 min and 1.43 BV of regenerant. At the end of the regeneration, all columns were flushed with demineralized (DI) water for 40 min at a flow rate of 5.20 m/h (equal to 3 BV). Here, the BV is the total amount of WAC and SAC or WBA and SBA, respectively. When stated, double volumes of the regeneration solutions were used in order to investigate the effect on the regeneration efficiency. For each experiment (one with tap water and one with RO permeate), new resins were used. The resins were conditioned prior to use by performing five single regenerations. In addition, for the WBA and SBA, an acid regeneration was conducted in order to remove any residual TOC, possibly left behind from the resin production process. This was followed by a double caustic regeneration. To examine the proper functioning of the resins, the baseline of the setup was examined, compared to the background TOC level in the lab-scale setup. The background TOC concentration was determined by flushing DI water through the complete demineralization setup. If the baseline gave sufficiently low TOC values in the effluent (50 ppb), the resins were double regenerated one last time before starting the experiments. During an ion exchange experiment, breakthrough was defined as the leakage of the first component (being TOC or sodium, from the SBA or SAC, respectively). Both TOC and conductivity were measured online at the outlet of the setup after the SBA column, and the experiment was stopped when a certain threshold was reached. The TOC threshold value was set at 100 ppb (±15 ppb). After each breakthrough experiment, a regeneration of the resins was carried out as described above, and a careful TOC mass balance was performed. The resulting TOC removal or adsorption efficiency of the WBA and SBA was calculated, as well as whether the mass balance was closed (in other words, if all TOC adsorbed was also coming off the resins during regeneration, in order to check if the regeneration was complete). By integrating the TOC breakthrough curve, the nonadsorbed TOC amount could be calculated. The integration was done by fitting a polynomial to the TOC breakthrough curve in Excel. When that amount is subtracted from the total TOC load (the incoming TOC concentration multiplied by the total volume of treated water), the adsorption efficiency can be determined by eq 1.

relativeadsorptionefficiency = =1−

adsorbed TOC total TOCload

nonadsorbed TOC total TOCload

(1)

This method determines the experimental adsorption efficiency (based on the volume treated at the breakthrough point), which will be lower than the theoretical capacity of the resin, as stated by the manufacturer (which is based on the full use of the ion exchange capacity). Some complex organic matter interacts strongly or irreversibly with the resins; these compounds might only be removed by application of a brine treatment.28 In order to determine the reversibility of the TOC adsorption and to investigate whether a normal regeneration procedure was adequate to remove the adsorbed TOC, the regeneration efficiency was calculated. The released TOC after regeneration was calculated by measuring the TOC concentration in the combined regeneration and backwash solution, taking into account the total volume. The regeneration efficiency can be determined by eq 2. regenerationefficiency =

released TOC adsorbed TOC

(2)

2.5. Detailed, Longer-Term Investigation of the TOC Removal for the Two Water Streams. From the previous experiments, the time before exhaustion of the resins could be determined. This allowed one to set up longer-term experiments with several cycles of normal operation and regeneration. Two long-term experiments (one for each feedwater stream) were then conducted to determine the (reversibility of the) TOC removal by ion exchange on the long-term. Nine cycles were carried out (with TOC mass balances constructed after every run and regeneration) to assess the effect of longer-term operation on the resin capacity. The TOC concentration and conductivity were monitored continuously in the effluent of the SBA column, and samples for Na+, Ca2+, Cl−, and other ion analyses were taken at the start and the end of each experiment. Cation analyses were performed on the effluent of the SAC column, and anion analyses were performed on the effluent of the SBA column. For tap water, nine identical cycles were carried out. For the RO permeate, the initial execution of the experiments was altered, due to the unexpected results of the breakthrough curve (see further), in order to provide more information about possible TOC removal. First, two identical cycles followed the breakthrough experiment, and second, a mixture of tap water and RO permeate, at five different ratios (0.1, 0.2, 0.3, 0.5, and 0.75), was treated with the demineralization setup. The ratio indicates the volume of tap water relative to the total volume. For each ratio, the total TOC concentration was measured and the percentage of low molecular weight neutral compounds (LMWN) was calculated with eq 3. %LMWN [LMWNROpermeate] × (1 − ratio) + [LMWNtap water] × ratio = [TOCROpermeate] × (1 − ratio) + [TOCtap water ] × ratio × 100%

(3)

The concentrations of LMWN in the tap water and RO permeate were determined by the LC-OCD analysis. 2.6. Simulation of Boiler Conditions in a Lab-Scale Setup. In order to determine the corrosiveness of each water 3745


Article


Figure 2. Schematic overview of the lab-scale setup in order to simulate boiler conditions of 500 °C and 60 bar.

