Development of a Resin Wafer Electrodeionization Process ...€¦ · improve the EDI technology, engineers at Argonne National Laboratory have replaced the loose IEX resin beads with

Development of a Resin Wafer Electrodeionization Process forImpaired Water Desalination with High Energy Efficiency andProductivityShu-Yuan Pan,†,‡ Seth W. Snyder,† Hwong-Wen Ma,§ Yupo J. Lin,*,† and Pen-Chi Chiang*,‡,§

†Energy Systems Division, Argonne National Laboratory, 9700 South Cass Avenue, Argonne, Illinois 60439, United States‡Carbon Cycle Research Center, National Taiwan University, 71 Fang-Lan Road, Da-An District, Taipei City 10674, Taiwan (ROC)§Graduate Institute of Environmental Engineering, National Taiwan University, 71 Chou-Shan Road, Da-An District, Taipei City10673, Taiwan (ROC)

ABSTRACT: Use of groundwater, reclaimed water, and otherimpaired water sources is a critical strategy for fresh surfacewater conservation. Because impaired water desalination byreverse osmosis is energy-intensive, a robust ion-exchangeresin wafer electrodeionization technology was developed toconserve energy during impaired water desalination. The looseion-exchange resin beads were immobilized and molded toform a porous resin wafer material and utilized for impairedwater desalination. In this study, the impaired waterdesalination using resin wafer electrodeionization wasevaluated along with various key performance indicators,including current efficiency, energy efficiency, and productiv-ity. Results suggest that resin wafer electrodeionization canimprove energy efficiency to >35% in comparison to that of reverse osmosis (normally ∼12%) for impaired water desalination.The energy consumption of resin wafer electrodeionization was found to be 0.354−0.657 kWh/m3 with a productivity of 20.1−41.3 L h−1 m−2 (i.e., 5.3−10.9 gal h−1 m−2) for impaired water desalination. To reclaim cooling water at thermoelectric plants inthe United States, a huge amount of energy usage (∼400 GWh/day) can be saved, if resin wafer electrodeionization was appliedinstead of commercial reverse osmosis. We conclude that resin wafer electrodeionization offers the potential for treating impairedwater as a source water, which should be viewed as a crucial component in the portfolio of water supply options.

KEYWORDS: Brackish water, Reclamation, Electrically driven, Thermodynamic, Current efficiency, Energy consumption

■ INTRODUCTION

A major part of the energy−water nexus concerns the energyefficiency and energy consumption for supplying and treatingwater.1,2 As suggested by the U.S. Department of Energy,2

development of innovative desalination techniques couldreduce the required energy for water treatment, therebyproviding the economic incentive for nontraditional waters.On the other hand, freshwater stress and scarcity are two of themost challenging emerging issues caused by climate change,rapid population growth, urbanization, and improved standardsof living.1,3 Therefore, desalination of impaired water, such asbrackish water and plant (process) water, could be a resource-efficient solution, especially in water-scarce regions. However,the current commercial process, i.e., reverse osmosis (RO), isgood for seawater desalination but may not suitable forimpaired water desalination because of the low energyefficiency of desalting low-salinity feedwater, e.g., total dissolvedsolid (TDS) of 2−10 g/L. For a cost-effective desalination,maintaining high performance in terms of both productivity(processing rate) and energy consumption is strategicallyimportant for the optimization of process economics.

Consequently, to achieve a high energy efficiency for impairedwater desalination, several innovative processes have beenproposed and developed, such as electrodeionization (EDI),4

forward osmosis,5 and (membrane) capacitive deionization.6,7

EDI has been commercially applied to produce ultrapurewater in the semiconductor and pharmaceutical industries.Electrically driven separation processes, such as electrodialysis(ED) and EDI, remove charged ions by applying an electricfield. Ion-exchange resins are incorporated into the EDI processchannel to provide a pathway for enhanced ion migration. Theion-exchange resins increase conductivity across the processchannel and allow the transport of ions toward the ion-exchange (IEX) membranes with low-conductivity processwater. The ion-exchange resin beads are continuouslyregenerated electrochemically by protons (H+) and hydroxylions (OH−) via a water splitting reaction in the applied electricfield. For this reason, chemical regeneration is not required in

Received: October 10, 2016Revised: February 2, 2017Published: March 2, 2017

Research Article

pubs.acs.org/journal/ascecg

© 2017 American Chemical Society 2942 DOI: 10.1021/acssuschemeng.6b02455ACS Sustainable Chem. Eng. 2017, 5, 2942−2948

Dow

nloa

ded

via

UN

IV O

F C

HIC

AG

O o

n Ju

ly 3

1, 2

019

at 1

6:26

:59

(UT

C).

