Nitrite desorption from activated carbon fiber during capacitive deionization (CDI) and membrane capacitive deionization (MCDI)

Nitrite desorption from activated carbon fiber during capacitive deionization (CDI) and membrane capacitive deionization (MCDI)

Accepted Manuscript Title: Nitrite desorption from activated carbon fiber during capacitive deionization (CDI) and membrane capacitive deionization (M...

806KB Sizes 1 Downloads 44 Views

Accepted Manuscript Title: Nitrite desorption from activated carbon fiber during capacitive deionization (CDI) and membrane capacitive deionization (MCDI) Authors: Chengyi Wang, Lin Chen, Shanshan Liu, Liang Zhu PII: DOI: Reference:

S0927-7757(18)31227-5 https://doi.org/10.1016/j.colsurfa.2018.09.072 COLSUA 22872

To appear in:

Colloids and Surfaces A: Physicochem. Eng. Aspects

Received date: Revised date: Accepted date:

24-7-2018 26-9-2018 26-9-2018

Please cite this article as: Wang C, Chen L, Liu S, Zhu L, Nitrite desorption from activated carbon fiber during capacitive deionization (CDI) and membrane capacitive deionization (MCDI), Colloids and Surfaces A: Physicochemical and Engineering Aspects (2018), https://doi.org/10.1016/j.colsurfa.2018.09.072 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Nitrite desorption from activated carbon fiber during capacitive deionization (CDI) and membrane capacitive deionization (MCDI) Chengyi Wanga,b, Lin Chena,b,*, Shanshan Liua,b, Liang Zhua,b a

Key Laboratory of Integrated Regulation and Resources Development of Shallow

College of Environment, Hohai University, Nanjing 210098, China

A

CC E

PT

ED

M

A

N

U

SC R

b

IP T

Lakes, Hohai University, Nanjing 210098, China

*Corresponding Author,

E-mail address: [email protected]

ED

M

A

N

U

SC R

IP T

Graphical abstract

PT

Abstract

In this study, nitrite desorption from activated carbon fiber (ACF) in capacitive

CC E

deionization (CDI) and membrane capacitive deionization (MCDI) was examined and the effects of the operation parameters (pH, voltage, temperature and flow rate) and co-

A

existing matters were deeply investigated. Desorption mechanisms were analyzed via Brunauer–Emmett–Teller (BET) and Fourier transform infrared spectroscopy (FTIR). Results showed that the final desorption ratio increased from 0 to 100% in MCDI and the enhancement was also observed in CDI that the ratio increased from 18.7 to 83.5% when solution pH increased from 2 to 10. Increasing the voltage and solution

temperature also contributed to the ion desorption both in CDI and MCDI, while the effect of flow rate was negligible. Generally, MCDI showed greater desorption performance than CDI due to the elimination of co-ions effect. However, it was interesting to find that when the voltage was in the range of 0.4-0.6 V, the desorption

IP T

ratio increased from 38.4% to 50.8% in MCDI which was lower than that in CDI (45.4% to 55.8%). One possible explanation was that the presence of membranes would

SC R

inevitably introduce additional resistance into the system and decrease effective voltage

especially at lower voltage. Compared to the desorption performance in MCDI when

U

the solution pH was 2, the greater desorption performance was observed in CDI which

N

was mainly attributed to the site competition. As for the influence of coexisting matters,

A

the presence of bovine serum albumin (BSA) posed an adverse effect for the ion

M

desorption both in CDI and MCDI. The inhibition effect was more serious in CDI, and

ED

pore blockage caused by BSA attachment onto ACF was the main mechanism. Therefore, this study would provide some referential advice for the investigation of ion

PT

desorption in CDI and MCDI.

CC E

Keywords: operation parameters; coexisting matters; desorption mechanism; activated carbon fiber.

A

1 Introduction

Capacitive deionization (CDI), a promising technology which was initially

appeared between the mid-1960 s and the early 1970 s, was firstly introduced by Caudle et al. [1]. CDI generally consists of adsorption stage and regeneration process [2]. Basically, when the electrical voltage is applied to electrodes in the adsorption process,

charged ions migrate towards the electrodes and are adsorbed onto electrode to form the electrical double layer. Once the electrode is saturated, the adsorbed ions would be released back into solution by reversing the voltage to regenerate electrode in the desorption process. Comparing to the conventional desalination techniques, CDI

IP T

enables ion removal at low pressure, low voltage [3], and no chemicals addition, hence secondary pollution could be avoided. However, once the electric potential is reversed

SC R

in the regeneration step, the electrode will desorb the adsorbed ions but meanwhile

attract the oppositely charged ions from the bulk solution. In fact, ion desorption and

U

adsorption occur simultaneously during this step, which would restrain the electrode

N

regeneration efficiency. In view of this phenomenon, the ion exchange membranes are

A

introduced to improve CDI performance, which is termed as membrane capacitive

M

deionization (MCDI). More specifically, cation and anion exchange membranes are

ED

placed in front of the negatively and positively charged electrodes, respectively, which only allow counter-ions to move from the bulk solution toward the electrode [4]. Due

PT

to the selective permeability property of membrane, the co-ions are blocked and the

CC E

desalination process is more efficient. Lee et al. [5] was the first group to introduce the membrane into CDI system in 2006 and they found that MCDI's desalination rate was about 19% higher than that in CDI and the maximum NaCl removal could reach up to

A

92%.

Current studies are mainly focused on adsorption behavior and desalination performance in adsorption process, and the ion adsorption has been extensively studied both in CDI and MCDI in recent years. Tang [6] investigated the optimization of sulfate

removal and discussed the effects of operating parameters (current, pump flow rate, ending cell voltage) on MCDI desalination performance. Kim and Choi [7] found an enhanced performance by 32.8%-55.9% with MCDI in comparison with CDI, which was greatly depended on the operating conditions (potential difference and flow rate).

