Ionic Liquid Electrolytes with Various Constituent Ions for Graphene-based Supercapacitors

Electrochimica Acta 161 (2015) 371–377

Contents lists available at ScienceDirect

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

Ionic Liquid Electrolytes with Various Constituent Ions for Graphene-based Supercapacitors Po-Ling Huang a , Xu-Feng Luo a , You-Yu Peng b , Nen-Wen Pu c, Ming-Der Ger b , Cheng-Hsien Yang a , Tzi-Yi Wu d, Jeng-Kuei Chang a, * a

Institute of Materials Science and Engineering, National Central University, Taiwan Department of Chemical and Materials Engineering, National Defense University, Taiwan Department of Photonics Engineering, Yuan Ze University, Taiwan d Department of Chemical and Materials Engineering, National Yunlin University of Science and Technology, Taiwan b c

A R T I C L E I N F O

A B S T R A C T

Article history: Received 25 December 2014 Received in revised form 12 February 2015 Accepted 13 February 2015 Available online 16 February 2015

Although activated carbon shows a higher maximum capacitance than that of graphene nanosheets (GNSs) in a conventional organic electrolyte, the latter material, characterized by high conductivity and a unique planar structure, is more suitable for use in an ionic liquid (IL) electrolyte for supercapacitors. IL electrolytes consisting of various cations (1-ethyl-3-methylimidazolium (EMI+) and N-butyl-Nmethylpyrrolidinium (BMP+)) and anions (bis(trifluoromethylsulfony) imide (TFSI ), tetrafluoroborate (BF4 ), and dicyanamide (DCA )) are systematically studied. Among them, BMP-DCA IL is found to be the superior electrolyte, in which the GNS electrode exhibits a capacitance of 235 F g 1 and a satisfactory rate capability within a potential range of 3.3 V at 25  C. This electrolyte is even more promising for elevatedtemperature applications. At 60  C, a symmetric-electrode GNS supercapacitor with BMP-DCA IL is able to deliver maximum energy and power densities of 103 Wh kg 1 and 43.3 kW kg 1 (based on the active material on both electrodes), respectively, which are much higher than 19 Wh kg 1 and 17.6 kW kg 1 for a control cell with a conventional organic electrolyte. ã 2015 Elsevier Ltd. All rights reserved.

Keywords: graphene ionic liquid electrolyte supercapacitor

1. Introduction With the upcoming depletion of fossil fuels coupled with growing concerns about pollution and global warming, the development of clean and effective energy conversion/storage systems that can meet future power requirements has become increasingly important [1,2]. Supercapacitors, including electric double-layer capacitors (EDLCs) and pseudocapacitors, are promising energy storage devices owing to their higher power density, wider operation temperature window, superior cyclic stability, and greater charge–discharge efficiency compared to those of batteries [3]. EDLCs, which are based on a storage mechanism of nonfaradaic charge separation at the electrode/electrolyte interface, are commonly used in practical applications because they are stable, inexpensive, and easily mass-produced [4]. Activated carbon (AC) electrodes and organic electrolytes are typically used in traditional EDLCs [5]. The development of a more effective

* Corresponding author at: 300 Jhong-da Road, National Central University, Taoyuan, Taiwan. Tel.: +886 3 4227151x34908; fax: +886 3 2805034. E-mail address: [email protected] (J.-K. Chang). http://dx.doi.org/10.1016/j.electacta.2015.02.115 0013-4686/ ã 2015 Elsevier Ltd. All rights reserved.

carbon material and a compatible electrolyte to further improve EDLC performance has attracted a lot of research attention. Increasing the energy density of supercapacitors to approach that of batteries is an important topic as it extends the application domain of supercapacitors [6–8]. One effective way to achieve this is to find an advanced electrode material. Traditional AC, despite its high surface area, typically has unsatisfactory electric conductivity (and thus is unfavorable for high-rate performance). Moreover, its small pores may not be easily accessible to electrolytes, leading to a low double-layer capacitance. Carbon aerogels, ordered mesoporous carbon, and hierarchical porous carbon are attractive new materials for EDLCs [9–11]. Nevertheless, their high cost and low yield limit widespread application. Graphene nanosheets (GNSs), characterized by a two-dimensional (2-D) architecture with large accessible area, have excellent conductivity and high double-layer capacitance and thus have become a recent research focus [12,13]. Moreover, it was reported that GNSs can be manufactured in ton quantities at low cost and high uniformity [14]. It is noted that to optimize an EDLC electrode, an appropriate design of the corresponding electrolyte (type and composition) is of great significance. However, this issue has not been well explored for GNS electrodes, and thus further investigation is needed.