results from the LC-OCD analysis, with the chromatograms, can be found in Figures S2 and S3. From the LC-OCD analysis, it was obvious that the DOC composition of tap water and RO permeate was completely different. The RO permeate, which had a DOC concentration of only 691 ppb, had a lower amount of biopolymers, humic substances, building blocks, and low molecular weight acids but had twice as many LMWN, compared to tap water which had a higher DOC concentration of 1.259 ppm. For the RO permeate, about 85% of the dissolved DOC components was LMWN, whereas this was only 20% for the tap water. The DOC measurements based on the IR detector after UV degradation of the carbon in the LCOCD analysis gave a slightly higher organic carbon concentration for both tap water and RO permeate compared to the measurements done with the Sievers online TOC analyzer, although the Sievers is also based on UV degradation of the organic carbon, in combination with persulfate (values in Table S1). Two possible explanations for this observation could be the contamination due to storage and transport or the fact that the Sievers online TOC analyzer makes use of a CO2selective membrane with a conductometric detection. Breakthrough experiments were conducted for both tap water and RO permeate. During these experiments, the conductivity of the produced demineralized water (the effluent of the SBA column) was lower than 2 μS/cm during the whole runtime. In addition, the concentrations of the most important cations (Na+ and Ca2+) after the SAC and anions (Cl−) after the SBA were lower than 0.5% of the initial concentration. Therefore, the TOC concentration measured after the SBA was used as the parameter to investigate breakthrough of the resins, as the TOC value always showed breakthrough first compared to the other parameters. In order to define a limit value for the TOC breakthrough, the VGB guideline for TOC was considered at a TOC concentration of 100 ppb. The breakthrough curves for tap water and RO permeate are shown in Figure 3. A rather normal breakthrough behavior can be observed for tap water, in contrast to almost immediate breakthrough for the RO permeate. The TOC breakthrough curve for tap water follows the general trend of an S-shaped breakthrough behavior, as seen in the literature.22,29 Despite the lower

quality in terms of organic acid formation, boiler experiments were performed. The investigation of the thermal stability, or in other words the susceptibility to thermolysis, of the different water streams was done in a mini-boiler. This lab-scale boiler consisted of 4 stainless steel 316 components (Swagelok, US), moreover a duplicate of SS-800-R-6 and SS-600-6-1. Heat transfer (500 °C) to the lab-scale boiler was accomplished by a Fluidized Sand Bath (FSB-3) containing fine aluminum oxide, which was fluidized by ambient air (Omega, US). A 4838 Parr temperature controller (Parr Instrument Company, US) regulated the FSB temperature, and the temperature feedback was provided by a thermocouple. In order to remove oxygen from the glass vessel, containing the desired solution, 1 ppm carbohydrazide was added. In addition, argon was sparged continuously into the glass vessel to prevent intrusion of residual oxygen. The oxygen concentration was measured at the start, in the middle, and at the end of each experiment with the WTW multimeter multi 3420 (VWR, US). In this way, the oxygen concentration during the experiments was kept below 20 ppb. From the glass vessel, the desired feed solution was sent through the stainless steel lab-scale boiler by a HPLC pump at a fixed flow rate. The residence time of the produced steam in the mini-boiler was 1.24 s. Upon leaving the mini-boiler and the FSB, the solution was immediately cooled by submerging the stainless steel tubing in water at room temperature. The back pressure (60 bar) was controlled by an adjustable relief valve (BPR) (Swagelok, US). The formation of organic acids was determined by performing an anion analysis with the 930 Compact IC Flex, as described above. The full lab-scale setup is shown in Figure 2. In order to establish a good comparison, new volumes of both feed waters (tap water and RO permeate) were again treated by ion exchange for 75 min. In this way, both feed waters had the same period of treatment, independent of the TOC breakthrough value.

3. RESULTS AND DISCUSSION 3.1. Difference in Breakthrough Behavior Linked to the TOC Composition of the Water Qualities. The detailed 3746


Article


Breakthrough Curves for Tap Water. The tap water was demineralized with the lab-scale IEX setup during nine successive cycles (each consisting of a normal operation and a regeneration mode), whereby the TOC concentration and the conductivity of the treated water were continuously measured after the SBA column. In addition, cation (after the SAC column) and anion (after the SBA column) analyses were conducted at the start and end of each experiment. For the sake of clarity, only the TOC breakthrough curves for runs 2, 4, 6, and 8 are given in Figure 4. The graphs for the other experiments, including the conductivity measurements, can be found in Figure S4.

Figure 3. TOC breakthrough curves for tap water and RO permeate. A TOC concentration of 100 ppb was set as the breakthrough point.