See

http

s://p

ubs.

acs.

org/

shar

ingg

uide

lines

for

opt

ions

on

how

to le

gitim

atel

y sh

are

publ

ishe

d ar

ticle

s.

EDI in comparison to the conventional ion-exchange columnprocess. Furthermore, EDI reduces the amount of energy usedfor impaired water desalination in comparison to the amountsused for ED and RO.8

Conventional EDI suffers from inconsistent performance anddifficult on-site maintenance because of the use of loose IEXresin beads. Loose IEX beads can result in channel formation,decreasing the rate of ion removal. With loose resins, the stackmust be handled with care in a controlled environment. Toimprove the EDI technology, engineers at Argonne NationalLaboratory have replaced the loose IEX resin beads with a resinwafer, which is composed of the original loose IEX resinsimmobilized and molded into a porous, solid matrix.9 Resinwafer electrodeionization (RW-EDI) improves ionic mobilityand also allows local pH control.10 RW-EDI has beensuccessfully applied to several process streams, including CO2capture,10 recycling of cooling water in power plants,11

production and recovery of organic acids,12,13 and deacidifica-tion of lignocellulosic biomass hydrolysate liquor.14−16 For thedesalination of impaired water using RW-EDI, the effect of keyoperating parameters on energy efficiency and productivity hasnot yet been clearly addressed.Balancing the energy efficiency and productivity to ensure

process economy is an essential task prior to the large-scaledeployment of a desalination plant. For the achievement of thisgoal, this article should be considered as a pioneering studyevaluating the performance of RW-EDI for impaired waterdesalination, in terms of current efficiency, energy consump-tion, energy efficiency, and productivity. The objectives of thisstudy were (1) to evaluate the effect of different operationconditions, including applied voltage, feed flow rate, and saltconcentration, on several key performance indicators, (2) tobalance productivity and energy consumption for RW-EDI, and(3) to compare the process performance of RW-EDI with thoseof the existing water desalination technologies, in terms ofenergy efficiency and productivity.

■ EXPERIMENTAL SECTIONResin Wafer Electrodeionization (RW-EDI). The RW-EDI

experiments were conducted in a modified, commercial electrodialysisstack (EUR2B-10) purchased from Ameridia Corp. (Somerset, NJ).

The RW-EDI stack contains the cell pair composite of an ∼3.8 mmthick dilute compartment and an ∼1.5 mm thick concentratecompartment. Cation-exchange membranes (Neosepta CMX, strongacid cation), anion-exchange membranes (Neosepta AMX, strongbasic cation), and bipolar membranes (Neosepta BP, thickness of 0.22mm) were purchased from Ameridia. Resin wafers were fabricatedusing anion-exchange resin beads (Purolite PFA444), cation-exchange(Purolite PFC100E) resin beads, a binder polymer, and a porosigen.The mixture was then heated to 120 °C for 1 h and cast in a mold toform a porous resin wafer.

Experimental Setup for Impaired Water Desalination. Weused a four-cell pair configuration RW-EDI device for the desalinationexperiments. As shown in Figure 1, each cell pair contained a diluatecompartment (ions depleted) and a concentrate compartment (ionsaccumulated), separated by cation, anion, and bipolar membranes. Thediluate compartment contains a porous ion-exchange resin wafer, witha 195 cm2 cross-section surface area. This combination was repeatedfor the multicell pair configuration. Bipolar membranes were used atboth ends of the stack to isolate the electrode rinse solution (2.5%Na2SO4) from the process fluid. The membranes were arranged tofacilitate unidirectional ion flow under an applied electric field. Ionscan move out only from the diluate compartments into concentratecompartments.

Simulated impaired water containing 5 g/L NaCl was fed into theRW-EDI stack. In each experiment, 2 L of synthesized impaired waterwas processed for 120 min. The concentrate tank started with 2 L of0.1 N NaCl (initial concentration). The flow rate of the concentratestream is fixed at 1.1 L/min in all experiments. A dc power source(XHR 100-10, Sorensen, AMETEK, Inc.) was used to apply a constantvoltage or current across the electrodes.

Key Performance Indicators. For a cost-effective desalinationprocess, maintaining high performance in terms of both energyutilization (such as current efficiency and energy consumption) andproductivity (such as processing rate) is strategically important formaximizing process economics. In this study, four key performanceindicators, including current efficiency (ηc), energy efficiency (ηe),energy consumption (ψc), and productivity (φ), were defined toevaluate the performance of the RW-EDI process for impaired waterdesalination.