372

P.-L. Huang et al. / Electrochimica Acta 161 (2015) 371–377

To increase the energy density (ED) and power density (PD) of a supercapacitor, increasing the cell voltage (V) is an effective strategy because both ED and PD are proportional to the square of V [3]: ED = (Ccell  V2)/2

(1)

PD = (Ccell  V2)/(2  t)

(2)

where Ccell is the specific cell capacitance and t is the discharge time. Therefore, an alternative electrolyte with a large potential window compared to that of a conventional organic electrolyte (such as propylene-carbonate-based solution, which has a potential window of 2.5 V) is highly desired. Ionic liquids (ILs) are excellent electrolyte candidates [15,16]. Besides having large potential windows, they possess intrinsic ionic conductivity and a high concentration of cations/anions. Using ILs can also eliminate the safety and environmental concerns raised by organic electrolytes, since ILs have excellent thermal stability, non-flammability, and non-volatility [17]. Similar drawbacks (safety and environmental issues) of the organic electrolyte in other electrochemical devices (such as Li batteries and dye-sensitized solar cells) have also been concerned [18,19]. A combination of GNS electrodes and IL electrolytes has shown promising results. 1-ethyl-3-methylimidazolium tetrafluoroborate (EMI-BF4) [20–24] and 1-ethyl-3methylimidazolium bis(trifluoromethylsulfony) imide (EMI-TFSI) [25–27] are the most commonly used IL electrolytes. However, it was found that their performance varies. For instance, the reported potential windows for EMI-BF4 range from 3.5 to 4 V and the measured GNS capacitances are between 90–230 F g 1 [20–24]. Moreover, most previous studies have focused on only one kind of

IL. The lack of a systematic comparison makes it difficult to tell which IL is most suitable for GNS electrodes. To have a further insight into this subject, the present study investigates various IL electrolytes. The effects of the constituent cations and anions on Ccell,V, ED, and PD are systematically discussed. The BMP-DCA IL is found to be a promising electrolyte for GNS-based EDLCs. 2. Experimental procedure GNSs were synthesized using a modified Staudenmaier method; the preparation details can be found in our previous papers [28,29]. Briefly, graphite oxide (GO), chemically oxidized from natural graphite (Alfa Aesar; particle size: 70 mm; purity: 99.999%), was reduced and exfoliated by heating (1  C min 1) to 300  C in an inert Ar atmosphere, yielding GNSs. Besides GNSs, AC (Kuraray, YP-50F) was used in this study for comparison. The microstructure of the GNSs was examined with transmission electron microscopy (TEM; JEOL 2100 F). X-ray diffraction (XRD; Bruker D8 Advance) was used to explore the crystallinity. In the analyses, the X-ray detector was scanned at a speed of 1 min 1. A Raman spectrometer (UniRAM MicroRaman) was employed to study the bonding structure of GNSs. The spectrum was excited by a diode-pump solid-state laser with a wavelength of 532 nm. The GNS thickness was characterized using atomic force microscopy (AFM; Veeco/DI NanoMan D3100CL). To make a supercapacitor electrode, a slurry, prepared by mixing 70 wt% graphene (or AC), 20 wt% Super P, and 10 wt% poly (vinylidene fluoride) in N-methyl-2-pyrrolidone solution, was pasted onto nickel foam and vacuum-dried at 110  C for 8 h. The active material loading was approximately 0.5 mg cm 2. EMI-TFSI, N-butyl-N-methylpyrrolidinium bis(trifluoromethylsulfony) imide

Fig. 1. (a) TEM image, (b) Raman spectrum, (c) AFM image, and (d) height profile across the marked line in (c) of synthesized GNSs.