TOC concentration in the RO permeate, compared to tap water, the TOC compounds present in the RO permeate appeared to have a very low affinity for the ion exchange resins. Closer investigation of the LC-OCD analysis gave a plausible reason for this phenomenon. As the largest part of TOC in the RO permeate was LMWN, the TOC affinity with the charged IEX resins was indeed expected to be low. Removal of neutral organic compounds is established by a physical adsorption behavior rather than ionic interactions.22 Therefore, early breakthrough was observed for the RO permeate, while the tap water contained less LMWN and more charged TOC compounds (humic and fulvic acids), resulting in a more gradual breakthrough, in relation to the exhaustion of the resins. According to the LC-OCD analyses, tap water had a LMWN concentration of 248 ppb, whereas for the RO permeate a value of 586 ppb was measured. Despite the presence of LMWN in tap water, these compounds did not lead to immediate breakthrough in contrast to the RO permeate. These findings suggest that the LMWN present in tap water could differ from the ones present in the RO permeate, leading to a different adsorption mechanism, as still some LMWN was retained by the resins during the tap water breakthrough experiments. TOC removal by the resins was on average 93% and 57% for tap water and RO permeate, respectively. This again indicates that the TOC compounds present in tap water had a much higher affinity for the ion exchange resins compared to the TOC compounds present in RO permeate. This is a confirmation of the LC-OCD analysis, as ion exchange is mainly based on the adsorption of charged compounds, in other words electrostatic interactions. A comparison between the regeneration efficiencies could not be obtained from these preliminary results, as for the first attempts to investigate the TOC release during regeneration online TOC measurements were used. Unfortunately, these online TOC measurements gave no representative results, due to the slow response in TOC concentration between two successive measurements. This method was carried out in an attempt to visualize the TOC release during regeneration. For the subsequent experiments, the released TOC was calculated by measuring the TOC concentration of the regeneration and the backwash solution, taking into account the volume of the solution. 3.2. Dynamic and Long-Term Performance of TOC Removal by WBA-SBA Resins. 3.2.1. General S-Shaped

Figure 4. Difference in TOC breakthrough for the tap water experiments in runs 2, 4, 6, and 8. A TOC concentration of 100 ppb was set as the breakthrough point.

It is clear that new resins show faster breakthrough than resins that have already been in use for several cycles. In other words, the absolute adsorption efficiency appeared to increase over the consecutive cycles. An explanation for this phenomenon can be found in the fact that, despite the thorough rinsing of the new resins before use, still some production process byproducts (monomers) were probably present in the resin matrix which leached out during the first couple of experiments. This leaching out resulted in a higher TOC value of the effluent (as also seen by the higher absolute TOC value of the first runs and thus a faster breakthrough). Once these byproducts are all leached out, they no longer contribute to the TOC concentration of the treated tap water (the effluent of the SBA) and longer cycles are observed. These byproducts also explain the difference in baseline TOC values between the consecutive runs. After the minor leaching problem in the first few runs, a stable IEX operation was observed in runs 7, 8, and 9. The average adsorption and regeneration efficiency were 96 ± 1% and 64 ± 6%, respectively. The equilibrium is shown in Figure 5. For each run, the adsorption and regeneration efficiencies were calculated using eqs 1 and 2 (Table S5). The adsorption efficiency was higher than 92% for each run, indicating that less than 8% of the incoming TOC in the tap water was not retained by the resins. Most likely, this nonadsorbed TOC consisted of LMWN (or other neutral TOC compounds). Over the consecutive runs, the regeneration efficiency decreased from 84% to only 56% (in run 9), indicating that the resins 3747