IP T

Zhao [8] also reported that the introduction of ion exchange membranes could effectively block the appearance of “co-ion” effect and thus dramatically decrease

SC R

energy consumption of MCDI to about four times of the CDI. For the enhancement of

ion adsorption, various considerations have been supported by experiments [9].

U

However, ion desorption behavior and electrode regeneration have not been

N

systematically studied in CDI and MCDI, which are considered to be important to

A

understand the reversibility of adsorption process. Importantly, the regeneration of

M

electrode is an economically and environmentally attractive direction specifically for

ED

the electrode used in CDI [10], and the regeneration of saturated electrodes can be achieved by applying the reverse voltage or short-circuiting the electrodes [11].

PT

To our knowledge, no relevant experiments about ion desorption in CDI and MCDI

CC E

had been reported. The objective of this study was to investigate the ion desorption behavior in CDI and MCDI, and batch experiments were performed to compare the effects of operating parameters (solution pH, applied voltage, temperature and flow rate)

A

on desorption performance. Additionally, the desorption performance of NO2- in a competitive environment was studied in the presence of BSA with different concentrations. The interaction between ions and ACF as well as the mechanisms for ion desorption in CDI and MCDI were elucidated via Fourier transform infrared

spectroscopy (FTIR), zeta potential of electrode and BET analysis. 2 Experimental 2.1 CDI/MCDI set-up The experimental system included the configurations of CDI/MCDI cells and the

IP T

detailed schematic diagram was presented in our previous report [12]. The CDI cell consists of ACF with the effective area of 0.02 m2, non-conductivity spacer

SC R

(thickness=1mm) and graphite sheet. The electrodes were made from ACF and they were attached onto graphite sheet. Additionally, a plastic plate with a thickness of 1 mm

U

was used as an electrode separation to form a fluid channel and prevent short circuit.

N

There was no difference between the CDI and MCDI processes except for the

A

introduction of ion exchange membranes. Specifically, the anion-exchange membrane

M

was attached onto the anode and the cation-exchange membrane was coated with the

ED

cathode closely. After assembly, all layers in the stack were compressed and placed in the plexiglass with a dimension of 25 × 20 cm2. Additionally, the whole system

PT

included influent tank, conductivity monitor (Shijiazhuang Ji Shen Electronic

CC E

Technology Co., Ltd, EC-1800), digital relay (Qin Yang Electric Co., Ltd, DH48S-S), peristaltic pump (YZ II 15, Longer Precision Pump Co., Ltd, China), paperless recorder (THTZ408R, Penghe Electronics Co., Ltd, China), effluent tank and potentiostat

A

(IT6720, Itech Co., Ltd, China). 2.2 Desorption experiments The experiments consisting of adsorption and desorption steps were conducted in a batch mode. In the adsorption process, the feed solution was pumped into CDI/MCDI

cell and then flowed back into feed tank driven by the peristaltic pump with a constant rate of 20 mL/min. The applied voltage was set at 1.2 V by a potentiostat to provide energy. The effluent conductivity of the effluent was monitored every 5 seconds using a conductivity monitor and the data was recorded in a computer. For each experiment,

IP T

the initial concentration of NaNO2 (Sinopharm Chemical Reagent Co., Ltd) was set as 1000 mg/L, and the ion adsorption amount was controlled at 120 mg/g through

SC R

controlling the charging time.

A series of desorption experiments were conducted after adsorption experiments

U

using an initial adsorption amount of 120 mg/g to investigate the effect of operational

N

parameters (pH, voltage, temperature, flow rate) on nitrite (NO2-) desorption behavior.

A

Firstly, the deionized water with 250 mL was used as the background solution and the

M

solution pH was carefully adjusted by adding a small amount of HCI or NaOH to

ED

control pH value in the range of 2-10. The applied voltage, solution temperature and flow rate were controlled at 1.2 V, 20 ℃ and 20 mL/min, respectively. The second group

PT

was to investigate the effect of the voltage on the ion desorption that the voltage

CC E

gradually increased from 0.4-1.2 V at a solution pH of 6, a temperature of 20 ℃ and a flow rate of 20 mL/min. Thirdly, the solution temperature was slightly increased from 20-50 ℃ using a thermostat water bath, and the solution pH, voltage and flow rate were

A

controlled at 6, 1.2 V and 20 mL/min, respectively. Finally, the flow rate was changed from 20-60 mL/min at a solution pH of 6, a voltage of 1.2 V and a temperature of 20 ℃. For these four experiments, the deionized water with 250 mL was used as the background solutions to conduct the desorption experiment.

Furthermore, to identify the effect of co-existing matter on ion desorption, the desorption experiments were conducted by using BSA (Sinopharm Chemical Reagent Co., Ltd) solutions with different concentrations (10, 20, 50, 100 mg/L) as background solutions. Solution pH, voltage, temperature and flow rate were controlled at 6, 1.2 V,

2.3 Analytics and characterization

SC R

The ion desorption amount at time t was calculated from Eq.(1):

IP T

20 ℃ and 20 mL/min, respectively.

QDes =Ct ×V

(1)

U

where QDes is the desorption amount of NO2- at time t, Ct and V represent the ion

N

concentration in solution and solution volume, respectively.

QDes QAbs

×100%

M

R=

A

The desorption ratio was calculated from the following equation: (2)

ED

where R is the desorption ratio, QDes and QAbs represent the desorption and adsorption

PT

amounts of NO2- ions, respectively.