Current efficiency (ηc) is a measure of how effective ions aretransported across the ion-exchange membranes for a given appliedcurrent. Typically, current efficiencies of >80% are desirable incommercial stacks to minimize electric energy costs. Low currentefficiencies indicate one of several operation problems: water splittingin the diluate or concentrate streams, shunt currents between the

Figure 1. Experimental setup of the resin wafer electrodeionization process (RW-EDI) for impaired water desalination: (1) concentratecompartment, (2) anion-exchange membrane, (3) dilute compartment, (4) cation-exchange membrane, (5) resin wafer material, (6) bipolarmembrane, and (7) electrode rinse.

ACS Sustainable Chemistry & Engineering Research Article

DOI: 10.1021/acssuschemeng.6b02455ACS Sustainable Chem. Eng. 2017, 5, 2942−2948

2943

electrodes, or back-diffusion of ions from the concentrate to thediluate. Therefore, the relationship of salt removal performance andthe process operation parameters can be clearly understood by currentefficiency, which can be calculated according to eq 1:17

∫η =

×× = ×

−×

zFn

I t

zQ C CIN MW

(%)d

100 161( )

tc

0

f i o

cp Cl (1)

where z is the valence of the ion, F is the Faraday constant (96500 s Amol−1), n is the mole number of the ion emigrated (moles), I is thestack current (amperes), t is the time interval (seconds), Qf is the feedflow rate (milliliters per minute), Ci and Co are the initial and outflowconcentrations (grams per liter) of chloride ions in solution,respectively, Ncp is the number of cell pairs, and MWCl is themolecular weight of chloride ion (i.e., 35.45 g/mol). It is noted thatthe performance of electrodeionization depends on both (1) theproperties of the materials (e.g., ion-exchange resins and membranes)in the system and (2) the operating conditions, such as the flow rate,applied voltage, and current.The energy efficiency (ηe), followed by the second law efficiency of

the desalination plant, is defined as the ratio of the minimum work ofseparation required for the desalination process to the actual workconsumed,18 as described in eq 2:

η = ×W

W(%) 100e

min

actual (2)

where Wmin is the minimum work of separation as calculated fromthermodynamics and Wactual is the actual work supplied to theseparation process. The theoretical minimum energy of desalination,realized when the separation occurs as a reversible thermodynamicprocess, is a function of recovery percentage and salinity of thefeedwater,19 while the practical energy required for desalination isstrongly dependent on the fundamental removal mechanism anddesalination method.20

Energy consumption (ψc, kilowatt hours per cubic meter) is theelectric energy used to produce a unit of purified water. In this study,only the electric energy directly used in the RW-EDI, as calculated byeq 3, was considered.

ψ = × UIQ

(kWh/m ) 16.7c3

p (3)

where U is the applied voltage (volts) and Qp is the flow rate ofproduced water (milliliters per minute).The processing productivity (φ) is reported as a function of total

membrane area (square meters). It is typically defined as the ratio ofthe feed processed rate to the total active cross-section membranearea, as calculated by eq 4:

φ =′

′− − V

t A(L h m )1 2 B

(4)

where VB′ is the volume of the feed solution in a batch operation(liters), t′ is the time to reach the target concentration of effluent(hours), and A is the total active cross-section membrane area in EDI(square meters).Analytical Methods. The conductivity of samples was measured

with a conductivity meter (SevenGo, Mettler Toledo). Theconcentration of chloride in samples was analyzed with an ionchromatograph (IC Plus system, Metrohm) with a Grace (Deerfield,IL) Allsep anion column and a conductivity detector. A mobile phaseof 3.2 mM sodium carbonate and 1.0 mM sodium bicarbonate wasprepared and introduced into the IC at a flow rate of 1 mL min−1.Moreover, sulfuric acid (100 mM) was used as the suppressor liquid. Acalibration curve between conductivity and chloride concentration wasestablished. Selected chloride concentrations were remeasured withthe IC equipment.

■ RESULTS AND DISCUSSIONPerformance Evaluation of Current Efficiency under

Different Conditions. In this study, the effect of feed flowrate, applied voltage, and effluent concentration on the currentefficiency of the RW-EDI process (ηc) was evaluated. Panels aand b of Figure 2 show the effect of feed flow rate and applied

voltage on current efficiency at different effluent NaClconcentrations, respectively. The feed flow rate had a significantinfluence on current efficiency. Current efficiency increasedwith an increase in feed flow rate from 410 to 625 mL/min.Increasing the feed flow rate increases the ion flux, wherecurrent efficiency is positively correlated. However, with afurther increase in the feed flow rate to 775 mL/min, thecurrent efficiency was reduced, revealing that excessive feedflow rates could decrease the removal ratio of salt. Theresidence time of ions in the stack decreased with an increasedfeed flow rate, which may have accounted for the decrease incurrent efficiency when the feed flow rate was >625 mL/min.For the RW-EDI, superficial velocity is one of the key

parameters for process design and control. The superficialvelocity is determined by the feed flow rate, porosity, anddimensions of the resin wafer, among which the porosity anddimensions of the resin wafer are fixed parameters; therefore,

Figure 2. Influence of (a) feed flow rate and (b) applied voltage oncurrent efficiency (defined by eq 1) for different effluent NaClconcentrations (operating condition, initial NaCl concentration of 5 g/L).