P.-L. Huang et al. / Electrochimica Acta 161 (2015) 371–377

(BMP-TFSI), EMI-BF4, 1-ethyl-3-methylimidazolium dicyanamide (EMI-DCA), and N-butyl-N-methylpyrrolidinium dicyanamide (BMP-DCA) ILs were used as the electrolytes. The ILs were prepared and purified following published methods [30,31]. Their water contents, measured using a Karl Fisher titrator, were 80–100 ppm. A three-electrode cell controlled with an AUTOLAB (PGSTAT 302 N) potentiostat was adopted to study the electrochemical properties. The reference electrode was a Pt wire placed in a fritted glass tube containing BMP-TFSI IL that had a ferrocene/ ferrocenium couple (Fc/Fc+ = 50/50 mol%) as a potential standard. The counter electrode was a graphite sheet, which was directly immersed in the electrolyte. Cyclic voltammetry (CV) was performed in an argon-filled glove box (Innovation Technology Co. Ltd.), where both the moisture content and oxygen content were maintained at below 1 ppm. Two-electrode symmetric cells (0.15 mg active material at each 0.3-cm2 electrode) were also assembled to evaluate the full-cell performance. Glassy fiber separators were used. A conventional organic electrolyte, consisting of propylene carbonate (PC, 99.7 wt%, Sigma–Aldrich) solvent and 1 M tetraethylammonium tetrafluoroborate (TEABF4, 99 wt%,

20 15 10 5 0 -5 -10 -15 -20

Alfa Aesar) solute, was also used for comparison. Electrochemical impedance spectroscopy (EIS) analyses were conducted for the two-electrode cells at a cell voltage of 0 V. The frequency range and the potential amplitude were 105–10 2 Hz and 10 mV, respectively. 3. Results and discussion A TEM bright-field image of a prepared GNS is shown in Fig. 1(a); the transparency indicates that the GO was effectively exfoliated during the thermal reduction reaction. The wrinkles on the GNS are related to the existence of residual functional groups, which cause local surface nonunifomity [32]. These wrinkles are supposed to alleviate re-stacking of GNSs, which is beneficial for preserving the electro-active sites for double-layer charging/ discharging. Fig. 1(b) shows the obtained Raman spectrum of GNSs. The D band (1360 cm 1) and G band (1585 cm 1), corresponding to imperfect sp2 carbon bonding and Ramanallowed phonon vibration of well-ordered graphite, respectively, were clearly observed. The defective sites (associated with the D signal) originated from graphite oxidation and GO exfoliation

(b) Current density (A g -1)

3.5 V increase scan rate

-2

-1

0

1

2

20 15 10 5 0 -5 -10 -15 -20

+ Potential (V vs. Fc/Fc ) Current density (A g -1)

Current density (A g -1)

-1

0

1

2

(d) 20 15 10 5 0 -5 -10 -15 -20

3.5 V increase scan rate

-2

-1

0

1

2

20 15 10 5 0 -5 -10 -15 -20

+ Potential (V vs. Fc/Fc )

(f) 20 15 10 5 0 -5 -10 -15 -20

3.3 V increase scan rate

-3

-2

-1

0

+ Potential (V vs. Fc/Fc )

2.5 V! increase scan rate

-2

-1

0

+ Potential (V vs. Fc/Fc ) Current density (A g -1)

Current density (A g -1)

-2

+ Potential (V vs. Fc/Fc )

(c)

(e)

3.7 V increase scan rate

!

Current density (A g -1)

(a)

373

9 8 7 6 5 4 3 2 1 0

0

EMI"TFSI BMP"TFSI EMI"BF4 EMI"DCA BMP"DCA

10

20

30

40

50

Scan rate (mV s ) -1

Fig. 2. CV curves of GNS electrodes at various scan rates (from 5 to 50 mV s 1) recorded in (a) EMI-TFSI, (b) BMP-TFSI, (c) EMI-BF4, (d) EMI-DCA, and (e) BMP-DCA IL electrolytes. (f) CV current density (at midpoint potential) as function of potential scan rate in various IL electrolytes.

374

P.-L. Huang et al. / Electrochimica Acta 161 (2015) 371–377

processes. As also shown in Fig. 1(b), emergence of a clear 2D vibration band at 2680 cm 1, which results from a second-order two-phonon vibration mode [33,34], suggests that the GNSs were few-layer graphene. Fig. 1(c) shows the AFM analysis results of a GNS. It was found that the thickness of the nanosheet was approximately 4 nm, confirming that there were 3–5 graphene layers [35,36]. These data are consistent with a measured BET surface area of 550 m2 g 1 for the prepared GNSs. Fig. 2(a) shows the CV curves of the GNS electrode recorded with various potential scan rates in EMI-TFSI IL. The stable potential window, which was defined to avoid irreversible IL decomposition and to have symmetric anodic and cathodic charge (difference <2%), was found to be 3.5 V (see Fig. S1(a) for details). It was found that the Ni substrate showed negligible current; therefore, the specific capacitance (C) of GNSs can be calculated according to the following equation:

Cell Voltage (V)