The pore size distribution of the ACF was determined using a surface analyzer

CC E

based on the N2 adsorption/desorption isotherm and the pore size (D) was calculated by Density Function Theory (DFT) method. The specific surface area was calculated using

A

the BET (3H-2000PM2) method and the t-plot equation was used to calculate micropore volume [13]. Before analysis, the ACF was outgassed at 200 °C for 4 h under the vacuum condition. The functional groups of the ACFs before and after usage which was correlated with the desorption mechanism were examined using pressed potassium bromide (KBr)

pellets containing 5% of the samples on a FTIR analysis spectrometer in the scanning range of 4000-400 cm−1. Notably, the selected ACF after usage was achieved under the operation condition of solution pH 6, voltage 1.2 V, temperature 20 ℃ and flow rate 20 mL/min. The samples were pulverized (100 mesh) and mixed with KBr before being

IP T

pressed into a disk. The point of zero charge (pHpzc) of the prepared ACF was a determination by using

SC R

potentiometric titration method [14] by using a zeta potential analyzer (SZ-100Z, Horiba, Japan).

U

3 Results and discussions

N

3.1 Characterization of ACF

A

3.1.1 BET

M

Similar to the adsorption behavior that was influenced by electrode properties

ED

including electrode structure, surface charge, functional groups, and all relevant experimental conditions [15], it was suggested that desorption behavior should also be

PT

affected by these factors. Therefore, in this study, the electrode ACF was characterized

CC E

by BET, Zeta potential and FTIR to clarify the desorption process in the regeneration stage.

The specific surface area and pore size distribution of ACF had a significant effect

A

on the ion storage and the results are presented in Table 1. It was obtained that the ACF had a BET specific area of 899.6 m2/g and the micropore surface area was estimated to be 617.8 m2/g. The total pore volume was up to 0.67 cm3/g, and 68.7% of the total surface area was located in the micropores of the ACF. These results suggested that

ACF exhibited quite high specific surface area and developed porous structure, which was favorable for ion storage. In addition, the average pore size was 3.0 nm based on the nitrogen adsorption/desorption isotherms. The smaller pore size was considered to be accessible for the ion storage, while it was difficult for organic molecules with larger

IP T

size to transport into inner pores. 3.1.2 Zeta potential

SC R

The pHpzc represented the pH at which the surface charge of the electrode surface

was zero, whose value was extremely important to evaluate the effect of solution pH

U

on the adsorption capacity of a given adsorbent material [16]. As depicted in Fig.1, it

N

was observed that the pHpzc of ACF was 4.6, indicating that the ACF surface developed

A

negative charge in solutions as pH value was higher than 4.6 and positive charge when

M

the solution pH was lower than 4.6. The oxygenated functional groups of ACF electrode

ED

underwent protonation and de-protonation reactions as aqueous solution pH changed, leading to the variation in the zeta potential. At a lower pH, the access of protons in the

PT

solution neutralized the negatively charged groups on the electrode surface and hence

CC E

the number of the positively charged sites increased, which correspondingly increased the zeta potential of the electrode. However, at higher pH value, hydroxyl had more chances to be adsorbed onto electrode, resulted in an enhanced negatively charge of the

A

electrode surface. Considering the electric property of ACF at different solution pH and the negative charge of NO2-, the ion desorption was partly determined by the electrostatic interaction between ions and the electrode surface charge. Hence, except for the electrostatic interaction produced by the applied voltage, the interaction between

electrode and ions was also a key parameter to examine the desorption performance. 3.1.3 FTIR Fig.2 presents the FTIR spectra of the ACF of anode before and after usage and it was noticed that there was four distinct peaks in the spectra of virgin ACF-1, including

IP T

1112, 1384, 1637, and 3447 cm−1. The broadest band at 3447 cm-1 belonged to the hydrogen-bonded O-H stretches in carboxyl and phenol groups [17]. The band near

SC R

1637 cm-1 was ascribed to the C=O stretching vibrations corresponding to the carboxyl

groups, and the peak in the region of 1384 cm-1 was attributed to C=OH stretches [18].

U

The peaks at 1112 cm−1 was described to the C=O stretching mode [19]. Comparatively,

N

when ions were adsorbed onto electrode surface as shown in ACF-2, the bands in the

A

regions of 3447 cm−1 and 1112 cm−1 shifted to lower wavenumbers to 3440 and 1101

M

cm−1, and the intensities of functional groups became weak. These two peaks

ED

represented the characteristics of carboxyl groups, and this shift implied that the carboxyl groups were the adsorption sites for the formation of the hydrogen bonds. The

PT

formation of hydrogen bond had been demonstrated by various reports and its effect on

CC E

ion adsorption was of great importance. Sivasamy et al. [20] assumed that HF and F− can interact with functional groups on the carbon surface via hydrogen bonding. Furthermore, Islam et al. [21] found that the band at the peak of 3410 cm−1 moved to

A

lower frequencies in a fluoride–adsorbed material, indicated the formation of hydrogen bonding between the protonated amine and fluoride. Under the action of electrical interaction, NO2- was adsorbed onto ACF surface in the adsorption process and the carboxyl groups were responsible for adsorption sites.

Specifically, the carboxyl groups contained a large π bond structure and led to the charge transfer from O atom in the O-H structure to the C atom, and thus resulted in an increased positive charge on the H atom [22-24]. Therefore, the ACF surface acquired positive charge which in turn attracted more NO2- by means of hydrogen bonding

IP T

through the electrostatic interaction [25]. To deeply understand the desorption process in CDI and MCDI, it was necessary

SC R

to elucidate the adsorption process and it was found that ions were adsorbed onto the electrode surface and bond with the carboxyl groups on the electrode due to hydrogen

U

bonding based on FTIR analysis. Therefore, the affinity between ions and carboxyl

N

groups was also considered as a crucial direction to investigate the desorption behavior

A

in the regeneration process. Additionally, this affinity was greatly affected by the

M

solution pH, which would be discussed next.