ACS Sustainable Chemistry & Engineering Research Article

DOI: 10.1021/acssuschemeng.6b02455ACS Sustainable Chem. Eng. 2017, 5, 2942−2948

2944

the feed flow rate is directly correlated with the superficialvelocity. In this study, a feed flow rate of 625 mL/mincorresponds to a superficial velocity of 2.17 cm/s. Therefore, itsuggests that a superficial velocity maintained at 2.17 cm/sshould be sufficient to achieve a high current efficiency forimpaired water desalination using RW-EDI.Current efficiency was not impacted by applied voltage in the

effluent NaCl concentration range of 1−3 g/L. In other words,the current efficiency should not be dependent on appliedvoltage at high effluent concentrations. There are enoughcharged ions being removed for efficient use by the appliedcurrent. In contrast, with a high cell voltage of 3.45 V, currentefficiency decreased significantly, especially at a relatively lowconductivity, e.g., <0.6 g/L. This indicates that excess appliedvoltage was utilized for water splitting. At a high applied cellvoltage, water splitting occurs at two locations: (i) the ion-exchange membrane surface, i.e., on both sides of ion-exchangemembranes, and (ii) the surface of cation- and anion-exchangeresin beads.21 The former reduces the extent of ion transferthrough the CEM and AEM. The latler helps regenerate theion-exchange resin beads. Therefore, to maintain a high currentefficiency, we suggest that the applied voltage should not begreater than 3.15 V per cell.In addition, it is important to note that in this evaluation,

part of the applied voltage was used for the electrode reactionsin cathodic and anodic compartments, which was much greaterthan the voltage across each call pair. For instance, in the caseof an applied voltage at 6.72 V in four-cell pair RW-EDIoperation, the actual measured cell voltage drop crossing all thecell pairs was only 2.22 V (or 0.56 V/cell). In other words,approximately 68% of the applied voltage was used in theelectrode reactions. In a commercial-scale stack of multiple cellpair ED or EDI operation (such as 100 cell pairs), thepercentage of voltage used in the electrode reactions will be<8%. Therefore, the applied electric field, such as voltage (orcurrent), should be carefully controlled to avoid excess watersplitting.Furthermore, it was observed that, in all cases, current

efficiency decreased as the effluent concentration of the dilutecompartment decreased. A decreased solution concentrationwould cause an increased resistivity of the solution within thestack. If a constant voltage was applied to the stack, the crosscurrent should decrease with the decrease in solutionconcentration because the resistance is inversely correlatedwith current according to the Ohm law. A limited cross currentthough the stack would result in a low rate of transport of ion,thereby leading to less efficient operation of salt removal. Inother words, it is expected that a high current efficiency couldbe easily achieved within an appropriate operating window ofsolution concentration. Because of sufficient ion conductivityacross the EDI stack, most of the applied electric energy couldbe efficiently utilized in the ion transport of salt.Balancing Productivity and Energy Consumption for

RW-EDI. A cost-effective process requires us to balanceproductivity and energy consumption. In this study, the effectof productivity on energy consumption was experimentallyinvestigated at an initial NaCl feed concentration (Ci) of 5 g/L,with a final dilute stream concentration (Co) of <0.5 g/L NaCl,as shown in Figure 3. With an initial feed concentration of 5 g/L NaCl, energy consumption increases from 0.354 to 0.657kWh/m3 with an increase in productivity from 20.1 to 41.3 Lh−1 m−2 (zone I), where energy efficiency decreased from 65 to35% (at 90% recovery). A further increase in productivity above

41.3 L h−1 m−2 will result in a significant increase in energyconsumption (zone II). In all cases, RW-EDI is an attractiveprocess for impaired water desalination because of its highenergy efficiency of ≤65%.In contrast, energy consumption in large-scale RO plants for

treating feedwater concentrations of 5 g/L NaCl at a recoveryof 50% was approximately 1.2−1.5 kWh/m3,22 correspondingto an energy efficiency of 10−12%. On the other hand, forconventional ED or EDI, the best achievable energyconsumption for treating brackish water was in the range of0.7−3.7 kWh/m3.23−25 These results suggest that RW-EDI ismore energy-efficient than RO and conventional EDI forimpaired water desalination.To balance the productivity and energy consumption of the