Article


anions in general, chloride was the first element which showed an elevated concentration; however, the Cl− concentration after the SBA was never higher than 0.36% of the incoming tap water Cl− concentration. These results prove the excellent performance of the lab-scale setup, as there was never a real breakthrough of the most common cations and anions. In addition, this proved that TOC was the limiting factor concerning breakthrough. In some cases, the concentration of a certain component was higher at the start of a run compared to the concentration at the end of the same run. This is most likely due to the small size of the lab-scale setup, since every small adjustment or manual operation led to a high fluctuation in conductivity. In an industrial demineralization unit, these fluctuations are less noticeable, due to the large buffer capacity. Thus, after the start of a run, the system needed time to reach equilibrium. From the anion analysis, it could be determined that there were indeed some resin manufacturing byproducts that were leaching out from the resins during the first runs; namely, formate and acetate were found in the effluent, although these compounds were not present in the initial tap water. Moreover, their concentrations decreased with an increasing number of runs. The decreasing presence of manufacturing byproducts over consecutive runs is most likely partially responsible for the increasing adsorption efficiency of the resin. Another reason for the increasing adsorption efficiency is the higher amount of available functional groups after consecutive regenerations, due to the fact that resins need some adjustment time during the first operation cycles. 3.2.2. Low TOC Removal Efficiency for RO Permeate, Leading to Early Breakthrough. To confirm the results of the breakthrough experiment, the RO permeate was demineralized with the lab-scale setup two more times to verify the obtained results. Since runs 1 and 2 gave the same results as the initial breakthrough experiment, another demineralization was performed with the other feedwater (tap water) in order to have certainty about the proper functioning of the resins and to confirm whether there were no irreversible changes in the resin properties. This confirmation run provided a good reproducibility of the tap water results presented in the previous section (General S-Shaped Breakthrough Curves for Tap Water) and, thus, made clear that the demineralization of RO permeate, to produce an industrial water quality with a low TOC concentration (100 ppb), was not possible with the current IEX resins used. The TOC breakthrough curves for the runs with RO permeate can be found in Figure S5. Further research was carried out to assess the reason for the rapid breakthrough of TOC with the RO permeate. A first hypothesis was that this was mainly due to the presence of LMWN in the RO permeate. These LMWN have been shown to have less affinity for WBA and SBA resins, compared to charged organic matter.22 LMWN were also present in the tap water, but in contrast to the RO permeate, the LMWN concentration in tap water was lower (less than half of the LMWN in RO permeate) and the removal of the other TOC fractions in the tap water by IEX was sufficient to produce a water stream with a TOC concentration of less than 100 ppb. A second hypothesis was formulated that the TOC in the tap water provided useful synergies in adsorption of the LMWN in the tap water to the IEX resin. The more complex and charged TOC compounds, such as humic and fulvic acids, could form complexes with low molecular weight organics.33 As such, adsorption of the LMWN onto the humic and fulvic

Figure 5. TOC breakthrough curves of runs 7, 8, and 9 show that the IEX setup was in equilibrium and provided a stable operation. A TOC concentration of 100 ppb was set as the breakthrough point.

were never completely regenerated. The incomplete regeneration can be explained on the one hand by the limitations of the lab-scale setup (no inert resin was present on top of the functional resin, and no preheated regeneration solution was used due to safety issues) and on the other hand by the strong affinity of TOC compounds for the SBA resin. Inert resin on top of the functional resin and a preheated regeneration solution have a positive effect on the regeneration efficiency as a compact bed mode (improvement of the contact between resin beads and solution) and a higher temperature increase the kinetics of the ion exchange. It is expected that the TOC fractions adsorbing on the SBA resin during the operation mode are strongly bound and are difficult to remove during normal regeneration. Most likely, it can only be removed by a brine regeneration solution, a mixture of NaCl and NaOH, in which the chloride ions have a higher affinity to the functional groups of the SBA resin, compared to the hydroxide ions used in a normal regeneration solution, leading to a higher regeneration efficiency. To further investigate the latter explanation, more research is needed to investigate the exact TOC adsorption and desorption on both WBA and SBA resins. The results indicate that, during the runs carried out, there was no difference between a single or double regeneration, and the regeneration did not have an obvious effect on the treated water volume. In addition, the incomplete regeneration had no effect on the adsorption capacity of the subsequent run. This suggests that when a complete regeneration could be performed the absolute adsorption efficiency of the resin could increase even more over the consecutive cycles. A cation and anion analysis was done at the start and the end of each run. The detailed results can be found in Table S6. For the cations in general, the first element which showed an elevated concentration at the end of a run was calcium, which is the opposite of what is normally expected, namely, breakthrough of sodium according to a typical affinity series for cations on the SAC. An explanation for this observation is that the inlet concentration of calcium is double the amount of sodium, as the affinity is regulated not only by the ion but also by its concentration. However, the Ca2+ concentration after the SAC was never higher than 0.32% of the incoming tap water Ca2+ concentration, so in fact, there was no real breakthrough. In general, breakthrough occurs when the effluent concentration equals 10% of the inlet concentration.30−32 For the 3748