ED

3.2 Effect of operating parameters on the ion desorption 3.2.1 Effect of solution pH

PT

The effects of solution pH on the desorption behavior of NO2- in CDI and MCDI

CC E

was firstly investigated at solution pH=2, 4, 6, 8 and 10, and the results are presented in Fig.3. For MCDI, it was clearly shown that the desorption ratio was approximately zero and there were almost no ions desorbed from electrode when the solution pH was

A

2. At pH 4, the ratio gradually increased to 47.2% within 35 minutes and kept steady indicating that desorption equilibrium was reached. Similar trends were also obtained at solution pH of 6, 8 and 10 that the desorption ratio followed an upward trend to increase to 94.2%, 100%, 100%, and the equilibrium was achieved within 30, 20 and

15 minutes, respectively. It was worth mentioning that when the solution pH was up to 8, the complete ion desorption was obtained and further increasing solution pH had a limited effect on the desorption ratio. However, the desorption rate was greatly accelerated and thus the equilibrium time was significantly shortened with the

IP T

increment of solution pH. Based on above analysis, it could be concluded that alkaline conditions were favorable for ion desorption and similar findings were also obtained in

SC R

CDI that the final ratio increased from 18.7% to 83.5% and significant decrement in equilibrium time was found when solution pH increased from 2 to 10.

U

Under the action of the electrostatic repulsion, the adsorption ions were repelled

N

and released back into the solution, and the electrical interaction played a major role in

A

ion desorption. Besides, ions were adsorbed and bond with the functional groups of the

M

electrode in the adsorption process which was demonstrated by FTIR analysis, implying

ED

that the affinity between ions and functional groups was a key factor affecting the ion desorption. As pH was increased from 2 to 10, the carboxyl concentration decreased

PT

significantly due to the deprotonation effect. Therefore, the affinity determined by the

CC E

strength of hydrogen bonding was significantly weakened due to the decrement in the concentration of carboxyl group on the electrode. Further, the deprotonation effect also affected the surface charge of electrode, which was reflected by the variation of zeta

A

potential of electrode as presented in Fig.1. When the solution pH was below the point of zero net charge (pH=4.6), the electrode surface was dominated by positive charges on their surface and provided the opportunity to coordinate with NO2-, thus the electrical repulsion produced by the applied voltage was slightly weakened by the

positive effect of the electrode surface. Hence, ion desorption was inhibited significantly at a lower pH. Conversely, as the solution pH increased over zero net charge, the surface charge developed negative value due to de-protonation reactions, in which the electrostatic repulsion between the electrode and negatively charged NO2-

IP T

was increased, leading to a rapid and complete ion desorption. 3.2.2 Effect of the applied voltage

SC R

The experiments were conducted to investigate the effect of the applied voltage on

the ion desorption behavior in CDI and MCDI, and the applied voltage was controlled

U

in the range of 0.4-1.2 V to prevent the electrolysis reaction of water [26]. Fig.4 presents

N

the changes in the desorption ratio in CDI and MCDI as the applied voltage was varied

A

in the desorption process. At the applied voltage of 0.4 V, the desorption ratio slightly

M

increased to 45.4 % and 38.4%, respectively, for CDI and MCDI with the operation of

ED

discharging time within 50 minutes. This incomplete ion desorption was partly due to the fact that the electrostatic repulsion at the voltage of 0.4 V was not sufficient to

PT

desorb the adsorbed NO2- completely. It was observed that the desorption ratio

CC E

presented a continuous increasing trend as the voltage rising and was maximized at 1.2 V, which reached to 94.2% and 72.6% in MCDI and CDI and the time to achieve desorption equilibrium was obviously shortened. It was suggested that the ion

A

desorption was highly dependent on the applied voltage and higher voltage would enhance ion desorption. In CDI/MCDI, the driving force for ion desorption was electrostatic interaction which was mainly produced by the applied voltage. Therefore, the improvement in the ion desorption was expected with the increment of the voltage.

3.2.3 Effect of temperature The desorption behavior of CDI and MCDI at different temperature (20, 30, 40, 50 ℃) are presented in Fig.5. Briefly, it was found that the desorption ratio in CDI and MCDI ascended with the temperature increasing, showing that positive effects that solution

IP T

temperature had on NO2- desorption. For MCDI, the complete ion desorption was obtained when solution temperature was 30 ℃ and further increasing temperature could

SC R

not pose any influence on the ratio, while the desorption rate showed an accelerating

tendency and the equilibrium time was achieved more rapidly within 28 minutes. More

U

significant enhancement was obtained when the solution temperature was increased to

N

50 ℃ that the equilibrium time was reduced to around 18 minutes. Regarding the CDI,

A

the enhancement was also observed that the desorption ratio exhibited a continuous

M

increasing trend from 72.6 % to 83.1% and the equilibrium time was shortened from

ED

34 to 24 minutes with the increment of solution temperature from 20 to 50 ℃. Previous studies reported that the adsorption process was an exothermic reaction and

PT

hence increasing the solution temperature would inhibit adsorption [27]. Therefore, as

CC E

a reverse reaction of adsorption, we assumed that the desorption experiments had the contrary results on the temperature-dependence in this study and the desorption process was endothermic in nature, which was accorded with the experimental results. Rising

A

the temperature aggravated ionic movement and enhanced the tendency of ions to escape from the electrode back into solution both in CDI and MCDI, leading to an increasing trend of ion desorption. Further, the electrode wettability which was mainly determined by the surface groups acted as a leading role in the ion

adsorption/desorption, and it was reported that greater hydrophilicity of the electrode enabled easier access of ions to the pores and strengthened the affinity between ions and electrode [28]. Increasing solution temperature would cause the transition from hydrophilic to hydrophobic of the electrode surface [29], which weakened the affinity

IP T

between ions and electrode and consequently NO2- desorption was accelerated. 3.2.4 Effect of flow rate