EDI process at a benchmark (e.g., effluent water quality of <0.5g/L NaCl), a two-dimensional contour plot presenting therelation of key performance indicators was constructed. Figure4 shows the effect of cell voltage and feed flow rate onproductivity and energy consumption based on the exper-imental data. For a constant productivity, energy consumptionis found to be positively correlated with applied voltage. Incontrast, there is an operating window for feed flow rate toachieve a minimum energy consumption. It is noted that asufficient feed flow rate could lead to a good fluid flowdistribution through the resin wafer material. In other words, arelatively low feed flow rate would result in a poor flowdistribution, thereby increasing the resistance and cell voltage.On the other hand, a relatively high feed flow rate reduces theresidence time of the feed stream in the EDI, which woulddecrease the removal efficiency and process productivity.The economic viability of a desalination plant is normally

considered along with the technical aspects. In general, energyconsumption is directly related to the operation cost, and theproductivity determines the plant size, capital cost, andoperation and maintenance cost (such as membrane replace-ment cost). Ideally, energy consumption should be minimized,while the productivity is expected to be maximized for systemoptimization. However, in reality, such a scenario may not existbecause the effect of different design factors and operatingparameters on energy consumption and productivity is quite

Figure 3. Dependence of productivity and energy consumption ondifferent initial salt concentrations to achieve a 0.5 g/L salt outflowconcentration. Productivity is reported as a function of totalmembrane area (square meters). The thermodynamic minimumenergy for 90% recovery at 15 °C is 0.23 kWh/m3 (as indicated by thedashed line).

ACS Sustainable Chemistry & Engineering Research Article

DOI: 10.1021/acssuschemeng.6b02455ACS Sustainable Chem. Eng. 2017, 5, 2942−2948

2945

complicated. In this study, by using Figure 4, the cell voltagesand flow rates at the desired energy consumption andproductivity can be determined throughout the graphicalpresentation. For instance, to achieve a target of energyconsumption of <0.85 kWh/m3 while maintaining a produc-tivity of >45.4 L h−1 m−2 (① → ③), the operating voltage percell and feed flow rate should be around 1.8 V (① → ②) and780 mL min−1 (③ → ④), respectively. It is noted that a techno-economic analysis of large-scale RW-EDI using field operationdata should be comprehensively performed to determine thebest achievable performance for balancing energy consumptionand productivity, which should be considered in our futureresearch.Relationship between Energy Consumption and

Energy Efficiency: Thermodynamic Viewpoint. From thethermodynamic point of view, the minimum energy require-ment for desalination basically is a function of recovery ratio(i.e., the percent of feed concentrated stream to outflow freshstream) and TDS concentrations of feed and outflow streams.In principle, the theoretical minimum energy of desalination isobserved when the separation occurs as a reversiblethermodynamic process,19 under which the theoreticalminimum energy would be equivalent to the actual free energyof mixing. In this investigation, the theoretical minimumenergies of desalination for impaired water of TDS 5 g/L atrecovery ratios of 50 and 90% were found to be 0.11 and 0.23

kWh/m3, respectively. In practice, the energy consumption of adesalination unit depends on several factors, such as themethod of desalination, the physicochemical properties of thefeedwater (salt concentration), the type of energy recoverysystem, the operating conditions, the geological location, andthe capacity of the plant. Existing technologies, such as RO19

and ED,26 are well established for seawater desalination.However, these technologies have not been optimized forimpaired water because they were operated far from thethermodynamic efficiency limit of desalination.In terms of energy efficiency for impaired water desalination,

the electrically driven processes should be superior to thepressure-driven processes as only charged ions were pulled outfrom concentrate stream by an electric field, instead of pressingall the water molecules through a semipermeable membrane. Aspresented in Table 1, significant differences in the energyefficiency (ηe) of RO were found to be 64.6 and 11.3% forseawater desalination and impaired water desalination,respectively. In other words, the energy efficiency of thesepressure-driven desalination processes decreases significantlywith the decreases in feedwater salinity. Additionally, ROgenerates a huge amount of brine waste, which would have agreat environmental impact on water, energy, and land. It isnoted that one of the best alternatives is to increase therecovery ratio of the desalination unit.2

Figure 4. Establishment of operation criteria for achieving low energy consumption and high productivity. The red line indicates an example forenergy consumption of <0.85 kWh/m3 and a productivity of >45.4 L h−1 m−2 (operating condition, initial concentration of 5 g/L NaCl and outflowconcentration of <0.5 g/L NaCl).