Article


organic matter) was for each run more or less the same, but the difference between the runs was the difference in TOC compounds present in the initial water, resulting in less or more TOC adsorption. In other words, when the mixture contained a high amount of LMWN, the relative adsorption efficiency was low, and when the mixture contained more charged TOC compounds, the adsorption efficiency increased. The regeneration efficiency for each experiment was very low and never higher than 73%. These lower efficiencies compared to the tap water experiments could be partially explained by the ambient temperature, which was approximately 4 °C lower, compared to the ambient temperature during the tap water experiments. The regeneration efficiency for the run with a tap water−RO permeate ratio of 0.1 was 105%, which is in theory impossible. Most likely, TOC adsorbed during the run with tap water, which was not removed during the regeneration, was rinsed out during the next regeneration. This implies that TOC, which was not removed during a regeneration, could be partially removed during a subsequent regeneration. This was not observed during the tap water experiments due to the high adsorption efficiency compared to the regeneration efficiency. In order to overcome this phenomenon, a stronger regeneration may be needed; otherwise, the resins become exhausted as less functional groups are available for the adsorption of organic matter in a subsequent operation cycle. Nevertheless, it is obvious that the percentage of LMWN in a mixture of tap water and RO permeate should even be lower than 30% in order to meet the VGB guideline of 100 ppb after IEX treatment. This can be achieved by further increasing the volume of tap water relative to the total volume. In other words, despite the higher total TOC concentration at higher ratios, the TOC adsorption is higher and this confirms the innovative statement that the TOC composition is more important than the TOC concentration in IEX demineralization. The cation and anion analyses proved again that there was no real breakthrough of ions during the different runs. The highest sodium concentration after the SAC was never higher than 0.74% of the incoming Na+ concentration, while the highest chloride concentration after the SBA was never higher than 0.6% of the incoming Cl− concentration. In some cases, the concentration of a certain component was higher at the start of a run compared to the concentration at the end of a run due to the instability of the setup (the low buffer capacity of the smallscale setup compared to industrial units). Again, some formate and acetate, as resin byproducts, leached out during the first run with the new resins. Attention needs to be paid when new resins are used as they need more than a thorough regeneration procedure to overcome the contamination of production process byproducts. 3.3. Importance of TOC Composition for Organic Acid Formation during Boiler Conditions. The initial TOC concentrations of the new volumes of tap water and RO permeate were 1.143 ppm and 304 ppb, respectively. In order to obtain a good comparison for the boiler experiments, the same amount of water volume was demineralized by the IEX setup for both water qualities. Both breakthrough curves during IEX are given in Figure 7. After the SBA, the TOC concentrations were 54 and 155 ppb for the tap water and the RO permeate, respectively. No complete TOC breakthrough was intended; as according to the VGB guideline, boiler feedwater is only allowed to have a maximum TOC concentration of 100 ppb. A volume of 250 mL

compounds could enhance their removal or enhance steric hindrance for the LMWN during passage through the resins. In addition, the concentration of LMWN differed between tap water and RO permeate, as discussed above. To assess whether such synergies exist and achieve better removal of the LMWN in the RO permeate, some trials were done in which tap water was added to the RO permeate in different ratios, namely, 0.1, 0.2, 0.3, 0.5, and 0.75 (Figure 6). Breakthrough experiments were performed with these different water qualities. The graphs including the conductivity measurements can also be found in Figure S5.

Figure 6. Effect of the addition of tap water to RO permeate on the breakthrough curve. A TOC concentration of 100 ppb was set as the breakthrough point.

The addition of tap water to RO permeate had a positive effect on decreasing the TOC concentration of the treated water. Despite the higher initial TOC concentration, a higher tap water−RO permeate ratio resulted in a lower TOC concentration of the treated mixture. As an example, the mixture with a ratio of 0.1 had an initial TOC concentration of 538 ppb and the mixture with a ratio of 0.75 had a TOC concentration of 1.305 ppm; however, after a treated volume of 80 L, the TOC concentration in the effluent was, respectively, 255 and 150 ppb for the different ratios. At first instance, this would indicate a positive synergy between the humic and fulvic compounds in the tap water and the LMWN in the RO permeate. However, even with the addition of tap water, it was still not possible to produce water after IEX demineralization with a TOC concentration less than 100 ppb. Despite the fact that addition of tap water to the RO permeate helped to decrease the TOC concentration in the effluent, the decrease in TOC concentration with increasing tap water−RO permeate ratios appeared to be merely the result of the dilution of the LMWN present in the RO permeate by the addition of tap water, and there was no synergistic effect of the tap water TOC on the removal of TOC from the RO permeate. All ratios showed TOC breakthrough at the same amount of treated volume (an average of 4 L); see Figure 6. Similar to the tap water experiments, the adsorption and regeneration efficiency for the runs at the different ratios were calculated for each run (Table S7). It is again clear that the TOC adsorption efficiency increased with increasing tap water−RO permeate ratios. The absolute adsorption efficiency of the resins (the amount of functional groups available on the resin and their capacity to adsorb 3749