SC R

Ion desorption experiments were performed to evaluate the effect of the flow rate

(20, 30, 40, 50 mL/min) on the desorption behavior of NO2- in CDI and MCDI, as shown

U

in Fig.6. It was reported that the ion adsorption was greatly influenced by the variation

N

of the flow rate in MCDI [30], while its effect on the desorption behavior was quite

A

limited as observed in this experiments. More specifically, the desorption ratio slightly

M

increased to 95.2% and the time of desorption equilibrium was around 24 minutes in

ED

MCDI when the flow rate was increased to 50 mL/min, while it was noticed that this enhancement was limited. Compared to the experimental result with the flow rate of 20

PT

mL/min, it was suggested that the flow rate had a negligible effect on ion desorption

CC E

during the regeneration process in MCDI. In contrast, a slight increment was observed in CDI that the value increased from 72.6% to 78.4% when the flow rate increased from 20 to 50 mL/min. Increasing the flow rate meant that the residence time of feed in the

A

spacer declined correspondingly and the desorbed NO2- could rapidly release from the spacer into bulk solution, which inhibited co-ions effect to a certain extent. Therefore, ion desorption in CDI was slightly promoted. However, it could also be found that the effect of flow rate was not so notable comparing to the effects of solution pH, voltage

and temperature which had been discussed previously. 3.3 Desorption performance comparison between MCDI and CDI To compare the desorption performance of CDI and MCDI under various conditions, the final desorption ratio was presented in Fig.7. As depicted in Fig.7 (a), it was clearly

IP T

shown that CDI had the ratio of 18.7% when solution pH was 2 which was superior to that in MCDI, and this minor desorption in CDI was likely caused by site competition

SC R

and ion displacement. Generally, during the desorption phase, the electrode reversed its polarity and changed to the negative electrical field. In this case, the electrode would

U

desorb the adsorbed NO2- and attract oppositely charged ions (H+) from the spacer

N

simultaneously. The adsorbed ions (H+) occupied adsorption sites and the NO2- were

A

repelled from electrode, leading to the substitution of NO2- by H+ especially at a lower

ED

presence of membranes.

M

pH. In contrast, in MCDI system, the ion displacement was prevented due to the

However, when the solution pH increased above 4, CDI showed significantly lower

PT

ion desorption than MCDI attributed to the effect of ion exchange membrane. For CDI,

CC E

the adsorbed ions on the electrodes were released back into the solution that accompanied by some newly-released ions moving towards and adsorbing onto the opposite electrodes with the application of reversed electric force, which was known as

A

“co-ion” effect [31]. Therefore, part of NO2- ions were attracted and moved towards the electrode leading to a slight decrement in the ratio, and thus the full regeneration of electrode as well as the complete ion desorption was impossible. However, in MCDI, this “co-ion” phenomenon could be eliminated, in which NO2- re-adsorption was

prevented due to the selective permeability property of membrane. A similar phenomenon was also observed in the case of different temperatures and flow rates that the desorption performance in MCDI was superior to that in CDI due to the elimination of co-ions effect, as presented in Fig.7 (c)-(d).

IP T

Fig.7 (b) evaluated the desorption performance between CDI and MCDI at the voltage 0.4-1.2 V, in which a significant discrepancy was obtained as the voltage varied.

SC R

More specifically, it was interesting to note that CDI had a greater desorption

performance compared to MCDI when the applied voltage was in the range of 0.4-0.6

U

V, which could be explained by the membrane resistance. The inclusion of membranes

N

would inevitably introduce additional resistance into the system, resulting in a greater

A

potential drop throughout the cell [32]. Considering that the membrane represented an

M

additional resistance within the system, the effective working voltage used for ion

ED

desorption was lower than that in a CDI process with the application of the same voltage, and this decrement in the effective working voltage was more significant especially at

PT

a lower voltage. Comparatively, when the working voltage was up to 0.8 V, the MCDI

CC E

presented better desorption performance due to the co-ion effect in CDI. The introduction of membranes effectively prevented the occurrence of co-ions effect and dramatically increased the current efficiency, which improved the ion desorption.

A

Combing the results of the co-ions effect and the additional resistance introduction, it could be concluded that the resistance addition was the dominant factor which hindered the ion desorption in MCDI when the voltage was lower than 0.6 V. As the voltage increased above 0.8 V, the co-ions effect became more significant and obvious

decrement in the desorption ratio was obtained in CDI. 3.4 Effect of co-existing organic matter To prevent the co-ions effect in the desorption process, the cation exchange membrane (CEM) was coated onto electrode to prevent NO2- to be re-adsorbed onto

IP T

electrode surface. Fig.8 presents the results of desorption behavior of NO2- in CEMCDI and MCDI by using BSA solution as background solutions with different

SC R

concentrations (10, 20, 50 and 100 mg/L). As shown in Fig.8, for the background solution of 10, 20, 50, 100 mg/L BSA, the desorption ratio for NO2- gradually decreased

U

to 88.2%, 83.3%, 78.4% and 72.8% for MCDI. Regarding CEM-CDI, a more

N

significant decrement in desorption ratio was observed and the value dropped to 70.7%,

A

65.5%, 54.2% and 40.5%, indicating that the presence of BSA led to less ion desorption

M

and the inhibition became more significant especially at higher BSA concentration both

ED

in CEM-CDI and MCDI. The initial solution conductivity was about 1900 μS/cm and the addition of BSA had no effect on solution conductivity and the initial operating

PT

current. However, the current in desorption experiment was both used to desorb the

CC E

adsorbed ions and adsorb BSA molecules, which decreased current efficiency and thus inhibited ion desorption. Additionally, it was well known that the “soft” proteins had the ability to adsorb any kind of ions although it was hydrophilic and electrostatically