Table 1. Energy Consumption and Efficiency of Reverse Osmosis Desalination Comparable to the Thermodynamic Limit

[salt] (g/L) energy consumption (kWh/m3)

water sourcetreatmentprocess inlet outlet

typical recovery(%)a

thermodynamiclimitc

state-of-the-artperformance

energy efficiency (%), as determinedby eq 2 ref

sea water RO 35 0.2 42b 1.02 1.58 64.6 29RO 33 0.3 52 1.04 2.30 45.2 30

impairedwater

RO 5 0.5 50−80a 0.11−0.17 1.2−1.5 9.2−11.3 22

RO 2 0.5 60 0.033 1.01 3.3 31RW-EDI 5 0.5 90 0.230 0.35−0.66 35.0−65.0 this

studyaRecovery was assumed on the basis of a report from the U.S. Environmental Protection Agency.2 bUsing a SW30XLE-400i RO membrane. cThethermodynamic minimum was assumed at 15 °C.

ACS Sustainable Chemistry & Engineering Research Article

DOI: 10.1021/acssuschemeng.6b02455ACS Sustainable Chem. Eng. 2017, 5, 2942−2948

2946

For impaired water desalination, from the thermodynamicpoint of view, electrically driven processes (e.g., RW-EDI)should inherently exhibit energy efficiencies higher than thoseof pressure-driven processes (e.g., RO) because of thedifferences in recovery ratios. The typical recovery ratio ofRO (a pressure-driven process) for impaired water desalinationis approximately 50−80%,2 while that of RW-EDI (anelectrically driven process) is >90%. As shown in Figure 5,the energy consumption of existing RO processes for impairedwater desalination (in the case of TDS 5 g/L feedwater) was inthe range of 0.9−6.9 kWh/m3, corresponding to a low energyefficiency of 2.1−16.3%. In this study, a high energy efficiencyof up to 65% for impaired water (TDS 5 g/L) desalinationusing RW-EDI, corresponding to an energy consumption of∼0.354 kWh/m3, was observed. In the case of feedwater withTDS 2 g/L, the energy consumption of the RW-EDI processwould decrease to 0.27 kWh/m3, as compared to that of theRO process that ranged from 0.60 to 3.03 kWh/m3. On theother hand, the recovery ratio of RW-EDI (i.e., >90%) issuperior to that of RO (i.e., 50−80%)2 for impaired waterdesalination, indicating a high productivity can be achieved byRW-EDI. This also could significantly reduce the amount ofconcentrated brine waste that is produced by a desalinationunit.For commercial operation, productivity is an important

factor that determines the footprint and, thus, dictates thecapital cost. Therefore, the balance between energy efficiencyand productivity in improving RW-EDI performance isessential. According to the results presented above, theperformance of RW-EDI is quite competitive with respect tothat of the commercial RO process for brackish waterdesalination in terms of energy consumption and recoveryratio. However, the breakthrough challenges for improving thestate-of-the-art RW-EDI process should be the removalmechanism by incorporating different phenomena in EDIsuch as ion adsorption/desorption on a membrane and resinbead and the water splitting reaction, thereby synthesizinginnovative materials for resin beads and membranes. Our nextgoal will be maintaining an energy efficiency of >50% (i.e.,energy consumption of <0.46 kWh/m3) at a high productivityof >45.4 L h−1 m−2 (i.e., ∼12 gal h−1 m−2).The Future of Impaired Water Desalination. Impaired

water desalination, such as cooling and process water, using

EDI offers the potential for an abundant source of fresh water.It should be viewed as a crucial component in the portfolio ofwater supply options. For example, thermoelectric powergeneration withdraws approximately 40% of freshwater forcooling (i.e., 196 billion gal/day in the United States in 2011)2

and presents challenges in terms of the reduction of freshwaterconsumption and preservation of water quality (especiallysalinity and temperatures). One major approach to reducingfreshwater consumption at thermoelectric plants is thereclamation of water from in-plant operations. Although ROis the dominant desalination process for seawater in the UnitedStates, it is nearing its practical limits.2,27 By considering thebest achievable performance of RO for brackish waterdesalination, i.e., 1.2 kWh/m3,22 the energy consumption forreclaiming cooling water at thermoelectric plants in the UnitedStates would be ∼890.2 GWh/day. If the current-state RW-EDIwere applied for cooling water reclamation instead, i.e., 0.657kWh/m3, a huge amount of energy would be saved (∼402.83GWh/day). The assumed average electricity consumption for aU.S. residential utility customer in 2013 was 909 kWh permonth;28 the saved energy would supply energy to ∼13.5million households.In this study, we used an innovative resin wafer material (i.e.,