4 3

i. EMI-TFSI ii. BMP-TFSI iii. EMI-BF4

2

iv. EMI-DCA v. BMP-DCA

1 0

i ii iii iv

0

v

100 200 300 400 500 600 700

Time (s)

C = Qm/DV (3)

Fig. 3. Galvanostatic (0.2 mA) charge–discharge curves of symmetric GNS cells with various IL electrolytes.

where Qm is the mass specific voltammetric charge integrated from both the anodic and cathodic CV sweeps, and DV is the potential scan range (i.e., 3.5 V  2). The obtained capacitances were 125, 116, 109, and 90 F g 1 at CV scan rates of 5, 10, 25, and 50 mV s 1, respectively. Fig. 2(b) shows the data measured in BMP-TFSI IL. Using BMP+ to replace EMI+ led to an increased potential window of 3.7 V (see Fig. S1(b)). This is associated with the cathodic decomposition potential of BMP+ being more negative than that of EMI+ [37,38]. The measure capacitance at 5 mV s 1 was 128 F g 1, which is close to that found for EMI-TFSI. However, as shown in Table 1, when the scan rate was increased to 50 mV s 1, the capacitance decreased more rapidly in BMP-TFSI than in EMI-TFSI (69 F g 1 vs. 90 F g 1). The lower ionic conductivity (and higher viscosity) of the former IL (as shown in Table S1) is unfavorable for electrode high-rate performance. The electrode CV behavior was also evaluated in EMI-BF4 IL; the obtained results are shown in Fig. 2(c). Within a potential window of 3.5 V, a more rectangular CV shape than those in Fig. 2(a) and (b) was observed. Moreover, the electrode capacitances (as shown in Table 1) were clearly higher than those measured in the TFSI-based ILs regardless of the potential scan rate (e.g., 145 F g 1 at 5 mV s 1). The constituent anions of ILs seem to predominantly govern the maximum capacitance of the GNS electrode. BMP-BF4 was also synthesized in this study; however, its melting point was 150  C. Since this compound is unsuitable as a capacitor electrolyte at room temperature, it was not further investigated in this study. The CV data recorded in EMI-DCA IL are shown in Fig. 2(d). Since the anodic decomposition of DCA–occurred at around +0.5 VFC/FC+ at the GNS electrode, the operational potential window was relatively narrow (i.e., 2.5 V) as compared to those found for EMI-TFSI and EMI-BF4 ILs. However, the maximum GNS capacitance increased to 225 F g 1 (at 5 mV s 1), which was considerably higher than those obtained for the previous ILs. As demonstrated in Fig. 2(e), replacing EMI+ with BMP+ results in an increase in the potential window to 3.3 V (for BMP-DCA) while the maximum capacitance remained as high as 235 F g 1. The aforementioned results (as summarized in Table 1) clearly indicate

that the IL anions play a key role in determining the maximum capacitance. This can be attributed to the fact that the anions are relatively small as compared to the cations (as shown in Table S2). Consequently, the former are more favorably adsorbed/desorbed at the electrode surface, dominating the electric double-layer capacitance. Moreover, the anion size increases in the sequence DCA– < BF4– < TFSI– (see Table S2), explaining the superior capacitances found in the DCA-based ILs. Fig. 2(f) shows the measured CV current density (at the midpoint potential) as a function of potential scan rate in various IL electrolytes. As illustrated, for the EMI-based ILs, the response current progressively increased with increasing scan rate, indicating a great electrode kinetic property. However, in the ILs with BMP+ cations, the CV current at a high rate (50 mV s 1) clearly deviated from the linear trend. The more bulky BMP+ spatially hindered the adsorption/desorption reaction at the electrode surface, especially at a high potential scan rate, and decreased the ion mobility in the electrolyte, leading to the poorer rate capability. Symmetric two-electrode full cells with various IL electrolytes were also assembled and tested. Fig. 3 shows the obtained galvanostatic (0.2 mA) charge–discharge curves within the cell voltages determined by CV (in Table 1). For all the cells, the charge and discharge branches are essentially linear and symmetric, reflecting ideal capacitive behavior and excellent reversibility of the IL electrolyte cells. The specific cell capacitance (Ccell) was calculated as: (4)Ccell = (I  t)/(V  m) where I is the applied current, t is the discharge time, V is the cell voltage, and m is the total active material mass (on both electrodes). The Ccell values for the EMI-TFSI, BMP-TFSI, EMI-BF4, EMI-DCA, and BMP-DCA cells are 30, 31, 36, 55, and 58 F g 1, respectively. These values are ideally about one-fourth of the specific GNS capacitances obtained from the 5 mV s 1 CV measurements, because a series connection of two capacitive electrodes in a cell makes the capacitance become half and because the total