A

repulsive [33]. Since BSA was a kind of soft protein [34], and it was able to bind with NO2- in spite of the electrostatic repulsion leading to the decrement in NO2concentration in solution. Moreover, a more serious decrement of the ratio was observed in CEM-CDI

compared to that in MCDI. It should be emphasized that the carbon properties were seemed to be quite complicated affecting CDI/MCDI desalination performance, in which pore size distribution of carbon electrode played a fundamental role [35]. The porous electrode was consisted of macropores where ions migrated and micropores

IP T

where electrical double layers were formed and ions are temporarily stored. For the effect of organic matter on the ion desorption, two determinative mechanisms were

SC R

suggested including the direct competition for the limited adsorption sites and indirect

competition due to pore blockage by organic matter, which was strongly dependent on

U

the molecule size. Considering the limited adsorption sites and capacity of the ACF, the

N

competitive effect between H+ and NO2- was inevitable and the charged ions (H+) were

A

continuously adsorbed onto the ACF surface and occupied more adsorption sites under

M

the action of the reverse voltage, leading to the NO2- substitution by H+, which was

ED

usually termed as the direct competition. Similar result about the occurrence of the ion competition and ion substitution was also demonstrated by Yingzhen Li et al. [36].

PT

Direct competition was particularly positive for desorption [37], while pore blockage

CC E

conversely inhibited the ion desorption [38]. As mentioned in BET analysis, the ACF used in this study had microporous features with an average pore size of 3.0 nm, which made these materials to exhibit limited ability for adsorption of large BSA molecules.

A

BSA was facile to be adsorbed onto electrodes for their hydrophobic or amphoteric character [39]. However, BSA molecules were easy to aggravate together to form clusters with the size distribution from 21 to 300 nm, which was rather larger than the pore size of ACF. Considering the discrepancy between electrode pore size and the size

of BSA aggregates, BSA molecules were unable to enter into inner pores and thus pore blockage was more prominent than site competition, which impeded the ion transport within the ACF. Additionally, with the increasing of BSA concentration, the pore blockage phenomenon was more serious which was attributed to the fact that BSA

IP T

molecules would aggravate more significantly to form larger size. However, this pore blockage could not be observed in MCDI due to the existence of membranes, which

SC R

prevented BSA aggregates to be adsorbed onto electrode surface. 3.5 Mechanism study

U

Desorption mechanism was deeply investigated in CDI, CEM-CDI and MCDI, and

N

the mechanism was briefly illustrated in Fig.9 summarily based on the experimental

A

results. First of all, the electrostatic interaction produced by the applied voltage was a

M

leading role in the ion desorption both in CDI, CEM-CDI and MCDI, and higher

ED

voltage was beneficial for the ion desorption. In addition, the properties of the ACF including the functional groups and surface charge was also a key parameter

PT

determining the ion desorption, which was highly dependent on the solution pH and

CC E

temperature. Generally, an increment in solution pH made the surface charge of ACF transfer from positively charge to negatively charge and weakened the affinity between ions and carboxyl group, which greatly enhanced the ion desorption. For the effect of

A

solution temperature, the increment caused the transition from hydrophilic to hydrophobic of the electrode surface, which weakened the affinity between ions and electrode and consequently NO2- desorption was accelerated. Comparing the desorption mechanism of CDI and MCDI as illustrated in Fig.9 (a)

and Fig.9 (b), it was found that co-ions effect was a significant factor inhibiting ion desorption in CDI. However, at lower pH condition, ion desorption was significantly hindered and no ions could be desorbed from electrode, which was mainly due to the electrical attraction produced by the positive surface charge of ACF and the stronger

IP T

strength of the hydrogen bond between carboxyl groups and ions. Comparatively, a slight desorption ratio in CDI was obtained when the solution pH was 2 and this result

SC R

was attributed to the site competition caused by H+ substitution. As for the effect of

BSA on the ion desorption as shown in Fig.9 (c), the co-ions effect and NO2- re-

U

adsorption was avoided in CEM-CDI. Due to the narrow pore size of ACF, BSA

N

molecules with larger size could not transport into inner pores of ACF and hence they

A

would attach onto electrode surface and result in the pore blockage naturally, which

M

inhibited ion desorption significantly. In MCDI as presented in Fig.9 (d), the co-ions

ED

effect and BSA attachment was eliminated due to the presence of membranes, leading to greater desorption performance compared to that in CEM-CDI.

PT

4 Conclusions

CC E

In this study, the effects of coexisting matter and operation parameters (solution pH, voltage, temperature and flow rate) on desorption behavior in CDI and MCDI were

A

investigated, and the following conclusions were obtained: It was found that the ion desorption was highly associated with the operating

parameters and the interaction between ions and ACF, and the desorption mechanisms were (a) electrostatic interaction produced by the applied voltage; (b) electrostatic interaction between ions and electrode which was affected by protonation and de-

protonation caused by pH variation; (c) pore blockage; (d) site competition; and (e) coions effect. Acidic solution greatly inhibited ion desorption both for CDI and MCDI, and the ions could not be desorbed when solution pH was 2. Alkaline conditions were favorable

IP T

to ion desorption and higher solution pH resulted in complete ion desorption due to electrode protonation and de-protonation. Increasing solution temperature enhanced ion

SC R

desorption attributed to the fact that the desorption process was an endothermic reaction. Additionally, it was found that the flow rate had a negligible effect on ion desorption.

U

When BSA was added into the background solution, the current was both used to

N

desorb NO2- and adsorb BSA molecules, which significantly decreased the current

A

efficiency. Additionally, in CDI, the pore size of electrode was rather smaller according

M

to BET analysis and therefore BSA molecules could not transport into inner pores,

Acknowledgement

ED

which naturally led to pore blockage and consequently inhibited ion desorption.