immobilized resin beads) to overcome the challenges in theconventional EDI process (i.e., loose-type resin beads). Forinstance, the conventional loose-type resin beads in the EDIprocess pose great difficulties in on-site installation andmaintenance. This also would lead to inconsistent performancebecause the fluid flow distribution might be fairly different aftereach single operation and/or maintenance. In other words,linear scale-up of the conventional EDI process may not beeasily achieved. However, using the resin wafer material, theaforementioned challenges and barriers in conventional EDIcan be avoided. Although the binding polymer used in the resinwafer might reduce the size of the interfaces between the cationand anion resins (resulting in a dead zone), high energyefficiency and productivity could be achieved. Compared toconventional loose-type resin beads, the immobilized resinwafer could enhance mass transfer between solid (resin bead)and liquid (feed solution) phases. It thus suggests that thedeveloped RW-EDI could provide high energy efficiency whilemaintaining high productivity for impaired water desalination.Our future research will focus on the improvement and

Figure 5. Box plot of energy consumption for impaired water desalination using RO and RW-EDI. Thermodynamic minimum at 50 and 90%recoveries for RO and RW-EDI, respectively (as indicated by red lines).

ACS Sustainable Chemistry & Engineering Research Article

DOI: 10.1021/acssuschemeng.6b02455ACS Sustainable Chem. Eng. 2017, 5, 2942−2948

2947

modification of resin wafer materials to prevent the formationof a dead zone.

■ AUTHOR INFORMATIONCorresponding Authors*Phone: +1-630-252-3741. Fax: +1-630-252-1342. E-mail:[email protected].*Phone: +886-2-2362-2510. Fax: +886-2-2366-1642. E-mail:[email protected] Chiang: 0000-0002-7531-6398NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTSThis work was created by UChicago Argonne, LLC, operator ofArgonne National Laboratory (“Argonne”). Argonne, a U.S.Department of Energy Office of Science laboratory, is operatedunder Contract DE-AC02-06CH11357. In addition, theauthors sincerely thank the Ministry of Science and Technology(MOST) of Taiwan (ROC) for Grants MOST 106-3113-E-007-002 and 104-2911-I-002-576.

■ REFERENCES(1) Water for Energy: Is energy becoming a thirstier resource?;International Energy Agency: Paris, 2013; p 33.(2) The Water-Energy Nexus: Challenges and Opportunities; U.S.Department of Energy: Washington, DC, 2014; p 262.(3) Buerkle, T. Making every drop count; Food and AgricultureOrganization of the United Nations: Rome, 2007.(4) Alvarado, L.; Chen, A. Electrodeionization: Principles, Strategiesand Applications. Electrochim. Acta 2014, 132, 583−597.(5) Luo, H.; Wang, Q.; Zhang, T. C.; Tao, T.; Zhou, A.; Chen, L.;Bie, X. A review on the recovery methods of draw solutes in forwardosmosis. J. Water Process Eng. 2014, 4, 212−223.(6) Dlugolecki, P.; van der Wal, A. Energy recovery in membranecapacitive deionization. Environ. Sci. Technol. 2013, 47 (9), 4904−10.(7) Bian, Y.; Yang, X.; Liang, P.; Jiang, Y.; Zhang, C.; Huang, X.Enhanced desalination performance of membrane capacitive deion-ization cells by packing the flow chamber with granular activatedcarbon. Water Res. 2015, 85, 371−376.(8) Zhang, Y.; Wang, L.; Xuan, S.; Lin, X.; Luo, X. Variable effects onelectrodeionization for removal of Cs+ ions from simulated waste-water. Desalination 2014, 344, 212−218.(9) Lin, Y. P. J.; Henry, M. P.; Snyder, S. W. Electronically andionically conductive porous material and method for manufacture ofresin wafers therefrom. U.S. Patent 7977395, 2011.(10) Datta, S.; Henry, M. P.; Lin, Y. J.; Fracaro, A. T.; Millard, C. S.;Snyder, S. W.; Stiles, R. L.; Shah, J.; Yuan, J.; Wesoloski, L.; Dorner, R.W.; Carlson, W. M. Electrochemical CO2Capture Using Resin-WaferElectrodeionization. Ind. Eng. Chem. Res. 2013, 52 (43), 15177−15186.(11) Gill, J. A synergistic combination of advanced separation andchemical scale inhibitor technologies for efficient use of impaired wateras cooling water in coal-based power plants. Technical Report; U.S.Department of Energy: Oak Ridge, TN, 2010; p 36.(12) Datta, S.; Lin, Y. P.; Snyder, S. W. Current and emergingseparations technologies in biorefining. In Advances in biorefineries;Woodhead Publishing: Cambridge, U.K., 2014; pp 112−151.(13) Arora, M. B.; Hestekin, J. A.; Snyder, S. W.; St. Martin, E. J.; Lin,Y. J.; Donnelly, M. I.; Millard, C. S. The Separative Bioreactor: AContinuous Separation Process for the Simultaneous Production andDirect Capture of Organic Acids. Sep. Sci. Technol. 2007, 42 (11),2519−2538.(14) Datta, S.; Lin, Y. J.; Schell, D. J.; Millard, C. S.; Ahmad, S. F.;Henry, M. P.; Gillenwater, P.; Fracaro, A. T.; Moradia, A.; Gwarnicki,Z. P.; Snyder, S. W. Removal of Acidic Impurities from Corn Stover