Article


the corrosive behavior due to organic acids should be rather similar for both streams. The only disadvantage of using untreated RO permeate is the high Cl− concentration, which can initiate pitting corrosion at metal surfaces. When IEX treated RO permeate was used, only formate was formed (11 ppb). On the basis of these preliminary findings, RO permeate treated with ion exchange led to less formation of organic acids under boiler conditions, compared to the treated tap water. Basically, some TOC is removed from the RO permeate, which decreases its corrosive properties after hydrothermolysis. These results prove that different TOC compounds are susceptible to hydrothermolysis and lead to the formation of organic acids to a different extent. The TOCs present in the RO permeate, which were mainly LMWN, did not lead to a high formation of organic acids compared to the TOCs present in the tap water. The larger TOC compounds abundantly present in tap water were humic substances and building blocks. These charged TOC components were apparently more vulnerable to hydrothermolysis than the small LMWN. When both water qualities were treated with IEX, the difference in organic acid formation during boiler conditions was minor. Nevertheless, the treated RO permeate had a higher TOC concentration (155 ppb) compared to the treated tap water (54 ppb). In other words, the composition of TOC is more important than its concentration for any predictions concerning corrosive behavior of a certain water quality in terms of organic acid formation.

Figure 7. TOC breakthrough curves for tap water and RO permeate after ion exchange treatment. A TOC concentration of 100 ppb was set as the breakthrough point.

was collected after the WBA for both treated water qualities and used as feed solution for the lab-scale boiler experiments. The difference in TOC concentration of the IEX treated RO permeate in the breakthrough experiments (600 ppb) and the boiler experiments (304 ppb) was due to a difference in TOC concentration of the biologically treated wastewater. The RO permeate used in the breakthrough experiments was produced from biologically treated wastewater with a higher initial TOC concentration, compared to the RO permeate used in the boiler experiments. The biologically treated wastewater toward the UF/RO pilot unit contained 28 and 20 ppm TOC, respectively. After the investigation of the TOC removal by ion exchange for both tap water and wastewater effluent after RO treatment, one question remained unanswered: namely, whether a higher TOC concentration, in boiler feedwater, results directly in a higher organic acid formation under boiler conditions. In practice, there is the VGB TOC guideline which is only a strict value, namely, 100 ppb, and does not take into account potential differences in TOC composition. In general, this guideline is established in order to prevent the occurrence of corrosion in steam-water cycles. Therefore, both feedwater streams and their respective IEX treated streams were subjected to hydrothermolysis, simulating boiler conditions at lab-scale, in order to investigate the possible formation of organic acids and to give an indication of their potential contribution to corrosion. In Table S9, the anion analyses are shown for untreated tap water and RO permeate, IEX treated tap water and RO permeate, and their respective streams after boiler conditions. In general, the salt concentration of the streams was much lower after boiler conditions. This is due to the fact that, at the beginning of the mini-boiler, the water solution evaporates, but the salts precipitate. As only the steam is eventually condensed at the end of the mini-boiler, the salt concentration decreases in the condensate. An exception was the treated tap water after boiler conditions, but since the outlet Cl− concentration (232 ppb) is higher than the inlet concentration of the treated tap water (17 ppb), this was probably a contamination of previous experiments. In terms of organic acid formation under boiler conditions, untreated RO permeate gave the same results as IEX treated tap water; 9 ppb formate and 25 ppb acetate were formed. As such,

4. CONCLUSION The purpose-built lab-scale demineralization setup provided a decent simulation of the demineralization unit at Monsanto Europe N.V. The demineralization of Water-Link tap water gave very good results, concerning TOC breakthrough, as a significant amount of tap water could be treated before a TOC value of 100 ppb was reached. In contrast, the demineralization of the RO permeate gave almost immediate TOC breakthrough, and the VGB guideline for TOC could not be met. In this study, a LC-OCD analysis explained the difference in breakthrough behavior: in the RO permeate, more LMWN were present, compared to tap water. These LMWN had little or no affinity with the functional groups of the resins, leading to a lower relative adsorption efficiency for the RO permeate (56%), compared to tap water (95%). Addition of tap water to the RO permeate had a positive effect on lowering the TOC concentration of the IEX treated water. Still, it remained impossible to produce water with a TOC concentration of less than 100 ppb. In fact, the decrease in TOC concentration of the treated water with increasing tap water−RO permeate ratios appeared to be merely the result of the dilution of the LMWN present in the RO permeate by the addition of tap water. The regeneration efficiency of the resins was inadequate due to lab-scale limitations (no inert resins and no preheated regeneration solution) and perhaps the strong affinity of TOC compounds on the SBA resin. Therefore, improvement of the lab-scale setup and more in-depth regeneration experiments are required to obtain more reliable results concerning TOC regeneration efficiencies. Finally, despite the fact that the RO permeate treated by ion exchange had a higher TOC concentration (155 ppb compared to 55 ppb for tap water), the treated RO permeate led to less formation of organic acids under boiler conditions (500 °C, 60 bar). These results show that the questioning of the VGB TOC 3750