PT

This study was mainly financially supported by National Natural Science Fund of

CC E

China (grant number: 51508153), Natural Science Fund of Jiangsu (grant number: BK20150813), Fundamental Research Funds for the Central Universities (2018B15014) and A Project Funded by the Priority Academic Program Development of Jiangsu

A

Higher Education Institutions. References [1] G.W. Murphy, Electrochemical demineralization of water with carbon electrodes, Scientific Reports, 4 (1965) 7397-7397. [2] K.S. Lee, Y. Cho, K.Y. Choo, S.C. Yang, M.H. Han, D.K. Kim, Membrane-spacer assembly for flowelectrode capacitive deionization, Applied Surface Science, 433 (2018).

[3] H. Jo, K.H. Kim, M.J. Jung, J.H. Park, Y.S. Lee, Fluorination effect of activated carbons on performance of asymmetric capacitive deionization, Applied Surface Science, 409 (2017) 117-123. [4] J. Chang, K. Tang, H. Cao, Z. Zhao, C. Su, Y. Li, F. Duan, Y. Sheng, Application of anion exchange membrane and the effect of its properties on asymmetric membrane capacitive deionization, Separation & Purification Technology. [5] J.B. Lee, K.K. Park, H.M. Eum, C.W. Lee, Desalination of a thermal power plant wastewater by membrane capacitive deionization, Desalination, 196 (2006) 125-134. [6] W. Tang, H. Di, C. Zhang, T.D. Waite, Optimization of sulfate removal from brackish water by membrane capacitive deionization (MCDI), Water Research, 121 (2017) 302-310.

IP T

[7] H. Li, Y. Gao, L. Pan, Y. Zhang, Y. Chen, Z. Sun, Electrosorptive desalination by carbon nanotubes and nanofibres electrodes and ion-exchange membranes, Water Research, 42 (2008) 4923-4928.

[8] Y. Zhao, Y. Wang, R. Wang, Y. Wu, S. Xu, J. Wang, Performance comparison and energy consumption

SC R

analysis of capacitive deionization and membrane capacitive deionization processes, Desalination, 324 (2013) 127-133.

[9] K.H. Park, D.H. Kwak, Electrosorption and electrochemical properties of activated-carbon sheet electrode for capacitive deionization, Journal of Electroanalytical Chemistry, 732 (2014) 66-73. [10] E. Bayram, N. Hoda, E. Ayranci, Adsorption/electrosorption of catechol and resorcinol onto high

U

area activated carbon cloth, Journal of hazardous materials, 168 (2009) 1459-1466.

[11] Y. Zhao, X.M. Hu, B.H. Jiang, L. Li, Optimization of the operational parameters for desalination

N

with response surface methodology during a capacitive deionization process, Desalination, 336 (2014) 64-71.

A

[12] L. Chen, C. Wang, S. Liu, Q. Hu, L. Zhu, C. Cao, Investigation of the long-term desalination

M

performance of membrane capacitive deionization at the presence of organic foulants, Chemosphere, 193 (2018) 989-997.

[13] K.S.W. Sing, Reporting Physisorption Data for Gas/Solid Systems-With Special Reference to the

ED

Determination of Surface Area and Porosity, Pure & Applied Chemistry, 57 (1985) 603-619. [14] J. Rivera-Utrilla, I. Bautista-Toledo, x. a, A. Ferro-García, C. Moreno-Castilla, Activated carbon surface modifications by adsorption of bacteria and their effect on aqueous lead adsorption, Journal of

PT

Chemical Technology & Biotechnology, 76 (2001) 1209-1215. [15] D.I. Oyarzun, A. Hemmatifar, J.W. Palko, M. Stadermann, J.G. Santiago, Adsorption and capacitive regeneration of nitrate using inverted capacitive deionization with surfactant functionalized carbon

CC E

electrodes, Separation & Purification Technology, 194 (2017). [16] M.H. Marzbali, M. Esmaieli, H. Abolghasemi, M.H. Marzbali, Tetracycline adsorption by H 3 PO 4 -activated carbon produced from apricot nut shells: A batch study, Process Safety & Environmental Protection, 102 (2016) 700-709. [17] T. Tsoncheva, N. Velinov, R. Ivanova, I. Stoycheva, B. Tsyntsarski, I. Spassova, D. Paneva, G. Issa,

A

D. Kovacheva, I. Genova, Formation of catalytic active sites in iron modified activated carbons from agriculture residues, Microporous & Mesoporous Materials, 217 (2015) 87-95. [18] P. Painter, M. Starsinic, M. Coleman, DETERMINATION OF FUNCTIONAL GROUPS IN COAL BY FOURIER TRANSFORM INTERFEROMETRY, Fourier Transform Infrared Spectra, (1985) 169241. [19] Y. Guo, D.A. Rockstraw, Physical and chemical properties of carbons synthesized from xylan, cellulose, and Kraft lignin by H 3 PO 4 activation, Carbon, 44 (2006) 1464-1475.

[20] D. Qiang, D.E. Wurster, Specific and non-specific interactions of procaine with activated carbon surfaces, Journal of Colloid & Interface Science, 358 (2011) 541-546. [21] M. Islam, P.C. Mishra, R. Patel, Fluoride adsorption from aqueous solution by a hybrid thorium phosphate composite, Chemical Engineering Journal, 166 (2011) 978-985. [22] Z. Sun, L. Chai, Y. Shu, Q. Li, M. Liu, D. Qiu, Chemical bond between chloride ions and surface carboxyl groups on activated carbon, Colloids and Surfaces A: Physicochemical and Engineering Aspects, 530 (2017) 53-59. [23] G.-X. Xu, X.-Y. Wang, Material structure, People Education Press, 1987. China Coal Society, 29 (2004) 443-447.

