Activated carbon nanofiber webs made by electrospinning for capacitive deionization

Electrochimica Acta 69 (2012) 65–70

Contents lists available at SciVerse ScienceDirect

Electrochimica Acta journal homepage: www.elsevier.com/locate/electacta

Activated carbon nanofiber webs made by electrospinning for capacitive deionization Gang Wang a , Chao Pan a,b , Liuping Wang a , Qiang Dong a , Chang Yu a , Zongbin Zhao a , Jieshan Qiu a,∗ a Carbon Research Laboratory, Liaoning Key Lab for Energy Materials and Chemical Engineering, State Key Lab of Fine Chemicals, School of Chemical Engineering, Dalian University of Technology, Dalian 116024, China b College of Science, Dalian Ocean University, Dalian 116023, China

a r t i c l e

i n f o

Article history: Received 27 October 2011 Received in revised form 20 February 2012 Accepted 20 February 2012 Available online 28 February 2012 Keywords: Polyacrylonitrile Electrospinning Activated carbon fibers Desalination Capacitive deionization

a b s t r a c t Activated carbon fiber (ACF) webs with a non-woven multi-scale texture were fabricated from polyacrylonitrile (PAN), and their electrosorption performance in capacitive deionization for desalination was investigated. PAN nanofibers were prepared by electrospinning, followed by oxidative stabilization and activation with carbon dioxide at 750–900 ◦ C, resulting in the ACF webs that were characterized by Xray diffraction, Raman spectroscopy, scanning electron microscopy and nitrogen adsorption. The results show that the as-made ACFs have a specific surface area of 335–712 m2 /g and an average nanofiber diameter of 285–800 nm, which can be tuned by varying the activation temperature. With the ACF webs as an electrode, an electrosorption capacity as high as 4.64 mg/g was achieved on a batch-type electrosorptive setup operated at 1.6 V. The ACF webs made by electrospinning are of potential as an excellent electrode material for capacitive deionization for desalination. © 2012 Elsevier Ltd. All rights reserved.

1. Introduction The reliable supply of clean fresh water is becoming increasingly difficult as the world develops quickly, and this has become a bottleneck issue for the sustainable development of some countries such as China. It is estimated that there is approximately 1.38 billion km3 of fresh water in the world, unfortunately, only 2.5% of this water resource is drinkable [1]. Seawater desalination is one of the important alternative ways to get fresh drinkable water. Up to now, several processes for desalination including membrane separation such as reverse osmosis (RO) and electrodialysis, and thermal vaporization such as multi-stage flash distillation (MSF) and multieffect distillation, have been widely explored and used. Among these processes, the RO and the MSF processes have been employed by the majority of water treatment plants (ca. 90%) worldwide for desalination of seawater [2]. The cost of seawater desalination is governed by a number of factors such as the desalination process and the production capacity of the seawater treatment plant [3]. In the case of the RO technique, about 1.5–2.5 kWh of electricity is normally consumed to produce 1 m3 of fresh water, while in the case of thermal distillation process, the same amount of electricity consumed will produce 10 times more of fresh water. In reality, only those rich countries such as Saudi Arabia can afford to run such

∗ Corresponding author. Tel.: +86 411 84986080; fax: +86 411 84986080. E-mail address: [email protected] (J. Qiu). 0013-4686/$ – see front matter © 2012 Elsevier Ltd. All rights reserved. doi:10.1016/j.electacta.2012.02.066