Table 1 Specific capacitances at various CV scan rates and stable potential windows of GNS electrodes measured in five kinds of IL electrolyte. Specific capacitance (F g 5 mV s EMI-TFSI BMP-TFSI EMI-BF4 EMI-DCA BMP-DCA

125 128 145 225 235

1

1

)

10 mV s 116 112 141 198 207

Stable potential window (V) 1

25 mV s 109 88 128 168 171

1

50 mV s 90 69 102 155 135

1

3.5 3.7 3.5 2.5 3.3

P.-L. Huang et al. / Electrochimica Acta 161 (2015) 371–377

Energy density (Wh kg -1)

100

10

EMI"TFSI BMP"TFSI EMI"BF4 EMI"DCA BMP"DCA

1

Table 2 Maximum ED and PD of symmetric GNS cells with various IL electrolytes.

EMI-TFSI BMP-TFSI EMI-BF4 EMI-DCA BMP-DCA

Energy density (Wh kg 1)

Power density (kW kg 1)

51 59 61 48 88

9.9 5.2 12.1 17.1 17.5

50

Cdl !

40

Ci Rs

W Ri

!

-Zlm (ohm)

active material mass has been doubled (since two electrodes are included). These results also confirm that assembly of graphenebased EDLCs incorporating IL electrolytes is rather practicable. The ED and PD of the cells at various charge–discharge currents were calculated according to Eqs. (1) and (2), respectively; the obtained Ragone plots are shown in Fig. 4. Table 2 summaries the maximum ED and PD of various cells. While the BMP-TFSI cell had a higher ED (59 Wh kg 1) than that of the EMI-TFSI cell (51 Wh kg 1, attributed to the cell voltage difference), the latter cell possessed a superior PD (9.9 kW kg 1 vs. 5.2 kW kg 1), which can be explained by the EIS data shown in Fig. 5. As demonstrated in this figure, the Nyquist plots of the cells are characterized by a single semicircle in the high-frequency region, a short 45 -sloped (Warburg) line in the middle-frequency region, and a nearly vertical line (associated with double-layer capacitive behavior) at low frequency. The impedance spectra can be fitted by the equivalent circuit shown in the figure inset, where Rs, Ri, Ci, W, and Cdl are the electrolyte resistance, interfacial contact resistance (between the electrode/ electrolyte interface), interfacial contact capacitance, Warburg impedance, and double-layer capacitance, respectively [39,40]. The Warburg portion is inconspicuous, indicating that ion transport in the electrolyte is fast. The Rs and Ri values for the EMI-TFSI cell are less than those for the BMP-TFSI cell (see Table 3), resulting in the better high-power performance of the former cell. Fig. 4 also shows that the EMI-BF4 cell outperforms the TFSI-IL cells, which is consistent with the previous CV results. Specifically, superior ED (61 Wh kg 1) and PD (12.1 kW kg 1) were obtained for the EMI-BF4 cell. As demonstrated in Fig. 4, the EMI-DCA cell has the lowest ED, which is due to the low cell voltage (2.5 V). Nevertheless, the maximum PD of this cell was as high as 17.1 kW kg 1, which is mainly attributed to the much lower Rs and Ri values compared to those of the other IL electrolyte cells (Table 3). With EMI+ replaced with BMP+ in this IL, since the cell voltage significantly increased (3.3 V) while the high capacitance remained, optimal ED (88 Wh kg 1) and PD (17.5 kW kg 1) were achieved for the BMP-DCA cell. This is the first study to reveal that BMP-DCA IL is a promising electrolyte for graphene-based EDLCs. It was also found that the BMP-DCA cell showed a satisfactory cyclic stability. After 2000 cycles, the capacitance retention ratio was approximately 88%, which is comparable to that (90%) of a control cell with a conventional organic electrolyte. Fig. 6 compares the charge–discharge performance of the AC and GNS symmetric cells with TEABF4/PC and BMP-DCA electrolytes. As shown, in the conventional organic electrolyte (Fig. 6(a) and (b)), the AC cell shows higher capacitance than that of the GNS cell at a low current condition (0.2 mA). This is associated with the

375

30 EMI-TFSI BMP-TFSI EMI-BF4

20 10 0 0

EMI-DCA BMP-DCA

10

20

30

40

50

ZRe (ohm) Fig. 5. Nyquist plots of symmetric GNS cells with various IL electrolytes.