Article


(10) VGB PowerTech e.V. VGB-Standard: Feed Water, Boiler Water and Steam Quality for Power Plants/Industrial Plants; VGB PowerTech Service GmbH: Essen, 2011. (11) Rosario-Ortiz, F. L.; Snyder, S.; Suffet, I. H. Characterization of the Polarity of Natural Organic Matter under Ambient Conditions by the Polarity Rapid Assessment Method (PRAM). Environ. Sci. Technol. 2007, 41, 4895. (12) Grefte, A.; Dignum, M.; Kroesbergen, J.; Cornelissen, E. R.; Rietveld, L. C. The Effect of Changing NOM Composition, Measured by LC-OCD, on NOM Treatability in Downstream Processes. In IWA Speciality Conference on Natural Organic Matter, CA, USA, July 27−29, 2011. (13) Baghoth, S. A. Characterizing Natural Organic Matter in Drinking Water Treatment processes and Trains, Ph.D. Dissertation, Delft University, Delft, The Netherlands, 2012. (14) Lawson, K. W.; Lloyd, D. R. Membrane Distillation. J. Membr. Sci. 1997, 124, 1. (15) Č uda, P.; Pospíšil, P.; Tenglerová, J. Reverse Osmosis in Water Treatment for Boilers. Desalination 2006, 198, 41. (16) Bolto, B.; Dixon, D.; Eldridge, R. Ion Exchange for the Removal of Natural Organic Matter. React. Funct. Polym. 2004, 60, 171. (17) Lamsal, R.; Walsh, M. E.; Gagnon, G. A. Comparison of Advanced Oxidation Processes for the Removal of Natural Organic Matter. Water Res. 2011, 45, 3263. (18) Luukkonen, T.; Tolonen, E. T.; Runtti, H.; Pellinen, J.; Hu, T.; Rämö, J.; Lassi, U. Removal of Total Organic Carbon (TOC) Residues from Power Plant Make-up Water by Activated Carbon. J. Water Process Eng. 2014, 3, 46. (19) Miller, J. E. Review of Water Resources and Desalination Techniques; SAND 2003-0800 Sand Report; Sandia National Laboratories: Albuquerque, NM, 2003; No. March, 1. (20) Katsoyiannis, I. A.; Gkotsis, P.; Castellana, M.; Cartechini, F.; Zouboulis, A. I. Production of Demineralized Water for Use in Thermal Power Stations by Advanced Treatment of Secondary Wastewater Effluent. J. Environ. Manage. 2017, 190, 132. (21) Grefte, A.; Dignum, M.; Cornelissen, E. R.; Rietveld, L. C. Natural Organic Matter Removal by Ion Exchange at Different Positions in the Drinking Water Treatment Lane. Drinking Water Eng. Sci. 2013, 6, 1. (22) Cornelissen, E. R.; Moreau, N.; Siegers, W. G.; Abrahamse, A. J.; Rietveld, L. C.; Grefte, A.; Dignum, M.; Amy, G.; Wessels, L. P. Selection of Anionic Exchange Resins for Removal of Natural Organic Matter (NOM) Fractions. Water Res. 2008, 42, 413. (23) Bazri, M. M.; Barbeau, B.; Mohseni, M. Evaluation of Weak and Strong Basic Anion Exchange Resins for NOM Removal. J. Environ. Eng. 2016, 142, 04016044. (24) Tan, Y.; Kilduff, J. E. Factors Affecting Selectivity during Dissolved Organic Matter Removal by Anion-Exchange Resins. Water Res. 2007, 41, 4211. (25) Pankratov, D. A.; Anuchina, M. M.; Borisova, E. M.; Volikov, A. B.; Konstantinov, A. I.; Perminova, I. V. Sorption of Humic Substances on a Weakly Basic Anion-Exchange Resin: Relationship with the Adsorbate Structure. Russ. J. Phys. Chem. A 2017, 91, 1109. (26) Bolto, B.; Dixon, D.; Eldridge, R.; King, S.; Linge, K. Removal of Natural Organic Matter by Ion Exchange. Water Res. 2002, 36, 5057. (27) Humbert, H.; Gallard, H.; Suty, H.; Croué, J. P. Performance of Selected Anion Exchange Resins for the Treatment of a High DOC Content Surface Water. Water Res. 2005, 39, 1699. (28) Purolite. Application Guide: Cleaning Methods for Fouled Ion Exchange Resins; P-000041-200PP-0615-R2-PCO; Purolite: Bala Cynwyd, PA, 2015. (29) Rodrigues, C. C.; de Moraes, D., Jr.; da Nóbrega, S. W.; Barboza, M. G. Ammonia Adsorption in a Fixed Bed of Activated Carbon. Bioresour. Technol. 2007, 98, 886. (30) Thornton, A.; Pearce, P.; Parsons, S. A. Ammonium Removal from Digested Sludge Liquors Using Ion Exchange. Water Res. 2007, 41, 433. (31) Pavlović, D. M.; Babić, S.; Dolar, D.; Ašperger, D.; Košutić, K.; Horvat, A. J. M.; Kaštelan-Macan, M. Development and Optimization