seawater desalination facilities. As such, new technologies are highly demanded to further decrease the desalination cost. Capacitive deionization (CDI) technology is an electrochemically governed method for removing salt from aqueous solutions, in which the carbonaceous electrodes are one of the key materials. When a potential is applied to the electrode, the ions in the solution will be charged, and moved to and adsorbed on the electrode to form an electrical double-layer. After the potential is removed, the adsorbed ions are quickly released back into the bulk solution [4]. The CDI process is an environmental friendly technology for seawater desalination because no contaminants are produced in the whole process, in which the energy consumption is low because the CDI process can be operated at a low potential and no electrolysis occurs at the electrodes [5]. New carbon materials such as carbon aerogel [6], carbon clothes [7], carbon nanotubes [8], graphite [9], and ordered mesoporous carbon [10] have been tested as electrode-active materials for the CDI process. Previous works have demonstrated, to some degree, that carbon aerogels are one of the most promising available materials for CDI [1]. Nevertheless, for carbon aerogels, the obvious shortcomings are their high resistivity and low mechanical strength [11]. In the present work, a CDI unit cell was constructed to evaluate the salt-removal efficiency of electrodes made of activated carbon fiber (ACF) webs prepared by electrospinning [12]. The nonwoven ACF webs were stabilized, carbonized and activated in CO2 before use, which led to ACF webs with a high specific surface area

66

G. Wang et al. / Electrochimica Acta 69 (2012) 65–70

and a relatively narrow pore-size. With the web-like ACFs as an electrode, it has more benefits than the traditional powder carbonaceous materials that must be mixed and combined with a polymer binder and an electric conductor such as carbon black [13], and the binder would increase the internal resistance and block some of the pores in the carbon materials, which subsequently resulted in lower adsorption capacity [14]. 2. Experimental 2.1. Electrode fabrication Polyacrylonitrile (PAN) (Mw = 150,000, Aldrich Co.) was dissolved in dimethyl formamide in a weight ratio of 10 wt.% by gently stirring for 4 h at 60 ◦ C, resulting in a PAN solution that was transferred into a 10 mL syringe with a capillary tip (0.8 mm in diameter). The electrospinning was conducted at a high positive voltage (22 kV) that was applied to the polymer solution via the syringe needle tip. The electrospun fibers were collected as a thin web on a rotating metal drum wrapped in aluminum foil, and the drum was rotated at a rate of 400 rpm. The electrospun fiber webs were stabilized in air at 280 ◦ C for 2 h with a heating rate of 1 ◦ C/min, and then activated at 750–900 ◦ C at a ramping rate of 5 ◦ C/min. When the activation temperature was reached, a CO2 flow of 150 mL/min was introduced into the reactor and continued for 0.5 h, then, the CO2 flow was replaced by flowing argon. The samples were cooled down to below 50 ◦ C in flowing argon before taken out from the furnace. The as-made ACF webs were denoted as ACFX, where X stands for the activation temperature. The electrospun ACF webs were examined using scanning electron microscopy (SEM, Hitachi, S-4700, Japan). The content of carbon, nitrogen and hydrogen in the samples was analyzed using a Vario EL III analyzer (Elementar, Germany). The specific surface area and pore-size distribution were evaluated using the nitrogen adsorption (ASAP2020, Micromeritics, USA). Before the measurement, the ACF webs were degassed at 250 ◦ C for 5 h under vacuum. X-ray powder diffraction (XRD) examination of the ACF web samples was performed on a D/Max-III type X-ray spectrometer with Cu K␣ radiation at 40 kV and 100 mA. Raman spectra of the samples were recorded on a LabRam-010 spectrometer with a resolution of 2 cm−1 , a 514.53 nm−1 laser beam with an intensity of 1000 mW and a slit width of 3.5 cm−1 . 2.2. Electrochemical measurements Electrochemical performance of the electrodes fabricated from the ACF webs was evaluated using cyclic voltammetry (CV). A threeelectrode cell assembly was utilized with an Hg/HgO reference electrode, a counter electrode of Pt, and a working electrode with a diameter of ca. 1 cm in 6 M KOH at 25 ◦ C. The potential was swept between −0.9 and −0.1 V at a scan rate of 2–50 mV/s. The gravimetric capacitance (C, F/g), i.e. the specific capacitance per mass weight carbon nanofiber, is calculated using the following equation [15]: 1 C= mV v



Fig. 1. Diagram of CDI experiment setup.