larger surface area of AC (1600 m2 g 1 vs. 550 m2 g 1 for GNSs). As the charge–discharge current increased (to 1 mA), the AC capacitance decayed quickly and became lower than that of GNS, which can be attributed to the lower electronic conductivity of AC. In contrast, with the IL electrolyte (Fig. 6(c) and (d)), the GNS cell clearly outperformed the AC cell regardless of the current applied. A large fraction of the AC area is attributed to internal cavities, which are not easily accessed by the viscous IL (50 cP), leading to the low measured capacitances. GNSs, characterized by a planar architecture with a large open surface area and high conductivity, are a unique electrode material for IL-based EDLCs. Synergistic capacitive performance was achieved by combining GNS electrodes and IL electrolyte. The effects of operation temperature on cell performance were also studied. Fig. 7 shows the Ragone plots of the GNS cells with BMP-DCA and organic electrolytes at 25 and 60  C. Since the IL electrolyte cell had a higher voltage (3.3 V vs. 2.5 V for the organic electrolyte cell) and higher capacitance (as depicted in Fig. 6), its plots were clearly in the upper right of the figure, reflecting superior ED and PD of the cell. Of note, increasing the temperature to 60  C significantly improved the performance of the IL cell. This is attributed to the temperature-induced ionic conductivity increase and viscosity decrease of the IL electrolyte, as shown in Table 4. In contrast, the organic cell performed similarly at the two temperatures. Although the conductivity and viscosity of the TEABF4/PC electrolyte also improved (but to a smaller extent, as shown in Table 4), this organic electrolyte was volatile and became unstable at an elevated temperature [41]. As a result, the cell performance did not distinctly improve with increasing temperature. It is worth notice that the IL electrolyte is also characterized by excellent thermal stability and Table 3 Rs and Ri for symmetric GNS cells with various IL electrolytes.

10

Power density (kW kg -1) Fig. 4. Ragone plots of symmetric GNS cells with various IL electrolytes.

Rs (ohms) Ri (ohms)

EMI-TFSI

BMP-TFSI

EMI-BF4

EMI-DCA

BMP-DCA

6.0 15.9

9.5 20.4

2.1 10.5

1.5 4.7

2.0 5.6

376

P.-L. Huang et al. / Electrochimica Acta 161 (2015) 371–377

2.5

0.2 mA 0.5 mA 1 mA

0.0

18 F g-1 21 F g-1

0

25 F g-1

100

Cell Voltage (V)

(b) Cell Voltage (V)

(a)

2.5

0.2 mA 0.5 mA 1 mA

0.0 17 F g-1 18 F g-1

200

0

Time (s)

100

200

Time (s)

(c)

(d) 3.3

0.2 mA 0.5 mA 1 mA

0.0

11 F g-1 14 F g-1

0

19 F g-1

100

200

Cell Voltage (V)

Cell Voltage (V)

20 F g-1

0.2 mA 0.5 mA 1 mA

3.3

0.0

43 F g-1 50 F g-1

58 F g-1

0 100 200 300 400 500 600

Time (s)

Time (s)

Energy density (Wh kg -1)

Fig. 6. Charge–discharge curves at various currents for AC ((a), (c)) and GNS ((b), (d)) symmetric cells with TEABF4/PC ((a), (b)) and BMP-DCA ((c), (d)) electrolytes.

(similar to those for other topics [44,45]) will be attempted in the future to take into account all the relevant variables and their interactions, aiming to find an optimal electrode/electrolyte combination for energy storage.

100

10

4. Conclusion

BMP-DCA (25 oC) BMP-DCA (60 oC) TEABF4/PC (25 oC) TEABF4/PC (60 oC)

1 0.1

1

10

Power density (kW kg -1) Fig. 7. Ragone plots of symmetric GNS cells with BMP-DCA and organic electrolytes at 25 and 60  C.

non-flammability [17], which makes it promising for EDLC applications, especially when safety and reliability are of major concern. The ED of 88 Wh kg 1 (or 103 Wh kg 1) and PD of 17.5 kW kg 1 (or 43.3 kW kg 1) at 25  C (or 60  C) shown in Fig. 7 are higher than those of the GNS-CMK-5/EMI-BF4 [22], GNS/BMIPF6 [42], poly(ionic liquid)-modified GNS/EMI-TFSI [25], and poly (ionic liquid)-modified GNS/PMP-TFSI [43] cells reported in the literature, indicating a great potential of the proposed GNS electrodes/BMP-DCA IL supercapacitor. A multivariate study Table 4 Ionic conductivity and viscosity of BMP-DCA IL and TEABF4/PC electrolytes. BMP-DCA