IP T

[24] F.-g. Zeng, K.-c. Xie, Theoretical system and methodology of coal structural chemistry, Journal of [25] B. Gu, J. Schmitt, Z. Chen, L. Liang, J.F. Mccarthy, Adsorption and desorption of different organic matter fractions on iron oxide, Geochimica Et Cosmochimica Acta, 59 (1995) 219-229.

SC R

[26] D.J. Lee, M.S. Kang, S.H. Lee, J.S. Park, Application of Capacitive Deionization for Desalination of Mining Water, Journal of the Korean Electrochemical Society, 17 (2014) 37-43.

[27] M. Mossad, L. Zou, A study of the capacitive deionisation performance under various operational conditions, Journal of hazardous materials, 213-214 (2012) 491-497.

[28] J.J. Lado, R.L. Zornitta, F.A. Calvi, M.I. Tejedor-Tejedor, M.A. Anderson, L.A.M. Ruotolo, Study

U

of Sugar Cane Bagasse Fly Ash as Electrode Material for Capacitive Deionization, Journal of Analytical & Applied Pyrolysis, 120 (2016) 389-398.

N

[29] H.J. Wang, X.K. Xi, A. Kleinhammes, Y. Wu, Temperature-induced hydrophobic-hydrophilic transition observed by water adsorption, Science, 322 (2008) 80-83.

A

[30] R. Zhao, O. Satpradit, H.H. Rijnaarts, P.M. Biesheuvel, d.W.A. Van, Optimization of salt adsorption

M

rate in membrane capacitive deionization, Water Research, 47 (2013) 1941-1952. [31] H. Li, C. Nie, L. Pan, Z. Sun, The study of membrane capacitive deionization from charge efficiency, Desalination & Water Treatment, 42 (2012) 210-215.

ED

[32] Z. Liang, H. Xu, Y. Wang, S. Yang, D. Ping, Using oxygen plasma treatment to improve the performance of electrodes for capacitive water deionization, Electrochimica Acta, 106 (2013) 494-499. [33] A. Kondo, F. Murakami, K. Higashitani, Circular dichroism studies on conformational changes in

PT

protein molecules upon adsorption on ultrafine polystyrene particles, Biotechnology & Bioengineering, 1 (1992) 889-894.

[34] N. Varga, V. Hornok, D. Sebők, I. Dékány, Comprehensive study on the structure of the BSA from

CC E

Extended - to Aged form in wide (2-12) pH range, International Journal of Biological Macromolecules, 88 (2016) 51-58.

[35] Z. Peng, D. Zhang, L. Shi, T. Yan, S. Yuan, H. Li, R. Gao, J. Fang, Comparative Electroadsorption Study of Mesoporous Carbon Electrodes with Various Pore Structures, Journal of Physical Chemistry C, 115 (2011) 17068-17076.

A

[36] Y. Li, C. Zhang, Y. Jiang, T.-J. Wang, H. Wang, Effects of the hydration ratio on the electrosorption selectivity of ions during capacitive deionization, Desalination, 399 (2016) 171-177. [37] F. Zietzschmann, E. Worch, J. Altmann, A.S. Ruhl, A. Sperlich, F. Meinel, M. Jekel, Impact of EfOM size on competition in activated carbon adsorption of organic micro-pollutants from treated wastewater, Water Research, 65 (2014) 297-306. [38] P.A. Quinlivan, L. Li, D.R. Knappe, Effects of activated carbon characteristics on the simultaneous adsorption of aqueous organic micropollutants and natural organic matter, Water Research, 39 (2005) 1663-1673.

[39] M. Elimelech, X. Zhu, A.E. Childress, S. Hong, Role of membrane surface morphology in colloidal fouling of cellulose acetate and composite aromatic polyamide reverse osmosis membranes, J.membr.sci, 127 (1997) 101-109.

Figure caption

IP T

Fig.1 Zeta potential of ACF at different pH Fig.2 FTIR spectra of ACFs

20 mL/min, temperature of 20 ℃ and voltage of 1.2 V.

SC R

Fig.3 Desorption ratio at different solution pH in MCDI and CDI with the flow rate of

Fig.4 Desorption ratio at different voltages in MCDI and CDI with the flow rate of 20

N

U

mL/min, temperature of 20 ℃ and pH of 6.

A

Fig.5 Desorption ratio at different temperatures in MCDI and CDI with the flow rate of

M

20 mL/min, voltage of 1.2 V and pH of 6.

Fig.6 Desorption ratio at different flow rates in MCDI and CDI with the temperature of

ED

20 ℃, voltage of 1.2 V and pH of 6.

PT

Fig.7 Comparison of desorption performance in MCDI and CDI (a) pH; (b) voltage; (c) temperature; (d) flow rate.

CC E

Fig.8 Desorption ratio in the presence of BSA with different concentrations in MCDI and CEM-CDI with the flow rate of 20 mL/min, temperature of 20 ℃, pH of 6 and

A

voltage of 1.2 V.

Fig.9 Proposed mechanism in the desorption process (a) in CDI; (b) in MCDI; (c) in CEM-CDI in the presence of BSA; (d) in MCDI in the presence of BSA.

A ED

PT

CC E

IP T

SC R

U

N

A

M

A ED

PT

CC E

IP T

SC R

U

N

A

M

A ED

PT

CC E

IP T

SC R

U

N

A

M

A ED

PT

CC E

IP T

SC R

U

N

A

M

A ED

PT

CC E

IP T

SC R

U

N

A

M

A ED

PT

CC E

IP T

SC R

U

N

A

M

A ED

PT

CC E

IP T

SC R

U

N

A

M

A ED

PT

CC E

IP T

SC R

U

N

A

M

IP T SC R U A

N SBET (m2/g)

ACF

899.6

PT CC E A

VT (cm3/g)

VM (cm3/g)

DP (nm)

0.67

0.46

3.0

ED

Sample

M

Table 1 Properties of ACF