Hydrolysate Liquor by Resin Wafer Based Electrodeionization. Ind.Eng. Chem. Res. 2013, 52 (38), 13777−13784.(15) Lopez, A. M.; Hestekin, J. A. Improved organic acid purificationthrough wafer enhanced electrodeionization utilizing ionic liquids. J.Membr. Sci. 2015, 493, 200−205.(16) Gurram, R. N.; Datta, S.; Lin, Y. J.; Snyder, S. W.; Menkhaus, T.J. Removal of enzymatic and fermentation inhibitory compounds frombiomass slurries for enhanced biorefinery process efficiencies.Bioresour. Technol. 2011, 102 (17), 7850−9.(17) Feng, X.; Wu, Z.; Chen, X. Removal of metal ions fromelectroplating effluent by EDI process and recycle of purified water.Sep. Purif. Technol. 2007, 57 (2), 257−263.(18) Sharqawy, M. H.; Zubair, S. M.; Lienhard, J. H. Second lawanalysis of reverse osmosis desalination plants: An alternative designusing pressure retarded osmosis. Energy 2011, 36 (11), 6617−6626.(19) Elimelech, M.; Phillip, W. A. The future of seawaterdesalination: energy, technology, and the environment. Science 2011,333 (6043), 712−7.(20) Porada, S.; Zhao, R.; van der Wal, A.; Presser, V.; Biesheuvel, P.M. Review on the science and technology of water desalination bycapacitive deionization. Prog. Mater. Sci. 2013, 58 (8), 1388−1442.(21) Keramati, N.; Moheb, A.; Ehsani, M. R. Effect of operatingparameters on NaOH recovery from waste stream of Merox towerusing membrane systems: Electrodialysis and electrodeionizationprocesses. Desalination 2010, 259 (1−3), 97−102.(22) Zhao, R.; Porada, S.; Biesheuvel, P. M.; van der Wal, A. Energyconsumption in membrane capacitive deionization for different waterrecoveries and flow rates, and comparison with reverse osmosis.Desalination 2013, 330, 35−41.(23) Mcgovern, R. K.; Weiner, A. M.; Sun, L.; Chambers, C. G.;Zubair, S. M.; Lienhard, J. H., V On the cost of electrodialysis for thedesalination of high salinity feeds. Appl. Energy 2014, 136 (C), 649−661.(24) Ortiz, J. M.; Exposito, E.; Gallud, F.; García-García, V.; Montiel,V.; Aldaz, A. Desalination of underground brackish waters using anelectrodialysis system powered directly by photovoltaic energy. Sol.Energy Mater. Sol. Cells 2008, 92 (12), 1677−1688.(25) Sun, X.; Lu, H.; Wang, J. Brackish water desalination usingelectrodeionization reversal. Chem. Eng. Process. 2016, 104, 262−270.(26) Strathmann, H. Electrodialysis, a mature technology with amultitude of new applications. Desalination 2010, 264 (3), 268−288.(27) Carter, N. T. Desalination and Membrane Technologies: FederalResearch and Adoption Issues; Congressional Research Service:Washington, DC, 2015; p 18.(28) Commercial average monthly bill by Census Division, and State.U.S. Energy Information Administration: Washington, DC, 2015.(29) MacHarg, J.; Seacord, T. F.; Sessions, B. ADC baseline testsreveal trends in membrane performance. Seawater DesalinationResearch 2008, 18 (2), 30−39.(30) Schallenberg-Rodríguez, J.; Veza, J. M.; Blanco-Marigorta, A.Energy efficiency and desalination in the Canary Islands. RenewableSustainable Energy Rev. 2014, 40, 741−748.(31) Bergman, R. Reverse osmosis and nanofiltration, 2nd ed.;American Water Works Association: Denver, 1999.

ACS Sustainable Chemistry & Engineering Research Article

DOI: 10.1021/acssuschemeng.6b02455ACS Sustainable Chem. Eng. 2017, 5, 2942−2948

2948