Conductivity (mS cm 1) Viscosity (cP)

TEABF4/PC

25  C

60  C

25  C

60  C

12.2

36.5

13.1

22.9

50.0

20.4

4.0

1.9

A combination of a GNS electrode and an IL electrolyte synergistically improved supercapacitor performance. The constituent anions of IL electrolytes determined the maximum specific capacitances of the GNS electrodes. The smaller the anion, the less the spitial hindrance of ionic adsorption/ desorption at the electrode, leading to a higher measured capacitance. It was also found that the Rs and Ri values for the cell incorporating a DCA-based IL electrolyte were lower than those found with BF4- and TFSI-based electrolytes. BMP+ has a more negative decomposition potential than that of EMI+; as a result, the IL with the former kind of cation possessed a wider potential window, which was beneficial for increasing the cell voltage. As a consequence, the BMP-DCA IL is the most promising electrolyte among those studied. A symmetric-electrode GNS supercapacitor with this IL electrolyte (V = 3.3 V) can provide an ED of 88 Wh kg 1 and a PD of 17.5 kW kg 1 at 25  C. The advantages of using IL electrolytes are even more pronounced at high temperature, since their conductivity increases and viscosity decreases while stability remains high. A significant cell performance improvemnet was found at 60  C (i.e., 103 Wh kg 1 and 43.3 kW kg 1). This GNS/IL supercapacitor has demonstrated attractive properties for high-energy and high-reliability energy-storage applications. Acknowledgements The financial support of this work by the Ministry of Science and Technology of Taiwan is gratefully appreciated.

P.-L. Huang et al. / Electrochimica Acta 161 (2015) 371–377

Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j. electacta.2015.02.115. References [1] S. Arico, P. Bruce, B. Scrosati, J.M. Tarascon, W. Van Schalkwijk, Nat. Mater. 4 (2005) 366. [2] P. Simon, Y. Gogotsi, Nat. Mater. 7 (2008) 845. [3] B.E. Conway, Electrochemical Supercapacitors: Scientific Fundamentals and Technological Applications, Kluwer-Plenum, New York, 1999. [4] G.P. Wang, L. Zhang, J.J. Zhang, Chem. Soc. Rev. 41 (2012) 797. [5] P. Simon, A. Burke, Electrochem. Soc. Interface 17 (2008) 38. [6] J. Yan, Z. Fan, W. Sun, G. Ning, T. Wei, Q. Zhang, R. Zhang, L. Zhi, F. Wei, Adv. Funct. Mater. 22 (2012) 2632. [7] Z. Tang, C.H. Tang, H. Gong, Adv. Funct. Mater. 22 (2012) 1272. [8] J. Yan, Q. Wang, T. Wei, Z. Fan, Adv. Energy Mater. 4 (2014) 1300816. [9] P. Hao, Z. Zhao, J. Tian, H. Li, Y. Sang, G. Yu, H. Cai, H. Liu, C.P. Wong, A. Umar, Nanoscale 6 (2014) 12120. [10] T. Cai, M. Zhou, D. Ren, G. Han, S. Guan, J. Power Sources 231 (2013) 197. [11] Y. Guo, Z.Q. Shi, M.M. Chen, C.Y. Wang, J. Power Sources 252 (2014) 235. [12] Y.B. Tan, J.M. Lee, J. Mater. Chem. A 1 (2013) 14814. [13] Y. Huang, J. Liang, Y. Chen, Small 8 (2012) 1805. [14] M. Segal, Nat. Nanotechnol. 4 (2009) 612. [15] P.J. Hall, M. Mirzaeian, S.I. Fletcher, F.B. Sillars, A.J.R. Rennie, G.O. Shitta-Bey, G. Wilson, A. Cruden, R. Carter, Energy Environ. Sci. 3 (2010) 1238. [16] Y.S. Li, I.W. Sun, J.K. Chang, C.J. Su, M.T. Lee, J. Mater. Chem. 22 (2012) 6274. [17] M. Armand, F. Endres, D.R. MacFarlane, H. Ohno, B. Scrosati, Nat. Mater. 8 (2009) 621. [18] A. Chiappone, F. Bella, J.R. Nair, G. Meligrana, R. Bongiovanni, C. Gerbaldi, ChemElectroChem 1 (2014) 1350. [19] Q. Lu, J. Yang, W. Lu, J. Wang, Y. Nuli, Electrochim. Acta 152 (2015) 489. [20] C. Liu, Z. Yu, D. Neff, A. Zhamu, B.Z. Jang, Nano Lett. 10 (2010) 4863. [21] K. Zhang, B.T. Ang, L.L. Zhang, X.S. Zhao, J.S. Wu, J. Mater. Chem. 21 (2011) 2663.