measured in a flow-through setup, as shown in Fig. 1. Electrodes were placed face to face at both sides of an insulating nylon net (thickness: 0.2 mm) with an area of 50 mm × 70 mm. The unit cell was prepared by fixing the electrodes with plexiglass on both sides. A flow channel was created by punching a 1-cm-diameter hole in the bottom of the end plate to let the solution run through a spacer to the other end plate. For each run, the solution was continuously pumped into the cell at a flowing rate of 5 mL/min using a peristaltic pump (Loner BT100, China). The total solution volume was maintained at 50 mL and the solution temperature was kept 25 ◦ C. During the measurement, the potential difference between the two electrodes was kept at a constant voltage of 1.6 V using a programmable DC power supply (PST-3202, Gwinstek, Taiwan). The changes of NaCl concentration in the solution were continuously monitored using an ion conductivity meter (ET915, eDAQ TECH, Australia). The electrosorptive capacity (Q) is defined as below: Q =

(C0 − C)VNaCl M

where C0 and C (mg/L) are the initial and final concentrations of NaCl, VNaCl is the solution volume (mL), and M is the mass of the ACF web electrodes (g). 3. Results and discussion 3.1. Morphology and structure analysis The white electrospun fiber webs were easily obtained under the electrospinning conditions as described above. After heat treatment, they are turned to black carbon materials with a very smooth surface, as shown in Fig. 2. The ACF webs were directly mounted to

V0 +V

I(V )dV

(1)

V0

where m is the mass of the carbon nanofiber (g), v is the scan rate (V/s), V is the sweep potential range during discharging (V), and I(V) is the corresponding current density (A/g). 2.3. Batch-mode electrosorptive experiment The ACF webs were used as CDI electrodes and directly attached to the graphite current collectors to make a CDI cell. The adsorption removal efficiency of ions on the ACF web electrodes was

(2)

Fig. 2. Photograph of the ACF web after heat treatment.

G. Wang et al. / Electrochimica Acta 69 (2012) 65–70

67

Fig. 3. SEM images of (a) ACF750, (b) ACF800, and (c) ACF900.

graphite foil supports with conductive epoxy strips when used in the CDI cell. The SEM images of the ACF webs activated at different temperatures are shown in Fig. 3. The nanofibers in the ACF webs have a regular and flexuous fibrous morphology, of which the diameter becomes smaller as the activation temperature increases from 750 ◦ C to 900 ◦ C. The average diameter of the nanofibers is 285 nm for the ACF webs activated at 900 ◦ C. The shrinkage in diameter is due to the reactions during the thermal stabilization and the activation steps. The XRD patterns of the ACF webs made at different activation temperatures are shown in Fig. 4, showing a strong peak between 20◦ and 30◦ for all samples. This can be attributed to the (0 0 2) diffraction of the graphitic crystallites. In addition, a broad peak around 2 = 43◦ for all the ACF webs can also be observed, which is due to the graphite basal plane. The large width of this peak and a complete absence of higher order peaks in the XRD patterns suggest a disordered amorphous structure for the ACF webs [16].

Nevertheless, as the activation temperature increases, the (0 0 2) peak becomes sharper and upshifts slightly, implying a little bit more ordered structure, though it is far from the perfect graphitized structure. The samples were further examined using Raman spectroscopy to get more detailed information about the structure of the ACF webs, of which the results are shown in Fig. 5. It is known that the D-band at 1350 cm−1 is associated with the disordered graphite while the G-band at 1582 cm−1 is associated with the ordered graphite. The D-band and G-band are also related to the finite particle size effect or lattice distortion and the vibrations in all sp2 bonded carbon atoms in a 2D hexagonal lattice [17]. As the activation temperature of the ACF webs increases from 750 ◦ C to 900 ◦ C, the structure of the ACF webs becomes more ordered, evidenced by the fact that the ratio of the integrated intensity of the D- and G-bands (ID /IG ) becomes smaller (Fig. 5, inset). The Raman results are in good agreement with the XRD results discussed above.

68

G. Wang et al. / Electrochimica Acta 69 (2012) 65–70

Fig. 4. The XRD patterns of the ACF webs.