guideline of 100 ppb is well founded. In practice, it is not only about how much TOC is present in the boiler feedwater but also about what kind of TOC is present. The LMWN present in the RO permeate did not lead to more formation of organic acids under boiler conditions. In this case, the inadequate demineralization caused no problem for the reuse of the wastewater effluent after RO treatment. Further research will be conducted on TOC removal by IEX from a variety of water qualities, and the effect on corrosion due to hydrothermolysis of organic matter will be investigated using the current lab-scale boiler setup as well as a newly developed first condensate sampling method. The aim is to establish a more well-founded TOC guideline for the boiler industry in the near future.

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.iecr.7b05059. Materials and methods; results and discussion; LC-OCD analysis (PDF)

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. ORCID

Evelyn De Meyer: 0000-0003-3079-2937 Marjolein Vanoppen: 0000-0002-4200-1613 Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The authors would like to thank Lanxess N.V. for providing the resins and Monsanto Europe N.V. for the financial and technical support. In addition, E.D.M. gratefully acknowledges the support of Research Foundation-Flanders for the Doctoral (PhD) Grant Strategic Basic Research No. 1S46516N.

■

REFERENCES

(1) Custodio, E. Coastal Aquifer Management in Europe. In Ressources et gestion des aquifères littoraux, Dix-huitièmes Journées Techniques du Comité Français d’Hydrogéologie de l’Association Internationale des Hydrogéologues, Cassis, France, March 15−17, 2012. (2) Asano, T.; Levine, A. D. Wastewater Reclamation, Recycling and Reuse: Past, Present, and Future. Water Sci. Technol. 1996, 33, 1−14. (3) Anderson, J. The Environmental Benefits of Water Recycling and Reuse. Water Sci. Technol. Water Supply 2003, 3, 1−10. (4) WWAP (United Nations World Water Assessment Programme). Wastewater: The Untapped Resource. The United Nations World Water Development Report 2017; UNESCO: Paris, 2017. (5) Ma, H.; Allen, H. E.; Yin, Y. Characterization of Isolated Fractions of Dissolved Organic Matter from Natural Waters and a Wastewater Effluent. Water Res. 2001, 35, 985−996. (6) Moed, D. H.; Verliefde, A. R. D.; Heijman, S. G. J.; Rietveld, L. C. Organic Acid Formation in Steam-Water Cycles: Influence of Temperature, Retention Time, Heating Rate and O2. Appl. Therm. Eng. 2014, 65, 194. (7) Watercycle Research Institute. OPIW 45: TOC in Process and Demin Water. In Phase 1 Inventory; KWR: Nieuwegein, 2010. (8) Maeng, W. Y.; Macdonald, D. D. The Effect of Acetic Acid on the Stress Corrosion Cracking of 3.5NiCrMoV Turbine Steels in High Temperature Water. Corros. Sci. 2008, 50, 2239. (9) Savelkoul, J.; Janssen, P.; Verhoef, H. Monitoring of First Condensate Corrosion (FCC) in Industrial Steam Systems. PowerPlant Chem. 2001, 3, 326−330. 3751


Article

Industrial & Engineering Chemistry Research of the SPE Procedure for Determination of Pharmaceuticals in Water Samples by HPLC-Diode Array Detection. J. Sep. Sci. 2010, 33, 258− 267. (32) Neumann, S. Working with Ion Exchangers: A Practical Guide; Lanxess-Lewatit: Pittsburgh, Pennsylvania, 2008. (33) Neale, P. A.; Antony, A.; Gernjak, W.; Leslie, G.; Escher, B. I. Natural versus Wastewater Derived Dissolved Organic Carbon: Implications for the Environmental Fate of Organic Micropollutants. Water Res. 2011, 45, 4227.

3752


Organic matter composition more important than ... - ACS Publications

Recommend Documents