377

[22] Z. Lei, Z. Liu, H. Wang, X. Sun, L. Lu, X.S. Zhao, J. Mater. Chem. A 1 (2013) 2313. [23] C. Cui, W. Qian, Y. Yu, C. Kong, B. Yu, L. Xiang, F. Wei, J. Am. Chem. Soc. 136 (2014) 2256. [24] L. Zhang, F. Zhang, X. Yang, G. Long, Y. Wu, T. Zhang, K. Leng, Y. Huang, Y. Ma, A. Yu, Y. Chen, Sci. Rep. 3 (2013) 1408. [25] T.Y. Kim, H.W. Lee, M. Stoller, D.R. Dreyer, C.W. Bielawski, R.S. Ruoff, K.S. Suh, ACS Nano 5 (2011) 436. [26] F. Zhang, J. Tang, N. Shinya, L.C. Qin, Chem. Phys. Lett. 584 (2013) 124. [27] S. Mitani, M. Sathish, D. Rangappa, A. Unemoto, T. Tomai, I. Honma, Electrochim. Acta 68 (2012) 146. [28] J.W. Wu, C.H. Wang, Y.C. Wang, J.K. Chang, Biosens. Bioelectron. 46 (2013) 30. [29] C.H. Wu, C.H. Wang, M.T. Lee, J.K. Chang, J. Mater. Chem. 22 (2012) 21466. [30] D.R. MacFarlane, S.A. Forsyth, J. Golding, G.B. Deacon, Green Chem. 4 (2002) 444. [31] F.F.C. Bazito, Y. Kawano, R.M. Torresi, Electrochim. Acta 52 (2007) 6427. [32] A. Bagri, C. Mattevi, M. Acik, Y.J. Chabal, M. Chhowalla, V.B. Shenoy, Nat. Chem. 2 (2010) 581. [33] A. Gupta, G. Chen, P. Joshi, S. Tadigadapa, P.C. Eklund, Nano Lett. 6 (2006) 2667. [34] A.C. Ferrari, J.C. Meyer, V. Scardaci, C. Casiraghi, M. Lazzeri, F. Mauri, S. Piscanec, D. Jiang, K.S. Novoselov, S. Roth, A.K. Geim, Phys. Rev. Lett. 97 (2006) 187401. [35] H.C. Schniepp, J.L. Li, M.J. McAllister, H. Sai, M. Herrera-Alonso, D.H. Adamson, R.K. Prud'homme, R. Car, D.A. Saville, I.A. Aksay, J. Phys. Chem. B 110 (2006) 8535. [36] H.B. Zhang, J.W. Wang, Q. Yan, W.G. Zheng, C. Chen, Z.Z. Yu, J. Mater. Chem. 21 (2011) 5392. [37] F. Endres, A.P. Abbott, D.R. MacFarlane, Electrodeposition from Ionic Liquids, Wiley-VCH, Weinheim, 2008. [38] S. Zhang, N. Sun, X. He, X. Lu, X. Zhang, J. Phys. Chem. Ref. Data 35 (2006) 1475. [39] K.P. Wang, H.S. Teng, J. Electrochem. Soc. 154 (2007) A993. [40] X. Liu, Y. Wang, L. Zhan, W. Qiao, X. Liang, L. Ling, J. Solid State Electrochem. 15 (2011) 413. [41] M.T. Lee, Y.S. Li, I.W. Sun, J.K. Chang, J. Power Sources 246 (2014) 269. [42] Y. Chen, X. Zhang, D. Zhang, P. Yu, Y. Ma, Carbon 49 (2011) 573. [43] J.P.C. Trigueiro, R.L. Lavall, G.G. Silva, J. Power Sources 256 (2014) 264. [44] F. Bella, N.N. Mobarak, F.N. Jumaah, A. Ahmad, Electrochim. Acta 151 (2015) 306. [45] I.H. Hwang, K.I. Hong, K.S. Jeong, W.D. Jang, RSC Adv. 5 (2015) 1097.