Both the XRD and Raman data have revealed that the electrospun ACF webs are essentially disordered materials. However, the graphitization of all these samples has been found to increase with increasing the activation temperature. Because of this, the conductivity of the electrospun ACF webs is expected to increase with the increasing the activation temperature. Higher conductivity is helpful for improving the desalination performance of capacitive deionization [18]. 3.2. Characterization of the ACF webs Fig. 6a illustrates the nitrogen adsorption and desorption isotherms of the PAN-based ACF webs prepared at different activation temperatures. According to the IUPAC classification, the isotherms of the ACF webs are typical type I, indicative of the domination of micropores in the porous structure. Although the isotherms of the three ACF webs made at 750 ◦ C, 800 ◦ C and 900 ◦ C are identical in shape, the specific adsorption quantity of nitrogen differs greatly, implying the difference in pore and texture structure. The specific surface area tends to increase with increasing temperatures up to 900 ◦ C. This trend is commonly observed for ACFs produced by CO2 activation [19]. The detailed texture data and the ultimate analysis results of the ACF webs are listed in Table 1. The specific surface area is noted to be directly proportional to the burn-off degree. Furthermore, all of the samples have a dual-mode pore structure, i.e. both micropore and mesopore are

present in the porous structures. As shown in Table 1, the volume distribution ratios of mesopore to the total volume are 26% for ACF750, 21% for ACF800, and 24% for ACF900, respectively. It is known that the efficiency of CDI strongly depends on the pore structure and surface property of electrodes, such as surface area, microstructure and size distribution of pores, chemical functional groups, and adsorption properties [20]. According to Ying and Yang [21], the mesopore can reduce electrical double-layer overlapping effect and increase electrosorption capacity. In our case, the ACF webs have well-developed mesoporous structure (see Fig. 6b) with an average pore diameter of ca. 2.0 nm, which is greatly beneficial to electrosorption [22]. The influence of the activation temperature on the C, H, and N content of the ACF webs is also shown in Table 1. As the activation temperature increases, the carbon content of ACF webs increases, while the nitrogen content of ACF webs becomes smaller, implying that nitrogen continues to be eliminated from the stabilized fibers during the activation step. Carbon monoxide is formed due to the reaction of the stabilized fiber with CO2 , resulting in the loss in the weight of the ACF webs [23]. These factors jointly result in the change in the morphology, size and texture of the ACF webs, evidenced by the SEM, XRD, Raman, nitrogen adsorption and elemental analysis as discussed above. 3.3. Electrochemical characteristics CV curves obtained at different scan rates with the ACF webs made at different activation temperatures are shown in Fig. 7. The CV curves show a typical ideal capacitor behavior with a nearly rectangular shape at different scan rates, though they distort a little bit at high voltage scan rates. The shape of the curves becomes more rectangular as the scan rate increases, while at the same time, the voltammetric currents increase. These imply that in the case of lower potential scan rates, the electrolytes have sufficient time to enter the micropores, which contributes greatly to the formation of double-layer capacitance [24,25]. Because of this, the electrosorption deionization becomes more efficient. The specific capacitance calculated from the CV curves is shown in Fig. 7d. It should be noted that the currents increase monotonically as the scan rate increases, indicating the dependence of voltammetric current on the scan rate. The capacitance values become smaller with the increase in the scan rate, which is not uncommon for the electrochemical systems. Generally, the larger the specific surface of porous carbons, the higher the capability for the accumulation of charges in the electrode/electrolyte interface that leads to higher gravimetric capacitance [26]. Correspondingly, the capacitance (at a voltage scan rate of 2 mV/s) increases as the activation temperature increases, evidenced by the fact that the capacitance is 172 F/g for ACF750, 190 F/g for ACF800, and 228 F/g for ACF900. Higher capacitance means more electrolyte ions have reached the pores of the ACF electrodes. As the activation temperature increases, the pores in the ACF web electrodes are better connected that makes more pores be accessible by the ions. Therefore, ACF900 web electrode can be expected to have a better performance than ACF750 and ACF800 web electrode in the capacitive desalination process. 3.4. CDI desalination performance Fig. 8 shows the electrosorptive removal curves of Na+ and ions on the ACF web electrodes. The mass of ACF750, ACF800 and ACF900 web electrodes are 0.35, 0.38 and 0.31 g, respectively. The calculated electrosorption capacity according to Eq. (2) is 2.74 mg/g for ACF750, 2.99 mg/g for ACF800, and 4.64 mg/g for ACF900. Table 2 lists the electrosorption capacity of different Cl−

Fig. 5. The Raman spectra of the ACF webs (inset: the ratio of integrated intensity of the D- and G-bands vs activation temperature).

G. Wang et al. / Electrochimica Acta 69 (2012) 65–70

69

Fig. 6. Nitrogen adsorption and desorption isotherms (a) and pore size distributions (b) of the ACF webs.

Table 1 The burn-off, porous parameters and ultimate analysis results of the PAN-based ACF webs activated at different temperatures. Sample

ACF750 ACF800 ACF900 a

Burn-off (%)

49.6 59.8 69.2

BET (m2 /g)

335 442 712

Vtotal (cm3 /g)

0.202 0.223 0.363

Vmeso (cm3 /g)

0.053 0.046 0.089

Daverage (nm)

2.41 2.02 2.04

Ultimate analysis (%, daf) C

H

N

Oa

70.53 72.54 76.33

1.87 1.88 2.01

16.81 14.43 6.64

10.79 11.15 15.02

By difference.

carbon electrode materials. Though the electrosorption capacity is obtained under different test conditions, the electrosorption capacity of ACF900 electrodes is consistently higher than that of other electrode materials, as shown in Table 2. It should be noted that the ACF webs made by electrospinning followed by air-oxidation stabilization and CO2 activation are binder-free

non-woven multi-scale carbon. This unique feature makes it possible to enhance the electrical conductivity that finally leads to better capacitive deionization performance. Of course, the improved electrosorption capacity might also be due to the pseudo-faradaic properties of the nitrogen functionality of ACF webs to some degrees. This needs to be further clarified in the future.

Fig. 7. CV curves for (a) ACF750, (b) ACF800, (c) ACF900 at scan rates of 2–50 mV/s, and (d) specific gravimetric capacitance as a function of the voltage scan rates.

70

G. Wang et al. / Electrochimica Acta 69 (2012) 65–70

Table 2 Comparison of electrosorption capacity of various carbon electrodes. Electrodes

Surface area (m2 /g)

Initial conductivity (␮S/cm)

Final conductivity (␮S/cm)

Electrode mass (g)

Applied voltage (V)

Electrosorption capacity (mg/g)

Activated carbon [27] Woven carbon fibers [28] Carbon aerogel [29] CNT [30] CNTs–CNFs [8] Mesoporous carbon [10] Graphene [9] ACF900 (the present work)

1260 1980 400–1100 129 211 844 14.2 712

420 – 100 – 100 50 50 192

109 – 18 – 40 25 39 122

1.3 – 131.8 15 0.85 2.0 – 0.31

1.5 1.0 1.2 1.2 1.2 1.2 2.0 1.6

3.68 1.87 3.33 1.73 3.32 0.69 1.85 4.64

Research Funds for the Central Universities (no. DUT11ZD120), the Research Start-up Fund in DUT (no. DUT12RC(3)04), the Postdoctoral Science Foundation of China (no. 20100470071), and the NSFC (nos. 20836002, 51102033). References

Fig. 8. Electrosorption behavior of the ACF web electrodes in CDI at 1.6 V.

4. Conclusions ACF webs are prepared from electrospun PAN by air-oxidation stabilization and CO2 -activation at different temperatures. The asmade ACF webs have high specific surface area and a dual-mode pore size distribution, and feature a non-woven multi-scale texture, which endows it with excellent electrochemical performance. It has been found that higher activation temperature leads to higher capacitance and electrosorption capacity. The salt-removal tests in a CDI unit cell with electrodes made of ACF webs show a high electrosorption capacity of 4.64 mg/g for ACF900 at 1.6 V and a flow rate of 5 mL/min of saline water, demonstrating that the ACF web electrodes made by electrospinning are of potential in electrochemical capacitive deionization for desalination of sea water. Acknowledgments This work is partly supported by the Dalian Science and Technology Bureau of China (no. 2010A17GX095), the Fundamental

[1] I. Villar, S. Roldan, V. Ruiz, M. Granda, C. Blanco, R. Menéndez, R. Santamaria, Energy Fuels 24 (2010) 3329. [2] R. Borsani, S. Rebagliati, Desalination 182 (2005) 29. [3] I.C. Karagiannis, P.G. Soldatos, Desalination 233 (2008) 448. [4] Y.J. Kim, J.H. Choi, Water Res. 44 (2010) 990. [5] M.A. Anderson, A.L. Cudero, J. Palma, Electrochim. Acta 55 (2010) 3845. [6] C.J. Gabelich, T.D. Tran, I.H.M. Suffet, Environ. Sci. Technol. 36 (2002) 3010. [7] M.W. Ryoo, J.H. Kim, G. Seo, J. Colloid Interface Sci. 264 (2) (2003) 414. [8] X. Wang, M. Li, Y. Chen, R. Cheng, S. Huang, L. Pan, Z. Sun, Appl. Phys. Lett. 89 (2006) 053127. [9] H. Li, T. Lu, L. Pan, Y. Zhang, Z. Sun, J. Mater. Chem. 19 (2009) 6773. [10] L. Zou, L. Li, H. Song, G. Morris, Water Res. 42 (2008) 2340. [11] K.Y. Foo, B.H. Hameed, J. Hazard. Mater. 170 (2009) 552. [12] V. Presser, L.F. Zhang, J. Niu, J. McDonough, C. Perez, H. Fong, Y. Gogosti, Adv. Energy Mater. 1 (2011) 423. [13] M. Wang, Z.H. Huang, L. Wang, M.X. Wang, F.Y. Kang, H.Q. Hou, New J. Chem. 34 (2010) 1843. [14] B.H. Park, J.H. Choi, Electrochim. Acta 55 (2010) 2888. [15] J. Yan, Z. Fan, T. Wei, J. Cheng, B. Shao, K. Wang, L. Song, M. Zhang, J. Power Sources 194 (2009) 1202. [16] A. Jänes, H. Kurig, E. Lust, Carbon 45 (2007) 1226. [17] G. Sun, X. Li, H. Yan, J. Qiu, Y. Zhang, Carbon 46 (2008) 476. [18] S. Nadakatti, M. Tendulkar, M. Kadam, Desalination 268 (2011) 182. [19] H. Tavanai, R. Jalili, M. Morshed, Surf. Interface Anal. 41 (2009) 814. [20] P. Xu, J.E. Drewes, D. Heil, G. Wang, Water Res. 42 (2008) 2605. [21] T.Y. Ying, K.L. Yang, J. Colloid Interface Sci. 250 (2002) 18. [22] Y. Zhan, C. Nie, H. Li, L. Pan, Z. Sun, Electrochim. Acta 56 (2011) 3164. [23] Y.J. Su, T.H. Ko, J.H. Lin, J. Appl. Polym. Sci. 108 (2008) 3610. [24] C. Kim, K.S. Yang, Appl. Phys. Lett. 83 (2003) 1216. [25] J. Chmiola, G. Yushin, R.K. Dash, E.N. Hoffman, J.E. Fischer, M.W. Barsoum, Y. Gogosti, Electrochem. Solid State Lett. 8 (2005) A357. [26] P. Simon, Y. Gogotsi, Nat. Mater. 7 (2008) 845. [27] Y.J. Kim, J.H. Choi, Sep. Purif. Technol. 71 (2010) 70. [28] M.W. Ryoo, G. Seo, Water Res. 37 (2003) 1527. [29] J.C. Farmer, D.V. Fix, G.V. Mack, R.W. Pekala, J.F. Poco, J. Electrochem. Soc. 143 (1996) 159. [30] K. Dai, L.Y. Shi, J. Fang, D. Zhang, B. Yu, Mater. Lett. 59 (2005) 1989.